Paul Mil

This first-of-its-kind systematic text treatment of the

economics of the modern firm:

• Focuses on the internal organization of the firm,

its boundaries, structure, and policies
• Offers a rigorous treatment of the material and
explains difficult or subtle points clearly and
logically
• Provides a balanced treatment of both incentive
problems and problems of coordination
• Discusses the key elements of economic theory
without presupposing an extensive economics
background and uses jargon-free language
• Uses up-to-date examples and boxed material to
introduce and illustrate every concept or theory,
showing how these apply in the real world
• Provides extended, managerial applications of the
theory to human-resources management,
compensation, corporate finance and
restructuring, and the vertical and horizontal
structure of the firm
• Features numerous learning aids, including "food
for thought" exercises, quantitative exercises,
extensive chapter summaries, and bibliographic
notes

ABOUT THE AUTHORS

Paul Milgrom, Professor of Economics at Stanford
University, has previously taught at Northwestern
University's Kellogg Graduate School of Management
and at Yale University in the School of Organization
and Management and the Department of Economics.
He was also the Ford Visiting Professor at the
economics department of the University of
California, Berkeley. He is the founding director of
the Stanford Institute for Theoretical Economics and
has lectured in Argentina, Belgium, Canada,
England, France, Israel, Italy, Japan, Spain,
Sweden, Uruguay, and throughout the United States.

Professor Milgrom's many publications include major

contributions in actuarial science, auctions and
competitive bidding, pricing strategies, securities
markets, and game theory and mathematical
economics. Recently his work has focused on
incentives and organizations, the economics of
manufacturing, theoretical explanations of historical
institutions, and complementarities. Much of his
work is joint with his co-author of this text, John
Roberts. (continued on back flap)
 ECONOMICS,
ORGANIZATION
AND MANAGEMENT

Paul Milgrom John Roberts

Stanford University Stanford University

Prentice Hall, Englewood Cliffs, New Jersey 07632

 Library of Congress Cataloging-in-Publication Data

Milgrom, Paul R.
Economics, organizations, and management / Paul Milgrom, John
Roberts.
p. cm.
Includes bibliographical references and index.
ISBN 0-13-224650-3
1. Managerial economics. 2. Organization. I. Roberts, John,
Feb. 11- 11. Title.
HD30.22.M55 1992
338.5'024658-<:lc20 91-41359
CIP

Editorial/production supervision: Alison D. Gnerre

Interior design: Donna M. Wickes
Cover design: Mark Berghash
Manufacturing buyer: Robert Anderson
Prepress buyer: Trudy Pisciotti

ii
- © 1992 by P<entice-HalJ, Inc.
A Division of Simon & Schuster
Englewood Cliffs, New Jersey 07632

All rights reserved. No part of this book may be

reproduced, in any form or by any means,
without permission in writing from the publisher.

Printed in the United States of America

10 9 8 7 6 5 4

ISBN D-13-224650-3

Prentice-Hall International (UK) Limited, London

Prentice-Hall of Australia Pty. Limited, Sydney
Prentice-Hall Canada Inc., Toronto
Prentice-Hall Hispanoamericana, S.A., Mexico
Prentice-Hall of India Private Limited, New Delhi
Prentice-Hall of Japan, Inc., Tokyo
Simon & Schuster Asia Pte. Ltd., Singapore
Ed1tora Prentice-Hall do Brasil, Ltda, Rio de I aneiro
 Dedicated to
Jan, Joshua, and Elana
and to
Kathy
with gratitude for their
patience, support and love
 CONTENTS

Preface xm

PART I THE PROBLEM OF ECONOMIC ORGANIZATION

0 1 DOES ORGANIZATION MATTER? 2

Business Organization 2
Crisis and Change at General Motors 2, Toyota 4, The Hudsons' Bay
Company 6, The Northwest Company 7
Organizational Strategies of Modern Firms 9
Salomon Brothers and the Investment Banking Industry 9
The Changing Economies of Eastern Europe 12
Recent History 12, Building Socialism 13, The Collapse
of Communism 15
Patterns of Organizational Success and Failure 16

0 2 ECONOMIC ORGANIZATION AND EFFICIENCY 19

Economic Organizations: A Perspective 19
Formal Organizations 20, The Level of Analysis: Transactions
and Individuals 21
Efficiency 22
The Concept of Efficiency 22, Efficiency of Resource Allocations 23, Efficiency
of Organizations 23, Efficiency as a Positive Principle 24
The Tasks of Coordination and Motivation 25
Specialization 25, The Need for Information 26, Organizational Methods for
Achieving Coordination 26
Transactions Costs AnalYsis 28
Types of Transaction Costs 29, Dimensions of Transactions 30, Limits of the
Transaction Costs Approach 33
\'
 Wealth Effects, Value Maximation and the Coase Theorem 35
The Value Maximization Principle 35, The Coase Theorem 38,
The Transaction Costs Approach versus Alternative
Views 39
Organizational Objectives 39
Profit Maximization 40, Other Goals and Stakeholders'
Interests 41
Modelling Human Motivation and Behavior 42
Rationality-Based Theories 42
Case Study: Coordination, Motivation, and Efficiency in the Market for
Medical Interns 43
Matching Problems and Failed Solutions 43, The Nation Intern
Matching Program 44, The Evolution and Persistence of Organizational
Forms 48

PART II COORDINATION: MARKETS AND MANAGEMENT 55

0 3 USING PRICES FOR COORDINATION AND MOTIVATION 56

Prices and Coordination 57

One Objective and a Single Scarce Resource 58, A Market-Clearing
Interpretation 60, Extensions and Difficulties 61
The Fundamental Theorem of Welfare Economics 62
The Neoclassical Model of a Private Ownership Economy 62, Scope of the
Neoclassical Model 68
Incentives and Information Transfer Under Market Institutions 71
Incentives in Markets 71, Informational Efficiency of Markets 72
The Neoclassical Model and Theories of Organization 73
Market Failures 73, Market Failures and Organization 77
Using the Price System within Organizations 78
Patterns of Internal Organization in Firms 78, Transfer Pricing in Multidi­
visional Firms 79

0 4 COORDINATING PLANS AND ACTION 88

The Variety of Coordination Problems and Solutions 90

Design Attributes 91, Innovation Attributes 92, Comparing
Coordination Schemes 93
Prices versus Quantities: Assessing Brittleness 94
Some Examples 94, A Mathematical Formulation and Analysis 96,
Constant and Increasing Returns to Scale 99
Economizing on Information and Communication 100
The Informational Requirements of Production Planning 100,
fudging Informational Efficiency 101, Planning with
Design Attributes 103
Coordination and Business Strategy 106
Scale, Scope, and Core Competencies of the Firm 106, Complementarities
and Design Decisions 108, Complementarities, Innovation Attributes, and
Coordination Failure 111
\!I Co11t1·11h
 Management, Decentralization, and the Means of CoonJinatio11 11:3
Centralization and Decentralization 114, The Role of Management in
Coordination 114

PART Ill MOTIVATION: CONTRACTS, INFORMATION, AND INCENTIVES 12S

0 5 BOUNDED RATIONALITY AND PRIVATE INFORMATION 126

Perfect, Complete Contracts 127

The Requirements of Complete Contracting 127, The Problems of
Actual Contracting 128
Bounded Rationality and Contractual Incompleteness 129
Bounded Rationality 129, Contractual Responses to Bounded Rationality
131, Effects of Contractual Incompleteness 13 3, Investments and Specific
Assets 134, Achieving Commitment 139
Private Information and Precontractual Opportunism 140
Bargaining over a Sale 140, Incentive Efficiency 143, Efficient Agreements
with Large Numbers of Participants 145, Bargaining Costs 147
Measurement Costs and Investments in Bargaining Position 147
The De Beers Diamond Monopoly 148, Investing in
Bargaining Advantages 149
Adverse Selection 149
Adverse Selection and the Closing of Markets 150, Adverse Selection
and Rationing 15 3
Signalling, Screening and Self-Selection 154
Signaling 154, Screening 156
Implications 159

0 6 MORAL HAZARD AND PERFORMANCE INCENTIVES 166

The Concept of Moral Hazard 167

Insurance and Misbehavior 167, Efficiency Effects of Moral Hazard 168, The
Incidence of Moral Hazard 168
Case Study: The U.S. Savings and Loan Crisis 170
The Savings and Loan Industry 170, Deposit Insurance and Risk Taking:
An Example 171, Incentives for Risk Taking with Borrowed Funds 173,
The Perverse Effect of Competition 175, Fraud in the S&Ls 176,
Who's to Blame? 176
Public versus Private Insurance 176
Other U.S. Government Insurance and Guarantee Programs 177,
Private or Public Insurance? 178, Moral Hazard in Private Life
Insurance 178
Moral Hazard in Organizations 179
Moral Hazard and Employee Shirking 179, Managerial Misbehavior 181,
Moral Hazard in Financial Contracts 183
Controlling Moral Hazard 185
Monitoring 186, Explicit Incentive Contracts 187, Bonding 189,
Do-It-Yourself, Ownership Changes, and Organizational Redesign 190
Contents vii
 lnfluenct' Activities and Unified Ownership 192
Unified Ownership and Selective Intervention 192,
Influencing Interventions 193, Influence Costs and Failed Mergers 193

T
P.\RT I\ EFFICIENT II\CENTI\'ES: CONTRACTS AI\D OWNERSHIP 205

0 7 RISK SHARING AND INCENTIVE CONTRACTS 206

Incentive Contracts As a Response to Moral Hazard 206

Sources of Randomness 207, Balancing Risks and Incentives 208
Decisions Under Uncertainty and the Evaluation of Financial Risks 209
Computing Means and Variances 209, Certainty Equivalents and Risk
Premia 210, Risk Premia and Value Maximization 211
Risk Sharing and Insurance 2 11
How Insurance Reduces the Cost of Bearing Risk 211, Efficient Risk Sharing:
A Mathematical Example 212, Optimal Risk Sharing Ignoring Incentives 213
Principles of Incentive Pay 21 i
Basing Pay on Measured Performance 214, A Model of Incentive
Compensation 215, The Informativeness Principle 219, The Incentive­
Intensity Principle 221, The Monitoring Intensity Principle 226, The Equal
Compensation Principle 228, lntertemporal Incentives: The Ratchet Effect 232
Moral Hazard with Risk-Neutral Agents 236
Problems with the Risk-Neutral Agent Scenario 237

0 8 RENTS AND EFFICIENCY 248

\rhen Distribution Affects Efficiencv 24-8

Efficiency Wages for �:mployment Incentives 250
The Shapiro-Stiglitz Model 250, A Mathematical Example: Comparative
Statics for Efficiency Wages 254, A Marxian View of Efficiency Wages 256,
Additional Aspects and Applications of Efficiency Wage Theory 257
Reputations as Contract Enforcers 259
The Elementary Theory: Reputations in Repeated Transactions 259,
Ambiguity, Complexity, and Limits of Reputations 264, The Advanced
Theory: Reputations Aided by Institutions 266
Rent-Set>king� Influence Costs and Efficient Decision Routines 269
Rents and Quasi-rents 269, Rent Seeking in the Public and Private
Sectors 270, Organizatior.al Design: Optimizing Influence Activities 273,
Influence Costs and the Legal System 277, Participatory Management 279

0 9 OWNERSHIP AND PROPERTY RIGHTS 288

The Conct>pt of Ownership 289

Residual Control 289, Residual Returns 290, Pairing Residual Control
and Returns 291
Tlw Coas•· Theorem Reconsidered 29;3
Ill-Defined Property Rights and the Tragedy of the Commons 294, Untradable
and Insecure Property Rights 297, Bargaining Costs and the Limits of the
Coase Theorem 300, Legal Impediments to Trade 301, Transaction Costs
\'Ill ( ,1111t1·11h
 and the Efficient Assignment of Ownership Claims 303, The Ethics of Private
Property 30 5
Predicting Asset Ownership 307
Asset Specificity and the Hold-Up Problem 307, Uncertainty and
Complexity 308, Frequency and Duration 310, Difficulty of Performance
Measurement 311, Connectedness 312, Human Capital 313
Ownership of Complex Assets 313
Owning Complex Return Streams 313, Who Owns a Public
Corporation? 314, Whose Interests Should Count? 315

PART V EMPLOYMENT: CONTRACTS, COMPENSATION, AND CAHEEHS 325

0 10 EMPLOYMENT POLICY AND HUMAN RESOURCE MANAGEMENT 326

The Classical Theory of Wages, Employment, and Human Capital 327

Wages and Levels of Employment 327, Human Capital 328, Defects of the
Classical Model 329
Labor Contracts and the Employment Relationship 329
Employment as a Relationship 329, Employment Contracts 329, Implicit
Contracts 332, Risk Sharing in Employment Relations 333, Borrowing and
Lending in Employment Relationships 338
Recruitment, Retention and Separation 338
Recruiting 339, Retention 344, Separations 347
Case Study: Human-Resource Policies in Japan 349
Hiring and Retention 349, Protecting Interests of Permanent Employees 3 50

0 1 1 INTERNAL LABOR MARKETS, JOB ASSIGNMENTS, AND PROMOTIONS 358

Internal Labor Markets 359

Labor Market Segmentation Patterns 359, Pay in Internal Labor
Markets 360
The Rationale for Internal Labor Markets 362
Long-Term Employment 363, Firm Specific Human Capital 363, Promotion
Policies 364, Pay Attached to Jobs 369, Internal Labor Markets as
Systems 3 71
Influence Costs, Incentives, and Job Assignment 375
A Job-Assignment Problem 376, Organizational Responses 378
Tenure and Up-Or-Out Rules 379
Tenure 380, Up-Or-Out Rules 382

0 12 COMPENSATION AND MOTIVATION 388

The Forms and Functions of Compensation 388

Differing Forms of Pay 388, The Obiective of Compensation Policy 390
Incentives for Individual Performance 391
What to Motivate? 391, Explicit Incentive Pay 392, Piece Rates 392,
Sales Commissions 396, Individual Incentive Pay in Other Contexts 399,
Eliciting Employees' Private Information 400, Implicit Incentive Pay 402
Performance Evaluation 403
Performance Evaluation with Explicit Performance Pay 403, Performance
Evaluation in Subiective Systems 404
Contn1ts IX
 Job Design 408
fob Design and Incentive Pay 41 0, fob Enrichment Programs and
Complementarities Between Tasks 411 , Responsibility and Personal
Business 41 2
Incentive Pay for Groups of Employees 413
Forms of Group Incentives 413, The Effectiveness of Group
Incentive Contracts 41 6
Pay Equity and Fairness 4 18
0 13 EXECUTIVE AND MANAGERIAL COMPENSATION 423

Patterns and Trends in Executive Compensation 424

CEO Compensation in Large U.S. Firms 424, Patterns and
Comparisons 425, Middle-Level Executives 427
Motivating Risk-Taking 429
The Puzzle 429, Managerial Investment Decisions and Human Capital
Risk 430, Inducing Risk Taking 431 , Paying for Investment Proposals 431
Deferred Compensation 432
Commitment Problems 433
Performance Pay for CEOs? 433
Setting CEO Pay 433, The Debate on Executive Compensation 434,
The Tasks and Temptations Facing Senior Executives 435, Value
Maximization and Incentives 436, The Evidence on Performance and
Pay 437, Does CEO Pay Affect Performance 441 , Implications and
Conclusions 44 3

PART VI FINA!\CE: INVESTl\lENTS, CAPITAL STRUCTURE, AND CORPORATE

CONTROL 447

0 14 THE CLASSICAL THEORY OF INVESTMENTS AND FINANCE 448

The Classical Economics of Investment Decisions 449

The Fisher Separation Theorem 449, Net Present Values 451 , Strategic
Investments as Design Decisions 454
Classical Analyses of Financial Structure Decisions 456
The Modigliani-Miller Analyses 456, The Allocation of Investment Capital
by Markets 459
Investment Risk and the Cost of Capital 460
Risk and Return 460, The Capital Asset Pricing Model 464, Expectations,
Asset Pricing, and Efficiency 466
Information and the Prices of Financial Assets 466
Forms of the Efficient Market Hypothesis 467, Evidence on the Efficient
Markets Hypothesis 469, Shortsighted Markets and Shortsighted Management
470, Implications of the New Theories for Organizations 473

0 15 FINANCIAL STRUCTURE, OWNERSHIP, AND CORPORATE CONTROL 482

Changes in Corporate Control: Patterns and Controversies 483

Corporate Control Changes in the 1 980s 483, The Rise of Debt 485, The
Debate 487, International Patterns of Financing and Ownership 489
X C1111t,·11h
 Financial Stru ct u re and Incentives 49 1
Confl.icting Interests: Managers versus Owners 491, Confl.icting In terests:
Current Lenders versus Other Capital Suppliers 494, Owners' Incentive for
Monitoring 496, Monitoring Incentives for Lenders 501, Default and
Bankruptcy Costs 502, Financial Structure, Incentives, and Value 505
Signalling and Financial Decisions 50.5
Debt and Equity 506, Dividends, Monitoring, and Signaling 507,
Objectives in Selecting Financial Structure 508
Corporate Control 508
The Mechanics of Control 509, Takeovers and Restructurings in the
United States in the 1980s 510, Takeover Defenses 515,
The Aftermath 520
Alternatives to the Publicly Held Corporation 521
Partnerships 522 , Charitable Activities and Not-for-Profit
Organizations 524

PART V I I THE DESIGN AND DYNAMICS OF ORGANIZATIONS 537

0 16 THE BOUNDARIES AND STRUCTURE OF THE FIRM 538

The Changing Nature of the Firm 539

Emergence of the Industrial Enterprise 539, The Development of the
Multidivisional Form 540, The Multiproduct Firm 542, Drivers of
Change: Complementarities and Momentum 543
The Internal Structure of the Firm 544
Advantages of the Multidivisional Form 544, Problems of Managing a
Divisionalized Firm 546
Vertical Boundaries and Relations 552
Advantages of Simple Market Procurement 553 , Advantages of Vertical
Integration 556, Alternative Vertical Relations 561
Horizontal Scope and Structure 568
Competitive Strategy and Organizational Innovation 569, Directions of
Divisional Expansion 569, Disadvantages of Horizontal Integration 572
Business Alliances 576
Keiretsu 579
0 17 THE EVOLUTION OF BUSINESS AND ECONOMIC SYSTEMS 585

The Present and Future of the Business Firm 586

Technological and Organizational Change in Manufacturing 586, The Service
Industries 588, Globalization of Economic Activity 589, Innovations in
Ownership, Financing and Control 590, Human Resources 591
The Present and Future of Economic Restructuring in Eastern
Europe and the USSR 591
The Communist and Capitalist Systems 592, Managing the Transition 593
The Future of Economics� Organization� and Management 594

Glossary 595

Index 607
Contt>nts XI
 PREFACE

Privatization, perestroika, hostile takeovers, leverage, mergers and spinoffs, executive

compensation, pay for performance, i nternal corporate reorganizations, strategic alliances,
employee ownership, the S&L debacle--every day the newspapers carry stories about
issues in economic organization. It is an exc iting time both for students of economic
organizations and for managers in these organizations, and an understanding of efficient
organization can contribute greatly to the economic health of the world.
This book addresses these issues-not as isolated phenomena but by using a consistent
framework of economic analysis. Although some current econom ics books include words
like "organization" or "the firm" in their titles, this is the first textbook to deal systematically
with firms and organizations as they really are and to acknowledge and analyze them in
their complexity. Simi larly, many management books are concerned with "organizations, "
but this is the first to adopt a thorough-going economic point of view and to use the
powerful insights of rigorous, relevant economic theory to derive the underlying principles
that are at work.
Outside of our teaching in the Economics Department and the Graduate School of
Business at Stanford University, courses based on the material treated in this book are
rarely found in either MBA or undergraduate economics programs. Some of our topics
are included in other economics courses-for example, principal-agent theory is now
occasionally taught cursorily in intermediate microeconomics courses, and the theory of
the vertically integrated firm is taught in some courses on industrial organization.
Nowhere else, however, are these treated with the other major issues of economic
organization in a unified and systematic way. Most business and management schools
offer courses that deal with many of the topics we cover, a few have courses that treat
most of these topics together, and a very few of the most progressive even do so using a
unified, economic approach, but without benefit of a textbook.
Our insistence on giving a unified treatment accounts for another unusual feature of
the book: It is part textbook and part research monograph . For example, the treatment
of problems of coordination in Chapter 4 goes well beyond anything that has yet appeared
in the scientific l iterature of the economics profession. Academic economists studying
organization have often focused too narrowly on the problem of incentives in organiza­
tions, or, even more narrowly, on incentives in relationships involving only two or
perhaps three people at a time. Successful business organizations, however, are complex
systems of mutually reinforcing parts, and our consistent emphasis on how the pieces of

xiii
 an efficient organization fit together into a coherent whole is unique to the economics
literature.
The uniqueness of this book also depends on how it is used. As an economics text,
our book is uniquely rich in applications of theory that a re drawn from real companies
and events. More than 1 00 examples a re woven into the text, illustrating how economic
pri nciples help to explain business institutions and practices, and every major idea is
introduced in the immediate context of a real-world example. As a management text, it
unrelentingly applies an economic perspective in which people make self-interested
choices and enter agreements only when they expect a mutual benefit. The power of
this economic perspective for explaining real-world institutions and phenomena surprises
many students and captures their interest. The economic slant gives management students
the conceptual "hooks" on which to hang the details of the cases they study, helpi ng
them to recogn ize puzzles and remember lessons long after the final exam or case analysis
has been graded and returned.
Why would we write a textbook for a subject where so few courses exist and with such
a seemingly small audience? Over the past decade the study of contracts and busi ness
organ izations has been one of the most active and fruitful areas of research in
microeconomics, and the pace of new research is quicken ing. With the successes of
Japan's unique organ izations, the creation of new institutions in the formerly commun ist
countries, the privatization of formerly publicly owned businesses in many countries
around the world, the major changes in corporate ownership and governance that are
still occurring, and the new emphasis on business alliances in Western countries, the
importance and currency of the subject of economic organization is undeniable. The
academic treatment of the principles of economic organization , however, has been
fragmented, in hibiting both professors who want to teach this subject and the students
who are eager to learn it. With the introduction of this book, a barrier has now fallen ,
and we expect a course based on the material covered here to become standard fare in
the business and econom ics curricula.
Although we have written this book for an audience of undergraduates and MBAs, it
is likely that doctoral students in various fields will find it useful as well, both for the
new research it contains and because there is nowhere else to turn for a unified and
systematic introduction to our subject. We also believe that practicing managers will find
it useful in organ izing thei r experiences and insights.

ORCANIZATI0N OF THE TEXT

We have organ ized the text into seven parts. The first deals with the fundamental
problems of economic organ ization , namely those of coordinating and motivating the
members of an organization to work in ways that are coherent and advance the common
interests of the organization's members. Chapter 1 , intended as a reading for the fi rst
day of class, illustrates these problems with a set of case studies. Chapter 2 develops the
economic perspective in more detail, introducing our focus on individuals and transac­
tions, the notion of efficiency, and the nature and classification of transacti on costs . It
also illustrates how these ideas are manifested in a wide variety of real institutions, both
public and private.
Part II is about coordination, both by the invisible hand of prices in markets and by
the quite visible hands of managers. Chapter 3 highlights the role of the price system
for bringing coherence to the diverse activities of people who may be unaware even of
one another's existence. We also explore the failures of the price system, but wi th an
unusual emphasis on the economies of scale (Chapter 3) and scope (Chapter 4) that are
fundamental for understanding the role of management in organizations. Alternative
sol utions to coordination problems arc studied in Chapter 4, where we compare the price
system to other systems of coordination and identify the situational factors that determine
X I \' Pn·fw 1·
the relative usefulness of alternative systems . In this chapter we also develop the
importance of complementari ties among activities and the need to look at successful
organ izations as coherent systems of mutually supporting parts.
Part III, which contains Chapters 5 and 6, introduces the problems of con tracti ng,
information, and incentives, and gives an informal treatment of their sol utiom. Chapter
5 deals with contracting with bounded rationality and informational differences and
incompleteness, and it also treats the problems of selecting partners and negotiati ng and
enforcing contracts. Adverse selection is treated here as well, along with signaling and
screening theories. Chapter 6 focuses on moral hazard and institutional responses to this
problem of post-contractual opportunism.
Part IV, consisting of Chapters 7 through 9, provides a careful, formal treatment of
some of the central methods of providing incentives efficiently. Chapter 7, which deals
with risk sharing and incentive contracts, introduces a set of theoretical principles for
efficient performance contracting and illustrates these principles with a series of sign ificant
applications. Whereas Chapter 7 focuses on situations in which issues of efficiency and
distribution can be treated separately, Chapter 8 investigates theories in which those
issues are irretrievably intertwined. Private ownership, the most pervasive and common
way to provide incentives for caring for assets, is the subject of Chapter 9.
The theory of Chapters 1 through 9 is put to the test in the next seven chapters. Part
V, consisting of Chapters 1 0 through 1 3, presents an economic treatment of the nature
of the employment relationship, examining explicit and implicit employment contracts,
compensation policies, and career paths. Chapter 1 0 reviews the classical theory of labor
markets and more modern theories that explain why long-term employment relations
predominate in every developed economy. Chapter 1 1 focuses on job assignments and
promotions and particularly on how firms resolve the tension between assign ing people
to the jobs that best suit them and using promotions as a reward for past performance.
The way compensation is determ ined puzzles almost everyone at some time in his or
her life, and that is the subject of Chapter 1 2. Chapter 1 3 reviews the facts and theories
about the controversial matter of executive compensation .
Chapters 1 4 and 1 5 (Part VI) treat financial decisions, particularly in investments,
capital structure, and corporate control . The classical theories of finance-present value
analysis, Modiglian i-Miller theory, the Capital Asset Pricing model, and the efficient
markets hypothesis-appear in Chapter 14, which is aimed mostly at undergraduates on
the presumption that MBA students see this material elsewhere. Chapter 1 5 introduces
students to explanations of capital structure based on information and incentive arguments
and to the contentious issues at the heart of the debate around corporate control and
governance.
The last part of the book comprises two chapters about the design, internal structure,
and dynamics of organizations, including an examination of the boundaries and scope
of business firms . One unden iable feature of organizational history is that the dominant
modes of organization have changed rapidly since the onset of the industrial revolution
and contin ue to change at a dazzling pace . In Chapter 1 6, we identify the problems that
major organizational innovations aimed to solve and the principles that have governed
past organizational choices. Chapter 1 7 looks ahead, emphasizing the dynamic processes
that drive organ ization change and peering into the future of capital ist firms, national
economies, and the study of the economies and management of organ izations.
Use in Courses
We have used earlier drafts of these chapters as teaching notes in our undergraduate and
MBA level courses at Stanford. Stanford quarters are 10 weeks long and involve 4 hours
per week of class time, plus a final exam period. The undergraduate course open s with
Chapter 1 on the first day, and proceeds to cover Chapters 2 through 9 and 14 through
Prefac1· X\'
 1 7 with varying degrees of thoroughness . Students find the techn ical material in Chapters
3 and 7 to be the most demanding reading, and it pays to spend a bit more time on
these chapters than on the others. We have found it useful to devote the last week of
the course to case studies-either those developed by students in term papers or business
school cases . Among the latter, the McDonalds case (Harva rd Business Review Reprint
744 1 0) and the Lincoln Electric Company (Harvard Busi ness School case 376-028) are
particularly appropriate.
In the MBA course, the principal focus of most class meetings is on case discussions.
Some class time is devoted to lectures about the more conceptually demanding
formulations, such as those in Chapter 7, but the other parts of Chapters 2 through 9
are used mostly as background readings that help students prepare the case analyses.
Chapters 1 0 through 17 are assigned as specific readings in connection with particular
case studies. Among the cases that we can recommend are the two classics mentioned
above plus Stanford Graduate School of Business cases Sony Corpora tion En ters the
En tertainment Business (S-BP265), and Marathon Oil Company's Last Stand (S-BP223),
and Harvard cases Analog Devices, Inc. A (9- 1 8 1 -00 1 ), Benetton SpA (9- 389-074 ),
Charles River Co. (9- 1 89- 1 79), Goodyear Restructuring (9-288-046), Lucky S tores, A, B,
and C (9-389-0 50, -0 5 1 , and -066), Merck & Co. , Inc. , A, B, and C (9-94 1 -005, -006,
and -007), Na tomas North America A and B (9- 1 84-03 1 and -03 2), RKO Wa rner Home
Video Inc. Incen tive Compensation Plan (9- 190-067), and Wash ington Post, A and B
(9-677-076 and -077). The Stanford cases are available from Case Services, Graduate
School of Busi ness, Stanford, CA, 94305-501 5, and the Harvard cases from HBS Case
Services, Harvard Business School, Boston, MA, 02 1 63.
In addition, both courses make significant use of material from the daily press,
especially The Wall Street Tournal and The New York Times' business section, as wel l as
The Economist and Business Week, as a basis for class discussion . It is remarkably easy
to find news stories almost every day that can be meaningfully and substantively analyzed
using the ideas from this book, and the use of such current material involves students
and raises their level of interest.

OUR DEBTS
This book evolved over several years of teaching courses at Stanford to undergraduates
and MBA candidates, and our first debt is to these students. They challenged and debated
us at every turn, comparing our economic perspective to competing ways of thinking
about organ izations . Without their comments, criticisms, and suggestions, this book
could never have been written . Among our students, Darryl Biggar, Sandro Brusco,
Xinghai Fang, Mark Fisher, Joshua Gans, William Lehr, Pino Lopomo, Jonathan Paul,
Scott Schaefer, and Joel Watson commented on parts of the manuscript, devised questions
and answers that we have borrowed, and generally made themselves very useful in the
preparation of this text. In addition, Stanford students Aaron Edlin, Rhonda Hollinberger,
Yongjae Lee, and Mike Saran made suggestions that we incorporated in the final
manuscript, and Chris Avery and Peter Zemsky collaborated with us in some of the
unpublished research that we draw on here.
Many of our colleagues and friends at Stanford and elsewhere gave generously of their
time to help us improve the manuscript in innumerable ways . For thei r help, we are
most grateful to Masahiko Aoki, Banri Asanum3, Connie Bagley, Jim Baron , Elaine
Bennett, Ted Bergstrom, Steve Durlauf, Dan Friedman, Nancy Gallini, Bob Gibbons,
Avner Greif, Jane Hannaway, Mark Hodes, Bengt Holmstrom, Chuck Horngren, Mike
Jensen, Pitch Johnson, Ed Lazear, John Litwack, Mark Lang, Susanne Lohmann,
Anthony Marino, Peter Newman, Yaw Nyarko, Masahiro Okuno-Fuj iwara, Myron
Scholes, Abraham Seid mann, Ken Si ngleton , Grace Tsiang, Karen Van Nuys, Mark
Wolfson, Barry Weingast, Robert Wilson , Ho-Mou Wu, and Mark Zupan.
X VI l ' n · fal'f•
 The entire manuscript was edited by Rachel Nelson and produced by Afoon Gnerre
of Prentice Hall, who labored hard to turn our sometimes incomprehensible jargon into
clear English prose. For their patience and skill, we owe them special gratitude. We also
thank Whitney Blake, who got this book project off to a healthy sta rt, and Kathleen
Much of the Center for Advanced Study in the Behavioral Sciences, who provided
remarkably speedy and insightful editorial advice. Debbie Johnston had the tedious but
crucial job of keyboarding the corrections to our man uscript. She did her work with
uncanny accuracy, freeing us to devote our time to other activi ties .
As we indicated above, large parts of what is reported here is new and comes from
our own research . So we also owe thanks to our resea rch collaborators, especially Bengt
Holmstrom, Dave Kreps, and Bob Wilson, and to those who provided financial
support for our research, including the National Science Foundation, the John Simon
Guggenheim Foundation, the Center for Economic Policy Research at Stanford, and
the Jonathan B. Lovelace professorship and the Robert M . and Anne T. Bass faculty
fellowship at Stanford. The fin ishing touches were put on this book while we were
Fellows at the Center for Advanced Study in the Behavioral Sciences.

Paul Milgrom
John Roberts

Pr<'fa,·1· xvii
 Part

I
THE PROBLEM
OF ECONOMIC ORGANIZATION

1
DoEs ORGANIZATION MATTER?

2
EcoNOMIC ORGANIZATION
AND EFFICI ENCY
 1
DOES ORGANIZATION MATTER ?

I 'I I
I

he principles of organization got more attention among us than they did

then in universities. If what follows seems academic, I assure you that we did not
think it so.
Alfred Sloan 1

BUSI NESS ORGANI ZATION

Crisis and Change at General Motors

When Pierre du Pont appointed Alfred Sloan to head General Motors in 1 92 1 , the
company was in crisis. The demand for cars had fallen in the 1 920 recession , and
although the firm already had huge inventories of unsold cars, the factory managers
contin ued to produce with abandon . In response to this falling demand, Ford Motor
Company had cut the price of its Model T by about 2 5 percent-a reduction that
GM with its higher costs could not match . Trying to hold the line on prices, GM
saw its sales fall by 75 percent between the summer·and fall of 1 920. By 1 92 1 , Ford
Motor Company's Model T held a 5 5 percent share of the U . S . automobile market,
compared to just 4 percent for Chevrolet and 1 1 percent for all of General Motors'
brands combined . Taking relentless advantage of the huge cost advantage that came
from producing a single product at very high volume, Ford was expanding its
production capacity in order to increase its already dominant position in the car
market.
Even apart from the immediate difficulties caused by the recession, General

1
Alfred Sloan , My Years With General Motors (Garden City, N. Y.: Doubleday, 1 964), p. 50.
Motors faced a fundamental long-term problem . It simply could not produce a car Docs O rgan i zation
that would offer more va lue at a lower price than the Model T, and it was squandering Matter?
the resources and capabilities it did have, as divisions like Cadil lac, Buick, Oakland,
Olds, and Chevrolet competed mostly with one another. What was needed first was
a new, more coherent marketing strategy focused squa rely on competing with Ford.
Sloan's plan was as follows: GM would design different cars for different segments of
the market. The Cadillac division would make luxury cars for the highest-income
buyers, and the other divisions each would serve successively lower-i ncome segments,
with Chevrolet making a model that would be sold for even less than the Model T.
The Model T would still appeal to some buyers, but Sloan bel ieved that most
customers would choose to buy cars that were either more luxurious or less expensive
than Ford's product.

PRODUCT DI VERSITY AND DESIGN COORDI NATION There was, however, one important
hitch: Carrying out this plan involved a combination of diversity of products and close
coordinati�n in design that exceeded anything that had ever been attempted before.
There would need to be a variety of new car designs, new dealersh ips, market
information about the customers in each new market segment, separate factories to
manufacture each type of car, and different supplies for each factory-a huge amount
of variety. At the same time, the many parts of the organization would need to be
coordinated in various ways. They would have to cover the different segments of the
market without competing too much with each other. They would need to share ideas
about how to improve products and reduce manufacturing costs, coordinate thei r
research and development efforts, cooperate with the supply divisions that produced
major components like bearings, radiators, and spark plugs, and standardize designs
enough to achieve economies of scale in parts production. Compared to Ford's one­
product strategy, Sloan's segmented-market strategy required that many more decisions
be made and much more information be continuously gathered and evaluated . The
organization used by Ford Motor Company would be no model for the new General
Motors.
The former organization of General Motors as essential ly a collection of car
companies and suppliers operating without any central direction was no model either.
The car divisions had failed to coordinate their parts designs, raising costs for all of
them. The accounting system, which allocated costs among the producing divisions,
was unable to keep accurate track of which decisions by which units raised costs and
therefore failed to guide divisional managers to economize. For example, the individual
divisions continued production even in the face of huge accumulated inventories of
unsold cars during the recession because the system failed to assign inventory holding
costs to the divisions.

GM's MULTI Dl \'ISIONAL STRUCTURE Sloan studied GM's organization and decided a
radical change was needed. The new organization would be a multidivisional structure
with a strong, professional staff in the central office. There would be no infringement
on the basic autonomy of the divisions in making operational decisions. Each separate
division would make and sell a car targeted for an assigned market segment. Each
would have its own managerial team with authority to make its own operating
decisions. Unlike other business organizations, GM's central office would not be
responsible for day-to-day operations. Instead, its primary roles would be to audit and
evaluate each division's performance and to plan and coordinate overall strategy. The
central office would also be responsible for the research, legal, and financial functions
of the corporation. It would survey market prices to make sure that the prices used
for internal accounting purposes were a good reflection of actual costs . This would
The Problem of enable the company to evaluate internal supply divisions on the basis of their
Economic profitability, as if each were a separate firm . 2
Organ ization Henry Ford, accustomed to being well informed about every important decision
in his company, was skeptical about GM's reorganization and especially about how
far its top management would be removed from its operations. He commented:

To my mind there is no bent of mind more dangerous than that which

sometimes is described as the "gen ius for organization . " This usually
results in the birth of a great big chart showing, after the fashion of a
family tree, how authority ramifies. The tree is heavy with nice round
berries, each of which bears the name of a man or an office . . . . It takes
about 6 weeks for a message from a man living in one berry at the lower
left-hand corner of the chart to reach the president or chairman of the
board . 3

However, Sloan's organization , which looked so cumbersome to Henry Ford , quickly

transformed GM into a fearsome competitor. From 1 927 to 1 937, Ford lost $200
million , whereas GM earned over $2 billion . GM's market share grew to 45 percent
in 1 940, whereas Ford's once commanding share shrunk to a mere 16 percent.
The creation of a multidivisional structure not only enabled General Motors to
compete successfully with its new strategy, it also set the stage for a continuing
expansion of the company's product line. In the years that followed, General Motors
added products ranging from trucks to kitchen appliances to its product portfolio.
Such an expansion would not have been possible using the older forms of business
organ ization. The multidivisional form that GM helped to pioneer has become a
standard organizational feature of the corporate world, enabling many companies to
produce a wide array of products. General Motors' new organization was well suited
to its needs, but GM was hardly the last automobile company to discover the
advantages of organizational innovation .

Toyota
In the early 1 9 50s, Toyota was a small automobile manufacturer serving the Japanese
market. Compared to its giant U. S. competitors, Toyota suffered from a drastic lack
of capital and a tiny scale that made it impossible to match its competitors' low
production costs. Although Toyota enjoyed much lower labor costs than the U . S .
firms a t the time, many other countries had even lower labor costs. However, none
had been able to parlay these low costs into a substantial competitive advantage in
what was then a high-technology, capital-intensive industry.

"'JL1sT-IN-Tt.t\tE'' l\hNUFACTL1 RING Like other automobile compan ies in Europe,

Japan, and the Eastern Bloc, Toyota tried for a time to mimic the advanced mass­
production techniques of its U . S . competitors. Soon , however, under the leadership
of Eiji Toyoda and Taiichi Ohno, it began to develop a distinctive approach that was
better suited to the scale and nature of its operations. One of the most fa mous of
Toyota's innovations was the development of the "kanban" or "just-in-time" (JIT)

2 Similar organizations were independently developed in the same era by the Du Pont Company,
Standard Oil of New Jersey, and Sears. The fascinating history of the emergence of the multidivisional
firm is told by Alfred Chandler in St rategy and Structure: Chapters in the History of the American Industrial
Enterprise (Cambridge, MA: M IT Press, 1962).
3 As quoted in Alfred Chandler, "The l listorian and the Enterprise, " Ninth Annual David R.
Calhoun, Jr. Memorial Lecture delivered at Washington University (March 29, 1988); and Alfred Chandler,
Strategy and Str11 ct11re.
manufacturing system. Th is system was ideally intended to eliminate all inventories Docs Organ i zation
from the production process. In traditional manufacturing industries, goods processed Matter?
on one machine would be held in a buffer inventory until the next mach ine in the
sequence was ready for its operation . Inven tories separating the successive stages
protect each machine's operations from delays or disruptions at adjacent stages of
production. However, inventory systems are subject to very large economies of scale,
so Toyota could never achieve cost pa rity (equality) with its larger competitors if it
relied on such a system .
In place of inventories, Toyota established a system of closer communication
and tighter coordination between successive stages of the production process, so that
each stage would be informed "just in time" when it had to deliver its product to the
next stage. Without inventories to buffer the disruptions caused by defective products
and broken machines, Toyota engineers had to work to improve the rel iability of every
step of the process. The same changes that reduced the number of interruptions in
the production process often reduced the number of defects in Toyota's cars as well,
as flaws were caught immediately rather than piling up in the in-process inventory.
The absence of inventories also meant that Toyota had to be linked more tightly to
its suppliers than were U . S . firms, communicating with them about day-to-day needs
and helping them to improve the reliability of their own systems. At the same time,
the need to repair broken equipment quickly caused Toyota to tra in its equipment
operators to carry out maintenance and repairs themselves. In contrast, ma intaining
and repairing machines was a separate specialty with a sepa rate job classification in
the United States, and when a machine broke down, its operator stood around waiting
until a repair specialist turned up to fix it.
Under Alfred Sloan's leadership, General Motors moved to take full advantage
of the large scale of its operations by including the same parts in many of its different
cars. By using the same chassis or engine or brakes in several models, GM
could afford to develop special ized manufacturing equipment for these components,
substantially reducing the production costs. In contrast, Toyota did not enjoy such
scale economies and instead emphasized improving the flexibility of the equipment
it did use, so that the same equipment could be quickly reset to produce different
models. Given this emphasis, it should come as no surprise that Toyota had become
a world leader in the use of industrial robots as early as the I 960s.
Because General Motors' specialized equipment was not easily adapted to the
production of radically new designs, GM had major redesigns of its models only about
once every 1 2 years in the 1 9 50s and 1 960s, whereas Toyota redesigned its vehicles
twice as often, introducing improvements with each new design . By the early 1 970s,
Toyota's technological prowess in the design and manufacture of small cars had earned
it a considerable share of the world ma rket. As its sales grew, Toyota built new
manufacturing plants that were much larger than those built earlier by the U . S. car
companies, allowing the company to enjoy many of the scale economies that remained
in the production process.

COORDINATION WITH OUTSIDE SUPPLIERS Another feature that distinguished the

Toyota organization from its North American competitors was Toyota 's great reliance
on outside suppliers. In contrast, GM was very highly vertically integrated. GM's
huge volumes and use of the same components for many models allowed the compa ny
to utilize fully the output of efficient-sized parts factories. Furthermore, ready access
to capital allowed it to produce these parts and components itself. In its early days,
Toyota could not achieve efficient sca le in the building of components, nor could it
afford to own its own parts makers. Therefore, unlike GM, Toyota chose to rely on
outside suppliers not j ust for its basic inputs like sheet steel, screws, and fabric for
 6
The Problem of seats, but also for the more complex components and systems, such as headlamps,
Economic brake systems, and fuel-injection systems. This also opened the possibility that its
Organization suppliers then could achieve larger scale by producing for other auto manufacturers
as well .
The J IT system necessitated close coordination between Toyota and its suppl iers.
The need was reinforced by the frequent redesign of the vehicles and the fact that the
suppliers were providing high-level components that had to fit together, rather than
simple standardized commodities . As a result, simple market arrangements with the
suppliers became problematic. Instead of seeki ng numerous suppliers for each part or
component and shifting business among them to induce price competition , as CM
did, Toyota built long-term relations with a much smaller number of suppliers. These
long-term relations facilitated communication and made the suppliers willing to face
the risks of investing heavily in both skills and machinery to meet Toyota's specialized
needs.

The Hudson's Bay Compan y

The history of the automobile industry provides a clear example of the importance of
a coherent organization that is well suited to the firm's size, capabil ities, and market
strategy. But the importance of the deta ils of business organization can also be seen
from centuries past.
On May 2, 1 670, the Governor & Company of Adventurers of England Trading
into Hudson's Bay was formed as a joint stock company by a royal charter of King
Charles I I of England. The charter gave the company a monopoly over trade in all
the lands draining i nto Hudson's Bay. 4 This is an i mmense area of 1 . 5 million square
miles, coveri ng much of Quebec, most of Ontario, all of Manitoba, much of
Saskatchewan a nd Alberta, the eastern part of the Northwest Territories, parts of
Minnesota, North Dakota, and Montana, and a bit of South Dakota. The area is
larger than 10 Japa ns, 1 5 United Kingdoms, or 30 states of New York. The company's
total legal monopoly nominally covered all trading of goods in the region, but in fact
the firm was in the business of trading European manufactures to the native peoples
for animal furs, and especially beaver pelts .
Now known as the Hudson's Bay Company (HBC), the firm is still in existence.
It is the world's oldest commercial entity that continues its original line of business .
I n the late eighteenth and early nineteenth centuries, however, it was in desperate
shape, being thoroughly beaten by a rival, the North West Company (NWC), that
operated under a minor legal deficiency (it was violating the royal monopoly) and
what should have been an absolutely debilitating technological inferiority. However,
the NWC used a m uch more effective organizational structure and strategy that put
it closer to its customers than the HBC, encouraged more flexible, effective responses
to changing conditions than the older company could manage, and gave its employees
stronger incentives for initiative a nd effort. This strategy and structure more tha n offset
the older firm's huge cost disadvantage. Only when the HBC mimicked key aspects
of the NWC's strategy and structure was it able to compete effectively.
HBC's ORGANIZATION The stock in the Hudson's Bay Company was owned by a
group of wealthy aristocrats in the United Kingdom. Management was provided from
London by a committee of the owners under the leadership of one of their number,

4 A well-researched a nd extremely readable source on these matters is Newman's two-volume history

of the H udson's Bay Com pan y . This is the source of most of the details re ported here. See Peter C.
Newman , Company of Adventurers and Caesars of the Wilderness ( London , New York, and Toronto:
Pen g uin Hooks , 1 988).
 7
the Governor. None of these worthies ever set foot in the area of the company's actual Docs Orga n i zation
operations during the seventeenth and eighteenth centuries, and they were often Matter?
mon umentally ignorant of the conditions that prevailed in the field . They recruited
employees to carry out the trade in North America and paid them a flat salary. Both
the general strategy that these employees were to implement and many of the finest
particulars about operating decisions were made in London (along with detailed rules
regarding the employees' conduct and behavior). Of course, communication between
London and Hudson's Bay was extremely slow in the days of sail, especially because
the bay is frozen solid for much of the year. Therefore, the lag between the reporting
of information and the receipt of a response from management was frequently as lo�1g
as 1 5 months.
The company's strategy was to build a few trading posts along the shores of
Hudson's Bay and to trade only there. Native tribes near the bay were the in itial
customers, but they rapidly became independent middlemen, obtain ing furs from
more distant peoples in return for goods obtained from the company. Th is pattern
had the usual inefficiencies of"double monopoly": The intermediaries took a sign ificant
markup for their services. However, the strategy fit well with the HBC's personnel
and pay policy and with its management's preference for control . In addition, the
employees in the field for the most part had no interest in leaving the relative safety
and comfort of their dismal "forts" and "factories" to risk the barren wilderness in
hopes of increased company profits but with no extra reward for themselves.

The North West Com p an y

Until the surrender of French claims in Canada to the Un ited Kingdom in 1 76 3, the
H BC faced intermittent competition from French Canadian coureurs de bois,
independent traders from the St. Lawrence who traveled by canoe to trade with the
native peoples directly on their home trapping grounds. However, the colon ial
authorities in New France frowned on such trade, fearing it would lure men away
from farming and foster too much independence, and they taxed the returns peavily,
quite often simply confiscating the furs. Indeed, the HBC itself had been establ ished
at the instigation of two such entrepreneur-explorer-traders, Radisson and Groseilliers,
who tired of losing their revenues to the French government.
Once British government and commercial law came to Canada, Scottish and
other English-speaking immigrants to Montreal establ ished an effective trade based
on the coureurs de bois model which competed with the Hudson's Bay Company. By
1 779 these traders had formed a partnership called the North West Company, but
even before then they had established permanent fur trade posts as far away as the
Athabasca River delta, 3,000 miles northwest of Montreal.
THE NWC's PROBLEM The North West Company (NWC) suffered under an immense
technological disadvantage, but they suffered heroically. Montreal and the Hudson's
Bay forts of the HBC were approximately the same sailing distance from the source
of trade goods and the market for furs in England . Montreal, however, was separated
from the prime fur country by an extra 1, 500 miles of trackless swamp, bare rock,
and impenetrable bush. The Nor'Westers' (North West Company traders') solution
was to move the trade goods to the north and bring the furs back in fragile birch-bark
canoes powered by French Canadian voyageur paddlers. This solution was possible
because the interlocking system of Canadian rivers and lakes permits travel from the
Atlantic to the Arctic and Pacific Oceans with very few, reasonably short land bridges.
Nevertheless, using this system presented mon umental challenges.
North West Company canoe brigades would leave the fur-trading posts in the
north as soon as the ice left the rivers, traveling across half the North American
 8
The Problem of continent, through absolute wilderness. along fast, rocky rivers and across storm-swept
Economic lakes, often having to lug their canoes and all the goods they carried around raging
Organization rapids . Thei r destination was the company's inland headquarters, first located at Grand
Portage on the northwest corner of Lake Superior, just south of the current border
between Canada and the United States, and later moved 40 miles north to Fort
William. There they would meet canoe brigades that had come up from Montreal ,
loaded with trade goods, via the Ottawa River and Lake H uron . The trade goods and
furs were exchanged, then the Nor'Westers would retrace their paths back to the fur
country, where they would spend the wi nter trading with the native peoples.
Meanwhile, the furs would be carried back to Montreal , where they would (if the
river was sti ll open) be loaded on ships for England. As a result, there was a mini mum
lag of 1 5 to 1 8 months between the NWC's purchase of trade goods in England and
its sale of the furs it traded for them, with 24 months a more common length of ti me.
In contrast, trading at its Hudson's Bay forts allowed the HBC to sell furs within four
to six months of buying the goods they traded for them . The difference in working
capital needs, combined with that in transportation costs, should have made the HBC
domi nant, for its costs for importing goods were half those of its rival.

THE SUCCESS OF THE NORTH WEST COMPANY In fact, by the end of the first decade
of the nineteenth century, the NWC had sei zed nearly 80 percent of the trade and
was immensely profitable, whereas the HBC was losing money at such a rate that its
officers considered getting out of the fur business altogether. A key factor in the
Nor'Westers' success was thei r strategy of building trading posts in the fur lands, which
put them close to their customers and gave them a market advantage. But at least as
important was their organizational structure, which embodi ed systems of i ncentives
and decision making that encouraged the effort, imagination, flexibility, and innovation
that were crucial to maki ng the market-oriented strategy work.
The HBC was rigidly hierarchical. Rules and controls from distant London
circumscribed every action and decision of its employees in the field, leaving little
possibil ity of flexibly responding to emergi ng conditions. In novation was discouraged
or even pun ished, and performance was rewarded only by the possibility of someday
gaining a promotion to the next rung of the bureaucratic ladder. Employees were
chosen for their ability to withstand the excruciati ng boredom of their indentured
terms of service by the frozen bay and for their willingness to work cheaply and follow
orders. They were disciplined by floggings for breaking the company's myriad
regulations.
Organized and managed this way, the HBC was no match for the NWC. The
NWC was a partnership, with two classes of partners who shared in the profits of the
enterprise. The senior, Montreal-based partners were responsible for acquiring trade
goods and financing and for marketing the furs at the London auctions. The "wi nteri ng
partners" ran the trade in the field. The two groups met annually at Grand Portage
or Fort William to exchange information , set policy, and divide the profits that arose
from their efforts. Operating decisions in the field were largely left to the individual
wintering partners on site, who could respond quickly and i maginatively and who
were well motivated by thei r ownership shares. Other employees were chosen for their
fit with the aggressive, entrepreneurial style that characterized the North West
Company and its partners. These people were given real responsibility, performance­
related pay, and a serious chance to become a partner.

Tm: HEOHC.\N IZATIO� OF I lt rosON'S 8.\r COMPANY The HBC ultimately responded
to the NWC's challenge, beginn ing in 1 809 when n ew owners gained effective control
of the company after its share price had fallen from £2 5 0 to £60. The response was
 9
simple: mimic their rival, build trading posts inland to compete directly with the Docs Organ ization
Nor'Westers, institute a profit-sharing scheme that allocated half the profits to the Matter?
officers in the field based on performance, give other employees more incentives and
more freedom of action, and recruit a new class of employees who would respond to
this new organizational strategy.
The Nor'Westers immediately saw the danger that the HBC's new strategy
presented, for they always were painfully aware of the cost disadvantage they faced .
Their immediate response was an attempted hostile takeover. They sought to buy a
controlling block of shares in their rival so as to gain access for themselves to the short
transportation route through Hudson's Bay. For once, being based in Canada rather
than London was a disadvantage, for before they could implement their strategy, the
shares in question had been purchased by the new, aggressive owner-managers and
their allies .
· The HBC managed to transform itself remarkably quickly from a feudal royal
monopoly · to an effective commercial competitor, and its inherent cost advantage
became decisive. The competition continued fiercely and even bloodily for a decade,
but by 1 820 the NWC essentially was beaten. The competition ended in a merger
that nom inally treated the two rivals as equals but in fact gave control to the victorious
HBC. However, the NWC's aggressive spirit survived in the merged company, which
expanded its range of successful operations in the next half century across the Rockies
to the Pacific and even to the Hawaiian Islands and Asia. Today, the HBC has left
the fur business, but it is a major real estate firm and the owner of the largest chain
of department stores in Canada.

ORGANIZATIONAL 8TRATECIES OF MODERN FIRMS

Providing incentives through compensation and ownership was an important element
in the NWC's challenge and in the HBC's successful response. Design of the
compensation and ownership structure continues to be an important feature of the
organizational strategies of modern firms. A current example from the financial­
services industry involves some interesting twists.

Salomon Brothers and the I nvestment Banking I ndustry

Salomon Brothers is a major investment bank headquartered in New York City. Like
many other investment banks, Salomon was originally organized as a partnership, but
it became a publicly traded corporation after the partners sold their ownership claims
in 198 1 to Phillips Brothers, a firm in the commodity-dealing business. In 1984, John
Gutfreund, the head of the Salomon subsidiary, became CEO of the parent company,
which had been renamed Phibro. The corporate name was subsequently changed to
the current Salomon Inc. During the mid- l 980s Salomon was reputed to be the most
profitable of all the Wall Street investment banks and, on a per employee basis, the
most profitable corporation in the world. Its stock sold in the neighborhood of $60
per share.
Investment banking emerged as a separate industry in the United States after
1 9 34, when the Glass-Steagall Act prohibited commercial banks, which accept deposits
and make loans, from underwriting securities as they had previously. Investment banks
arose to take over the underwriting function. 5 Modern investment banks, including

5 Underwriting the issue of a new security fi rst involves the underwriter's agreeing on a price for the
security with the issuing entity. The underwriter pays this price for the security, then markets it to investors.
The underwriter thus bears the risk that the security will not be saleable at the agreed price.
 10
The Problem of Salomon, are involved in two main activi ties: corporate fi nance, and sales and tradi ng.
Economic The former i ncludes underwriting new securities issued by corporations, governments,
Organization and not-for-profits; advising on, helping with , and organizing financing for mergers,
acquisitions, divestures, and financial restructurings; and advising on corporate
fi nancial policy. Sales and trading involve the trading of fi nancial instruments (stocks,
bonds, options, warrants, mortgage-backed securities, and so on) both for resale and
for the firm's own investment account, and the sale of these i nvestments to (large)
individual investors and institutions.

BOND TRADING AND THE CULTURE OF SALOI\ION BROTHERS Salomon Brothers always
has been particularly strong in "fixed i ncome" sales and trading. The general public
tends to think of Wall Street in terms of buyi ng and selling stocks . However, in the
1 980s the real sales and trading action was in bonds, spurred by the volatility of
interest rates that marked the period combined with the huge growth of the U . S .
government debt and the debts of U . S. corporations and i ndividuals. Because the
price of bonds and other securities that pay a fixed rate of i nterest varies inversely with
market interest rates, 6 volatile rates meant that bond prices could swi ng wildly, creating
opportunities for speculation (buying or selling bonds in the hopes of realizing large
profits when prices change, but risking comparably huge losses) and arbitrage (the
process of buying and selling to make a nominally riskless profit by taki ng advantage
of price disparities between markets or equivalent assets). The growth of i ndebtedness
meant that there were a great many bonds to trade, especially after Salomon traders
developed new securities that packaged together home mortgages to form tradable
fi nancial instruments. Thus, even before the explosion of "junk bond" activity centered
on Michael Milken of the rival investment bank of Drexel Burnham Lambert, huge
profits were being made in bond sales and trading. Salomon traders were the leaders
in this busi ness.
Bond trading is a frenzied business in which immense sums of money are at
risk. Traders will buy hundreds of millions of dollars worth of bonds in a day, acting
on very limited i nformation and under extreme time pressures from competition.
Each trader has a pai r of telephones and spends much of the day with one in each
ear, screaming orders. Tiny movements in interest rates and bond prices result in
gigantic gains and losses. The risks are monumental , but the profits that come from
successful trading are commensurate. The sort of people who are attracted to this
work are an especially individualistic, risk-taking, competitive lot. The leaders of the
firm, including Gutfreund , are former bond traders, and they have set the style of the
corporate culture at Salomon Brothers. In this culture, performance as measured by
profits generated is the sole source of power, prestige, status, and respect. 7

THE PERFORI\IANCE- PAY SYSTEI\I Salomon's compensation system for its employees,
from the clerical staff through the approximately 1 50 managing directors and the more
senior executives, i nvolves a base salary plus an annual bonus. The bonus is determined
in a fashion that approaches an individual piece-rate system . Almost every transaction
is separately priced, wi th charges for cred it risk a nd overhead , and the resulting profit

6
Consider a bond with a maturity val ue of $ 1 00 that pays $ 5 in interest per yea r-a 5 percent rate
of interest. Now suppose that interest rates rise to I O percent, so that newly issued bonds with a maturity
va lue and price of $ I 00 must pay $ I O per year in interest. Then i nvestors will be unwilling to pay as much
as $ I 00 for the old bond, and its price will fall to a level where it again becomes an attractive investment.
For a ten-yea r bond with a maturity val ue of $ I 00 to yield I O percent, its price must fall to $74. 28.
- This culture is described very entertain ingly in the best seller by l\l ichael Lewis, Lia r's Poker
( New York: W. W. Norton & Co. , 1 989).
 11
is credited to the individuals or departments involved . The individual bonuses then Docs Organ ization
are determined on the basis of performance eval uations in which these calculated Matter?
profit contributions play a key role. However, the bonus payments actually arc
negotiated; the bonus is not si mply computed by formula from the profit contribution.
Because the bonus payments can be extremely large-million-dollar bonuses
are not uncommon and the bonus is commonly two thirds of aggregate earnings­
the bonus system is a key incentive device in the firm. This performance-pay system
is quite elaborately defined and extensively applied by the standards of most businesses.
It has very effectively encouraged Salomon's people to work extremely hard and to
take great risks to increase both their own and their departments' profits. By doing so,
it has rei nforced the firm's fiercely competitive corporate culture.
The system has not, however, encouraged cooperation between departments.
For example, if corporate bond traders obtained information that might be valuable
to Corporate Finance, they might not bother to pass it on because there was nothing
in it for them. It also has caused managing directors to focus excessively on their
individual departments and accounts, rather than developi ng the firm as a whole and
building its long-term success. Finally, the performance-pay system has resulted
sometimes in highly dysfunctional attempts to "steal" other departments' profits.

STOCK OWNE RSHI P In May 1990 Salomon Brothers attempted to respond to these
drawbacks by reform ing its bonus scheme. Myron Scholes, a finance professor on
leave at the firm from Stanford University, was instrumental in designing the solution .
Under the new scheme, once the individual bonus has been determined, a fixed
percentage is withheld and used to buy stock in the firm. This stock is purchased in
the open market, so that the capitalization of the firm will be unchanged . The stock
then is held in a trust for the employee, who will not be able to withdraw it for five
years. In effect, the value of each employee's current bonus is tied to the overall
market value of the firm five years in the future. The plan covers all employees,
although there are different percentages applied to those at the managi ng director
level and above. It is estimated that in five years the employees will own 20 percent
of the firm.
The plan won acceptance because of the incentive effects it embodies. The
explicit aims were to change the culture, to encourage a long-run perspective and
cooperation, and to al ign the employees' interests with one another and with those of
the stockholder owners. It was also expected that this scheme would influence favorably
the type of people who would be attracted to work for Salomon . In the long run, this
"self-selection" or "clientele" effect on recruitment and turnover could be very
important. Of course, the plan puts a major new element of risk into the employees'
personal incomes because they no longer receive cash that they can invest as they
choose in a safely diversified portfolio. Instead, they are locked into a single stock
whose price might move up or down . One element in the acceptance of the plan was
that the company was sell ing essentially for the book val ue of its assets at the time the
program was introduced; therefore, an upward movement in its stock price was
relatively likely. Nevertheless, the firm also attempted to offset the increased risk to
some extent by specifying that it would buy 1 5 percent more stock for the employees'
trust than the bonus pool would have generated .
Many companies attempt to provide incentives to employees, and especially to
senior executives, through stock ownership. Employees may be allowed to buy stock
at a reduced price, and executive bonuses are often paid in stock or in options that
give the right to buy the company's stock in the future at a fixed price. Salomon's
technique of using the trust helps ensure that the employees will actually hold on to
the stock rather than sell it soon after receiving it. The intent is to ensure that
 12
The Problem of employees will be concerned with the fi rm's long-run performance. 8 The trust also
Economic has tax advantages, because any appreciation of the price of the stock that occurs
Organization while it is being held in the trust is free of capital gains taxation. 9
Several years' experience will be needed before we fully can judge the effectiveness
of this innovation in compensation. However, the system did pass a key market test:
As of 1990, announcement of the plan appeared to have raised the price of the 1 00
million shares of outstanding Salomon stock by between $ 1 . 50 and $2 per share from
a base of $2 1 . Thus, the plan raised the market value of the firm by between 7 and
10 percent, and the present value of the firm's profits was estimated to have risen by
at least $ 1 5 0 million as a consequence of adopting the plan. Because the plan's direct
cost to the fi rm is, if anything, slightly higher than the original cash-only bonus
(because of the promise to buy the extra 1 5 percent more stock for the trust), investors
in the market estimate that the new incentives' effect on revenues will be even greater
than the direct profit effect.

THE CHANGING ECONOMIES OF EASTERN EUROPE

Salomon Brothers' pay reforms are an example of experimentation with ownership
structure and incentive plans at the level of the individual enterprise. A much more
dramatic experiment at a macro level was being carried out in the late 1980s and
early 1990s in Eastern Europe, where whole economies were being redesigned.

Recent History
The world was shocked and the Western capital ist democracies were stunned in 1 957
when the Union of Soviet Socialist Republics launched Sputnik, the first space
satellite, and then quickly followed this triumph by putting the fi rst man in space.
Barely a dozen years before, the Soviet Union had emerged from the Second World
War having suffered the deaths of 1 4 million of its citizens, both in battle and through
deprivation, and the occupation and destruction of large portions of its territory. Before
the war the Soviet economy had been rapidly industrializing under Stalin's Five Year
Plans, but the progress was uneven and the Soviet economy was still relatively
backward . During the war, the Soviets were forced to depend heavily on food and
mun itions sent by their allies, especially the United States.
The immediate postwar years saw the outbreak of the Cold War between the
Soviet Union and its former allies. This conflict initially went very well for the
communist side, with the installation of Soviet-dominated communist regimes in the
Eastern European countries occupied by the Red Army and the adoption of
communism in China, North Korea, and North Vietnam. Suddenly, the Soviet
Union had demonstrated scientific and engineering prowess that seemed to exceed
that of the West, and soon the Soviet leader, Nikita Khrushchev, was boasting that
the economic might of the Soviet Union would bury the West. Communism, as an
alternative to capitalism, not only seemed to enjoy its claims of ethical superiority but
also seemed capable of producing remarkable economic progress.
By 1 990, however, communism had been rejected as a total failure throughout

8 It is possible, especially for sophisticated investors such as these people, to offset the lock-in effects
to some degree by adjusting their portfolios. A simple way would be to sell Salomon stock short, contracting
to deliver the stock at a fixed price five years hence. This would lock in a price that the employee would
be able to get for the stock in the trust; therefore, the future price of the stock would no longer be of
concern . The firm no doubt would be very displeased with an employee who shorted its stock.
9 If the employee had received the stock outright at the time the bonus was earned, held it for five
years, then sold it, the increased value over the period would have been a capital gain and been taxed by
the federal (and perhaps the state and local) government.
most of Eastern Europe, and even the Soviet Union itself was struggling to crea te a Docs Organ i zation
market economy after 70 years of commun ist organ ization. The system simply had M atter?
not produced. Living standards had been flat or falling for a decade in the Soviet
Union and in most of the members of the Soviet bloc. There were constant shortages
of food, housing, and all other sorts of consumer products. As a result, workers, with
the implicit consen t of their managers, took several hours off each day to wait in line
at shops. Luck, influence, and patience were the means by which the available goods
were rationed; the money that had to be paid for them was at most a secondary
consideration: "We pretend to work, and they pretend to pay us . " The goods that were
available were relatively expensive, shoddy, and often unsafe. For example, more
than 2, 000 television sets a year exploded in Moscow alone, causing injuries and
fires. Agricultural output, wh ich had grown immensely throughout the rest of the
world, had stagnated under state control and collectivization, and the Soviet Union
had become dependent on grain imports from the West to feed its people. Meanwhile,
Soviet agricultural officials annually blamed the repeated crop failures on yet another
year of unusually bad weather.
Communist industry was both technologically backward and monumentally
inefficient; measured total factor productivity in Soviet industry actually fell in many
major sectors in the decades following the triumph of Sputnik, and Eastern Europe
faced the most serious industrial pol lution problems in the world. Maintaining the
Cold War military competition with the United States and NATO was devouring an
immense percentage of the already low Soviet gross national product (GNP), nearly
bankrupting the country. Instead of the "workers' paradise" and equal ity that were
promised, communism had delivered a repressive, totalitarian government and
institutionalized privileges for the pol itically favored.

Building Socialism
Soviet communism found its intellectual ongm in the writings of Karl Marx, a
nineteenth-century classical econom ist. However, Marx provided no blueprint for
organizing an economy along communist lines. For Marx, the triumph of communism
was simply foreordained by the inherent contradictions in the capitalist system . The
actual task of designing the first commun ist system fell initially to Nikolai Lenin, the
first leader of the Soviet state that emerged after the Revolution of 1 9 1 7, and then to
Joseph Stalin, who ultimately succeeded Lenin and led the Soviet Union from the
late 1 920s until after the Second World War. It was Stali n's vision that defined the
basic form for the Soviet economy: state socialism and central planning under
the direction of the Communist Party.
The system that Stalin developed replaced private ownership of the means of
production by collective-or, rather, state-ownersh ip of land, buildings, machi nes,
and other capital. Factories belonged to the communist state as representi ng the
proletariat, and individual farms were forcibly socialized into collective farming
enterprises. Private business essentially ceased to exist, except on the smallest scale.
Workers were guaranteed jobs and had latitude in their career choices, but their
mobility was limited and being unemployed became a criminal offense.
Prices were set by the government and left unchanged for extended periods.
They therefore bore little or no con nection to costs and could not serve to ration
demand when shortages appeared. Prices also were not allowed to direct resources to
their highest-value uses, increasing supplies of goods where there was shortage and
favori ng purchases from low-cost producers. Instead, a system of detailed central
planning and extensive vertical information flows was instituted to replace the
decentralized decision making, horizon tal communications, and price-guided coordi­
nation that are characteristic of market systems . Central planners in Moscow decided
 14
The Problem of how much of what goods were to be produced in each period by which factories. The
Economic plans directed where inputs were to be obtained, where outputs were to go, and what
Organization prices were to be paid. Similar systems were instituted in Eastern Europe after the
communists gained control there. In Poland at one point even the production of
pickled cucumbers and the number of hares that would be shot by hunters were
included i n the plan, although later the planning was less pervasive. In all cases,
investment decisions were centralized, with capital allocated by the state. As
ci rcumstances changed and unforeseen contingencies arose that rendered the original
plan infeasible, i nformation would be passed up to the central planners who would
attempt to redefine the plan and coordinate activities.
The initial intent of the socialist system was to replace economic incentives with
political and moral appeals to the workers' patriotism and socialist consciousness. Pay
was di vorced from supply and demand-"From each according to his abilities, to
each according to his needs"-and the aggregate amount of GNP to be devoted to
consumer goods was determined by the planners so as to leave enough extra resources
to finance the planned i nvestments. However, whether the means of realizing the
plan were administrative orders or designed economic i ncentives, individual self­
interest intervened . This led to the sort of difficulties epitomized by the familiar (if
possibly apocryphal) story of a factory meeting its target of I 0, 000 kilograms of nails
by producing a si ngle nail weighing 1 0, 000 kilos. Later, when the planning and
incentive systems were refined, incentive problems still arose, especially with regard
to quality. A worker in a Baltic television man ufacturing facility told of the rush at
the end of each month to meet production targets and earn bonuses:

We never use a screwdriver in the last week. We hammer the screws in.
We slam solder on the connections, cannibalize parts from other televisions
if we run out of the right ones, use glue or hammers to fix switches that
were never meant for that model . And all the time the management is
pressing us to work faster, to make the target so we all get our bonuses. 1 0

The 2, 000 exploding televisions a year in Moscow are now understandable.

THE RATCHET EFFECT A particular i ncentive problem that proved to be of fundamen­

tal importance was the ratchet effect. The central planners were never as well informed
about the productive capability of any particular fa ctory as were the factory managers.
The planners would set targets for input usage and output levels and then use the
information gathered from experience to judge what would be possible in future
planning periods. Those factories making their quotas were rewarded in various ways;
those failing were punished. From the point of view of the factory manager, the
i ncentives i nherent in this system were especially perverse. By exceeding the quota
this period, the manager could expect to be "rewarded" with a higher quota for the
indefinite future, as performance expectations are ratcheted up or raised to a higher
level . As a result, there was no reason ever to exceed quota. Instead, the incentives
were to meet the quota barely, or even-provided a good excuse were available­
deliberately miss it somewhat so that the next period's quota would be lower and
easier to meet. In particular, th1=re was an inexorable tendency to hoard resources, to
hold back on effort and output, and to underreport capabilities.
These tendencies were intensi fied by the experience of factories and managers
who responded to experiments with stronger and better incentives by increasing
production, only to be accused of having defrauded the state before. These perverse

1
° Clive Cook, "A Survey of Perestroika, " The Economist, April 28, 1 990, 6.
 15
incentives led to low productivity in the economy. They also burdened the planners Docs Organization
with systematically distorted information . The task of planning even the key resource Matter?
flows in the economy would have been formidable if the information being conveyed
was accurate. Huge amounts of information needed to be transferred. Even employing
strong simpl ifying assumptions about the structure of technology, the computational
problems were extreme, and the problems of responding to unforeseen, emergent
events were overwhelming. The factory managers' misrepresentation of productivity
information and their secret hoarding of resources made the task impossible.
Beginning in the 1960s there were many attempts to reform the communist
economies. Moves were made towards decentralization designed to give more decision­
making authority to local planners a nd individual factories and to provide better
incentives. None of these attempts were truly successful, although Hungary's liberal iza­
tion .moves following 1 968 had positive effects.

The Colla p se of Communism

The collapse of communism in Eastern Europe was dramatically rapid . There had
been a history of worker uprisings and attempted breaks with Moscow-in East
Germany and Hungary in the 1950s, in Czechoslovakia in the 1 960s, and in Poland
in the late 1970s and early 1980s . All had been more or less suc�essfully squashed,
often with the use of Soviet troops. Then in 1 989 and 1 990, within a single year
following signals from Soviet leader Mikhail Gorbachev that the Soviet Union would
no longer intervene in the internal affairs of the Eastern European countries,
noncommunist governments appeared, the Berlin Wall came down and travel
restrictions were removed, independent presses evolved, multiparty elections were
held and capital ist-oriented parties were elected, communist parties were disbanded
or re-formed as social ist parties, East and West Germany were united under the West
German model, and experiments in moving to market economies began in the other
Eastern European countries. These changes were most radical in Poland, which quite
simply decided to restructure itself as a free-enterprise economy as soon as possibl e.
In the Soviet Union Gorbachev's policies of Glasnost (openness) and Perestroika
(restructuring) were originally meant simply to reform communism and make it work
better . 1 1 Even with these intentions, the pol icies met strong resistance from the
entrenched party functionaries and bureaucrats. On the other side, "radicals" and
noncommunists won election victories and began exerting pressure for the elimination
of communism and its replacement by a market system.
THE CHALLENGES OF ADOPTING A MARKET SYSTEM These economies face great
problems in moving to more ma�ket-oriented systems. They must determine property
rights and decide who will be allowed to own the currently state-owned enterprises,
how title will be transferred, and what prices will be charged . They must set up capital
markets and create banking, financial, and monetary systems . They have to design
meaningful accounting systems so that firms can be valued and their performances
judged. They need to redraft their laws to allow for new forms of economic
organizations, new patterns of ownership, arid new sorts of transactions. They have
to find managers who can operate in a market system and compete in a world market.

1 1 George Shultz, the former U.S. Secretary of State, is reported to have been told by his Soviet
counterpart, Edvard Schevardnadze, that the Chernobyl nuclear reactor accident made the Soviet leadership
more fully appreciate the need for Glasnost. The Kremlin leaders were unable to get reliable, timely
information on the crisis through their bureaucratic channels. Instead they found that the Cable News
Network (CNN) broadcasting by satellite from the United States was their best source of information on a
major event in their own country. This drove home to top government leaders that their closed system had
to be opened up for the benefit of the country.
 16
The Problem of They have to educate their populations to the new rules of the game and gain
Economic acceptance for these rules. They must decide on competition and regulatory policies
Organization and find a way to deal with the fact that simply privatizi ng the giant, inefficient state
firms will yield a system of giant, inefficient private monopol ies. They must decide
how much to wea n their industries from state subsidies and develop tax systems to
finance government activities. They have to decide whether a nd when uncompetitive
firms will be allowed to fail and create social-service and support systems to handle
the human costs of the dislocations that thei r economies are sure to face, both during
and after the transition .
The great difficulty in all this is that these tasks are all interdependent and i n
need o f coordination. Private enterprise will not work without the discipline of
potential failure. Output markets are of no use without input markets, i ncluding ones
for capital, through which producers can obtai n the resources they need. Neither of
these are of much value without well-defined property rights and mechanisms for
contract enforcement. All the parts have to come together and fit reasonably well for
the system to work. The worst outcome might be some halfway compromise. Consider
the joke told in response to criticisms that Poland's "cold-turkey" foreswearing of
communism and adoption of free enterprise is too much, too soon: It was decided i n
a certai n country to change the side o f the road o n which traffic drove from the left
to the right, but there was concern about maki ng too radical and rapid a switch; so
as an experiment only the trucks changed sides for the first year.
PATTERNS OF
0RCAl\l ZATIONAL SL'CCESS AND FAILLRE
Henry Ford's account of organization missed the mark. The study of organization is
not about how berries are arranged on a tree of authority but about how people are
coordinated and motivated to get thi ngs done.
In these few, brief historical accounts, several patterns have begun to emerge.
First, and most fundamentally, organization and business strategy can be as important
as technology, cost, and demand in determining a firm's success. Despite its superior
technology, greater resources, and scale advantages, Ford Motor Company under
Henry Ford's management lost its battle with Alfred Sloa n's General Motors. General
Motors, in turn, lost market share to a smaller and technologically weaker Toyota ,
which labored under the same kinds of disadvantages. The Hudson's Bay Company
suffered many defeats in its competition with the fledgli ng North West Company.
The successful competitors in these stories gained advantage partly from the strategies
they adopted in their markets, but a large part of their advantage also came from their
innovative organizational structures and policies and especially from the match of
their strategies and their structures .
Which aspects of organization matter? I n these examples, incentives are one
important element. The Hudson's Bay Company employees were not much inclined
to show initiative and judgment when they could be flogged for their errors but got
no share of any extra profits they earned for the company. At General Motors in
1 920, the failure to charge divisions for the cost of the i nventories they accumulated
was responsible for a huge inventory build-up leading to a financial crisis. Salomon
Brothers, with its many independently operating traders, originally had the incentives
partly right. However, its emphasis on individual performance evaluation discouraged
employees from cooperating and from sharing information with one another. Finally,
the communist countries, with their ideological commitment to econom ic equality,
ran afoul of incentives at every turn.
Another important shared feature of successful orga nizations in these examples
 17
is the tendency to place authority for decisions in the hands of those with information. Docs Organ i zation
Salomon Brothers, with two telephones and enormous discretion in the hands of its Matter?
traders, is an extreme example. General Motors, with its multidivisional o rganization,
placed product and marketing decisions in the hands of divisional managers. Toyota's
decision to give responsibility and authority for machine repair and maintenance to
those who operate the machines led to improved reliability. And the North West
Company gained advantage over its larger competitor by operating as a partnership
with the wintering partners in the field making timely decisions based on up-to-date
information about local market conditions.
Although delegating authority to those with the information needed to make
good decisions is an important part of good organization design, it is of little use
unless the decision makers share the organization's objectives. We have already
mentioned incentives as a way to align individual and organizational objectives . The
additional point we want to make here is that incentives a re especially important when
more initiative is expected from employees. The same point, viewed from a different
angle, is that delegation of authority is much more valuable when those being
empowered have also been given incentives to work for the organization's objectives.
It is no accident that the North West Company both relied on the judgments made
by traders in the wilderness and made them partners who shared in the profits . Nor
is it an accident that the traders at Salomon B rothers whose decisions can have a
multimillion-dollar effect on the company's holdings a re provided with both info rma­
tion and the incentive to use it well . Similarly, in Eastern Europe, giving more
discretion to factory managers without also providi ng fo r decentralized ownership
could not resolve the problems.
In the language of economics, incentives and delegated authority a re comple­
ments: each makes the other more valuable. Evaluating complementarities-how the
pieces of a successful organization fit together and how they fit with the company's
strategy-is one of the most challenging and rewarding parts of o rganizational analysis .
In our General Motors example, the reason for the multidivisional structure was to
carry out Sloan's new market-segmentation strategy. The delegation of decisions to
divisional managers was also combined with improved accounting information to help
evaluate those decisions and with coordination from the central office to ensure that
the parts of the organization were not working at cross purposes. At Toyota, an
organization based on very low inventories was natu rally vulnerable to disruptions at
any stage in the production process a nd to simple changes in plans. If low inventories
were to be achieved, the rest of the organization had to emphasize reliability, quick
response to machine breakdowns, quick and close communications with suppliers,
stable production plans, and other features that substitute for the buffer function of
inventories. Toyota's emphasis on reliability fit with an emphasis on quality in its
marketing, and its use of flexible equipment, necessitated by its initially small scale,
fit well with its more frequent redesign of its several models.
I n this book, our economic analysis of organizations is based on elaborating the
several ideas that these brief histories suggest. We study coordination: what needs to
be coordinated, how coordination is achieved in markets and inside fi rms, what the
alternatives are to close coordination between units, and how the pieces of the system
fit together. We also study incentives and motivation: what needs to be motivated,
why incentives are needed and how they are provided in markets and firms, what
alternative kinds of incentive systems are possible and what needs to be done to make
incentive systems effective. In the last chapters, we see in more detail why these
aspects of organization do matter, when we apply the principles to make a detailed
study of a few important functions of the business enterprise.
 18
The Problem of E.\ E ttCI SES
Economic
Organization Food for Thought

1 . In fast-food chains, some decisions about standards are made centrally and
others are left to the individual outlet managers. Who typically makes wh ich ki nds of
decisions? Why? Can you th ink successfully about the fast-food business by dividing
the issues between coordi nation and motivation?
2. Arm ies in battle have especially severe organization problems. What kinds
of decisions arc made centrally and which a re left to commanders in the field? What
principles dictate the division? Can you th ink successfully about the problems of
military organization by dividing the issues between coordination and motivation?
3. In late summer of 1 99 1 , it was revealed that beginning in February of that
year, the Salomon Brothers' managing director in charge of its deal ings in the markets
for U . S . Treasury securities had secretly violated government rules designed to prevent
any one buyer from purchasing more than a specified share of the new gm·ernment
debt obl igations being aucti oned at any one time. Moreover, sen ior executi\'es at the
firm had learned of these violations of the law but had fai led to act on them. In the
resulting scandal, four of Salomon's top executives (including Gutfreund) were asked
to resign , and the future of the firm was imperiled.
One theory was that the employee who violated the rules was motivated by his
competitive, aggressive nature: He bitterly resented the government's attempt to limit
his actions and was determined to show them that he was smarter and tougher than
they . Another was that he was motivated by greed: He \\'anted to monopolize the
market in these government bonds and reap the rewards. In any case, his actions, and
those of the senior executives who failed to stop him and report his wrong-doing, were
very costly to the firm: The stock price, wh ich had risen by almost a half in the
preceding year, fell back to where it had been before, and important customers ceased
dealing with the firm.
Does th is episode mean that the incentive scheme described in the text was
fundamentally Aawed?
 2
ECONOMIC ORGANIZATION
AND EFFICIENCY

[People] i n general, and within lim its, wish to behave econom ically, to make their
activities and their organization "ef-{i.cient" ra ther than wasteful.
Frank Knight 1

This book is concerned with the problems of designing and managing efficient
economic organ izations. The preceding chapter gives examples of the range of
problems and institutions that we are considering. We now explore the source and
nature of these problems and the approach we take to analyzing them .

ECONOM IC ORGANIZATIONS : A PERSPECTIVE

Economic organ izations are created entities within and through which people interact
to reach individual and collective economic goals. The economic system consists of
a network of people and organizations, with lower-level organizations linked together
through higher-level organizations.
The highest-level organization is the economy as a whole. While it is somewhat
unusual to think of an entire economy as an organization, this perspective is useful
because it emphasizes that the economic system is a human creation and because
many of the problems that smaller, more formal organizations face exist at the
economy-wide level as wel l. As an organization , an economy can and should be

1 "Review of Melville J. Herskovits' 'Economic Anthropology,' " fournal of Political Economy, 49

(April 1941 ), 246-258, quoted by Oliver Williamson in "Mergers, Acquisitions, and Leveraged Buyouts:
An Efficiency Assessment, " Working Paper Series D, Economics of Organization, no. 30, Yale University
School of Organization and Management (1987). (Emphasis added by Williamson. ) 19
 20
The Problem of evaluated based on its performance relative to possible alternative arrangements. The
Economic current experiments in thoroughgoing reform of entire economic systems in Eastern
Organization Europe are prime examples of the importance of this perspective.

Formal Organizations
At the next level are the entities more traditionally regarded as organizations and the
ones that are our main concern: corporations, partnerships, sole proprietorships, labor
unions, government agencies, universities, churches, and other formal organizations.
A key characteristic of the organizations at this level is their independent legal identity,
which enables them to enter binding contracts, to seek court enforcement of those
contracts, and to do so in their own name, separate from the individuals who belong
to the organization.
ORGANIZATIONS .\ND CoNTR\CTING This ability to enter contracts is critical to one of
the major approaches to the economic analysis of organizations. In this view, which
was first suggested by Armen Alchian and Harold Demsetz, an organization is regarded
as a nexus of contracts, treaties, and understandings among the individual members
of the organization . The firm itself is then a legal fiction that enters relatively simple,
bilateral contracts between itself and its suppliers, workers, investors, managers, and
customers. Without a legal entity that can contract with them individually, these
people would have to fashion complex, multilateral agreements among themselves to
achieve their aims.
The contracting approach to organization theory emphasizes the voluntary
nature of people's involvement in (most) organizations: People will give their allegiance
only to an organization that serves their interests. Furthermore, along with the ability
to enter contracts come the possibilities for reform, redesign, and abandonment of
the organization by rearranging contractual terms. This approach also facilitates
accepting the fuzziness of organizational boundaries and the fact that organizational
forms blend together. Markets and hierarchies-sometimes regarded as the major
discrete alternative ways of organizing economic activity-are actually just two extreme
forms of organizational contracting, with voluntary bargaining characterizing markets
and strict li nes of authority characterizing hierarchy.
THE ,\RCI HTECTl'RE OF ORG.\N IZ.\TIO'\iS Although the legal aspects of organization
are important, a full description of organizational architecture involves many more
elements: the patterns of resource and information flows, the authority and control
relationships and the distribution of effective power, the allocation of responsibilities
and decision rights, organizational routines and decision-making processes, the
methods for attracting and retaining members and resources, the means by which
new ideas and knowledge are generated and diffused throughout the organization , the
adaptation of the organization's routines to reflect and implement organizational
learning, the organization's expressed objectives and the strategies and tactics employed,
and the means used to unify the goals and behavior of the individual members of the
organization and the objectives of the organization as a whole. Various parts of our
analysis focus on each of these and on how the pieces fit together to yield a coherent
pattern .
Once our focus becomes the elements of organizational architecture, defin ing
a formal organization simply by its ability to contract as a disti nct legal entity can
become quite inappropriate because it can easily misidentify the effective boundaries
of the organization. Consider, for example, the Sony company, which is known
for its innovative consumer electron ics products, but also manufactures broadcast
eq uipment, computer components, and sem iconductors and owns one of the three
largest record companies in the world (Sony Records, formerly CBS Records) and a
 21
major movie and television production and distribution operation (Columbia Pictures Economic
Entertainment). Sony in fact consists of the Sony Corporation, the parent orga nization Organization and
based in Tokyo, plus subsidiary corporations around the world . Each of these Efficiency
subsidiaries is a separate legal entity able to enter contracts on its own, and being an
employee of Sony Corporation of America or of Sony GmbH in Germany docs not
make one an employee of Sony Corporation. Thus, the legal-entity approach might
point to viewing each subsidiary as a separate organization . Yet this accords neither
with the way the world sees Sony nor with the way it sees itself and manages its affairs.
DISCRETION AND AUTONOMY FROM INTERVENTION In these circumstances, a useful
way to look at the defining boundaries of an organization is in terms of the smallest
unit that is functionally autonomous in that it is largely free from intervention by
outside parties in its affairs and decisions, over which it then enjoys broad internal
discretion. Within a firm, the rightful decision makers-usually senior management­
collectively have broad legal rights to order that activities be conducted as they see fit
and to require that their directives be followed. Private outside parties cannot
countermand these orders. Courts and government regulators may be able to intervene
in some ways, but the discretion of even these public agencies is generally limited .
They cannot interfere except under prescribed conditions, and then the measures they
can take are limited compared to those that are available within the firm .
Using this approach, the companies that make up Sony constitute a single
organization, even though they are separate legal entities. The senior managers of
Sony Corporation have the power to intervene largely as they wish in the operations
of the subsidiaries, even if they in fact rarely choose to do so. Thus, the subsidiaries
are not separate organizations by this test. Meanwhile, no outside private party has
the legal right to specify how Sony will be run, what products it will make, what
prices it will charge, what pay and job assignments it will offer, how it will divide its
activities among its subunits, what investments it will make, and so on . Thus, Sony
stands as a separate organization.

The Level of Analysis : Transactions and Individuals

The most fundamental unit of analysis in economic organization theory is the
transaction-the transfer of goods or services from one individual to another. The
way a transaction is organized depends on certain of its characteristics. For example,
if one kind of transaction occurs frequently in similar ways, people develop routines
to manage it effectively. If a transaction is unusual, then the parties may need to
bargain about its terms, which raises the costs of carrying out the transaction.
The ultimate participants in transactions are individual human beings, and their
interests and behavior are of fundamental importance for understanding organizations.
People are fundamental first in the sense of being indivisible decision makers and
actors; it is people-not organizations-who actually decide, vote, or act. The actions
of individuals determine the behavior and performance of organizations. Furthermore,
only the needs, wants, and objectives of individuals have ethical significance.
Economic organizations are judged only on the basis of how well they serve
people's intended purposes. Finally, it is people who ultimately create and manage
organizations, judge their performance, and redesign or reject them if th is performance
is found inadequate.
In analyzing how organizations emerge, how they are structured, how they
function, and how economic activity is divided among them, we adopt the position
put forward in the opening quotation: that people will seek to achieve efficiency in
more than just the day-to-day conduct of their economic affairs. Efficiency also must
exist at a systemic level, in the organization of people's activities and in the design ,
 22
The Problem of management, a nd governance of the i nstitutions they create. Before we explore this
Economic position more fully, we need to be more precise about what we mean by eff iciency.
Organization
EFFICIENCY
The goal of any economic organization, i ncluding the economic system as a whole,
is to satisfy the wants and needs of individual human beings. We j udge economic
performance in terms of this goal. This approach does not i mply an exclusive concern
with materialistic achievement. If mil itary dominance or national prestige is the
priority, then an economy that serves these goals could be performing well by our
standards. If the population were united in believing that the purpose of all human
activity should be to glorify the deity, then a n economic system that supported this
consecration would be a good one. For purposes of our discussion, however, we
assume that people primarily are concerned with regular economic goods and services.
The economic system then is judged on how well it satisfies the economic needs of
the population.
Th is approach obviously requires that we ascribe preferences to i ndividuals.
Indeed, we assume that people are equipped with measures of their welfa re (called
utility functions), that they like one situation better than a nother if a nd only if it gives
greater utility, a nd that their economic goal is to maximize this measure of satisfaction.
We explore this assumption in more detail later i n the chapter.
The problem of scarcity, however, means that trade-offs typically have to be
made. Increasing the utility of one person may mean having to give less to another.
How then are we to measure performance? What does it mean to consider how well
peoples' i nterests a re served when these i nterests are possibly i n conflict?

The Concept of Efficiency

• The partial solution is to focus on efficient choices or options, by which we mean
ones for which there is no available alternative that is universally preferred in terms
of the goals and preferences of the people involved. More precisely, if individuals are
sometimes i ndifferent about some of the available options, then a choice is efficient
if there is no other available option that everyone in the relevant group l ikes as least
as much a nd at least one person strictly prefers. Turning the definition around, a
choice is ineff icient when there is an alternative possible choice that would help one
person without harming any other.
Note well that the efficiency criterion can never be applied to resolve ethical
questions about when it is justified or worthwhile to help one person at another's
expense. In some situations, such questions cannot be avoided, but efficiency will
not help. Instead, appeals to other criteria that explicitly trade off one individual's
welfare agai nst another's are needed.
Note too that the efficiency or inefficiency of a choice is always relative to some
specific set of i ndividuals whose interests are bei ng taken i nto account and also to
some specific set of available options. This is important to remember. It is distinctly
possible that a particular choice from a given set of alternatives will be efficient relative
to the i nterests of a given group of people, but not when some larger affected group
is considered. Similarly, a choice may be efficient when all the constraints that delimit
the set of available options a re recognized, but not when the removal of some of these
makes more options available. Thus, i n applying the concept of efficiency it is
necessary to be clear about whose interests are counted and what alternatives a re
considered to be feasible..
Efficiency of Hesource Allocations Economic
Organization and
Efficiency can be defined and applied at many levels, depending on the kind of choice
Efficiency
being considered. Our fi rst application-and the most common one in economic
analysis-is to compare alternative allocations of resources . An allocation of resources
A is inefficient if there is some other available allocation B that everyone concerned
likes at least as well as A and that one person strictly prefers. (In this case the allocation
A is sometimes said to be Pareto domina ted 2 by allocation B . ) An in"efficient allocation
is wasteful: by making better use of the available resources it would be possible to
make some people better off without hurting anyone else. If, on the other hand, no
allocation exi�ts that is unanimously preferred to A, then the given allocation A is
efficient (or Pareto optimal). An effi cient allocation of resources is thus one such that
there is no other available allocation that makes someone better off without making
another person worse off.
Although the reasons for wanting an efficient allocation of resources are obvious,
efficiency by itself is a weak performance criterion. First, there are typically many
efficient allocations of a given collection of resources. Therefore, requiring efficiency
does not pin down a unique outcome. However, we see later that under certain
conditions, efficiency is a good predictor because it then implies sharp restrictions on
the choices that can be made. Second, having all the potential benefits of economic
activity go to an insatiable, completely selfish person would be efficient because any
reallocation of resources would hurt this person and so could not gain unanimous
support. Thus, to note that an allocation is efficient is hardly to recommend it on
ethical grounds!
H owever weak efficiency may be as a predictor or as an ethical criterion, actually
achieving efficient allocations is extremely demanding. At a macro level, not only
must all goods be produced at the lowest possible cost, but the right mix of outputs
must be forthcoming, the right levels of savings and investment must be provided,
and, i n general, there must be no way to increase consumer satisfaction by any
reallocation of society's resources. The task of computing an efficient allocation for a
complex modem economy is clearly beyond the limits of feasibility. Even with only
two people, it may be impossible to determine whether there is any way of rearranging
their activities to help one without hurting the other.
Moreover, even if an efficient allocation has been identified, it also is necessary
to ensure that the people involved do their parts in bringing it about. The problem is
that there often will be ineffi cient allocations that are better for one person or subgroup
than is the target efficient allocation, and these people may be able to effect the
inefficient outcome that they prefer. 3 Despite this difficulty, efficiency is both an
important device for organizing ideas and a useful criterion for evaluating performance.

Efficiency of Organizations
Efficiency of outcomes or allocations is not, however, the key concept for the study
of organizations. Instead, we are concerned about the efficiency of the organizations
themselves. We assume that the fundamental objects that people care about are the

2 Vilfredo Pareto was an Italian economist and sociologist who is credited with developing this
criterion for problems with multiple objectives.
3 There will necessarily be a third allocation that is efficient and that Pareto dominates the inefficient
one: otherwise, the latter would actually be efficient. However, as we will see, identifying and reaching
this efficient allocation may be a problem.
 24
The Problem of outcomes the organizations generate, and that organizations are to be judged on the
Economic basis of these outcomes . There are, however, several ways to do this.
Organization A simple way is to make the comparison outcome by outcome. Consider two
contracts, routines, decision processes, organizations, or economic systems, say X and
Y, which both might be used in a variety of circumstances. Suppose that, in each
such ci rcumstance, Y always yields outcomes that are viewed by all the people involved
as being at least as good as those that X produces, and that sometimes Y yields results
that at least one person definitely prefers to the outcome under X. In th is case, X is
inefficient because Y does better. In contrast, a contract, routine, process, organization,
or system is efficient in this sense if there is no alternative that consistently yields
unanimously preferred results. In particular, an organization that always yields efficient
outcomes is itself efficient.
This outcome-by-outcome comparison of organizations is quite demanding: To
declare an organization inefficient, there must be another that would do better in
every possible ci rcumstance. Consequently, the resulting notion of efficiency is weak
because it is easy to pass the test of not having a better alternative. Thus, there might
be many efficient organizations using this criterion . Later we refine the efficiency
notion, for example, by specifying that an organization is inefficient if there is another
that does better for each person on average across the circumstances in which the
organization operates. Such specifications narrow the class of orgc1nizations that meet
the test of efficiency. Because it is now easier to find another mechanism that is
unan imously preferred, it becomes more difficult to pass the test of there not being a
preferred mechan ism .
Regardless of which specification we employ, if achieving efficiency of alloca­
tions is problematic, realizing the systemic efficiency of organizations clearly is even
more demanding. However, the notion of efficiency still provides a key organizing
pri nciple.

Efficiency as a Positive Principle

So far, we have emphasized efficiency as a normative concept, a criterion by which
group decisions-including both resource allocation decisions and organ ization
structure decisions-are to be evaluated. However, the quotation at the begi nn ing of
this chapter points in another direction . If people seek efficiency in their activities
and in the ways they arrange their affairs, then efficiency can become a positive
concept, with explanatory and predictive power, as well as a normative, prescriptive
one.
There is good reason to expect that people will seek out and settle upon efficient
choices. After all, by the definition of efficiency, if an inefficient situation is reached,
then someone could propose an alternative that everyone would prefer. If the parties
can bargain together effectively and can effectively implement and enforce any
agreements they reach, they should be able to realize these gains. Inefficient decisions,
whether about resource allocations or organizational arrangements, are thus always
vulnerable to being overturned . Efficient arrangements are much less vulnerable
because any proposal to change an efficient arrangement will always be opposed by
someone. For this reason , we expect to find inefficient arrangements being supplanted
over time, while efficient ones survive. We summarize this argument in the following
efficiency principle.
The Efficiency Principle: If people are able to bargain together effectively
and can effectively implement and enforce their decisions, then the
outcomes of econom ic activity will tend to be efficient (at least for the
parties to the bargain).
 25
Much of our analysis of organizations is based on the efficiency princi ple. We Economic
try to understand existing arrangements as efficient choices, and we interpret changes Orga nization and
in these arrangements as efficiency-enhancing responses to changes in the environment Efficiency
within which the arrangements exist.
In pursuing this agenda, it is important to keep in mind the quali fications that
bargaining, implementation , and enforcement should be effective. A major focus of
the analysis in this book is the task of giving specific meaning to the notions of effective
bargaining, implementation , and enforcement, and of identifying and analyzi ng
cond itions under which these premises might or might not be expected to hold . What
factors influence the possibility and likelihood of efficient bargaining? What factors
promote or impede implementation of plans and enforcement of bargains ? The answers
to these questions provide much of the basis for understanding actual arrangements
and for analyzing possible changes.
In using efficiency as a positive concept for predictive purposes, it is especially
important to be clear about the set of individuals whose interests are being taken into
account in determining economic arrangements. (Recall that efficiency is always
defined relative to a specific set of individuals and options . ) Suppose a large group of
people might be affected by some choice, but only a relatively small subgroup of them
are able to communicate with one another effectively, and that they alone can reach
agreements and implement and enforce them . Then the appropriate concept for
predicting what arrangements will be chosen is efficiency relative to the small, effective
group. The interests of people who do not take part in the decision about what choices
are made are unlikely to be fully reflected in the choices . For normative, valuative
purposes, efficiency relative to the larger group may sti ll be an appropriate criterion,
but it is unlikely to be predictively powerful.

Tl IE TASKS OF COORDINATION AND MOTl \'ATION

A fundamental observation about the economic world is that people can produce
more if they cooperate, specializing in their productive activities and then transacting
with one another to acquire the actual goods and services they des i re. The problem
of organization then arises because when people are specialized producers who need
to trade, their decisions and actions need to be coordi nated to achieve these gains of
cooperation, and the people must be motivated to carry out their parts of the
cooperative activity. Both the existence of formal organ izations and the specific details
of their structures, policies, and procedures reflect attempts to achieve efficiency in
coordination and motivation.

Specialization
Adam Smith's famous example of the pin factory vividly shows the benefits of
cooperation and specialization and the correspond ing need for coordination . Smith
described how in his time (the late eighteenth century) the various stages of pin
manufacturing were carried out by different people, each of whom specialized in a
single task-pulling the wire, straightening it, cutting it to appropriate lengths,
sharpening the point, attaching the head, and packaging the finished product-and
how the resulting volume of output was many times greater than it would have been
if each person involved had done all the stages alone. The crucial point, however, is
that such specialization requires coordination . A single person producing pins alone
turns out something useful. The time and efforts of the specialists are wasted unless
they can be sure that both the people at each of the preceding stages are doing their
parts in generating semifinished materials in the appropriate amounts and in a timely
way, and that those at the latter stages of manufacturing are prepared to take what the
people before have produced and turn it into a fi nished product.
 26
The Problem of The principles of specialization and coordination apply both to small, simple
Economic economies and to large, complex ones. Robinson Crusoe did not face a coordination
Organization problem when he was alone on his desert isla nd, but once Friday entered the story,
there were opportunities for ga in through specialization and exchange. This meant
there was also a need to coordi nate the two men's actions to ensure that all the
necessary tasks were done and that they were not needlessly repeated by both Crusoe
and Friday. In a modern economy, the va riety of tasks that are carried out is
unfathomably complex. Somehow, each of these jobs must be accomplished, in the
appropriate amounts, using the appropriate methods, at the right time, in the right
order, and by the right numbers of the right people. Coordinating the billions of
people and their choices among the infinite possibilities facing them is a mind­
boggli ng problem . Moreover, even if a solution were somehow found that was
reasonably well adapted to currently prevailing conditions, the problem would not be
solved once a nd for all. Conditions at every imaginable level and on every possible
dimension are constantly changing, and adaptation to these changes is necessary.

The Need for Information

A key problem in achieving effective coordination and adaptation is that the information
needed to determine the best use of resources and the appropriate adaptations is not
freely ava ilable to everyone. Efficient choice requires i nformation about individual
tastes, technological opportunities, and resource ava ilabilities. No single person in
society has all-or even a signi ficant fraction-of the needed i nformation. Instead,
information is localized and dispersed throughout the economy. Even if the relevant
information were generally available, determining what should be produced, for
whom , by whom, and using what methods and materials is an overwhelmingly large
and complex problem . Because this information is local ized and dispersed, however,
no one has the knowledge needed to make these calculations, even if they m ight be
feasible i n principle.
Two solutions are possible. Either the dispersed information must be transmitted
to a central computer or plan ner who is expected to solve the resource-allocation
problem, or else a more decentralized system must be developed that involves less
information transmission and, correspondingly, leaves at least some of the calculations
and decisions about economic activity to those with whom the relevant i nformation
resides. The trick with the first option is to make timely decisions while keeping the
costs of communication and computation from absorbing all the available resources.
The challenge of decentralization is to ensure that the separately made decisions yield
a coherent, coordinated result.

Organizational Methods for Achieving Coordination

Different organizational structures achieve coordination in different ways and with
differing results. As we discuss in Chapter 1 , the original strategy of the H udson's Bay
Company was overly centralized . As a result, decisions were not timely and the
company made poor use of local i nformation . In contrast, the structure of General
Motors was initially overly decentralized , and the company suffered from a consequent
lack of coordination. In the case of Smith's pin factory , the solution was a single firm
whose owner-ma nager specialized in providing coordination. He or she hired workers
and assigned them to the different tasks, set the levels at which each was to perform ,
tracked performa nce and the external environment, and adjusted plans as needed.
This individual also probably owned the capital equipment that was bei ng used,
collected the sales revenues, and pa id the bills. Another solution m ight have been a
cooperative of pinmakers, where the workers would jointly have decided on the levels
 27
of activity and the task definitions and assignments and would then have shared in Economic
the costs and revenues they generated . The older, highly decentralized system of Organization and
individual craftsmen, each producing pins alone, sacrificed the gains from specializa­ Efficiency
tion but reduced the need for coordination. Yet another solution would have been to
organize each stage of prod uction as a separate firm and let the transactions between
stages be intermed iated by the market. Th is last alternative may sound far-fetched if
we picture the person at each stage in the pin factory as a separate firm, buying input
from the person/firm on one side and selling output to the next person/fi rm in the
production line. However, the striking differences between GM and Toyota in their
reliance on independent suppliers (see Chapter 1 ) indicate that this kind of alternative
is genuine. Finally, within some modern corporations, products are sold by one
division to another using transfer prices, and the division managers are judged on
their individual division's profitability. With this system, the firm's internal organization
mimics the market in many ways.
COORDINATiON THROUGH A SYSTEM OF MARKETS AND PRICES A thoroughgoing use of
the market is one possible solution to the problem of coordinating economic activity.
At the extreme, all transactions could be between separate individuals on an arm's­
length basis, and there would be no firms or other organizations apart from the market
system itself. The opposite extreme would be complete elimination of the price system
under a regime of explicit central plann ing, with all decisions being made within a
single (presumably multilevel) organization. Of course, no economic system approaches
either extreme. Even in their most centralized versions, the centrally planned
communist economies left many decisions to individual consumers who made their
choices in part in response to prices. The market economies feature firms that interact
with one another through markets but within which activity is explicitly coordinated
through plans and hierarchical structures.
In fact, the system of markets and prices is often a remarkably effective
mechanism for achieving coordination . Day in and day out, without any conscious
central direction, it induces people to employ their talents and resources so effectively
that the shortages and ration ing which are familiar to residents of planned economies
are deemed newsworthy events when they occur in market economies. As a practical
matter, the advantages of the market system over socialist planned economies seem
clear.
As we see in the next chapter, it is even possible to argue formally that no
system can solve the coordination problem more effectively than a system of markets
coordinated by prices. A mathematical model is used to show that in economies with
certain characteristics, the allocations generated by a price system are always efficient
for society as a whole. Moreover, under certain conditions, the price system
achieves this result while economizing on information demands-the system requires
transmitting less information than any other system capable of ensuring efficient
outcomes (see Chapter 4). In an ideally functioning system of markets, all that anyone
needs to know is his or her own capabilities and tastes and the prevailing prices .
There is no need to transmit detailed information about preferences, technological
possibilities, resource availabilities, and the like that would be needed to achieve a
central ized solution because the prices summarize all the relevant information.
Furthermore, when conditions change, detailed local knowledge of these changes
need not be transmitted to achieve effective responses. Instead, the changes in prices
again convey all the information that is actually needed for people to respond
effectively.
I NCENTl\'ES I!\ MARKETS The strength of a market system with private property lies
not just in its providing the information needed to compute an efficient allocation of
 28
The Problem of resources in an efficient manner, however. At least equally important is the manner
Economic in which it accepts individually self-interested behavior, but then channels this
Organization behavior in desi red directions. People do not have to be ca joled, artificially i nduced,
or forced to do their parts in a well-functioning market system . Instead, they are
simply left to pursue their own objectives as they see fit. Yet, at least in the right
circumstances (which we explore in Chapter 3), people are led by Adam Sm ith's
"invisible hand" of impersonal market forces to take the actions needed to achieve an
efficient, coordinated pattern of choices. Workers, sel fishly attempting to maximize
their own i ndividual welfare, are led to select the trai ning, careers, and jobs where
their talents a nd energy are most valuable. Producers, pursui ng only private profits,
are led to develop the goods and services on which consumers put the highest values
and to produce these goods and services at the lowest possible costs. The owners of
resources and capital assets, seeki ng only to increase their own wealth, are led to
deploy these assets i n socially desirable ways. Finally, consumers, seeki ng only to
satisfy thei r individual wants and needs, are led to do so in the way that puts the least
strai n on society's resources for the level of satisfaction achieved.
All of this is based on a particular theory of markets, which posits that competition
is ubiquitous, firms have little market power, a nd the only goods that are of significance
are those that are traded in active markets. The incentives provided by real markets
do not always align so nicely with the realization of social objectives . Large firms or
ca rtels may set prices inefficiently high, leading to i nefficient resource allocations.
Externalities and missing markets for some goods may lead to additional distortions.
The quality of goods may be hard to verify, leading some consumers to make mistakes
in their choice of goods and some firms to neglect quality control in the hopes that
their actions will go unnoticed. As we see in the examples of Chapter 1 , similar
failings plague other forms of orga nization as well. Organ izations either must rely on
individuals to ignore their own self-i nterests, with unsurprisingly disappointing results,
or else they must devote ingenuity and resources to bring coherence between individual
self-interest and the social or organizational objectives.
For example, Salomon Brothers' complex system of attributing profits and paying
bonuses linked to performance is an attempt to generate within the fi rm the sort of
i ncentives that are provided automatically by the market. The transferring of partial
ownership to the employees is i ntended to give them the i ncentives that owners have
in order to care about the long-term value of their asset, the fi rm . The manager and
workers in the Soviet factory that produced the si ngle giga ntic nail were respondi ng
to inappropriate incentives. Their jobs were made easiest by adopting a socially
wasteful producti on plan, and their incentives were to mini mize their efforts while
meeting their poorly specified quota . Such a gross i nefficiency would never arise in
a ma rket system , although, as we see in later chapters, more subtle difficulties can be
expected.

TRANSACTION COSTS ANALYSIS

If ma rkets can perform so well, why then do we so often see the price system
supplanted, with economic activity being organized within and among formal
hierarchical structures usi ng expl icit pla nning and directives? More simply, why are
there firms? What is their economic function? And what determines which transactions
are mediated through ma rkets a nd which are brought within a formal organization
a nd made under centralized direction?
These fundamental questions were first posed by Ronald Coase. According to
Coasc, there are costs to carryi ng out transactions, and these transaction costs differ
depending on both the nature of the transaction and on the way that it is organized.
Furthermore, as suggested by the efficiency principle, the tendency is to adopt the
 29
organ izational mode that best econom izes on these transaction costs. Tims, transactions Economic
tend to occur in the market when doing so is most efficient, and they arc brought Organ ization and
within the firm or some other formal organization when doing so mi nimizes the costs Efficiency
of carrying them out.
This is a simple, but profound, idea . However, Coasc was not very explicit
about the origin and nature of these transaction costs, and without a systematic
understanding of these issues, the idea is not very useful . Consequently, much of the
research in the economics of organization has been devoted to giving substance and
content to his idea . In fact, transaction costs are the costs of running the system: the
costs of coordi nating and of motivating. Thus, under the hypothesis that organizational
structure and design are determined by minimizing transaction costs, both aspects of
the organization problem affect the allocation of activity among organizational forms.

Typ�s of Transaction Costs

Different organizational forms and institutional and contractual arrangements represent
different solutions to the problems of coordi nation and motivation . These problems
give rise to transaction costs, which manifest themselves differently in different
contexts.
COORDINATION COSTS U nder a market system, transaction costs associated with the
coordination problem arise from the need to determine prices and other details of the
transaction, to make the existence and location of potential buyers and sellers known
to one another, and to bri ng the buyers and sellers together to transact.
As an example of these coordination costs, think about the problem of exchangi ng
financial assets such as stocks and bonds . These transactions mostly take place through
organized financial exchanges, like the New York, London, and Tokyo stock
exchanges. Very few markets function more efficiently than the organized financial
markets, and yet the amount of resources absorbed in their operation is clearly
significant. Large buildings, immense communication and computational power, and
the talents of thousands of gifted people are employed in setting prices and carrying
out transactions. If the often breathtaking incomes of investment bankers and security
dealers are any indication of the social costs involved in employi ng them in this
industry, the transaction costs of running these markets are very large.
In other markets, transaction costs associated with coordination include the
resources that sellers expend on market research to determine buyers' tastes, on
advertising and marketi ng expenditures to make the product or service known, and
on managerial decisions determining the prices to charge. On the buyers' side, they
also include the time spent searching for suppliers and for the best prices. More subtly,
the transaction costs also include the lost benefits that are not realized because the
matching of buyers and sellers is imperfect and worthwhile transactions fail to occur.
The transaction costs of coordination through hierarchies-whether private or
governmental-are primarily the costs of transmitting up through the hierarchy the
initially dispersed information that is needed to determine an efficient plan, using the
information to determine the plan to be implemented, and then communicating the
plan to those responsible for implementing it. These costs include not only the direct
costs of compiling and transmitting information, but also the ti me costs of delay while
the communication is taking place and while the center is determining the plan.
Because this communication can never be perfect, there are also transaction costs of
maladaptation that occur because decision makers have only insufficient or inaccurate
information .
MoTI\'.\TION COSTS The transaction costs associated with the motivation problem
are primarily of two kinds. The first type of costs are those associated with informational
 30
The Problem of incompleteness and asymmetries-situations in which the parties to a potential or
Economic actual transaction do not have all the relevant i nformation needed to determine
Organization whether the terms of an agreement are mutually acceptable and whether these terms
are actually being met. For example, the potential buyer of a new car may have
difficulty determining whether the seller's claims about its economy and reliability are
correct, and may wonder why the seller would want to get rid of the vehicle. Or a
sales manager may have difficulty in determining whether a salesperson i n the field
is actually devoting full time and honest effort to the company's business, or instead
is pursuing private interests on company time. In such circumstances, mutually
adva ntageous transactions may fail to occur, because one or the other party fears being
victimized, or costly arrangements will be made to protect against opportunistic
behavior.
The second type of tra nsaction costs connected to the motivation problem arise
from imperfect commitment-the inability of parties to bind themseh-es to follow
through on threats and promises that they would like to make but which, having
made, they would later like to renounce. As an example, consider a manufacturer
seeking to have a supplier make a large investment to meet the manufacturer's specific
needs . The suppl ier must be concerned that-all promises to the contrary not
withstanding-once the investment is sunk the manufacturer will try to force a lower
price and other concessions on the suppl ier, who will then have little recourse.
Recognition that threats and promises may not be kept deprives them of their
credibility. Thus, far-sighted people will not rely upon them, and again there will be
missed opportunities or a necessity of expending resources to facilitate commitment
or protect against opportu nism. The manufacturer would gain if it were possible to
commit not to behave opportun istically, because then the supplier would be more
willing to make the investment. Achieving such commitment may be difficult, and
so the investment may not be made or costly measures might need to be put i n place
to defend the supplier.
These problems affect both market and nonmarket organizations, although their
nature and impact may differ between organizational forms . Therefore, one form may
be better adapted tha n another for a specific transaction.

Dimensions of Transactions
According to the transaction costs approach , the va riety of ways of orga rnzmg
transactions found in the world reflects the fact that transactions differ in some basic
attributes. Five kinds of transaction attributes play important roles in our analysis:
1. the specificity of the i nvestments required to conduct the transaction
2. the frequency with which similar transactions occur and the duration or period
of time over which they are repeated
3. the complexity of the transaction and the uncertainty about what performance
will be required
4. the difficulty of measuring performance in the transaction
5. the connectedness of the tra nsaction to other tra nsactions involving other people
ASSET SPECIFICITY One important dimension on which transactions differ is the
nature of the i nvestments that the parties to a transaction must make. When an
individual consumer buys bread from a baker, neither party makes any investment
with that particular tra nsaction in mind. The baker may invest in a store and an oven,
but he or she uses those assets to supply many different customers. In contrast, when
a subcontractor makes wing assembl ies for a particular model of Boeing airliner, it
may invest in setting up a production line to make those specific assemblies. Such an
 :n
investment is called a specific investment because it would lose much of its value Economic
outside of the specific use of providing wings to Boeing. The subcontractor would not Organization and
want to make the- investment unless it has a firm order from its customer, or at least Efficiency
reasonable assurance that an order will be forthcoming. For the same reason , an
employee may not want to invest in learning the systems of a declining company
where the prospects of conti nuing employment arc poor. Transactions that require
specific investments normally also require a contract or practice to protect the investor
against early termination or opportunistic renegotiation of the terms of the production
relationship.
FREQl 1 ENCY ANO DURATION Some transactions are one-time affairs as, for example,
when a homeowner buys a house from its previous owner. Others are repeated
frequently, involvi ng some of the .same parties under more or less sim ilar conditions
over a long period of time.
In the first case, one expects the parties to use whatever general mechanisms
are available in the community to control their transaction . In particular, they likely
will resort to a standard form contract, with any disputes between them to be resolved
in court.
In the case of parties who interact frequently, one expects quite a different sort
of mechanism that is specifically geared to the particulars of their relationship. For
example, disputes between a supervisor and worker in a factory are rarely resolved in
courtrooms. Instead, factories may set up a special grievances com mittee involving
the union or other worker representatives, or an ombudsman may be used to hear
complaints and attempt to mediate a solution . The special purpose institution is
worthwhile because it can be tailored to the particular ci rcumstances of the factory,
keep down the cost of resolving disputes, and be conti nually improved in light of the
circumstances in the particular factory. Generally, when sim ilar transactions occur
frequently over a long period of time involving some of the same parties, the one who
interacts repeatedly may find it valuable to design and introduce low-cost routines to
manage the transaction.
Frequency and duration also have another effect. Parties involved in a long,
close relationship with frequent interactions have many opportunities to grant or
withhold favors to one another. The ability to reward faithful partners and punish
unfaithful ones in a long-term relationship greatly reduces the need for any kind of
formal mechanism to enforce agreements between them. The parties can also develop
understandings and routines that reduce the need for explicit planning to coordinate
their actions. These practices can sometimes eliminate the need for formal, detailed
agreements, both because the parties understand what is expected of them and because
they have no need to document those understandings for outsiders to enforce. The
cost savings that result can be considerable.
UNCERTAINTY AND COMPLEXITY The standard way for two parties to organize a market
transaction is to write a contract specifying what is expected from each. If the product
is wheat, then the contract may simply specify that a fixed amount of a standardized
grade of the grain (for example, Manitoba # 1 Northern hard wheat) will be delivered
at a particular date (say, April 1 , 1992) and place (say, Winn ipeg) for a specified price
(for example, $5 Canadian per bushel). The basic contract is then very simple.
In contrast, a contract to build a power plant for an electric utility is very
complicated. The utility's esti mate of demand may change during construction, and
the cost and availability of different kinds of fuel may change as well. The environmental
impact of the facility may be unknown at the outset, and the cost of providing the
necessary envi ronmental and health and safety safeguards may be unpredictable. The
right way to proceed, the length of time to take, and whether to finish the project at
 32
The Problem of all are decisions that will have to be made later, after the contract is signed and
Econom ic execution begins. If the project is changed or delayed or terminated, there will need
Organization to be some way to determine what payments ought to be made.
Uncertainty about the conditions that will prevail when a contract is being
executed together with complexity of the task make it impossible, or at least
uneconomical , to determine in advance what should be done in every possible
contingency, so the contract that is written will generally be less determinate than in
a simpler setting. Rather than specifying how much of what is to be del ivered when ,
the contract may specify who has the right to make which decisions and within what
limits.
Return ing to our example of a maker of wing assemblies for an aircraft, the
contract between the aircraft manufacturer and the supplier m ight deal with uncertainty
about future aircraft sales by specifying that the supplier will provide whatever number
of assemblies the buyer requires according to a particular pricing formula. In return,
the buyer may promise in the contract to purchase wing assemblies only from that
supplier as long as the supplier is able to meet the demand. The buyer might also
promise to supply advance estimates of demand that are within a fixed percentage of
the actual quantities, to subsidize the purchase of specific assets used in the production ,
and so on .
Generally, when uncertainty and complexity make it hard to predict what
performance will be desirable, contracting becomes more complex, specifying rights ,
obligations, and procedures rather than actual performance standards.

DIFFICLTLTY OF PE.RFORl\1ANCE MEASLTREMENT Even when the desired performance is

perfectly predictable, it may be difficult or costly to measure actual performance. For
example, a person who employs a lawyer in a divorce proceeding may have no idea
whether the negotiated settlement is really a good one, or whether a better lawyer
could have negotiated a better deal . Similarly, the low output of a group of workers
in a factory m ight be due to low effort, to poor materials, or to inferior methods used
by the company. When a taxi that has been driven by several drivers over a period of
time breaks down , the owner may be unable to tell which (if any) of the drivers has
abused the car or failed to get maintenance when needed, or whether instead the
breakdown is due to poor design or plain bad luck. Of course, if the taxi doesn 't break
down i mmediately, but hard use now makes future problems more likely, the cost of
that abuse is nearly impossible to measure.
As these examples suggest, it is hard to provide effective incentives unless one
can measure performance accurately. If the lawyer's performance could be accurately
evaluated, or the factory workers', or the taxi drivers', then the consumer, factory
manager, or taxi company owner could hold those parties responsible for their
performance. That would presumably lead to more effort and better results.
When measuring performance is difficult, people commonly arrange their affairs
to make measurement easier or to reduce the importance of accurate measurements.
In our taxi example, the taxi may be assigned to j ust one driver, so that responsibility
for any evident damage can be more easily assigned. Or, the taxi may be driver owned,
so that any loss from abuse or poor maintenance (including even losses that are not
immediately detectable) comes straigh t out of the owner/driver's pocket. Other
attributes of the transaction determine which of these possible solutions is best, or
even whether any of them are workable.

Co��ECTED�ESS TO OTI IEH TH.\NS.\CTI0NS Finally transactions differ in how they

'
are connected to other transactions, especially those involving other people. Some
transactions arc largely independent of all others. For example, an office's decisions
about when to buy new typewriters, where to keep its files, and which supplier to use Economic
for general office supplies hardly need to be coord inated. Organization and
Other transactions arc much more interdependent. When rail roads were Effici ency
introduced in the Un ited States in the nineteenth century, the various railroad
companies failed to coordinate their choices of track gauges (the size of the rails and
distance between them). Because rail cars adapted to one gauge of track cannot be
used on track laid to other gauges, the result was that goods being shipped long
distances had to be unloaded and reloaded onto different cars at several points in the
journey. Standardization on any one of the various gauges that were actually adopted
would have been much more efficient. A similarly costly situation is still present in
Europe, where Span ish rail gauges do not match those used in France. The eventual
standardization of the rail gauges in the Un ited States resulted in much quicker and
less expensive shipping of goods and contributed to the development of the western
parts· of the country.
For a more modern example, suppose a computer maker is design ing a new
model of computer. It cannot deliver a working model until all the relevant components
such as the memory chips, central processor, power supply, and so on are all available
in sufficient quantity to begin assembly. In addition, the operating system for the
computer and some of the appl ication software must be ready; otherwise, there is
nothing useful that the finished computer can actually do.
The manufacturers of the various parts and the software developers in this
situation need to have their activities closely coordinated. There is little advantage to
rushing the completion of a plant to assemble the computer, for example, if the other
parts of the system will not be ready to go. Similarly, the amounts of the various
components that the different suppl iers are able to del iver must line up with one
another, for there is little value to having a larger number of keyboards than disk
drives . The capacities and capabilities of the machine components must be compatible
with one another, and the design tolerances need to be coordinated. Failures to align
capacities or to match design tolerances or to have components ready on time can be
much costlier than failures to adopt the best possible design and introduction date.
When the costliest potential mistakes are of these kinds-rather than, say,
failures to make effective use of local resources-we say that the transactions display
design connectedness. Design connectedness is j ust one extreme; the relative costs of
different kinds of mistakes can in general have any relationship.
One way that firms respond to close connectedness is to strengthen central
coordination mechanisms. This may mean that there are more meetings among the
people involved in the individual transactions, or that the managers in charge spend
more time on oversight, or some combination of those things. A second way fi rms
respond is to reduce the number of different people involved, so that fewer people
need to be coordinated. The particular way that close connectedness is managed
depends on other attributes of the connected transactions.

Limits of the Transaction Costs Approach

This transaction costs approach is appealing, and we adopt it later for some of our
analyses. However, the approach cannot be correctly applied to all problems in
economic organization because, without additional conditions, its fundamental
argument-that economic activity and organ izations are arranged so as to minimize
transaction costs-is problematic. There are two main problems.
First, it is not generally true that the total costs of an economic activity can be
expressed as the sum of production costs and transaction costs, where the former
depend only on the technology and the inputs used and the latter depend only on the
way transactions are organized . Production and transaction costs generally depend
 34
The Problem of both on the organization and on the technology, which makes the conceptual
Economic separation between production and transaction costs troublesome. If production is lost
Organization due to delays in planning, is it the result of slow planning or of a technology that
cannot adapt quickly to late changes in the plan? A more subtle example can be seen
in the semiconductor industry. 4 Integrated circuit production is marked by increasing
returns to scale and very strong learning curve effects, so that costs are lower the larger
the volume of production within any facility, both in any single time period and in
aggregate over time. Thus, efficiency in production would require that any particuiar
design of a circuit be produced by a single manufacturer. For a long time, however,
it was standard procedure for a company that developed a new chip to give the design
to a second firm that would compete with it in producing and selling the chip. It was
even common to assist this "second source" in setting up production.
This way of organizing the production of integrated circuits sacrifices production
cost efficiencies for other advantages: Without a competing second source, potential
users of the new integrated circuit would be reluctant to adopt it for fear that once
they had become locked into its use, the supplier would exploit its monopoly position.
Creating a competing second source is an effective way to achieve commitment,
leading to increased demand. 5 Are the extra costs incurred a "production cost," arising
because an inefficient technology is used that does not take full advantage of economies
of scale, or a "transaction cost," incurred to satisfy customers that the terms of the
transaction are secure? There are no unassailable answers to these questions. The
lesson is that although the costs of transacting are real, they are not always easily
separated from other kinds of costs.
The second problem is not with the concept of transaction costs per se, but with
the notion that efficient institutions would minimize them. For example, according
to Coase's postulate, employment relations can be understood as minimizing the total
transaction costs involved. But why should employers minimize total transaction
costs in designing their employment, compensation, promotion, supervision, and
performance review systems, rather than simply the categories of costs that they
themselves must bear? Some of the transaction costs surely will be borne by the
employees; why should we expect the employers' choices to take proper account of
these? In fact, why would they not push all the transaction costs onto the employees?
A standard answer to these questions is that competition would force employ­
ers to take account of the costs to employees. In Chapter 8, we argue that this stand­
ard answer is of limited application. But even when it does apply, having to rely
on competition or other external forces to bring about efficiency would critically
weaken the theory because its range of potential application would then be severely
narrowed.
A more general version of this second problem is that because there are typically
many quite different efficient solutions to any resource-allocation problem, efficiency
alone may not be a strong enough criterion to give very specific predictions or clear
explanations. Too many different patterns of organization might be compatible with
efficiency for it to be a useful concept.

4 See Andrea Shepard, "Licensing to Enhance the Demand for New Products," Rand /ournal of
Economics, 18 (1987), 360-68, and Joseph Farrell and Nancy Gallini, "Second-Sourcing as Commitment:
Monopoly Incentives to Attract Competition," Quarterly /ournal of Economics, I 03 (1988) , 673-94.
5 The break with this practice came when Intel chose not to second-source its 80386 microprocessor.
The market for this chip was judged to be secure cvcn without second-sourcing because Intel's performance
was secured by its need to compete with its own earlier chip designs, which continued to be produced by
other manufacturers.
 It turns out that one simplifying assumption, that is, the condition of no wealth Economic
effects discussed in the next section, takes care of this problem completely. 6 When Organization and
this condition is satisfied, only one pattern of behavior is consistent with efficiency, Efficiency
and that is the pattern that maximizes the total value created in the transaction.

\VEALTl I EFFECTS, VALUE MAXIMIZATION,

AND THE COASE THEOREM
In many economic decisions, the choice actually made depends on the decision
maker's wealth. A poor person (or a poor country) may not have the resources to
pursue some courses of action that a richer one could. Even when the same alternatives
are affordable, a poorer person might still make different trade-offs than a richer one.
For example, the poor person might be reluctant to take financial risks that a rich
person would welcome. The changes in choices resulting from increased wealth are
known as wealth effects.

The Value Maximization Principle

Although wealth effects can sometimes be significant, this is not always the case. In
fact, the formal analysis of problems in the economics of organizations is greatly
simplified when wealth effects can be ignored entirely. Moreover, it is precisely in
ignoring wealth effects that such key management concepts as "creating value" become
unambiguously defined.

No WEALTH EFFECTS We say that there are no wealth effects for a certain decision
maker with respect to a set of possible decisions when three conditions hold. First,
given any two alternative decisions y 1 and y2, there is a definite amount of money $x
that would be sufficient to compensate the decision maker for switching from y1 to y2
(or from y2 to y 1). Second, if the decision maker were first given an additional amount
of wealth, then the amount needed to compensate the decision maker for the switch
from y 1 to y2 would be unaffected. Third, the decision maker must have enough
money to be able to absorb any wealth reduction necessary to pay for a switch from
the less preferred to the more preferred option.
None of these conditions can be expected always to hold. For some people, for
example, there may be no amount of money that they would accept as compensation
for a serious risk of loss of life or limb, or for being forced to live far from family and
their childhood homes, or to live in a culture where they cannot exercise their religious
beliefs. Nevertheless, the condition that there is some monetary amount that would
compensate for a change of circumstances holds widely for many of the most common
kinds of business decisions.
To examine the implications and applicability of the second condition, suppose
that a corporation suddenly finds itself richer on account of an unexpected increase
in the value of its assets. If there are no wealth effects, then the price the corporation
would demand for its goods and the returns it would demand from its planned
investments would remain unchanged. In this example, the absence of wealth effects

6 Like other modelers, economists use the terms assumption and assume in a different sense than
they are often used in ordinary discourse. In everyday conversation, assuming something connotes believing
it is true. Making an assumption in an economic model carries no such connotation. An assumption is
merely a working hypothesis used to abstract from the complexity of the real economic world. The purpose
of making an assumption may be to derive a good approximate prediction or to highlight a single force or
effect for closer study and better understanding. Assumptions are used in both ways in this text.
 36
The Problem of seems likely to hold, at least over a broad range of wealth levels. As a second example,
Economic however, suppose a worker wins the jackpot in the state lottery. If there are no wealth
Organization effects with respect to current consumption choices, then the winner would not buy
a new house, or quit his or her job, or do any of the things lottery winners normally
do. Here the assumption of no wealth effects seems particularly inappropriate.
The third condition connects the decision maker's initial wealth and the changes
in the nonfinancial situation that are being considered. The amount by which a
worker's pay could be cut to offset his or her psychic gain in being allowed to shift
working hours to miss rush-hour traffic is probably small relative to income, and the
condition is likely to hold in this case. On the other hand, holding a worker in a
nuclear reactor facility financially responsible for the effects of his or her mistakes is
apt to run into wealth constraints, and so the assumption of no wealth effects will be
inappropriate.
In general, these examples suggest that the assumption of no wealth effects is
most restrictive-least likely to be valid and most likely to lead to incorrect
conclusions-when the decision makers are individuals and when large cash transfers
or significant changes in personal living conditions are involved. When the sizes of
the cash transfers are small relative to the decision maker's financial resources,
assuming that there are no wealth effects (or that they are small enough that they can
safely be ignored) is more likely to be a good approximation to reality.
THE EQU\'ALENT VALL1E INDEX The utility function of a decision maker who shows
no wealth effects with respect to a set of decisions can be represented very simply.
Let x represent the decision maker's monetary wealth and let y be a list of all
the other influences or characteristics associated with decisions that affect his or her
preferences: social approval, job assignment, effort exerted on the job, and so on. An
important example involves uncertain income and expenses, where x is interpreted as
the certain, unconditional amount of money that will be received (or the average,
expected amount) and y reflects a risky component of income. In general, the utility
function takes the form u(x,y), where x and y may interact in complex ways. If there
are no wealth effects, however, then there is always a cash equivalent value v(y) that
can be assigned to the list y and the decision maker's utility function can be written
in the form u(x,y) = x + v(y). In other words, by adding the cash equivalent value
v(y) to the decision maker's wealth x, we obtain an index of personal welfare, which
we may call his or her value index. 7 The importance of the value index is that, when
it is valid, a related index-the total value of the affected parties-is an appropriate
measure of welfare changes for group decision making. 8 We state the matter as follows.
The Value Maximization Principle: An allocation among a group of
people whose preferences display no wealth effects is efficient only if it
maximizes the total value of the affected parties. Moreover, for any
inefficient allocation, there exists another (total value maximizing) alloca­
tion that all of the parties strictly prefer.

- To see how the three preceding conditions relate to this formula, note first that the change in x
necessary to compensate for a change from y1 to y2 is easily computed to be � x = v( y2)- v( y1), because
for any given initial wealth M, M +�x+v(y2) = M+v(y1) . The change in xto compensate for switching
from y2 to y1 is then just - �x. The calculated amount �xis independent of the initial wealth, 1\1, as the
second condition requires. With a general utility function u(x, y), there might be no amount � that would
make the equality u(M +�.y2) = u(M, yi) hold, as the first condition requires, and even if there were such
an amount, its magnitude would generally depend on M. Finally, as long as the initial wealth M is larger
than the largc�t possible difference in the v( y) values between two alternative )'S, so that making the transfer
� x cannot require more money than the individual has, then the third condition holds.
� The total value is also sometimes called the total (consumer and/or producer) surplus.
 :n
Economic
Organization and
Efficiency

Figure 2.1: Any point like A on a line of lower

total wealth is Pareto dominated by some point
B on the line of highest total wealth. Points like
B on the line of highest total wealth are undom­
1's Wealth inated.

The Logic of Value Maximization

To establish the principle for a particular example, consider an investment
decision involving two people with utility functions that satisfy the condition of
no wealth effects: U/x,y) = x + v;(y), i= 1,2, where y represents inputs to be
provided by the parties. 9 The investment generates total cash income of P(y).
We may think of v;(y) as representing the personal cost investor i bears in
supplying the agreed inputs. In that case, v;(y) is a negative number for positive
levels of y. The income P(y) will be divided between the individuals, with x1
being paid to individual 1 and x2 being paid to individual 2, so that x1 + x 2 =
P(y). For any particular allocation (x1 ,x2 ,y), the total utility or value of the two
parties is [x1 + v 1(y)] + [x2 + vz(y)], which (because x1 + x2 = P(y)) is equal
to P(y) + v 1 (y) + vz(y). The total value depends on y alone and not on the
profit shares x. As we vary the profit shares x1 and x2 , the individual utilities of
the two parties change, but the total utility remains fixed.
Figure 2.1 illustrates the situation. Each line depicts the possible levels of value
for the two parties for any fixed investment decision y, as the profit shares x are
varied. The fact that the lines are straight and run at 45 ° to the axes reflects the
fact that total value is independent of its distribution between the parties; it is
possible to move utility or value from one party to the other (via the xs) without
affecting the totals. It is clear from the diagram that any point like A on a line
that corresponds to lower total value is Pareto dominated by some point like B
on the line of highest total value. It is also clear from the figure that no point
on the line of highest total value is Pareto dominated by any point on that line
or any other line. Consequently, an allocation (x1 ,x 2 ,y) is efficient if and only
if y maximizes the total value: P(y) + v 1(y) + vz(y).
The intuition of this result is simple: With more total it is always possible to
distribute it in a way that makes everyone better off. A full mathematical proof
of the proposition is developed in the exercises at the end of the chapter. 10

9 We can think of y as being the pair (y ,y ), where y, is i's contribution. Then writing v;(y) rather
1 2
than V;(y;) allows the possibility that each individual's costs depend on both parties' contributions. At the
same time, writing v;(y) allows that V; depends only on Y;-
IO In the presence of wealth effects, an allocation that maximizes the sum of utilities is still efficient,
but there may also be efficient allocations that do not maximize the sum of utilities.
 38
The Problem of .\PPL\1:--JC \'ALLIE l\lnl\llZ,\TIO:\Although this discussion is set in the context of two
Economic people making an investment, the principle itself is much more general. When
Organization preferences take the form we have described, then any decision (x,y) is efficient if and
only if y is chosen to maximize the total value of the parties. Importantly, the efficiency
of the choice (x,y) does not depend on the selection of x, which determines only how
the fruits of the joint enterprise are shared. When the value maximization principle
applies, the issue of the distribution of value is completely separable from the issue
of how value is created. While this separation is not always realistic (see Chapter 8),
it is often reasonable and it does always simplify the analysis of economic organization
problems. For that reason, it is a good starting point for thinking about organizations.
The abstract model we have described has wide application because the variable
y can be given so many different interpretations. The model can apply to people in
a community who must decide on the resources y to be devoted to parks, libraries,
or some other public good. It can also apply to overtime assignments (y is then the
identity of the person assigned to work overtime), to water rights (y identifies the owner
of the rights), to the brand of computer to be used (y names the brand), or to the
location of the new office building (y specifies the location).

The Coase Theorem

In practical business situations, the way the benefits of an agreement are divided
among the parties will depend, of course, on what assets each brings to the bargaining
table, how patient each party is, what outside opportunities are available, and so on.
Nevertheless, if the parties engage in efficient bargaining, that is, if they reach an
agreement from which there is no possibility of further mutual gain, and if the value
maximization principle applies, then regardless of what cash changes hands at the
time of agreement, y will be chosen to maximize the total value of the parties to the
agreement. Only the distribution of the costs and benefits will be affected by the
strength of their relative bargaining positions, and this distribution will show up in
the xs. This conclusion is summarized in a proposition that is also due to Coase:

The Coase Theorem: If the parties bargain to an efficient agreement (for

themselves) and if their preferences display no wealth effects, then the
value-creating activities (y) that they will agree upon do not depend on
the bargaining power of the parties or on what assets each owned when
the bargaining began. Rather, efficiency alone determines the activity
choice. The other factors can affect only decisions about how the costs
and benefits are to be shared (x).

This celebrated proposition is the foundation of the transaction costs approach

to the theory of the firm and other economic organizations. With the assumption of
no wealth effects, the Coase theorem and the efficiency principle mean that all real
activities are determined to maximize the total value of the parties, taking into account
the costs of organization (transaction costs) along with all other kinds of costs. For
any given plan of production specifying who is to make what with which resources
(and thus the aggregate production costs that will be incurred), if we think of transaction
costs as the costs of managing the transactions (including the costs of writing contracts,
supervising workers, enforcing contracts, and resolving disputes), then the efficient
organization for that plan is the one that minimizes transaction costs.
To understand the significance of this perspective, it is helpful to contrast it
with other perspectives that have been vigorously advocated.
The Transaction Costs Approach versus Alternative Views Economic
Organization and
The transaction costs approach differs sharply from the Marxian approach. According Efficiency
to Marxian theorists, organizational arrangements are a reflection of underlying power
relationships and class interests. In contrast, according to the transaction costs
perspective, the choice of a firm's organization (y) does not depend on the a priori
distribution of power between the owners of capital and the laborers. For example,
in Yugoslavia, where employee control of firms has been the norm, labor hired
managers who organized the factories in similar ways and instituted similar kinds of
controls (even over labor) as those seen in capitalist firms.
Applied to relationships among firms and between firms and their customers,
the transaction costs approach suggests that business arrangements should be understood
as attempts to increase the total wealth available for sharing among the parties.
Scholars trained in the Harvard schoof of industrial organization and antitrust would
try instead to explain the arrangements as attempts by the firms to increase their ability
to manipulate prices for the products they sell or the inputs or labor they buy. For
example, resale price maintenance-the practice by manufacturers of limiting
contractually the rights of distributors and retailers to set prices-has been attacked
by members of the Harvard school as anticompetitive. The transaction costs approach
suggests that it in fact may be efficiency promoting. Adherents of this approach might
argue that, without resale price maintenance, discounters would free-ride off the
service and customer education provided by other dealers, relying on them to provide
product support but then grabbing the sales for themselves. This would drive the
amount of such support below efficient levels.
The correct way to decide between competing hypotheses is to confront them
systematically with detailed evidence. The observation concerning management
methods in the Yugoslav firm represents an instance of this, but it is by itself
insufficient. We report more of the relevant evidence later in this book.
In any case, in application, it is important to remember that the Coase theorem
and its various implications depend on restrictive hypotheses regarding preferences
and, perhaps more importantly, on the ability to make unlimited transfer payments
between the parties. The implications do not hold when some of the parties have very
limited capital with which to make payments. Thus, although it would be reasonable
to apply this analysis to study the terms of a contract between General Motors and
Toyota, it would be a mistake to apply it uncritically, for example, to study land
tenure in a developing country or the institution of slavery in the pre-Civil War
American South.

ORGANIZATIONAL OBJECTIVES
It is traditional in economics and management texts to assume that firms seek to
maximize profits and, more generally, to ascribe well-defined goals to organizations
and to presume that the organization acts in pursuit of these objectives. Occasionally
we employ such a hypothesis in this text, when it is convenient to treat the organization
as a purposeful entity. More often, however, we do not presume that organizations
per se have goals that they seek to realize. Rather, we are careful to treat organizational
decisions and actions as the outcomes either of strategic interplay among self-interested
people responding to incentives designed to influence their behavior, or of collective
or managerial attempts to compromise the interests of the parties affected by the
decisions. Only when the value maximization principle applies is there an objective
that we can ascribe to the firm that is implied by considerations of efficiency alone.
 Profit J\laximization
40
The Problem of
Economic
Organization The goal most commonly ascribed to firms in economic analyses is profit maximization.
It might seem at least that the self-interested owners of a firm would unanimously
favor such a goal. In that case, they would attempt to design the organization to
motivate its managers and employees to pursue profits. In fact, we often conduct our
analyses presuming that this is the case, but the reader should remember that there
are many reasons why owners might have other objectives.
First, to the extent that one of the owners is also a customer of the firm or one
of its input suppliers , that owner might prefer that the firm not maximize profits in
dealing with him or her, but instead favor the owner with better prices or terms. This
would be a potential problem in a firm where ownership is shared between employees
and outside investors, with each owning part of the claims on the firm. The inside
employee-owners might prefer that the firm's managers adopt policies to protect
workers' jobs, pay high wages, and provide many on-the-job benefits. Meanwhile, the
outside investor-owners might prefer that the value of their investment in the firm be
maximized.
Second, many of the decisions of the firm involve expenditures and receipts
that are both uncertain and spread out over time. In such cases, it is common to
assume that people are interested in the expected value of the discounted stream of
utility they receive over time in the various uncertain future circumstances (see
Chapter 1 2). However, just as owners differ in the likelihoods they place on the
various ways future events might unfold and in the relative weights they place on
income accruing in the more or less distant future, they also will disagree on which
plans maximize the expected present value of profits. This is especially likely to be a
problem when the firm is considering investing in a new process or product whose
benefits and costs are unknown.
A partial solution to this difficulty is achieved if there are so-called complete
and competitive markets (a concept investigated in more detail in Chapter 3). Then
maximizing the market value of the firm is an appropriate goal to which the owners
would agree. In this context, having complete and competitive markets means that
any individual can use the financial and insurance markets to move income across
time and shift it between different uncertain events, all at given prices. In doing so,
the individual achieves whatever patterns of receipts he or she may like and can afford.
In such circumstances, it is best to make the firm's value (evaluated at the given
market prices) as large as possible, because this provides the largest amount to invest,
and this amount can then be invested in light of the person's own preferences and
beliefs about the future (see Chapter 1 4). However, markets are almost surely not
complete in this sense. Thus, there are disagreements among owners about the optimal
course of action for the firm, and market value maximization does not always win
unanimous approval.
Third, to the extent that those making the decisions are not the only claimants
on returns from the concern, they may wish to maximize their portion of the flows,
rather than total returns. For example, suppose the owners are stockholders with
limited liability who are at risk for the firm's obligations only up to the amount of
their investments, but some of the firm's financing is via debt that must be repaid
before the owners get any returns on their investments. In that case, the owners might
prefer to make investments that are excessively risky rather than to maximize the value
of the firm (including both the value of its debt as well as its equity). When things
go well, the debtors are paid their contracted amounts and the equity holders keep
the remainder. When things go badly, the debt is not repaid in full and some of the
losses arc thus shifted onto the debtors. Riskier investments shift more of the downside
 41
on to the debtors and leave the upside to the stockholde rs. Therefore, they might be Economic
preferred by the latter group, even if the increased risk reduced the val ue of the debt Organization and
by more than it increased the value of the stock. We explore this particular conflic t Efficiency
again in Chapter 5 in the context of the savings and loan industry.

Other Goals and Stakeholders' I nterests

Of course, there are many organizations where profit maximization is clearly not the
goal. M utual insurance companies reduce the premiums paid by their policyholders
through dividends (distributi ons of part of the excess of receipts over costs). Customer­
owned cooperatives are intended to sell to their members at lower prices than profit
maximization would yield. Clearly, assuming profit maximization is likely to be
inappropriate for such firms. H owever, in each of these cases the customers (the
policyholders and the co- op members) are the nominal owners, and usefu l analysis
can be conducted using the assumption that the organization is structured to attempt
to serve the{r interests. Similarly, in the case of a firm that is entirely employee owned
such as Avis, the automobile rental company, or Egged, the Israeli intercity bus
company, an examination of the employees' interests should give insight to the policies
that the firm will pursue.
Even if we are willing to assume that a privately- owned firm will be designed
to serve the interests of the owners, however, in whose interests is a research university
run? S urely, "in the public interest" is an inadequate answer. The public consists of
a multitude of people whose interests may be in conflict. S tudents may prefer faculty
to be selected largely for teaching ability; employers may want teaching that gives
students professionally applicable training and research that is geared to quick industrial
application; alumni booster clubs might prefer that more resources go to generating
winning athletic programs; taxpayers might want subsidies held down and only local
students admitted. Generally, not-for-profit organizations have no owners in the usual
sense. In such circumstances, predicting organizational form and behavior requires
careful analysis of who has the power to design the organization, who can make
decisions, and who can influence these decisions and their implementation.
A similar difficulty arises in firms in which the interests of stakeh olders other
than the nominal owners are given legitimacy. For example, a survey of the presidents
of I 00 maj or Japanese firms asked what the obj ectives of companies should be, and
what their actual goals were. 1 1 In each case, the pursuit of shareholders' profit came
a very distant fourth on the list, garnering only 3 . 6 percent of the responses. Asked
the questions of to whom companies should belong and to whom they actually do
belong (with multiple answers allowed), senior executives listed shareholders first
among those entitled to ownership, but employees first among the actual owners. A
survey of Japanese middle managers indicated that they viewed stockholders as onl y
fourth in a list of those to whom companies should belong, with employees first.
Shareholders came third on the middle- level managers' list of the effective owners of
the firm, behind employees and management. 1 2
Management in other countries also often appears to be concerned with the
interests of the nonowner stakeholders, including employees, suppliers, customers,
the communities in which the firm is located, and those for whom the natural
environment is affected by the firm' s actions. To the extent that the nominal owners
cannot enforce exclusive attention to their interests in management's decisions, the

11 Reported in the Nihon Sangyo Shimbun , July 5, 1 990. We are grateful to Masahiro Okuno­
Fujiwara of the University of Tokyo for this and the next reference.
12 Reported in Nihon Sangyo Shimbun , April 2 3 , 1 990. Multiple responses were allowed.
 42
The Problem of actual pol icies and practices of the firm are going to represent a pol itical compromise
Economic mediated by management.
Organization In summary, the assumption that organ izations are maximizing entities with
well-defined goals is one that should be made very cautiously.

MODELING HUI\IAN MOTIVATION AND BEHAVIOR

A central premise of economic analysis is that people (as opposed to organ izations)
do have well-defined interests describable by individual util ity functions, and that they
seek to maximize thei r util ity. Although this assumption is far from uncontroversial,
it in fact has virtually no empirical content in light of the limitless factors on which
individual utility could depend.
For example, a sufficient concern both for the well-being of others and for social
approval could rationalize apparently extreme self-sacrifice. There is no doubt that
such factors are both real and, in some cases, immensely important. Although a
narrow calculation of personal costs and benefits might usually be enough to prevent
soldiers from deserting in the face of battle ("Am I more likely to be shot by the enemy
or by a firing squad?"), sacrificing one's life for one's comrades seems hard to explain
without appeal to altruism or to an exceedingly high regard for others' opinions of
one's courage. Such factors are important even in the more mundane problems that
concern us here, such as motivating workers to provide honest effort or providing
incentives for borrowers not to skip out on thei r creditors. Furthermore, important
features of many organizations can be best understood in terms of deliberate attempts
to change the preferences of individual participants to make these factors more salient.
As a result, organizationally desired behavior becomes more l ikely. This is clearly an
element of leadership as it is usually understood, and it has much to do with practices
of orga nizing semipermanent groups of workers and encouraging them to interact
socially as well as at work.

Rationalit y -Based Theories

While admitting all this and, in particular, the possibility of intrinsic motivation , we
adopt the view that many institutions and business practices are designed as if people
were entirely motivated by narrow, selfish concerns and were quite clever and largely
unprincipled in their pursuit of their goals. This model of human behavior incorporates
the sort of rational self-interest usually assumed in standard economic models, such
as that of a consumer making utility-maximizing purchases at given prices. But it goes
further. It posits that people will be very sharp in discovering even subtle ways in
which they can advance their interests and that they will be fundamentally amoral ,
ignoring rules, breaking agreements, and employi11g guile, manipulation , and decep­
tion if they see personal gain in doing so.
As we shall see, this assumption does have bite: It often serves to give sharp,
testable predictions and explanations. Moreover, · even though it is an extreme
caricature to regard people as amorally motivated solely by narrow self-interest, the
predicted institutions and practices are often not very sensitive to this caricature. A
bank has guards, vaults, and audits because it would otherwise be robbed; this
explanation of practices is unaffected by the observation that many hon est people
would not rob an unguarded bank. Sim ilarly, organizations design reward schemes
so that employees find it in their personal interests to work to advance the organization's
goals . The fact that many employees would not instantly abandon the organ ization's
interests even if their inccnti\'CS were removed is not particularly important for our
analysis .
At the sa me time, we do not automatically make the hyperrationality assumptions
common in some economic analyses: that people are capable of instantaneous, Economic
unlimited, perfect, and costless calculation, that they can effectively and effortlessly Organ ization and
forecast all possible eventualities and the full im plications of any information or Efficiency
decision, and that they completely optimize in all si tuations. These assumptions arc
not j ust counterfactual; they also prevent understanding of many important clements
of organi zations. For example, organizations regularly employ routines-standard
operating procedures and rules of th umb-for obtaining information, making and
implementing decisions, and carrying out tasks. Using routines economizes on scarce
and valuable decision-making resources, although it sometimes means that the
decisions that are made are not the best ones that could have been reached if the
problems were subjected to a full analysis rather than treated routinely. Organizational
learning occurs when these routines are modified in response to new knowledge.
Non� of this makes much sense if we assume hyperrationality.
Paradoxically, the very imperfections in the rationality of people and in the
adaptability ·of organizations denied by many simple economic theories are necessary
in proving that the rationality-based theories are descriptively and prescriptively useful.
With perfect rationality, one would rarely expect to observe two organizations in
substantially the same circumstances making substantially different choices, so there
would be no possibility of testing what kinds of organizations perform best. For
example, to test whether the commonly observed scheme of paying commissions to
compensate insurance agents is an efficient system of sales incentives, we would want
to compare these firms to other firms that do not provide such incentives to evaluate
their performances, or possibly to the same firm using different practices at different
times. It is untenable to adhere too closely to tenets of individual and organizati onal
rationality and at the same time to claim an empirical basis for the- theory. A more
defensible position, suggested by Richard Nelson and Si dney Winter, is that people
learn to make good decisions and that organizations adapt by experimentation and
imitation, so that there is at least "fossil evidence" available for testing theories.
Nevertheless, theories based on perfect rationality and adaptability are surprisingly
successful in generating explanations and specific predictions about observed institu­
tions and business practices, and they form the main focus of this book.

CASE STUDY : COORDINATION, MOTIVATION, AND

EFFICIENCY IN THE MA RKET FOR MEDICAL I NTERNS
The evolution of the system under which graduating medical students in the United
States are matched with hospitals seeking interns and residents provides a striking
illustration of a number of the themes and concepts introduced in this chapter. 1 3

Match ing Problems and Failed Solutions

The practice of new M. D. degree holders taking internships in hospitals as a clinical
stage in their medical education appeared in the Uni ted States about the beginning
of the twentieth century. It gave the interns practical training and the hospitals cheap
labor. Until graduates of foreign medical schools began seeking U. S. internships in
significant numbers in the 1 970s, the number of positions open in hospitals for interns
always exceeded the number of students seeking internships, so competition for interns

13 This section is based on two papers by Alvin E. Roth, "The Evolution of the Labor Market for
Medical Interns and Residents: A Case Study in Came Theory, " foumal of Political Economy, 92 (December
1 984), 99 1 - 1 0 1 6, and "A Natural Experiment in the Organization of Entry Level Labor Markets: Regional
Markets for New Physicians in the U. K. , " American Economic Review, 8 1 (June 1 99 1 ), 4 1 5-40 See also
Roth's l ess technical discussion, "New Physicians: A Natural Experiment in Market Organization, " Science,
250 ( December 1 4, 1 990), 1 524-28.
 44
The Problem of was intense. Of course, students differ in their overall attractiveness and in their
Economic specifi c appeal to different schools, so the competiti on was more intense fo r some
Organization than for others. M oreover, hospitals seeki ng several interns might be concerned with
the particular mix of students they got, and not j ust with the overall quality of the
individuals they attracted. They would consequently have preferences over groups of
students. At the same time, students may have differing individual preferences among
the hospitals at which they might i ntern. I n sum, the fi nal matching of students and
hospitals is of great concern to all.
Various methods were used to match students and hospitals over the years. First
was a scheme similar to that used in U . S . college and graduate school admissions
today. Students would apply to hospitals that were seeking interns. The hospitals then
ranked their applicants and offered positions to some number while telli ng others that
they were alternates on a wait-list. The students then had to decide whether to take
one of their firm offers or to wait to see if a hospi tal they preferred would make them
a firm second-round offer. The problem was that the hospitals began competing by
making their offers earlier and earlier, sometimes as early as only half- way through
the students' medical school studies. This was very bad from the point of view of the
hospitals, because at that stage they could learn so li ttle about the students' ultimate
performance and interests. H owever, each hospital still had an indi vidual incentive
to accelerate its offers to try to recruit the best students. The process was also very
disruptive for the medical schools and the students' studies.
The eventual response was for the medical schools to agree not to give out
information on their students until late in their careers. This moved the recruiting
into the students' fi nal year of medical school, but a new problem arose. Students
who had been accepted by one of their less-preferred hospi tals but wait-listed by one
they liked better would hold off responding to the first hospital's offer as long as
possible, hoping that they would clear the other hospital's wait-list. Again, this was
individually optimal behavi or by all concerned, but it meant that there were student­
hospital matches that were missed but that were mutually preferred to those that
actually occurred. This in turn led to students' reneging on their acceptances.
Attempts were made to overcome these problems. The chief one was for the
hospitals to agree on the earliest date for making offers, for them to shorten the time
given to students to make their deci si ons (down to as little as a twelve-hour period
from midnight to noon!) and to limit communication between the hospi tals and
students during the period that offers were outstanding. Perhaps unsurprisingly, these
measures proved futile, and the turmoil continued.

The National I ntern Matching Program

I n 1951, a centralized scheme was introduced by the hospitals and medical schools
on an experimental basis to coordinate the matching of students to internships. The
workings of the experimental version were adj usted in light of students' observations
that they would do better under its rules to misrepresent their preferences among
hospitals, and the revised scheme was formally instituted in 1952. U nder i t, students
seeking internships and hospitals seeking interns first exchange information with one
another, much as before. Then the students rank the hospitals in wh ich they are
willing to be employed and the hospitals similarly rank the groups of student applicants
whom they are willing to take. These rankings are then submitted to a central office,
which uses a specific rule (algorithm) to match students and hospitals. This new
system, called originally the National I ntern Match ing Program (NIMP), ended the
turmoil that had marked the market for interns. I t has remained in place, drawing
th e volunta ry pa rticipation of the vast bulk of hospitals and medical students (up to
95 perce nt of stude nts). However, begin ning in the mid- 1 970s there was an increasing
 45
number of married couples needing two internships in the same vicinity, and large Economic
numbers of these people sought and found matches outside the NIMP system. Organization and
The actual algorithm used to match students and hospitals is somewhat Efficiency
complicated, and understanding its detailed workings is not necessary for what follows.
However, the basic idea can be explained simply for the (unrealistic) case in which
each hospital has space for only one intern and there is an equal number of students
and hospitals. 1 4
The algorithm works through rounds in which it seeks to match students and
hospitals on the basis of their submitted preference orders. In essence, each hospital
at each round offers its position to its most preferred applicant. Students offered a
place are then tentatively matched with their favorite among the hospitals making
them an offer. Their names are then stricken from the submitted rankings of all those
hospitals which they ranked lower than the one with which they are tentatively
matched, and the process is repeated using the hospitals' revised rankings. Note that
.
generally some students will have moved up in some of the hospitals' revised rankings,
because other students whom these hospitals actually liked better have been (tentatively)
matched elsewhere. This means that students who have moved to the top of some
hospital's rankings will now get offers from that hospital. If they were already tentatively
matched with some other hospital but prefer the new offer, the old tentative match
is broken and they are (still tentatively) assigned to their preferred choice. The process
then continues until everyone is tentatively assigned, at which point the assignments
are made final and announced. 1 5
EFFICI ENCY AND STABILITY Consider now the general case where the hospitals have
multiple slots for interns and there is no presumed equality between the numbers of
students and available spaces. Even in this context, the matches that the N IMP
generates can be shown to be effi cient in the usual sense: There is no way to reassign
the students to the hospitals so as to make one of the students or one of the hospitals
better off without hurting some other hospital or student. This was not necessarily
true of the older systems it replaced. Moreover, the NIMP avoids many of the
transaction costs that marked the older systems. The process is relatively straightforward,
and most students and hospitals have accepted its results rather than trying to go
around it to find mutually preferred matches.
However, in the present context, monetary transfers and "side payments" are
not permitted. Thus, efficiency cannot be equated with value maximization because
compensation cannot be paid. There may then be many different efficient patterns of
matches. In light of this, it is especially significant that the matching generated by
the NIMP can be shown to enjoy an extra property that is much stronger than
efficiency in this context: the NIMP matches are stable. Specifically, there will never
be a hospital-student pair where the student prefers the given hospital to the one to
which the algorithm assigned him or her and the hospital would rather have the given
student than one of the students actually assigned to it. Thus, recontracting between
individual students and hospitals cannot upset the outcome of the process, even when
the hospitals are free to drop people who have been assigned to them and students
are free to ignore the assignment they have been given.
Stability means that the proposed assignment is not only efficient for the group
as a whole, it is also efficient for every subgroup, even if that group ignores the effects

14 We are indebted to Robert B. Wilson of Stanford University for this interpretation.

1 5 Note that each student gets a more preferred hospital whenever his or her tentative match is
changed. Thus, with only a finite number of hospitals, the algorithm eventually must cease generating
changes, at which point it stops.
 46
The Problem of Table 2. 1 An Example of Student and Hospital Preferences
Economic
Organization
Ranking Student Hospital
Alice Barbara Charlie Hopkins Stanford Yale
(A) (B) (C ) (H) (S) (Y)

1st y s s A A B
2nd s y y B B A
3rd H H H C C C

of their decisions on non-members. Regardless of what group of students and hospitals

form to seek alternative matches, the members can never find a set of matches that
involves only members of the group that is better for all of them than the match
initially proposed by the NIMP. Stability is a very demanding condition. The fact
that the NIMP matches are stable helps to explain the persistence of the program.
It may seem remarkable that the hospitals and medical schools were able to
devise an efficient, stable system. What is perhaps more remarkable is that this scheme
is ideal from the hospitals' point of view (and quite the opposite from the students'
perspective).
If a hospital wants some specific number n of interns, then the NIMP algorithm
actually assigns that hospital the n students it ranks highest among all the students it
could ever get at one or another stable assignment (no matter how computed)! So if
one stable assignment gives a hospital its first and fourth choices and another gives it
its second-ranked and third-ranked students, the NIMP actually gives it its first- and
second-ranked students. At the same time, each student is assigned to the hospital he
or she ranks lowest among all those to which he or she could be assigned at some
stable assignment.
All this presumes that the students and hospitals submit lists reflecting their
actual, true rankings. In this regard, the NIMP consistently claims that there is no
advantage to misrepresenting preferences. In fact, no student or hospital can gain by
misrepresenting its first choice. However, it is not individually optimal for students to
list their complete true preferences independent of the hospitals' rankings and the lists
being submitted by other students, or for a hospital with more than one slot to fill to
submit an honest list independent of the students' lists and those of the other hospitals.
AN EX..\!\IPLE OF SUCCESSFUL STRATEGIC l\1tSREPRESENTATIO:\l To see how misrepresen­
tation could theoretically be beneficial, consider an example that is remarkably simple
compared to the real situation faced by actual students. 16
Suppose there are three students, Alice, Barbara, and Charlie, and three
hospitals, Hopkins, Stanford, and Yale. Each hospital wants only one student, so by
the results claimed earlier, they have no reason to misrepresent their preferences. The
assumed rankings are given in Table 2. 1 .
At the first round, Hopkins and Stanford offer positions to Alice, the most
preferred student, and Yale offers one to its favorite, Barbara. Alice is tentatively
matched with Stanford, which she prefers to Hopkins, and Barbara is tentatively
matched with Yale. Thus, both Alice and Barbara are removed from Hopkins's
rankings, moving Charlie to the top (by default as he is the only one left). At the next
round, Stanford and Yale repeat their offers to Alice and Barbara, respectively, because

16 We are indebted to Jeremy Bulow of Stanford University for this example.

 47
Table 2. 2 Hospital's Revised Rankings at Economic
Round III of the Algorithm Organ ization and
Efficiency

Ranking Hospital
H s y
1 st B A A
2nd C B C
3rd C

each is at the top of the corresponding school's list, and H opkins make s an offer to
Charlie. All the students and hospitals are now matched, and these matches are the
ones actually announced. Al ice and Barbara end up with their second-fa vorite
hospitals, but these two hospitals get their first picks.
N ow suppose that Barbara misrepresents her preferences and submits a list that
still ranks Stanford first, but then puts H opkins rather than Yal e in second place.
Tracing thr ough the workings of the computer al gorithm is much more compl icated
in this case (and you may prefer to skip the next two paragraphs, which describe the
pr ocess), but it turns out that this misrepresentation will work to Barbara's advantage.
S he ends up being assigned to her first choice, S tanford.
At the first round, Alice is again tentatively matched with Stanford and Barbara
with Yale. N ow, however, since Barbara claims that she views Yale as third best, her
name is not removed from H opkins's list. Alice's name is, however, still taken off
H opkins's list, which now lists Barbara first and Charlie second. At the next round,
H opkins and Yale make offers to their now-fa vorite, Barbara, and S tanford repeats its
offer to Al ice. Barbara claims to like H opkins better than Yale, so her tentative match
with Yale is broken, and she is matched provisionally with H opkins. As a resul t, she
is removed from Yale's ranking. The revised rankin gs of the hospitals going into the
third round are given in Table 2. 2.
At the third stage, each hospital again makes an offer to the top-ranked student
on its revised listing: H opkins to Barbara, and S tanford and Yale to Alice. The tentative
matches set by the computer are Barbara again with H opkins, but Alice with her top
choice, Yale. Thus, Alice is dropped from S tanford's list, moving Barbara to the top.
N ow, at the fourth stage, Barbara gets offers from both H opkins and S tanford, while
Alice again is Yale's choice. The computer now matches Barbara with S tanford, her
first choice, and removes her from H opkins's list. Consequently, at the fifth round,
the computer matches Alice with Yale, Barbara with S tanford, and Charlie with
H opkins. This is the match that is announced.
Barbara's misrepresentation allowed her to get her first choice rather than her
second, as she would have gotten by being honest. (I ncidentally, it also moved Alice
from her second-ranked hospital to her first. ) B y downgrading the hospital that ranked
her first and would offer her first-round admission, Barbara manages to stay on H opkins's
list after being first matched with Yale, while Alice is dropped from H opkins's list.
Then Barbara is offered admission by H opkins, which all ows her to eliminate herself
from Yale's list. This leads Yale to take Alice, opening the spot at S tanford for Barbara.
Figuring out an advantageous preference misrepresentation was not obviously straight­
forward, since it depended on knowing the rankings of both the hospitals and the other
students as wel l as the workings of the algorithm, but one existed.
I n fact, medical students do spend time and energy trying to learn what the
 48
The Problem of various h ospitals are looking for, what their historical patterns of rankings have been,
Economic and what other students are likely to do, h oping thereby to figure out an ad vantageous
Organization way to misrepresent their preferences. H owever, as the example suggests, figuring out
a worthwhile strategic misrepresentation is subtle and difficult. It is quite possible that
students eventually give up trying to game the system and report accurately.

The Evolution and Persistence of Organizational Forms

The logic introduced in this chapter suggests that the efficiency and stability of the
NIMP h ave contributed crucially to its survival, j ust as the failure of the earlier systems
can be attributed to their inefficiency and instability. Wh en an inefficient arrangement
is in place, there is a general interest in supplanting it with one that will make
everyone better off. When the arrangements are unstable in the sense we h ave
discussed here, they are particularly fragile because pairs of agents have both the
incentive and the ability to subvert their workings and presumably will do so.
This argument is supported by consideration of the mechanisms that have been
used in the U nited Kingd om to address the problem of matching new physicians and
hospitals there. The situati on in the U nited Kingd om is more complex than in the
U nited States, because there are seven regional markets, each of which has tried its
own matching algorithms after a period of instability similar to that experienced in
the U nited S tates before 1951. Eight of these different algorithms have been analyzed
formally, of wh ich two were found to be stable and six were not. The two stable ones
are still in use, while four of the six unstable ones have been discarded.
As noted earlier, an i ncreasing number of student couples in the U nited S tates
are find ing internships for themselves outside the NIMP. In terms of the theory
proposed here, a reason for this is that the NIMP algorithm is not guaranteed to
prod uce stable matches when couples look for assignments together. Th us, the
recontracting problems that beset the older systems are effectively reappearing. In fact,
no algorithm is guaranteed to find stable assignments in these environments, because
none need exist! This suggests that instability in this market may become endemic.
 49
Economic
Organization and
Efficiency
SUMMARY
There are economi c organizati ons at many levels, from the economy as a whole to
firms to units wi thi n them. Within the theory, the firm is disti nguished from other,
smaller units by its status as a legal enti ty able to enter into bi nding agreements wi th
indi vi duals. This power makes it unnecessary for the indivi duals to enter into a
complex multilateral contract to organize their transactions and consequently makes
it more li kely that efficient arrangements can be negotiated.
The basic unit of analysis in economic organization theory is the transacti on,
where goods or services are transferred from one person to another. An important
focus of the analysis i·s on the behavior of the individuals who transact. The main
tasks of economic organi zation are to coordinate the actions of the various indivi dual
actors so that they form a coherent plan and to motivate the actors to act in accordance
wi th the plan.
We evaluate organizations on the basis of how well they satisfy the wants and
needs of people, that is, on the basi s of their efficiency. Since organizations are partially
designed, one can also explain features of organizati ons as attempts by the organization
desi gners to achieve efficiency. The relative successes of different ki nds of organization
provide some of the main evidence for theories about which kinds of organization are
most efficient in parti cular envi ronments.
Using efficiency as a positive principle requires taki ng care about whose interests
are being served and what kinds of arrangements are feasible. A small group that is
able to bargain among its members may decide on arrangements that are efficient for
themselves but that would be regarded as inefficient if the group could be enlarged.
Efficiency in this sense is used to make predictions only and not to evaluate the social
desi rabi lity of the agreed arrangements. In additi on, arrangements that appear to be
wasteful may still be efficient in the positive sense if there is no feasible alternati ve
for the group that all would prefer.
The expansi on of production i n modern economies has been accomplished in
large measure through specialization , according to which any one i ndividual performs
only a tiny fraction of the ki nds of tasks required to make what he or she uses.
Increased specialization implies that people become more reliant on the work of
others, and the need for coordination increases. The two extreme alternative ways to
coordinate are to communicate informati on to a central planner who makes all the
important decisi ons or to provide indivi duals wi th the information and resources they
need to make decisions that fit in with the overall plan. Both extremes are mere
caricatures. Real economies all use a mix of these two approaches.
Transacti on costs are the costs of negotiating and carrying out transactions. They
include coordination costs, such as the costs of monitoring the environment, planning
and bargaini ng to decide what needs to be done, and motivation costs, such as the
costs of measuring performance, providi ng incentives, and enforcing agreements to
ensure that people follow instructions, honor commitments, and keep agreements.
The way the transactions are best organi zed and managed depends on the basi c
attributes of the transacti on. Five attributes have been identified as especially important.
The first is asset specificity. When parties are called upon to make larger specific
investments, they generally seek to organize in ways that safeguard those investments.
Second, when one party is involved in frequent, sim ilar transactions over a long
du ration , it is likely to pay that party to set up more specialized mechanisms or
procedures to reduce the costs of transacting. Frequent transacting between two or
more people over 3 long horizon allows the parties to develop understandings, reducing
 50
The Problem of the need for explicit agreements, and to grant or withhold favors, reducing the need
Economic for outside enforcement of agreements. Third, uncertain ty about the circumstances
Organization in which a transaction will occur and the complexity of the decisions that will be
required make it hard to forecast exactly what performance will be required. This
undermines the effectiveness of simple contracts and leads parties to contract over
decision rights and procedures rather than over specific aspects of performance.
Fourth, the costliness of measuring performance makes it difficult to provide perfor­
mance incentives, leading the parties to seek organizations where measurement and
incentive issues are of less importance. Fifth, when a transaction is closely connected
to other transactions, that is, where failures of close fit among the transactions are
quite costly compared to failures to make best use of local resources, coordination
mechanisms tend to be strengthened, either by increasing managerial oversight or
by arranging frequent meetings among the people responsible for the individual
transactions.
In its simplest form, transaction costs theory holds that organizations are designed
to minimize the total costs of transacting. The two problems with this simple theory
are that the costs of transacting are not logically distinguishable from other costs, and
that efficiency itself does not always imply total cost minimization. However, there is
one special case where the latter problem, at least, disappears.
When individual preferences are free of wealth effects, that is, when everybody
regards each outcome as being completely equivalent to receiving or paying some
amount of money and when there are no a priori restrictions on monetary transfers,
the efficient allocations are precisely those that maximize the total value and divide
it all among the participants. This conclusion is known as the value maximization
principle. The Coase theorem holds that when there are no wealth effects, all decisions
about productive activities and organizational arrangements are unaffected by the
wealth, assets, or bargaining power of the parties. Only the decision of how benefits
and costs are to be shared is affected by these factors. This view contrasts, for example,
with the Marxian view that organizations reflect underlying power arrangements and
class interests and not the desire to maximize total wealth. When there are no wealth
effects, an efficient organization acts as if it were an individual with a well-defined
objective to maximize total value.
In the general case, the value maximization criterion does not describe how
organizations behave. Organizations then may serve a variety of conflicting individual
interests, rather than maximizing a single overall organizational objective. This is
especially true of public organizations, like universities, with their ever-shifting balance
among different social interests, but it is also true in varying degrees about business
firms, where even the owners may have divergent interests.
While we do not attribute motives to organizations in general, we do attribute
them to people. In the theories treated in this book, people are self-interested and
opportunistic, and successful organizations must channel that self-interest into socially
beneficial behavior.

• BIBLIOGRAPHIC NOTES
As with so many of the central concerns of economics, the problems of economic
organization and organizations find their first treatment of lasting significance in
Adam Smith's The Wealth of Nations. Although organizational issues were not
a major focus of mainstream economists after Smith, important insights come
from the writings of Karl Marx in the nineteenth century and especially from
 51
Frank Knight and John Comm ons in the first quarter of the twe ntieth cen tury. Economic
Comm ons in particular cham pioned the treatment of the transaction as th e Organ ization and
fu ndamental unit for analysi s, while Knight specifically addre ssed the organizati on Efficiency
of firms, and of econom ic activity m ore generally, in efficiency terms.
Ronald Coase is rightly viewed as the originator of transaction cost eco nom ics,
and m uch of what we have reported in this chapter is an outgrowth of his classic
19 37 paper in which he first developed the idea that economizing on transaction
costs would determine the organization of econom ic activity and the division of
activity between firms and markets. His 1960 paper, which develops the Coasc
Theorem, is another classic. I t has done much to make econom ists aware of the
power of using value maxim izati on and efficiency as positive, explanatory
principles. These papers were speci fically cited when Coase was awarded the
Nobel prize in Economics in 1991.
The im portance of dispersed, local inform ation for economic organization was
accentuated by Friedrich H ayek in his contribution to the debate about market
systems versus central planning that followed the establishm ent of the centralized
comm unist system in th e USSR.
Am ong m ore recent contributions, the contractual approach to organizations was
championed by Arm en Alchian and H arold Dem setz in seeking to explain the
role of hierarchy and supervision in the firm in incentive term s. Kenneth Arrow's
i nfl uential little book develops his treatment of organizations as arising when
markets fail. Oliver Wil liamson's writings have played a maj or role in developing
transaction cost economics. His 1985 book gives an excelle nt overview of his
approach, which identi fies asset specificity, frequency, and uncertainty as the key
dimensions of transactions and which also accentuates the lim its of hum an
rationality. This latter them e was first introduced into economics by H erbert
Sim on, and has been developed in an evolutionary direction by Richard Nelson
and Sidney Winter. Yoram Barzel, building on the contributions of S tephen
Cheung, has emphasized the measurement costs dim ension of transaction cost
economics. The connectedness dimension and the notion of design connected­
ness are intr oduced for the first time here.
The sur vey papers by Bengt H olmstrom and Jean Tirole and by William son are
valuable supplements not only for the questions in thi s chapter, but for many of
the issues addressed thr oughout the book. Our paper on bargaining and influence
costs is an integrated presentation and critique of the basics of transaction cost
economi cs.

• REFERENCES
Alchian, A. , and H. Dem setz. "Production, Information Costs, and Econom ic
Organization," American Economic Review, 62 (1972), 777-95.
Arrow, K. J . The Limits of Organization (New York: W. W. Norton, 1974).
Barzel, Y. "Measurement Costs and the Organization of M arkets," fournal of
Law and Economics, 25 (1982), 27-48.
Cheung, S. N. S. "Transaction Costs, Risk Aversion, and the Choice of Contractual
Arrangements," fournal of Law and Economics, 12 (1969), 23-42.
Coase, R. "The Nature of the Firm," Economica, 4 (19 37), 386-405.
Coasc, R. "The Problem of Social Cost," fournal of Law and Economics, 3
(1960), 1-44.
 52
The Problem of Commons, J. R. "Institutional Economics," American Economic Review, 2 1
Economic ( 1 93 1 ), 648-57.
Organization Hayek, F. A. "The Use of Knowledge in Society," American Econom ic Review,
3 5 ( 1 945), 5 1 9-30.
Holmstrom, B. R. , and J. Tirole. "The Theory of the Firm, " Chapter 2 in R.
Schmalensee and R. Willig, eds., Handbook of Industrial Economics (New York:
North-Holland, 1989).
Knight, F. H. Risk, Uncertainty and Profit (London: London School of Econom­
ics, 1 92 1 ).
Marx, K. , Capital (Harmondworth, UK: Penguin Books, 1 976).
Milgrom, P. , and J. Roberts. "Bargaining and Influence Costs and the Organiza­
tion of Economic Activity," in J. Alt and K. Shepsle, eds. , Perspectives on Positive
Political Economy (Cambridge: Cambridge University Press, 1990).
Nelson, R. R. and S. Winter. An Evolu tionary Theory of Economic Change
(Cambridge, MA: Harvard University Press, 1 982).
Simon H. Models of Man (New York: John Wiley & Sons, 19 57).
Smith, A. An Inquiry into the Nature and Causes of the Wealth of Nations
(Oxford: The Clarendon Press, 1976).
Williamson, 0. The Economic Institutions of Capitalism : Firms, Markets,
Relational Contracting (New York: The Free Press, 1 98 5).
Williamson, 0. "Transaction Cost Economics," Chapter 3 in R. Schmalensee
and R. Willig, eds. , Handbook of Industrial Economics (New York: North­
Holland, 1 989).

EXERCISES

Food for Thought

1. One of the main tenets of economic analysis is that people act in their own
narrow interests. Why, then, do people leave tips in restaurants? If a study were to
compare the size of the tips earned by servers in roadside restaurants with those
frequented mostly by locals, what would you expect to find? Why?
2. For most large Japanese firms, a majority of the voting shares are owned
by the company's major lenders (banks and insurance companies), its customers and
suppliers, and related firms with whom it has longstanding relationships. How would
this ownership structure affect the objectives that these fi rms seem to pursue?
3. Would you expect the organization of agriculture in developing countries
to be arranged in a way that maximizes the total wealth of the farmers, workers, and
lenders? If arrangements do not maximize total wealth, what kind of variations would
be most likely? Explain your answer.
4. In California's fruit farms, farm workers who pick fruit are commonly
organized into teams that are paid according to the number of trees that are cleanly
picked. The teams themselves decide how to divide the pay among their members.
What attributes of this transaction account for this arrangement?
5. Cable television companies lay cables to individual households in the
communities they serve to carry the television signal. How specific is this investment?
What kind of arrangements would you expect the cable companies to make with local
communities about the pricing and taxation of cable services?
 53
I Mathematical Exercises I Economic
Organization and
Efficiency
I . S uppose four families share a common stretch of beach and they are
considering a program of improvements, including a stairway and a play structure for
children. The value of spending y in total for the improvements is 5y -½ y2 for famil ies
# I and #2, 7y - ½ y for family #3, and 4y - y for family #4. What is the efficient
2 2

level of expenditure on beach improvements?

2. Continuing with the situation hypothesized in problem I , show that if the
cost of the impr ovements is to be shared equally, then family #4 will be unwilling
to bear its share of the cost. What is the largest improvement that all the families
would agree to if the cost of improvement must be shared equally? Demonstrate that
the ,r esulting expenditure is inefficiently low by finding an alternative level of
expenditure and pattern of cost sharing that the families unanimously prefer.
3. (Mathematical proof of the value maximization principle. ) S uppose that
there are N individuals. Person n's utility when outcome y occurs and he or she
receives cash compensation of xn is given by the utility function xn + vn (y). S uppose
that decision y generates net profits P(y), which may be a positive number if we think
of y as representing an investment or a negative number if we think of y as representing
some public good, such as parks or roads, at a total cost of - P(y). The profits are
divided among the individuals with individual n receiving x n (or paying - xn ). The
payments must add up to the amount available, that is, P(y) = x 1 + . . . + xN .
Pr ove that an allocation (y,x 1 , • • • , xN ) is effi cient if and only if y maximizes the total
value P(y) + v 1 (y) + . . . + vN(y). [Hint: There are two things to prove. First, you
must show that if an allocation does maximize the total value, then there cannot be
an allocation that Pareto dominates it. Second, if the allocation does not maximize
the sum, then you must show that there is another allocation that dominates it. To
show the latter, take any y with a higher total value and show that the x n s can be
chosen so that the gain in total wealth is divided equally among the participants.]
4. (Characterizing a utility function when there are no wealth effects.) Suppose
that a decision maker's preferences are such that for any two decisions y and y', there
is an amount of cash compensation or a cash payment such that y combined with
cash compensation of C(y,y') would be j ust as good, from the decision maker's
perspective, as y' with a zero-cash compensation. Further suppose that this amount
C(y,y') does not depend on the level of other cash payments made to or by the
decision maker. Finally, suppose that the decision maker prefers more money to less.
Fix any possible decision y and define v(y) = C(y, y). S how that with this definition,
the utility function x + v(y) represents the decision maker's preferences; that is, the
decision maker will prefer an allocation (x, y) to another all ocation (x' ,y') if and only
if x + v( y) > x' + v(y'). [Hint: Argue first that the decision maker is indifferent
about having (x, y) or (x + v(y),y). Therefore, for any utility function that represents
the decision maker's preferences, U(x, y) = U(x + v( y),y) and, similarly, U(x' ,y') =
U(x' + v( y' ),y) . ]
 Part

II
COORDINATION : MARKETS AND
MANAGEMENT

3
UsINC PRICES
FOR COORDINATION
AND MOTI VATION

4
CooRD INATI NC PLANS
AND ACTIONS
 3
USING PRICES
FOR COORDINATION
AND MOTIVATION

lwl hich of these systems /central planning or competitive markets] is likely to be

more efficient depends mainly on the question under which of them we can expect
tha t fuller use will be made of the existing knowledge. And this, in turn, depends on
whether we a re more likely to succeed in putting at the disposal of a single central
a uthority all the knowledge which ough t to be used but which is in itially dispersed
a mong many differen t individuals, or in conveying to the individuals such
additional knowledge as they need in order to enable them to fit their plans in with
those of others.
-Friedrich Hayek 1

All economies fa ce th e basic problems of determining what is to be produced, by and

for wh om, using what meth ods and resources. Even in the simplest, most primitive
societies, food must be gathered, hunted or grown, cl othing must be made, shelter
must be provided, and so on. Of course, in primitive societies it is possible for people
to do many or all of these things for themselves. H owever, as civilization advanced
people realized that they can make more of all these produced goods using the same
limited resources if people specialize their activities and if production occurs on a
larger scale th an would be needed to serve any single person. For example, a person
wh o specializes in building shelters and wh o builds many of them can acquire
specialized tools to make the j ob easier that would hardly be worth wh ile if only one
shelter were to be built. This person can also acquire experience and specialized
training to improve his or her proficiency. Such a person can build more shelters,
faster, with less waste of matenals and better-quality construction than can individuals
building shelters only for their own needs, with little or no prior experience.
With specialization comes a keen need to plan and coordinate people's activities
so th at effective use can be made of the limited resources at their disposal . People
wh o build shelters for th emselves may have a good idea what is needed, when, and

1
"The Use of Knowledge in Society," American Economic Review, 3 5 ( 1 945), 5 1 9-30.
 57
where, but people who build shelters for o thers need to be to ld these th ings. If shelters Using Prices for
are to be built by specialists, how many builders will be needed? How many tools of Coordination and
each kind should be prepared? How much lumber? And how many people should be Motivation
co bblers, farmers, and so on? How much land should be planted with crops and how
much used for grazing, or left fallow, or saved for parks? Even in the sim plest
communities, eco nomic life demands a large measure of coordinatio n amo ng people.
In the modern economy, the scope of activities that need coordinatio n is great.
The amo unt of oil that should be pumped from Saudi Arabian oil fields depend s on
the characteristics of the fields, of course, but it also depends on how much gaso line,
heating oil, and other products will be used by drivers, homeowners, and industries
at hundreds of millio ns of separate locatio ns throughout the world. It depends o n how
much oil is pumped in other places from Alaska to the North Sea and from Texas to
Siberia as well. The changing availability of oil affects the need for refineries and
drilling equipment, for cars, gas statio ns, and home insulation, for public transportation
and highways, for electric cars, electrical generatio n plants, and research into synthetic
fuels, amo ng many other needs and activities. The pro blems of and possibilities for
organizing economic activity on a world scale are daunting, yet they are carried out,
every day, with no effective central control by any government or cartel and hardl y
any thought by the participants about their roles in the system as a whole.
If it seems surprising that the global economic system could work at all without
some central coordinatio n, it is surely even more surprising that it functio ns so
smoothly so much of the time. In most Western natio ns, it is co nsidered newsworthy
when drivers cannot simply drive to the local pump to fill their cars' tanks with
gasoline. In co ntrast, in the centrally planned economy of the Soviet U nio n, it is
newsworthy when consumers find adequate supplies of milk and meat in the stores.
Therefore, before we discuss the eco nomics of how managed organizatio ns work in
Chapter 4, we first devote this chapter to the workings of unmanaged markets, in
which decision makers acting without central directio n and with perhaps all too l ittle
social co nsciousness achieve the effective coordinatio n that often escapes more tightly
controlled systems.

PRICES AND COORDINATION

Our principal purpo se in this sectio n is to begin developing the central economic
model that has been used to study how markets can achieve a high degree of
coordination without central planning: the neoclassical market model. This model is
based on a particular conceptio n of eco nomic organization. S pecifically, the eco nomy
consists of co nsumers/resource suppl iers, who se needs and wants the organization tries
to satisfy, and productive units (firms) that purchase resources (including labor services)
from co nsumers, make the products co nsumers desire, and are owned by consumers
(either directly or indirectl y).
The needs of co nsumers can be satisfied in a variety of ways. If wheat for bread
is in short supply this year, oats, rye, or corn can be substituted to produce a similar
product. Clothing can be made from co tton, wool, or various synthetic fibers and, in
many cases, consumers might not care if the garments are delivered this week or next.
Even identical goods and services can be produced in many ways. Homes can be kept
warm using electricity, fuel oil, natural gas, or solar power, or better insulatio n can
substitute for energy use. Goods can often be transported to a destinatio n by truck,
train, ship, barge, or airplane. Similarly, each of the eco nomy' s resources can be put
to many different uses. Corn can be used to make food or fuel. Workers can be
trained to be clerks, carpenters, o r computer pro grammers. The possibil ities are almost
endless.
Given the complexity of the pro blem of allocating reso urces, no single person
 58
Coordination: could possibly determine an efficient allocation. To do so, one would have to first
Markets and comprehend and account for all the possible activities and claims on resources that
Management could and should be considered . Then one would have to identify feasible plans for
all these activities consistent with the resources available and with the technological
opportunities. Finally, from among these, one would have to identify an efficient
plan. Even this last step alone demands too much to be practicable. There are simply
too many facets of the economy-the different resources available, the diverse
individuals with particular skills and needs, the many new technologies, and so on­
that must be understood in order to check whether a particular proposed allocation is
efficient. Even if the decision maker were aided by banks of powerful computers so
that computation were not a problem, there would still be the need to assemble all
the relevant information about production possibilities, resource availability, and
individual tastes for the computers to process. Collecting the necessary data, ensuring
their accuracy, and keeping them continuously up to date would be an impossible
task. To organize production well, local decisions must rely to a large extent on
knowledge of local circumstances. For example, decisions about how to respond when
a worker gets sick or a machine breaks down are usually best handled by those directly
involved.
The economic organization problem, as seen through the neoclassical lens, is
to provide people throughout the economy with the information they need to make
decisions that are coherent (that is, part of an efficient overall plan) and to motivate
them to carry out their parts of the plan. The neoclassical model is a formal
mathematical model consisting of a set of equations and inequalities that represents
the way markets work. It can be used to prove that a system of properly determined
prices can solve the organization problem. Under certain circumstances, prices provide
people with all the additional information about the economy that they need in order
to make efficient use of the available resources. Furthermore, if individuals and firms
act out of pure self-interest, taking prices as given and maximizing their individual
utilities or profits, they will be motivated to undertake exactly those activities that lead
to efficiency.
Our purpose in th is chapter is to study how far a simple price system can go in
solving the coordination problem of economic organization. The following discussion
develops the main idea in two steps. First, we solve the optimal organization problem
for an example (road maintenance) in which there is a single objective (saving lives)
and a single constrained resource (hours of work-crew time). We then turn to the
more general theory, in which there are many consumers, each with his or her own
objectives, and many limited resources, each of which is valuable and needs to be
conserved. In the neoclassical model, the market simultaneously reconciles these
conflicting objectives, directs resources to production, and motivates firms to produce
the right products.

One Objective and a Single Scarce Resource

Suppose that you are in cha rge of the Department of Highway Safety for the
government. Your job is to save lives by directing the resources at your disposal to
projects that reduce the number of fatal highway accidents. You are limited by the
number of hours available from the work crews that carry out the projects.
Reviewing your department's performance last year, you have constructed a
table of some of the projects that you considered. Table 3. 1 might be an excerpt from
a much longer table, but it shows all the projects that were actually carried out. Thus,
a total of 3, 000 crew hours were expended, the maximum number available.
Last year, projects were not selected in the way that maximizes the number of
lives saved. If project 2 had been eliminated and the crew hours freed were used to
 59
Table 3. 1 Life-Saving Projects (Estimates) Using Prices for
Coordination and
Motivation
Project Crew Lives Lives per Project
ID No. Hours Saved 1 ,000 Crew Hours Accepted?

5 800 4 5. 00 No
4 900 3 3. 33 Yes
1 800 2 2. 50 Yes
6 500 1 2 . 00 No
2 1 , 300 2 I . 54 Yes
3 700 1 I . 43 No

carry out projects 5 and 6, an estimated three additional lives could have been saved
(four lives saved by project 5 and one by project 6, instead of the estimated two lives
from project 2). The supply of crew hours has therefore not been used efficiently.
Two problems make the selection operation difficult. First, because the projects
are not all available for review at once, they cannot be compared to select the best
ones . Second, due to the number and complexity of projects to be reviewed, staff
members in several regional offices are involved in making the estimates and acceptance
decisions. At the beginning of the year, you may have a pretty good idea about what
kinds of projects will be available. However, you have no way to know what the
specific projects will be, when they will be proposed, or which offices will evaluate
them. You need to find a way to coordinate decisions made by different offices at
different times during the year to assure that the most productive projects are chosen.
What can you do to help your staff members make well-informed decisions?
DETERMINING THE OPTIMAL PRICE To understand how to solve this problem, let us
start with an even easier one. Suppose you knew in advance which projects would be
proposed during the year. In order to maximize the number of lives saved subject to
the constraint on available labor time, you would want to start by carrying out those
projects that have the highest return-per-unit input, and then move from the highest­
return projects to those with lower returns until the crew hours were exhausted.
Otherwise, shifting labor from a project that saves a low number of lives (per crew
hour) to one that saves a higher number would increase the number of lives saved
without using any more crew time. Thus, you would rank the projects according to
the number of lives saved per thousand crew hours required. This index of life-saving
efficiency is shown for last year's projects in Table 3 . 1 .
If the projects last year had been selected according t o this index, you would
have undertaken projects 5, 4, 1 , and 6, at which point the 3 , 000 crew hours would
have been exhausted. No other combination of projects in this example could save
more lives given the limited availability of crew time. Note that these are precisely
the projects for which the lives saved per thousand crew hours is greater than or equal
to two. Your problem therefore has a very simple solution that involves communicating
remarkably little information: You could simply have directed the different offices to
carry out projects if and only if their indices of lives saved per 1 , 000 crew hours were
at least two. If the offices obeyed instructions, the best allocation would have been
achieved.
This example illustrates a general lesson: No matter what the list of projects,
there is always a number P such that if all projects for which the index of lives saved
per 1 , 000 crew hours is at least P are accepted and all projects with lower indices are
rejected, the selected group of projects is the one that maximizes the estimated number
of lives that can be saved with the available resources. Here the number P is a price,
 60
Coordination: expressed in terms of lives rather than dollars, marks, or yen. The unit in which a
Markets and price is expressed is not important for this analysis. What is important is that all the
Management costs and benefits from a project be expressed in terms of a common unit so that they
can be compared. In our example, I ,000 crew hours has a "price" of two lives. This
means that it pays to undertake (only) those projects for which the benefit in terms of
lives saved exceeds the cost of the crew hours, when the crew hours are valued in this

By calculating P and instructing your staff members to base their decisions on

way.

that value, the result will be coordinated project selection decisions both among offices
and over the course of the year. There is no need for more detailed communication

problems. The price P provides each staff member with all the information he or she
or coordination among your staff, nor does the timing of the proposals create any

Of course, in actuality you do not know in advance what the value of P should
needs in order to make the decisions that best accomplish the objectives.

be. It will depend on the projects that actually become available during the year. If

have been accepted. If you set it too low, you will accept too many projects early in
you set the value too high, you will wind up rejecting some projects that ought to

the year and run out of resources for better projects later in the year.
One possible solution is to estimate P on the basis of last year's experience. This

If the distribution of projects varies considerably from year to year, however, an

may often work quite well; historically-based systems are commonly used in budgeting.

estimate of P based on the previous year's experience is likely to be inaccurate and to

lead to an inefficient selection of projects . Still, it is important to recognize that any
system for selecting projects will sometimes yield the wrong answers if decisions must
be made before all the alternatives are known. A great advantage of the price system
is that it eliminates all the other sources of waste that might otherwise be present even
if all the project proposals were known in advance. More specifically, it eliminates
the lack of consistency of decisions over time for a single decision maker and the lack
of coordination among the several staff decision makers.

A Market-Clearing Interp retation

The method of determining the optimal price can be viewed as resulting from a
market. This is an important idea because it gives us our first hint about how
decentralized markets lead to the efficient use of resources.
Suppose that the project evaluators in the previous example were able to bid for
the crew hour resources in an ordinary market, each with the objective of maximizing
the number of the lives he or she saves minus the cost of the crew hours used. Their
demands are plotted in Figure 3. 1 . Quantities are placed on the horizontal axis,
whereas the supply of crew hours is assumed to be fixed and is represented by the
vertical line at 3, 000 crew hours. To determine how quantities depend on prices, find
the relevant price on the vertical axis and look across to find the corresponding
quantity.
Looking back at Table 3. 1 we see that at any price higher than 5 . 00 (lives per
thousand crew hours), none of the projects would be "profitable. " That is, the cost of
the crew hours used in terms of lives would exceed the benefit, even for the most
cost-effective projects. Consequently, when the price exceeds 5 . 00, there is a zero
demand for crew hours. At any price below 5. 00 but above 3 . 3 3, project 5 would be
the only profitable project, so the 800 crew hours required by project 5 are recorded
in Figure 3 . I as the demand for prices in that range. At prices below 3. 3 3 but above
2. 50, projects 5 and 4 are both profitable. They require a total of I , 700 crew hours,
which is therefore the demand corresponding to prices between 3. 3 3 and 2. 50 in
 61
Demand for Supply of
� 5 Crew Hours / Crew Hours
Using Prices for
:::, Coordination and
0
Motivation
:::c
:r: 4
u
3
§
,-
Q) 2
C.
1/J
g! 1
:::i
Figure 3. 1 : The projects that are profitable at a
price like 2 where supply equals demand are just
.8 1.7 2.5 3.0 4.3 5.0 the ones that ought to be undertaken to maxi­
Crew Hours (Thousands) mize the number of lives saved.

Figure 3 . 1 . As the price continues to fall, the demands of the various projects are
added in the order l isted in the table.
As the figure indicates, two (lives per thousand crew hours) is a market-clearing
price. Simply, at this price the quantity of crew hours demanded equals the quantity
available. At higher prices the demand falls short of the available supply, whereas at
sufficiently lower prices there is a shortage of crew hours. At the market-clearing
point, each evaluator finds it most "profitable" in terms of the specified objectives to
undertake precisely those projects that are part of the optimal life-saving plan.

Extensions and Difficulties

In the preceding example, prices enable the different regional offices to work
independently and yet make coherent, well-coordinated decisions. The example has
many special features, however. First, there is a single scarce resource-crew hours­
that limited the department's ability to save lives. Second, there is a nice match
between the size of the available labor supply and the amount needed to complete
the most efficient projects: Carrying out the best projects uses up exactly the available
3,000 hours. Third, the staff shares a common objective so there is no concern that
some office might pursue idiosyncratic or personal goals.
Relaxing the first of these special assumptions is relatively easy. Suppose that,
in place of the single constraint on the number of crew hours available, the department
were limited in both the budget for materials and the number of hours available from
work crews. In that case, we would conclude that two prices-one for materials and
the other for work crew hours, both expressed in terms of lives-can be used to
characterize the optimal plan.
As we mentioned earlier, an important characteristic of prices is that they can
be expressed in virtually any units, provided that all costs and benefits are expressed
in the same units. In the actual economy, prices are most commonly expressed in
money terms. In this case, the plan that saves the greatest number of lives can therefore
be characterized as the one that selects just those projects for which the dollar cost
per life saved is less than the "dollar value" of a life.
The second special feature of the match between the size of the labor supply
and the amount needed can also be removed at this point. Referring to Table 3 . 1 ,
suppose that project 6, instead of requiring 500 crew hours to save one l ife, required
1 , 000 hours and saved two lives. The index of lives saved per 1 , 000 crew hours
remains two . There is no longer enough labor to carry out this project, however,
given that all higher-valued projects are still to be completed . If projects can be scaled
down with both the resources used and the number of lives saved scaled down
 62
Coordination: proportionately, then an augmented version of the price system still works. You would
Markets and want to undertake project 6 at a 50 percent level, using 500 hours and saving one
Management life. With knowledge of the price only, however, the office considering this project
would not know that it should undertake only part of it. Thus, for marginal projects,
additional communication is necessary to achieve coordination.
If the crew hours involved in individual projects are few relative to the aggregate
supply of crew time, this inadequacy of prices may be only a small problem; most
decisions can still be price guided, and only those that are close calls will require
additional communication. If the projects are both large and indivisible, however, in
that each must be completed in full or not at all, then a price system may be unable
to identify the optimal solution. For example, the optimum may entail undertaking
a low-valued project that fits within the available labor constraint while forgoing a
higher-valued project. Communication of prices alone will then fail to allow decision
makers to identify the correct selection of projects.
It is also possible to dispense with the third special feature of the single, shared
objective in the context of this example. However, we instead turn to a more general
treatment in which the multiplicity of individual objectives is a central feature.

THE FUNDAMENTAL THEOREM

OF WELFARE ECONOMICS
We now turn to an analysis of markets in an economy that has many consumers and
producers with varying individual objectives and many goods and services. Remarkably,
a price system can sometimes solve the coordination problem in this context as well.
Briefly, we conclude that if ( 1 ) each productive unit knows the prices and its own
individual production technology and maximizes its own profits at the prevailing
prices, (2) each consumer knows the prices and his or her own individual preferences
and then maximizes utility given the prevailing prices and his or her income, and (3)
the prices are such that supply equals demand for each good, then the allocation of
goods that results is efficient: There is no other allocation consistent with the available
resources and technological opportunities that the consumers would unanimously
prefer.
There are two remarkable aspects to this conclusion. First, knowing only local
information and the system-wide prices is enough for each producer and consumer
to make the choices required for coherence and efficiency. There is no need for
central planning or extensive sharing of information; coordination is automatically
achieved. Second, each consumer and each firm is asked only to pursue its own
interests fully and diligently; firms maximize profits and consumers maximize personal
utility. Nobody is asked to use information in any way contrary to h is or her narrow
self-interests, and yet the whole system of behavior is coherent so that the resulting
allocations of goods and services are actually efficient. The precise statement of this
much celebrated conclusion has come to be known as the fundamental theorem of
welfare economics. This theorem has a long, rich history, which dates back to its
intuitive formulation in the work of Adam Smith, and includes a series of mathematical
formulations developed in the nineteenth and twentieth centuries. The following
formulation is a variant of a model created originally by Kenneth Arrow and Gerard
Debreu.

The Neoclassical Model of a Private Ownership Econom y

Tm: hDl\'ID l l \L CoNSl !\IER Our analysis begins with a single consumer, one of
1

many who participate in the economy. This consumer can supply his or her own
labor to employers for wages. The co ns umer may also o wn stocks of consumer goods, Using Prices for
land, commodities, or industrial eq uipmen t. Le t E be a lis t of numbers denoting the Coordi nation and
amoun ts of the vario us goods, labor, land, commodities, and so on that the consumer Motivation
owns. This list E is called the consumer's resource endowment.
The concept of a list (or vector) is quite important in our formal development.
Each list specifies an amount for every conceivable good or service in the economy,
that is, an ythin g that any co nsumer or firm might ever want to buy or sell. Let G
denote the number of different goods an d services in the eco nomy. A typical
en dowmen t list E = (E i , E 2 , . • . , Ee) has components specifying the amounts of
farm land, wheat, labor hours, shoes, and so o n that the consumer owns-one
component for each of the G goods. For most consumers, many of the elements of
the list are zero. The average consumer does not own rolls of newsprint, electrical
gem�rating equipment, or magnetic resonance imaging machines.
Lists can be added together. For example, if E is your e ndowment and E' is
your nei ghbor's endowment, then E + E ' = (E 1 + E ; , E 2 + E�, . . . , Ee + E�) is a
new list that specifies how much of each good the two of you have toge ther.
What do co nsumers do with their endowments? Basically, they can sell some
amo unts of the different goods in the market and keep some to consume. Let S be a
list of the quantities that the consumer sells of each good or service traded in the

more of any good than he or she owns. If the consumer sells only labor and uses the
market. Of course, S 1 $ E i , S 2 $ E 2 , and so on. That is, the co nsumer cannot sell

income to buy other goods, then only the labor entry in the list S will be different
from zero. Still, we take S to be a complete list, with an entry for each item that the
consumer might possibly buy or sell.
In addition to the list S, there is a list B indicating the quantities that the
consumer buys. This list has an e ntry for each item that the cons umer might possibly
buy or sell. The entry corresponding to the consumer's own labor will be zero (the
consumer does not buy his or her own labor), and there may be zeroes for many
other goods as well.
Of great importance to the consumer is the list P of prices at which the various
goods trade. Again, there is an entry for each good, each kind of labor-everything
that is traded in the market. We use the notation PB to denote the cost of buying the
types and amo unts of goods in the list B at the prices listed in P. We can compute
PB by multiplying the price for each good by the quantity of that good purchased to
get the amo unt spent o n the individual good, and then adding these amo unts up over
all the goods purchased. The summation notation for this eq uation is PB =
L�= 1 P;B i . S imilarly, PS = L�= 1 P;S; denotes the income the co nsumer earns from
selling the quantities of various items listed in S.

economy. Let F; be the fractio n of firm j that the consumer owns. If firm ; earns a
Each cons umer may also o wn shares of the various business firms in the

pro fit and pays a dividend D;, then the consumer will find his or her income

receives in the form of dividends. If there are / firms in all, then FD = L�= 1 F;D;-
augmented by the amount F;D;- Let FD denote the total income the consumer

INDI\"IDUAL CONSUMPTION PLANS A consumption plan for a consumer is a pair of

lists B and S indicating what the consumer plans to buy and sell. The plan is affordable
at prices P if PB $ PS + FD, that is, if the consumer generates e nough income from
dividends and from selling labor and other goods to pay for the goods he or she plans
to buy. The co nsumer might like to live in a mansion on M alibu Beach and drive a
red Porsche, but these purchases may be unaffordable.
In the neoclassical model, the u tility or satisfaction a consumer gets from
 64
Coordination: Table 3 . 2 A Description of Consumer Behavior
Markets and
Management
Consumer # 1
Goods Units Price Endowment Sales Purchases Consumption
Labor Hours 15 2 , 600 2, 000 0 600
Bread Loaves 1 0 0 1 00 1 00
Autos (new) Number 8, 000 0 0 1 1
Autos ( used) Number 4, 000 1 1 0 0

carrying out a plan depends only on the list of goods C that he or she con sumes. This
utility is den oted by U(C) . 2 Implicit in this n otation is the assumption that con sumers
care only about their own consumption and n ot about the firms' methods of production
or the consumption e'n j oyed by others. The model also assumes that the con sumer
chooses the plan that leads to the highest level of utility among those affordable at
the given prices.
Table 3 . 2 shows the kind of data we use to describe the activities of the consumers
in the econ omy. The first three columns in the table represen t factors that are common
to all agen ts: the goods in the economy, their units of measurement, and the prices.
The next four columns represen t one of the collections of lists we have described.
This one is for con sumer 1. The reader should imagine that the table contin ues with
more rows below the last one, showing additional goods that can be bought an d sold,
an d more column s, showing the data for additional consumers.
According to the table, consumer 1 is endowed with 2,600 hours of labor for
the year, perhaps con sisting of 50 hours per week for 52 weeks. Accordin g to the plan,
the consumer wi ll sell 2,000 hours of labor to generate income and will consume 600
hours in the form of leisure. That is, for 600 hours during which the con sumer might
possibly have been working, he or she will in stead attend a concert or go on a family
picn ic, or do any number of things. Because the price of labor hours is 15, this sale
of labor will generate an income of 30,000 (2, 000 x 15 ) that consumer I can spend
on other goods. According to the table, the con sumer will also sell a used car,
generating proceeds of 4, 000 that can be spen t on other things. The table also shows
how the income will be spent. A new car will consume 8, 000 units of the income,
1 00 loaves of bread will consume an other I 00, and so on.
As the table illustrates, the person's con sumption of each good can be computed
by startin g with the endowmen t, addin g the quantity of the good that the consumer
buys, and then subtracting whatever he or she sells. That is, con sumption is equal to
C = E + B - S. The consumer normally will be a buyer of some goods an d a
seller of others. In either case, consumption of each good is determined by the formula
just described.
The fundamental theorem makes one additional assumption about consumer
tastes called local nonsatiation: S tartin g from any consumption bundle or list, there
is always some good or service of which the consumer would like to have a little bit
more. One consequence of local n onsatiation is that utility-maximizing con sumers will
always spend all of their income on something. This fact is expressed mathematically as
follows:
PB = PS + FD (3 . 1)

2 Note that we do not assume here that there are no wealth effects (Chapter 2) . The theorem does
not require this assumption.
 65
The left-hand side of the eq ua tion is the aggregate amoun t the consumer spends on Using Prices for
purchases of all the different goods, a nd the right- hand side is the total income from Coord i nation and
sales and dividends. Motiva tion

THE FIRM We turn next to the description of firms, or producers, in the economy.
Each firm has some set of possible activities to which it can devote resources. The
firm may be a ble to produce many different products, each in several different wa ys
using a variety of inputs. What the firm does is summarized by what we call a
production plan. S uch a plan consists of a list O of outputs produced and another list
I of inputs used. The plan is techn ically feasible if it is possible to produce the outputs
0 from the inputs I using the technology availa ble to the firm. The set of technically
feasible plans for the firm is denoted by T. The fact that a plan (I,O) is feasible is
represented by the mathematical statement, (I, O) E T, which we read as "the plan
(I,O) is in T, the technologically feasible set for thi s firm. "
Comm only, firms and consumers will be on opposite sides of a ny market
transaction. For example, consumers typically sell labor and buy bread, whereas a
bread-making firm will buy la bor (and other inputs) a nd use it to make bread to sell.
Thus, inputs for a firm often correspond to sales by consumers a nd outputs often
correspond to purchases. In other cases, purcha ses and sales occur between firms,
with one's output being another's input.
The neoclassical model assumes that firms are motivated by profit alone. Given
the prices P, the firm's total income or revenue from the sale of its outputs is PO.
The cost of its inputs, including the wages paid to workers and all other expenses, is
PI. Therefore, the firm chooses the technically feasible production plan that ma ximizes
its profit PO - PI. The dividend D paid by the firm is assumed to be equal to its
profit PO - PI.

ECONOMIES AND ALLOCATIONS N ow, suppose there are many consumers in our
idealized economy, each with his or her own tastes for consumption. In addition,
each consumer may have a different endowment. For example, one person ma y be
endowed with the labor of a computer programmer and a nother with the labor of an
a utomobile mechanic. One may have inherited wealth, whereas a nother is a pauper
with only his or her own la bor to sell. A wide range of possibilities exists. In our
notation, each individual consumer is identified by a social security number. For
example, if n is the social security number identifying one particular consumer, then
E n is the consumer's endowment, S n is his or her list of sales, and B n is the list of
things the consumer purchases. S imilarly, there ma y be many firms with differing
technologies and ownership. Firms are represented by their employer ID numbers.
Firm j's plan is represented by the lists of outputs Qi a nd inputs Ii it will use, which
must lie in its own technically feasible set, Ti.
Formally, a private ownership economy in our model consists of the following
elements. There is a set of consumers N, together with a utility function LJ n a nd an
endowment E n for each consumer n . There is also a set of firms f, together with a
technically feasible set Ti for each firm j, and owne rship shares p i for each consumer
n in each firm j. An alloca tion for the economy is a consumption plan for each
consumer and a production plan for each firm that together are feasible. A feasible
set of plans is one with the following three properties:

1 . Each firm is able to make the prescribed outputs from the prescribed inputs.
That is, (Ii,Oi) E Ti.
2. Each consumer has a vailable the goods he or she is asked to deliver. That is,
5 n ::;; E n .
 66
Coordination: 3. The total amounts of each good planned for delivery to consumers and firms
Markets and does not exceed what is available in the economy. That is, LB n + LI; s
Management LS" + LO;.

An allocation is efficient (or Pareto optimal) if there is no other allocation that Pareto
dominates it, that is, no other allocation that all consumers view as at least as good
as the given one and some consumer strictly prefers.

PRICE FORMATION The model we are describing does not include, within its formal
structure, any description of the mechanism by which prices are set and adjusted with
changing conditions. Instead, it simply postulates that there are prices for each good
and service that are publicly known and at which everyone believes they can transact.
In fact, there is no one single mechanism by which prices in actual economies
are set. There is inste�d some combination of mechanisms, including stores setting
posted prices for the goods they offer, commodities being sold at auction, negotiators
setting wages, and so on. Which of these processes is used will vary across situations,
but the results of all of them should be responsive to economic conditions.
When supplies of a good are less than the amounts that buyers want to purchase,
for example, stores find it easier to sell their inventories of the good without special
discounts and sales. If bidders are worried that there may be no more units for sale
tomorrow, they place higher bids today. If labor negotiators feel secure that high
wages will not lead to layoffs or temporary shutdowns of production, they will press
for better pay . The price paid by buyers in all these circumstances tends to rise. As
prices rise, buyers become less eager to buy and sellers rush to provide more goods,
helping to close the gap between the quantities the sellers offer for sale and the buyers
bid to purchase.
When there are excess supplies of some good, the opposite process takes place.
Stores with unsold inventories put the goods on sale, bidders may find themselves
bidding unopposed, and labor negotiators may worry about possible layoffs. All these
factors lead to a reduction in the prices paid by buyers.
As long as there is any gap between what sellers want to supply and what buyers
demand at the prevailing prices, there is pressure for prices to change to reduce the
gap. Prices are constantly adjusting in response to these pressures, always aiming for
a balance between supply and demand. Nevertheless, to assess the potential of the
price system, we make the same kind of simplification we used to study the problem
of saving lives. In the life-saving example, we supposed that the prices are known and
set so as to balance the supply of crew hours with the demand for them, and we
showed that the resulting decisions were optimal. Here, we suppose that the publicly­
known prices are set exactly as necessary to balance supply (the quantities offered for
sale) with demand (the quantities that buyers wish to purchase). This point, at which
prices have no pressure to change, is called a competitive equilibrium.

COI\IPETITI\'E EQl 1 ILIBRIUI\I A competitive equilibrium for the economy in this model
consists of a price list P that contains a nonnegative price for each good, a consumption
plan B" and S 11 for each consumer n, and a production plan (I i,Q i) for each firm ;.
Furthermore, these lists must satisfy three conditions. First, each consumer's consump­
tion plan must maximize his or her utility. That is, it must give the consumer at least
as much satisfaction as does any other affordable plan. Second, each firm must
maximize its profits. That is, each firm's production plan m ust generate profits that
arc at least as high as those for any other techn ically feasible plan. And third, at the
given prices, the quantity demanded of each marketed item must be equal to the
 67
quantity offered for sale. Again using the sym bol � to represent a summation, the Using Prices for
statement "supply equals dem and" is expressed as: Coordination and
Motivation
:z:s n + L· Qi = :Z:B" + :Z:Ji. (3 . 2)
n n i

The left-hand side of Equation 3 . 2 is a supply list consisting of the total sal es
by c onsumers plus the outputs by firm s of the various goods. For example, the q uantity
of oats supplied in the left-hand list is the sum of the quantities sold by individuals
plus the quantities produced by farms. The right-hand side of the equation is the
demand list eonsisting of the total purehases of the various goods. For example, the
quantity of oats purchased in the list is the total of the quantities purchased by
individuals, by farmers for horse feed, by bakers for bread, by breakfast-cereal
manufacturers, and so on. Each purchaser is classi fied as either a consumer, who is
maximizing the utility of consum ption, or a producer, who is maximizing profits.

THE THEOREM N ow that we have identified the basic elements involved in the
fu ndamental theorem of welfare econom ies, we can form ally state the theorem as
follows:

The Fundamental Theorem of Welfare Eeonomies: If (P,B,S,I,O) are

the price lists and plans of a com petitive equilibrium, then the resulting
allocation is effieient.

A logically equivalent way to state the theorem is as follows. If (B,S,I,O ) is a

set of c ompetitive equilibrium plans and (B ' ,S' ,I' ,O') is any other set of plans that
all consumers like at least as well and that at least one c onsumer strictly prefers, then
(B' , S ' ,I' ,O') is not feasible.
N otice that it is only allocations, not priees, that we describe as effieient. Prices
are the key to one system that guides people's plans and aetions. I t is the plans and
actions alone that have physical, psychological, and ethical significance; the prices
are therefore j udged solely in terms of the plans and ac tions that result. Further, the
allocation's efficiency is j udged solely in term s of the c onsumers' preferences; pr ofits
per se do not count. This is in keeping with the view that organizations are in
themselves without signifieance, but rather are created entities that exist to serve
human needs.
J ust as in our highway safety exam ple, prices in the neoclassical m odel serve to
inform the parties about what they should do. Consumers and producers do not need
to know why prices have changed to determine how to respond to changing
circumstances efficiently. For exam ple, if fuel oil is in short supply or if a valuable
new use for copper makes i t desirable to ec onomize its old uses, the prices for these
goods will increase and the quantities demanded will adjust to acc omm odate the m ore
limited supply.
U nlike our highway safety example, however, prices in the neoclassical m odel
do m ore than j ust inform people; they also motivate them . Given the prices in a
com petitive equilibrium, the plan ealls for each consumer to buy only what he or
she thinks best, and it ealls on the firm to produce only what is m ost profitable.
Despite the variety of objectives, behavior is coherent enough so that no resources are
wasted. This is the real meaning of the welfare theorem.
Finally, given prices, there is no conflict am ong the owners of any firm about
what it should do. The only effect the firm's ac tivities have on the owners is to
contribute to their inc ome, and all owners prefer more inc ome to less; therefore, all
will favor profit maximization.
 68
Coordi nation: Scope of the Neoclassical Model
Markets and
Management The mathematical model that we have just described is quite general and allows for
a large number of interpretations. The welfare the orem is true for all of them. For
example, it might appear that the model allows no role for time, but this is not the
case. Perhaps people will not want to consume all that they own today but will want
instead to save for tomorrow. Likewise, a firm's input decisions may involve capital
equipment whose usefu l life extends over several months or years. The neoclassical
model accommodates both of these possibilities quite nicely. The goods in the model
can be identified so that "oranges-today" and " oranges- next-month" are different items,
and the consumer may evaluate them differently. There may then be a storage firm
that can use oranges- today and warehouses-today as inputs in order to produce oranges­
next-month and warehouses-next-month as outputs.
This particular interpretation permits an extension of the model in which
consumers can acquire skills and change their future labor endowme nts. For example,
individuals might have direct access to a technology which allows them to combine
their endowme nt of raw labor time today with instruction to yield an endowment of
skilled computer programmer time tomorrow.
As another example, it might appear that because " oranges-today" could be one
of the goods in the model, we must assume that the same price prevails for oranges­
today in Orlando and Montreal. H owever, because the model is abstract, we could
regard oranges in Orlando and M ontreal as different goods. There may be a trucking
firm that can take oranges-today-i n-Orlando and trucks-today-in-Orlando as inputs
and produce oranges-tomorrow-in-M ontreal and trucks-tomorrow-in-M ontreal as
outputs.
The model can also allow for uncertainty. The method here is to consider the
various possible fu ture realizati ons of the uncertainty in the world as defining different
distinc t "states of the world, " and then to again reinterpret the notion of a good so
that, for example, "umbrellas-whe n-it-is-raining" are distinguished from "umbrellas­
when-it-is-sunny." U sing this interpretation allows for uncertainty in production and
for risky research and development (R&D) activities. For example, certain amounts
of input or R&D investment today may yield high level s of output tomorrow if a
fortunate state of the world should occur, and low output if the firm's productivi ty
turns out to be low or the R&D effort is unsuccessfu l.
Finall y, it might appear that the assumption that firms maximize profits precludes
applying the model to situations in which managers are limited i n the amount of
information they can (realistical ly) process. This, too, is incorrect: Such situations
can be accommodated by carefu lly limiting the set of feasible plans. For example, if
a sin gle manager is unable to compute the best way to allocate the labor and equipment
at his or her disposal, then the plan that uses one manager and labor and equipment
i n that way is simply regarded as infeasible. If a better use for the same labor and
equipme nt can be found whe n two managers work together, then the plan that uses
two managers and the superior allocati on of labor and equipment is regarded as
feasible. Because the limitations on the feasible activities of firms are stated abstractly,
any limits on what a firm can do that can be expressed as technical limits or as limi ts
on available resources can be accommodated by the neoclassical model. The
fundamental the orem of welfare economics appl ies for all such limits.
The ne oclassical model omits any explicit treatment of the man y important
dimensions of tran saction s described in Chapter 2. Nevertheless, the model provides
a useful poin t of departure by itemizing i n such an encompassing way so many of the
th in gs that have mistaken ly been regarded as obstacles to the successful operation of
 69
a price system. In purely in tellectual terms, the neoclassical model is such a m aj or Using Prices for
accomplishment in a tradition of thi nking extending back to Adam S mith that it is Coordination and
well worth studying and understanding. Motivation

Proof of the Fundamental Theorem

of Welfare Economics

there is no other set of consumption and production plans (B' ,S' , I' , O ') that is
In order to prove the theorem, we must show that in a competitive equilibrium

feasible, that makes no consumer worse off, and that is strictly preferred to the
original plans by at least one consumer. This is accomplished by first developing
the mathematical properties that such an alternative plan must have. We will
then deduce tha t these properties contradict the premise that the original plans
constitute a competitive equilibrium. Thus, we conclude that if the allocation
is not effi cient, it did not arise from a competitive equilibrium. As we noted,
this is equivalent to the original statement of the theorem.

under (B ' ,S' ,I' ,O' ) to the competitive equilibrium consumption plan. B ecause
Suppose the consumer with index i strictly prefers his or her consumption plan

the competitive plan is (by definition) the consumer's most preferred plan among
those that are affordable, that consumer must have found his or her part of the
alternative plan unaffordable at the competitive equilibrium prices P. We write
this conclusion formally as follows: The cost of the purchases under the
alternative plan for this consumer exceeds the income from sales under the
plan. Mathematically,

(3 . 3)
Every other consumer n is no worse off under the new plan. Then it must be
that the net cost of the purchases minus the sales for each consumer n under
the alternative plan must be no less than under the original plan, that is, no
less than the dividends received:
PB "' - PS"' 2: DF" ( 3 . 4)
Otherwise, according to our assumption that consumers are locally nonsatiated,
the consumer could have done better than at the original plan B", S" by instead

- PB"' on something he or she likes.

beginning with the plan B" ' and S" ' and then spending the excess PS"' + DP

Let us now add up incomes and expenditures over all the consumers. In view
of Expressions 3 . 3 and 3 .4, we find that:

2-PB"' > 2-(PS" ' + DP) (3. 5 )

n n

That is, using the dividends and price list of the given competitive equili brium,
the total expendi tures by consumers under the new plan strictly exceeds their
total income.
A similar analysis applies to firms. N o firm ; can be earning a higher profit
under the new plan because a competitive plan is the one that maximizes profits
at the prices P. That is,
pQi - pJ i 2: pQ i ' - pJi ' (3. 6)
 70
Coordination: Therefore, the total profits of all firms combined must have been at least as
Markets and high under the old plan as under the proposed ( feasible) alternative:
Management
L(PQi - P/ i) 2?:. L (PO i ' - PJ i') (3. 7)
I I

We have assumed that the firms pay out their profits to their shareholders in
the form of dividends. The total dividends received by consumers under the
original plan must therefore be equal to the total profits earned by all the firms:

LDF" = L (PQi - p/i) (3. 8 )

n I

( Formally, Equation 3 . 8 can be derived by summing Equation 3 . 1 over all

consumers n and then substituting the result into the market-clearing Equation
3. 2. )
Because the alternative plan 1s required to be feasible, it must produce enough
net output to meet all the planned uses for all of the goods. This requirement
is represented by:

LB "' + LJi' s LS"' + LOi' (3. 9)

n i n /

Observe that, unlike the others, Expression 3 . 9 is an inequality among lists.

This means that there is enough net output of each good to meet the planned
uses for that good.
To complete the proof, we must show that these inequalities cannot simultane­
ously be satisfied. Doing so will prove that there cannot be any feasible alternative
plan that is at least as good for each consumer and better for at least one
consumer. The proof is accomplished by stringing together the equations and
inequalities ( 3 . 7), (3. 8), ( 3 . 5), and ( 3 . 9) in that order. We first write the proof
as expression ( 3. 1 0) and then explain it. Notice the crucial role played by prices
in this argument: All the following inequalities are stated in money terms.
L (POi' - PJ i') s L (PQi - p/ i) ( a) by ( 7 )
I I

= LDF"
n
( b ) by (8)

< LPB " ' - LPS"' (c) b y ( 5 ) ( 3 . IO)

n n

- p[ ;s·· - ;s·· ] (d) by the distributive law

:'5 P [ f O i ' - f ii' ] (e) by ( 9) because P is non negative

The inequality (a) says that total profits under the alternative set of production
plans, evaluated using the given competitive equilibrium prices, will be no
higher than those under the profit-maximizing plans. According to (b), the latter
are equal to the total div idends paid to the firms' owners. Because consumers
prefer their consumption under the alternative plan, the new consumption plans
must be unaffordable at the competitive equilibrium prices, which is the meaning
of inequality (c). In (d), we use the distributive law to regroup terms, expressing
net consumer expenditures under the alternative plan as the cost of net consumer
purchases under that plan, all using the original prices. For the alternative to
be feasible, however, net consumer purchases cannot exceed the sum of every
firm's net production of the various goods. Therefore, the value of total net
 71
consum ption at the competitive equil ibrium prices (or indeed at an y nonnegative Using Prices for
prices) cannot exceed th e value of net production; this is asserted by inequal ity Coordination and
(e). H owever, that total value is j ust the profi ts under the alternative plan with Motivation
which the argument began.
This string of equations and inequalities implies the nonsensical proposition
that the profi ts of the firms under the new plan are stric tly less than the profi ts
of the firm under the new plan. Beginning with the hypothesis that the
c ompetitive equilibrium plan is not efficient, we deduced a patently false
conclusion. Thus, th e original hypothesis must be false: The c ompetitive
equilibrium plan is efficient.

INCENTIVES AND I NFORMATION

TRANSFER UNDER MARKET INSTITUTIONS
As we noted previously, the price system not only directs resources to efficient use, it
also has other desirable properties. Given the prices, consumers are asked only to do
what they perceive to be in their best interests and firms are asked only to do what is
best for their owners. The market harnesses these selfi sh motivations and directs them
to a socially efficient outc ome. Thus, if the firms and c onsumers do take prices as
given, the price system does not just provide a means to address th e coordination
problem. I t also provides the proper motivation, and so achieves a fairly complete
solution to the overall ec onomic problem. M oreover, it does so while putting low
demands on the amount of information transmission that must occ ur. We examine
both these properties in a preliminary manner here.

Incentives i n Markets
The neoclassical model assumes that producers and c onsumers take prices as given to
them. If they do, then the coordination problem is solved. But will they fi nd it
individually optimal to take prices as given? This question cannot be addressed within
the model itself because it does not contain any discussion of how prices are formed
and th us of what opportunities there might be for influencing them.
If, in fact, th e prices are set by the market participants themselves, it might be
expected that they would reflect any market power that the participants have. I n that
case, the efficiency result may not hold because the exercise of monopoly and
monopsony power deflects prices from th eir c ompetitive levels. M oreover, even if the
prices are not set directly by market participants but rath er by some mechanism
(perhaps an auctioneer or planner) that gath ers information from the participants and
then announces the appropriate prices, any individual participants whose information
affects the price will generally have incentives to distort the information they provide
so as to influence prices to their benefi t.
A vast amount of recent research in economics, both theoretical and experimen­
tal, has focused on the behavior that is induced by the incentives that arise under
various market institutions. Related studies try to determine when th is behavior wil l
lead (approximately) to the efficient outcomes identified by the neoclassical model.
The theoretical work employs many different approaches, but the c ommon conclusion
is that in most ec onomies with a sufficiently large number of participants, competition
between agents will eliminate monopoly power and result in essentially competitive
prices and outcomes. Furthermore, the experimental work indicates th at the number
of participants necessary to make a market tolerably competitive need not be
unrealistically large.
M oreover, even if market incentives are not able to induce fully efficient results,
market institutions when c ombined with private property can still be powerful engines
 72
Coordination: for directing individual self-interest to produce widespread ec onomic advance and
Markets and welfare gains. As Adam Smith noted long ago, individuals pursuing onl y their own
Management selfish aims are led, as if by an "invisible hand," to promote the general welfare. The
comparisons of Easter n and Western Europe in the postwar period, the experience of
the newly ind ustrialized nations of Asia Pacific, and agricultural reform in China are
all recent evidence of the validity of this insight.

Informational Efficiency of Markets

Friedrich H ayek' s quotation that began this chapter contrasts the informational features
of a system of markets with those of a system of central planning. The main difference
is that the market system does not require transmission of detailed information about
resource availability, consumer preferences, or technological opportunities. In contrast,
a centrally planned system would seem to need such information transmission in
order to compute an efficient allocation (or even a feasible one). In a market system,
decisions about the use of resources are left to the individual consumers and firms
with whom the local knowledge of preferences, endowments, an d production
possibilities resides. Only the relatively small amount of information represented by
prices and by offers to buy and sell is transmitted. Indeed, a general theoretical
proposition (which we examine in detail in the next chapter) shows that the competitive
market system involves the minimal information transmission consistent with deter­
mining an efficient allocation of resources.
In actual market systems, more information is transmitted than simply prices.
Firms seek to discover consumer preferences and plans so that they can tailor their
product designs and forecast the demands that will be put on their pr oduction facilities.
They also seek forecasts of macroec onomic conditions, and they attempt to disc over
their competitors' investment, production, and marketing plans. At the same time,
they advertise the prices, characteristics, and availability of their products and that
they are interested in hiring people with particular skills. They do this because
consumers and potential employees want this information. As well, even in a market
ec onomy there are always government regulatory pr ocedures that usually work through
detailed rules and directives rather than thr ough price signals. S till, prices provide
much of the information that is needed, and market systems do achieve effective
coordination with much less communication of nonprice information than centralized,
planned systems use.
PRICES AND SocIALISl\l In the period after the Russian Revolution of 1917, there was
a maj or debate in economics concer ning the possibility of running a socialist ec onomy
efficiently. One of the first relatively precise statements of the efficiency of the
outc omes of price-taking behavior in markets was developed in the course of this
debate by Abba Lerner. Strikingl y, however, Lerner developed his arguments to show
how a socialist system, with collective ownership of the means of pr oduction, could
use prices to allocate resources as efficiently as could a capitalist market ec onomy.
According to Lerner, managers of socialist firms could be directed to take prices as
given and to determine quantities to equate prices to marginal costs, just as would be
done in a competitive firm. The resulting quantities would then have the same
efficiency under the socialist regime as those elicited under the market system.
(Presumably, however, the differing pattern of ownership of resources under socialism
would affec t what the market-clearing prices and quantities would be. ) Thus, there
would be no need for the central planners to attempt the overwhelming task of
determin in g the com plete allocation to be implemented. I nstead, they need " only"
determ ine and ann ounce th e righ t prices.
It would take us too far afield to discuss the econ omics of market socialism.
One point needs to be made, however. Al though properly selected prices may be able Using Prices for
to solve the coordination problem in a system without private ownersh ip, and al though Coordination and
much theoretical work has gone into the study of planning procedures for determ ining Motivation
prices to support an efficient all ocation of resources, attempts to determine the proper
prices in actual plann ed economies and to adapt them to changing circumstances
have in practice encountered maj or difficul ties. I t is not j ust that such determ inations
are directl y costl y, al though they certainly are. Perhaps more importantl y, changing
prices typicall y helps some individual s and groups and hurts others, and these
distributional consequences mean that the determination of prices by governments
becomes a political decision.
The consequences of distorted prices are often trul y striking. For example, in
Poland in the earl y 1 980 s, government bread subsidies l ed to severe shortages, as
farmers found it more profitabl e to feed bread rather than grain to their barnyard
animal s. Simil arly, in the Soviet Union in 1989, a personal computer that sold for
$3, 500 in the West coul d be sold to an industrial buyer for $145, 000 at the wil dly
inaccurate official rates of exchan ge. 3 When administered prices like these fa il to adj ust
to reflect changing circumstances, they cease to be useful guides to behavior; bread in
Pol and and computers in the S oviet Union were not being directed to their best uses.
I n this respect, a private property market system in which prices are determined by
individual s using economic criteria, rather than by planners and pol iticians worried
about the pol itical implications of distributional effects seems to have advantages.
THE NEOCLASSICAL MODEL
AND THEORIES OF ORGANIZATION
The fundamen tal theorem of welfare economics and the related resul ts on incentives
and informational efficiency represent an intell ectual triumph that, at a practical
l evel , is hel pful for thinking about what a system of prices can accompl ish. There are
few economists who woul d argue that the neoclassical model presen ts even an
approximatel y complete and accurate description of the way an y modern economy
works, however. Rather, the model represents an attempt to determine the scope of
the hypothesis that the "invisibl e hand" of markets and prices is sufficient to guide
the economy to an effi cient outcome.
We are interested in organizations, how they arise, and how they can be
effi ciently run. As Alfred Chandl er has observed, new organizations-especiall y
firms-historically often were organized when peopl e found that market outcomes
were inefficient. This observation hel ps to guide our study. If the competitive
equil ibrium of the neocl assical model actuall y did provide a good and compl ete
description of how markets work, there woul d be no need for other economic
organizations. Pol itical organizations might still arise as peopl e attempt to capture
larger shares of the benefits of j oint production or to bring more concern for equity
in to the system, but organizations aimed at improving economic efficiency woul d be
unnecessary. But where markets do not l ead to efficient outcomes, other institutions
may emerge in both the private sector and the publ ic sector to remove, avoid, or
mitigate whatever stumbl ing bl ocks are preventing simple markets fr om achieving
efficiency. Therefore, we look for market failures to expl ain nonmarket economic
organization.
Market Failures
We have al ready noted one aspect of market fa ilure, namel y, the possibil ity that firms
exercising market power will set prices at l evel s other than the competitive l evel and

' As reported in The Economist, August 5 , 1 989, p. 44.

 74
Coordination:
Markets and Demand
Supply �
Management

10
Q)
(.)

�
-- supply

Figure 3 . 2: With economies of scale, the

supply curve is discontinuous. The result in
this case is that there is no price at which the
1 00 200 quantity supplied equals the quantity
Quantity demanded.

thereby distort resource allocations. There are other ways in which markets may fail
to reach efficiency.
INCREASING RETURNS TO SCALE One of the less obvious limitations of the competitive
equilibrium is that, for some economies, it may be logically impossible for a
competitive equilibrium to exist. In other words, there may be no prices at which
supply is equal to demand simultaneously for all goods. This failure of existence is
especially likely when there are significant economies of scale in some production
processes, so that it is less expensive per unit to produce many units than to produce
few.
For example, suppose that consumers are willing to pay as much as $ 1 6 per
unit up to 1 00 units for some good A, but have no use for any additional units beyond
this amount. Suppose that the technology for making the product requires buying a
machine costing $ 1 , 000 and then using additional labor and material inputs costing
$ 5 for each unit produced, up to the machine's capacity of 200 units. This is an
example with returns to scale because the average cost of production per unit falls
continuously from $ 1 ,005 if the firm produces just one unit to $ 1 0 if it produces 200
units.
Figure 3. 2 shows the supply and demand curves for this particular market. At
any price less than $ 1 6, consumers will want to purchase 1 00 units. At higher prices,
they will want to purchase zero units. On the supply side, if it behaves as a competitive
price taker, in the long run the firm will want to produce exactly 200 units at any
price exceeding $ 1 0 and zero units at lower prices. (A price of $ 1 0 or more makes
buying the machine and operating it at capacity worthwhile; a lower price would not
induce the firm to buy the machine, although once it is installed the firm will produce
at capacity as long as the price exceeds the short-run marginal cost, $5.) There is no
price at which the quantity the firm wishes to supply is equal to the quantity that
consumers wish to buy (where the supply curve intersects the demand curve). Because
there are economies of scale, the supply curve is not continuous.
There is a value-maximizing solution in this example. In it, the firm produces
1 00 units and transfers them to the consumers. This yields $ 1,600 in value to
consumers at a cost of $ 1 , 500. Any other plan leads to less value being created.
Moreover, as long as the consumers pay the firm an amount between $ 1 5 and $ 1 6
per unit, both sides arc better off than when the good is not produced. The problem
is that prices alone cannot guide the firm to produce 1 00 units. The firm needs to
know not only how valuable the good is to others in the society but also precisely how
many units the consumers are willing to buy. Just as in the highway safety example
 75
earlier in the chapter, the fi rm n eeds information about quan tities in addition to Using Prices for
information about prices. Coord ination and
Of cour se, this failure of the price system in our th eo retical mode l is n ot the Motivation
same as a failur e of firms in the real world. Farmers may base their plan tin g on price
information and price forecasts alone, but producers subj ect to scale economics and
limited markets do keep in touch with their customers, using their sales forces to get
information about quantities, qualities, desired product attributes, an d much more.
Information processing for production plann ing is don e in various ways in
differen t countries. In centrally planned economi es, the state planne rs provide guidance
in a centralized way. In North America, firms are organized to coordin ate the affairs
of their various divi sions. Because most firms make only limi ted use of the price
system for organizing their internal affairs, it is reasonable to assume that the systems
used- in organizations require closer coordination th an a price system could ordinarily
provide. In :Western Europe, Japan, and Korea, there is more of a sense of partnersh ip
between the government and private industry than there is in N orth America. A
national in dustrial policy an d systems of indicative planni ng may be devised to h elp
coordinate activiti es among firms, both directly and by giving them common
information on which to base expectations. In the developing countries of the world,
a wide variety of approaches h as been examined to coordinate attempts at development,
with mi xed results. The use of other organizational arrangements to replace the price
system for purposes of coordination is discussed in more detail in Chapter 4.
EXTERNALITIES There are various conditions under which the outcomes of market
equilibria are not effi cient because one or more of the premises of the fundamental
theorem of welfare economics are not sati sfied. We have j ust seen that a maj or failure
can occur when increasing returns imply that competitive market-clearing prices fail
to exist. To the extent that increasing returns are a source of market power, it may
be that the market failures associated with imperfect competition are unavoidable.
Externalities provide another example.
Externalities are positive or negative effects that one economic agent's actions
have on another's welfare that are not regulated by the system of prices. Their presence
means that ineffi cient levels of externality- bearing activities may result. For example,
if a h omeowner paints h is house in pink and purple stripes, his consumption behavior
might make his neigh bors quite unhappy. Smoke from a nearby factory, which spoils
the consumers' environmen t and threatens th eir h ealth; investments in fa ctories by a
local employer, wh ich benefit others by creating j obs and raising property values;
inventions that provi de a foundation for other inventors to build upon-all these are
examples of consumption and production behavior that affect people and firms other
than th ose making the deci si ons. Ineffi ciencies occur wi th externalities because the
decision makers are not taking full account of all the costs and benefits associated
with their ch oices, namely, th ose th at accrue to other people.
MISSING MARKETS The market failure associated with externalities may also be seen
as a matter of missing markets: The externalities correspond to goods (or " bads") that
individuals would want to buy or sell because they affect utility or production
possibilities. B ecause these goods are not traded in competitive markets, h owever, no
prices are attached to them and so the market system fails to guide th eir allocation.
To illustrate furth er the power of the abstract formulation of the neoclassical model,
we could think of havi ng all the competitive markets that were needed, in which case
the conclusion of the fun damental theorem would con tinue to h old. Formally, th e
market model could be expanded to make person A' s consumption of a beautifully
lan dscaped h ouse one of the goods. Another good could then be created for each of
person A' s neigh bors, wh o must also, in a way, "con sume" A's h ouse. Th ese goods
 76
Coordination: would be joint products: If A consumes such a house, all his or her neighbors would
Markets and have to consume their corresponding goods as well. Because these are separately
Management priced goods, however, the neighbors would make market offers to buy and sell them
independently of one another and of A's choices. In a competitive equilibrium, their
enjoyment of the enhanced beauty of the neighborhood would lead them to buy the
goods. This would increase the total amount paid for the collection of goods
corresponding to A's landscaping and would bring the higher level of beautification
that efficiency requires but that would not result were A to have to pay for it with
only his or her own money, time, and energy.
Of course, such markets do not typically exist. Furthermore, if they did exist,
they could not reasonably be assumed to be competitive because there would be only
a single buyer of each of the extra goods. We see in Chapter 9 that the efficient
outcome may be achievable without all these extra markets. Certainly, however, the
absence of these mqrkets means that a system of impersonal market transactions
mediated only by prices will not generate efficiency.
In fact, with this wide conception of possible markets, missing markets are the
major source of market inefficiencies. We alluded earlier to the possibility of
interpreting goods in the neoclassical model as being differentiated by the date at
which they are available and by the particular realization of the uncertainties in the
world that have occurred by that date (the "event" or "state of the world"). Of course,
consumers do care about their future consumption, and there is always uncertainty
about endowments, tastes, and technology. Therefore, the neoclassical model effec­
tively assumes that there are competitive markets in which transactions can be made
to buy and sell any good for future delivery at every future date, with the prices on
these markets being contingent on the uncertain event as well. 4
For example, many homeowners who have property near the San Andreas fault
are concerned about being able to repair their houses after an earthquake. In the
context of the neoclassical model, there would be markets in which homeowners
could contract today for housing reconstruction to be done tomorrow, next year, or
22 years from now. Furthermore, for each date there would be separate markets, open
today, for contingent delivery of these services depending on the fault's movements
and its effect on the housing stock.
Clearly, such an extensive set of markets does not exist. Without these markets,
the version of the fundamental theorem stated earlier does not apply because the
assumption that everything of interest to consumers and firms is priced in a competitive
market is not met. Although it is possible that enough insurance markets may exist
to resolve homeowners' needs, this conclusion seems problematic. Some of the reasons
a set of complete, contingent futures markets does not exist will be discussed in
Chapter 5. For now it is enough to note that the absence of markets can lead to
inefficiencies and market failure. It is precisely this sort of market failure that leads
people to seek alternative arrangements to meet their economic needs.
SEARCH, 1\1.\TCHINC, AND COORDINATION PROBLEJ\IS In the neoclassical model, every­
one is assumed to know what the prices are and where and when goods can be bought
and sold. Finding a willing buyer or seller at the going price is assumed to be
unproblematic. In reality, however, potential buyers and sellers may not even know
of each other's existence, let alone the exact specifications of the goods and services
they seek or have to offer or the terms at which they are willing to buy or sell.

4 Actually, it is enough that there be securities and insurance markets that allow transfers of
purchasing power across time and between possible future events. This point was first made by Kenneth
Arrow, and it forms the basis for much of the modern theory of financial markets.
 77
Acquiring this information necessitates costly search by one or both sides. Consumers Using Prices for
need to search to find the best prices, firms search for potential employees, workers Coord i nation and
search for jobs, and companies expend resources finding suppliers who can meet th eir Motivation
needs and informing potential customers about their products. This search and
communication absorb resources. In addition, because these activities arc costly,
people do not continue searching until all the best matches are found. The difficulties
described in Chapter 2 in matching medical interns and hospitals are indicative of
the market failures that can arise when highly decentralized market approaches are
used. The NIMP, a more centralized solution, came into being to overcome these
inefficiencies.
The coordination problems that arose in the market for medical interns can also
manifest themselves at an economy-wide scale. In the neoclassical model, people and
firms. assume that they will be able to buy or sell as much as they want in the market
at the going prices. In fact, this may be reasonably descriptive of the actual situation
faced by a small investor buying and selling securities on one of the organized financial
exchanges. In most other markets the assumption is less accurate. People and firms
are concerned about their ability to buy and, especially, sell what they want. If workers
do not expect that they will be able to sell their labor, they are likely to cut back on
their purchases. They may become discouraged and not bother looking for work.
Meanwhile, if firms anticipate that demand will be low, they will not hire people to
produce goods that they do not expect to be able to sell. Then there are few jobs, as
workers feared. Further, because low employment means workers' incomes are low,
demand is weak, just as firms expected. The pessimistic expectations are self­
confirmi ng: People limit their purchases because they expect others to do so, and they
turn out to be right. It is possible that, i n the same physical circumstances, there
could be also be optimistic self-fulfilling expectations, in which firms hire and workers
spend more freely because they expect others to buy their goods or labor.
The upshot is that, contrary to the conclusi on of the neoclassical model, there
could be multiple possible levels of economic activity that are internally consistent,
with some having inefficiently low employment and output. In such a situation, it is
possible that no firm has any incentive to alter its prices or wages even though sales
are low and workers unemployed. Thus, the usual mechanism that neoclassical
economists expect to overcome these problems may be ineffective. A version of this
sort of coordination failure may also explain some of the problems of the economy
of the Soviet Union. Constant shortages of goods meant there was little incentive to
work hard and earn more because there was nothing to buy, and the resulting low
levels of production meant that the shelves were empty.

Market Failures and Organization

In the traditional view, the presence of externalities, unemployment, or other market
failures serves to justify special policies by the government, �hich can intervene to
set right the mischief of markets. The problem with this view is that when markets
fail, governments are not the only ones that can step in to set things right-individuals
and firms can also take action. When a manufacturer sets up a computer department
instead of buying computer services from a vendor, when an electric utility enters
into a long-term contract with a coal supplier rather than purchasing coal as needed
on spot markets, or when a group of homeowners bands together to provide a recreation
facility for their common use, the decision makers are exercising their option to use
nonmarket organizational forms to meet their needs. Private parties as well as
governments can make arrangements to replace the simple price system when they
are not satisfied that ordinary market arrangements have worked well.
 78
Coordination: USING THE PRICE SYSTEM W ITHIN ORGAN IZATIONS
t\ larkets and Even though formal organizations may be seen as a response to failures of the price
Management system, many large organizations in fact make extensive internal use of price systems
to help provide coordination and motivation. We have already noted a limited version
of this in the Salomon Brothers example from Chapter 1 , where the profits and losses
on each transaction are calculated and attributed to the individuals and groups
involved. A much more thorough use of prices is found in the management systems
of many multidivisional firms and a variety of other large organizations. These
organizations have sought to decentralize, locating the responsibility for making many
decisions at the levels where much of the relevant information resides and where the
actual operations take place. Decentralization creates an acute need to ensure that
the various people making key decisions have the proper incentives and that the
resulting plans are compatible and coherent. Senior managers have found that many
of the virtues of the price system in providing coordination and motivation hold within
their organizations as well as in the market. Consequently, they have partially re­
created the operation of the market within their organizations, using financial controls
and performance measurement and introducing internal transfer pricing for transactions
between units in the organization.

Patterns of I nternal Organization in Firms

The first modern firms were organized functionally in a centralized fashion. The Ford
Motor Company in the 1920s is an example (see Chapter 1 ). The head office in such
firms oversaw and directed all activity, with one department responsible for finance,
another for production, and still others for personnel, purchasing, logistics, sales, and
marketing for the whole organization. However, this form proved to be ill suited for
coordinating activity in large, multiproduct firms operating over broad geographical
areas. Head-office decision makers were too far removed from crucial local knowledge
about production and market conditions. Too much time and information were lost
in communication between the head office and the field, and the central managers
were overwhelmed with the number, size, and complexity of the decisions they faced.
These problems had nearly destroyed the Hudson's Bay Company in the eighteenth
century (see Chapter 1), but they became especially widespread in later times with
the emergence of more large firms. The larger the firms became, the more these
problems hindered effective management.
At the opposite extreme were the holding companies that emerged in the
nineteenth century. These were extremely decentralized collections of entirely separate
firms under common ownership. The head office of a holding company took little or
no management role, simply collecting the profits of the constituent firms. This
system worked reasonably well as long as there was no need to coordinate across units,
but it could not work when there were gains to making coordinated decisions across
units about investment, production, or marketing.
Several firms that faced these problems in the decades after the First World War
independently introduced similar solutions to it. They moved away from functional
or holding company organization and toward a multidivisional organization. This
essentially involves creating mini-companies-divisions-within the firm, each of
which is responsible for a particular product, market, region, or technology. The
divisions might in turn be organized functionally, or further subdivided in a variety
of ways, but the key was that a broad set of the decisions relevant to a particular unit
was made the responsibility of a single division manager. However, in comparison to
the extremely decentralized holding companies, the multidivisional firms had relatively
strong central offices to coordinate the divisions' activities.
 79
The companies that pioneered this form of organization in the United States Using Prices for
included Du Pont, which set up market-based divisions to separate its explosives Coordination an<l
business from its fertilizer business; Scars Roebuck, which established separate divisions Motivation
to handle stores in different regions of the country; Standard Oil of New Jersey, which
had grown by encompassing increasingly diverse business activities in oil exploration,
production, transportation, refining, and retail marketing; and General Motors, which
(as we describe in Chapter 1) was created as a combination of independent automobile
manufacturers. Each of these firms created structures to allow a measure of
independence to the divisions while providing a central office that could coordinate
their overlapping activities. In particular instances divisionalization thus could represent
increased decentralization, as at Du Pont, or increased central coordination and
control, as at GM .
. Today, the multidivisional firm is the dominant organizational structure in large
manufacturing firms throughout capitalist economies. Many nonmanufacturing firms
and some n�t-for-profit organizations, including a number of universities, also employ
key features associated with the divisionalized form.
Divisions can be defined in a variety of ways, even within a single firm. For
example, a firm could have a defense products division (market or customer defined),
a small electric motors division (product based), a biotech division (technology based)
and an international division (geographic). Senior management and the head office
staff retain primary roles in a number of areas, including raising outside capital,
allocating resources among divisions, appointing and evaluating divisional managers,
centrally coordinating the firm's overall policies, and setting its strategic direction. In
the most decentralized multidivisional firms, all other activities and decisions are
the province of the divisional managers, including R&D, design, engineering,
procurement, personnel, manufacturing, marketing, and sales. In less decentralized
firms, some of these functions reside with the head office, and divisional managers'
autonomy is circumscribed on some important dimensions by centrally determined
policies and directives.

Transfer Pricing in M ultidivisional Firms

The performance of divisions and their managers is always measured at least partly in
financial terms. Figures are developed measuring divisional cost, revenue, profit, and
investment performance, and these figures are used in judging performance and in
allocating rewards. They are also crit ical input for decisions determining where to
allocate corporate capital, what products and investments to develop and back, and
which executives to promote.
Within any decentralized organization, products and services are frequently
supplied by one division to another. Moreover, smaller responsibility centers are often
created within divisions, with financial measures employed to judge the performance
of these units and their managers. These responsibility centers can be particular plants
or offices or even smaller subunits. There is constant movement of products and
services between responsibility centers. Therefore, the use of financial performance
measures necessitates pricing these transferred goods and services. These transfer prices
are of crucial importance to the corporat ion and to the divisional managers.

Tl IE I\IPOHT.\NCE OF TR.\NSFEH PRICES From the point of view of the division, the
prices paid and charged on interdivisional transactions can be the single most important
determinant of the unit's measured financial performance. Consider an integrated
petroleum company that has one division that produces crude oil, another that
transports the oil to the firm's refineries, and a third that refines the crude oil into
petroleum products. Crude is "sold" by the production division to the transport
 80
Coordination: division, which then "resells" it to the refining division. The price in the first of these
Markets and transactions determines the revenues of the production division and is a major element
Management of the costs of the transportation division. The price in the second transaction
determines the revenue per unit of the transportation division and is a major element
of the costs of the refineries. Given the volume actually transferred, these transfer
prices do not affect overall corporate profit, but they do determine the apparent
performance of the various divisions.
However, if the division managers have autonomy over the determination of
the quantities they buy and sell, either internally or in dealings with outsiders,
corporate profits can also depend critically on the transfer prices. If the refinery
managers are judged on their division's profitability, and if the transfer price the
managers pay to the transport division is too high, they may seek to buy crude from
other suppliers. If the transfer price paid by the transport division to the crude producer
is too low, the latter's managers may decide to sell their output into the market, rather
than transfer it within the firm. Both of these transactions may adversely affect the
firm's overall profitability when the firm's profit is maximized by buying and selling
internally. Even when outside sales and purchases are not an issue, if each division
is free to determine how much it will buy and sell within the firm, a misspecified
transfer price may make a division unwilling to transact the quantities that are needed
to maximize total profits. This occurs because the transfer price makes the marginal
units bought or sold unprofitable for the division, even if they are profitable for the
corporation as a whole.
Badly chosen transfer prices can also misdirect corporate decisions, even when
the divisional managers do not have the power to decide where or how much to
purchase and sell. Altering the transfer prices can make an activity appear either
highly profitable or grossly unprofitable. Unless the central executives and staff are
especially careful in interpreting divisional performance measures, they may decide
that a given manager is doing a much better or worse job than in fact is the case.
More seriously, they may incorrectly conclude that a particular activity is highly
profitable and should be expanded or that it is unprofitable and should be dropped.

TRANSFER PRICES AND MARKET PRICES There is one special case where transfer prices
that are right on all counts are remarkably easy to determine: when there is an outside
market for the good or service, that market is perfectly competitive (and that would
remain so even if the firm joined the market), and there are no additional costs or
benefits to the corporation as a whole in using the market instead of transacting
internally. In this case, adopting the outside market price as the transfer price both
directs divisional quantity decisions to maximize corporate profits and provides the
right signals regarding performance and investment. It makes no difference to corporate
profitability whether the goods and services are bought and sold internally or externally,
so long as all transaction quantities maximize divisional profits at the given market
prices. Furthermore, these quantities are the ones that actually maximize corporate
profits, and the competitive market prices also serve as the best available indicators of
the marginal profitability of expansion or contraction. (We provide a formal proof of
these claims at the end of this section. )
Of course, it is rare that these conditions will all be satisfied completely.
Competitive markets for perfect substitutes for the firm's products and services are
likely to be found only for standardized commodities. Thus, an international grain
dealer or the manufacturer of common agricultural chemicals might find this condition
satisfied, but certainly an automobile manufacturer is unlikely to find a perfectly
competitive market for engines or transmissions that will fit into the company's designs.
More often, firms will have to rely on internal standard cost estimates to set
 81
Using Prices for
Coordination and
Motivation
Transfer Pricing at Bel/core
A typical example of the negative effects of setting the wrong transfer
prices comes from Bcll core (formerly known as Bell Labs), the research arm
of AT&T, which as American Telephone and Telegraph was previously the
telephone monopoly in the U nited S tates. Bellcore sells its research to
governmentally regulated regional and local telephone operating companies.
Bellcore discovered in the late 1 980s that its highly talented and well-paid
research engineers and scientists were typing their own letters, memos, and
research papers or negotiating with outside typists, risking the security of
internal communication s, while the typing pool that was supposed to do this
work was laying people off because there was not enough for them to do.
The cause of this gross misall ocation of personnel was that the transfer
prices paid for typing services were based on a flawed internal cost accounting
scheme, which allocated too large a share of fixed company overhead to the
typing unit. At their peak, the transfer prices for typing reached as high as
$50 per page! The scientists and engineers, who were also subj ect to financial
incentives, naturally chose not to use the service. As they withdrew their
business, the price needed to recoup the fixed costs assigned to the typing
pool rose ever higher.

This example is drawn from Edward Kovac a nd Henry Troy, "Getting Transfer Prices
Right: What Bellcore Did," Harvard Business Review (September-October 1 989), 1 46-54.

transfer prices. S till, a competitive market for a similar good may exist that can be
used to help set the internal price. 6 The B ellcore case (see box) is a good example.
Outside typing services were available, but they were less secure for sensitive
documents, less conveniently located, and perhaps slower. Nevertheless, it was the
comparison of the $50 per page price for internal typing services with the much lower
outside price that indicated that something was seriously wrong.
As we see in Chapter 1 6, when nearly perfectly competitive markets for inputs
are present, there is little reason for a firm to be verticall y integrated, supplying its
own inputs. For that very reason, the most common case for transfer pricing is where
there is no market for the desired input or, if the market does exist, it is far from the
competitive ideal. There may be few competitors in the market, so that prices are too
high on account of monopoly power. M ore importantly, there may be severe
transaction costs to using outside markets, which an se because of informational
differences (see Chapter 5).
Without a well-fu nctioning outside market, both determining whether the
transaction should occur within the firm or acr oss firm boundaries and establishing
the appropriate price to charge internally become complicated tasks. Division managers,
seeking to improve the profitability of their own divisions, have an interest then in
manipulating the transfer price, perhaps by assigning overhead costs to products for
which the buying division has no alternative source in order to infl ate the profits
reported on goods for which it faces competition.

6 J. R. Gould, "Internal Pricing in Firms When There A re Costs of Using an Outside Market,"
fournal of Business (1 964), 61-67.
 82
Coordination:
E
Markets and
Management

-
C

� P3 f--------'---�:;__----'----

Figure 3. 3 : When there is an outside market,

the difference in supply and demand within the
firm at the market price is optimally accommo­
Y1 dated by outside purchases or sales without ad­
U nits justing transfer prices.

Transfer Pricing ll'ith a Competitive

Outside Alarket
We claimed earlier that when a competitive outside market for its internal
products exists with a market price of p, the firm maximizes total profits by
setting its internal transfer price equal to p. There are two ways to go about
establishing this claim. The first way is intuitive and graphical and has the
advantage that it shows how large the losses are from using a different policy.
The second is more abstract but has the advantage that it unifies ideas by
connecting the analysis explicitly to two of the main ideas of economic analysis:
the fundamental theorem of welfare economics and the principle of total wealth
maximization.
A GRAP! IICAL THEATI\IENT Figure 3. 3 illustrates the argument. In the figure,
the prices P i , P2 and P 3 represent three possible levels of the market price. The
upward sloping line is the marginal cost and supply curve of the selling division
and the downward sloping one is the demand curve of the buying division. Only
by coincidence would the price that clears the internal market , P3 , happen to
coincide with the price determined on the outside market.
Suppose first that the outside price is low, at P 1 . If the transfer price is set at
the level P 3 at which internal supply and demand are equalized, what will the
firm's profit be? From basic microeconomics, the demand function is the
marginal revenue product of the buying division from acquiring units of the
transferred product and the supply function is the selling division's marginal
cost function. Consequently, the total profit enjoyed by the two divisions will
be the area between the supply and demand curves, which is the area of triangle
ADE. If, instead, the supply division produces x 1 and the buying division uses
y 1 , purchasing the excess y 1 - x 1 from outside suppliers in the market, then the
divisional profits will correspond to the areas of the triangles ABP 1 for the supply
division and CEP 1 for the buying division, so the total profit will be higher by
the area of triangle BCD.
Similarly, if the outside price is the high price P2 at which the supply x2 exceeds
the demand y 2, selling the extra supply on the market at price P2 increases total
profits by an amount equal to the area of triangle DFG when compared to
transacting internally at the price and quantity where supply equals demand.
 A FOHI\I.\L APPHO.\CI I Here, we prove the claim by constructing an arti ficial Using Prices for
economy and applying the fundamental theorem of wel fare econom ics and the Coordination and
value-maximization pri nciple to it. The exercise ill ustrates that the very same Motivation
logic establ ishing that prices guide efficient resource allocations in markets can
also be appl ied internally with in firms, just by rega rd ing divisions or managers
as if they were consumers.
In the artificial economy, there are two goods and three consumers. The two
goods are ( 1 ) the good that is transferred within the firm, and (2) money, which
by definition has a price of one. The first two consumers are the two divisions
(or their managers). They care only about how much of the internally produced
good they buy or sell and how much money they are paid for it. We define the
uti lity of each of these two consumers to be the total profit earned by the
· corresponding division . Let's denote the selling and buying divisions by the
symbols S and B. If Ys is the number of units sold and x5 is the total amount
received for it, we can write the selling division's utility as v5(y5) + x5 , where
v5(y5) is a negative number representing the cost of producing Ys units. For the
buying divisi on, utility can be written in the same general form as vB (YB ) + xB ,
but xB is a negative number, representing the amount paid, and vB (Y B ) is a
positive number, representi ng net profits before any payment for the transferred
goods. The third consumer in our arti ficial economy represents all the other
consumers in the actual market, so we designate the third consumer by the
symbol M. Consumer M has utility vM(Y M ) + xM , where we specify that vM (YM ) =
PYM · Consequently, M is willing to buy or sell any amount of the product at
price p.
What does the competitive equil ibrium of this artificial economy look like?
Fi rst, it must have price p, because M would want to buy or sell unbounded
quantities at any other price. Given the price p, consumer S sells the amount
that maximizes its utility v5(y5 ) + PYs and consumer B buys the amount that
maximi zes vB(YB ) + py8 . (Recall that YB is a negative number. ) At a market
equilibri um, supply must be equal to demand; that is, YM + YB + Ys = 0. Notice
that M's utility is zero regardless of how much it buys or sells at the market
price p, so this market-clearing equation is consistent with maxim ization by
consumer M. Since this is a competitive equi libri um, by the fundamental
theorem of welfare economics, the allocation it specifies is Pareto efficient. Also,
by construction, the value maximization principle appl ies, so an allocation is
efficient if and only if it maximizes the total value for all the three parties. Since
M's utility is always zero, regardless of the decisions YB and Ys of the two divisions,
the competitive equilibrium choices maximize the total utility of consumers B
and S; that is, they maximize the total profit of the firm. If the firm sets a transfer
price different from p, the result could only be lower total profits.
Notice that we have used the value-maximization principle and the fundamental
theorem of welfare economics here to characterize efficient behavior for the
divisions rather than for the actual economy as a whole. By replacing the actual
consumer sector by the arti ficial construct M whose welfare was unaffected by
the firm's decisions, we expl icitly omitted their welfare from the calculation .
The argument given here appl ies even if the fundamental theorem does not
apply to the actual economy, for example, because there is a problem of
monopoly power or externalities or missing markets. The kind of reasoning
displayed here is quite important in the positive economic analysis of organiza­
tions and should not be confused with reasoning about the actual efficiency of
the economy as a whole.
 84
Coordination:
Markets and
Management
SUMMARY
Even in the simplest situations, a team cannot work effectively unless its members
act in a coordinated fashion. In a modern economy, the problem of supplying the
right goods and services at the right times and places and to the right people, when
production takes place over a geographically widespread area, requires a major feat of
coordination. Moreover, the individuals involved must be motivated to do their parts
in the coordinated plan.
Prices can sometimes be used to achieve effective organization in large-scale
decision making, where decision makers are asked to value limited resources by placing
a price on each resource used. In our central example, there was a price in lives for
crew hours used on any project because those crew hours could be used in other
projects to save lives. Furthermore, with knowledge only of the price and the
characteristics of their individual projects, a decentralized staff could make decisions
that taken together constitute an efficient plan for the whole organization. The prices
that fulfill this function are market-like prices; they can be determined as prices at
which the supply of each resource is equal to the demand.
The Arrow-Debreu neoclassical general equilibrium model is capable of
representing in fine detail the complex activities of consumption and production by
firms and decision makers throughout the economy. A competitive equilibrium of
the model is a set of prices and an allocation of goods such that consumers are
maximizing their respective utilities, firms are maximizing their profits, and the
quantities that sellers wish to supply at the given prices are the same as those the
buyers wish to purchase. The fundamental theorem of welfare economics holds that
the competitive allocation of goods is an efficient one. At a competitive equilibrium,
prices provide consumers and firms with all the information they need to know what
to do, and they do not ask firms or consumers to do anything but maximize their
own profit or utility. In that sense, markets can theoretically resolve both the
coordination and the motivation problems. Experimental markets seem to verify that
even with moderately small numbers of participants, the behavior in markets is
approximately as predicted by the neoclassical model.
The neoclassical analysis also points to several problems for real economies.
First, for economies with increasing returns to scale, there may not exist any prices
at which supply equals demand, in which case prices alone cannot coordinate and
motivate appropriate choices . Economies of scale are also associated with imperfect
competition, which damages the efficiency of economic performance. Second, there
may be externalities or missing markets, so that individual decision makers may find
that prices do not accurately reflect social costs or that certain desired trades cannot
be made. As we see in Chapter 5, some of the missing markets are absent for good
reasons and are not easily restored.
Historically, one of the major reasons for the growth of the firm and certain
other nonmarket institutions was that unregulated markets could not achieve efficiency.
But nonmarket forms raise new problems of coordination, planning, and control. As
we have seen, prices can be of great value even in internal organization to evaluate
performance and guide managerial decisions.

• B IB LIOC BAPH IC NOTES

The neoclassical model of a private-ownership economy and the fundamental
theorem of welfare economics represent the culmination of two hundred years of
 85
economic research, beginning wi th Adam S mi th's Wealth of Nations ( 1 776). An Using Prices for
account of the hi storical development of the theory is found i n Roy Weintraub's Coordination and
article. The versi on and i nterpretati ons presented here are largely due to Kenne th Motivation
Arrow an d Gerard Debrcu, both of whom won Nobel prizes partly for thi s work.
The classical treatment is found i n Debreu's Theory of Value ( 1959), but this
volume is qui te abstract and highly mathematical. M ore accessible treatments are
found i n some intermediate and graduate microeconomic theory books and i n
Werner Hildenbrand and Alan Ki rman's book. Arrow's testimony to the U . S.
Congress (reprinted i n the H aveman-Margolis volume) is an especially clear
expositi on of the key ideas. His idea of viewing externalities as mi ssi ng markets
is developed i n the paper in the Margolis volume.
Theoretical and experimental explorati ons of incentives i n markets and the
question of whether the competitive equilibri um approxi mates the outcomes of
imperfectly competitive behavior are acti ve areas of current research. Brief
i ntroductions to these subj ects can be found i n the articles i n The New Pa/grave
by J ohn Roberts and Vernon S mi th. The low i nformati on requirements of the
price system were accentuated by Friedrich H ayek in the course of the debate on
the possi bility of runni ng an efficient soci ali st economy. Abba Lerner's proposal
for market socialism and his development of the relati on between pri ce- taking
behavi or and efficiency are published i n The Economics of Control. The
formalization of H ayek's ideas is primari ly due to Leonid H urwicz, who gi ves an
i ntroducti on to this topic in hi s paper in American Economic Review.
M ost i ntermediate microeconomics textbooks treat the problems that the price
system has with increasi ng returns, imperfect competiti on, and externaliti es. The
focus on mi ssi ng markets is agai n due to Arrow. S ee agai n hi s Congressi onal
testimony referenced earlier. The i mportance of search was first developed by
George S tigler. H e was cited for this work when he won the N obel prize. The
possi bi lity of coordi nati on fai lures explaini ng unemployment is another active
area of current research. Peter Diamond has been a leader i n thi s work. Perhaps
the most complete modeli ng along these li nes i s Roberts's paper i n American
Economic Review, but it is qui te difficult and there is at present no easily accessi ble
expositi on of thi s work. The closest approximati on may be the paper by Russell
Cooper and Andrew J ohn. The idea that organizati ons are a response to market
failures is developed i n Arrow's book.
Transfer pricing is covered i n most texts on managerial accounti ng. The classic
i n the field i s by Charles H orngren and George Foster.

• REFERENCES
Arrow, K.J . "Political and Economic Evaluati on of S ocial Effects and Externali­
ties," i n The Analysis of Public Output, J . M argolis, ed. (New Y ork: Columbia
U niversi ty Press, 1970).
Arrow K. J. "The Organizati on of Economic Activity: Issues Pertinent to the
Choice of Markets versus Nonmarket Allocati on," i n Public Expenditures and
Policy Analysis, R. Haveman and J. M argolis, eds. (Chicago: M arkham, 1970).
Arrow, K.J . The Lim its of Organization (New Y ork: N orton, 1 974).
Arrow, K.J. and G. Debreu "Existence of an Equi li brium for a Competi tive
Economy," Econometrica, 22 (1954), 265-90.
Cooper, R., and A. J ohn. "Coordi nating Coordi nati on Failures in Keynesian
M odels," Quarterly Journal of Econom ics, 1 0 3 (August 1988), 44 1-64.
Debreu, G. The Theory of Value (New York: Wiley, 1959).
 86
Coordination: Diamond, P. A. "Aggregate Demand Management in Search Equi librium, "
Markets and fournal of Political Economy, 90 (October 1982), 88 1 -94.
Management Hayek, F. "The Use of Knowledge in Society, " American Economic Review, 3 5
( 1 945), 5 1 9-30.
Hi ldenbrand, W. , and A. P. Kirman. I ntroduction to Equilibrium Analysis

Horngren, C. T. , and G. Foster. Cost Accounting: A Managerial Emphasis, 7th

(Amsterdam: North-Holland, 1 97 5).

ed. (Englewood Cliffs, NJ: Prentice Hall, 1 99 1 ).

Hurwi cz, L. "The Design of Mechani sms for Resources Allocation, " American
Econom ic Review 63 (May 1 973), 1-30.
Lerner, A. P. The Economics of Control (New York: Macmillan, 1946).
Roberts, J. "Large Economies, " in The New Pa/grave: A Dictiona ry of Econom ics,
J. Eatwell, M. Mi lgate and P. Newman, eds. (London: Macmillan, 1 987),
volume I II, 1 3 2-3 3 .
Roberts, J . "Perfectly and Imperfectly Competitive Markets, " in The New Pa/grave:

Macmillan, 1 987), volume Ill, 837-4 I .

A Dictionary ofEconomics, J. Eatwell, M. Milgate and P. Newman, eds. (London:

Roberts, J . "An Equilibrium Model with Involuntary Unemployment at Flexible,

Competi tive Prices and Wages, " American Econom ic Review, 77 ( 1 987), 856-
74.
Smith, V. L. "Experimental Methods in Economics, " in The New Pa/grave: A
Dictiona ry of Econom ics, J. Eatwell, M. Milgate and P. Newman, eds. (London:
Macmillan, 1987), volume I I , 24 1 -49.
Stigler, G. "The Economics of Informati on, " fournal of Political Economy, 69
(June 1 96 1 ).
Weintraub, R. "On the Existence of Competitive Equilibrium: 1 9 30- 1 954, "
fournal of Economic Litera ture, 2 1 (March 1 983), 1- 39.

EXERCISES

Food for Thought

1 . Airline companies often try to allow for passengers who fai l to show up for
fl ights by a practice called overbooking-they sell more seats than are actually available
on the flight. When all the passengers with reservations do show up, i t becomes
necessary to "bump" some passengers to a later fli ght. What criteria mi ght be used to
choose which passengers to bump? How are passengers li kely to react to your proposed
criteria? In recent years in the United States, it has become common to offer bonuses
to passengers who will volunteer to be bumped. Evaluate this alternative from the
perspectives of both equity and efficiency.
2. In many countries of the world, an attempt is made to ensure the welfare
of the poorest families by subsidizing the cost of basic necessiti es. For example, many
countries subsi dize the price of bread, reducing it far below the market price. What
are the disadvantages of this approach to caring for poor families? Can you identi fy a
better approach?
3. In California, for historical reasons, city dwellers pay as much as twenty
times more for water than farmers pay. What consequences would you expect to
follow from such a large difference?
 87
4. In many colleges and universities, space in dormitories and other university Using Prices for
subsidized housing is assigned by a complicated system ba sed on seniority and a Coordination and
lottery. Why is a price system not used to assign rooms to those who value them Motivation
most? What would be the advantages and disadvanta ges of such a system?

I Mathematical Exercises I
1 . In the transfer pricing problem studied in Figure 3 . 3 , when the market
price is P i , the gain from using the outside market rather than just buying and selling
internally within the firm was shown to be equal to the area of triangle BCD.
Reconstruct the graph and show what portion of this gain results from increased
purchases by the buying division. Show, too, the portion that results from reduced
output by the selling division. If the firm has been using an internal transfer price of
P3 with no outside purchases or sales and adopts the market price of P i , allowing
outside purchases and sales, how are the measured profits of each division affected?
2. Consider an economy with two kinds of goods that people value. We call
these two goods "money" and "manna. " There are also two types of people. The first
type-"manna lovers"-value a combination of x 1 units of money and y 1 units of
manna according to the utility function x 1 + (3y 1 - yr). The second type-"money
lovers"-value x2 units of money and y2 units of manna according to the utility
function x2 + (2y2 - YD- Each type of consumer is endowed with one unit of manna
and ten units of money. Use the value-maximization principle to determine how
manna must be allocated among the two types of consumers at any efficient allocation
in this economy. What must the price of manna be (in money units) in order for
there to be a competitive equilibrium in this economy? Use the first part of the
problem to show that the competitive equilibrium allocation is efficient.
3. Consider an economy with two kinds of goods, called Xs and Ys, and two
kinds of people, called "X-lovers" and "¥-lovers. " An X-lover who consumes x units
of good X and y units of good Y enjoys utility of 2/n(x) + ln(y). A Y-lover who consumes
x units of good X and y units of good Y enjoys utility of ln(x) + 2/n(y). There are equal
numbers of X-lovers and ¥-lovers in the economy and each is endowed with three
units of each kind of good. If the prices of good X and good Y are each one per unit,
how many units of each good will be supplied or demanded by each kind of person?
Use your answer to show that there is a competitive equilibrium at which the price
of each good is one. Show that, in this case, the competitive equilibrium allocation
maximizes the total utility of all the people in the economy. Argue that this implies
that the allocation is efficient.
 4
COORDINATING PLANS
AND ACTIONS

[I] f there really were some basic in trinsic advantage to a system which employed
prices as plann ing instrumen ts, we would expect to observe many organ izations
opera ting with this mode of control, especially a mong m ultidivisional business firms
in a competitive environ ment. Yet the a llocation of resources within private
companies (not to men tion governmen tal or nonprofit orga n izations) is almost never
con trolled by setting administered transfer prices on com modities and letting self­
in terested profit maxim ization do the rest. The price system as a n alloca tor of
in ternal resources does not pass the market test.
Ma rtin Weitzman •

{I] t is surely importa nt to inquire why coordination is the work of the price
mechanism in one case and of the entrepreneur in a nother.
Ronald Coase2

[M]odern business enterprise took the place of market mechanisms in coordina ting
the activities of the economy and allocating its resources. In many sectors of the
economy, the visible hand of management replaced wha t Ada m Smith referred to as
the invisible hand of market forces.
Alfred Chandler3

In this chapter we explore some of the many ways in which economic coordination
is achieved in economies and within organizations other than through a highly
decentralized system of prices and markets. We examine the characteristics of different
specific sorts of coordination problems and of the mechanisms used to solve them,
and we develop elements of theories to help us understand which solutions are efficient
m which situations. Because the subject matter of this chapter has received less

1 "Prices versus Quantities," Review of Economic Studies, 4 1 , October 1 974, 477-9 1 .

2 "The Nature of the Firm, " Economica, 4 , 1 937, 386-40 5.
3 The Visible / land: The Managerial Revolution in American Business (Cambridge, l\1A: The
88 Belknap Pres� of l larvard University Press, l 977), p. l .
 89
scrutiny from econom ists than other parts of the theory of organizations, the Coord inati ng Plans
developme nt here is less complete and the conclusions more tentative than elsewhere and Actions
in th is text.
In Chapter 3, we saw that a price system could sometimes solve the fundamental
and immensely complex problem of coordinating the plans and actions of all the
diverse decision makers in a modem economy. According to the central result
developed there in the fundamental theorem of welfare economics, if prices on a
complete set of competitive markets are set so that the quantities of each good supplied
and demanded are equal, then the resulting allocation of resources is effi cient.
M oreover, the price system achieves this remarkable feat of coordination without
requiring communication among individual decision makers of anything more than
the summary information about the economy embodied in the prices and without
requiring any individual to do other than what he or she deems to be in his or her
own best interests. We also saw that sometimes the price system can be used inside
firms to obtain similarly effi cient results. And even though actual markets do not fully
meet the assumptions of the theorem, the evidence across nations and over the years
is overwhelming that a decentralized system of prices and markets based on private
ownership is an extremely effective mechanism for solving the coordination problem.
Yet, as M artin Weitzman argues in the quotation that opens this chapter, formal
organizations make at most quite limited use of prices to coordinate their internal
activities. Indeed, as we have suggested already in Chapters 2 and 3 , organizations
can be thought of as arising and supplanting the market when they offer more efficient
mechanisms for coordinating economic activity and motivating people to carry out
the resulting plans. Given this, it would be somewhat surprising to see them rely very
heavily on an internal price system. Instead, managers more usually formulate general
strategies, make these operational by specifying quantitative goals, develop specific
plans to realize these goals, and then direct people to carry out their specified
roles using the resources they have been allocated. Routines are developed, and
administrative processes and procedures are instituted to guide activity. All this is done
in telephone conversations and meetings and is embodied in memos and spreadsheets.
The language used is not that of prices but rather of technological, organizational and
individual capabilities, quantitative performance levels, specific plans and budgets,
and detailed work assignments and operations schedules. And when changing
availabilities and capabilities of people and physical resources within the organization
require adaptation, the signals that indicate this need and guide the organization' s
responses are rarely prices.
Even in market systems, there is extensive use of means of coordination besides
prices. Governments in particular favor giving direct orders that specify particular
actions to be taken. They set quantity limits on the pollutants that a vehicle or factory
can emit, on the amounts of impurities that foods and medicines can contain, on the
speed that drivers can select, and on the minimum number of years children must
spend in school. They command resources directly, as in a system of compulsory
military service. They provide goods and services without explicitly pricing them:
roads, police services, health care, food for the needy, and so on. In some countries,
including Japan and S outh Korea, they target industries for expansion and new
technologies for development and then coordinate explicitl y the realization of these
plans. In the centrally planned communist economies, governments have attempted
to coordinate the finest details of resource allocation through quantity plans and
orders. And, during World War II, government planners in the most market- oriented
economies directed the production and use of a variety of high-priority goods, including
not j ust steel and rubber, but also sugar and meat.
M oreover, firms often do not arrange their dealings with one another as simple
 90
Coordination: market transactions. For example, they may work cooperatively with their suppliers
Markets and to develop together the specifications for the inputs they seek, and they negotiate
Management complex requirements contracts under which one firm has the power to direct how
much the other will supply to it. Less formally, they may share information on plans
and estimated requirements. They set up joint ventures and enter other sorts of
alliances, they design complex royalty agreements and franchise contracts, and they
put law firms on retainer, paying them even if they do not use their services. Within
these relationships, they may exchange large amounts of information and often
formulate joint plans.
In Chapter 3 we explored reasons why a price system would not yield an efficient
outcome and so might be replaced: increasing returns, externalities, missing insurance
and futures markets, excessive search costs, and the possibility of unemployment
equilibria. In this chapter we focus mostly on situations where, in principle, the price
system could be used and the fundamental theorem of welfare economics would hold,
and yet other mecha.nisms for coordination are actually employed.
Planning and coordinating economic activity never come for free. It takes real
resources to plan-people with offices, files, data banks, and the computing and
communication equipment to support them. In addition to the planners' time,
planning demands time from production people who must fill out forms, complete
reports, and answer the planners' queries. At the end of the planning process, errors
inevitably still occur, both because the prices or plans are based partly on guesses and
partly on erroneous, incomplete, or misleading information and because miscalcula­
tions and mistakes occur.
In actual economies, a loose mix of systems is used to coordinate and manage
the various kinds of activities. What determines which system is or ought to be used
in any particular set of circumstances? How can we account for Weitzman's observation
that prices are often ignored in internal decision making, so that "the price system as
an allocator of resources does not pass the market test"? To answer these questions,
we need to study coordination problems and systems in more detail.

THE VARIETY
OF COORDINATION P ROBLEMS AND SOLUTIONS
Robinson Crusoe, living alone before meeting Friday, took care of all his own needs.
Gathering and preparing his own food and securing his own shelter, Crusoe spent no
effort coordinating his activities with those of anyone else. The need for coordination
comes from specialization, in which various tasks are divided among a group of
people, each of whom relies on the others for part of the job. As noted in Chapters
2 and 3, specialization creates the opportunity for enormous increases in productivity.
People who specialize in a job can prepare specialized tools, gain specialized training,
develop specialized methods, and exploit their accumulated experience to get more
done, more quickly and with fewer resources.
The kind of coordination that is most effective depends on the nature of the
task. It is helpful to distinguish among several kinds of problems to understand the
kinds of solutions that are used. The most general kind of coordination problem we
consider is called a resource allocation problem. This is a problem of allocating a
fixed set of resources among various possible uses. The term resource can be interpreted
broadly enough to classify virtually every kind of important economic or business
decision as a resource allocation problem. In this chapter, however, it is useful to
distinguish particular attributes of resource allocation problems that make one system
of coordination or another especially effective.
 91
Design Attributes Coordinating Plans
We arc especially in terested in problem s with desi gn attributes. Th ese arc problem s and Actions
in which ( I ) there is a great deal of a priori information about the form of the optim al
solu ti on, th at is, about h ow th e variables should be related, and (2) failing to achie ve
the right relation ship among the variables is gen erally more costly than are other kinds
of errors, including especially sligh t misspecifications of th e overall pattern, as long
as the individual pieces fit. In this discussion, the word design is a general term
describing a system in which the pieces must fit together in a predictable way, th us
narrowin g the search for efficient decisions. Two common kinds of problems with
design attributes are synchronization problems and assignment problems.
SYNCHRONIZATION PROBLEMS An extreme example of a synchronization problem
arise's in the sport of crew, in which it is crucially important that each rower make
his or her stroke at precisely the same moment. The coxswain sol ves the synchronization
problem by determining a rh ythm for th e crew and calling out the signal for each
stroke. Like most centrally directed solutions, synchronization has the disadvantage
that the centrally made decision cannot be fully responsive to information of the
others in the system. In this case, the coxswain can only guess h ow tired the individual
crew members are. This could be an important disadvantage if the coxswain pushes
the crew too hard early in the race, leaving them too weak for a strong finishing
sprint. H owever, the great advantage of the system is it synchronizes th e actions of
the crew, making their individual efforts much more effective. The costs of not setting
quite the right pace are very small compared to th ose of failing to have everyone
pulling in unison.
Th ough it seems ridicul ous to contemplate using prices in th is context, it is
illuminating to see what a price solution for this problem would be and wh y it would
not work well. In this application, a price system would entail the coxswain telling
each rower the "price" of effort, that is, h ow valuable a unit of extra effort at this
moment would be to the team. Then, each rower would ch oose his or h er own
action, taking full account of his or her own ph ysical condition and the summary
information supplied in th e form of prices by the coxswain. In principle, if th e
coxswain could determine the righ t prices and could costlessly communicate them to
the crew, and if the crew could make the right decisions based on that information,
the resulting level of effort would be j ust righ t. In practice, a system of prices would
fare badly for various reasons. First, it would be too difficult or costly for the coxswain
to obtain the relevant information from the rowers and then to determine the prices.
Second, communicating the prices back to the rowers would be too difficult and too
slow. Third, the crew migh t respond inaccurately to th e prices, failing to ach ieve
coordination even if the prices were set correctly. Finally, small errors that disturbed
the synchronized rh ythm of the crew would be very costly. The price system has the
same advantages in th is application as in others with multiple producers. It takes full
account of information about the individual condition of each producer, but that
advantage comes at too high a cost in this synchronization problem.

AsSIGNI\I ENT PROBLEMS Similar diffi culties arise in assi gn ment problems, in which
there are one or more tasks to accomplish and there is a need for j ust one person or
unit to do each. The coordination pr oblem is to ensure that each task is done and
that there is no wasteful du plication of effort. For example, if someone is seriously
inj ured in an automobile accident, there typically is a need for one ambulance at th e
site of the accident as soon as possible. In practice, someone calls for an ambulance
and then a central dispatcher assigns a particular ambulance to drive to the site. Even
 92
Coordination: if the dispatchers are in constant communication with the units, they may not be
Markets and fully aware of all the relevant circumstances about the location and condition of the
Management ambulance, of surrounding traffic, or of the training and experience of the crew. In
principle, the dispatchers could try to determine a system of prices to determine which

they j ust select one and send it. If the prices were set incorrectly, more than one
ambulance could provide the best service in the most timely fashion, but in practice

ambulance might rush to the scene, or no ambulances might respond. A system of

prices performs poorly in assignment problems because it often leads to unnecessary
duplication or costly delay.
It is interesting to contrast the way decisions about the uses of ambulances are
made with decisions about how many ambulances to keep available, what equipment
they should have, and how the ambulance staff should be trained. These latter
decisions, made in relatively unhurried circumstances, take account of people's wages
and salaries as well as the prices of equipment, vehicles, and training; they also utilize
estimates of the benefits of additional ambulance capacity. At higher levels of decision
making, people may also take account of the opportunity value of funds used for
ambulances, which might otherwise be used to hire additional police officers or to
build up an emergency fund or to buy additional park land.
DESIGN PROBLEMS AND ORGANIZATIONAL ROUTINES Our crew and ambulance exam­
ples combine several features: a sense of urgency about the decision, the extreme
dependence of the optimal course of action on particular circumstances (Are the
opponents ahead or behind? Is there an accident, and if so, where?), and the coxswain's
and central dispatcher's substantial knowledge about the form of an optimal decision.
Together these make central control an attractive alternative to a decentralized
approach, as through a system coordinated by prices. However, when design problems
arise repeatedly and call for largely the same solution each time, it may be unnecessarily
expensive to solve each of them anew by centralized direction. Instead, established
organizations set up routines that guide decentralized solutions to the recurrent design
problems. For the most common kinds of demands made on the organization, no
upper-management discretion needs to be exercised because those who first become

higher-level personnel. If someone notices that paper for the copy machine is getting
aware of the demand know what to do and who to notify, without ever involving

low, then (depending on the established routine) he or she either simply orders more
and leaves a sign telling others that paper has been ordered, or else he or she notifies
the designated individual in the office whose job is to order more paper. With well­

parts. If the environment of the organization changes, however, then the same routines
established routines, each part of the organization can rely on the others to do their

that were effective may become counterproductive in the new environment, and new
routines will need to be devised if the organization is to continue to achieve its goals.

Innovation Attributes
Highly decentralized decision making, whether guided by prices or by organizational
routines, will perform poorly whenever the optimal resource allocation depends on
information that is not available to any of the people at the operating levels of the
organization. Coordination problems with this innovation attribute are most com­
monly present when the organization is trying to do something that is outside its

adopting a new approach to manufacturing. If a firm wants to consider replacing its

experience, such as introducing a new kind of product, entering a new market, or

system of buffer inven tories of work-in-process at each work station with a just-in­
time system (in which production is synchronized so that each station receives the
necessa ry inputs just in time to process them), then the current production and
inventory managers may not have the experi ence to give a wel l-informed assessment Coord i nating Plans
of the costs and bene fits. Th e deci si on cannot be properl y made usi ng just the and Actions
i nformati on already avai lable i n the system.
When i nnovati on attri butes are present, effecti vely solvi ng the coordinati on
problem commonly involves someone gatheri ng or developi ng the needed i nformati on
and then commu nicati ng it to deci si on makers i n the organizati on. Thi s task mi ght
be taken on at his or her own i nitiati ve by someone at an operati ng level who has
recognized that the problem exists. In other cases, the task will require more resources
than l ower-level personnel are able to muster on their own, and hi gher-level deci si on
makers will have to become i nvolved, allocati ng resources and assigni ng people to get
the needed i nformati on.
In any case, the fact that a deci si on requires gatheri ng new i nformati on from
outside the organizati on does not imply that the deci si on ought to be centralized.
People i n operati ng positi ons may be best equi pped to combi ne new knowledge with
local knowledge, or they may be better motivated to make the new system successful
if they are also the ones who designed it. An effective way to handle thi s problem
may be to educate members of the organizati on about the proposed alter nati ves and
to i nvolve them i n the final deci si on. What is certai n, however, i s that a simple price
system relyi ng only on the responses of i ndi viduals usi ng their local knowledge cannot
reli ably achi eve an optimal plan i n these circumstances.

Com p aring Coordination Schemes

The sample coordi nation problems we have described differ from one another, and
the soluti ons that people actually adopt vary wi dely. S ome deci si ons may be extremely
urgent, with little time to process i nformati on. Others may require extremely close
synchronizati on or coordi nati on of assignments, with little tolerance for faults. S ome
require onl y maki ng effective use of i nformati on that is already i n the organizati on,
whereas others require the infusi on of new i nformati on.
Coordi nati on systems also vary. Certai n centralized command systems demand
little upward communicati on of local knowledge and yet they still arrive qui ckly at
reasonable plans and commu nicate clearly what the operati onal personnel are expected
to do. Others require much more upward commu nicati on, but are correspondi ngly
more responsi ve to local knowledge. Decentraliz ed systems emphasize communi cati ng
informati on to support local deci si ons, with the required volume of commu nicati on
bei ng between the two centralized extremes. When these vari ous systems fail, they
fail i n predictably di fferent ways. The pri ce system may work too slowly or lead to
duplicati on, or it may require too much i nterpretati on of i nformati on by deci si on
makers. The coxswai n's call and the ambulance di spatcher's i nstructi ons are quick
and their meani ngs are clear, but they are i nsensiti ve to local knowledge.
There are more aspects to the coordin ati on problem than just what the means
of coordi nati on will be. The manufacturi ng example pr ovides an illu strati on of how
a system could be designed to reduce the value of close coordi nati on while still
capturing gai ns to speci alizati on : The system of buffer in ventories elimi nates the need
to synchronize producti on but requires costly holdi ngs of i nventories. 4
CRITERL\ FOR COMPARING SYSTEMS We use three criteria to compare how well
di fferen t systems perform i n solvi ng vari ous ki nds of economi c problem s. First, if all
the informati on required by the system were reported, if all reports were made hon estly

4 Inventory costs include the interest on funds used to finance inventories, losses due to spoilage,
theft, obsolescence, the cost of space in the storage area, and certain other costs that come from the
tendency of inventories to conceal the effects of production performance that varies over time.
 94
Coordination: and accurately, and if informati on processi ng were perfect and costless, could thi s
Markets and system achieve an effici ent decisi on? Sec ond, j ust how much c ommunication and
Management information does the system requi re to achieve its purpose? Are there other systems
that could do equally well with less communication? Third, how brittle is the system?
That is, if some of the desired i nformation i s missing or inacc urate, how badly will
the system's performance deteri orate?
To c ompare alternative coordinati on systems in different settings, we rely on a
cost- benefit pri nciple: The system that should be adopted in any specific situation is
the one that maximizes the net benefit, after properly acc ounting for all these ki nds
of costs. Unfortunately, the present state of knowledge does not allow us to make
specific statements about which system will work best in any particular ci rcumstance.
N evertheless, there is much of value to be learned by studying some specific problems
of coordination and control and seei ng how the optimal soluti on depends on the
details of the problem.

PRICES VERSUS QUANTITIES : ASSESSING BRITTLENESS

I n order to study more closely the effecti veness of different approaches to coordi nation,
we look first at the standard ec onomic problem of allocating scarce resources. The
overall obj ecti ve is to compare a system of prices wi th a system of centralized quantity
planning in which the coordinator instructs the production uni ts how much to produce
and wi th what resources. H owever, in this context it is not sensible to i gnore how
prices are set (as is done i n much of Chapter 3). Thus, we compare two systems: one
in which the central coordinator simply specifi es the production units' quantities, and
another in which the center attempts to guide the units' decisi ons via price signals,
c ounting on the units' man agers to respond by picki ng the appropriate quantities. The
price-based system recalls the proposal for market socialism discussed in Chapter 3.
The problem could also be thought of as that of a firm attempti ng to c oordi nate
intern al producti on decisions.
Because the informati on available to the planner is fixed in thi s analysi s, we
use two criteria to compare the performance of the alternative systems. Fi rst, when
the planner' s information is perfect, does the system permi t the planner to achieve an
efficient outcome? Sec ond, when the planner's i nformation is imperfect, to what
extent does the system performance fa ll short of the performance wi th perfect
i nformation? That is, how brittle is the system?

Some Exam p les

Suppose the planner knows the benefits accruing to any level of output from a
producti on unit and wants to ensure that the efficient level of output is actually
produced. U nfortunately, the planner may not be perfectly informed about production
costs. I nstead, it relies on its estimate of these costs, which may be wrong.
PERFECT l:'>JFOR!\IATION A sample calc ulati on is i llustrated in Table 4. 1 . On the basis
of the planner's cost estimate, total benefits minus total costs are maximized if either
5 or 6 units are produced. I f the planner directs the fi rm to produce 6 uni ts, the
estimated net benefit will then be 38 (total benefits of 58 mi nus the total cost of 20).
The net benefit is also 38 if 5 units are produced, but it is no more than 36 for any
other output quanti ty.
Still assumin g that there is in fact n o error i n the cost estimates, the planner
could acc ompli sh the same thing by setting a price of 8 and telling the firm to produce
the q uan tity that is m ost profi table at that price. The marginal benefit and the marginal
cost of the sixth un it are both equal to 8. (In terms of supply an d deman d, the
marginal ben e fit schedule tabulates the prices at wh ich the quantity demanded 1 s
 95
Table 4. 1 Coord i nating Plans
and Actions
Planner's Error
Cost Estimate Scenario
Number Total Ma rginal Total Ma rginal Net Total Marginal N et
of U nits B enefits Benefit Costs Cost Benefit Costs Cost Ben efit

4 40 5 35 17 23
5 50 10 12 7 38 20 3 30
6 58 8 20 8 38 24 4 34
7 64 6 29 9 35 29 5 35
8 68 4 39 10 29 35 6 33
9 70 2 50 11 20 42 7 28
10 70 0 62 12 8 50 8 20
In the error scenario, the planned quantity of 6 leads to a net benefit 34 out of a possible 3 5, but
the planned price of 8 leads to a net benefit of only 20.

equal to the listed num ber of units, a nd the marginal cost schedule ta bulates the sam e
information a bout the quantity supplied. ) At a price of 8 the firm can maximize its
profits by producing either 5 or 6 units, and the outcome of the process maxim izes
the total net benefits. 5
Thus, the answer to our first question is "yes" for both the price-directed and
qua ntity-directed system s: If the pla nner has perfect information, then both system s
result in a n efficient outcome. Thus, we m ove to the second question: H ow is
performance degraded by imperfect information?
I MPERFECT INFORMATION To exam ine this question, suppose that the decision maker's
estimate of the cost is wrong a nd that the actual cost characteristics are those given
in the E rror Scenario columns of Ta ble 4.1. The firm will know that these a re its
true costs when it actually comes to carry out production, but it cannot inform the
planner of the true costs before the price or qua ntity directive is announced.
The price or quantity to be a nnounced is still determ ined on the ba sis of the
planner's (incorrect) estimate. That is, the qua ntity is set at 6 and the price a t 8. Also,
in both cases, the socially efficient quantity to produce is 7 because that maximizes
the total benefits minus the actual tota l costs. With a quantity of 7, the actual net
benefit would be 35 (64 total units of benefit m inus 29 total costs).
If the plan specifies an output of 6 units, then the net benefit will be only 34,
so there is a loss of 1 unit of net benefit on account of the estima tion error in this
system (where a quantity is specified). In the sam e circum stances, suppose that the
planner had tried to control the producer using a price system and had fixed a price
of 8. In this ca se, given its a ctual costs, the firm maximizes its profits by producing
to the point where its marginal cost is equal to 8, that is, by producing IO units of
output, yieldin g a net benefit of only 20. 6 The firm has responded to the price and
its actual costs as effi ciency requires that it should, that is, by setting output so that
ma rginal cost is equal to price. The incorrect price ha s led to a n inefficient quantity,

5 This differs from a standard monopoly problem in that the firm here is forced to take the price as
given and is not permitted to manipulate the price by varying its output decision.
6 In this case, the firm could also maximize its profit by setting its output at 9 because it is just
indifferent about producing the last unit. In the interests of clarity and simplicity, we will always use the
output level where the price is equal to the marginal cost to characterize the firm's decision.
 96
Coordination: Table 4. 2
t\ larkets and

Planner's Error
t\lanagement

Cost Estimate Scenario

Number Total Marginal Total Marginal Net Total Marginal Net
of Units Benefi ts Benefi t Costs Cost Benefit Costs Cost Benefit

4 42 5 37 17 25
5 50 8 12 7 38 20 3 30
6 58 8 20 8 38 24 4 34
7 66 8 29 9 37 29 5 37
8 74 8 39 10 35 35 6 39
9 82 8 50 11 32 42 7 40
10 90 8 62 12 28 50 8 40
In this example, the marginal benefit does not depend on the quantity produced. As a result, the
price control leads to no loss at all in the error scenario. In contrast, a quantity control leads to a loss of 6.

however. In this particular example, the loss on account of the mistake is 1 5 units,
a much larger loss than for the quantity-directed system.
In this example, the quantity system works better, but this is not always the
case. Compare the example in Table 4. 1 with that in Table 4. 2 , in which the costs
are the same but the benefit levels differ. The marginal benefit in Table 4. 2 is constant
at 8. In both scenarios, if the price is specified to be 8, the quantity that maximizes
the firm's profits is also the quantity that is socially efficient. Therefore, the use of
prices to coordinate behavior avoids any loss on account of imperfect cost estimates.
I f instead the planner had fixed the quantity at 6, the best level based on the estimated
costs, and if the actual cost function were that shown in the error scenario in the
table, then 6 units of net benefit \vould be lost. (Six units give a net benefi t of 34,
whereas the actually efficient choice of 1 0 yields a net benefit of 40. ) In the second
example, the approach of fixing prices performs better than fixing quantities.
The flat marginal benefi t curve is one extreme. At the other is a marginal benefit
schedule that is vertical at the relevant point, as in Table 4. 3.
In this example the right quantity is 6 units, independent of which cost scenario
prevails. The quantity system achieves the optimum, with a net benefit of 36. A price
system based on the incorrect estimate would involve setting the price at 8, the level
of (incorrectly) estimated marginal cost at the efficient quantity. However, facing a
price of 8 and the costs given in the error scenario, the firm will select a quantity of
9 or I O units, both of which give it profits of 30 (9 X 8 - 4 2 and 1 0 X 8 - 50). The loss
from using prices is either 1 8 = (36 - 1 8) or 26 = (36 - 1 0).
The third example recalls our discussion of ambulances. It makes little sense to
call out a price and see how many ambulances arrive because the number of
ambulances that will show up is too dependent on irrelevant information. If you know
that you want one ambulance, you should order one ambulance. In the second
example, because the marginal value of each unit is 8, the correct price will be 8
regardless of the cost function. if you know that the marginal value of output will be
8, then the best form of control is to fix the price at 8. These extreme cases are
informative, but we still want to know: What general principle guides the choice in
intermediate cases, such as that in Table 4. 1 ?

A l\ lathernatical Formulation and Analysis

To provide a more general view of when prices work relatively well, we switch to an
algebraic formulation of the problem. As in our precedin g examples, we study the
 97
Table 4. 3 Coordinating Plans
and Actions
Planner's E rror
Cost Estimate Scenario
Number Total Marginal Total M arginal N et Total M arginal N et
of Units Benefits Benefit Costs Cost Benefi t Costs Cost B enefi t

4 40 5 35 17 23
5 50 10 12 7 38 20 3 30
6 60 10 20 8 40 24 4 36
7 60 0 29 9 31 29 5 31
8 60 0 39 10 21 35 6 25
9, 60 0 50 11 10 42 7 18
10 60 0 62 12 -2 50 8 10
In this example, the marginal benefit drops from 1 0 to O at 6 units. The optimal quantity is 6 for
either cost scenario. There is no loss if a quantity system is used. In constrast, a price system leads to a
loss of 26.

case where the marginal benefits and marginal c osts are linear functions of the output
quantity. U nlike those examples, however, we do not restrict outputs to integer
amounts. Also, as in the examples, the slopes of the linear functions are assumed to
be known, but the decision maker is unsure about the intercept of the c ost function,
so that all the marginal costs might turn out to be higher or lower than antici pated
by some fixed (but unknown) amount. Suppose the losses associated with the systems
of fixed prices and fixed quantities are measured in the same way as in the preceding
examples, as the difference in the net benefi ts achieved under the particular system
and the net benefi ts that would be achieved with correct information. In that c ase, it
turns out that the ratio of the losses incurred depends on the slopes of the marginal
benefi t and marginal cost fu nctions, as follows:

Loss from a price control Slope of margical benefi t

2
= [ ] (4. l )
L oss from a quantity c ontrol Slope of marginal cost
Acc ording to Equation 4 .1, a price- based system of coordination leads to smaller
losses than does a quantity system when the slope of the marginal benefi t function is
less than the slope of the marginal c ost function, and the reverse ranking holds when
the slopes are reversed. Thi s qualitative conclusion matches those we had obtained
in the discrete examples. (It does not quite match quantitatively because outputs were
restricted to integer amounts in the examples. )
I n the theory of c ompetitive markets where fi rms base their quantity decisi ons
on market determined prices and their own costs, the corresponding ratio is the slope
of the firm's demand functi on to the slope of its marginal cost function. From the
competitive firm's point of view, demand is infinitely elastic; that is, the slope of the
demand function is zero. Consequently, according to Equation 4. 1, its choices are
best guided by the system of prices: Efficiency cannot be improved by regulating a
competitive firm.

Derivation of the Formula

Figure 4.1 shows how the formula in Equation 4. 1 is derived. As usual, the
horizontal axis i ndicates quantities of output and the vertical axis indicates prices
(money per unit of quantity). The downward sloping line (MB) is the marginal
 98
Coordination: benefit curve, whereas the two upward sloping lines correspond to the marginal
Markets and cost curve as estimated by the planner (MC) and in the error scenario (MC').
Management Total benefit corresponding to any output is then the area under the MB curve,
and the total cost in either scenario is the area under the relevant marginal cost
curve. The marginal benefit and estimated marginal cost curves intersect at the
point a = (Q, P), where Q is the efficient quantity and P the corresponding
price if the planner's estimates are correct.
In the error scenario, the actual marginal cost is less than the planner's estimate
by the amount d. The actual efficient quantity is then Q' . If the planner specifies
a quantity of Q, what will be lost? The area of the trapezoid QacQ' in Figure
4. 1 represents the benefit of increasing output from the level Q specified by the
planner to the optimal level Q', while the area of QbcQ' represents the cost of
the extra output. The difference, represented by the area of the triangle abc, is
the net loss of welfare on account of the planner's mistaken quantity choice.
Using the formula for the area of triangles, the welfare loss is equal to ½
d(Q' - Q) .
For comparison, suppose the planner instead specifies a price of P and the firm
responds by producing the profit-maximizing quantity Q". Compared to the
optimal quantity choice Q', choosing Q" brings extra benefits represented by
the area of the trapezoid Q' ceQ" and extra costs represented by Q' c{Q". The
extra costs exceed the extra benefits by an amount equal to the area of the
triangle efc. Calculating as before, the area is ½ D(Q" - Q' ) .
Notice that the triangles abc and efc are similar triangles. (The marked angles
at b and f arc alternate interior angles and are therefore congruent. The same
applies to the alternate interior angles at a and e.) It then follows that (Q" - Q')/
(Q' - Q) = D/d: The altitudes of the triangles from vertex c are in proportion
to the bases. Also, letting SMB denote the magnitude of the slope of the MB
curve (the curve is negatively sloped) and SMC that of the MC curve, the
diagram reveals that d = SMC X (Q" - Q) and D = SMB X (Q" - Q). Hence,
Did = SMBISMC. Taking the ratio of the areas of the two shaded triangles
leads to:
2
½D(Q" - Q') = D X
(Q" - Q') = [
Slope of marginal benefit ]
(4 _ 2)
½d(Q" - Q ) d (Q" - Q ) Slope of marginal cost
which establishes E quation 4. 1 .

Figure 4. 1: The left-hand and right-hand

shaded triangles show the losses from quantity
and price controls in the error scenario. The
Q Q' ratio of the areas is given by Equation 4. 1 .
 99
Table 4.4 Coordinating Plans
and Actions
Planner's Error
Cost Estimate Scenario
Num ber Total Marginal Total Marginal Net Total Marginal N et
of U nits Benefits Benefit Costs Costs Benefit Costs Costs Benefit

4 40 32 8 28 12
5 50 10 40 8 10 35 7 15
6 58 8 48 8 10 42 7 16
7 64 6 56 8 8 49 7 15
8 68 4 64 8 4 56 7 12
9 70 2 72 8 -2 63 7 7
10 70 0 80 8 - 10 70 7 0
With constant returns to scale, prices alone cannot guide the output decision.

Constant and I ncreasing Returns to Scale

Our discussi on so far has assumed that the marginal cost of producing increases with
the total number of units being made. Treating this as the long-run marginal cost,
the assumption is that there are decreasing returns to scale. I n such circumstances, if
demand is large, then it is generally optimal to divide production among a number
of units to keep total costs relatively low. The price system has certain extra advantages
in that setting because the optimal price depends only on the aggregate or average of
the producers' supply curves, while efficient quantity planni ng would require the
planner to know the costs of each producer indivi dually. N ot surpri si ngly, the
advantages of decentralization in economizing on information gathering and communi­
cations are greatest when efficient production requires that many producers be
coordinated. We return to study how the need for communications varies among
systems in the next section.
First, however, let us consider the important situation i n which individual
producers enj oy increasing returns to scale and decreasing average costs. In this case,
it is ineffi cient to divide production among many small firms. Fewer firms could
produce the same output at a lower total cost. M oreover, as we saw in Chapter 3, the
criterion of maximizing profits at given prices can never be sufficient in that case to
stimulate the firms to produce optimal quantities. Wi th given prices, revenues grow
in proportion to output but, with increasing returns, costs per unit fall as output
increases. Thus, if the given price ever exceeds average cost, the firm would seek to
increase output without bound because each additional unit of output increases total
profit. The price system in these circumstances fails to meet our first cri terion. It
could not support an efficient plan even if there were no chance of any error in the
planner's estimate of the firm's cost functi on.
The dividing li ne between the cases of increasing and decreasing returns to scale
is the case of constant returns to scale, where the slope of the marginal cost function
is zero and marginal cost equals average cost at all output levels. Applying Equation
4. 1 to the case where the slope is nearly zero, the loss fr om a fixed- price choice is
much greater than the loss from a system that fixes quantities. Our next example,
presented in Table 4. 4, makes the reason clear.
S etting a correct price of 8 in this example provides no guidance whatever to a
profit-maximizing producer. Every level of output leads to the same level of producer
profits (zero). If the price is set incorrectly, however, the results are dramatic. If the
 100
Coordination: price is set to 8 when the marginal cost of production is 7 (the error scenario), then
Markets and a profit-maximizing producer would expand output to the limit of its capacity.
Management Presumably, it will also be delighted to expand its capacity. Similarly, if the price is
8 but the marginal cost is 9, the firm will drop its output to zero because it loses
money on each unit produced. With constant returns to scale, the producer's responses
to small price changes are too extreme for prices to be an effective instrument of
controlling production.
This conclusion points out one of the most important limitations of the
fu ndamental theorem of welfare economics. The theorem asserts that a competitive
equilibrium allocation is efficient. That is, if the prevailing prices and allocation are
such that supply is equal to demand for every good at the given prices, then the
allocation is effi cient. The problem is that, as we saw in Chapter 3, with increasing
returns to scale, there may be no prices at which supply is equal to demand.
Furthermore, with ronstant returns to scale, the price that equates the quantities
supplied and demanded gi ves the producer no guidance about the appropriate level
of production; every production level leads to an equal level of profits. As applied to
Table 4. 4, the theorem merely asserts that if the quantity produced at a price of 8
happens to be the one where supply equals demand (in this case, 6 units), then the
outcome is an effi cient one.
If several suppliers enj oy constant returns to scale over the relevant range of
production levels, then an efficient plan demands that the entire order be produced
by the one supplier with the lowest total cost and that the quantity ordered be adj usted
so that the marginal benefit associated with the last unit is equal to that supplier's
marginal cost. One way to accomplish this feat of coordination is for the suppliers to
engage in competitive bidding, where each bid reflects one supplier's cost and the
buyer purchases as many units as desired at the lowest quoted price. This helps to
explain the common business practices of competitive bidding and requirements
contracting (in which the seller agrees to supply as many units as the buyer may
require at the quoted price) and suggests that these practices ought to be most common
for goods produced with constant or increasing returns to scale.
Competitive bidding, of course, is a kind of market organization in which firms
set prices. I n the short run, the prices bid by suppliers guide the buyer's demand. In
the long run, the prices paid by buyers guide the suppliers' capacity decisions.
Therefore, even when constant returns to scale exist, prices still have a role to play
in determining the allocation of resources.

ECONOM IZING ON INFORMATION AND COMMUNICATION

So far in this chapter we have ignored the important costs of gathering, organizing,
storing, analyzing, and communicating information in a form that is usefu l for
decision making. We now explore the extent and limitations of the idea, as suggested
by H ayek in the quotation at the beginning of Chapter 3, that a system of prices is a
particularly good way to economize on communication and information processing.

The Informational Requirements of Production Planning

The basic intuition regarding the informational effi ciency of a price system is perhaps
best illustrated by studying the problem of minimizing the cost of producing a given
amount of total output in a firm with several production facilities or in an economy
as a whole. Using a system in which the planned allocation is centrally determined
and then implemented through quantity controls requires that the planner have huge
a mounts of detailed information a bout each individual production unit. Even finding
a plan that is techn ologica lly feasible requires the center to know a great deal about
 101
the capabilities of each un it, so that it is not asked to do the impossible. Then Coordinating Plans
determining how much each facility should effi cien tly produce out of a given desired and Actions
total output requires kn owing the m arginal costs in each facility. S ince this inform ation
is not likely to be kn own in itially by the cen ter, it must be communicated from the
individual units. The massive plann in g bureaucracies that developed in the Sovi et
Union an d other communist coun tries are proof of j ust how costly it is to gather and
process that much information.
In contrast, a price-guided, decen tralized system does not req uire communication
of such large amoun ts of information. Individual production units will respond to
given prices on the basis of their local kn owledge of their techn ological capabilities
and their costs, each producing where the common price eq uals its marginal cost.
This mean s the total output is produced at min imum total cost.
· This treatmen t, however, ign ores the problem of finding the right prices that
call forth the desired total output. As we n oted earlier, determining the right prices
requires that the price setter kn ow on ly the total supply function, and this would seem
to be less demanding than kn owing the total production possibilities for each individual
factory. But there is still a question of how the cen ter would come to have this
information. I n a market system, we coun t on the forces of supply and demand to
adj ust prices. An anal og in the planning con text would be to think about an iterative
procedure. The cen ter would ann ounce a candidate price, the individual units would
respond with ten tative production plans, and then the cen ter would adj ust the price
up or down depending on whether too little or too much output was forthcoming.
Then what needs to be communicated at each roun d is the sin gle price and the output
levels, one per plan t.
An analogous scheme could be considered for quan tity plannin g. S uppose the
cen ter ann ounced a tentative output level to each producer, each responded by stating
its marginal cost at this output level, and then the quan tities were adj usted up for
those plan ts ann ouncing low marginal costs an d down for those whose ann ounced
marginal costs were high. If the individual plan ts all ann ounced the same marginal
cost level, then the corresponding division of production among them would be
efficient. Note, however, that each roun d n ow requires more commun ication than
under the price- based system: an output level for each plan t an d its marginal cost. I f
there are N plants, price- based plannin g in volves commun icating N + 1 n umbers: the
price from the center to the plan ts, an d the N plan ts' outputs back to the cen ter. The
quan tity system requires commun icating 2N n umbers. Even in this simple context,
the price system requires comparatively less communication of information .

Judging Informational Efficiency

To make a formal assessmen t of the relative communication requiremen ts of any
particular system, we use the idea in the last paragraph of comparin g the sizes of the
messages used. H owever, since we wan t to compare more than just the sort of iterative
procedures for plannin g production that we discussed earlier, we need to take a little
differen t approach than used there.
THE I I URWICZ CRITERION The one widely applied approach to comparin g the
informational req uirements of differen t systems is due to Leon id H urwicz. The key
idea is to con sider how much information it takes to determine whether a particular
plan is efficien t. We then think of the planning system as based on broadcasts of
augmen ted plan s to producers and con sumers. An augmen ted plan con sists of the
plan itself-input and output levels of each good for each producer and amoun ts of
each good received or supplied by each consumer-plus possibly additional information
used to check the effi ciency of the plan. B y "broadcasts, " we mean that any information
 102
Coordination: that is commu nicated is made available to everyo ne. U po n receiving the broadcast,
Markets and each individual producer or co nsumer evaluates the plan using local information and
Management then replies with a message, which we may take to be a "Yes" or a "No. " In terms of
the iterative procedures discussed previously, it is as if the center were annou ncing
both the prices and quantities, with the firms responding "Yes" if their marginal costs
at the annou nced output equalled the price. The whole system must be co nstructed
so that if everyo ne replies with "Yes," then the resulting plan is an efficient one.
In this framework, the Hurwicz criterion holds that one system operates with
less commu nication than another if the first broadcasts fewer additional variables
(besides the plan itself). A system is then informationally efficient if no other system
uses less extra i nformatio n than it does to verify that a given plan is efficient.
This criterio n is an imperfect measure of informatio n used by the system. Its
most significant drawback is that it does not accou nt for how quickly different systems
find an effi cient allocatio n or for how much informatio n they communicate in the
process. Instead, it focu ses only on the amou nt of information that is used to check
whether a proposed resource allocation is effi cient. So far, however, the Hurwicz
criterio n is the o nly measure of commu nicatio n requirements which has been
extensively and successfully analyzed.
THE INFORMATIONAL EFFICIENCY OF THE PRICE SYSTEM Despite its drawbacks, the
Hurwicz criterion does allow us to capture one aspect of the idea that the price system
pl aces particularly light demands on commu nication. The following theorem, which
is due to Hurwicz, gives the minimum amou nt of information that mu st in general
be commu nicated in addition to the plan to permit verificatio n of the plan's efficiency.
This amou nt is what is commu nicated in the price system.
Informational Efficiency Theorem: Suppose that there is no a priori
informatio n about the optimal resource allocatio n, so that given what any
single producer or consumer knows, any allocatio n of society's limited
resources might still be efficient. Su ppose too that each producer is
u niquely well informed about its own productive capabili ties and each
co nsumer alo ne knows his or her own preferences and what amou nts of
various goods he or she initially owns, so that no single agent alone has
the informatio n needed to compute an effi cient allocation of resources.
Then any system capable of supporting an effi cient resource allocatio n
using augmented plans must commu nicate, in addition to the plan, at
least one additional variable for each separate good or resource, minus
one.
In particular, when a competitive equ ilibrium exists, the price system, which
commu nicates exactly one additio nal variable (the price) for each good or resource
after the first, 7 achieves economic efficiency with minimal commu nicatio n. The price
system is then informationally efficient by the H urwicz criterion.
A:-,.i I NTUITION FOR THE INFOR!\IATIONAL EFFIC I ENCY Tl lEOREI\I To u nderstand why this
result is true, co nsider the problem of verifying whether a single producer's allocatio n
is efficient. Producers cannot tell on the basis of their local knowledge and the
proposed plan alone whether their own part of the plan is efficient becau se they do
not know how valuable the resources might be in other uses. Because the producers
are un iquely well informed about their own capabilities, no body else can tell if the
plan is even technologically feasible for the producer, let alo ne whether it is consistent

7 The first good is taken to be the numeraire. The prices of all other goods can then be expressed
in units of the first good. llistorically, the numeraire good was often gold or silver.
with overall efficiency. Therefore, the broadcast message m ust convey enough Coordinating Plans
i nformati on to the local producers to allow them to verify the optimali ty of thei r and Actions
i ndividual parts of the plan.
I n order for the producer's speci fied part of the plan to be consistent with overall
efficiency, the margi nal rate at which th e producer transforms any particular kind of

faci lity i n the econom y that uses thi s i nput i n produci ng the same good. (If, for
input i nto any particular kind of good m ust be the same as at any other producti on

example, some other factory had a lower margi nal rate of transformation, then the
same output could be achieved wi th fewer resources by increasi ng slightly the
producti on of thi s factory and reduci ng the producti on of the other by an equal

know what these econom y-wide margi nal rates of transform ati on are. If there is only
amount.) In order to determi ne whether these condi ti ons hold, the producer m ust

each product. If there are many ki nds of i nputs, then there m ust be at least one
one kind of i nput, then there is j ust one relevant margi nal rate of transformati on for

addi ti onal margi nal rate of transformation for each additi onal ki nd.
O ne num ber per good produced and one num ber for each i nput after the first
then define a lower bound for the am ount of communicati on that is needed to verify
the effi ciency of any proposed plan. As we saw earlier in thi s chapter and in the
fu ndamental theorem of welfare economics i n Chapter 3, thi s am ount of comm unica­
ti on is sufficient as well when the num bers are (relative) prices. In that case, each
producer looks at the pr oducti on levels speci fied for it, determi nes whether its plan

If everyone responds "Yes" and if the plans are chosen so that the supply equals
maximizes its profits at the gi ven prices, and if i t does, then it responds with a "Yes."

demand for every kind of good, then the plans are effi cient by the fu ndamental
theorem of welfare economics.
APPLYING THE THEOREM Of course, we do not normally thi nk of the price system
i n term s of a broadcast plan. H owever, suppose each consumer and producer has a
uniquely best choice at the competitive equilibrium prices. Then havi ng prices
"announced" and havi ng each i ndi vi dual agent "respond" with the correspondi ng

identi fied by the theorem. If the announced prices are the com peti ti ve equi librium
am ounts that he or she wants to buy or sell i nvolves j ust the amount of comm unicati on

ones, the quanti ties balance and an effi cient allocati on results. U nder the usual
concepti on of how a com petitive m arket system works, if the i nitial prices are not the
competi tive equili brium ones, then the quantity responses wi ll not balance. With
supply not equal to demand, there will be pressures for prices to change. The new
prices call forth adjusted quanti ties as people respond to the new rates at which they
can buy and sell, and the system gropes its way along, perhaps ultim ately reachi ng
some equilibri um.
M ore often, however, this fram ework and the theorem are applied to explici t
production planni ng problem s, either within firm s or at an i ndustry or economy-wi de
level. One of these arises in consideri ng si tuati ons where the coordi nati on problem
has desi gn attributes.

Planning with Design Attributes

Notice that the hypotheses of the i nformati onal effici ency theorem rule out problem s
i n which there are design attri butes. For deci si ons with design attributes, there is a
priori i nformati on about the nature of any effi cient choice, and it may be possi ble to
verify th e optimality or near- optimality of th e plan with less i nform ati on than that
communicated by a system of prices. In our exam ple of the coxswai n guidi ng a crew
of rowers, it was known in advance that synchronizati on of the rowers' strokes would
characterize any optimal soluti on and that fai lures of synchronizati on would be very
 104
Coordination: costly to the crew. The coxswain system uses much less communication than does a
Markets and system of prices, and its failures when the coxswain's judgment is imperfect are much
Management less costly.
Problems in which synchronization is important arise frequently in business.
For example, if an automobile company is designing a new car, all the parts must be
designed and production facilities made ready by the target date for introducing the
product. To synchronize, the new product team agrees on a product introduction date
and communicates that date to other relevant parties, rather than marginal values for
early completion.
It is not only timing that is communicated in this way in product development.
When a new car is being developed, the engine must be designed to pull a car of a
particular weight, the brakes must be designed to stop a car of that weight, and the
chassis must be designed to carry that weight safely. Similar statements can be made
about other attributes of the car, such as its physical dimensions. The development
team coordinates its activities by agreeing on goals and objectives and on how to
realize these. The language the team uses is not generally the language of prices.

USING PRICES IN DESIGN DECISIONS Let us emphasize that the avoidance of prices
for coordinating activities in synchronization problems and other sorts of design
problems is not due to any theoretical impossibility of applying a price system. Rather,
it is due to ( 1 ) the unreasonable information demands that the price system makes in
these kinds of problems and (2) the brittleness of the system, that is, the high cost of
asynchronous behavior. To illuminate the information demands, let us consider once
more how a price system would operate for a synchronization problem.
For simplicity, suppose that there are ten suppliers who must coordinate the
timing of their supply activities, and five possible completion dates. Recall that goods
are described not only by their physical characteristics but also by their time and place
of delivery. This is a crucial point for understanding how markets and other
organizations work; skis rented for two weeks in the winter are quite a different thing
from skis rented for two weeks in the summer, and we should not be surprised to see
sharp differences in the rental prices of the two. Applying that perspective to the
synchronization problem, we find that each of the ten suppliers can supply any one
of five different goods, according to whether the physical product is delivered at dates
one, two, three, four, or five. Therefore, there are 50 inputs in all (five for each of
the ten producers), and a price system would require a price for each of them-50
separate prices.
If the number of suppliers, possible delivery dates, or component designs are
large, then the number of prices that need to be specified is also large. Nevertheless,
the fundamental theorem of welfare economics does apply. If all the prices are set so
that the coordinator would want to buy one unit of one design of each input at some
particular date T and if, given the prices, each supplier finds it most profitable to
deliver one unit of the corresponding design at date T, so that markets clear, then the
allocation (the list of dates at which supplies of each type of component become
available) is guaranteed to be efficient.

A BETTER WAY Determining 50 prices to solve this synchronization problem is

unnecessary and wasteful. All the coordinator actually needs to know to check whether
a proposed introduction date is optimal is whether the total marginal cost of introducing
the product a bit earlier-taking into account the extra costs incurred by each
component supplier-is equal to the marginal benefit of doing so. To check this, it
suffices for the coordinator to know the marginal cost of a speed-up for each supplier,
which is just ten numbers. The greater the number of possible delivery dates, the
 1 05
Coordinating Plans
and Actions

Optimal
Figure 4.2: The optimal date of product intro­
Date
duction is the date at which the marginal benefit
of faster introduction is equal to the sum of the
· Date of I ntrod uction marginal costs incurred by the two supply units.

greater the discrepancy between the minimum amount of necessary communication

and the amount required by the price system.
Figure 4. 2 illustrates an extreme case, for which there are only two suppliers
but the product might be introduced on any of a continuum of possible dates. The
downward sloping curves show the marginal cost to each component supplier of
speeding up the product introduction by a small amount, given any particular target
introduction date. The downward slant means that speeding up the i ntroduction grows
increasingly costly for each supplier as the planned introduction date is moved up.
The total marginal cost of a speed-up for the two suppliers is determined by adding
the costs for the indivi dual units. The upward sloping curve in the figure is the
marginal benefit curve. Its upward slope means that longer delays grow increasingly
costly for the firm. The optimal introduction date is determined at the point where
the marginal cost of additional speed-ups to the two suppliers is equal to the marginal
benefit.
In this problem, to verify whether any particular list of prices leads to optimal
choices, the planner would have to know what costs each supplier would incur for
each possible delivery date-an infinite list of information-in order to tell which
date a profit-maximizing supplier will pick. To verify the actual optimal date, however,
all the coordinator needs to know is the marginal benefit of a speed-up and the two
suppliers' marginal costs: three numbers in all. Taking advantage of special knowledge
about the problem, the coordinator can drastically reduce the amount of communica­
tion required.
What about the cost of mistakes? In many real synchronization and product
design problems, the most costly sorts of errors are failures of synchronization (as
when the late availability of one component delays a large project) or fit (as when one
component's incorrect tolerances cause the product to fail), and these are the ones
most to be avoided. In contrast, as long as the parts fit together, small variations in
the design itself are less crucial. Thus, it matters less whether the drain hole in a car's
oil pan is 2 cm or 2. 5 cm than that the hole and the plug are both the same size.
Because our definition of design attributes includes two characteristics­
predictable elements of fit and a high cost of small errors of fit-we may conclude
that the price system performs poorly for design problems on both the communication
and the brittleness criteria. The usefulness of this general conclusion, however,
depends on an assessment of how often real problems have both of the characteristics
that we have included in our defini tion of design attributes.
If prices should not be used to coordinate design decisions, how should
 106
Coordination: coordination be achieved? The answer is that the design variables themselves should
Markets and be communicated. Each rower in the crew must know the intended stroke rate and
Management the timing needs to be communicated to the rower. Each member of the product
introduction team must know the introduction date and the weight and other key
design parameters of the new car. The ambulance driver must be told which crisis to
attend, when, and where. This form of control minimizes communication by the
Hurwicz criterion and reduces the cost of error associated with more indirect methods.
The appendix to this chapter contains a more formal treatment of design decisions,
establishing that communicating the design variables is informationally efficient for
the class of design decisions, just as communicating prices is informationally efficient
for the class of resource allocation problems treated in the informational efficiency
theorem.

COORDINATIO!\ AND BUSINESS STRATEGY

Strategic business decisions present complicated problems. Good strategic decision
making virtually always requires that effective use be made of line managers' knowledge
about how the operations actually work and what capabilities the business has, but
they may also involve using knowledge about new technologies, new markets, new
business partners, or new forms of organization about which the line managers'
knowledge may be limited. In our lexicon, strategic decisions often have innovation
attributes. In addition, especially in manufacturing industries, but in some service
industries as well, there are important economies of scale that mitigate against
completely decentralized decision making. Finally, as we argue later, business strategy
decisions commonly have important design attributes. There are predictable elements
of fit in any good strategy that make it important to coordinate the actions of various
parts of the organization closely. All these factors work against using prices or other
very decentralized means of coordination and favor direct communication and other
more systematic, centralized control systems.

Scale, Scope, and Core Com petencies of the Firm

As we have already seen, when there are scale economies, the efficient level of output
in a firm cannot be determined by prices alone. Operational scale itself is a design
variable. Depending on the volume of sales that a firm anticipates, it will adjust its
production capacity and the size of its sales force and secure supplies and distribution
equipment and facilities (such as trucks and warehouses)--all tailored to the expected
scale of its operations. If the actions taken by the marketing, production, personnel,
distribution, and procurement managers are to be coherent, then all these people
need a shared vision of the intended scale of operations.

SCALE A'.'JD STRl 'CTrnE The anticipated scale of a firm's operations predictably affects
more than just the scale of each part. As the GM and Toyota examples from Chapter
1 illustrate, it also determines the degree of specialization the firm should adopt. With
larger operations, a firm may be able to afford more specialized equipment, more
distribution outlets located nearer to customers, a larger number of plants, training
programs for its employees tailored to particular circumstances, and so on. A smaller
firm, operating without specialized equipment, may be more likely to rely on suppliers
for many more of its components because the suppliers may be better positioned to
enjoy economies of scale of their own by serving many firms. Thus, a larger firm
with more specialized capital equipment may find it profitable to be more vertically
integrated than would its smaller competitors. By definition, economies of scale in
production allow a firm to reduce its costs compared to small-scale production, and
 107
these costs are an importan t elemen t in determ in ing the prices to be charged. Lower Coord inating Plans
marginal costs, other th in gs equal, lead th e firm to charge lower prices, which and Actions
increases the potential product deman d, which in turn supports an increased scale of
production.
Operational scale is a design variable because it meets the two condition s. I t
has predictable impl ications for the various parts of the organization, and man y of
the mistakes associated with incorrect percepti ons of scale by parts of the organization­
for example, havi ng too few raw materials or components to keep an expensive factory
operating at full capacity-can be very high . Thus, firms that are large enough to
assign different management functions to different decision makers take special pains
to forecast market growth, competitors' plans, techni cal changes, i nput availabilities,
and so on, all so that they can use those forecasts to plan the growth (scale) of their
own operations and coordi nate based on these plans. By making sure that its managers
share common expectations about what i t is trying to do, the firm takes an important
step toward �oordinating their plans and behavior.
EcONO:\II ES OF SCOPE E ven when a firm operates at too small a scale in any i ndividual
product market to enj oy si gnificant economies of scale, it may still enj oy them in
producing components that are used in each of several products. For example, a firm
like G eneral Electri c may enj oy economies of scale in producing small electric motors,
using those motors to make food processors, hair dryers, fans, vacuum cleaners, and
various other products. A firm like Casio may enj oy economies of scale i n the
manufacture of liquid crystal displays (LCDs), using them to produce calculators,
wristwatches, electr onic address books, and other products.
In these sorts of circumstances, the firms are sai d to enj oy economies of scope;
that is, they can produce their several products together at less cost than could a group
of single- product firms. Unsurprisingly, economies of scope entail all the same needs
for coordination that economies of scale do. The problems are often harder, however,
because coordination in planning is required among the managers responsible for
different products. As Casio grew, for example, a forecast of large sales in the market
for calculators led to falling costs for LCDs, maki ng it more pr ofitable to enter the
market for wri stwatches that use LCDs.
CORE COMPETENCIES When a firm introduces new products relatively frequently,
one very important kind of scale economy that it may enj oy is at the level of product
development. That i s, a firm may acquire generalized expertise in the important skills
that are required to design and market new products in a set of related markets or in
usi ng a set of related technologies. For example, a computer maker may develop
expertise in micropr ocessor design, display technologies, memory chips, operating
systems, computer manufacturing, data communications, networki ng, and so on­
skills that it expects to be able to apply over and over again as i t continues to introduce
new products. Scale economies at this level are so important in modern management
theory that a new name has been coined for them: core competencies of the firm. In
a dynamic environment, a firm' s capacity to introduce new products and to manufacture
them effi ciently can be even more important than are the economies of scale it
achieves in making its existing product line. In that setting, a strategy of developing
scale economies translates into one of bui lding the core competencies of a firm.
In an abstract sense, core competencies are j ust another kind of shared
component, but there is an important practical difference in that the cost of building
the competency is shared with a series of products that does not yet exist. Investments
in new manufacturing technologi es today may actually raise the costs of today' s
products if the old manufacturing system is well understood and well implemented.
The gains to be enj oyed on account of the new system will come over a longer period,
 108
Coordination: as the firm learns to use the new system, refines its methods, and adjusts the rest of
Markets and its operations to take full advantage of the capabilities of the new technology. In such
Management a case, managers need to plan for the demands of generations of products not yet even
imagined. They also need to keep the price of the current products low enough to
sustain a strong volume of sales, even if the company appears to be losing money on
each sale, recognizing that these losses are actually an investment in the capabilities
needed to produce profitable products in the future. This is just another illustration
that when there are economies of scale, individual production and pricing decisions
cannot be evaluated in isolation.
More controversially, the same ideas may be applied at the level of national
industrial planning-a process about which many economists are skeptical but which
has been used successfully in Japan and South Korea. The key first step in successful
planning is to identify (groups of) industries to promote, which "fit" together with
each other and with the country's existing competencies and advantages. For example,
in Japan, the push to develop high-definition television (HDTV) in the 1 990s is firmly
grounded in that nation's strong positions in semiconductors, consumer electronics,
and display technologies. To coordinate Japan's drive into HDTV, the Japanese
national broadcasting company, NHK, has announced a set of technological standards
that serves both to focus cooperative development efforts within groups of Japanese
firms and to intensify competition among groups.

Complementarities and Design Decisions

Complementarities among a set of activities are an important source of design
attributes. The standard definition of complementarity in economics is market oriented.
Two inputs to a production process are said to be complements if a decrease in the
price of one causes an increase in the demand for the other. In order to be able to
employ the concept of complementarity usefully to study choices of levels of various
internal activities as well as levels of input purchases, we introduce an alternative,
more inclusive, definition: Several activities are mutually complementary if doing
more of any one activity increases (or at least does not decrease) the marginal
profitability of each other activity in the group. 8
For example, where there are declining marginal costs due to learning or other
kinds of economies of scale in producing a component, then the activities of producing
various products using those components are complements. If General Electric's
marginal cost of producing small electric motors declines with increasing volume,
then the activities of producing electric fans and food processors are complementary
because producing more fans makes it cheaper and therefore more profitable to
produce more food processors as well.
Complementarities lead to predictable relationships among activities. A decision
to increase the level of one activity will raise the profitability of any contemplated
increases in levels of any complementary activities. Thus, high levels for all the
elements of a group of complementary activities go together. This predictability is one

8 In mathematical terms, the complementarity relationship among a group of activities can be

characterized as follows. Let x = (x1, . . . ,x.) be the levels at which the activities are conducted and let
,r(x) be the resulting profits. If ,r is a smooth function, then the activities are mutual complements if for
all i,t,;, a 2,r/ax,ax, � O; an increase in the ;i1i activity raises the marginal return to the ith activity.
When a change occurs that makes any one of the complementary activities more profitable or less
costly and encourages the firm to do more of that activity, then the marginal returns to the other activities
are also increased, leading to more of those activities as well. The increased levels of the complementary
activities further increase the marginal returns to the first activity, possibly leading to another round of
increases in it and all the related activities.
 1 09
of the two defi ning features of design attributes. The second feature is the high cost Coordinating Plans
of failure to match or fit the parts together. This feature is a separate aspect of th e and Actions
problem that is not logi call y implied by the conditi ons of complementarity alone.
H owever, we say that a group of activities is strongly complementary when raisi ng the
levels of a subset of activities in the group greatly i ncreases the returns to raising the
levels of the other activities. When a group of acti viti es is strongly complementary,
design attributes are always present.
The box entitled "M odern M anufacturing Strategy" describes one particular
business strategy that illustrates the complementarities between particular aspects of
product strategy, manufacturing policy, equipment choice, pers onnel and compensa­
tion policies, supplier relations, accounting methods, and more. What is most
important to notice about the strategy is the coherence of i ts parts. The parts of the
strategy that are connected are mutually supporting because they call for complementary
activities. Part of the success of the modern manufacturing strategy i n the 1 980s and
1 990s grows out of the way it takes advantage of the new technologies of the time,
such as rapid, low-cost communications, highly flexi ble equipment, and computer­
aided design, which make many of the elements of the strategy more effective and
less costly.
Other manufacturing companies in other times have succeeded using other
strategies, als o well fi tted to the available technology. In the fi rst part of the twentieth
century, for example, Ford M otor Company adopted an entirely different but equally
coherent strategy that was appropriate for its time, based on producing a single product
(the M odel T Ford) in high-volume fa ctories using specialized equipment and rigidly
discipli ned labor paid accordi ng to the j ob bei ng done. This strategy reduced cos ts
and i ncreased the quality of the automobile so much that Ford was eventually able
to capture more than 50 percent of the automobile market.
The continuing changes in technology and factory pri ces i n the modern era,
includi ng the falling costs and growi ng reliability of highly automated equipment that
works with no direct labor involvement, are likely to lead to further revisions in
manufacturi ng firms' strategi es.

Modern Manufacturing Strategy

O ne kind of strategy that has taken on a new i mportance in the la te 1 980s
and 1 990s is the "modern manufacturi ng" strategy, which is actually a whole
group of similar strategies. The particular vari ant we describe involves
produci ng a wide range of related products for customers with specialized
needs. The fi rm strives to remain the quality leader in i ts markets with
frequent new product introductions and frequent improvements of existing
products.
Market behavi or of this kind has implications for a wide range of
acti vities i n the organizati on. At the level of manufacturing, the small quanti­
ties demanded of each product prevent the company from setting up
specialized production lines for each, forcing it to mai ntain a high level of
flexibility. B ecause the demand for each product is small and because there
are important economies of scale in inventory systems, the company will fi nd
it profi table to organize in a way that avoids holding i nventories of fi nished
products. I t will eschew keeping local product warehouses and i nstead may
 1 10
Coordination:
l\ larkets an d
l\ lanagement
even ship directly from the factory to the customers, perhaps using air freight
if speed is important. In order to serve customers quickly, communications
behveen the sales staff and the factory need to be closer than for a company
that relies on finished goods inventories. Strong communications are always
a key variable when products are to be completed "just in time" to meet
customer demands. Since the average inventory level for a product is directly
proportional to the production run, the company will want to use more
frequent, smaller production runs to make its products. The use of frequent,
small production runs forces the company to reset its production line more
frequently to change the product being made. This makes it more profitable
to use flexible production equipment, which can be quickly and easily
changed over from task to task.
Frequently introducing new products adds another dimension to the
strategy. Obviously, it increases the need for product designers and design
engineers. It also means that equipment originally purchased to make one
product will surely need to be switched to producing other products, further
increasing the importance of flexibility. And it increases the importance of
communications between the design team and the manufacturing department,
since the product must be designed to be manufacturable with existing
equipment.
The modern manufacturing strategy also affects personnel and compen­
sation policies, supplier relations, and accounting systems in predictable
ways. Short production runs mean that factory workers will frequently be
changing tasks, moving from making one product to another. This tendency
is compounded by the frequency of product redesigns and new product
introductions. Flexible manufacturing thus requires multiskilled people who
can do more than one task, and companies that adopt this strategy will often
compensate workers on the basis of the skills they acquire, rather than on
the basis of the particular job that they are presently assigned to do. In a
system that is based on just-in-time manufacturing and low le\·els of buffer
inventories, the rate of \Vork at each step of the process must be in constant
balance; if the molding equipment runs faster than final assembly even
temporarily, inventories will accumulate. Consequently, systems tend to
avoid piece-rate compensation, which would encourage workers to work at
their own varying paces.
Frequent product redesign makes it important that the firm has accurate
information about the cost of producing various alternati\'e designs, both in
order to make well-informed design choices and to price the product correctly.
Since several products are being made on the same machine, the correct
allocation of machine costs becomes an important focus of the accounting
system. The variable that limits the use of machines is generally time, and
so the increasing tendenq among firms that use the modern manufacturing
strategy is to assign machine costs to the individual products on the basis of
the machine time that the product requires.
To the extent that the firm's strategy allo\\'s it to use general-purpose
manufacturing equipment, it has a greater ability to rely on subcontractors
to make its product in times of high demand and to invest in equipment that
it cannot keep busy. Another alternati\'C for the firm is to sell its excess
capacity to other firms, that is, to become a subcontractor itself.
 111
Coordinating Plans
and Actions

Optimal Prod uct

Variety Curve �

Figure 4. 3: There may be several coherent

combinations of batch size and product variety,
Batch Size but only one will generally be optimal.

Complemeritarities, I nnovation Attributes.,

and Coordination Failure
When complementarities are present, the various aspects of a firm's strategy must be
aligned properly for the firm's strategy to be optimal. Alignment alone is not enough
for optimality, however. The parts of the strategy that Henry Ford adopted in
introducing the Model T fit together well, and for a period of time the strategy worked
effectively. However, as described in Chapter 1 , Ford's strategy was overtaken by
Alfred Sloan's strategy at General Motors, which was overtaken in turn by a variant
of the modern manufacturing strategy introduced by Eiji Toyoda and Taiichi Ohno
at Toyota.
Even in the absence of explicit strategic planning, the various managers in a
firm may be able to adapt their choices to each other's, learning how best to pursue
the firm's objectives in their own narrow domains. Figure 4. 3 illustrates the possibilities
for, and a potential problem with, an adaptation of this kind.
The figure portrays the decisions of two managers-a manufacturing manager
and a marketing manager-in the same firm. The marketing manager chooses the
number of product varieties to offer to customers, and the manager of manufacturing
decides how large the batch sizes should be, that is, how many of one item to make
before switching production over to another item in the product line. Suppose that
each seeks to make the decision that maximizes aggregate firm profits.
Offering more products allows the firm to tailor its offerings more closely to
customer needs. Then it can choose to receive a higher price for its products or to
increase the number of units it sells. This benefit of increased variety must be weighed
against certain costs. Holding the batch size fixed, increasing the number of varieties,
raises the level of inventories proportionately. The larger the batch size, the costlier
are increases in product variety and the smaller is the optimal number of products.
This relationship is depicted by the optimal variety choice curve in Figure 4. 3 .
Similarly, given the number of different products being produced, there is an
optimal batch size. Lower batch sizes increase the frequency with which the firm
must incur the set-up costs involved in switching production from one product to
another, but they also reduce average inventories and their costs. When the number
of varieties is increased (holding total unit sales constant), the marginal inventory cost
of increasing batch size rises. For this reason, increases in product variety lead to
smaller optimal batch sizes. This relationship is depicted by the optimal batch size
curve in the figure.
DEC[NTH.·\ LIZED DECISIONS AND COORDINATION FAILURE Suppose that each manager
sets the variable he or she individually controls to maximize the firm's profits, given
 1 12
Coordination: what the other manager is expected to do. Then the marketing manager's choice will
Markets and lie along the optimal variety choice curve in the figure at the point determined by
Management the batch size this manager expects the other to select. Similarly, the production
manager will select the point on the optimal batch size curve corresponding to the
product variety he or she anticipates. Points where the two curves cross are ones where
each manager's choice is the right one to maximize firm profits, given the other
manager's choice.
As the figure shows, when the optimal variety choice is a decreasing function
of the batch size and the optimal batch size is a decreasing function of the number
of varieties, there can be several different combinations of product variety and batch
size at which each manager's choice maximizes the firm's overall profits, given what
the other is doing. Each of these combinations represents a coherent strategy. The
lower right-hand intersection point can be likened to Henry Ford's strategy with the
Model T-low variety and large production runs. The upper left-hand point can be
likened to a "niche market" version of the modern manufacturing strategy-many
varieties produced in small batches. Of course, the actual strategies involve many
more choices than are illustrated in the figure, making the choice problem much
more difficult.
Still, it is possible that the managers would find a coherent combination on
their own, even without any explicit attempt to coordinate their decisions. For
example, each might observe what the other is currently doing and adjust optimally to
this choice. Such a decentralized process might converge without any communication
between the managers (although it is not guaranteed to do so). In any case, it is sure
to take time to reach coherence in such an uncoordinated fashion, during which the
firm is operating inefficiently. Having the two managers meet and share their
information would, on the other hand, seem very likely to achieve coordination on
one of the coherent patterns.
There is, however, no reason to suppose that the particular coherent combination
the managers arrive at will be the best one. In general, only one combination is
actually profit maximizing, although this diagram does not contain enough information
to determine which it is. 9 Ford's strategy was best for the auto industry in the early
part of the century, while many industries now find that the modern manufacturing
strategy of a broad product line combined with small batch sizes is optimal. Could
the operating managers, using only the information they might normally be expected
to have gained from their own experience, somehow settle on the best one? The
answer is quite likely "No." Even if the managers pooled their information and
attempted to make a coordinated decision, they may still not have enough information
to determine which of the coherent strategy combinations is best.
I:\:\()\ .\TION ATTR IBlTES .\ND STR.HEC IC CooRDINATIO'.\ The key difficulty is that as
the environment with its demand conditions and the costs and productivity of different
technologies changes, the two curves may change more or less smoothly, and, with
them, the precise description of each of the coherent strategies. Yet a small change
in the curves may hide the fact that the relative profitability of the different coherent
patterns has changed and so th� actual profit-maximizing strategy has shifted radically.

9 Determining which is best requires looking at a third factor, the total profits accruing to any pair
of choices. The two curves represent "ridge lines, " traces of the highest points on the profit surface in the
relevant east-west or north-south direction. Depicting the height (the total profits) along the ridge lines
graphically requires a third dimension. If this third dimension were shown, it would reveal that the profit
surface here is like a mountain with three peaks, one at each of the intersections of the curves. Any of
these might be the highest , so that the intersection point under it is the strategy pattern that actually
maximizes profits.
For exam ple, gradual changes over the years in productio n techno lo gies and customer Coord i nating Plans
needs have no t radical ly altered the general form of the coheren t pattern correspo nding and Actions
to mass productio n. Y ct in many industries these cum ulative changes have made the
modern manufacturing strategy much more profi table than the mass production
strategy. This means that profit maximization wo uld require a radical shift in strategy.
The operatin g managers may be able to use their familiarity with demand and
cost characteristics to keep up with local movements in the coherent patterns,
margin ally adj usting their decisio ns to track the changes in what is best. B ut recognizing
that a radical strategic shift is desirable is likely to require informatio n that neither
manager has, and the local adj ustments they are likely to make are never going to
track the glo bal shifts in which strategic configuratio n is best. A shift to much smaller
variety and larger batch sizes might enable the firm to use highly specialized equipment
that is better tailored to the particular products being made, leading to much lo wer
co sts. EstiTT1 ating the benefi ts of such a change would require knowledge abo ut
technologies and equipment that may be completely unfamiliar. Similarly, a shift to
a much wider product line and smaller batch sizes may require much more flexible
equipment and completely revised relatio ns with suppliers, which might again be well
beyo nd the range of the productio n manager's experience. Also, the marketing manager
may not know how much customers wo uld be willing to pay for more highly tailored
products, or by how much sales would fall if a narrower but cheaper product line
were offered.
As in this example, the overall strategic choice in the presence of complementari­
ties is frequently a design decisio n with innovatio n attributes. The pieces of the strategy
have to fit together, and the informatio n needed to identify and choose between
alter native patterns is unlikely to be available freely within the organization. In these
circumstances, informatio n must be acquired from outside the organization and
assessed, a coordinated decision must be made o n the design variables, and then this
decisio n must be communicated to the affected parties. All this presents a role for
central coordination and, in particular, for to p management. We discuss this in the
next sectio n.
We emphasize again that the example we have been discussing involves o nly
two choice variables, and yet complementarities still lead to potential failures of
coordination. In general, the problem of selecting amo ng multiple coherent policies
becomes more difficult as the number of variables to be adapted increases and as
the strength of the complementarity relatio ns between pairs of decisio ns grows.
U nderstanding the complementarities in the system makes it easier to identify
po tentially profi table changes of strategy and to anticipate the scope of the changes
needed to implement a new strategy.

MANAGEMENT, DECENTRALI ZATION.

AND THE M EANS OF COORDINATION
We have seen a variety of situatio ns in this chapter in which the price system, the
quintessential decentralized coordination mechanism, is less than ideal because it is
too brittle and so does not deal well with imperfect informatio n or because it requires
too much communicatio n. When the price system fails, the search for effi ciency
means that other mechanisms must be adopted. Sometimes these supplement the
market's operatio ns, but often they involve the creation of formal organizatio ns that
largely supplant the market over a range of activities.
The coordinatio n problem is not solved by merely putting a no nmarket form of
organizatio n in place, however. Instead, it is transformed into a problem of man­
agement. The ideas in this chapter are useful for thinking abo ut the tasks of
 1 14
Coordination: management and, in particular, the role of senior management and its staff in
Markets and hierarchies . But first we need to set our terminology more precisely.
Management
Centralization and Decentralization
We have used the terms centralized and decentralized frequently throughout this
chapter, and although we have used them in a consistent way, we have not been
explicit about exactly what we mean by them. We now need to be more precise.
Think about a situation in which there is a set of individuals who have various
decisions to make and actions to perform. The individuals here could actually be
groups if each group can usefully be thought of as performing actions and having
information. A particular decision is then decentralized if it is left to the individuals
alone to make. In contrast, a centralized decision is one that is made at a higher level
and communicated to or imposed on the individuals. This higher level might be
thought of as an individual who has the power to make the decision, as in a managerial
hierarchy or under state planning. Alternatively, it could be the whole of the set of
individuals meeting and acting collectively, as in a meeting or through a referendum.
The centralization of a decision is a matter of degrees, depending on the level of the
hierarchy involved or the number of different managers who must approve the
decision.
The price system can be thought of as being fully decentralized in this sense.
Individuals decide on their own how much to offer for sale and how much to try to
purchase. Even the prices are determined in a largely decentralized fashion by
individual decision makers. At the opposite extreme is the traditional assembly line,
which, in principle, operates in a completely centralized fashion. Individuals on the
line have little or no discretion about what they do and when and how they do it.
In complex organizational decisions, however, neither decentralization of all
aspects of the decisions nor complete centralization is likely to be optimal. Crucial
information always resides with individuals, so centralizing all aspects of a decision
requires that this local information be communicated upward to the central decision
makers, or else ignored. Both are costly. But leaving all decisions to the individuals
who actually take the actions risks these decisions being uncoordinated. The problem
is then to determine just what aspects of the overall decision ought to be left to the
various individuals, what information should be communicated to assist individual
decision makers, which parts of the decision should be centralized, who should make
the centralized decision, and what information sources they should use. In a system
with both centralized and decentralized decisions, the centralized decisions serve to
define the parameters of the decentralized ones and to put constraints on the local
decision makers.

The Role of Management in Coordination

The key role of management in organizations is to ensure coordination. The survival
and success of the organization is crucially dependent on achieving effective
coordination of the actions of the many individuals and subgroups in the organization,
on making sure that they all are focusing their efforts on carrying out a feasible plan
of action that will promote the organization's goals, and on assuring that the plan is
adjusted appropriately to remain feasible and appropriate as circumstances change.
Ensuring that the members of the organization are properly motivated is very important
as well, and, as we see in later chapters, motivation problems can influence the
effectiveness of coordination mechanisms. But incentives become an issue only once
there is a feasible plan to be carried out and a pattern of behavior that needs to be
followed.
 1 15
The first step in achievin g coordination is the organizational design decision Coordi nating Plans
discussed earlier. This involves determining which decisions are to be centralized and and Actions
which are to be left to indivi duals and operating un its, who should make the cen tralized
decision s, and what information will be transmitted upwards to support the centralized
decision making an d back down to guide those who will implement the plan. The
con cepts developed in this chapter are useful here.
First, design variables need to be determined in a cen tralized fashion by a senior
manager, perhaps supported by staff, or by meetings amon g local managers. The
benefits of this centralization tend to be greatest when there are many divisions or
departments that need to be coordinated. Timing and scale are examples of
design variables about which explicit agreement must be reached. In the modern
manufacturing strategy discussed previously, the breadth of product lines, the flexibility
of equipment, and the frequency of design changes are design variables that affect the
planning of several differen t departments and groups within the firm such as the
marketing department, the manufacturing department, the engineering department,
the accounting department, and the pr oduct design group. Decisions about designs
need to be set and then communicated to those whose plans are affected.
While design decisions are best managed in a centralized way, most of the
details of any plan are not design decisions. Efficient organization takes advantage of
the knowledge of local managers and workers by allowing them considerable autonomy
in implementing their parts of the plan. The modern manufacturing strategy calls for
flexible equipment, but it does not specify what equipment the manufacturing
department should use, how workers should be trained, what hours to work, and so
on. Having management send the right messages focuses the search of individual
decision makers within the firm and adds coherency to the individual efforts, adding
enormously to the pr oductive ability of the organization. Of course, the organizational
design pr oblem is itself a design problem, and must be solved in a centralized fashion.
Second, in deciding what and how to communicate, the costs of information
transfer and the brittleness of the corresponding planning system become important.
We have seen that sometimes prices are the efficient means to guide decisions, but
that in other cases design variables and quantities should be used to guide local
decisions. Again, the decision about how to communicate is a design pr oblem that
should be centralized.
Third, when complementarities lead to multiple possible coherent patterns of
local decisions, then efficient choice among these requires that the decision be
centralized. Thi s need becomes even more acute when the choice involves innovation
attributes, with the information necessary to identify the different coherent patterns
and to evaluate their profitability not being available in the organization. I n these
circumstances, someon e must track the external environment, collect and evaluate
the information relevant to the strategic choice, and communicate the new strategic
pattern to the operating levels if a change is called for. M aking sure that this task is
done is an assignment problem, which itself must be solved centrally. Then
the determination and communication of the new strategic design again call for
centralization.

SENHm M.\N.\CEI\IE.NTS Hou: Senior management and its staff usually take on these
roles of determining and communicating strategy in firms. Yet, again, much of this
can and should be decentral ized to take advantage of local knowledge. For example,
senior management need not work out all the details of how to implement a modern
manufacturing strategy. Instead, having determined to make the switch, it might
sii nply an nounce to the operating managers and workers that the company is going
to accentuate quick turnaround, low inventories, flexibility, and meeting customers'
 11 6
Coordination: needs in product design and that it is to achieve this reorientation by a certain date.
Markets and Ideally, each division or department manager could then base his or her own plans
Management on the assumption that the other departments would be ready as planned, determining
the myriad details for implementing the plan either individually or working in groups
with other managers.
These same ideas have been expressed in many ways by senior managers at
various firms, but the following comment by Hewlett-Packard co-founder David
Packard seems to capture the main points particularly well:
Early in the history of the company, while thinking about how a company
like this should be managed, I kept getting back to one concept: If we
could simply get everybody to agree on what our objectives were and to
understand what we were trying to do, then we could turn everybody loose
and they would move along in a common direction. 1 0
Planning at Hewlett-Packard meant sharing information until an agreement was
reached about "what our objectives were" and "what we were trying to do. " Once an
agreement about that was reached and communicated so that everybody understood,
a key part of senior management's job was largely done. It was time to "turn everybody
loose" to use their local information and individual capabilities to implement the
plan. The plan would serve to lend coherence to the efforts, so that everyone "would
move along in a common direction. "

10David Packard, quoted in M. Beer and B. Spector, Human Resources at Hewlett-Packard , Harvard
Business School Case #9-482- 1 2 5, 1 982.
 117
Coordinating Plans
and Actions

SUMMARY
Much of the growth of productivity in the modern world has come from specialization.
People acquire specific training to do specialized jobs using equipment and tools that
are particularly well suited to the task. With the growth of specialization comes a need
for coordination so that sufficient amounts of the desired products are produced by the
right methods, using the best resources.
The price system is one of the main methods used in a market economy to
coordinate specialized activity. According to the fundamental theorem of welfare eco­
nomics, ifmarket-clearing prices can be found and certain other conditions are satisfied,
the market allocation is guaranteed to be efficient. The price system is not the only
system with these properties, however, which raises the question: What systems of
coordinatio� are most effective and under what circumstances? To evaluate effective­
ness, we introduced three criteria: ( 1 ) Can the system identify an optimal or efficient
decision in the problem if perfect information is available to the coordinator? (2) How
much communication does it require to identify an optimal decision? (3) If the informa­
tion used is i mperfect, how badly does the system performance deteriorate? That is,
how brittle is the system?
In assessing the market system, we found that when there are economies of scale
in production, market-clearing prices generally fail to exist, so the fundamental theorem
of welfare economics does not apply. Generally, in these settings, a simple price system
cannot guide an optimal resource allocation, even though other systems may be able
to do so.
Second, the brittleness of the price system varies with circumstances. When there
are only slightly declining returns to scale (compared to the rate of decrease in marginal
benefits from additional production), the price system tends to be quite brittle. That is,
small errors in estimating costs can lead to large inefficiencies in actual production
decisions compared to those suffered by a system in which a central planner sets quantity
targets. When the relationships of scale are reversed, the price system tends to perform
better than a system of centrally established production targets.
In general resource allocation problems, in which nothing is known a priori about
the optimal allocation, no system can identify an optimal resource allocation using less
information than a system of prices uses, where communication is measured using the
Hurwicz criterion. Consequently, in these general environments, if a system of prices
works, then it economizes on costly communications.
Coordination problems in which there is a priori information about how the parts
of the decision must fit together and in which small failures of fit are very costly are
said to have design attributes. Centralized setting of design variables tends to reduce
both the cost of errors and the amount of communication and search necessary to
identify an optimal decision.
The scale of a firm's operations is a design variable. It serves to coordinate deci­
sions about the scale of various component operations, the degree of specialization of
equipment and people, the location of facilities, and so on. Forecasts of a firm's planned
growth are therefore an important part of coordination in a business organization.
Economies of scale that are achieved in the manufacture of components of several
products increase the need for coordination because the optimal scale of each product
is an increasing function of the anticipated scale of the other products that use the same
components. In this case, it is said that the products are complements in production and
that there are economies of scope. When the shared input to making different products
are design and manufacturing capabilities of the firm, then these shared inputs are core
 1 18
Coordination: competencies. The manager of any particular product, who focuses narrowly on the
Markets and costs and benefits in developing just that product, may neglect the product's important
Management contribution to building the firm's core competencies. Apparently unprofitable invest­
ments or product decisions may actually be profitable if they preserve and extend a
firm's competencies, enabling it to introduce more profitable products in the future.
Complementary activities are ones for which increases in the level of some of the
activities raise the marginal profitability of the others. When complementary activities
are undertaken, they must all be done together to achieve maximum effect. Strongly
complementary activities are a common source of design attributes. In our boxed appli­
cation, we identified the wide range of complementary activities that together constitute
a modern manufacturing strategy.
Complementarities commonly give rise to a variety of coherent strategies. Individ­
ual managers all acting with the firm's interests in mind may get stuck with a coherent
strategy that is not the best one for the firm. It takes conscious effort and centralized
decision making in these cases to determine and focus attention on the proper strategy.
When the price system is supplanted by a formal organization, solving the coordi­
nation problem becomes a key task of management. The first step is to solve the organi­
zational design problem of determining which decisions are to be centralized and which
ones are to be left to individual decision makers, who is going to make the centralized
decisions and what information is to be communicated upwards or obtained from
outside the organization to support these decisions. Also to be determined are how
the centralized choices that delimit and guide the decentralized decisions are to be
communicated to the individual decision makers. This situation is itself a design prob­
lem that must be solved in a centralized fashion. Usually, solving this problem in firms
is the task of senior management, as is the solution of the design problem of setting and
communicating strategy.

• BIBLIOGRAPHIC NOTES
The comparison of alternative systems of economic control was an important sub­
ject in economics following the Russian Revolution, when there was a debate about
the ability of socialist economies to be as productive as capitalist ones. Among
Friedrich Hayek's important contributions to the debate were his emphasis on the
difficulty of communicating to a central planner all the information he or she
would need to develop and implement an effective plan, compared to the much
more meager information needed to run a price system. Leonid Hurwicz formal­
ized and developed this idea, culminating in the theorem c ited in the text. These
analyses were generally optimistic about the potential of prices for coordination.
The first formal studies of the brittleness of alternative systems of economic control
were made by Martin Weitzman, some of whose work is reported here in a
simplified form. Patrick Bolton and Joseph Farrell have recently compared the
costs of centralized and decentralized control systems, emphasizing the relative
quickness of centralized responses to crises and the ability of centralized systems
to avoid duplication of efforts. The treatment of design and innovation attributes
is our own contribution, new in this book, and the treatment of modern manufac­
turing as a designed system is drawn from our American Economic Review article.
The importance of complementarities among different individual's decisions was
accentuated by Jeremy Bulow, John Geanakoplos, and Paul Klemperer and Drew
Fudcnberg and Jean Tirole in the context of industrial competition. Our Economet­
rica paper shows how to analyze situations marked by complementarities, but it is
technically very difficult. Coordination failures induced by strong complementarit­
ics have been an important theme in the recent literature on Keynesian macroeco-
 1 19
norn ics (surveyed by Russell Cooper and Andre w J ohn), but thi s theme h as not Coordinating Plans
before been systematically appli ed to problems in the the ory of organization. and Actions

• HEFEHENCES
Bolton, P. , and J. Farrell. "Decentralizati on, Duplication and Delay," Journal of
Political Economy, 98 (August 1 990), 80 3-26.
Bulow, J. , J. Geanakoplos and P. Klemperer. "Multimarket Oligopoly: S trategic
Substitutes and Com plements," Journal of Political Economy 93 (1985), 488-511.
Cooper, R. , and A. J ohn. "Coordinating Coordination Failures in Keynesian
M odels," Quarterly Journal of Economics, 10 3 (August 1 988), 441-64.
Fudenberg, D. , and J. Tirole. Dynamic Models of Oligopoly (Chur, Switzerland:
Harwood Academic Publishers, 1986).
Hayek, F.A. "The Use of Knowledge in S ociety," American Economic Review, 35,
(1945), 519-30.
Hurwicz, L. "The Design of Mechanism s for Resource Allocation," American
Economic Review 63 (M ay 1973), 1-30.
Milgrom, P. , and J. Roberts. "Rationalizability, Learning, and Equilibrium in
Games with Strategic Complementarities," Econometrica, 58 (1990), 1255-77.
Milgrom, P. , and J. Roberts. "The Economics of M odern M anufacturing: Tech­
nology, S trategy, and Organization," American Economic Review, 80 (J une 1990),
511-28 .
Weitzman, M . , "Prices vs. Quantities," Review of Economic Studies, 4 1 , (October
1974), 477-91.

EXERCISES

Food for Thought

1. During World Wars I and II, even the m aj or market economies switched to
centralized forms of planning to allocate m any scarce resources. Are there reasons based
on effi ciency to prefer such centralized planning in wartime? Are there other reasons?
2. For many years, S wiss craftsmen made the fi nest watches, which were
powered by winding a spring by hand. The S wiss lost ground to competitors in Japan
and elsewhere as m ore accurate, less expensive battery- powered electr onic watches
were perfected . What were the core competencies of watchmaking in the era of S wiss
dominance and how were they changed by new technologi es? What other products
are made by the new watchmakers?
3. In 1987, S ony Corporation, a leader in consumer electronics (televisions,
compact disk and audio cassette players, home vid eo recording and playback machines),
purchased the CBS Record Company. In 1989, S ony purchased Colum bia Pictures
Entertainment, one of the giants of the m ovie and TV production industry. In 1990,
M atsushita, the world leader in consumer electronics with its Panasonic, National,
and Quasar brands, purchased MCA Corporation, which makes m ovies, television
sli ows, and music recordings. Why would these compani es want to expand into these
particular ind ustries?
 120
Coordination:
Markets and
Management

Q)
(.)

&

a Figure 4.4: An error scenario in which the

I I
Quantity planner is mistaken about marginal benefits.

Mathematical Exercises

1. Consider the problem of supplying an input to a division of a firm or to a

firm in a planned economy. Suppose the executive or planner knows the marginal
cost function for the input. This is represented by the fixed supply curve in Figure
4. 4. However, the planner is uncertain of the benefit. Her best estimate of the

the price should be P and the quantity supplied Q. Consider an error scenario in
marginal benefit curve is represented by the curve MB in the figure, and on that basis

which the actual marginal benefit is higher, as indicated by the curve MB '. Mark on
the graph the quantity that would be produced using a price control in the error
scenario and draw a vertical line at that quantity level. Identify the triangles with areas
that correspond to ( 1 ) the loss suffered under the error scenario when Q is announced

when the price P is announced and the division chooses its quantity to maximize
under a quantity planning system and (2) the loss suffered under the error scenario

profits. Show that the ratio of these losses with an uncertain benefit curve is just the
inverse of the ratio with an uncertain cost curve as derived in Equation 4. 1.
2. A new product is being introduced in competition with another firm.
Having the product ready for production in t months will result in net profits after
the introduction of 1 44 - t2 if O :5 t < 1 2 but will generate no profits at all if t ?: 1 2.
Three departments must be ready before production can begin, and getting ready
quickly is a costly matter. Department 1 can be ready in t months for a cost of
3( 1 2 - t), department 2 for a cost of 4( 1 2 - t), and department 3 for 5( 1 2 - t). Until
all three departments are ready, production cannot begin. ( 1) What is the optimal
date t* at which to introduce the new product, that is, the time that maximizes the
net profit after product introduction minus the cost of readying the three departments?
(2) What will the firm's profit be if the project is mistakenly hurried and all three
departments are asked to deliver one month earlierthan the optimal date? One month
later? (3) What will the firm's profit be if department 1 and department 3 both are
asked to be ready at the optimal time, but department 2 is asked to be ready one
month earlier than the others? One month later? (4) Is the timing of product
introduction in this example a design decision?
3. A small community must decide how much land to set aside and improve
for public parks, to be used for playgrounds, ball fields, and so on. Suppose there are
one hundred families in the community and that there are no wealth effects on the
families' preferences. That is, family n's utility for x" dollars of personal savings and
y acres of park land is at x" + v"(y). If a plan to set aside and improve y acres of land
at cost c· y is proposed with family n paying t" in taxes to finance the project, what
conditions must be checked to test the efficiency of the plan? What is the minimal
amount of communication required to check whether the proposal is efficient?
 121
Coordinating Plans
and Actions

MATHEMATICAL APPENDIX:
A FORMAL MODEL OF DESIGN DECISIONS
In this appen dix we study a mathematical model of the problem of introducing a new
product, for example, a new model car. Our principle purpose is to show that this
problem indeed has the characteristics of a design decision and to establish that the
informationally effi cient way to handle such a problem is to announce the design
attributes (rather than prices).
There are N system components that go into making the product, each of which
is produced in a single, separate facility or department. In the case of a car, these
would include systems like headlamps, engines, brakes, and so on, each of which is
itself a complex product with many parts and many steps involved in its manufacture.
We establish units so that exactly one unit of each component goes into making the
final product. For example, a car requires two headlamps, so we regard a pair of
headlamps as being one single unit in our formulas.
Suppose the corporation makes available a set of corporate resources for the
project, and that there are k different kinds of corporate resources. There are 1 units x
x x
of the first type of resource, 2 of the second, . . . , and k of the last. The vector x
= (x 1 , • • • , x�) describes the corporate resources available fo r the project.
The allocation of these resources among the N units is denoted by the list (x1 ,
. . . , xN), where xn = (x1 , . . . , x,:) denotes the vector of resources allocated to plan
and produce the nth component. Of course, the total amount of each resource
allocated to the various departments cannot exceed the amount available. The factory
that makes component n is ready to begin production at date t n and with capacity y n .
In addition to the corporate resources, the factory may hire outside workers, suppliers,
and contractors to do parts of the j ob. The total amount of outside expenditures
required to design and prepare to make system n by date t n at capacity y n using
corporate resources xn is en (xn , y n , t n , z n ). This function e n is a cost function and z n
is a parameter of the cost function that is known only by the local manager. (This
means that only the local manager knows how costly it is to meet the capacity and
timing obj ectives with a given allocation of centrally allocated resources.) We assume
that it costs more to get ready earlier and to build additional capacity; that is, en is
decreasing in t n and increasing in y n .
Because manufacturing a unit of the product requires one of each of the system
components, the rate of sales cannot exceed the minimum of the production capacities,
and sales cannot begin until the last of the system components is ready for production.
Let us therefore write the revenues as:

G ross Revenue = R[Min (y 1 , . . . , y N), Max (t 1 , . . . , t N )] (4. 3)

Revenues depend on two numbers: the amount that is produced and the date at which
it is available. The first of these is equal to the minimum of the outputs of the various
components and the second is the date at which the last component is ready. N aturally,
R is increasing in the total available capacity and decreasing in the time to introduction
of the product.
An organization seeking to maximize profits (gross revenues less the total costs
incurred in all units) would want to select the allocation of the corporate resources
(x), the unit capacities (y) and the readiness dates (t) to solve:
N
Maximize R[Min(y 1 , • • •, y N), Max (t 1 , • • •, t N )] - � e n(x n , y n , t n , z n) ( 4. 4)
n= I
 122
N
Coordination: subj ect to L x" :s x
Markets and n = I

Management wh ere the inequality constraint reflects the limited availability of corporate resources.
Let us suppose that the profit expression in Equation 4. 4 is concave and that
th e optimum involves positive amounts of each resource being used by each unit.
Th en, a plan (x, y, t) is optimal if and only if th ere exists a vector of prices p*-one
price for each resource-and numbers t * and y* such that the equality constraints
y" = y*
t" = t* } for all n. (4. 5 )
ac nlax" = p*

and adding-up conditions

"'v N -
�n = I x" = X

( 4. 6 )
N
L n = I a C nlay" = aRJay
L� = 1 a C)at" = aRJat
are all satisfied. The equality constraints say that each unit's capacity and readiness
date are th e same, and that the impact on costs of changing the amount of any of the
corporate resources allocated to a unit is the same as at any other unit. The first
adding-up condition says that the corporate resources are fully utilized. The second
says that the marginal cost of adding more capacity in each unit j ust balances the
marginal revenue from the added capacity. Finally, the last says that the marginal
costs of having the output from every unit available a little sooner just balance the
additional revenue that results.

Prior I nformation and Efficient Plannin g

I n this formulation, it is clear that the choice of the timing and capacity variables has
one of the characteristics of a design decision. Even without knowing the actual cost
and revenue fu nctions, we know in advance that at an optimum all the times t" will
be equal and similarly all the capacities y" will be identical. There is a priori
information about how these variables fit, and one consequence of th is is that Hurwicz's
theorem does not apply. A price system is not guaranteed to be informationally efficient.
N evertheless, H urwicz's basic framework still applies. We can still proceed by
identifying the minimum amount of communication necessary to verify the optimality
of a proposed plan (x, t,y) and checking whether a proposed coordination system
ach ieves that minimum. For th is case, the relevant extension has been derived by
Fumikata Sato, wh o has established that, in addition to communicating the Nk + 2
numbers that comprise the plan itself (x, t, y), any system that can verify the plan's
efficiency must communicate at least k + 2N additional numbers. 1 1 H ere is a system
that achieves that bound . The central coordinator announces prices, an introduction
date, and a capacity (k + 2 numbers). Each pr oduction department n responds by
reporting what resources it wishes to purchase (k numbers) and its marginal costs for
additional capacity and earlier factory readiness (2 numbers). This information allows
one to check Equations 4. 5 and 4.6 and so to verify whether the plan is effi cient. It
requires the communication of N(k + 2) numbers by the N departments plus k + 2
numbers by the coordinator, so this system verifies the effi ciency of an allocation with
a minim um of communication.

1 1 Fumikata Sato , "On the Informational Size of Message Spaces for Resource Allocation Processes
in Economics with Public Goods," fournal of Economic Theory, 24 (February 1 981), 48-69.
 By compa rison, any attem pt to guide capacity choices and factory readi ness Coordinating Plans
dates using a price system would fare quite badly. For example, suppose there were and Actions
T � 2 possible delivery dates. Describing a full plan would involve the Nk numbers
that give the al locati on of the k different corporate resources to the N plants, plus a
description of the ready capacity for each of the N plants for each of the T dates, the
further NT numbers. Then there would need to be a price for each of the k resources,
and a price for capacity for each of the plants at each possible date. In sum,
an augmented plan under a price system would involve Nk + NT + k + NT =
k(N + 1) + 2NT num bers. Thi s exceeds the minimum required. Indeed, if the number
of possible introduction dates T is large, the amount of additional communication is
similarly large, being equal to 2NT - 2(N + 1).

The High Relative Cost of Failures of Fit

The other maj or attribute that defines a design problem is that failures of fit are more
costly than are failures to find the ri ght design. To see how this condition arises in
the present model, let us fi rst consider the cost of an error in which the parts fit but
the design is not optimal. Speci fically, let t* be the optimal introduction time and
suppose that all the tns are set equal to some different but coordinated date t* + E. To
estimate the cost of a small error, we substitute this expression for each t n in the
obj ective function and then take the derivative with respect to E. U sing the third
adding up condi tion in Equation 4. 6, we find that this derivative is z ero at E = 0.
That is, the marginal cost of an error of this kind is zero. This kind of small design
error is not at all costly. Intuitively, the reason is that the product introduction date
is optimally set so that the margi nal cost of seeki ng a slightly earlier i ntroduction is
just equal to the marginal benefit. A small change from t* to t should have a z ero
effect on profits, to a first- order approximati on. Si milarly, setting all the y ns equal to
y* + E, the derivative with respect to E is zero. A small error i n setti ng y leads to a
cost that is zero to a first-order approximation.
By comparison, small errors of fit among the variables are much more costly.
Suppose, for example, that all the variables but one, say y n , are set to their optimal
values. The cost of a small error in y n depends on whether y n is set too high or too
low. 1 2 When y n is set too low, the cost is approximately (to first order) equal to the
sIL:e of the deviation times the left-hand derivative of profits with respect to y n evaluated
at the optimum. When yn is set too high, the cost is the right- hand derivative times
the size of the deviation. These derivatives are:
. . aR ac n ""' ac;
£ h and d en vabve = - - _
Leit- = £.J;# n -1. (4. 7 )
ay ay n ay
ac n
R ight-hand derivative =
ay n
As long as capacity is costly, these derivatives are not z ero. Errors of fit incur first­
order costs.
The intuitive reason for thi s conclusion is straightforward. When y n is set too
large, the extra capacity is useless. The loss incurred from this error is the cost of that
extra capacity, as confirmed by the expression for the right-hand derivative. When yn
is too small, the limited capacity forces a reduction i n the level of output with a
marginal value of aRJay but also saves costs of aC/ayn . This difference appears as the
fi rst expression for the left-hand derivative in Equation 4. 7. U sing the third adding­
up condition in Equation 4. 6, this conditi on can be transformed into the second

12 This is because the total profits are not a differentiable function of y" at the optimum point.
 1 24
Coordination:
expression for the left-hand derivative. According to the second expression, the loss
Markets and is just the same as if a smal l amount of useless extra capacity were built by al l the
Management other departments, because the shortfall in capacity at department n renders the extra
capacities of the other departments unproductive.
The most important message, however, is not the particular forms that these
costs take. Rather, it is that decisions concerning timing and scale are, as we have
claimed, decisions with design attributes. Explicit coordination of these decisions by
either a central coordinator or by committee meetings, rather than decentralized
decisions l oosely coordinated by prices, is predictably the norm for decisions of these
kinds.
 Part

III
MOTIVATION : CONTRACTS,
INFORMATION, AND INCENTIVES

5
BouND ED RATIONALITY
AND PRIVATE I NFORMATION

6
MoRAL HAZARD
AND PERFORMANCE INCENTIVES
 5
BOUNDED RATIONALITY
AND PRIVATE INFORMATION

IAI ny attempt to deal seriously with the study of econom ic organ ization m ust
i
come to terms with the combined ram ifcations of bounded rationality a nd
opportun ism in conjunction with a condition of asset specificity.
Oliver Williamson 1

The preceding two chapters dealt with the problem of coordination. This chapter and
the next several (Chapters 5 through 9) are concerned with the other central problem
of economic organi zation and management: motiva tion. Motivation questions arise
because individuals have their own private interests, which are rarely perfectl y aligned
with the interests of other indivi duals, with the groups to which the individuals belong,
or with society as a whole. The coordi nation problem is to determine what things
should be done, how they should be accompl ished, and who shoul d do what. At the
organizational level, the problem is also to determine who makes decisions and with
what information, and how to arrange communjcations systems to ensure that the
needed information is available. The motivation problem is to ens ure that the various
individuals involved in these processes willingly do their parts in the whole undertaking,
both reporting information accurately to allow the right plan to be devised and acting
as they are supposed to act to carry out the plan.
Our analysis throughout these chapters is based on the assumption that people
will do only what they perceive to be in thei r own indivi dual interests. As we have
al ready admitted in Chapter 2, this is a caricature of actual human motivation and
behavior. Yet it is a powerful analytic simplification. Given this assumption, the

1 26 1
The Economic Institutions of Capitalism (New York: The Free Press, 1 985), p. 42.
 127
motivation proble m becomes one of arran ging affairs so that, as far as possible, selfish Bounded
in dividual actions take proper accoun t no t only of how the decision maker is affected Rationality and
by a decision, but of how others are affected as well. Private Information
This unnatural s tate of affairs is achieved primarily by compacts amo ng people,
who recognize their mutual interes ts and agree to modify their behavior i n ways that
are mutually beneficial. These agreements may encompass the sort of actions each is
to take, any payments that might How from o ne to another, the rules and procedures
they will use to decide matters in the future, and the behavior that each might expect
from the others. Usi ng the language that has become s tandard in economics, we refer
to these agreements as contracts, regardless of whether they have the legal s tatus of
contracts. These agreements are voluntary, they will be accepted only if the parties
find them individually and mutually advantageous, they can be crafted to suit
individual needs and circumstances, and in actual relationships, they may perform
the same functio ns that formal co ntracts do, and more. I n fact, co ntracts may be
completely· unarticulated and implicit, with no power of law behind them whatsoever.

PERFECT COMPLETE CONTRACTS

I n principle, a perfectly fashio ned complete contract could solve the motivatio n
problem. I t would speci fy precisely what each party is to do i n every possible
ci rcumstance and arrange the distri butio n of realized costs and benefits i n each
contingency (includi ng those where the co ntract's terms are violated) so that each
party individually finds it optimal to abide by the co ntract's terms. If the original plan
were an efficient one, then a complete co ntract could implement the plan leadi ng to
an effici ent outcome. I f we start our analysis from this idealized perspective, we find
that motivation problems arise only because some plans cannot be described in a
complete, enforceable contract.

The Requirements of Com plete Contracting

What would be involved in reaching and enacting a complete co ntract? First, the
parties must each foresee all the relevant contingencies that might be important to
them in the course of the contract and to which they might want to adapt the
contractually specified actio ns and payments. Moreover, they must be able to describe
these contingencies accurately so that they can unambi guously determi ne before the
fact just what possibili ties are bei ng discussed. They mus t also be able to know after
the fact which of the particular circumstances they considered beforehand has now
actually occurred. S eco nd, they must be willing and able to determine and agree
upon an efficient course of actio n for each possible co ntingency, as well as the
payments that are to accompany the actions. Third, o nce they have entered the
contract, they must be happy to abide by its terms. This has at leas t two elements.
First, the parties must not mutually desire to renegotiate the contract later. Otherwise,
the anticipatio n that they will renego tiate may deprive the original agreement of its
credibility and may prevent it from guidi ng behavior as it should. S eco nd, each party
must be able to determine freely whether the contract's terms are bei ng met, and, if
they are bei ng violated, each must be willing and able to enforce the agreed
performance.
APPLYING TO COLLEGE: AN EXAMPLE To clarify j ust how severe the demands of
complete co ntracting are, consider what would be involved in your entering such a
contract with a university to become a student seeking a degree there. First, bo th you
and the universi ty's representatives must be able to anticipate all the various
circumstances that might be relevant to the relatio nship in the future, and together
you must be able to describe them unambiguously. This i ncludes what subj ects yo u
 1 28
Motivation: might want to study in any particular term (which might depend on your experiences
Contracts, in previous classes and a host of other factors) and what faculty might be available to
Information, and teach courses. Other potentially relevant contingencies include the condition of the
Incentives j ob market fo r graduates with different degrees and training when you fi nish; the costs
to the univer sity of serving meals in the student cafeteria; whether an earthquake
might damage the campus buildings and, if so, which ones and how much; whether
a war will break out and whether you will want or need to fight; whether a dr ought
might dry up the campus lake on which you might have wanted to sail; and whether
a new scientifi c discovery might render what you have learned irrelevant. The list
could be infi nite. Each of these potentially could affect what being a student is worth
to you, what having you as a student is worth to the university, and what courses of
action are effi cient in your dealings with each other.
Next, you and the university must figure out and agree on what to do in each
of these distinct circumstances and what cash should pass from you to the university
(or vice versa) in each event. Should you continue with your planned maj or if the
professor you hate is the only one available to teach a course that is required for this
maj or? If so, what tuition should you be charged? Should you be expected to take the
same courses and pay the same fees if you turn out to be a star football player as you
would if you were to be a mediocre cellist in the university orchestra or a brilliant
master of the electronic games in the student union? What if you become disabled?
Or one of the university's faculty members wins the N obel prize, making your degree
more prestigious? Or if the university is found liable for overcharging the government
on research contracts? Or if the school's fund-raising efforts are much more successful
than anticipated and it is suddenly richer than anyone ever dreamed?
It is precisely this sort of complete contracting perspective that underlies the
most expansive interpretations of the theory of competitive markets, as described in
Chapter 3 . The goods traded in complete, competitive markets include state-dependent
commodities like "one seat in PoliSci I O I if Professor J ones is teaching it in the spring
term of my sophomore year. " The competitive theory requires even more: The
exchanges in competitive markets occur at publicly quoted prices; they inv olve
commodities with well-specifi ed, generally observable attributes and qualities; the
identity and personal characteristics of the buyer or seller are irrelevant; and it is
costless to determine if the agreements are being kept and to force compliance if they
are not.

The Problems of Actual Contracting

In actual transactions, enacting and enforcing a complete and perfect contract is
fraught with problems. Limited foresight, imprecise language, the costs of calculating
solutions, and the costs of writing down a plan-collectively, the bounded rationality
of real people-mean that not all contingencies are fully accounted for. Your contract
with the university does not provide explicitly for the kinds of contingencies we
described earlier, and many other contracts are similarly incomplete. In complicated
relationships, contingencies inevitably arise that have not been planned for and, when
they do, the parties must fi nd ways to adapt. These adaptations introduce the possibility
of opportunistic behavior, including reneging. For example, a university may change
its course requirements, or raise its tuition unduly, or cancel its plans to build more
student housing, leaving the student with little option but to meet the new requirements,
pay up, and spend another year living with his or her parents.
Fear of opportunism may deter parties from relying on one another as much as
they should for efficiency. For example, a farm family may be reluctant to rely on
 129
union labor to harvest a crop, feari ng that a strike threat during harvest season would Bounded
leave it too vulnerable to demands for wage increases. They might plant a smaller Rational i ty and
crop to allow it to be harvested by friends and family, even though a larger crop Private I nformation
harvested by organized labor might be more efficient. In general terms, incomplete
and unenforceable contracts lead to problems of imperfect commitment. For example,
if the farmer sees that the union is not actually committed to carry out the harvest
on the originally contracted terms, it will be reluctant to deal with the union.
Even if a contingency can be foreseen and planned for and contractual
commitments can be enforced, one of the bargainers may have relevant private
information before the contract is signed that interferes with the possibility of reaching
a value-maximizing agreement. In the used car market, for example, sellers have
better information about their cars than do potential buyers, which makes the buyers
skeptical about quality. Newspapers are filled with advertisements claiming that the
current car owner is "moving and must sell"-claims that are designed to make it
credible tha·t it really is a good car being sold at this bargain price. Fear that such
claims are not true and that the owner wants to sell the car because it is a lemon
probably sours many a worthwhile deal. This source of inefficiency is called adverse
selection, representing the idea that the selection of cars offered in the market is
determined in a way that is adverse to the interests of the buyer.
Even if there is no private information before the agreement is made, there may
be inadequate information afterward to tell whether the terms of the agreement have
been honored, or acquiring that information may be costly. This opens the possibility
of self-interested misbehavior, and the recognition of this moral hazard problem limits
the contracts that can be written and enforced. Real contracts are not perfect.
Consequently, the parties' individual interests under actual contracts will not
necessarily be properly aligned, and this leaves room for self-interested behavior to
thwart the realization of efficient plans. The motivation problem is then to overcome
these difficulties to the extent possible. This involves recognizing that what can
actually be accomplished is constrained by individuals' self-interested behavior and
then designing the most efficient plans that recognize these incentive constraints. It
also involves designing systems that better align individual interests, so that the
constraints are looser and the available options richer. Doing so requires a detailed
understanding of the various causes of imperfect contracting, of the consequences of
the various sorts of contracting difficulties, and of the responses that have emerged.

BOUNDED RATIONALITY
AND CONTRACTUAL INCOMPLETENES
The idea of foreseeing and unambiguously describing every contingency that might
possibly be relevant to the agreement between you and the university is obviously
ridiculous. No one could conceivably foresee every eventuality in such a complex
environment. Moreover, no human language could possibly be both rich enough and
precise enough to describe all the eventualities, even if they could be foreseen. Indeed,
even writing down the descriptions of those events that can be foreseen and described
would likely take more time than you would actually spend getting your degree!

Bounded Rationality
Real people are not omniscient nor perfectly far-sighted. They cannot solve arbitrarily
complex problems exactly, costlessly, and instantaneously, and they cannot communi­
cate with one another freely and perfectly. Instead, they are boundedly ra tional, and
they know it. They recognize that they cannot possibly foresee all the things that
 t:30
i\lotivation: might matter for them, they understand that communication is costly and imperfect
Contracts, and that understandings are often flawed, and they know that they are not likely to
lnformation, and find the mathematically best solution to difficult problems. They then act in an
lncenti\·es intentionally rational manner, trying to do the best they can given the limitations
under which they work. And, they learn.
L'.'\FORESEE'.'\ CIRCDISTA'.'\CES As we mentioned earlier, with bounded rationality,
contracts are not complete. Inevitably, contingencies will arise that have not been
accounted for because they were never imagined at contracting time. Often, these
gaps in the contracts are unimportant because what the parties ought to do and the
costs and benefits they receive are largely unaffected by the particular circumstances.
Sometimes they can have massive consequences for the parties' welfare, however. For
example, in all likelihood many of the American companies that bought television
network advertising time for the 1 980 Moscow Olympic Carnes ne\·er considered the
possibility that the U. S. team would boycott the games. Without U. S. athletes to
cheer for, Americans were much less interested in watching the games, and the
advertising time was of little value. Similarly, Westinghouse and the electric-power
utilities to which it sold enriched uranium fuel for their nuclear reactors apparently
did not foresee the possibility of a sharp escalation in world uranium prices when they
contracted for these sales. When this actually occurred (allegedly as a result of the
formation of an international cartel among the uranium-producing countries), it
forced Westinghouse's costs up and led it to repudiate the contracts.
COSTLY CALCL'LATIO'.'\S .\'.'\D CO'.\TR..\CTI:\G Even when contingencies are foreseen,
they may appear so unlikely that it is not worthwhile to describe them in detail and
to agree about what to do if they should arise. This is most likely the case when the
contingencies seem very improbable, when there is little experience \\·ith comparable
situations that would guide the planning about what to do should they arise, when
the opportunity costs of the parties' time spent writing the contract rather than doing
productive work are high, and when the contingencies seem unlikely to cause large
disputes if they should occur. Yet these calculations themselves are subject to error.
A recent example comes from a General Motors' panel stamping plant in Flint,
Michigan. In an attempt to indicate trust in the work force and reduce regimentation
and rigidity, the plant's management changed the work rules to allow employees to
leave early if they had met their day's production targets. Soon some workers were
leaving by noon, having done their jobs in half the time previously allotted. Outraged
at paying full-time for half a day's work, management increased the production targets.
The workers were left bitter, feeling that management reneged on a promise. They
viewed the contract as: "Do your work at your pace, and we'll pay you the same
amount as before for the same amount of output. " Trust and industrial relations at
the plant have deteriorated. Management certainly could have foreseen that the
workers might increase their pace significantly, and even double it. But apparently
management did not think it worthwhile to specify what the response would be to a
doubling of the pace or to communicate it clearly. 2
THE htPRECISIO'.'\ OF L.\'.'\Gl'.-\r.E A further source of contractual incompleteness is
that the natural languages in which contracts are written are inherently imprecise.
This means that statements describing any reasonably complex situation must be
somewhat ambiguous. For example, the "commercial practicability" doctrine in
contract law means that a firm that signs a contract is required to perform as agreed

2 Details can be found in C.A . Patterson, "UA\\' and Big Three Face l\1utual l\1istrust as Auto
Talk Heats Up,", The Wall Street Journal (August 29, 1990), IA
 1:u
only when performance is reasonably possible. Just what ci rcumstances would justify Bounded
application of this provi sion is i nherently uncl ear, however. In the uranium ca se, Rationality and
Westinghouse's lawyers argued that, foll owing the huge price increase for uranium Private In formation
ore, delivery of the processed uranium was "commercially" impossible and therefore
not required. Not surprisi ngly, the other side disagreed.
More generally, there is a difficult trade-off arising from the ambiguity of
language: Adding numerous specific provisi ons to cover behavior in more distinct
eventualities means that there are more boundaries that actual circumstances may lie
near and more question about which provisions apply. For thi s reason, adding many
detailed provisions to a contract can make di sputes even more likely.

Contractual Responses to Bounded Rationality

People design their contracts recognizing that they cannot possibly be perfectly adapted
to all possible future circumstances. Instead, they structure the best contracts they
can. One possibility is to write i nflexible contracts with blanket provisions that are to
apply very broadly. Thi s minimizes the costs of describing eventualities and actions
and leaves little room for ex post uncertainty about what behavior is requi red. Thi s
in fact may be efficient for relatively simple transactions that are quickly concluded,
so there is little possibility of a change i n circumstances before the contract expires
that would alter what actions are appropriate. Contracts reached under these
circumstances are called spot market contracts because they govern goods or services
that are to be exchanged "on the spot. " For more complex transacti ons that extend
over time, an i nflexible specification of the acti ons to be taken is likely to be too
unresponsive to changing conditi ons.
RELATIONAL CONTRACTS Another possibility is relational contracting, whi ch does
not attempt the impossible task of complete contracting but instead settles for an
agreement that frames the relationship. The parties do not agree on detai led plans of
action but on goals a nd objectives, on general provisions that are broadly applicable,
on the criteria to be used in deciding what to do when unforeseen contingencies ari se,
on who has what power to act a nd the bounds limiting the range of actions that can
be taken, a nd on dispute resolution mechanisms to be used if disagreements do occur.
For example, companies entering a relationship to collaborate on a joint research a nd
development project do not attempt to figure out precisely what they wi ll do i n every
detail as the uncertain project evolves. Instead, they each agree to give their best
efforts to developing the project, to share the costs and benefits, to consult wi th one
another as new developments occur, a nd to bargain in good faith when disputes arise.
Such contracts can in fact work quite effectively, at least when the potentia l conflicts
are not too great and the parties are not inclined to be too opportunistic in thei r
dealings wi th one another.
The contract that governs faculty employment at Stanford University is a nice
example of relational contracting. The written porti on of the contract is the Stanford
Faculty Handbook. 3 The document contains fairly explici t provisions governing such
matters as the accrual of credit towards sabbati cal leave, the distri buti on of rights to
writings and patents resulti ng from professors' work, and the limitations on outside
employment. It also states policies regarding scientific fr;ud, sexual harassment, and
the role of a faculty member i n setting a spouse's salary. On many of the most
important issues to a facul ty member, however, such as the determination of annual
salary and teaching assignments a nd the criteria for tenure, it is ei ther silent or consists

3 Interestingly, neither author of this text had seen this document when he agreed to join the
Stanford faculty.
 132
Motivation: largely of speci fic procedures to be followed in making decisions and the mechanisms
Contracts, for appeal.
Information , and Employment contracts, which typically delegate authority to the employer to
Incentives direct the employee's actions rather than describing the work to be done in every
contingency, are a response to the necessity of incomplete, imperfect contracting.
When an employee is hired, he or she (implicitly) agrees to follow the employer's
directions, so long as they fall within certain bounds that may be quite vaguely
defined. There is no detailed bargaining about what precise actions the employee will
take in various circumstances, as there would be in a world of complete contracting,
although the contract (often embodied in an employee handbook) may specify some
broad expectations and rules. Similarly, the employee and employer are not required
to bargain with the employer over assignments and pay ex post, once the tasks that
need to be carried out become known: The expectation is that the employer will tell
the employee what to do. The employee's ultimate defense against unreasonable
demands is to quit, and the employer's defense against refusal to take orders is to fire
the employee. All this economizes on contracting costs.
In general, in situations where reasonably complete contracts are too costly or
impossible, actual contracts are relational. They serve to structure a relationship and
set common expectations, and they establish mechanisms that will be used to make
decisions and allocate costs and benefits. This pattern of agreeing on process and
procedure rather than on actions is mirrored in corporate charters. These charters
specify such matters as the procedures for selecting directors and officers and, in very
broad terms, their powers and the range of decisions that they may make without
consulting the stockholders, but they do not go into further detail. The nature and
role of contract and business law has a similar explanation. The law sets a common
framework for private contracting, establishing a basis for expectations about what
should or will be done in events not explicitly considered in the contract. It also sets
certain default provisions that will be applied if the contract does not explicitly provide
otherwise. It thereby economizes on the costs of contracting. At an even higher level,
the pattern is seen again in national constitutions, which typically state broad general
pri nciples to be used in deciding issues and describe the mechanisms that are to be
employed, but do not attempt to foresee the actual decisions that will be confronted.

hlPLICIT CONTRACTS An important adjunct to incomplete written contracts are the

unarticulated but (presumably) shared expectations that the parties have concerning
the relationship. The importance of these shared expectations has led to their being
labeled implicit contracts. For example, Stanford faculty members' implicit contract
with the uni versity involves their having a say in hiring decisions, their salaries being
set to reflect not just the work that the market sees and rewards but also their provision
of services to the institution, and their being provided with offices, some secretarial
support for research, and an equal opportunity to .hunt for parking spaces.
To the extent that the expectations actually are shared and commonly understood,
implicit contracts can be a powerful means of economizing on bounded rationality
and contracting costs. In this regard, corporate culture-seen as a shared set of values,
ways of thinking, and beliefs about how things should be done-is a key aspect of the
implicit contract. This view suggests why changing a corporation's culture may be
difficult: Doing so means breaking old contracts and implementing new ones, all
without the benefit of being able to discuss the terms of either contract very explicitly.
Implicit contracts, by their very nature, cannot be easily enforced in a court of
law. There is no document, and indeed there may never even have been any oral
statement of the contract. Thus, implicit contracts must rely on other enforcement
mechanisms. These mechanisms arc discussed in detail later in the chapter.
Effects of Contractual I ncompleteness Bounded
Contracts are meant to protect people by aligning incen tives. When contracts are Rationa lity and
incomplete, the alignme nt can be imperfect. Contracts can also be seen as a Private Information

mechanism to achieve binding commitments that the parties can bank on in their
planning. When contracts are incomplete and imperfect, however, they have only
limited effectiveness for achieving commitment.
Concern with the possibility of being disad vantaged by self-interested behavior
that an incomplete, imperfect contract does not adequately control may prevent
agreement being reached in the first place. It may also ineffi ciently limit the extent
of cooperation that can be achieved.
COMMITMENT AND RENEGING Achieving commitment can be very valuable because
it can affect other's expectations about your behavior and thereby the behavior they
adopt. For _example, in 1066 William the Conqueror burned his fleet behind his
invasion army, cutting off its only means of retreat and committing his men to
fighting. Cortes repeated this stratagem in his invasion of Mexico. This commitment
deterred the opposition, who then themselves retreated . M ore recently, when Apple
Computer introduced its Macintosh computers, it built a highly specialized plant to
produce them and designed the plant in such a way that it would be very diffi cult to
adapt it to any other use. Further, it publicized this fact heavily. By making it more
expensive to drop the M acintosh than it would otherwise have been, Apple committed
itself to this product and market. This could have affected the behavior of Apple's
employees, who knew that they had to succeed in this market; of Apple's competitors,
who would see that there was little point trying to drive Apple out of the market; and
of its potential customers, who could count on Apple's continued presence in the
market and its support.
Commitment runs into two sorts of problems. The simpler and more obvious
one is that one side or the other may try to renege on the deal. A customer may
simply not pay for services rendered, or a supplier may refuse to deliver the goods
that were contracted for, perhaps because the costs of completing the contract are
higher than was anticipated. Reneging is especially problematic with incomplete
contracts because what should be done in various circumstances is left unstated or
ambiguous and thus open to differing interpretations. Then it may be easy for you to
claim that what you now want to do is what was agreed, and that no reneging is going
on at all. The other party may be unsure about whether you really believe this, and
so may be reluctant to label you a cheat. M oreover, even if the other party fully
believes you are not living up to the agreement, the ambiguity of the contract means
that it may be very hard to establish for outsiders who is misbehaving and what actual
behavior is appropriate. Westinghouse argued in court that its refusal to supply
uranium fuel to the electric utilities was appropriate contractual behavior. The utilities
argued that Westinghouse should have to supply them at the stated prices, even if it
had to absorb immense losses. Astronomical amounts were paid to lawyers in the fight
over whether Westinghouse or the utilities were in the right.
When the reneging takes the form of not carrying through on the agreed actions,
it may obviously affect efficiency. H owever, often the problem with reneging is not
that it impedes effi ciency directly, but rather that it affects performance ind irectly.
For example, a customer's paying or not paying is simply a matter of whether a
monetary transfer is made from one party to the other. U nder the hypothesis of no
wealth effects from Chapter 2, this sort of transfer is without direct significance for
effi ciency. Whether the payment is mad e is obviously not a matter of ind ifference to
the individ ual parties, however, and the fear of getting cheated may prevent an
effi cient transaction from ever occurring in the fi rst place.
 134
Motivation: Ex. PosT RENEGOTIATION The second commitment problem is related but more
subtle. In some circumstances, it may be advantageous for both parties to renegotiate

may not be so once actions have been taken or further information revealed. If the
Contracts,
Information, and the contract ex post because what was efficient when the contract was first entered
Incentives
parties understand at the time they are crafting the original agreement that they will
later face these incentives, however, they may not be able to draft the contract in a
way that generates the desired behavior.
For example, many firms use stock options to motivate executives. These options
give the executives the right to buy the firm's stock at a specified price in the future.
The idea is that having th� options will motivate the executives to work to get the
market price of the stock above the price specified in the options so that they can
make money buying the stock cheaply. However, the options will not have much
effect on motivation if the price at which the option can be exercised is too high
relative to the market price, because the goal of getting the stock price higher than
the exercise price will seem unreachable. Thus, the exercise price is typically set
moderately higher than the current market price. Now suppose that after options are
issued, the price of the firm's stock falls drastically. Suddenly, the options are essentially
worthless to the executives and, thus, are worthless as incentives. It might then make

precisely this. If the executives forecast that this will happen again in the future,
sense to issue new options reflecting the lower stock price, and many firms have done

however, they will not be as worried about a fall in the stock's price as they otherwise
would be. This dulls their motivation to improve the stock price.
A complication is that in some contexts it will turn out ex post that to maximize
total returns the original terms should not be followed, and yet one party or the other
may have an incentive to insist on the inefficient completion of the contract. For
example, suppose a department store chain has contracted with a knitting firm to
deliver 1 0,000 cotton sweaters in various colors. The sweaters are worth $20 a piece
to the store, and they will cost the knitter $ 1 0 each to produce. The agreed price is
$ 1 5, and meeting the order will absorb the full capacity of the knitter. Then the
knitter gets an order from another chain for 20, 000 polo shirts. Each shirt is worth
$ 1 5 to this chain and can be produced for $5, but the knitter cannot produce both
the sweaters and the shirts. Efficiency requires that the shirts be produced rather than

sweaters is only $ 1 00, 000. Thus, the original contract should be abrogated. If this
the sweaters because this yields a total value of $200, 000, whereas the surplus on the

efficiently agreed that the shirts should be produced instead of the sweaters. If the
possibility had been foreseen in the original bargaining, then the parties would have

contract was simply to deliver the sweaters at the agreed price, however, then the first
chain may insist on delivery. In this case, a possible solution is for the knitter to
renege on its contract and pay damages of $ 50, 000 to the first chain for its lost profits.
However, this kind of solution is problematic if the values of the various options to
the different parties are not commonly known to all, because then the appropriate
level of damages is hard to determine.

Investments and Specific Assets

Although even relatively simple contracts can be subject to various problems and
disputes, the most commercially important impacts of imperfect commitment arise
when significant investments are required, because the amounts are significant and
because the benefits may accrue over an extended period of time. In economic
language, an investment is an expenditure of money or other resources that creates
a potential continuing Aow of future benefits and services. The potential Aow itself is
called an asset. The kind of assets that are most familiar are physical ones, like houses
or machines, which provide a potential Aow of housing or manufacturing services
 Bounded
Rationality and
Private Information
Terrorism, Parenting, and Commitment
The governments of man y countries have announced that they will not
bargain with terrorists, even when hostages have been taken and their lives
are threatened. Th is policy is designed to prevent terrorism: I f potential
terrorists expect that the world's governments will n ot deal with them, even
to save the lives of their citizens, then there is less gain to kidnapping or
other terrorist acts.
The problem with the policy is that it requires commitment to be
credible. If hostages actually are taken, then the pressures to save their lives
by negotiating with their captors are immense, no matter what policies were
announced beforehand. This is especially so in countries with popularly
elected governments and open mass communications, where the pain and
grief of the victims' relatives is seen on television and where the governments
are particularly responsi ve to citizens' concerns. Terrorists clearly understand
the incentives to renege on the policy, and this l ed them to take many
hostages from among the citizens of democratic countries but few from
undemocratic ones.
S imilar, if less dramatic, commitment problems arise repeatedly for
parents raising children. For the children's own good, parents want them to
do various things: avoid dangerous play, study hard, save money. To
encourage the desired behavior, parents use threats of punishment, ranging
from "If you play on the road, you'll get a spanking you'll n ever forget" to
"U nless you save the money yourself, you can't have a new dress for high
school graduation. " Of course, the parents' pain in meting out the pun ishment
may be at least as great as the child's in receiving it. The temptation is to
forgive and omit the punishment because once the misbehavior has occurred
punishment brings pain with no immediate gain.
B oth these examples show how concern with the future effects of n ot
following through with the promised respon se to misbehavior can stiffen
resolve and lead the government or parent to absorb the costs of carrying out
the promise. Reputation concerns sometimes lead them to carry out their
threats, even when they have an otherwise attractive temptation to renege.

over time, or fi nancial ones like savings bonds, which yield a flow of cash returns.
Many assets are not physical, however. Patents and copyrights are assets that allow
their owners to demand royalties or exclusive rights to reproduce books or software
programs for a certain period of time. Investments in education can create an extremely
important asset, human capital, which leads to an increased flow of future income
as well as flows of equall y important, if perhaps less tangible, benefits.
The kind of investments that are most problematic in terms of the incentives
they create are investments in specific assets, that is, assets that are most valuable in
one specifi c setting or relationship. When the value maximization principle appl ies,
the specificity of an asset is measured as the percentage of investment value that is
lost when the asset is used outside the specifi c setting or relationship.
An important special case of specifi c assets are cospecialized assets. Two assets
are cospecialized if they are most productive when used together and lose much of
their value if used separately to produce independent products or services.
For example, consider the electrical power generation industry. M ost electrical
 1 36
Motivation: generating plants are located either near the customers they serve or near a source of
Contracts, energy. Coal-burning electrical plants, for example, tend to be located either near
Information, and coal mines or near cities. The plants near cities most often receive their coal by
Incentives railroad. Special rail lines are built to the mine in order to ship out their coal. These
feeder rail lines are assets that are cospecialized with the coal mine itself . If the rail
lines were not used to ship coal from the mine, there would be a great loss of economic
value. Because coal is bulky and has a low value per unit of weight, it would probably
be economically infeasible to ship the coal by means other than rail, so the mine
without the rails is of greatly reduced value. Meanwhile, there is not much other use
that could be made of a rail line that runs between a railroad junction and a coal
mine other than to ship coal. Thus, absent the coal, the rails also are of little value.
The total value of the mine and the rail line in their separate, alternative uses is much
lower than is the value of the two assets when used together to provide coal to utilities.
In the case of a mine-mouth electric plant, which is located away from the city
it serves and near the mouth of a particular coal mine, the mine and the plant are
cospecialized assets. The mine-mouth plant relies on the output of the mine for its
coal and would lose much of its value if the mine were closed. The mine would lose
much or all of its value if its only customer-the generating plant-were closed. The
plant and the mine in this case are cospecialized because they are far more valuable
if used together to support the production of electricity than if the coal were diverted
to other uses.
The problem that arises when investments are made to create cospecialized
assets is that much of the value of the investment depends on the behavior of another
asset owner with his or her own selfish interests. This opens the possibility of various
sorts of (ex post) opportunistic behavior that endangers the investment. Mine owners
who invest in expanding their mine, for example, may find themselves at the mercy
of the electric utility, which might later threaten to use the plant only for backup of
its nuclear or hydro-powered plants. It is not only the mine owners who need to worry
about its specific investment, however. The utility owner might worry that after the
plant is built, the mine owners will try to raise their price for coal. Once a mine­
mouth plant is built, there is no alternative source of coal supply for the plant, so the
utility owner may become vulnerable to demands for a price increase.

THE HOLD-UP PROBLEM The general business problem in which each party to a
contract worries about being forced to accept disadvantageous terms later, after it has
sunk an investment, or worries that its investment may be devalued by the actions of
others, is called the hold-up problem. The party that is forced to accept a worsening
of the effective terms of the relationship once it has sunk an investment has been held
up.
Clearly, if contracts were complete, the hold-up problem would not arise: The
parties could specify the full range of circumstances that might arise and could agree
on the behavior to be followed in each of these. The usual mechanisms for contract
enforcement would then prevent ex post opportunism. In the present situation, you
would expect that the mine and the utility, faced with the potential hold-up problem,
would at least enter into a long-term contract, so that the coal price level could be
determined in advance. But at what level should the price be set? The coal mine
owner might be worried about increasing costs of labor and mining equipment. Should
the contract include an escalator clause, which automatically adjusts the price of the
coal when some index of mining costs rises? Should the price of coal be tied to the
spot market price of coal in the general area?
How reliable must the coal supply be? If the mine owner guarantees an
uninterrupted supply of coal, it will be in a weakened bargaining position when the
coal miners' contracts come up for renewal. If not, the utility may worry that the Bounded
mine owner will be too willing to weather a long strike. The mine and the utility Rational ity and
have different interests. Private I nformation
What about future expansions? If the demand for energy increases, is the mine
obliged to supply an increased amount of coal at the same price per ton? If its average
and marginal cost curves slope upward, then the additional coal will be more expensive
to provide. Should the price negotiations be left for later? Later, the mine will be a
monopoly supplier; now, it is j ust one of several possible sources of coal for a new
plant whose location has yet to be determined.
The hold-up problem that the coal-burning electric plant illustrates is that large,
specific investments make asset owners vulnerable to opportunistic behavior by their
contracting partners. This is not a problem in standard market theory where contracting
is perfect and alternative suppliers are readily available. Perfect contracting implies
that the parties can write adequate protections into the contract to prevent the kinds
of problem� that we have described. Besides, if there were many suppliers for all
inputs, then there would be no need for extensive contracting. There would be no
scope for price gouging because the prices would be set by a competitive market.
There would be no concerns about supply interruptions because there would be
alternative supplies of an identical or nearly identical input, and no worries about
demand curtailment because there would be alternative uses for the asset. It is the
specificity of assets together with imperfect contracting that lies at the core of the hold­
up problem. Concern about these problems may lead to inefficiencies as firms, fearing
that their investments will leave them vulnerable, refuse to make the efficient
investment.
In a series of articles, Paul Joskow has investigated the nature of coal supply
contracts to the electric utility industry. 4 His empirical research reveals that coal mines
supplying mine-mouth plants are most frequently owned by the utility they serve and,
in all other cases, have a long-term contract to supply the utility. According to Joskow,
these contracts are typically quite complex, involving escalator pricing clauses of
various kinds, and they do a good j ob of tracking the actual variations in the cost of
supplying coal.
Another example of cospecialized assets can be found in the automobile industry.
I n the 1 920s General Motors purchased their automobile bodies from an independent
firm, the Fisher Body company. As the technology of automaking advanced and the
car companies moved from wooden bodies to metal ones, General Motors began to
design a new automobile plant to assemble their cars. In order to improve the reliability
of supply and to reduce shi pping costs, GM asked Fisher to build a new auto body
plant adjacent to the new GM assembly plant. The plants would have no need for
shipping docks; bodies would be transferred on the production line right from the
Fisher plant to the General Motors plant. Fisher refused to make the requested
investment, perhaps for fear that the new plant, so closely tailored to GM's needs,
would be vulnerable to demands that GM might later make. The issue was eventually
resolved by vertical integration: General Motors purchased Fisher Body. 5
The hold-up problem is an example of postcontractual opportunism. It arises

4 Paul Joskow "Vertical Integration and Long-Term Contracts: The Case of Coal-Burning Electric
Generating Plants, " Journal of Law, Economics, and Organization, l (Spring 1 985), 33-80; "Contract
Duration and Durable Transaction-Specific Investments: The Case of Coal, " American Economic Review,
77 (March 1987), I 68-85 ; and "Asset Specificity and the Structure of Vertical Relationships: Empirical
Evidence, " Journal of Law, Economics, and Organization, 4 (1988), 95-11 7.
5 See Benjamin Klein, Robert Crawford, and Armen Alchian, "Vertical Integration, Appropriable
Rents, and the Competitive Contracting Process," Journal of Law and Economics, 2 I ( I 978), 297-326.
 138
i\1oti\·ation: Table 5 . 1 The possibility of grabbing
Contracts, creates a Prisoners' Dilemma
Information , and game.
Incenti\'es

Firm A
Grab Don't

Firm Grab - 1, - 1 - 2, 3
B Don't 3, - 2 2, 2

because contracts are incomplete and imperfectly specified, so that the parties to the
contract can exploit loopholes to gain an advantage over one another.
A. MATHEMATICAL EXA�IPLE OF THE HOLD-UP PROBLDI A numerical example will
help clarify the logic of the hold-up problem. Suppose that there are two firms, A
and B, which have joined together to take advantage of a business opportunity. Each
firm is called on to make an investment in an asset that is entirely worthless outside
of the joint venture. These assets are therefore completely cospecialized.
To keep matters simple, suppose that the project requires that each party make
an investment. The cost of the investments is 2 for each party, and the gross return
from the investment is 8, yielding a net benefit of 4. To capture the possibility of ex
post opportunism, suppose that the division of the gross benefits between the two firms
can be affected by costly actions that each takes, but that it is not possible to make

possible actions are "grab" or "don't grab," where grabbing costs 3 . If neither firm
an enforceable contract governing these actions. Again, for simplicity, suppose the

2. If both grab, the gross benefits are again shared equally, but each firm absorbs the
grabs, then the gross returns are shared equally and each party enjoys a net benefit of

additional cost of 3 associated with grabbing. This yields each firm a net payoff of
4 - 2 - 3 = - l . Finally, if one firm, say A, grabs and the other does not, then
it gets all of the benefits. The grabber's payoff is then 8 - 2 - 3 = 3, and the other's
payoff is - 2. These payoffs are shown in Table 5. 1 ; the payoff table has the form of
a Prisoners' Dilemma game. 6

to do that: It is costly and creates no value. If contracting about grabbing is impossible,

Clearly, if the firms could contract about grabbing, they would contract never

however, what should we expect the outcome to be?

The analysis begins by considering the decision facing firm A, assuming that
the specific investments have already been made. Firm B is supposed to be making
its decision at the same time, so there is no possibility that what B chooses will be

grab. What are A's options? If A grabs, then A receives a payoff of 3, as computed
influenced by A's actual choice. Suppose then that A hypothesizes that B will not

earlier. If A refrains from grabbing, then it gets only 2 . Thus, if A thinks B will
refrain, its own selfish interests are best served by grabbing the returns for itself. In
other words, A's highest payoff in the first column of Table 5. l is achieved by
grabbing. Now suppose that A hypothesizes that B will grab. Then by grabbing as
well it gets a payoff of - l, whereas its payoff is - 2 if it refrains. Once again, A does
best by grabbing. Because A has the same selfishly best action no matter what it

6 The Prisoners' Dilemma is the most famous and most studied strategic ga me. For a lively
introduction to the subject, see Avinash Dixit and Barry Nalebuff, Thinking Stra tegical/)' ( New York: W. W.
Norton , 1 99 1 ).
expects B to do, a selfish A will always grab. A parallel analysis of B's decision indicates Bounded
that a selfish B would do li kewise, leading to payoffs of - 1 for both. Rational i ty and
So far, the analysis is the same as for any Prisoners' Di lemma game; the Private I nformation
difference here, however, is that the firms do not have to play. I f the firms recognize
the incentives that they will face after the specific investmen ts are made, they will
want to allow for these in deciding whether to invest at all. In this example, it turns
out that the threat of opportuni sm completely destroys the incentives to invest: If the
firms cannot commi t not to attempt to grab more than their share of the returns, no
investment takes place.
This mathematical example is so stark that one might be tempted to believe that
simple morality or business ethics could overcome the problem, and indeed it
someti mes does. Much of the problem is that what is honest and fair is rarely as
obvious in the rich settings of reality as it someti mes seems in classroom examples,
and it is tempti ng for people to think that what serves thei r own interests is honest
and fair. It is· too risky to rely on others to act consi stently contrary to their own selfish
interests unless there is something that commi ts them to that ki nd of behavior.

Achieving Commitment
We have already menti oned relational and impli cit contracts as a response to
contractual incompleteness. These contracts serve to set expectations and establish
decision processes to deal with the inevi table unforeseen circumstances while avoi ding
some of the difficulties of writing expli ci t detai led contracts. When cospecialized assets
are a factor, a common response is for the same person or firm to own both. After
all, there is little concern that a firm wi ll hold itself up! As we see later, however,
thi s ownership soluti on may also incur certai n costs that li mi t its usefulness.
Another approach is to attempt to achieve commitment by noncontractual
means. Thi s may be worthwhile, but costly. William the Conqueror and Cortes used
up valuable resources-perfectly good fleets. Apple presumably spent extra on
constructi on costs for its overly specialized plant and in return got a facility that may
become obsolete sooner than it otherwise would. Yet such expenditures may be the
cost of commitment.

THE ROLE OF REPUTATIONS In other situations, concern with one's reputation may
be an effective check on ex post opportunism, overcoming the temptations to renege
or renegotiate. It may even achieve the same results as actual commi tment. Bargaining
with terrori sts or not punishi ng naughty chi ldren establishes without a doubt that there
is no commitment and invites further challenges. Renegoti ating executive pay contracts
makes less credible any claims that bad future performance will not be rewarded. Not
payi ng legi ti mate bills or not fulfilling obligations also results in a reputati on for
untrustworthiness. In a world of costly and incomplete contracting, trust is crucial to
reali zing many transactions. Thus, the concern with getti ng a bad reputation that
reduces future possibilities for profitable transacti ons can li mi t reneging. Effectively,
it removes the incentives for opportuni stic behavior by creating a cost offsetting the
short-term gai ns of opportunistic behavi or.
As we see in Chapter 8, the value of a reputation-and thus the costs incurred
in building and maintaini ng a good one-depends on how often it will prove useful.
This in turn is related to the frequency of simi lar transacti ons, the horizon over which
similar transactions are expected to occur, and the transaction's profi tability. The
incenti ves to build and maintain a reputation are larger the more frequent the
transaction, the longer the hori zon, and the more profitable the transaction. One
implication is that, in relational contracting, where there is an issue of which party
should have the discretion to direct activities in unforeseen events, it should be the
 1 40
Motivation: o ne with the most to lose from a damaged reputatio n. This is likely to be the o ne
Contracts, with the longer horizon, the more visibility, the greater size, and the greater frequency
Information , and of transactio ns. In the employment relatio nship with a stable, lo ng-lived firm, this is
Incentives more likely to be the firm than the emplo yee.

PRIVATE INFORMATION
AND PRECONTRACTUAL OPPORTUNISM
Once whatever co ntingencies are going to be accounted for are established, the parties
to a po tential co ntract must reach agreement o n a course of actio n and o n the
payments to be made. This involves some sort of bargaining.
Bargaining is a very complex process. It is also central to economic life. It was
lo ng traditio nal in eco nomics to suggest that the outcomes of bargaining among small
numbers of people are essentially indeterminate. S uch hard-to-describe features as
bargaining strength, credibility, guile, patience, and strategic insight wo uld combine
with initial co nditio ns to determine the outcome. The effi ciency principle would
suggest that an efficient agreement should be reached, but predicting the actual result
was too much for eco nomic analysis.
Yet, with the hypothesis of no wealth effects, the value maximizatio n principle
means that the value-creating aspects of the agreement are determinate. Only the
mo netary transfers wo uld be determined by bargaining strength and the like. Recall,
however, that efficiency is defined relative to a set of feasible outcomes. What is
feasible-and thus what is efficient-turns out to depend crucially o n the informatio nal
co nditio ns. When the costs and benefits of different plans to each party are kno wn to
that party alone, or when the likelihood of different possible outcomes are private
information, then these informational asymmetries can prevent any agreement
from being reached, even when an agreement would be efficient under complete
informatio n. Furthermore, even if an agreement is reached, it will not typically be
efficient when j udged by the standards of complete informatio n, especially when the
parties have an optio n not to participate.

Bargaining over a Sale

The po ssibility of failing to reach an agreement can arise in the simplest problems
with private informatio n abo ut values. Co nsider two people, a buyer and a seller. The
seller owns a unit of some good in which the buyer is interested. Each is privately
informed about the value that he or she places o n having the good. The buyer believes
that the seller values the good either at $2 or that she finds it worthless. The seller
believes that the buyer values the good at either $1 or $3. The buyer assigns a
pro bability of 0. 2 to the seller's valuing the good at $2, and correspo ndingly, a
pro bability of 0. 8 that it is worth $0 to her. The seller assesses a pro bability of 0. 2
that the buyer's valuatio n is $1 and a pro bability of 0.8 that it is $ 3 (see Table 5 . 2).
Of course, the seller kno ws what the good is actually worth to her, and the buyer
knows what it is worth to him.
If the actual valuatio ns are $1 for the buyer and $2 for the seller, it is efficient
for the good to remain with the seller-this maximizes the to tal val ue. This occurs
with pro bability .2 X . 2 = . 04. I n all other circumstances, 96 percent of the time,
efficiency demands that a sale occur and the good be transferred to the buyer.
I f the valuations were known, there would not be much difficulty. Both parties
wo uld know what the good was worth to the oth er, and if there is a value gain from
transferring the good, they ought to be able to settle on some paymen t from the buyer
to the seller that makes both parties better off.
 141
Table 5. 2 Efficient Outcomes with Different Possible Valuations Bounded
Rationality and
Seller's Value Private I nformation

$0 $2
Buyer's Value (Probability = . 8) (Probability = . 2)

$1 Trade No Trade
(Probability = . 2)
$3 Trade Trade
(Probability = . 8)

INFORl\lATIONAL ASYl\1METHIES AND STRATEGIC M ISREPRESENTATION With the actual

tryi ng to misrepresent their valuati ons to get a better pri ce. If the buyer's actual value
i nformational asymmetries, however, there are potential problems with the bargainers'

is $3, he might like to insist that it is only $1 in hopes of paying a lower price. If he

trade occurs) he cannot be expected to pay more than $ 1 . S o, for example, by refusi ng
can convince the seller that the good is worth only $1 to him, then (provided that

any offer at a higher price, he may gain. Similarly, if the seller' s value is actually $0,
she might try to insist that she actually values the good at $2 so she would get a better
price. Of course, misrepresentati on of this sort carries the risk of the sale falling
through because the parties convince each other that their valuations are $1 and $2,
so that they believe there is no basis for mutually beneficial trade. This may be a risk
worth taking, however.
To prevent this sort of misrepresentation, each party must do at least as well
behaving straightforwardly and, in effect, " confessing" his or her actual valuation, as
he or she does misrepresenti ng it. Further, because trade is voluntary, neither can be
forced to trade if he or she does not find it personally advantageous. These conditi ons
constrain the set of arrangements that are feasible by requiring that each party get at
least some minimum amount of surplus from the transacti on, where the particular
amount depends on the party's actual valuation. 7 It may then be possible that there
is simply not enough surplus to go around. This can prevent trade from occurring
even though the buyer actually values the good more than the seller does. With
pri vate information, the full- information efficient solution may no longer be feasi ble.
EFFICIENT TRADE MAY BE UNATTAINABLE To illustrate this possibility, let us take the

exchange when the parties represent their values i n particular ways. If there is no
point of view of a mediator with the power to suggest the prices that will govern

pattern of prices that a neutral medi ator could identify to promote effi cient trade, then
there is none that the parties, with their additional interests, could find for themsel ves.
The problem that must be solved is to discourage the buyer from falsely claiming
a low value (in the hope of getting a low price) and the seller from falsely claimi ng a
high value. A seller who claims a value of 2 must be offered a price of at least 2, or
she will refuse to participate. (Such limitations on what is feasi ble that arise from
individuals' options not to participate are called participation incentive constraints.)
Offeri ng a higher price in this instance only encourages the seller to make this claim.
Therefore, as Table 5 . 3 shows, part of any solution to this problem must entail setting
a price of 2 when the seller claims a value of 2 and trade takes place. S imilarly, to

7 This surplus is sometimes called an informational ren t, that is, an excess return (rent) that is
received because of the individual's private information.
 142
Motivation: Table 5. 3 Prices That Best Encourage
Contracts, Honest Reporting and
Information, and Efficient Trade
I ncentives

Buyer's Value
1 3
Seller's 0 p
Val ue 2 No 2
Trade

discourage the buyer from claiming a value of 1 , the price should be set at 1 when
that value is claimed and trade takes place.
Now, the problem is to find a price, p , to be paid when the buyer says he has

be truthful. Consider the seller first. If she reports honestly, she gets $ 1 when the
a high value ($ 3) and the seller says her value is low ($0) that will encourage both to

buyer claims a value of $ 1. If the buyer is honest, this will happen 20 percent of the
time. The seller gets p the rest of the time, where p is the price when the buyer's

honestly is . 2 X 1 + . 8 X p . If she were to claim her actual value for the good is
value is $3 and the seller's is $0. Thus, the seller's expected payoff when she plays

$2, she would not trade when the buyer says his value is low, which would happen
20 percent of the time again. She would be left with the good in this circumstance,
but this is worthless to her. The rest of the time she would get the price of $2 that
we calculated earlier-the minimum consistent with her being willing to trade if $2
were her actual valuation, as she claims it is. Thus, her expected payoff from
dissembling about her valuation is . 2 X 0 + . 8 X $2 = $ 1 . 60. Honesty pays if the
first of these two expressions is larger, that is, if . 2 + . Sp � 1. 60. This incentive
compatibility constraint must hold if the seller is to find it worthwhile to report

A similar computation can be done for the buyer. If he reports honestly, he

honestly. It implies that p must be at least $ 1 . 7 5.

pays a price of $2 when the seller says her value is $2 and a price of p when she says her
value is $0, garnering a benefit of . 2 x (3 - 2) + . 8 x (3 - p) = $2. 60 - . 8 x p. If
the buyer dissembles, he trades only when the seller says her value for the good is
$0, in which case he pays $ 1 and gets the good that is actually worth $3 to him. This
would occur 80 percent of the time if the seller is straightforward about her
announcement, so his expected payoff from dissembling is . 8($3 - $ 1 ) = $ 1 . 60. To
make it worthwhile for the buyer not to dissemble, we must have $2. 60 - . Sp �
$ 1 . 60, or p � $ 1 . 2 5.
Thus, to induce the seller to be straightforward, the price she gets when she
says her value is $0 and the buyer says his is $ 3 must be at least $ 1. 7 5, but in the
same circumstance he cannot be asked to pay more than $ 1 . 2 5. The conclusion is
that it is impossible to find any prices that always make it in the parties' individual
interests to report truthfully and to trade whenever trade is, in fact, value increasing.
l
ACHIE\'INC EFFICIENCY DESPITE TI IE INFORI\L\TJONAL AsY \t\lETRIES We conclude that,
at least in this particular example with private information about values, it is not
possible for trade to occur precisely when the buyer's value exceeds the seller's. 8 The

8 You might object that we have not really shown this because we assumed that it was necessary to
get the buyer and seller to reveal their true valuations rather than to misrepresent what the good was worth.
Perhaps there is some scheme under which they do misrepresent, but trade nevertheless occurs just when
the buyer's actual value exceeds the seller's. However, a remarkable result cal led the revelation principle
special characteristic of this example is that the risks to the seller of exaggerating her Bounded
value to an honest buyer and to the buyer of misstating his value to an honest seller Rationality and
arc both low: The sale will be consummated wi th 80 percent likelihood anyway. Private I nformation
If the probabilities were reversed in the example, so that the low value for the
buyer and the high value for the seller both occurred wi th high probabi lity (0. 8 each),
then it would be possible to get correct revelati on and sti ll have trade occur just when
the buyer's value exceeds the seller's. For example, consider the incentives that exi st
if p is set to $ 1 . 50 in Table 5. 3 . If the buyer reports truthfully and the seller dissem­
bles when her actual value is $0, then the seller's expected payoff is . 8 X O + . 2 X
$2. 00 = $. 40, whereas telling the truth pays . 8 X $ 1 . 00 + . 2 X $ 1 . 50 = $ 1 . 1 0.
Honesty, in this case, is the best policy for the seller, and the same can be shown to
be true for the buyer; the incentive compatibility constraints are met.
The crucial difference when the probabi lities are reversed like this is that now
there .is a high probability (. 8) that the buyer will in fact be deali ng with a seller with
whom there· are no gai ns to trade unless the buyer's value is high, and similarly, the
seller is likely to be dealing wi th a low-value buyer and so should not trade unless
her value is low. This makes it more costly to dissemble, because dissembling now
runs a hi gher risk of preventing trade. If the hi gh-value buyer (or the low-value seller)
is straightforward, he always gets to trade. If he tries to mani pulate the pri ce to hi s
advantage, then 80 percent of the ti me he does not get to trade and misses out on
these benefits. This makes it easier to meet the i ncentive compatibili ty constraints­
they do not eliminate the full-information effici ent outcome.

I ncentive Efficiency
We have seen that private informati on someti mes, but not always, makes it impossible
to achieve the full-i nformation effici ent outcome. Recall that we have defined
effici ency relative to a set of people and a set of feasible outcomes. If incentive
constraints are unavoidable, then we do not label the outcome as i nefficient merely
because a better outcome could be achieved in the absence of incentive constraints.
Instead, a mechani sm for determining allocations is effici ent if there is no alternati ve
method that is feasible in the presence of incentive constrai nts and that all the relevant
parties prefer to the original mechani sm. In much of the economics literature,
mechanisms that are effici ent when incentive constrai nts are recogni zed in this way
are called incentive efficient. Accordi ng to the efficiency princi ple, we ought to expect
the bargainers to find incentive-efficient mechanisms. The key question then becomes
how much value the parties can create for themselves when faced with incentive
constraints.

INCENTIVE EFFICIENT BARGAINING WITH PRI VATE INFORMATION: AN EXAMPLE Another

example of bargai ning indicates what can be accomplished. Suppose agai n we have
a buyer and a seller, but now each party's value for the good might take on any value
between $0 and $ 1 with equal probability. Of course, the buyer knows the true value
B that he places on the good, and the seller knows her value S . Suppose the two
adopt the following simple rule to govern their bargaining: They will each simultane­
ously name a price; i f the price named by the seller is less than that named by the
buyer, the sale is made at a price midway between the prices; and if the seller's
demand exceeds the buyer's offer, no trade is made and the two bargainers are
commi tted to walk away.

shows that this is not the case: Any pattern of outcomes that can be achieved using any mechanism that
respects incentive constraints can in fact be achieved by a mechanism in which the buyer and seller report
their values and are given incentives to do so honestly.
 144
Motivation: If each side named the actual value of the good to him or her, trade would
Contracts, immediately occur whenever there are gains from trade, that is, whenever the buyer's
Information, and value B exceeded the seller's value S. I t cannot be in the two parties' individual
Incentives interests, however, to reveal their actual values: If the buyer is going to bargain
"honestly," it is advantageous for the seller to play strategically and announce a higher
asking price than her true value. Although this increases the chances that trade will
not occur, the trades that are lost are ones that were hardly profitable anyway. For
example, suppose the buyer is forthright but that a seller whose value is $. 50 reports
a val ue of $. 52 instead. The only trades that will be lost on account of this
misrepresentation occur in situations in which the buyer's val ue lies between $. 50
and $. 52 and therefore in which the price lies below $. 51, so at most $. 0 I of profit
is lost. O n average, when a loss occurs, it amounts to j ust $.005. Moreover, the
chance of loss is j ust the 2 percent probability that the seller's value lies in the specified
range, so the total expected loss from trades missed on account of the misrepresentatio n
is . 02 X $. 00 5 = $.0001. Against this is a price increase of $. 0 I that is enjoyed
whenever the buyer's val ue is at least $. 52, which occurs 48 percent of the time. So,
the expected extra profits are then $. 0048, for a net expected gain of $.0047 ( = $. 0048
- .0001).
Similarly, if the seller is forthright, the buyer gains by understating his valuatio n.
From the buyer's point of view, claiming that the value is lower than it actually is
involves a chance of blocking a trade that would be profi table, but as before the cost
of a small misrepresentatio n is tiny. More often, the misrepresentatio n wil l lead to a
reduced price for the buyer, and this effect is by far the larger o ne: Small misrep­
resentatio ns always pay.
In actuality, the calculatio ns we have just made are not the most relevant o nes.
If it is going to be in the seller's interest, for example, to misrepresent her value, then
the buyer ought to account for that in determining his own misrepresentation. Actually
calcula ting the bid and ask fu nctions that the two parties would use if they correctl y
forecast one another's choices is somewhat complicated, but the result that emerges
is that the two wil l systematically bid different amounts than their true values. As a
result, trade will occur if and only if the buyer's val uatio n exceeds the seller's by
1/4. 9 This is suggestive of a typical pattern: When incentive constraints are important
i
in bilateral bargaining, trade takes place only if the gains from trade are suffciently
large.
One might ho pe that some other bargaining arrangement would yield greater
efficiency, but this is actually not possible. In the example co nsidered, the specified
bargaining rules actually maximize efficiency subject to the constraints imposed by
the private information. For example, the parties could have initially decided to
co ntinue bargaining as lo ng as one or the o ther wished to do so or until an agreement
was reached. The ineffi ciency will still be manifested, however, either in the form of
costly disagreements or as a costly delay until an agreement is reached.

9 To calculate the strategies that the parties will adopt if they correctly forecast one another's
incentives to adopt strategic behavior, suppose that the seller expects the buyer's bid to be a linear function
b = 8 1 + 82 B of his true valuation and, in a corresponding fashion, that the buyer expects the seller's
asking price to be a linear function s = CJ 1 + CJ2S of S. Suppose too that the parties actually do decide to
use rules of this form to decide how much to bid and ask. (Of course, correct revelation of their values
occurs when they pick 8 1 = CJ 1 = 0 and 82 = CTz = l . ) Then it is possible to calculate, for given assumed
values of the CJ s, how the buyer's expected payoff depends on his choice of 8 , and similarly, how the
0

seller's payoff depends on her choice of CJ for any assumed values of 8. Maximizing each party's payoff
with respect to its choices (by setting derivatives equal to zero) and then requiring that each correctly
forecasts the other's choices yields the conclusion that b = ( l / 1 2) + (2/3)B and s = (l/4) + (2/3)S. Then
b > s only if B > S + 1/4.
 1 45
The existen ce of private i nformation frequently mean s that some value­ Bounded
maxim izing plans cannot be realized. S ometimes the pote nti al gai ns from transacting Rationality and
are missed. Incentive-efficient arrangem en ts m inim ize these losses, as in th is exam ple, Private Information
where trade does not occur if the gains are small but docs if they are sufficien tly large.

Efficient Agreements with Large Numbers of Partici p ants

Although the problem of incentives and private information can be serious even in a
bilateral relationship, it becom es much more serious when the num bers of people
who must agree begin to grow. When there is a diversity of interests, even moderate­
sized groups often find it impossible in practice to reach a unanimously acceptable
decision. In theory, with a large enough set of participants, it can be a virtual certainty
that there are gains to be realized from an agreement, and yet there may be no
agreement that meets the incentive and parti ci pation constraints, making it impossible
to realize th� potential gains.
INVESTMENT PROJECTS TO CREATE PUBLIC GOODS: AN EXAMPLE To illustrate the
underlying theoretical problem, we consider a simple decision about whether to
undertake an investment proj ect to create a public good. The cost of the pr oj ect is $ 1
in total, no matter h ow many people are benefitted. Each indi vi dual either wants to
have the proj ect done, placing a value of $2 on it, or is i ndifferent about it, getting
a value of $0 from its being undertaken. The probability that the project is valued
posi tively by any single individual is p, a num ber strictly between 0 and 1. Agai n,
each indi vidual knows his or her actual valuation, but this value is private information.
Individuals are free to drop out of the gr oup, but if they do so they do not get the
benefits of the proj ect.
If anyone wants the proj ect done, value maxim ization demands that it should
be carried out because the cost is $ I and the benefit is at least $2. If the group consists
of a single individual, that person will surely execute the proj ect. S uppose, however,
that there are several people. Anyone forced to pay m ore towards the costs of the
proj ect than it is worth to him or h er can drop out of the group and avoid paying.
Thus, no one can be forced to pay m ore than $2, and someone who claim s his or
her value is $0 cannot be m ade to pay anything. In essence, everyone must agree to
the plan for it to be undertaken. The danger is that people will try to free ride off the
others, falsely claiming that they value the proj ect at $0 and hoping that someone
else will pay for it.
Suppose that if m 2::: 1 out of the N people in the group say their values are
high, then the proj ect is built, with the costs being assigned randomly to one of the
m people, or that the costs are split equally am ong the m peopl e voting for the proj ect.
If there are two people in the group (N = 2), then we can express the incentive
constraint that a person with a value of $2 be willing to acknowledge that as follows:
p($2 - $. 5 ) + (1 - p)($2 - $ 1 ) 2::: p($2) + ( 1 - p)($0)
The first term on the left is the probability p that the other person will say his or her
value is high (assuming honest reporti ng by that person), times the gain the first
indivi dual receives in this case. This gain is the $2 he or she gets from having the
proj ect don e, less the expected cost to him or her, which is $. 5 0. The second term
is the probabili ty that the oth er person will report a value of zero, tim es the gain in
this case, which is the benefit of the proj ect less its cost. S o the left-hand side is the
expected return from reporting honestly. If the person dissembles, then the proj ect i s
built and financed only when the other person says his or her value is $2 (which
occurs with pr obability p), in which case that other person bears all the cost. The
right-hand side is thus the expected return to dissembling. The inequality says that
 1 46
being forthright is worthwhile. This inequality must hold if the project is to be built

Solving for p, we see that p cannot be too large: p � 2/3. If p is too high, it
Motivation:
Contracts, whenever it is value maximizing.
Information, and
Incentives appears too likely that the other person will volunteer to pay, and the indivi dual under
consideration cannot be induced to report his or her true valuation. It is precisely
when it is likely that others would be willing to finance the project even without one
person's participation that the person is most tempted to claim that he or she does not
value the project and, as a result, that an efficient project will be delayed or left
unbuilt.
The way to induce correct revelation is to reduce the probability that the project
is carried out when the reported gains seem small. Let q be the probability that the
project is carried out when only one person states that his or her value is $2. Then
the incentive constraint becomes

p($2 - $! ) + (1 - p )q($2 - $ 1 ) � pq($2) + ( 1 - p)($ 0 )

Now, if p > 2/3 , q must be less than 1 to induce truthful reporting. For example, as
p approaches 1 , q must approach 3/4. Thus, 2 5 percent of the time when someone
volunteers to pay, the project is not carried out. In other words, people will sometimes
free ride and not volunteer to pay for the project, even when it is actually individually

If there are three people, the incentive constraint (calculated assuming that the
worthwhile to pay for it themselves.

project will be built whenever it is value maximizing) gets even tougher to meet. In
this case, it is

p2 ($2 - $!) + 2 p( l p)($2 - $ !) + ( 1 - p) 2($2 - $ 1 )

� p 2($2) + 2 p( I - p)($2) + (1 - p)2 ( $ 0 )
which requires p to be less than about 0. 44. If p exceeds this value, the incentive
constraint cannot be met if the project is to be built whenever its value exceeds its
direct cost.
As N gets large, the incentive constraint becomes more and more difficult to
meet-free riding becomes more and more tempting if the others are expected to
report honestly. The essential intuition is that dissembling is costly only when the
others' reports make the given individual pivotal-the project is built if he or she
admits to having a high value for it, and not otherwise. As N gets large, it becomes
a virtual certainty that someone else will value the project highly and, assuming
honest reporting, will volunteer to pay for it. The cost of free riding, which is the
chance that the misrepresentation prevents the project from being constructed, then
becomes vanishingly small, and so misrepresentation cannot be prevented.
For example, when N = 3 and p = 1 /2, the incentive constraints mean that
the best that can be done (in terms of maximizing total expected value) is that the
project is undertaken when all three or any two of the people say their values are
high, but it is undertaken less than 90 percent of the time when only one person
admits to having a high valuation (that is, the person is pivotal). Yet, this one person
is individually willing to pay enough to finance the project.
As we discussed earlier, this analysis implies that it is not reasonable for people
to assume that others will all report honestly. That recognition increases their own
individual returns to honest reporting. The precise calculations are rather complex,
but the conclusion for this example is that as the number of people involved in the
decision becomes large, assuming that everyone behaves similarly, the probability that
someone will volunteer to contribute to the project tends to a limit of approximately
 147
two chances in seven, even though it is esse ntially certain that funding the proj ect is Bounded
value m aximizing. Rational ity and
Private Information
Bargainin g Costs
Contracting req uires reaching a mutually beneficial agreement. We have j ust seen
that private information about the values of different options can prevent reaching an
agreement on value-maximizing plans. One way to think of this is in terms of
bargaining costs, the transaction costs of reaching mutually acceptable agreements.
Bounded rationality imposes bargaining costs, even in the absence of any strategic
behavior. It takes time and effort to imagine and list contingencies, to determine
efficient courses of action, and to settle on divisions of costs and benefi ts. In fact,
bounded rational ity means that it may be impossible to carry out these tasks completely
and perfectly. The resulting incomplete contracts then give rise to further transaction
costs: The costs that are incurred in achieving commitment by noncontractual
methods and the ineffi ciencies that result from attempts to protect against imperfect
commitment.
When private information is a factor, there are additional costs. These can be
substantial. They can prevent agreement or delay its realization considerably. I n either
case, simple value maximization computed without regard to the constraints imposed
by incentives is a misleading basis for prediction.
In some bargaining situations, as we have already seen, private information may
completely prevent agreement even though there are signifi cant gains to be had from
trade. Before studying this problem further, we first examine why people may choose
to become privately informed and the special problems this poses.

MEASUREMENT COSTS
AND INVESTMENTS IN BARGAINING POSITION
In many exchanges, some important characteristic or quality of the good being traded
is unknown to both the buyer or seller. For example, when an oil company buys
rights to extract the petroleum on a piece of land, neither the company nor the seller
typically knows how much oil will be found. Similarly, neither the purchaser or the
seller of a piece of equipment, or both, may be unsure about how durable the
equipment may be. Although it is conceivable that such a situation could sometimes
be unproblematic-the parties might make a deal based on expected values-that is
not always the most likely outcome. M ore often, it is possible and profi table for one
or both parties to gain a bargaining advantage by investing in information about the
quality of the good. For example, by becoming informed the buyer can better avoid
paying a higher price for the good than what it is actually worth. M oreover, if one
side has become informed, it is very likely to be in the other side' s interest to absorb
the costs of becoming informed as well because otherwise it will be subj ect to adverse
selection or opportunistic misrepresentation. Thus, both sides may be led to incur the
costs of measuring the good's worth.
These information costs are wasted from the standpoint of society as a whole.
It is true that the information gathering will lead to a negotiated price that reflects the
value more precisely, but that does not increase total wealth. A different price benefi ts
one party only at the expense of the other. The good still ends up being transferred,
j ust as efficiency required and j ust as it would have been had the information not
been gathered. The costs of gathering that information are subtracted from the parties'
total wealth; there is no offsetting benefi t unless the information affects a productive
decision or the allocation of the asset. In line with the value maximization principle,
 148
Motivation: you might therefore expect businesspeople to try to devise institutions that reduce this
Contracts, source of waste.
Information, and
Incentives The De Beers Diamond Monopoly
The De Beers group, which runs the remarkably successful world diamond cartel,
uses an unusual method to sell diamonds that has been interpreted in just this way­
as an attempt to reduce excessive expenditures on information. 10 Buyers indicating an
interest in purchasing particular amounts of particular sizes and qualities of stones are
offered a packet of stones, called a "sight," that roughly corresponds to their indicated
interests. The stones are graded and sorted on the basis of their gross characteristics
only, without any attempt by De Beers to estimate their value closely. The sights are
offered on a strict take-it-or-leave-it basis, according to a pricing formula based on
these gross characteristics. Absolutely no negotiation over the price is permitted. Also,
the content of the packet is not negotiable (except for correcting errors in grading).
Furthermore, failure to take the sight as offered results in the buyer's being excluded
from buying from De Beers in the future.
Although elements of this arrangement may represent De Beers' exploitation of
its monopoly position (it controls over 80 percent of the world diamond supply), there
is little reason to expect that even a monopolist would choose an inefficient distribution
scheme. From a value-maximization perspective, it makes sense to look for a competing
explanation of these details in terms of efficient exchange.
Just such an explanation has been offered. If bargaining about the price and
contents of a packet were allowed, then there would be incentives for the buyers to
spend resources in finely examining each stone to determine the most valuable gem
that can be cut from each and to estimate the values of the uncut stones that way.
De Beers would have to do the same to protect itself; that is, it would have to abandon
its system of using only rough grading of the uncut stones. To minimize examination
costs and thereby maximize total value, however, a careful examination should be
made only by the final buyer, who must actually cut the stone, and not by the seller
or the other potential buyers. Moreover, it does not much matter which diamond
retailer purchases which stone. What is most important for efficiency is that the
transaction be made and that wasteful information gathering be minimized. Forbidding
negotiations eliminates the incentive for buyers to incur examination costs, and the
practice of withdrawing the opportunity to buy in the future prevents the "leave it"
option from being used as a bargaining device to try to improve future offers. Moreover,
buyers benefit from knowing that the stones they are purchasing have not been "picked
over" by other potential buyers, so that examining the stones is not necessary for self­
protection. Both sides enjoy savings from this system of trade.
Similar arguments have been advanced to explain the practices of prepackaging
fruits and vegetables in grocery stores rather than selling them individually, of supplying
warranties with products, of "block-booking" a set of movies from a given studio rather
than allowing theater owners to pick and choose the films they want, and of paying
percentage royalties rather than a fixed fee to book authors. In each case, the specified
practice reduces or eliminates the incentives to incur costs inspecting individual items
even when the efficient outcome is for trade to occur. The practices are then interpreted
as examples of the value maximization principle.

IO Roy Kenney and Benjamin Klein, "The Economics of Block Booking," Journal of Law and
Economics, 26 (1983), 497-540.
 1 49
I nvestin g in Bargainin g Advantages Bounded
These measurement costs are an example of the larger class of expen ditures made to Rational ity and
gain an advantage in bargaining or trade. Many such expenditures are typically Private I nformation

unproductive, yet they are incurred because they have private value in shifting the
distribution of the gains from trade. Another example comes from the international
arms control arena. In the U nited S tates there have been repeated arguments for
fu nding of enormously expensive weapons systems and the S trategic Defense Initiati ve
("Star Wars") on the basis that they provide " bargaining chips" in negotiations with
the S oviet U nion on arms control. The idea seems to be that these systems are being
developed and the corresponding expenditures made not because deployment of the
systems is actually worthwhile but so that they can be bargained away and leave the
U nited S tates with an adequately mighty arsenal.
· These investments in bargaining position carry direct costs that the bargainers
may wish they could all avoid. In this regard, the time and energy so often spent
haggling, posturing, and delaying agreement in attempts to influence the terms of the
deal are related wastes. M oreover, to the extent that the potential contracting parties
forecast that these costs will be incurred, they may be forestalled from ever attempting
to reach an agreement.

ADVERSE SELECTION
Besides the incentive to misrepresent valuations, a second sort of incentive problem
arising from precontractual informational asymmetries is known as adverse selection.
The term was coined in the insurance industry. The selection of people who purchase
insurance is not a random sample of the population, but rather a group of people
with private information about their personal situations that makes it likely they will
receive a higher-than-average level of benefi t payments under the insurance policy.
For example, if an insurance company were to issue an individual health insurance
policy that covers the medical costs associated with pregnancy and delivery, it is a safe
bet that the policy would be purchased disproportionately by women planning to bear
children in the near fu ture. Childbearing plans are a privately known, u nobserved
characteristic of the insurance buyer that has a huge effect on insurance costs. The
offering of comprehensive individual pregnancy and delivery coverage is subj ect to
such severe adverse selection that such coverages are no longer available in the U nited
S tates, where most health insurance is privately provided.
In many countries, the inadequacy of private health insurance has led govern­
ments to nationalize the provision of health care, although with mixed results. In the
U nited S tates, the private sector has responded by developing a new practice-group
health insurance-that partially substitutes for the missing market in health-insurance
contracts. In modern practice, most coverage for pregnancy and delivery benefi ts is
supplied through employer- provided medical i nsurance plans, where coverage is
automatic or compulsory for all members of the employee group. Because participation
is not voluntary and because pregnancy and delivery benefi ts are packaged with other
health-care benefi ts, the insurance company is able to insure a cr oss-section of the
community and thus avoid adverse selection.
A second example of adverse selection arises in connection with warranties on
automobiles. New cars typically come with a warranty under which the manufacturer
pays for any problems (other than routine maintenance) that arise with the car over
a certain period of time, such as one year or 12, 000 miles. S ome auto makers have
experimented with selling an optional extended warranty covering nonroutine expenses
over a longer period, such as fi ve years or 50, 000 miles. Insurance theory predicts
 1 50
Motivation: that the extended coverage would most often be purchased by people who expect to
Contracts, put their cars to hard use, such as driving on poorly maintained roads while towing
Information, and a trailer or other load in extreme weather conditions. Those who expect to use their
Incentives cars to carry passengers only, on well-maintained roads and in ordinary weather
conditions, would be less likely to purchase the extended warranty. Notice that an
extended warranty that provides coverage for all buyers is relatively much less
susceptible to this problem, and auto makers have recently been adopting such
coverages. Thus, even though extended warranties have not been an especially popular
option, several automobile manufacturers have begun to include long warranties as a
standard feature on their cars. This universal coverage works better than individual
coverage for the same reason that preg nancy coverage offered through employer­
pr ovided health coverage is economically viable whereas pregnancy benefits in
individual policies are not.
Adverse selecti on is a pr oblem of precontractual opportunism; it arises because
of the private information that the insurance consumers have before they have
purchased the insurance contract, when they are weighing whether the purchase is
beneficial. A woman considering a health-i nsurance policy with generous benefits for
labor and delivery has private information about her family plans. A driver who buys
extended warranty coverage has private information about how the car will be used.
These problems are additional to the moral hazard pr oblem, according to which the
availabil ity of insurance will actually change the insured' s unmonitored behavior. I t
is moral hazard-not adverse selection-if a family with insurance coverage for the
expen ses of pr�gnancy and childbirth is induced to have more children or to seek
more extensive (an d expensive) prenatal care, or if a car owner with an extended
warranty i s induced to take less care of the car.
Adverse selection is incompatible with the neoclassical account of markets from
Chapter 3 . There, in essence, all trades are with an impersonal market, and no one
cares who actually takes the other side of the transaction. This is not the case with
adverse selection. I n the neocl assical model, each firm has a definite set of technically
feasible plans. Each plan is described in terms of the things the firm must buy and
sell in order to carry i t ou t. When there i s adverse selection, the resources needed to
provide insurance depend not only on h ow much i nsurance is sold but also on the
unobserved characteristics of the buyers. The firm cannot verify the technical feasibility
of its production plan without knowing the mix of unobserved characteristics its
customers will have. These depend not only on the firm's plan, but also on the
insurance products offered by its competitors, on the prices of its products, and on
the mix of characteristics in the population at large. In the neoclassical model, none
of these things are permitted to affect the technical feasibility of a pr oducti on plan.

Adverse Selection and the Closing of Markets

·w hen the problem of adverse selection is especially severe, there may be no price at
all at which the quantity of a good supplied to the market by sellers is equal to the
quantity demanded by buyers. The problem is that the price must be the same to all
buyers, no matter what the costs of serving them, because the costs are not observable
by the seller. H owever, the only buyers who will pay any given price are those whose
pri vate information leads them to believe that the price is advantageous for them.
These wil l tend to be those who-like the women pl anning to have children and
co nsideri ng i nsurance wi th materni ty benefits-are most expensive to serve. I f there
arc any adm inistrati ve costs i n selling the product, th en the price will have to rise so
high for the seller to break even that n ot even those val uing th e product the most will
find i t worthwhi le to bu y. Any lower brea k-even price will a ttract only th ose wh o cost
more to serve tha n the price. Thus, the market completely collapses, an d it may do
 151
so even when there would be gains from trade in the absence of private information .
For example, the costs of buying private maternity coverage would be prohibitive,
Bounded
Rationality and
exceeding the cost of paying directly for the medical bills. Private Information

A I\IATI IEI\IATICAL EXAMPLE OF ADVERSE SELECTION Suppose that a company offers

an insurance policy for sale. Buyers' differing characteristics imply that they will
receive differing expected benefits under the policy, so let x denote the expected
benefit or claim payment. If the characteristic x were observed, then the insurance
company would charge a higher price to consumers with higher values of x to offset
its higher expected costs. We assume that x cannot be observed by the insurance
company, however, so that the same price P must be offered to all potential purchasers.
Suppose that, in addition to the expected receipts x, a buyer gains some value v from
the pure risk reduction that the insurance pol icy provides. A buyer will purchase the
insurance policy if the price is less than the value received: P :5 v + x. The set of
buyers who · will purchase at a price P therefore is all those whose characteristic x
exceeds P - v.
Now we come to a critical point: How does the insurance expense that the
company expects to pay depend on the price P? If there were no adverse selection,
the cost of insuring an average customer would not depend on the price charged , any
more than the cost of producing an average toaster depends on the selection of people
who purchase it. Insurance, however, is different from toasters. For an illustration ,
let us suppose that the distribution of claim amounts x in the population at large is
uniform between zero and i, so that there are an equal number of people with any
given level of expected claims x. Given this simplifying assumption, the average
insurance expense is equal to the average of the amounts paid to the consumer with
the lowest value of x who buys insurance (x = P - v) and the consumer with the
highest value of x who buys insurance (x = i); this average is equal to (P - v + i)/
2. Because the people who expect relatively low collections from insurance are
deterred from purchasing as the price is raised, and only those for whom x :2::: P - v
are willing to buy, the average amount of the insurance expense for those customers
who do buy is an increasing function of the price P, as the formula shows .
To see how this effect alters market outcomes, we describe the outcome in this
hypothetical insurance market by a pair (x, P), where P denotes the market price and
x the expected level of claims of the marginal customer. That is, customers who
expect losses greater than x buy insurance, while those who expect lower losses do
not. The average level of claims in this market is (x + x)/2 per customer. We suppose
that the insurance company incurs a claims administration cost of c for each dollar
of claims that it pays, so the average level of costs incurred by the insurer is
(x + i)( l + c)/2 . Thus, the insurance company is willing to supply insurance to the
group of buyers with losses of x or more if the price it receives is at least equal to the
Ps(x), where:
Ps(x) = ½(x + i)( l + c) (5. 1)
Also, the buyers will be just those whose expected claims are at least x if the price is:
Pb(x) = x + v ( 5 . 2)
These two equations are plotted in Figure 5 . 1 , using the assumptions that i( l + c) >
i + v, and that x( l + c)/2 > v, that is, that ci > v. As the Figure shows, there is
then no intersection of the two lines. For any given level of the price on the vertical
axis, moving horizontally to the Pb line identifies the corresponding level of expected
claims of the marginal insurance buyer. Then, moving vertically to the Ps line shows
the price Ps(i) the insurer would need to break even selling to that set of buyers. Since
 152
x(1 + c)
Motivation:
Contracts, V + X
Information, and
Incentives

½ x(1 + c)

Figure 5. 1 : With adverse select10n, there may

"
X x be no price at which the insurer is willing to
serve the set of customers who wish to buy in­
Expected Claims surance.

the resulting vertical move is always upwards in this example, there is no price that
the insurer could charge that would break even, given the set of buyers that the price
attracts. A market equilibrium in this case involves the insurer setting the price so
high that nobody wants to buy, that is, the market breaks down.
The key condition for market breakdown is that ci > v. For an insurance market
to exist, the cost to the insurance company of administering the claims of the claimant
with the highest expected benefits (which are ci) must be no more than the value (v)
that claimant expects to enjoy from the risk reduction. According to our analysis, for
a market to exist, it must be economical to provide insurance to those with the highest
expected claim payments.
Because providing insurance is costly, you might wonder whether the market
fails to exist only when it would in the full-information case, that is, only when it is
so costly to provide insurance for other reasons that, adverse selection aside, providing
coverage is uneconomical. If an employer or government could provide mandatory
coverage to the whole population, its average cost per person would be ( 1 + c)i/2, and
the average benefit received by the insured people would be v + i/2, so the average
benefit exceeds the average cost whenever v > ci/2. For the provision of insurance
coverage to increase total equivalent wealth, the cost of administering claims for the
i
average person must be less than the benef t the average person enjoys on account of
risk reduction. Whenever ci > v > ci/2, insurance is desirable (according to the
principle of total wealth maximization) but private individual insurance markets
cannot survive.
The situation modeled is exactly the sort in which you might expect to find
group insurance offered as an alternative to individual insurance. When adverse
selection is not too severe, the advantage of individual insurance over group insurance
is that it allows people to tailor their purchases to their individual tastes. When adverse
selection destroys the individual insurance market, however, group insurance provides
a good private sector alternative. The failure of the market for individual insurance
leads to alternative arrangements and practices in the private sector. In some naive
economic analyses, the failure of simple markets is always a valid reason for government
intervention. As the insurance example illustrates, however, the private sector is
continually innovating practices that avoid market failures, and these practices must
be evaluated in order to decide if a government intervention can improve matters.
The model developed earlier is a variation of the pioneering adverse selection
model of George Akcrlof, who went further to show that the volume of trade in
markets suffering adverse selection is inefficiently low, even in cases where adverse
se lec tio n does not entirely destro y the market. Akerlofs model is widely kno wn as the Bounded
lemons model, named after its original application to the market for used cars where, Rationality and
Akerlof claimed, the poorest quality cars ("lemo ns") are the o nes most often offered Private Information
for sale.
Adverse Selection and Rationing
As we have mentioned a number of times, standard market theory is based partly o n
the premise that prices adjust u ntil supply is equal to demand. When demand exceeds
supply, the sellers will find that they can raise prices without lo sing sales. When
supply exceeds demand, buyers can profit by holding out for a lo wer price. In that
way, prices are dri ven by market forces toward the level where supply equals demand.
When there is adverse selectio n, however, changing the price affec ts not only
the revenues of the selling firm bu t also its costs of supplying the product. As we saw
in the insurance example, the average claims mad e against the insurance company
may be an increasing func tio n of the price charged . I n a similar way, the interest
rates that a bank charges can affect the selectio n of customers who apply for loans:
Only those cu stomers with risky investments may be willing to p ay high interest rates.
In that case, the bank may find it unprofitable to raise its interest rate to take ad vantage
of excess demand. Indeed, a higher loan interest rate may actually lead to a lower
return for the bank, net of the co sts of default.
For example, suppose there are two sorts of borrowers, A and B, who are
indistinguishable to the bank. Both have access to a single investment opportu nity
that requires an initial investment of $1,000,000 and which generates an expected
return of 10 percent. Type A's projects are safe: They pay off $1,100,000 for sure.
Typ e B borrowers have risky projects that have equal chances of yielding $900,000
or $1,300,000. Suppo se that if the low payoff occurs, all the bank can collect is the
$900,000: The borrower has no additional collateral. Suppo se too that there are
enough type As in the popu latio n that the bank can make mo ney lending to all
prospective borrowers at 5 percent interest, provided it does not get an ad verse selectio n
of borrowers, and that competitio n leads 5 percent to be the interest rate o n loans.
Both types will be happy to borrow, and the bank in fact does no t suffer ad verse
selectio n.
Now assume that money becomes tighter in the economy and the cost of funds
to the bank increases, so that the bank would need to receive just over 10 percent to
be profitable while lending to a random selec tio n of borrowers. What happens if the
bank raises its interest rate to (j ust over) 10 percent? The safe, type A borrowers will
no longer borro w, becau se they are certain to pay more to the bank than they c an
earn from their i nvestment project. However, the risky type Bs may still be happy to
borrow: If thi ngs go well, they collect $1,300,000, pay the bank $1,100,000, and
make $200, 000. If things go bad ly, they pay the bank only $900, 000, netting zero for
themselves, with the bank losing $100,000. The B s thus expect a net pro fit of $100,000
on average, with the bank experiencing an offsetting expec ted loss. In this example,
the bank's net earnings from its loans fall when it raises the interest rate, becau se the
"quality" of its group of borrowers declines.
When mo ney is scarce, the bank may often find it more pro fitable to use the
opportu nity to improve the quality of its loan portfo lio rather than to charge a higher
rate of interest. Joseph Stiglitz and Andrew Weiss, u sing this sort of reaso ning, have
created an ad verse selec tion model to account for credit ratio ning and no nmarket
means of credit allocatio n. 11 When there is an excess demand for loans, the bank

11 They have also shown that , in �ome circumstances, high interest rates can induce a firm to choose
riskier investments than it otherwise would. The argument is similar to one we will develop in the next
 154
.Motivation: may prefer rationing credi t rather than rai sing the i nterest rate that it charges to
Contracts, borrowers. The same sort of analysi s can be used to explai n why firms use termi nations
Information, and rather than wage cuts when they fi nd their wage expenses are too high: Wage cuts are
Incentives di sproportio nately likely to lead to qui ts by the ablest workers, who frequently have
the best outside job opportuni ties.

SIGNALING, SCREE!\Il\G. Al\D SELF-SELECTION

Sometimes the pri vate information respo nsi ble for adverse selectio n may be di scovered
by expendi ng effort and attention. Aptitude and achi evement tests, for example, may
reveal i nformatio n about prospecti ve employees, as may backgroun d checks and
references from former employers. Some employers even resort to usi ng po lygraph
tests. There are obvious costs and limi tations to these measures, however, which limit
their use and leave co ntracti ng still subj ect to informational asymmetries.
Often i n si tuatio ns marked by preco ntractual pri vate i nformation, some of the
pri vately i nformed peo ple would gai n if they could make their i nformation known.
Moreover, the uni nformed wo uld gai n from learni ng thi s i nformation. For example,
a worker who knows he or she is especially productive would like prospecti ve employers
to know thi s as well, because i t would result i n a better job assignment and higher
pay, and the prospecti ve employers woul d be glad to get the i nformation that would
allow them to identi fy a to p prospect. Similarly, a firm that has low co sts might like
potential competitors to know thi s fact so that they will recognize th at i t will be a
tough competi tor and they will be less likely to enter its markets. A firm that has
developed a new product whose quali ty is exceptionally high but no t easily determi ned
except through extensi ve experience wi th the product has a simi lar i ncentive. It would
like to make the actual quality known to prospecti ve buyers, who wo uld themselves
be pl eased to recei ve the i nformation. (Of course, they wo uld also like to know if the
quality is bad, so that they can avoid the new product. )
The difficulty is that there may be no simple, direct way to reveal the private
information. I n particular, self-servi ng claims are not likely to be credi ble. It is equally
easy to say "I am unusually highly moti vated, energetic, creative, and dedicated, and
you should therefore hire me," or "You better not thi nk about entering our market,
because we are commi tted to i t and will fight fi ercely to defend i t, " or "New S platto
brand is the best ever, so buy it today! " whether the claim is true or false. If such
claims were believed and acted on, everyo ne woul d be tempted to make them, and
those who were foolish enough to believe would soon lear n the error of their nai vety.
However, there is still a mutual gai n to be had if the desired types of tl, e privately
i nformed agents can make their i nformation known. Thi s creates i ncenti ves to fi nd
credi ble ways to convey the information. O ne way is for the uninformed to attempt
to i nfer the private i nformation from observable actions (or verifiable statements) of
the i nformed. This leads to two different, but closely related, classes of strategies.
These are usually called signaling and screen ing; \\'ith the difference between them
dependi ng on whether the i nformed or uni nformed party takes the lead.

Sig naling
I n si gnaling, the privately informed parties take the lead i n adopti ng behavior that,
pro perly interpreted, reveals their i nformatio n. The best-kno\\'n exam ple of labor

chapter to explain why some savings and loan associations (S&Ls) in the Un ited States undertook ,·cry risky
in\'cstmcnts in the 1980s. When these in\'estmcnts failed, the S&Ls became insol\'cnt, leading to the "S&L
crisis" that is costi ng American taxpayers hundreds of millions of dollars. This moral ha:::ard effect is
additional to the adverse selection effect described earlier: Both effects imply that a bank that ra ises its
interest rates will find that the quality of its loan portfolio decli nes.
 1 55
market signaling of productivity vi a educational attainmen t is due to Michael S pen ce, Bounded
who was the first to explore these sorts of problems systemati cally. Rationality and
In Spen ce's sim plest model, workers are either of high or low productivity. They Private Information
pri vately know their own producti vity, or at least are better informed about it th an are
prospective em ployers. High-productivity workers can produce, for example, $50 per
hour of output ne t of other costs of production , an d under competitive labor markets
this is what they would earn if they were kn own to be highly productive. Low­
productivity workers can pr oduce only $20 per hour. If there is no way to distinguish
between them, members of both groups will be paid the same amount, equal to the
expected productivity of a ran domly chosen worker. If the percentage of high­
productivity workers in the population is, say, 30 percent, then the wage will be
(. 3 X $50) + (. 7 X $20) = $29. It is obvious that the high-pr oductivity workers
would like to be recognized because they are being paid less than the value of their
marginal pr_oduct, and employers who are overpaying the l ow- pr oductivity workers
and are unable to identify the high-productivity ones also would like to have the
information revealed.
N ow suppose that, for some reason, high-ability workers choose to acquire a
distinctly higher level of educational attainment (years of schooling, difficulty of the
courses taken, cl ass standi ng) than do low-pr oductivity workers, and further suppose
that educational attainment is observable and verifiable. If this pattern were recognized,
then employers could infer individuals' productivity from their educational achieve­
ments, and education would be a signal for productivity. Educated workers would be
seen to be high-productivity workers and would be paid accordingly, whereas those
who were relatively uneducated would be (correctly) perceived as being of low
productivity and would earn the corresponding wage. The issue then is whether this
sort of situation is sustainable.
The answer depends on two conditions. Fi rst, the level of education that is
taken as signalin g high productivity must be such that the l ow-productivity workers
are unwilling or unable to attain it, even if by acquiring more education they could
mislead employers into thi nking that their own productivity was high and be paid
accordingly. I f this self-selection constraint is not met, then the low-productivity
workers would attain as much education as high-ability workers, and the education
signal would convey no information. The second condition is that failure to obtain
the particular level of education should accurately signal that the person is not highly
productive; it should not be the preferred choice of highly-productive workers. This
is a second self-selection constraint.
For these two conditions to hold, it is clearly necessary that achieving a given
level of educatio n be cheaper for the high-pr oductivity workers than for the low. If
the costs are aligned in this fashion, then there will be a level of education such that
the more able will be willing to pay the costs of acquiring it and the less able will
not, even though gaining this level of education would lead the market to infer that
their ability is high and to offer them higher wages.
The self-selection constraints ensure that the signals are credible. When they
hold, employers can rely on the worker's signal because it will not be in the interest
of those who do not have the unobservable characteristic (high productivity) to mimic
those who do.
A MATHEMATICAL EXMIPLE OF SELF-SELECTION CONSTRAINTS M athematically, the
two self-selection constraints in this example are
$50 - CL X E u < $20
and
$50 - C u x E u > $20
 1 56
Motivation: where CL and CH are the costs (on a basis comparable to the wage ) of a unit of
Contracts, education to the low- and high-productivity workers respectively, and E L and E l l are
Information, and the levels of education chosen by each type. The first inequality says that low­
Incentives productivity workers who mimic the behavior of high-productivity ones by choosing
their level of education are worse off doing so, even receiving the high wage, than
when picking E L, being recognized as low productivity, and being paid the $20 wage.
The second inequality says that high-ability workers are better off signaling than not.
If Cl l :s CL, these inequalities cannot both be satisfied. Only if Cu < CL can the
inequalities be met and education be a credible signal for productivity.
Furthermore, for any C1 1 < CL there will be a level of education that the high­
productivity workers would obtain to signal their private information. For example,
if CH is $ 1 0 and CL is $20, then E 1 1 = 2 and E L = 0 will do. The low-productivity
worker then earns a net of $50 - $20 X 2 = $ 1 0 by mimicking and $20 by not
attempting to pass as a high-productivity worker, and the high-productivity worker is
better off getting the two units of education (thus receiving a net of $30) than not
doing so (and getting $20). Of course, other levels of E l l and E L will also do: For
example, an E H level of a little bit more than 1 . 5 and E L of zero.
SIGN.\LING ,\ND EFFICIENCY Notice that we did not assume that educational attainment
affects on-the-job productivity, although this can be incorporated into the analysis
without altering the conclusion substantially (and doing so helps motivate the whole
analysis). All that is required is a negative relationship between a person's cost of
acquiring education and his or her productivity. According to some, this relationship
might arise because productivity involves the ability to focus on boring tasks, to sit
still and take directions, and to show respect for authority, and these same abilities
are required to succeed in the educational system. It could also arise because inherent
intellectual ability affects success both in school and on the job.
Note too, however, that although the signaling leads to the effective revelation
of the private information, this does not come for free. Because of our assumptions,
there is in fact no social gain to revealing the information. Thus, the expenditure on
education in this example is a pure social waste! Even when we allow that education
may improve productivity directly or that signaling may be valuable because it permits
a better matching of workers to tasks, we will still typically find some excessive
signaling-overinvestment in education-from a full information efficiency point of
view. In fact, as we see in a problem at the end of the chapter, the high-productivity
workers may even be worse off with the signaling than they would be if it could
somehow be eliminated (although they are still better off signaling their private
information than not doing so and being thought to be low productivity).
Signaling has been used extensively to explain a wide variety of phenomena.
Examples include setting low "limit" prices as a signal of low costs to deter entry by
potential competitors; the use of seemingly uninformative advertising with newly
introduced "experience" goods (ones whose quality is not immediately evident prior
to purchase) to signal their quality; the offering of product warranties and money-back
guarantees as a signal of product quality; the paying of dividends by corporations to
signal financial strength, eve11 though dividends are tax-disadvantaged relative to share
repurchases as a means of distributing money to stockholders; and the choice of the
amount of debt versus common stock a firm issues as a signal of its financial prospects.

ScrPPnirw
t,

Screening refers to activities undertaken by the party without private information in

order to separate different types of the informed party along some dimension. This is
often done by offering a variety of alternatives, each intended for one of the
 1 57
various types of informed parties, whose choices then effectively reveal their private Bounded
information. As we mentioned earlier, the self-selection constraints are crucial in Rationality and
achieving screening: People will choose the alternative intended for them only if it is Private Information
the best for them in the set of alternatives offered, given what they know. An early
application of the screening principle is the following one, which provides an
explanation of the positive relationship that has often been noted between earnings
and age or experien ce.
SCREENING ANO AcE/WACE PROFILES It is empirically well establ ished that pay tends
to increase with age and experience. M any factors no d oubt contribute to this. S ome
economic factors that have been suggested are: increasing skill and ability (human
capital) with greater experience, greater responsibility being given to more senior
employees who have demonstrated their abilities, and better matching of individual
talents to j ob assignments as these talents become recognized. There is evidence,
however, that even after accounting for these effects and controlling for them
statistically, a positively-sloped age/wage profile remains: Given two workers with the
same j ob assignment, the same responsibilities, and the same prod uctivity, the older
one will tend to be paid more. Why should this be so?
O ne possible answer is a general social preference or norm for paying old er,
more experienced workers more and younger ones less. The economic forces working
against such a norm would be very strong, however. It would be in the interest of
profit-maximizing firms to make attractive offers to bargain- priced young workers from
firms with such a policy, and it would also be in the interest of these workers to
accept.
An alternative, economic explanation due to J oanne Salop and S teven S alop is
based on screening designed to reduce employee turnover. Having employees resign
is costly for all firms, but it may be especially costly when, for example, the firm
invests significant amounts in training workers. In such circumstances, the firm would
have a special interest in attracting workers who are less inclined to change j obs. The
difficulty lies in identifying those candidates with a proclivity for moving.
The screening solution is to design the employment contract so that only the
desired types of workers are attracted. Suppose the firm offers to pay relatively low
wages initially, with higher- than-market wages after the employee has been with the
firm for an extended period of time. Such a positively-sloped experience-wage profile
screens the population of potential employees because it is more attractive to workers
who intend or expect to stay with the firm than to those who are less inclined to stay
or who know that they will have to quit after only a short time in the j ob and who
therefore see themselves as unlikely to collect the high wages that come with long­
term employment. N ote that if, in the early stages of their employment, workers in
the firm are gaining experience or training that is valuable elsewhere, effective
screening may require paying wages that are initially well below the going market
level in order to offset the value of the training.
PERFORMANCE PAY ANO SCREENING As another example, suppose that the workers or
managers who are likely to be most productive in a particular firm tend also to be
those with the best outside j ob opportunities. Offering a performance-based pay system
amounts to offering a menu of different contracts because it allows employees to
determine their compensation by how hard they choose to work. Ind eed, paying a
wage that is based on measured performance tends to attract and retain the most
productive j ob applicants and to discourage the least productive, to the employer' s
benefit.
Figure 5. 2 illustrates the l ogic of this screening analysis. The workers' ability
and likely contribution to the firm are measured on the horiz ontal axis, whereas their
 158
Motivation:
Contracts,
Information , and
Incentives

>,
<U
a..

w
Figure 5 . 2 : Compared to paying the fixed wage
W*, basing pay on performance attracts more
A' A* A"
productive workers to the firm and discourages
Abil ity and Contribution less productive job applicants.

pay is shown on the vertical. The curved line shows the relationship between expected
productivity on this job and any likely outside job opportunities. Our assumption that
more productive workers tend to have better outside opportunities is represented by
the fact that this curve is upward sloping.
If a fixed salary or wage of W* is offered, workers of ability up to A* will find
it attractive to accept the firm's offer because that wage is more than they could earn
in outside jobs. More productive workers (those with ability above A * ) would decline
the firm's offer. Now suppose the firm were to introduce a performance pay system
that is just sufficient to keep the wages of a worker of productivity A* unchanged.
This is shown by the upward-sloping straight line. Higher performance, measured
along the horizontal axis, results in higher pay. Workers of lower productivity than
A* would be paid less, and some of them would now find it more attractive to accept
jobs elsewhere. According to the figure, the workers with abilities below A' would no
longer be attracted. In addition, workers with abilities in the range above A" who were
not previously attracted to the firm now will be. They now find employment in this
firm attractive because they can expect to earn more under its performance pay system
than in their outside job opportunities.
We will see in Chapters 6 and 7 that basing pay on measured output can also
help resolve postcontractual moral hazard problems within the firm by providing
incentives for workers to perform well, even when it is impossible to monitor how
hard they are working. The monitoring problems make it futile to attempt to contract
directly on the workers' providing a high level of effort. If their pay does not depend
on their output, they then have little direct incentive to exert more than minimal
amounts of effort. However, tying pay to productivity provides incentives to raise
output. This effect reinforces the screening effect of differentially attracting people
who are inherently more productive.

l\lENl 1S OF CONTH,\CTS .\ND EFFICIENCY The logic of screening is also helpful for
analyzing the decisions of a company that is setting prices for a whole line of related
products, such as office copiers or computers. In this case, a menu of options is
offered to customers, who must choose among the various products on the basis of
features and price. The prices and features for each model must be chosen recognizing
that they will affect the demand for other products in the line. For example, a low
price on the most basic model will attract customers who would otherwise buy a more
expensive model with more features. The mathematical expression of the problem is
for the seller to set prices and target customer groups to maxi mize some objective,
such as profits, subject to a set of self-selection constraints. These constraints express
 1 59
the requirement that prices an d produc t features be se t so that eac h group of customers Bounded
prefers to buy the product targeted for that gr oup rather than some other one. Rationality and
An other application is in offerin g a m enu of contracts to salespeople, where the Private Information
amount of salary and the percentage com mission on sales vary inversely across the
men u. Th is h as been done, for example, at IBM. The idea is that those salespeople
wh o kn ow that sales in their territories will be especially responsive to their efforts will
select high-c om mission, low- salary contracts and be motivated to exert extra effort,
whereas those whose private information indicates that increasing sales will be difficult
will make the opposite choice. The latter group then will face less risk in their inc omes
and so will not have to be paid as much in expectation to compensate for bearing
risk.
Similar considerations enter into the design of insurance c ontracts, where
different policies are designed for different risk classes of buyers. In this context, self­
selection is. achieved by varying the extent of c overage offered. This may result in
low-risk customers receiving less than full insurance coverage, whereas high-risk
individuals purchase full c overage (but pay a higher amount per unit of insurance).
The lack of ful l insurance for low-risk individuals is a cost of the informational
asymmetry.
I n general, achieving self-selection via screening will help overc ome the
informational asymmetry, but may be costly and will still leave opportunities for value
creation unrealized because they are unattainable under the incentive constraints.

IMPLICATIONS
Bounded rationality and prec ontrac tual private information mean that c ontracts are
incomplete and, at best, only c onstrained efficient. These represent maj or c osts of
market-like, arms'-length transactions.
In certain situations, some of these c osts can be avoided by bringing the
transaction under unified governance. We saw this in the hold-up problem: If a single
firm owns both the cospecialized assets, the problem disappears. Similarly, the
inefficiencies of bargaining between a firm and a supplier over provision of an input
can be mitigated by the firm's integrating vertically to produce the input itself. There
is no longer a need for contracting, there are no c ommitment problems with a single
decision maker, and the private information that may have marked the negotiations
between the supplier and the firm is not an issue any more. This solution can create
other difficulties, however.
For example, if a firm chooses to make some input for itself in order to avoid
the problems of bargaining with a supplier, the firm' s owner may not have the time
or the necessary expertise to supervise the input's production. The owner must then
hire a manager. But with his or her limited time and expertise, the owner may not
be able to tell if the manager is doing his or her j ob well. This opens the possibility
of postcontractual opportunism, with the manager not working as hard as the owner
would like or manipulating reports to his or her advantage. Chapter 6 deals with this
phenomenon of moral hazard.
 160
Motivation:
Contracts,
Information, and
Incentives SUMMARY
Because real people are only boundedly rational, complete contracts that specify what
they will do in every conceivable circumstance are impossible to negotiate and write.
People's ability to make plans and contracts are limited by the existence of unforeseen
circumstances, the costliness of deciding in advance what would be optimal to do in
every foreseeable contingency, and the difficulty of writing down descriptions of
contingencies and actions with enough precision to make writing such a contract
worthwhile.
To respond to these difficulties, people write contracts that are quite different
from those portrayed in ordinary market theory. They enter into relational contracts
that specify procedures to govern how decisions will be made and disputes resolved
and leave unspecified much of the substance of the relationship. Some of the con­
tract terms are implicit and largely unenforceable in courts of law, representing
understandings that exist and change with changing circumstances. These understand­
ings are enforced by a reputation mechanism: Those who violate the understandings
lose their ability to command the trust of others.
These responses, however, are only an imperfect substitute for complete contracts
and do not prevent people from behaving opportunistically to seize a greater share of
the fruits of the organization for themselves. Despite attempts at commitment, people
sometimes renege on promises or renegotiate deals in a way that undermines the
initial intent of the contract. Fear of suffering losses when a trading partner reneges
may make otherwise profitable business deals sour or may cause firms to spend
resources protecting themselves from opportunism by others-expenditures that are
socially wasteful because they do not create value for society as a whole.
Problems of reneging are especially severe when large specific investments are
required in the relationship. An investment is an expenditure of resources to create
an asset, which is a potential future flow of benefits and services . The specificity of
an asset is the fraction of the investment value that is lost when the asset is switched
from its intended use to its next best use. Highly specific assets create the potential of
a hold-up problem , in which one trading partner reneges on its contractual obligations
to attempt to extract better terms from the owner of the specific asset, whose bargaining
power is impaired by the losses he or she stands to suffer if forced to redeploy the
asset to another use. The inability of one partner to commit not to hold up the other
may prevent valuable investments from being made.
Within the firm, a reputation mechanism may help to overcome this problem.
Employees are encouraged to invest in firm-specific assets, such as knowledge of the
firm's objectives, routines and methods of operations, its products, customers, and
organizational capabilities. The firm may protect these investments, treating employees
well, in order to encourage investment by new workers and continuing investments
by existing ones.
In addition to problems of bounded rationality and commitment, another source
of inefficiency in contracting 1 s private information at the time the contract is being
negotiated. In the simple case of bargaining over the sale of a good, the seller may
have an incentive to exaggerate the cost of producing the good in order to obtain a
higher price and the buyer may have an incentive to understate the good's value,
trying to obtain a lower price. These distortions may prevent the parties from reaching
any agreement at all, even if the value-maximizing outcome would be for trade to
occur.
When these distortions caused by incentives are unavoidable, we must recognize
 1 61
them in deciding which trading arrangements are most efficient. An arrangement or Bounded
mechanism is incentive efficient if there is no other arrangement or mechanism that Rationality and
leads to a higher expected payoff for all parties, given the need to provide incentives Private Information
to the parties to act in whatever fashion is planned.
The costs of providing incentives in bargaining can be significant even when
there are onl y two bargainers, but they grow even larger when there are many. The
problem is a free-rider problem: Any one bargainer may fi gure that a misrepresentation
or refusal to contribute will have little effect on the final decision but will keep his
or her costs of participation low. This effect grows with the number of bargainers and
makes multiparty bargaining an especially difficult matter.
Another kind of cost that parties suffer during bargaining are measurement costs,
which are the costs incurred to obtain an informational advantage. Institutions and
practices can often be designed to minimize the opportunities for measurement or to
curtail the advantages that it confers. For example, the unusual practices used by De
Beers in the wholesale marketing of diamonds seem designed to reduce both the
opportunity for and value of measuring the quality of diamonds.
Private information can also operate to block the efficient functioning of ordinary
markets when the cost of providing services depends on privately known characteristics
of the buyer or when the benefits of purchasing depend on similar characteristics of
the seller. For example, the cost of providing pregnancy and maternity insurance
benefits depend in part on the private plans of the purchaser regarding childbearing.
The result is that the premium charged affects the selection of customers who buy,
and this adverse selection affects the seller's costs. For example, a high-priced pregnancy
and maternity plan would attract onl y buyers who planned a pregnancy soon, raising
the insurer's average cost. In this case, the consequence of adverse selection is that
there is no effective private market for individual pregnancy and maternity coverage.
Similarly, i n banking, raising the i nterest rate charged to borrowers may only lead to
a poorer quality of loan applicants. Banks may instead respond by rationing credit,
seeking to raise the average quality of the loans they make.
Besides rationing, the other responses in markets to selection problems include
screening and signaling. Signaling is the attempt by certain individuals to communicate
their private information in a credible way. For example, talented workers may acquire
education as much for the value of the credential as for its direct contribution to their
productivity. Screening is the design of products or contracts to attract only a desirable
portion of potential customers, workers, and so on, or the design of menus of contracts
to sort these parties into distinct groups. For example, offering some jobs with incentive
pay schemes and others with fixed wages may tend to sort the workers. With each
group of workers seeking the job with the highest likel y pay, the most productive
workers will tend to select the incentive pay jobs while the others opt for fixed wages .

• BIBLIOGRAPHIC NOTES
Nobel Laureate Herbert Simon has been the major proponent of recogni zing the
implications of bounded rationality for economic behavior. In particular, the
explanation of the employment relation as a response to bounded rational ity is
due to Simon. Oliver Williamson has built extensively on Simon's insights. Victor
Goldberg and Williamson have, in particular, emphasized relational contracting,
and this concept has also received significant attention from legal scholars. The
survey by Ol iver Hart and Bengt Holmstrom contains a valuable discussion of
the implications of incomplete contracting. Williamson and Benjamin Klein,
Robert Crawford, and Armen Alchian were the first scholars to emphasize and
elaborate the great importance of the hold-up problem for the analysis of business
 1 62
�lotivation: institutions and practices. The importance of the general issue of commitment
Contracts, and renegotiation, both in contracting and in other institutions, has been
Information, a nd repeatedly emphasized by Jean Tirole.
Incentives The importance of precontractual informational asymmetries has been one of the
major themes of economic research for the last two decades. Early important
contributions came from George Akerlof (the "lemons" problem with adverse
selection), Michael Spence (signaling), Michael Rothschild and Joseph Stiglitz
(adverse selection and screening in insurance markets), and Stiglitz and Andrew
Weiss (rationing as a response to adverse selection in credit markets). The
revelation principle was developed independently in various articles by a number
of scholars, including Allan Gibbard, Roger Myerson, Milton Harris and Arthur
Raviv, and Robert Rosenthal. Myerson and Mark Satterthwaite characterized the
effects of incentive constraints with private information on the possibility of
achieving value-maximizing trade. The effect of increasing the number of
participants was developed by Rafael Rob, George Mailath and Andrew Postle­
waite, and V. V. Chari and Larry Jones building on earlier work by Roberts.
The first complete signaling analysis of entry deterrence is due to Milgrom and
Roberts. Building on ideas of Philip Nelson, we also developed the model of
prices and uninformative advertising as signals of product quality. Sanford
Grossman developed the analysis of warranties as signals of quality. The first
model of dividends as a signal is due to Sudipto Bhattacharya. George Baker,
Michael Jensen, and Kevin Murphy give an exposition of the use of performance
pay as a screening device. Joanne Salop and Steven Salop developed the screening
model of the age/wage profile. An exposition of pricing under self-selection
constraints is given by Robert Wilson.

• REFERENCES
Akerlof, G. "The Market for Lemons: Qualitative Uncertainty and the Market
Mech anism," Quarterly fournal of Econom ics, 84 ( 1 970), 488- 5 00.
Baker, G. , M. Jensen, and K. Murphy. "Competition and Incentives: Practice
vs. Theory, " f ournal of Finance, 43 ( 1 988), 59 3-6 1 6.
Bhattacharya, S. "Imperfect Information, Dividend Policy, and 'The Bird in the
Hand' Fallacy," Bell fournal of Econom ics, 1 0 (Spring 1 979) , 2 59-70.
Chari, V. V. , and L. Jones. "A Reconsideration of the Problem of Social Cost:
Free Riders and Monopolists, " Discussion Paper, Northwestern University, 1 99 1 .
Gibbard, A. "Manipulation of Voting Schemes: A General Result," Econometrica,
4 1 ( 197 3), 5 87-602.
Goldberg, V. "Relational Exchange: Economics and Complex Contracts,"
American Behavioral Scientist, 2 3 (January/February 1 980), 3 3 7- 5 2.
Grossman, S. "The Informational Role of Warranties and Private Disclosure
About Product Quality, " Journal of Law and Econom ics, 24 ( 1 98 1 ), 46 1-83.
Harris, M. and A. Raviv. "Allocation Mechanisms and the Design of Auctions, "
Econometrica, 49 ( 1 98 1 ), pp. 14 77-99.
Hart, 0. , and B. Holmstrom. "The Theory of Contracts, " Advances in Economic
Theory: Fifth World Congress, T. Bewley, ed. (Cambridge: Cambridge University
Press, 1 987).
Klein, B. , R. Crawford, and A. Alchian. "Vertical Integration, Appropriable
Rents, and the Competitive Contracting Process, " fournal of Law and Economics,
2 1 ( 1 978), 297- 326.
 Ma ila th, C. a nd A. Postle waite. "Asymmetric Bargaining Pro ble ms with Many Bou nded
Agen ts," Review of Econom ic Studies, 57 ( 1 990), 3 5 1-67. Rational i ty a nd
M i lgrom, P. a nd J. Roberts. "Limit Pricing a nd Entry Under In complete Private I n forma tion
Informa tion: An Equ ilibrium Analysis," Econometrica, 50, ( 1 982), 443-59.
Myerson, R. B. "Mecha nism De sign by a n Informed Principal," Econometrica,
5 1 ( 1 983), 1 767-97.
Myerson, R. B . , a nd M. Sa tterthwaite. "Effi cient Mecha nisms for B ila tera l
Trading," Journal of Economic Theory, 2 3 ( I 983), 265-8 1 .
Nelson, P. "Ad vertising as Informatio n," Journal of Political Economy, 84 ( 1 974),
729-54.
Rob, R. "Pollu tio n Cla im Settlements U nder Private Information," Journal of
Econom ic Theory, 47 ( 1 989), 307- 3 3 .
Ro berts, J. "The I ncentives for Correct Revela tion of Preferences and the Nu mber
of Consumers," Journal of Public Econom ics, 6 ( I 976), 3 59-74.
Rosenthal, R. W. "Arbitratio n of Two-Party Disputes U nder U ncertainty," Review
of Economic Studies, 45 ( 1 978), 595-604.
Rothschild, M. a nd J. Stiglitz. "Equilibrium in Competitive Insura nce Markets:
An Essay on the Eco nomics of Imperfect Informa tion," Quarterly Journal of
Economics, 80 (November 1 976), 629-49.
Sa lop, J. and S. Salop. "Self-Selectio n a nd Turnover in the La bor Market,"
Quarterly Journal of Economics, 90 (November 1 976), 6 1 9-28.
S imon, H . A. "A Formal Theory of the Employment Rela tio nship," Econometrica,
19 ( 1 9 5 1 ), 293-305.
S pence, A. M. Market Signalling: Information Transfer i n Hiring and Related
Processes (Cambridge, MA: Harvard U niversity Press, 1 973 ).
Sti glitz, J.E. and A. Weiss. "Credit Ra tioning in Markets with Imperfect
Informatio n," American Econom ic Review, 7 1 ( 1 9 8 1 ), 393-409.
Tirole, J. The Theory of Industrial Organization (Cambridge, MA: The M IT
Press, 1 988).
Williamso n, 0. Markets and Hierarchies: Analysis and Antitrust Im plications
(New York: The Free Press, 1 97 5 ).
Williamson, 0. The Economic Institutions of Capitalism (New York: The Free
Press, 1 98 5 ).
Wilson, R. Nonlinear Pricing (Oxford U niversity Press, 1 992).

EXERCISES

Food for Thought

1. We are all familiar with the experience of feeling anger when we have been
cheated or otherwise mistrea ted, a nd ma ny of us ha ve go ne ou t of our ways to exact
retribu tion for such inj u stices, even when we expect never to see the malefactor again
and wc do not expect a nyone else to be a ware of our pu ni shing the tra nsgressio n.
Apparently, we have a taste for vengea nce. Most of u s also seem to ha ve a conscience,
a preference for behaving honestly a nd fairly. Indeed, Ro bert Fra nk [Passions Within
Reason: The Role of the Emotions (New York: W.W. Norton, 1 988)] has suggested
that early huma ns who had such preferen ces would have had a n evolu tio nary
adva ntage, so tha t such preferences should have become widespread as people who
 16-1
Motivation: were biologically "hard-wired" with these traits were more successful in competing to
Contracts, reproduce. What effect would the possibility that people might have such preferences
Information, and have on the analyses in this chapter of commitment and bargaining?
Incentives 2 . In common parlance, we often speak of "signaling intentions . " For example,
a firm might cut prices to signal its intent to compete fiercely to defend a market
against a competitor's expansion. The idea is to deter the competitor, who would not
want to challenge the first firm if it will react aggressively. The difficulty is that if the
competitor does sink costs in expanding, the first firm may not find it worthwhile to
fight. There is thus an issue of credibility of the signal. This problem arises whenever
the signaler still has an option after the fact about what actions it will take. What
means are available to give credibility to signals of intent?
3. One of the most rapidly growing health concerns of much of the developed
world is the spread of Acquired Immune Deficiency Syndrome (AIDS), with the
number of reported victims growing annually at alarming rates. One of the few
effective drugs to alleviate the symptoms of AIDS is AZT. Although this drug is
inexpensive to produce, it is sold for a high price, earning large profits for its
manufacturer at the expense of the already suffering AIDS victims. As a result, there
has been a public outcry to void the patent on AZT and to allow it to be produced
competitively, at lower cost. What damage might such an action cause? What
protections exist for an inventor that might keep a government from seizing intellectual
property? Are there any practical alternatives available to deal with the very real
medical-social problem that avoid the damage you have identified?

Analytical Problems

1. In the model of an insurance market with adverse selection, suppose that

the parameters entail ex < v < x( l + c)/2. Show graphically that there is a price at
which the insurer breaks even and some people are willing to purchase insurance.
What is the formula for that price?
2. Suppose that, in the context of the example in the text of the Spence model
of education as a signal, the high-ability workers are relatively more numerous, making
up 80 percent of the work force, but that the costs of acquiring education and the
productivities are as in the text (that is, is C I I = $ 1 0 and CL = $20, while wages are
$ 50 for the high-ability workers and $20 for the low). Determine how large EH, the
level of education chosen by the high-ability workers, must be to be a credible signal
if the level chosen by the low-ability workers is 0. Show that the high-ability workers
would be better off if there were no signaling (for example, if C11 were to rise to $20).
3. The units of two neighbors in a condominium housing development, Able
and Baker, overlook a small courtyard. The board of the condominium association is
considering whether to plant a tree in the courtyard. The cost is $40. Everyone agrees
that Able values having the tree either at $ 1 00 or at $0. The probability that the tree
is worth $ 1 00 to Able is p, a number between zero and one. Similarly, Baker values
the tree at either $70 (which occurs with probability r) or $0. Able and Baker are the

of them how much having the tree planted is worth to him or her personally. If both
only people who will be able to see and enjoy the tree. The board decides to ask each

equally between the two neighbors. If only one of them expresses a desire to have the
say that they value the tree then the board will have it planted and the cost shared

tree planted, then it is still planted but that person is charged the whole cost. If neither
claims the tree is desirable, it is not planted.
Derive the incentive compatibility constraints for each individual, and solve for
the values of p and r that make it incentive compatible to offer to help pay for the
 165
tree exactly when it is actually valuable to the individual concerned. Now let qA be Bou nded
the probability that the tree is planted when only Able offers to pay, and q 8 be the Rationality and
probability that it is planted when only Baker offers to pay. If p = 0. 7 and r = 0. 8, Private Information
what are the largest values of qA and q8 consistent with correct revelation of willingness
to pay?
4. A used-car salesperson and a potential buyer negotiate over the price of an
old car. The seller believes that the buyer values the car at $ 500 with probabi lity x
and at $ 1 , 000 with probability ( 1 - x). The buyer believes that the car is worth $2 50
to the salesperson with probability ( 1 - x) and $750 with probability x. Under what
circumstances should trade occur? How does the probability of efficient trade depend
on the value of x? If a mediator who wants to maximize the expected gains from trade
specifies that trade should occur at a price of p when the seller claims a value of $2 50
and the buyer claims a value of $ 1 , 000, derive two conditions on p that ensure that
both parties. will report honestly. Solve for the value of x that makes both incentive
constraints exactly hold as equalities. Note that for lower values of x, trade is more
likely to be efficient but it cannot always be attained. Explain why.
5 . A seller values a good at either $0 with probability x or $ 1 0 with probability
( 1 - x). The buyer does not know what the good is worth, but knows that he values
the good at $ 1 0 more than the seller does. Thus, it is certainly efficient for trade to
occur, no matter what the seller's actual valuation of the good. Show that it is
compatible with individual incentives for trade always to occur at a price of $ 1 0, no
matter what value the seller announces. Suppose now instead that the good is worth
only $5 more to the buyer than the seller. Is it possible to find prices (as functions of
the seller's claimed value) so that the seller will honestly report her true value and
both sides will be willing to trade at these prices? Why or why not?
6. Suppose X is an intermediate good in the production of a final good Y.
Firm A has a patent on good X, which it produces at zero marginal cost. Firm B is
considering buying a machine which it can use to turn up to 1 0, 000 units of X into
Y on a one-for-one basis. The machine costs $30, 000, once purchased it cannot be
resold, and it wears out completely once 1 0, 000 units of Y are produced. Firm B can
sell good Y for $5 per unit. Firm A offers to supply good X to firm B at $ 1 per unit
once B has bought the machine.
Suppose A and B can write a binding contract setting the delivery price of good
X. Will B buy the machine? What profits are earned by each firm? Suppose instead
no such contract is possible. Is A's promise credible? Will B i nvest in the machine?
What are the resulting profits?
Now suppose that A can enter into a licensing agreement with another firm,
C, under which C can produce good X, incurring a marginal cost of $c per unit.
Suppose that A gives C the lice�e, and that the cost, c, is $ 1 . 0 1 . Is A's price promise
to supply B at a price of $ 1 now,credible? Does A gain from giving away its technology?
Explain why or why not. Suppose the cost c were higher than $ 1 . 0 1 , but less than
$2 . What would you expect to happen now? What if c were less than $ 1 ?
 6
MORAL HAZARD
AND PERFORMANCE INCENTIVES

IT] ncentives are the essence of economics.

Edwa rd P. Lazea r 1

The problem with corruption is tha t it tends to become the Problem of

Corruption. Moral issues usually obscure practica l issues, even when the moral
question is a rela tively small one a nd the practical ma tter is very grea t.
2
fa mes Q. Wilson

Suppose that yo u are traveling along a highway when a dashboard l ight comes on
indicating that your car is overheating. There is a service station nearby, so you drive
your car there. The traditional analysis of this market situation is simple: There is
some price to be paid to repair the problem; either you pay it and the car is fixed, or
you decide to take your chances and decline to get the car serviced.
That account might reflect what would happen, but there are other possibilities.
After beginning to work on the car, the mechanic might say, "Your radiator is shot.
A new one will cost $500 . " If you are like most drivers, you have no idea whether
the mechanic is being truthful or not. After all, it may be in the mechanic's interest

1 "Incentive Contracts, " in The New Pa/grave: A Dictionary of Economics; Vol. 2 J . Eatwell, M .
Milgate and P . Newman, eds. (London: The Macmillan Press, 1987), 744-48.
2 "Corruption Is Not Always Scandalous, " in Theft of the City: Readings on Corruption in America,
J. A. Gardiner and D. J. Olsen, eds. (Bloomington: Indiana University Press, 1968), p. 29. Copyright ©
1968 by the New York Times Company. Reprinted by permission.
1 66
 1 67
to sell you a radiator, especially at that price. You face the same problem that decision Moral I Iazard and
makers in organizations frequently face: When those with critical information have Performance
interests different from those of the decision maker, they may fail to report completely Incentives
and accurately the information needed to make good decisions. If the mechanic is lying
to you, then both your interests and society's interest in efficiency are harmed. Your
interests are harmed because the mechanic is profiting at your expense, and society's
interests are harmed because productive resources have been wasted: A radiator that
could have been repaired has instead been discarded.
Suppose that you agree to buy the radiator, wait an hour while it is installed,
and then proceed down the highway another 1 00 miles toward your destination. You
notice the overheating indicator on your dashboard lighting up again. Pulling into
another service station, you learn that the new radiator was not installed correctly and
it will cost you another $3 5 and another hour of waiting to have the job done right.
When buyers cannot easily monitor the quality of the goods or services that they receive,
there is a tendency for some suppliers to substitute poor quality goods or to exercise too
little effort, care, or diligence in providing the services. Once again, both you and
society are harmed. You, because you paid and waited twice for the same service,
and society, because resources were wasted: It took two hours of mechanic time (and
waiting time) to do a job that could have been done in one.

THE CONCEPT OF MORAL HAZARD

These problems with the sale and installation of your radiator are examples of moral
hazard, which is the form of postcontractual opportunism that arises because actions
that have efficiency consequences are not freely observable and so the person taking
them may choose to pursue his or her private interests at others' expense. The
possibility of this sort of misbehavior is ruled out in the neoclassical model of markets
considered in Chapter 3 , where it was (somewhat implicitly) assumed that the
transactions that people undertake are simple exchanges of goods and services with
specifi c, well-understood, observable attributes, and that parties to a transaction can
costlessly verify whether the terms of the transaction are being met.

Insurance and M isbehavior

The term moral hazard originated in the insurance industry, where it referred to the
tendency of people with insurance to change their behavior in a way that leads to
larger claims against the insurance company. For example, being insured may make
people lax about taking precautions to avoid or minimize losses. If the necessary
precautions were known in advance and could be accurately measured and recorded,
then an insurance contract could specify which precautions must be taken. However,
frequently it is not possible to observe and verify the relevant behavior and thus it is
not possible to write enforceable contracts that specify the behavior to be adopted.
(The contract could call for the desired behavior, but how could the insurance
company tell if the contract terms had been met?)
The kinds of moral hazard associated with insurance arise frequently in daily
life. For example, you are likely to be much more careful in driving a rented car if
you are fi nancially responsible for all damage to the car than if you have purchased
the Collision Damage Waiver and so are insured against the costs of dents and scrapes.
Similarly, if you are covered under health insurance or belong to a Health Maintenance
Organization (HMO), so that you are insured against all or most of the costs of visits
to the doctor, you are likely to make greater use of medical services of all kinds: doctor
 1 68
Motivation: visits, emergency room visits, prescription drugs, prena tal care, a nd so on. I n each
Contracts, case, the fact tha t you are insured alters your behavior in wa ys that are costly to the
Information, and m surer.
I ncentives
Efficiency Effects of Moral Hazard
Although the term moral hazard has negative connota tions, not all of the cha nges i n
behavi or occasioned by insura nce are socially undesirable, because some social
interests may not be represented in the bargain between the insurer and the insured.
For example, increa sed prenatal visits may result in healthier mothers a nd babies,
consistent with society's goals. M oreover, the insura nce compa ny may not suffer any
losses from moral haza rd if it sets the insurance premium high enough to cover the
extra costs. Still, moral hazard does impair people's a bility to make mutually beneficial
agreements and does often interfere with efficiency.
M oral hazard in insura nce presents a n effi ciency problem because the extra
benefits enj oyed by the insured on account of his or her cha nged behavior will often
not be worth the costs. This happens because the insured decision maker does not
look at all the costs and benefits associated with his or her decisions. M oreover, the
inherent nature of insura nce makes this inevita ble. In the car rental example, you
bear the full costs of the care you exert, but if you are fu lly insured, then being careful
brings you no extra benefits. I n the health care example, you get the benefits of the
ex tra trea tment you seek when insured, but bear little or none of the costs. Thus, you
will tend to go to the doctor for minor ailments whose trea tment seems worth your
time, even if the total costs may exceed the benefits you receive.
The incentives to alter behavior would still not be a problem if it were easy to
determine when the behavior were appropriate a nd to prevent excessive use. This sort
of monitoring is often impossible, however, or a t least very costly. There is no cost­
effective way for the car rental compa ny to observe the care you take. I n the medical
context, you will often lack the expertise to j udge whether a particular visit is necessary,
a nd it will not generally be in the doctor's interest to report that you have made a n
unnecessary visit. For this rea son, moral hazard is a n information problem: The
difficulty or cost of moni toring a nd enforcing appropriate behavior creates the moral
hazard problem. These diffi culties mean that contracting is incomplete because there
is no point to writing a contract specifying particular behavior when the desired actions
ca nnot be observed a nd consequently the contract ca nnot be effectively enforced.
I n terms of sta ndard economics, the insura nce has lowered the cost to you of
something you value (doctor visits, not bothering to exert great care), and you thus
" buy" more. M ore particularly, you now pa y less tha n the full costs of extra units and
so you purchase a n inefficiently large amount. Putting the issue in these terms suggests
that the behavior is not especially wicked a nd that the negative implications in labeling
it "moral hazard" may be somewhat misplaced. M ore significa ntly, it suggests that
this sort of behavicr-a nd the efficiency losses it induces-might be quite widespread.
S uch is indeed the case.

The Incidence of Moral Hazard

M oral hazard problems may arise in a ny situation in which someone (who may be a
supplier, a customer, a n employee, or anyone else) is tempted to take an inefficient
action or to provide distorted information (leadi ng others to take inefficient actions)
because the in dividual's interests are not aligned with the group interest a nd because
the report cannot easily be checked or the action accurately monitored. These problems
are pervasive both in ma rkets a nd in other forms of orga niza tion. Some doctors in
the Un ited S ta tes, for example, in an attempt to protect them selves from malpractice
 1 69
Moral Hazard and
Performance
I ncentives
Hidden Actions or Hidden Information?
Although moral hazard and adverse selection usually seem quite distinct in
textbook discussions, in practice it may be quite difficult to determine wh ich
is at work.
A radio story in the summer of 1 990 reported a study on the makes
and models of cars that were observed going through intersections in the
Washington, D. C. area without stopping at the stop signs. According to the
story, Volvos were heavily overrepresented: The fraction of cars running stop
signs that were Volvos was much greater than the fraction of Volvos in the
total population of cars in the D.C. area. This is initially surprising because
Volvo has built a reputation as an especially safe car that appeals to sensible,
safety-conscious drivers. Volvos are largely bought by middle-class couples
with children. How then is this observation explained?
One possibility is that people driving Volvos feel particularly safe in
this sturdy, heavily built, crash-tested car. Thus they are wil ling to take risks
that they would not take in another, less safe car. Driving a Volvo leads to
a propensity to run stop signs. This is essentially a moral hazard explanation:
The car is a form of insurance, and having the insu rance alters behavior in
ways that are privately rational but socially undesirable.
A second possibility is that the people who buy Volvos know that they
are bad drivers who are apt, for example, to be paying more attention to their
children in the back seat than to stop signs. The safety that a Volvo promises
is especially attractive to people who have this private information about their
driving, and so they disproportionately buy this safe car. A propensity for
running stop signs leads to driving a Volvo . This is, of course, essentially a
self-selection story: the Volvo buyers are privately informed about their driving
habits and abilities. Unless this selection imposes costs on Volvo, however,
it is not adverse selection
Both stories are at least plausible. How would you go about testing
which, if either, is correct? What other explanations seem plausible?

suits, practice "conservative medicine, " ordering tests and procedures that may not be
in the patient's best interests and, in any case, are surely not worth the costs (which
are borne by the patient or the insurer-not by the doctor making the decision). Some
firms may find it most profitable to make shoddy or unsafe products when qual ity is
not easily observed. Security brokers may "churn" their clients' portfolios, encouraging
them to trade more frequently than they really should because each additional trade
generates commissions for the brokers . Automobile dealers may fail to mention the
poor resale values or the higher than average repair costs of the cars they sell. Rented
apartments may be less well maintained than owner-occupied ones because the renters
do not get the full benefits of their efforts at maintenance. All of these examples are
drawn from ordinary market experience.
Within organizations, an office employee may spend time during the day
studying for an accounting exam , thinking about a new business idea that he or she
hopes to pursue, or chatting on the telephone with friends when there is work waiting
to be done. Factory workers may call in sick during hunting or fishing season , and
when on the job they may exert the least care and effort they can get away with .
Managerial employees may exaggerate the difficulty of their assignments in order to
 1 70
Motivation: make their performance appear more impressive, or they may denigrate others'
Contracts, performance in order to improve their own chances of getting a particularly desirable
Information, and assignment or promotion. Division executives may adopt policies that lead to high
Incentives current performance that wi ll be rewarded by bonuses and promotions, even though
these policies will ultimately destroy the long-term profitability of the divisions they
will have left behind. Seni or executives may pursue their own goals of status, high
salaries, expensive "perks, " and job security rather than the stockholders' interests, and
so they may push sales growth over profits, treat themselves to huge staffs and corporate
jets, and oppose takeovers that would lead to their dismissal but would increase the
value of the firm.
These examples do not involve insurance explicitly, but they have the crucial
feature of insurance: The decision makers do not bear the full impact of their decisions.
The doctor who orders extra tests benefits in terms of the reduced likelihood of
successful malpractice suits but does not pay for the costs of the tests or suffer the
discomfort they cause. The employees get paid whether they work hard or not, or at
least they do not suffer a decrease in pay equal to the full lost value of what they
could have produced. The difficulties of monitoring are what prevents them from
bearing the full costs and benefits: The stockbroker's client does not have the expertise
to tell if a trade is good for him or her or just for the broker; the employee's supervisor
cannot freely determine whether he or she is thinking about company business or
personal matters; and the stockholders cannot easily evaluate whether a particular
executive action is in their interests or not.
THE PRINCIPAL-AGENT RELATIONSHIP Each of these examples can be cast in terms
of an agency relationship. This term has come to be used in economics to refer to
situations in which one individual (the agent) acts on behalf of another (the principal)
and is supposed to advance the principal's goals. The moral hazard problem arises
when agent and principal have differing individual objectives and the principal cannot
easily determine whether the agent's reports and actions are being taken in pursuit of
the principal's goals or are self-interested mi sbehavior. Agency relations in this sense
are pervasive: The doctor is the agent of the patient, the worker is the agent of the
firm, the CEO is the agent of the owners, and so on. As we will see later, however,
moral hazard problems also arise in relationships where neither party can be considered
the agent of the other but rather each is on an equal footing (as in a partnership).

CASE STUDY: THE U.S. SAVINGS AND LOAN CRISIS

The key factors giving rise to moral hazard problems--divergent interests, decision
makers being insured against some of the consequences of their actions, and monitoring
and enforcement being imperfect-all feature centrally in one of the most spectacular
moral hazard problems of recent times, the United States "savings and loan crisis. "

The Savings and Loan Indust ry

Savings and loan associations (S&Ls) are for-profit financial institutions that borrow
money from the public in the form of deposits and then invest it by lending it out
again, much like banks. The deposits of individual depositors in an S&L are insured
by a U . S. federal government agency-until 1 990, the Federal Savings and Loan
Insurance Corporation (FSLIC). If, for some reason, the S&L could not repay the
deposits, the FS LIC would. Government-provided insurance for bank deposits was
instituted in the United States in the 1 930s to protect depositors agai nst bank failures.
This insurance was also intended to reduce the likelihood of bank failures by
eliminating "bank runs, " which arise when depositors become fearful that their deposits
may not be repaid, rush to withdraw their funds, and thereby bring on the fai lure
 171
they feared. The S&Ls, as well as the not-for-profi t credit union s, were also pro vided Moral Hazard and
with deposit in suran ce. The fu nds to pay for claims again st the FSLIC came from Performance
charges levied on the in sured S&Ls. The size of these prcmia were not lin ked to the I ncentives
riskiness of the S&Ls' portfolio of loans and other investments.
THE CRISIS IN Tl fE 1 980s Traditi onally, the S&Ls were strictly limited in how they
could invest their fu nds, with their primary investments being residential mortgage
loans to local individuals secured by the homes they owned. During the 1980s many
S&Ls turned to riskier investments, including loans on commercial real estate an d
high-yielding but very risky corporate borrowing called "j unk bonds." As the commercial
real estate market collapsed in several parts of the country, borrowers ceased payments
on many of their loans, and the S&Ls were left holding property they could n ot rent
or sell. Later, defaults· by some corporations on their j unk bonds undercut the val ue
of all' high-risk debt, further reducing the S &Ls' assets. As well, a plague of fraud
spread through the industry. Thi s proved a devastating combination: Over 500 savings
and loans slipped into bankruptcy. The FSLIC' s reserves were inadequate to cover its
promises to protect depositors, and U.S. taxpayers are now having to foot the bill,
which is measured in the hundreds of billions of dollars! What led the S&Ls to make
such risky investments? What led to the increase in fraud? Could it all have been
prevented? Who is to blame?
THE CAUSE: MORAL HAZARD The very design of the deposit insurance program,
together with lax regulation, led to a costly problem of moral hazard in the management
of the savings and loans. In brief, deposit insurance and low capital requirements (the
amount of the S&L owners' own money at risk) encouraged excessive risk taking by
rel ieving the S&Ls of the responsibility for poorly performing investments while
allowing them to gain when the investments prospered. The insurance also relieved
the depositors of the usual responsibility of investors to keep tabs on those who hold
their money. This encouraged both risk taking and fraud. Insurance could be
economically provided only so long as other regulatory policies were able to prevent
the S&L managers from making reckless investments and engaging in self-dealing. In
the early 1980s, however, the regulations controlling the sort of investments the S&Ls
coul d make were relaxed. At the same time, the amount of insurance afforded to each
depositor was increased and the resources devoted to enforcing the relaxed regulations
were reduced. The whole system inevitably broke down.

Deposit I nsurance and Risk Taking : An Exam p le

The S&Ls made risky investments in part because the government insurance scheme
made those investments profitable for the owners of the S&Ls. To see how insurance
creates these incentives, l et us study an example. To make the calculations simple,
we set the interest rate pai d to depositors at zero. The principles we deduce from this
example apply to any other interest rate as wel l.
Suppose the owner of the S&L has full authority over how to invest the deposits.
The owner can choose between two possible investments, labeled "safe" and "risky. "
Actually, both investments in our example have some uncertainty about their returns
but, as Table 6.1 shows, the safe investment has less variation in its returns and can
never actually lose money.
The safe investment requires an initial outlay of $100 and return s either $100
or $110, each with a 50 percent probabil ity. We call it "safe" because it always returns
at least the i nitial outlay of $ 1 00. O n average, it does even better, returning the initial
outlay plus $5 more.
The risky investment is a lemon. There is a 5 0 percent chance that it will return
only $65, far less than the initial outlay of $100. O n average, it loses $5. This is a
 172
Motivation: Table 6. 1 Description of Investment
Contracts, Opportunities
Information, and
Incentives
Safe Risky

Initial outlay 100 100

High return 1 10 125
. . . probability . 50 . 50
Low return 100 65
. . . probability . 50 . 50
Expected return (gross) 105 95
Expected return (net) 5 -5

bad investment that, in a well-functioning system, would not be undertaken because

it wastes social resources. As we shall see, however, the way the S&L is financed and
insured can create differences of interests between the various parties whose resources
are at stake, and lead the managers to undertake the risky investment.
The funds an S&L has for lending and investing come from two main sources:
deposits and the capital supplied by owners. United States federal regulations in the
1 980s required the owners of an S&L to provide capital equal to about 3 percent of
the total value it invests. For our example, we suppose that of the initial outlay of
$ 1 00 required for the investment, $97 comes from the depositors (insured by the
FSLIC) and $ 3 comes from the owners of the S&L. The depositors have first priority
on any proceeds from the investment. This means that if the proceeds from the
investment are more than $97, the depositors must be paid in full. If they are less
than $97, the depositors get whatever money is available, the owners get nothing, and
the FSLIC pays the difference between the available funds and depositors' $97 claim.
SHARING THE RISKS Now let us examine the distribution of the costs and benefits if
the safe investment is made. The term gross return refers to all the income received
from the investment, without regard to the initial outlay. In Table 6. 2, the gross
return earned by the owners is $ 1 3 if the investment works out well and $3 if it works
out badly. Net return refers to the gross return minus the initial outlay. Because the
owners have an initial outlay of $3, their net returns are either $ 1 0 or $0. In
expectation, they get $ 5. In any event, the depositors just get their money back,
netting zero (the interest they were promised), and the FSLIC neither receives nor
disburses any funds.
When the safe investment is made, the FSLIC is never needed to "bail out" the
S&L, that is, to help it meet its obligations. The situation is quite different when the
risky investment is made, however. Table 6. 3 shows how the returns on the risky
investment might appear to the various parties.

Table 6. 2 Analysis of the Safe Investment

Depositors Owners FSLIC Total

Initial outlay 97 3 0 100

High return (gross) 97 13 0 1 10
Low return (gross) 97 3 0 100
Expected return (gross) 97 8 0 105
Expected return (net) 0 5 0 5
 1 73
Table 6. 3 Analysis of the Risky Investment Moral Hazard and
Performance
Incentives
Depositors Owners FSLIC Total

Initial outlay 97 3 0 1 00
H igh return (gross) 97 28 0 125
Low return (gross) 97 0 - 32 65
Expected return (gross) 97 14 - 16 95
Expected return (net) 0 11 - 16 -5

The final column of Table 6. 3 repeats the description of the risky investment
already given in Table 6. 1 . The first three columns show how the initial outlay and
returns are divided among the three parties. Notice that when returns are high, the
owners of the S&L enjoy exceptionally high profits. In contrast, when returns are low
and there is not enough money to pay the depositors' claims, the FSLIC bears
exceptional costs to pay off the depositors. All the owners of the S&L lose is their $3,
with the FSLIC absorbing the rest of the loss. The S&L owners benefit from risk
taking, whereas the FSLIC suffers from it.
THE WINNERS AND LOSERS The bottom line of Table 6. 3 further clarifies the matter.
Although this investment is a lemon overall, with an expected net return of - $ 5 , it
generates an expected net return of $ 1 1 to the owners. This is far more than the $ 5
the owners could expect to get from the socially preferable, safe investment. The
FSLIC, which bears the losses when the investment returns are too low to cover
deposits, now suffers an expected loss of $ 1 6, the difference between the owners'
expected returns and the total returns on the investment. The depositors, who are
insured, always get all of their money back.
According to the bottom line of Table 6. 3 , the expected net returns of all the
parties add up to the total expected net returns of the investment. Because the
depositors always get zero, each dollar of expected loss imposed on the FSLIC shows
up as another dollar of expected profit for the owners! In choosing among investments
with equal expected returns, the owners will prefer the riskier ones because they expect
to profit most when the FSLIC's expected losses are largest.

I ncentives for Risk Takin g with Borrowed Funds

The financial motivation for the S&L owners to make risky investments is now clear:
The riskier the investment, the higher the expected losses for the FSLIC and the
greater the expected profits for the S&L. Still, our story of how the government
botched its regulation of the S&Ls is incomplete. The problem is that the owners
would have the same motivation to make risky investments even if the FSLIC were
taken out of the picture! With the FSLIC gone, the losses would fall on the depositors
instead of the government agency. Still, the owners would be the ones to benefit from
the risky investment if things go well, and someone else would be left to bear the
losses. Thus it might appear that if the government eliminated deposit insurance, the
only thing to change would be that risks are shifted from the government agency to
the depositors. So long as investments are financed by borrowing, the borrowers will
always have an incentive to undertake riskier investments than the lenders would want.
Of course, the whole point of a savings and loan institution is that depositors leave
their money there for the S&L to invest; an S&L always is a borrower from its
depositors, so the problem always exists. It may seem, therefore, that we have unfairly
identified the FSLIC as being the root of the problem.
 1 74
Ivlotivation:
Contracts,
Information, and
Incentives
The Case of Seapointe Savings and Loan
S eapointe S avings and Loan was founded in 1985 in Carlsbad, California, a
suburb of S an Diego. Like other S&Ls, Seapointe's primary business was to
make home mortgage loans. According to its business plan: "At no time will
management presume to outguess the marketplace nor risk the net worth of
the institution in an attempt to 'make a killing' for the sake of short-term
earnings. "
When Seapointe failed in 1986, however, it h ad never even hired a
loan officer. Instead, it had sold " naked call options. " That is, it had sold a
promise to deliver $10 billion of bonds that it did not own, at the buyer's
option, at a given future date. Selling call options is a common practice, but
selling them without actually owning the bonds makes the options "naked";
it exposed Seapointe to the risk of an actual cash l oss if bond prices were to
rise. And rise they did, so that S eapointe was forced to pay the difference
between what the bonds cost at the delivery date and the price it had promised.
S eapointe lost $24 million on this one transaction-75 percent of its assets­
leaving the FSLIC responsible to repay the institution' s depositors. If S eapointe
had won its bet and bond prices had fall en, the buyer would not have
exercised its option to demand delivery of the bonds at the pr omised price,
and Seapointe would have kept the amount it was paid for its promise.
Seapointe would have 'made a killing' for its owners.

Source: Charlotte-Ann Lucas, "How an S&L Gambled Off Its Deposits Within a Year , "
San Francisco Examiner (December 2 , 1 990), A- 1 .

.MONITORING BoRRO\' r ERS To gain a deeper understanding of the issues, we must

look beyond the S&L industry to find related business pr oblems. The problem of
depositors who are, in effect, lenders to the S&L is by no means unique in business.
Lenders exist thr oughout the business world. H owever, they take precautions to ensure
that their money is not squandered, stolen, or put at unnecessary risk by those who
have borrowed it. A bank that lends you money will ask you about your financial
condition and about what you intend to do with the loan proceeds. It will run a credit
check, demand collateral , and often require regular payments of the interest and part
of the principal. If yours is a home loan, it will demand a legal interest in the property,
and if it is a car loan, the bank will keep title to the car until the loan is repaid. Those
who lend to firms frequently impose similar conditions. They examine the firm's
financial condition and credit history, put restrictions on how their funds may be
used, and often require business plans, collateral, and periodic financial statements.
Correspondingly, careful scrutiny by depositors is the mechanism by which an
unregulated and uninsured S&L might be kept from making irresponsi ble investments
or defrauding its investors.
Why, then, did S&L depositors not take the same precautions as other lenders?
Because doing so was costly and, anyway, the deposits were insured! The insurance
itse lf m ade the de positors willing to supply huge sums to S&Ls with out the usual
checking of creditworthiness or m onitoring of performance that accompanies oth er
large loans. For the FSLIC to pr otect itself against huge losses, it thus needed to
 1 75
regulate the S&Ls, m onitoring their activities, restrictin g their in vestme nts, and Moral H azard and
ensuring that they maintained adeq uate capital both to guard against unlucky Performance
investment outcomes and to ensure that the owners would suffer a significan t loss if Incentives
the organization should fail. The government failed to do this through the 1980s an d
suffered from the resulting moral hazard problem.

The Perverse Effect of Com p etition

The moral hazard problem in the S&L industry was actually i ntensified by the effects
of competition. N ormally we think of competition, which tends to drive out those
executives who are unwilling to take the profit-maximizing actions, as prom oting
effi ciency. In the context of the S&L industry in the 1980s, however, competition
had a. perverse effect. M any conservative S&L executives had no choice but to gamble
on risky investments if they were to survive in the circumstances we have described.
Think · about how the system works. The problems may have all begun with a
few S&Ls that directly saw the chance to exploit the deposit insurance system by
moving into more risky investments. To do so, they needed to expand their deposit
base. The only quick and sure way to attract substantial new deposits is to offer higher
interest rates to depositors. Thus, S&Ls seeking an influx of money to expand
investments offered higher interest rates than did their competitors. The government's
increasing the amount that was insured (from $40, 000 per account to $100, 000) made
these higher rates very effective, as large investors-including many firms-deposited
their money in the aggressive S&Ls.
N ow, the other S &Ls began to feel the heat. Deposits were being drawn out of
their doors and given to their competitors. To stay in business, some of these others
also raised the rates paid on their deposits. For some, given their operating costs, these
rates were higher than they were able to pay using just their normal, relatively safe
investments in residential real estate. Therefore, they too were driven to riskier
investments which, if they worked out well, would enable the company to pay the
promised interest rates and still make a profit.
Despite the spiraling competitive pressure, some S &Ls may have held out,
making only safe investments. They either offered lower interest rates to depositors
and so faced a crisis of falling deposits, or they matched the competitive, higher
interest rates and suffered losses as the income from their loans fell short of what was
needed to pay their costs and other obligations. This became especially significant in
1979 and 1980 when a change in monetary policy by the Federal Reserve (the U . S.
central bank) caused interest rates to shoot up throughout the economy. M any S &Ls
had 111uch of their money tied up in long-term, fixed-rate mortgage loans, the rates
on which suddenly were less than what they were having to pay for their deposits. As
well, the collapse of the real estate market in Texas when oil prices fell in the mid­
l 980s had a special adverse effect on the S&Ls in that state.
M any of the endangered S&Ls became prey for aggressive entrepreneurs, who
bought the failing companies for low prices and tried to make them profitable by
radical means-offering very high rates on $100, 000 "j umbo" deposits and investing
in risky commercial real estate, "j unk bonds," and other similar ventures. Investors
continued to place their deposits with these financially troubled companies because
the deposits were federally insured.
It is easy to see how competitive pressure forced ont many of the more
conservative S &L executives throughout the industry. Those who were unwilling to
make the risky investments were often driven out of business. The big loser was the
FSLIC-and the taxpayer.
 1 76
Motivation: Fraud in the S&Ls
Contracts, The story of the savings and loan industry as told here leaves out many important
Information , and details. Risky investments that went bad did not alone deplete the capital of so many
Incentives
S&Ls. Outright fraud was also responsible. News accounts report that the top officers
at savings and loan institutions made loans to themselves, their other companies, their
friends, and their family members at reduced interest rates and without adequate
collateral. They paid large dividends to investors and generous salaries to themselves
and their relatives, even as their firms were sliding into bankruptcy. They concealed
bad loans by lending more money to the borrowers so that they could afford to make
payments on older loans. These activities are tantamount to stealing funds from the
S&L, or, because of the insurance, from the taxpayers. Government investigators
have found that there was fraud in at least 2 5 percent of the cases of S&L bankruptcy.
An analysis of the problem of fraud in the savings and loan industry would be
quite similar to our analysis of the adoption of risky investments. Throughout the
economy, people entrust their funds to the management of others. They protect
against fraud and against excessive risk taking in the same sorts of ways: by monitoring
performance, hiring auditors, writing restrictive rules into the organization's charter
about what activities are allowed, and so on. They retain enough control of the
management's activities to dismiss errant executives. The depositors at a federally
insured savings and loan did not engage in these costly activities because the deposits
were backed by an agency of the U. S. government. No private investor has an
incentive to protect the federal insurance agency against fraud; the government
regulators have to do that for themselves.

Who's To Blame?
This example of the S&L crisis is remarkably rich, for it involves moral hazard on
the part of three distinct groups. First, and most obvious, are the S&L owners who
took excessively risky investments or committed fraud. Second are the depositors who
failed to monitor the S&Ls because their deposits were insured. The third group
consists of the politicians-in both the legislative and the executive branches-who
favored the industry at the expense of the general taxpayer. These politicians raised
the amount of insurance provided by the FSLIC, thereby making it easier to attract
large deposits. They relaxed the regulations on the S&Ls and did not provide for an
offsetting increase in monitoring. Furthermore, when the S&Ls were fi rst headed for
financial trouble, politicians blocked the regulators from intervening to protect the
FSLIC and the taxpayers. Possibly some of these politicians were motivated by a
genuine belief that the actions being taken were truly in the general interest. However,
the huge campaign donations made by some S&Ls to various of these politicians
certainly raise the possibility that the politicians were pursuing their own interests and
expected to get away with it because the public's monitoring of them is so imperfect.

PUBLIC VERSUS PRIVATE INSURANCE

The savings and loan crisis is fundamentally the result of moral hazard that arises
from the existence of the deposit insurance provided by the FSLIC. Yet this insurance
provides valuable social benefits as well. Small savers need not lose sleep worrying
about the safety of their deposits, nor do they have to incur the very real costs of
monitoring the institutions to which they have entrusted their life savings. The
desirability of deposit insurance, combined with the obvious problems that the
government-provided program experienced, has led some commentators to suggest
that such insurance should be privately provided. There are clear difficulties in
designing and implementing such a program, and the failure of a number of small
 1 77
private deposit in surance fu nds in the state of Rhode Island in 1 991 may have removed Moral Hazard and
some of the allure of this idea. S till, it is clear that government insurance programs Performance
do seem to have inordinate difficulties with moral hazard. Incentives

Other U.S. Government Insurance and Guarantee Programs

The crisis in the savings and loan industry is symptomatic of problems with man y
other government insurance and guarantee programs i n the U nited S tates. Together
these programs are estimated to involve insurance and loan-guarantee commitments
of more than $5 trillion, which is almost twice the U . S. national debt and five times
the level of yearly federal government spending. 3 H ere are just a few examples.
THE PENSION BENEFIT GUARANTY CORPORATION The Pension Benefit Guaranty Corpo­
ration (PBGC) was established by the Employee Retirement Income S ecurity Act of
1974 (ERISA) wi th the intent, according to its advocates, of ensuring that promises
of retirement benefits made to working people by their employers would be honored.
The PBGC is obliged to take over plans that are terminated without sufficient funds
to pay the promised benefi ts. It collects what it can from the company that terminated
the plan and uses the proceeds to pay the pensioners. To finance the remainder, it
collects a tax called an insurance premium that is imposed on other pension plans.
To avoid transferring huge pension liabilities to the new government agency,
ERISA provided certain minimum funding standards for pensions and held employers
liable in part for their pension promises. The act allowed dramatic underfunding of
certain kinds of plans, however, particularly multiemployer plans run by trade unions
for their own members, but also some corporate pension plans. Predictably, many of
these plans aggressively expanded benefi ts beyond what the limited funds available
could possibly justify. Later, some of these plans shut down, saddling the PBGC with
the liability to pay the promised benefits.
THE FEDERAL CROP INSURANCE CORPORATION The Federal Crop Insurance Corpora­
tion (FCIC) was established in 1939 to protect farmers against crop losses caused by
the vagaries of weather. Yet the FCIC has relatively few inspectors and has been raked
by fraudulent claims. S ome farmers have defrauded the FCIC by claiming crop losses
in one name and selling the harvest in another. S ome crop insurance policies have
been sold after the claimed loss occurred. In one case a claim was filed for crops that
were planted 30 days after the freeze that had triggered the insurance payment. 4
Even when fraud is not a factor, moral hazard arises. Insured farmers are
tempted to take greater risks by planting less hardy or more water-hungry crops than
would otherwise be prudent. If the weather turns out to be warm or rainfall turns out
to be plentiful, the farmers profit; if not, the government insurance program pays.
THE GOVERNMENT NATIONAL MORTGAGE ASSOCIATION The Government N ational
Mortgage Association (GNMA) exists to make it easier for homeowners to obtain
mortgage loans. It does this in several ways, most prominently by providing insurance
to lenders against defaults on the mortgages they write. Along with the development
of mortgage-backed securities (see Chapter 1) came mortgage brokers who received
commissions of $30 to $45 per $1,000 of the loans they wrote, while maintaining
capital of as little as $0. 30 per $1, 000 of outstanding mortgages.
It is profitable for these brokers to write very risky mortgages because they receive
a large commission but have little capital to lose in the event of a default; GNMA

3 "Government Waste: Where's Nanny?" The Economist (January 6, 1 990), 3 l .

4 Bruce Ingersoll, "Crop-Insurance Fraud and Bungling Cost U. S . Taxpayers Billions," The Wall
Street fournal (May I 5, 1 989), A- 1 .
 1 78
Motivation: picks up the tab. To protect itself, CNMA specifies limits on the ratio of the loan
Contracts, amount to the appraised val ue of the property, so that it will have adequate collateral.
Information, and Appraisals are subjective, however, and unscrupulous brokers have on occasion vastly
Incentives exaggerated the value of properties in order to j ustify large l oans, leading to default
and large l osses paid for by the taxpayer while the borrower and broker profit.
STUDE!'ff LOA.NS Student loans provide another example of how guarantees affect
costs. Federal government guarantees enable many students to obtain loans on more
favorable terms than would otherwise be available. Normally, when banks make loans,
they take measures to ensure that the loans are coll ectible and they are aggressive in
collecting from debtors. The incentives to ensure collectibility are bl unted by
government guarantees. U nsurprisingly, a huge proportion of guaranteed student loans
are never repaid. Just tracking who the debtors are often exceeds the government' s
limited capacity to administer these loans.

Private or Public I nsurance ?

Al though moral hazard problems are present in both government and private sector
insurance programs, the problems do seem to be less severe in the private sector. In
part, this is because private corporations cannot sustain such huge losses for long
without going bankrupt, and they often cannot rely on the taxpayers to pay for their
financial ineptitude. B ut the more l imited difficulties with moral hazard in private
insurance is not due entirel y to any special merit of private insurance programs; it is
partly a resul t of the private sector's unwillingness to undertake socially desirable
insurance programs in which the costs associated with moral hazard are high.
The money-making obj ective of private companies is quite different from the
obj ectives of a government agency. N o bureaucrat is well positioned to eliminate an
unprofitable deposit insurance program that protects the life savings of small depositors.
Even a legislature would have difficulty making such a decision. Nor is it clear that
elimination of such a program is social ly desirable. In contrast, most private firms
would have little trouble deciding to terminate the program if the losses being suffered
were large.
O n average, private-sector insurance programs do seem to be better managed
than are their government counterparts. S ome of the losses suffered by the various
government programs could have been avoided by having more inspectors or tighter
regulations. The " output" of inspectors, however, is not easily measured and so,
especially during periods of large budget deficits, there is an attractive short-term
economy to be achieved by reducing the number of inspectors on the government
payroll. Short-sightedness and poor management of this sort certainly seem to have
been a factor in the savings and loan crisis.
Yet moving insurance to the private sector is no cure. Moral hazard can manifest
itself there in some remarkable ways as well.

�loral Hazard in Private Life I nsurance

I n a life insurance contract, the insured's designated beneficiary is paid an agreed­
upon sum of money when the insured person dies. All life insurance contracts issued
in the U nited S tates have certain standard provisions, one of which deals with death
by suicide. Essentially, life insurance contracts provide that the insurance company
will pay off on the policy to the beneficiary after the insured person dies by suicide
only if the su icide occurs after a certain period of time has elapsed from the time the
policy was issued. I n the U nited S tates, this exclusion period is always either 1 2 or
24 m on ths. Life insurance statistics show that the suicide rate is lowest in the twelfth
and twen ty-fourth mon ths after a policy has been issued and highest in the thirteenth
 1 79
and twen ty- fifth. Th e inference seems un mistakable: People postpone their suicides Moral Hazard and
to allow their beneficiaries to collect the life insurance proceeds. Performance
Incentives
MORAL HAZA HD I N ORGANIZATIONS
Moral hazard was first identified in the insurance context, and some of its most
spectacular manifestations are still found there. For understanding organizations,
however, it is important to recognize the point made earlier-that moral h azard is a
very common phenomenon that affects a wide array of transactions and that attemp ts
to deal with moral hazard account for many of the particular institutional arrangements
we see, both in markets and within organizations. Indeed, the very boundary between
these two forms of organization is often a resp onse to moral hazard concerns.

Moral Hazard and Emplo yee Shirking

An important instance of moral hazard arises in employment relationships, where
employees may shirk their responsibilities. Frederick Taylor, the "father of scientific
management, " once wrote: "Hardly a competent worker can be found who does not
devote a considerable amount of time to studying j ust how slowly he can work and
still convince his employer that he is going at a good pace. " 5
Evidence of the importance of moral hazard in the employment relationship is
the frequency wi th which firms give employees incentive or performance contracts.
These arrangements tie the employee's compensation to vari ous measures of perfor­
mance, and are meant to motivate effort, creativity, care, diligence, loyalty, and so
on. Examples include pay tied to output, such as piece rates for manufacturing workers
or bonus clauses that reward unusually large numbers of touchdown passes caught by
football players; pay linked to sales, such as salespeople's commissions; pay linked to
productivity improvements, as under "Scanlon Plans"; and various ways of linking pay
and pr ofits, including employee stock ownership plans, the Japanese practice of paying
workers an annual bonus tied to firm profitability, and many executive compensation
schemes. When well designed and well administered, these sorts of arrangements can
be effective in promoting the desired behavior. Although clear communication to
employees of what it is that the employer values is partly resp onsible for this effect,
direct financial incentives are the key.
To see that these arrangements are evidence of moral hazard, note that the firm
is not paying directly for what the employees are supplying but instead uses a proxy
for it. What is actually being supplied are such things as the employees' intellectual
and physical effort. What is paid for are the results of these inputs-sales and
touchdown passes, for example. The amount and quality of the employees' efforts are
difficult to monitor directly, whereas the results of their efforts may be more easily
observed. Thus, rather than trying to pay for unobservable effort directly, the firm
attempts to motivate employees to choose to work harder or better by rewarding
outcomes that are more likely when they behave in the desired way.
AIR TRAFFIC CONTROLLERS: AN EXAMPLE 6 Air traffic controllers are charged with
maintaining air safety by keeping airplanes in flight at specified, safe distances fr om
one another. They use radar to track flights and radio to direct the pilots.
Federal government employees in the U nited S tates in the 1970s, including air
traffi c controllers, were covered by a disability program, the Federal Employees'

5 Frederick Taylor, The Principles of Scientific Management (New York and London: Harper, 1929).
6 This section is based on Michael E. Staten and John Umbeck, "Information Costs and Incentives
to Shirk: Disability Compensation of Air Traffic Controllers, " American Economic Review , 72 (December
1982), 1023-37
 180
Motivation: Compensation Act. If they were unable to work because of a disability that was the
Contracts, result of their j obs, they were entitled to a receive a fixed percentage of their pay for
Information , and the duration of their disability. This tax-free payment could be as high as 7 5 percent
Incentives of the base salary. Given the tax rates of the period, a disabled worker might actually
receive a higher take-home income under the disability program than when working.
I n order to collect on a disability claim , the injury had to be shown to be both
disabling and work related. The injury did not need not to be physical; stress-related
disorders that prevented the employee from working would qualify. To control
unj ustified claims, the inj ury report had to be supported by a statement from a
physician certifying the disability, another from the employee's supervisor describing
the events leading up to the injury, prior symptoms, and the work environment, and
first-hand reports from coworkers.
Air traffic controllers, whose j obs were viewed as unusually responsible and
stressful, would have had a relatively easy time making a claim for disability based on
nervous or emotional disability. Moreover, certain changes in 1 972 and 1 974 in the
rules governing disability claims made claiming disability even more attractive for
controllers. The 1 972 changes provided for retraining for second careers for those air
traffic controllers who were found to be disabled, even if the disability were not job
related. The 1 974 rule changes made monitoring false claims generally more difficult
and made catching a fake stress-related claim especially difficult. If moral hazard were
a problem among controllers, then these changes should have led to increased
incidence of fake disabilities ("punching out") and an increase in claims, especially
in the number of psychologically based claims.
In fact, the number of disability claims did rise with the initiation of each
program, more than doubling in 1 974 and continued to rise. The largest percentage
increase following the 1972 change was in psychological and psychiatric illness, such
as stress-related disabilities; it was largest by a factor of three. (Unfortunately, the data
did not permit identification of the mix of claims following the 1 974 change. )
More striking, however, was the apparent impact of the 1 974 change on job
performance. A controller who wanted to fake a claim for stress-related disability
needed to show the disability was job-related to collect, and the examiners were
directed to look for specific events on the job that either could have contributed to
the stress or that were sym ptoms of the disability. This created an incentive to
manufacture on-the-job incidents that could have caused the stress and that might
also indicate that the employee was no longer capable of doing the job. The natural
candidate here was a "separation violation, " in which planes for which the controller
was responsible came too close to one another.
The Federal Aviation Authority keeps track of two sorts of separation violations:
System Errors and Near Mid-Air Collisions. The former represent any violations of
the standard separation requirements; the latter are much more serious and directly
life threatening. Because either sort of violation would do equally well for the purposes
of filing a claim, a controller who did not want to cause unnecessary danger would
be much more willing to generate a minor violation than a near collision. In fact,
the number of Systems Errors j umped significantly after the 1 974 change, but there
was only a small, statistically insignificant change in the number of Near Mid-Air
Collisions. Furthermore, the increase in Systems Errors tended to occur not when
traffic was particularly heavy, as you might otherwise expect, but when it was relatively
light and the controller could cause the "needed" violation at minimal risk.
Finally, a controller considering punching out had to decide when in his or her
career to do so. Various factors, especially eligibility for retraining and an effective
dependence of the amount of disability pay on years of service, made it much more
 181
attractive to punch out after five years of service. Before the 1974 changes, controllers Moral Hazard and
with less than five years' experience (who presumably were relatively inexperie nced Performance
and more prone to mistakes) and, to a lesser extent, those with more than ten years' I ncentives
experience (who were more likely to suffer from actual "burn- out") were responsible
for most of the System Errors made. By 1 976 personnel in the five-to-te n-year range
were committing over 50 percent of the errors, although this group accounted for less
than 30 percent of the total numbe r of controllers. All this, the n, is striking evide nce
of moral hazard.

Managerial M isbehavior
The senior executives of corporations are charged with ad vancing the interests of the
stockholders, who are the owners of the company. They are supposed to be overseen
in thi s duty by the board of directors, who are elected by the stockholders and who
are empowered to represent them in voting on maj or corporate decisions and setting
the executives' compensation. Thus, both the executives and the board members are
considered the age nts of the stockholde rs.
Over 50 years ago, Ad ophe Berle and Gard ner Means argued that the dispersed
holdings of stock across a multitude of small investors had created an effective
separati on of ownership and control, with no individual stockholder having any real
ince ntive to monitor managers and ensure that the officers and board were running
the firm in the owners' interests. 7 Although this claim long remained highly
controversial, evidence that has accumulated indicates that managers ofte n do fail to
promote the interests of the stockholders effectively.
The problem typically is not that the e xecutives are lazy and do not work hard
enough. Corporate executives put in remarkably l ong hours of very inte nse effort.
Rather, the complaint is that they pursue goals other than maximizing the long-run
value of the firm. Critics claim that executives invest firms' earnings in low-value
projects to expand their empires when the funds would be better distributed to the
shareholders to invest for themselves. The y are alleged to hang on to badly performing
operations whe n other teams of managers could run them more profitably or eve n
when the operati ons are irredeemable losers. With the connivance of their hand­
picked board s, they pay themselves e xorbitantly and lavish expensive perquisites upon
themselves. They resist attempts to force more profitable operations, especially by
resisting take overs that threaten their j obs. All these alleged misdeeds serve the interests
of the managers themselves (and perhaps the interests of other concerned constituencie s,
such as employees), but not the interests of the firms' owners.
HOSTILE TAKEOVERS AND MANAGERIAL MISBEHA \'IOR During the 1 980s a wave of
hostile take overs occurred in the U nited S tates and to a lesser exte nt in the U nited
Kingd om and Canada. A hostile takeover is the acquisition of enough of the shares
in a company to give a controlling ownership interest in the firm, where the offer to
acquire the firm is opposed by the target company's executive s and directors. S uccessful
hostile takeover attempts generally resulted in re placement of the target firms' se nior
management and the naming of new board s of directors. The buyers in these
transactions were called "corporate raiders" (as well as many less complime ntary things)
by the managers of the target firms who fought to maintain the companie s'
independence. In this context, "independence" means the firms' continuing under
their curre nt managers and board s with the existing ownership. This ownership was

7Adophe Berle and Gardner Means, The Modern Corpora tion and Private Property (New York:
MacMillan, 1 932).
 182
Motivation: typical ly quite diffuse-i nd ividual smal l shareholders, plus pension plans, insurance
Contracts, compani es, and mutual funds.
Information, and Many observers have interpreted the hostile takeovers as a correcti ve response
Incentives to managerial moral hazard: The takeovers, i t is claimed, were intended to di splace
entrenched managers who were pursuing their own interests at the expense of the
stockholders. Whether thi s is the case or not, the huge profi ts that were generated in
these transactions raise questions about how effectively incumbent managers were
maxi mizi ng the values of the companies they ran.
The prices paid for the stock of firms in hostile takeovers in this peri od on
average represented a 50 percent premium over the target' s original market value. For
example, j ust before M obi l Oil launched i ts bid for Marathon Oil i n 1981, Marathon' s
stock was trading at $63. 75 a share. M obil offered $85, and eventually raised its offer
to $126 before Marathon was acquired by U.S. Steel (now called U SX). Similar
premia were paid in hundreds of cases. I n aggregate from 1977 thr ough 1986,
shareholders selling their stocks in hostile takeovers realized an estimated gain of $346
billion (in 1 986 dollars). 8 The takeover premia appear to be evidence of managerial
incompetence or moral hazard to the extent that this original market value represented
the discounted profi t stream that savvy investors expected the firm to generate under
its original management, whereas the takeover price refl ected the firm's value under
the new ownership.
The takeover premia are not conclusive evide nce that the managers were poor
steward s, however. Perhaps there was systematic overbidd ing by raiders who suffered
delusi ons about their managerial abi liti es or were using other peoples' money to
expand thei r own empires. Perhaps the new owners expected to reap gains at the
expense of other stakeholders (especially current managers and workers), where these
pri vate gains are not increases i n efficiency but the retur ns to violating explicit and
implicit promises. Or perhaps the stock market was systematically underestimating the
target firms' prospects before the takeover attempts were launched.
We examine these issues i n more detail in Chapter 1 5 , but one feature of the

a poison p ill pr ovi si on wi thout an approving shareholder vote.

takeover wave does seem to be a clear manifestati on of mi sbehavi or: The ad option of

POISON PI LLS Poison pills are takeover defenses. 9 They involve creati on of special
securi ties that give certain ri ghts to their holders i n the event that a raider acquires
more than a specified fraction of the shares in the firm. M ost commonly these ri ghts
are to buy shares in the target (or, if the takeover occurs, the acquiri ng) firm at very
low prices. They work as takeover defenses because they vastly i ncrease the cost of
acquiring the firm. Although they are often labeled as "Shareholder Rights Plans" by
managers trying to sell them to their shareholders (who receive some or all of the
special securiti e.s), they in effect remove the ability of the owners of the firm to sell
thei r shares to a buyer that their nomi nal agents-management and the board-d o
not like.
If the shareholders ad opt such a scheme, i t is largely their business, and i t may
well be in their best i nterest: Takeover defenses, if not strong enough to make takeovers
impossible, may improve the stockholders' bargaining posi tion and raise the price they

8 See Michael Jensen, "Takeovers: Their Causes and Consequences," Journal of Economic
Perspectives, 2 (1988), 21-48.
9 For a discussion of takeover defenses, see J. Fred Weston, Kwang S. Chung, and Susan E. Hoag,
Mergers, Restructuring, and Corporate Control (Englewood Cliffs, NJ: Prentice Hall, Inc., I 990), Chapter
20.
ultimately receive if an acq uisition does occu r. B oards of directors can adopt poison Moral Hazard and
pills without shareholder approval, h owever, and they often did so during the 1 980s Performance
when they ( or management) became nervous about possible take overs. Doing so seems Incentives
to be simply an expropriation of the shareholders' property by those who are supposed
to be looking out fo r and servin g their in terests. M oreover, the empirical evidence is
that adopting a poison pill typically reduces the firm's share value, an d the firms that
have adopted poison pills tend to be ones where managers and board members hold
very few of the company's shares. 10 This evidence further supports the view that the
adoption does not serve shareholder interests.

Moral Hazard in Financial Contracts

A common sort of moral hazard problem arises when different individuals have
differing claims on the financial returns from an investment. We have already seen
an instance of this in the savings and loan crisis, but other examples abound and
account for many elements of the form of financial contracts.
DEBT, EQUITY, AND BANKRUPTCY M any firms are financed by a combination of debt
and equity. The debt holders-banks, the purchasers of the firm's bonds, input
suppliers who offer credit-are lenders. They provide cash in return for a promise to
be repaid a fixed amount ( perhaps with interest) at a later date. The equity holders
get to keep whatever profits are left after paying the debt obligations. In a corporation,
the equity is lodged with the stockholders, who elect the board of directors to represent
their interests in setting policy and in hiring managers to run the firm. In a partnership
or sole proprietorship, the partners or owners are the equity claimants. Absent serious
managerial moral hazard, we should expect that the firm will be run in the interests
of the equity holders. This is not necessarily consistent with the interests of the firm' s
creditors.
We already saw one form of a conflict of interest between equity and debt in
the savings and loan example. Equity holders will favor riskier investments than the
firm's creditors would want. The reasoning is exactly the same as was developed there:
The equity holders win big if the investments work out, whereas the debt holders j ust
get their promised fixed payment, and if the investment loses money, some of the
loss may fall on the creditors who are not fully repaid.
As we also noted in the savings and loan example, lenders take measures to
protect themselves against the potential moral hazard problem that arises if the firm
is run to maximize the value of its stock. They do credit checks, they demand
collateral, they monitor performance (in part by requiring ongoing repayment), and
they may structure the loans so that they can demand immediate repayment if they
get nervous about the firm's ability to pay. M oreover, in some countries, the firm' s
bankers are normally named to the board of directors of the corporation, where they
can more easily monitor their investments. As well, the bond holders may insist on
covenants in the debt contract that limit the sort of actions the firm can take and the
amount of additional borrowing it can undertake.
Despite all this, sometimes the loan cannot be repaid, or at least a scheduled
payment is missed. In these circumstances, the lenders can force the firm into
bankruptcy. Bankruptcy can be seen as an institutional arrangement to protect the
value of assets. Once a firm is forced into involuntary bankruptcy, the creditors gain
many of the decision rights that normally belong to equity. This prevents more of the
resources available for meeting the debt obligations from being squandered. It thus

10 Michael Jensen, "Takeovers: Their Causes and Consequences," 21-48.

 184
Motivation: makes people more willing to lend money than they otherwise would be and encourages
Contracts, the efficient allocation of financial capital. Moreover, to the extent that managers lose
Information, and in a bankruptcy-because their jobs, their perks, and their pensions may disappear­
I ncentives the threat of bankruptcy may serve as a check on managerial moral hazard vis-a-vis
stockholders' interests.
Under U. S. tax laws, the payments that a corporation makes as interest on its
debt are tax deductible, whereas dividend payments on its stock are not. Because both
are payments by the corporation for the use of capital, this might suggest that the firm
would gain by financing itself overwhelmingly with debt. The attendant moral hazard
problem is one reason why this would not be automatically attractive, however. As
the fraction of the firm financed by debt increases, there is a growing incentive for
equity holders and the managers who represent them to take risks. This means that
at very high levels of debt to equity, the firm will have to pay very high interest rates,
put up extremely large amounts of collateral, and accord lenders extensive control
rights if it is to persuade them to lend to it at all. Although this is far from a complete
explanation of firms' decisions about how to finance themselves, it is an element. We
will see more on this topic in Chapters 1 4 and 1 5.
OtL AND GAs TAX-SHELTER PROGRAl\1S 1 1 In the United States in the early 1 980s,
many oil and gas exploration and development operations were organized through
limited partnerships, which are hybrid contractual arrangements mixing elements of
the forms of both corporations and partnerships. There are two classes of partners in
a limited partnership: the general partners and the limited partners. The limited
partners are in a position very like that of the shareholders in a public corporation.
They take no role in managing the partnership. Rather, they simply provide the cash
as investors to finance its operations, and they enjoy limited liability: Their financial
liabilities are limited to the amounts they invest. The general partners are like the
partners in a regular partnership. They make all the managerial decisions about the
partnership's operations, and they have unlimited liability for the partnership's debts:
Their personal wealth can be seized by creditors if the partnership defaults on its
debts.
The federal tax laws that prevailed in the early 1 980s partially accounted for the
popularity of this organizational form in oil and gas exploration. The partners could
often save on taxes if the limited partners paid all the costs of exploring for oil (which
were tax deductible when the costs were incurred), whereas the general partners paid
the costs of completing wells in which oil is found (which were "capitalized costs" for
tax reporting purposes). The general partners and the limited partners would then
share any revenues enjoyed when oil was pumped from producing wells.
This tax avoidance scheme is beset with moral hazard problems that arise from
the difference in interests it creates between the general partners and the limited
partners. The most fundamental of these results because each bears a different kind
of expense and receives only a share of the revenues. If a well is found to have oil,
the general partners have to decide whether to bring it to completion so it will produce.
If they decide to do so, they bear 1 00 percent of the cost of completing the well but
typically receive only 2 5 percent of the oil revenues, with the rest going to the limited
partners. Suppose that after the exploration costs have been sunk, a well is found to
have enough oil that the well-completion costs will be just 50 percent of the resulting

1 1 This section is based on Mark Wolfson, "Empirical Evidence of Incentive Problems and Their
Mitigation in Oil and Gas Tax Shelter Programs, " in Principals a nd Agents: The S tructure of Business, J.
Pratt and R. Zeckhauser, eds. , (Boston: Harvard Business School Press, 1985), 1 01-25.
 1 85
revenues, so that the partnership as a whole would profit from completing the well. Moral Hazard and
Despite the fact that completing the well would maximize total value, the general Performance
partners would not find it in their individual interests to complete the well: Their 2 5 Incentives
percen t share of the reven ue is not enough to cover their 1 00 percent share of the
cost. Furthermore, it would be very hard for the limited partners, with no role in the
management of the partnership and probably no expertise in the oil business, to ensure
that their interests are being given proper weight in the general partners' decisions.
A second conflict of interest arises when, as was often the case, the general
partners are involved in several exploration efforts at the same time and in the same
area but have differing shares in different projects. As an extreme example, suppose
the general partners have another exploration project on an adjoining tract that they
own outright. In that case, by shifting their drilling on the partnership's tract towards
the boundaries of their own tract, the general partners can acquire valuable information
about the likelihood of finding oil on different parts of their private holdings, with
the cost of that information acquisition being borne by the limited partners. Similar,
if less severe, problems arise when the general partner is involved in several limited
partnerships but has differing interests in each. The general partner may be led to
distort his or her allocation of time, effort, attention, and resources among the
partnerships, favoring the ones in which he or she has the greatest interest. Again, it
would be very difficult for the limited partners to monitor this sort of behavior.
A third conflict often arises when the general partners or their affiliates sell
equipment or services to the limited partnership. The problem is that the general
partners have an incentive to overcharge on these transactions because the limited
partners pay the bills, but the general partners, who make the decisions, collect the
money.
All these conflicts are clearly recognized in the industry, and the prospectuses
for the limited partnerships often discuss the incentive problems very clearly and
candidly. We return to this example in Chapter 7, where some of the means used to
offset the incentive problems are discussed.

CONTROLLING MORAL HAZARD

In order for a moral hazard problem to arise, three conditions must hold. First, there
must be some potential divergence of interests between people. Conflicts of interest
will not always arise, nor will they arise on all dimensions: Different individuals'
interests may naturally be quite well aligned in particular circumstances. However,
conflict will occur often, if only because scarcity of resources means that what one
person gets another cannot have. Second, there must be some basis for gainful
exchange or other cooperation between the individuals-some reason to agree and
transact-that activates the divergent interests. Up to this point, simple market
arrangements would work: Divergent interests are a factor in almost all exchanges,
and yet exchanges are often made successfully without being troubled by moral hazard.
The critical third requirement is that there must be difficulties in determining whether
in fact the terms of the agreement have been followed and in enforcing the contract
terms. These difficulties often arise because monitoring actions or verifying reported
information is costly or impossible. However, they could also arise even when both
parties know that the contract has been violated but this fact cannot be verified by
third parties (such as a court or arbitrator) who would have enforcement powers. This
means that the normal market solution will be problematic, because the parties will
not be able to write enforceable contracts covering all the crucial elements of the
transaction. These three conditions suggest ways to deal with the moral hazard
problem.
 186
Motivation: Monitoring
Contracts,
The first remedy is suggested by the third condition: I ncrease the resources devoted
to monitoring and verification. S ometimes the idea is to prevent inappropriate behavior
Information , and
Incentives
directly by catching it before it occurs. For example, U . S. corporations are not al lowed
to publish financial statements until they have been verified by indepen dent auditors,
pr ospectuses describing in vestments for which funds are sought from the public must
be approved by the Securities and Exchange Commission, and health care insurers
may have patients obtain a second opinion on a physician's recommendation for some
expensive treatment if they are concerned that the treatments may be unnecessary. I n
other situations, monitoring is inte nded to decrease the probability of getting away
undetected with the socially inefficient, self-interested behavior. I n this case, the
results of monitoring are the basis for rewards or penalties. For example, workers are
often required to punch a time clock, and their pay is reduced or other punishments
are imposed if they arrive late or quit early. M onitoring may also be used to support
a system of rewards for good behavior.
The payment of cash rewards is i tself sometimes subject to a moral hazard
problem of renegi ng. The party who is supposed to pay the reward may misrepresent
the outcome of the monitoring, claiming that the other person' s behavior was not
appropriate and no reward is due. This is likel y to be especially easy when the criteria
for j udging performance are hard to describe or measure precisely, so that evaluations
will tend to be subjective. S ometimes, the need to maintain a good reputation is
enough to control this temptation. (Reputation effects are discussed in more detail in
Chapter 8. ) I n other circumstances, the efficacy of monitoring may depend on
generating evidence verifi able to a court that can enforce payment of the agreed
rewards.
A more subtle but related commitme nt problem arises whe n punishment is due
but carrying out the punishment is costly for the party who is supposed to do it. For
example, if company policy requires that an employee who breaks certain rules must
be fired, and a valued, hard-to-replace employee is caught in a minor violation of the
rules, then the employer may be loathe to carry out the punishment and lose the
employee's services. Of course, if the worker foresees that the firm will be unwilling
to punish transgressions, the threat of punishme nt is empty.
COMPETING SOURCES OF INFORMATION Although monitoring requires developing
sources of information about the agent's truthfulness and performance, this does not
always require direct expen ditures of resources. One possibility is to rely on competition
among di fferent parties with conflicting interests to develop the needed information.
I n everyday life, competing sellers will ofte n happily compare the rel ative merits of
their own products against comparative defects in the competing product which the
other seller would be unlikely to emphasize. The same phenomenon can occur within
organizations, as for example when the navy and air force vie for responsibility for
some military mission, each emphasizing its own advantages compared to i ts
competitor. The danger with relying on compe ting information provi ders is greatest
when they have some common interests that are in opposition to the decision maker's.
For example, neither of two sellers of asbestos insulation was likely to emphasize the
health hazards of asbestos before these became widely known.
MoNITOHING BY MARKETS Managerial moral hazard is freque ntly alleviated by
monitoring provided for free by markets. Man agers of firms in reasonably competitive
product or input markets who do a poor j ob of generatin g profits will face a greater
probabi lity of fai lure. The fear of unemploymen t an d of carryin g a reputation for
havin g led a firm into ban kruptcy may then provide managerial in cen tives. S imilarly,
 1 87
the "m arket for corporate cont rol " provides incenti ves by th reate ni ng bad corporate
man agers with loss of their j obs foll owi ng a take over or a successfu l proxy fight. 12
Moral I fazard and

Ex p licit I ncentive Contracts

Performance
Incentives

In some situati ons, monitori ng actual behavi or or the veracity of reports may be simply
too expensi ve to be worthwhile. As we me nti oned earlier, however, it may still be
possible to observe outcomes and to provide i nce nti ves for good behavi or through
rewardi ng good outc omes. For example, e ven if it is impossible to monit or the care
and skill exerted by machine maintenance personnel, it may still be possible to
measure the percentage of time that machi nes break down. In fact, if the breakdown
rate of machines were c ompletely determi ned by the performance of the mai ntenance
worker, basi ng pay on that rate would be a perfect sub stitute for basi ng it on care and
effort. The same would be true even if other factors (such as the machi ne s' i nherent
quality, the · i ntensity and nature of their use, and the care e xerted by the machine
operators) also influe nce the breakdown rate, provi ded it were possible to c ontrol
preci sely for the effects of these other determi nants of breakdown.
U nfortunately, perfect connecti ons between unobservable acti ons and observed
resulti ng outcomes are rare. M ore ofte n people's behavi or only partially determi nes
outcomes, and it is impossible to isolate the effect of thei r behavi or preci sely. For
example, a firm's total sales depend not only on the efforts of the sales force but also
on a host of other factors: the price and advertising policy of the firm, competitors'
prices and promoti ons, and other conditi ons that affect customers' demands. Rewardi ng
on the basi s of results therefore makes the salespe ople's incomes depe ndent on random
and uncontrollable factors. A simi lar effect arises whe n the outc omes are fully
determi ned by the person's effort but are not measured preci sely, instead being only
estimated or measured with some unknown, random error. Agai n, i ncome s become
subject to random variati ons.
THE PROBLEM OF RISK-BEARING M ost people dislike havi ng their i ncomes depe nde nt
on random factors. They are risk averse, and would rather have a smaller i nc ome
whose magnitude is certai n than an uncertai n i ncome that is somewhat large r on
average but is subject to unpredictable and uncontrollable variability. The ri sks created
by incenti ve contracts are c ostly to these pe ople. They are not as well off with a risky
income as they would be recei ving the same expected level of pay for certai n, and
they thus have to be pai d more on average to convi nce them to accept these ri sks.
From the employer's perspecti ve, this extra i ncome is a cost of usi ng i ncenti ve pay.
M oreover, this cost can be a real one to society, one that can reduce overall
efficiency. The employer often is more tolerant of risk and better able to bear it than
are employees. I n the extreme case, where the employer i s a well-financed and widely
held corporati on whose stockholders keep their wealth in broadly di versi fied portfoli os,
the st ockholders can be assumed to be risk neutral-concerned mostly with expected
returns and virtual ly indiffere nt ab out variability in the net earnings of the firm,
especially vari ations of the magnitude of an i ndi vi.dual worker's performance pay.
Tyi ng workers' pay to their j ob performance means that a source of the vari ability of
earni ngs is transferred from the owners to the workers: When thi ngs go well on the
j ob, some of the extra returns accrue to the workers, and whe n thi ngs go badly, the

12 The stockholders in a corporation have the right to elect the directors and to vote on certain
ma jor pol icy decisions at stockholders' meetings. Few stockholders ever attend these meetings, however.
To allow for this, they are permitted to give their proxy to someone else to cast their votes on their behalf.
Typically, the proxy is given to management. In a proxy contest, rival groups will attempt to win stockholders'
proxies so that they can elect different directors or prevent management and the current directors from
enacting a policy change that the group opposes. See Chapter 1 5.
 1 88
Motivation: impact on the owners is cushi oned by the lower levels of i ncenti ve pay. H owever,
Contracts, transferri ng ri sk from the owners (who care li ttle about the risk and benefi t li ttle from
I nformation, and i ts reducti on) to the workers (who may strongly disli ke beari ng ri sk) means that the
I ncentives total costs of the given amount of risk i n the system are increased.
RISK COSTS AND I NCENTIVE BENEFITS Desi gni ng effi cient i ncentive contracts i nvolves
balanci ng the costs of risk beari ng agai nst the benefits of improved i ncentives.
Insulati ng ri sk-averse employees' pay from vari ati ons i n measured j ob performance
mi nimizes the costs of risk beari ng, but i t also elimi nates monetary performance
i ncentives. Shifti ng risk to the employees strengthens their i ncenti ves because their
pay now depends on actual performance, but the costs of risk beari ng ri se as well.
The effi ci ency pri nci ple suggests that observed contracts will tend to be effi ci ent,
subj ect to the constrai nts i mposed by observabi li ty problems.
O ne implicati on of thi s analysi s is that i t is i neffi cient to use contracts that make
ri sk-averse employees bear avoidable ri sks unless the contracts also provi de useful
i ncenti ves. 1 3 For example, consider a fi rm whose ri sk- neutral owners want to maximize
profi ts and whi ch has a problem moti vati ng producti on workers to be productive.
Because the owners care about profi ts, one possi bility i s to provide i ncenti ves for
everyone by payi ng bonuses based on profitabi li ty. This exposes workers to i ncome
vari ati ons ari si ng not j ust from thei r own producti vi ty, however, but al so from all the
other factors i nfl uenci ng profi ts that are beyond their control: i nput prices and
avai labili ty, the efforts of the sales force, the quali ty of executive decisi ons, variati ons
i n demand and in the i nterest rate the firm has to pay on i ts debts, the acti ons of
competi tors, and so on. In thi s case, i t may be preferable to use an i ncentive plan
based not on profi ts but on direct measures of the contri buti ons made by i ndivi dual
workers or work group, such as the volume of output, the number of defects, the
number of days absent from work, and so on. Even these measures expose the workers
to ri sk, because productivi ty is not completely under thei r control, but they do i nsulate
them from some unnecessary ri sks.
The basic idea behi nd i ncentive contracts is that of achievi ng goal congruence:
An appropriately designed reward system causes self-i nterested behavi or to approximate
the behavi or the desi gner wants. Alternati vely, we can thi nk of a well-desi gned
incentive scheme as removi ng the conflict of i nterests by effectively alteri ng i ndi vidual
obj ecti ves, aligni ng them more closely wi th those of the desi gner. We wi ll usually
thi nk of i ncenti ves as altering rewards to i ncrease the benefits associ ated wi th the
desired behavi or; for example, moti vating employees' i nterest i n profit seeki ng by tyi ng
thei r pay to profi tabi li ty. H owever, behavi or can also be modi fied throu gh j ob desi gn,
employee i nvolvement programs, and the provisi on of a better work environment, all
of which reduce the unpleasantness of work and lower the costs to employees of
providi ng effort. Requi ri ng office workers to be at thei r desks duri ng certai n hours can
be seen i n similar terms. Because they have to be at the office, they may as well do
their j obs, although if they were free to be elsewhere, they would fi nd other things to
occupy thei r time.
We give a (relati vely) si mple mathematical example of what i s conceptually
i nvolved i n desi gni ng an effici ent i ncenti ve contract i n the appendi x to thi s chapter.
In Chapter 7 we exami ne thi s issue i n much more detai l and develop a number of
pri nci ples that can be used to understand and evaluate actual con tracts and to guide
contract design. Also, in Chapter 12 we examine managerial issues that arise i n usi ng
ince ntive pay in organ izations.

n The poi nt is closely related to the adage that people should be held responsible only for things
under their control. Actually, the adage with this phrasing is misleading, as seen in Chapter 7.
 1 89
Bonding Moral Hazard and
In some industries, it is common to req uire the posting of bonds to guarantee per­ Performance
formance. The bond is a sum of money that is forfeited in the event that inappropriate Incentives
behavior is detected. For example, contractors often must post a bond that they lose
if the proj ect is not completed by the agreed date and in the agreed manner. S imilarly,
the capital provided by the owners of a bank or an S&L acts like a bond because in
the event of losses the capital must be paid out to meet obligations. In the early 1 970s,
Electronic Data Systems Corporation (EDS)-Ross Perot's computer service company
that was later acq uired by General M otors-req uired trainees who resigned within
three years of j oining the firm to pay the firm $12, 000. 14 This bonding was designed
to prevent employees from receiving costly training without doing substantial work for
the firm. The $ 12, 000 amount was comparable to an engineer's annual salary at the
time. '
Posting a bond can be a very effective way to provide incentives, but the problem
is that people often will lack the financial resources to post a sufficiently large bond.
This is especially the case when the gains from cheating are large and the probability
of getting caught is small, so that the bond would have to be large to give an adequate
incentive. These ideas are examined more carefully in Chapter 8, but one application
that sheds light on the puzzle of positively sloped age/wage profiles can be discussed
here.
AGE/WAGE PATTERNS, SENIORITY PROVISIONS, AND MANDATORY RETIREMENT As noted
in Chapter 5, pay tends to increase with age and experience, even after controlling
for productivity. In Chapter 5 we offer an explanation for this pattern based on
inducing self-selection to reduce employee turnover. B onding as a deterrent to
employee shirking has been suggested as an alternative explanation by Edward Lazear.
S uppose that the firm can fire any workers detected shirking. We may think of
workers who shirk as receiving some valuable benefit, such as a reduced level of stress

If workers were to post bonds of sufficiently greater value than these benefits, and if
or more time to pursue personal interests, which cannot be taken away from them.

being caught cheating resulted in losing the bond, then they would not cheat. Their
value to the firm would be increased by the bond and, with competition among
employers, so too would be the amount they would earn. In any case, when it is

achieved. If the gain from cheating is substantial, however, or the likelihood of getting
efficient for workers not to cheat and shirk, the bond may allow effi ciency to be

caught is small, workers may not be able to afford to post a big enough bond, and
the potential efficiency gain would be lost.
Suppose the firm in this circumstance makes a credible promise to the workers

and thus of what they could earn elsewhere. If the firm pays workers less than their
that, late in their careers, it will pay them more than the value of what they produce

marginal products early in their careers, then the value of lifetime earnings and the
firm' s total outlay need not be affected by this scheme. As the years of high pay draw
near, however, the high promised wages serve as a bond that the worker would forfeit
by dishonest behavior or shirking: The wage pattern duplicates the effect of a bond.
Therefore, the observed pattern of wages might be explained by a need to make
workers value their j obs in order to ensure honest, hard-working behavior.
S trikingly, a mandatory retirement provision will be necessary for efficiency

produce j ust equals the private cost to them of continuing working. If they were paid
under this scheme. For efficiency, people should retire when the value of what they

14 Doron Levin, Irreconcilable Differences: Ross Perot versus General Motors (New York: Plume,
1989), p. 46.
 190
Motivation: their marginal products, they would choose to retire at the efficient date, when the
Contracts, extra income j ust balances the increasingly high costs of continuing work. With wages
Information, and late in life exceeding marginal productivity, however, some people will want to
Incentives continue working too long because their pay exceeds the social value of their output.
They will not retire voluntarily at the appropriate date. Thus, mandatory retirement
is necessary for efficiency. Furthermore, at the start of their careers, workers would
be happy to sign contracts agreeing to a mandatory retirement date, even though once
it arrives they will be unhappy about being forced to stop working and earning.
This scheme also necessitates some sort of mechanism to make the firm's
promise credible. The danger is that the firm will renege on its agreement by letting
senior workers go once their pay exceeds their current productivity. Again , a concern
with reputation may work here, but it is perhaps less l ikely to be effective when workers
have few other employment options and so the incentives to avoid a firm that has
cheated are weak. In this case, a seniority rule in layoffs can play a useful role. If the
firm wants to lay off a senior worker who is earning a lot, it must first lay off all the
more jun ior people whom it is paying less than they are worth .

Do- It-Yourself, Ownership Chan ges,

and Organizational Redesig n
Moral hazard in agency settings can sometimes be overcome by eliminating the agent
and having the principals act on their own behalf. This is often impossible, however­
you cannot very well be your own surgeon, for example-and in any case it sacrifices
the gains of specialization .
In market settings, changing ownership patterns to bring the affected transactions
within a single organization can help overcome some moral hazard problems. (We
have already noted in Chapter 5 that unified ownership can be a response to
inefficiencies arising from bounded rationality and private information . ) For example,
if a firm and a supplier are frequently involved in complex transactions that are marked
by such manifestations of moral hazard as possible cheating on quality, it may be
efficient for one firm to acquire the other. In that case, the differing interests (each
firm's own profit) become merged, and many of the incentive problems are overcome.
A simple mathematical example illustrates how this can be effective.

INCENTl\'ES AND 0\'\ 'NERSHIP PATTERNS: AN EXAl\tPLE Suppose that two firms face a
joint opportunity that requires both of them to make some investment. Let us denote
the measured value of the investments by firms A and B by MA and M8 . These
measured amounts need not be the assets' actual values, however, and this is where
moral hazard enters because the firms' wills individually decide the actual amounts
they will invest. For example, a physical asset m ight be valued at $ 1 00 in the accounts
because it cost $200 a year ago and had an estimated useful life of two years. However,
the actual value of that asset today might be only $ 5 0 because new machines using
a new process have made the old process obsolete. Similarly, the value of a year of
an employee's time might be recorded at an amount equal to his or her wage, but
the wage might not accurately reflect the e mployee's value to the company. The
employee might be a young engineer, straight out of school, who is expected to go
through a period of learning and low productivity (when one subtracts the cost of
mistakes). Or, the employee might be an up-and-coming executive who has successfully
managed other special projects and is more val uable to the company than others of
the same rank.
Let VA and V8 be the actual value to the two companies of the resources they
invest in the venture. The expected reven ue from the venture is assumed to be 1 . 5
 191
Table 6.4 Company A's Profit Calculation Moral Hazard and
Performance
Incentives
B's investment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VB
A's investment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VA
Total investment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VA + VB
Total expected revenue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 . 5(VA + VB) - 600
A's expected revenue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0. 75 x (VA + VB ) - 300
(a one-half share)
A's net expected profit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0. 7 5 x VB - . 2 5 x VA - 3 00
(revenue minus investment)
A's profit-maxi mizing choice . . . . . . . . . . . . . . . . . . . . . . . . . . VA = 500
B's e�pected choice VB = 500
A's expected loss . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . $50

times the actual value of the total investment minus $600, that is, I. 5 X (VA +
VB) - 600. Higher real investments lead to more revenue for the project on average.
Suppose the two companies write a contract according to which each will invest
$ 1 , 000 in the project. This can only refer to measured investments, because even if
the companies were able to distinguish the values of the assets being used, there would
be no objective way for a court to verify the level of unmeasured investment in order
to enforce the agreement. As equal partners, the companies agree to divide the
revenues generated by the venture equally. If the companies were to invest so that
measured and actual values were the same, that is, so that MA = MB = VA =
VB = $ 1 , 000, then the investment would be profitable for both companies. The total
revenues generated by the project would be 1 . 5 x (VA + VB ) - 600 = $2, 400,
which is more than the total investment of $2, 000. Each company would expect to
receive $ 1 , 200 in revenues from an investment of $ 1 , 000, for a net profit of $200.
Now, suppose that each company is free to choose its investment in the project
so that, even though the measured value is exactly $ 1 , 000, the actual value can be
anything between $500 and $ 1 , 500. What choices wi ll the companies make? Will
the venture be a profitable one?
Let's look at the matter from company A's perspective. Whatever amount VB
company B actually invests, company A's expected profits will be its share of the total
revenues minus its investment (see Table 6. 4).
According to Table 6. 4, each extra dollar of value that fi rm A invests in the
venture yields $ 1 . 50 in extra revenue, but A's share of that $ 1 . 50 is just $0. 75.
Therefore, A loses $0. 25 on each extra dollar of investment. A's profit-maximizing
choice is to make the minimum investment of $ 500, as indicated in the seventh line
of the table. If A understands this calculation (and expects that B does too), then it
will expect B to invest only $500 as well. In that case, the total expected revenues of
the venture will be just $900, whereas the required investment will be $ 1 , 000. Each
firm wi ll expect to lose $50, so the investment wi ll not be made . An apparently
profitable busi ness opportunity wi ll be lost because A expects B to take a free ride on
their investment and B expects A to do likewise.
Of course, if the two firms were to merge, and the decisions about investment
were brought under a single individual who pays the entire costs and collects the full
benefits, then the problem described here would be eliminated .
Tt IE DETERI\I INANTS OF OWNERSHIP As our examples related to moral hazard in
employment have shown, a transaction that is "integrated" or brought "in house" does
 192
Motivation: not automatically align incentives. The problem is that integration only transforms a
Contracts, self-interested manager who formerly worked for the supplier into a self-interested
Information , and manager who works for the fi rm. The basic ince ntive problem may still need to be
Incentives solved.
The upshot is that merger does not always eliminate the incentive problem that
exists between separate fi rms. Compounding the problem is the fact that there are
additional unavoidable costs to bringing previously separate activities under common
direction. An important component of these are the infiuence costs, which increase
with the increased pote ntial for central control of activities in an integrated organization.
Although the infiuence activities that give rise to influence costs are a form of moral
hazard, they are of such importance in unde rstanding organizations that they deserve
separate treatment.

I NFLUENCE ACTIVITIES AND UNIFIED OWNERSHI P

What costs are involved in bringing two separate organizations under unified direction?
Why can't the merged entity do everything the separate compone nts did and more?
What are the limits, if any, on the efficient size of organizations? Why isn't all
economic activity organized in a single firm?
From our discussion in Chapter 2, it is clear that the answer must be that
bringing everything within a single organization involves inefficiently high transactions
costs of some sort. But what are they? I n fact, until recently, little attention has been
give n to the task of identifying the transactions costs of internal, nonmarket
organization. This is a subtle matter. I n actual organizations, much time and inge nuity
is spe nt overcoming transactions costs: Witness the example of the development of
the multidivisional form in Chapter 1. M ore over, a strikingly simple idea-the policy
of selective interve ntion-undercuts many of the possible candidates that come to
mind as distinctive disabilities of unified control.

Un ified Ownership and Selective I ntervention

Suppose it is efficient for two parts of a big organization to be independent and operate
as separate entities. The n, in the original organization the center could direct the two
units to conduct their transactions at arm's length as if they were not both part of a
single structure. For example, whe n market transacting works well, why not replicate
its operation within the fi rm, using internal, transfer pricing? Meanwhile, where there
are efficiency gains to be had from deviating from the patterns of transactions that
would occur in the market, why not have the central manageme nt selectively intervene
in the operations of the component units to e nsure that the gains are realized?
Following a policy of selective interve ntion consiste ntly ought to mean that the
unified organization can do everything the separate pieces could do, and do so at least
as well. There would the n be no bound on the efficient size of the organization. Why
then is all activity not brought under a single fi rm? The logical answer must be that
adhering to a thoroughgoing policy of selective intervention is impossible. But why
should this policy be infeasible? Influence activities provide part of an answer. 1 5
I nflue nce activities arise in organizations whe n organizational decisions affect
the distribution of wealth or other be nefi ts among members or constitue nt groups of
the organization and, in pursuit of their selfi sh interests, the affected individuals or
groups attempt to influence the decision to their benefi t. The costs of these influe nce
activities are influence costs.

15 A broader discussion of influence costs is found in Chapter 8.

 1 93
Infl uencing Interven tions Moral Hazard and

The fundamental difficulty with the policy of selective intervention is that it requires
Performance
Incentives
that there be a decision maker with the power to intervene wh o collects information
with which to make decisions: These th ings can by them selves impose costs on the
organization. The most obvious costs are the decision maker's salary and the cost of
providing information to support the decision-making system, including the time that
lower-level decision makers spend reporting information to the decision maker. Often
more important is that individuals and units within the organization may have selfish
reasons to seek unproductive interventions, and they may expend resources trying to
infl.uence the decision maker to bring them about. Even when the attempts fail, th e
resources expended in these influence activities represent a cost that brings no offsetting
gain. ·When they do succeed in influencing the central decision maker to intervene
inappropriately, there are further costs in bad decisions being made and implemented.
Finally, if the organization recognizes these possibilities and adj usts its structure,
governance, policies, and procedures to control attempts at influence, these deviations
bring further costs. All of these are elements of influence costs.
As is evident, the magnitude of influence costs depends on the existence of a
central auth ority, the kinds of procedures that govern decision making, and the degree
of h omogeneity or conflict in the interests of organization members. All th is is treated
in more detail in Chapter 8. Here, we focus on how influence costs limit the optimal
scope of formal organizations.
Wh en two previously separate organizations are brough t under a common,
central management with the power to intervene, the scope for influence increases
and influence costs increase. For example, members of one unit can try to influence
top management to transfer resources from the other unit to theirs. They can argue
that they have better investment opportunities and so can better use the fu nds being
generated in the oth er division, or that they h ave more valued uses for th e most
talented people now assigned to the oth er group, or that all marketing, or pr oduction,
or research and devel opment (R&D) sh ould be consolidated in a single unit (theirs!)
rather than remaining inefficiently spread over several units. The other group will
have a similar incentive to defend itself and even to counterattack. It can argue that
other units sh ould be required to purchase its outputs, even th ough the outside market
may provide superior or cheaper substitutes, because doing so h elps cover corporate
overhead or hel ps build the firm's core competencies, or it may complain that equity,
morale, and ultimately pr oductivity demand that its members be paid as well as th ose
in another group wh ose members may be especially productive or may h ave particularly
valuable skills or knowledge. Large amounts of time, ingenuity, and effort may go
into these attempts at influence, and huge amounts of the central executives' time
can be consumed dealing with them.
Of course, none of this would occur if there were no central auth ority with the
power to make the proposed changes. Thus, alth ough the merged organization may
be able to ach ieve things that were not possible before, it also suffers costs that were
not present when the parts were separate.

Influence Costs and Failed Mergers

Th is logic gives insight into the great frequency with which corporate mergers and
acquisitions apparently fail. In a study of the diversification records of 3 3 large U . S .
corporations between 1950 and 1986, Michael Porter found that fu lly 60 percent of
the acquisitions i n new fields of business by these firms were later divested, and 61
 194
Motivation: percent of the firms ended up divesting more of their acquisitions than they kept. 1 6
Contracts, Although not all divestitures of previous acquisitions necessarily represent failures,
Information, and even the sophisticated firms in Porter's sample had real problems making acquisitions
Incentives work.
There are obviously major problems involved in attempting to integrate two
different organizations with their own unique histories, their own ways of doing things,
their own reporting and control systems, their own pay and benefit schemes, and so
on. To focus on a pure case in which these factors should be of minimal concern,
consider a pure conglomerate merger in which one firm acquires another with the
intent of running it as a completely separate division, intervening in its operations
only when there are clear gains to doing so. Even here, influence costs present
problems that may cause the merger to fail.
TENNECO'S ACQUISITION OF HOUSTON OIL AND MINERALS A well-documented example
is the 1 980 acquisition of Houston Oil and Minerals Corporation by Tenneco, Inc.,
which was then the largest conglomerate in the United States. 1 7 Houston's business
was finding, developing, and bringing petroleum and mineral deposits into production.
The company was very aggressive and quite successful before its acquisition by
Tenneco. Tenneco's stated intent was to run Houston as a separate company,
maintaining the entrepreneurial, risk-taking style that had marked it as an independent
concern. In particular, it planned to maintain a separate compensation and reward
system at Houston that would provide unusually large individual payoffs to professionals
for successful discovery and development of petroleum reserves. (Several Houston
explorationists had become wealthy with the bonuses they earned from successful
explorations, and such packages were common among smaller firms in the industry.)
Yet Tenneco had great difficulty developing such a plan, and it ultimately failed to
do so. Within a year, more than a third of Houston's managers, a quarter of its
exploration staff, and almost a fifth of its production people had left the company for
better opportunities elsewhere. This severely hampered operations, and ultimately it
became impossible to maintain Houston as a distinct unit within the firm.
Tenneco's costly failure to institute the intended reward policy apparently
resulted from a concern for equity in pay across the organization. The Tenneco Vice
President of Administration was quoted in The Wall Street Journal as saying: "We
have to ensure internal equity and apply the same standards of compensation to
everyone. " Meeting this perceived need was very costly. It contributed to the exodus
and to the ultimate failure of the acquisition. Failure to meet this need might well
have been even more costly, however. Tenneco's 1 00, 000 employees, jealously looking
at the huge bonuses that would have been paid to the few hundred Houston
professionals, might have consumed large chunks of their superiors' time with their
jealous complaining and their attempts to get some of these funds for themselves.
Given the relative sizes of the two groups, the overall impact on productivity at
Tenneco could have been disastrous.

1 6 Michael Porter, "From Competitive Advantage to Corporate Strategy, " Harvard Business Review
(May-June 1 987), 43-59.
1 7 This example is discussed by Oliver Williamson in The Economic Institutions of Capitalism (New
York: The Free Press, 198 5), p. 1 58. The primary source is George Cetschow, "Loss of Expert Talent
Impedes Oil Finding by New Tenneco Unit, " The Wall Street Journal (February 9, 1982), A-1. The
quotation in the next paragraph is from this story.
 1 95
Moral I Iazard and
Performance
I nccntives
SUl\lMARY
The term moral hazard originated i n the insurance industry, where it referred to the
tendency of people who purchase i nsurance to alter their behavior in ways tha t are
costly to the insurance compa ny, such as taki ng less care to preve nt a loss from
occurri ng. Within economics, the term ha s come to refer to any behavior under a
contract that is inefficient, arises from the di fferi ng i nterests of the contracti ng parties,
and persists only becau se one party to the contract ca nnot tell for sure whether the
other i s honoring the contract terms. M oral hazard problems arise fre quently i n
principal-agent relationships, where o ne party (the "agent") i s called upon to act on
beh alf of another (the " pri ncipal"), because the agent's interests commonly differ from
the principaCs and the pri ncipal cannot evaluate how well the agent ha s worked or
whe ther the agent ha s been honest.
The savings and loan crisi s in the U nited States illustrate s the problem of moral
hazard and how it i s most often dealt wi th i n ordi nary busi ness transacti ons. The
difference of interests be tween the owners of the S&L and the federal i nsurance age ncy
(the FSLIC) arose becau se the owners be ne fited from ri sky i nve stments when they
turned out well, but the costs of failures are borne by the insurance age ncy. When
the age ncy failed to monitor and contr ol the S&L, th is led the S&L manageme nt to
make ri sky inve stments or even to engage in frau d in a wa y that was costly to taxpayers.
Competition for depo sitors' funds only i ntensi fied this effect, increasing the i nterest
rates paid on deposits and forcing m ore conservati ve S&L management to find higher­
yieldi ng-a nd hence usually ri skier-i nvestments.
The relati on of depositors to their S&L is sim ilar to the relationship of lenders
to a ny other kind of firm. N ormally, lenders protect their money by imposing controls
and requiring reporting by a nd audits of the borrower. What distinguished the S&L
case is that the de positors, bei ng insured, had little reason to m onitor the savings
institutions, and the federal government, whose money wa s at ri sk, di d not monitor
on its own behalf, in part becau se powerful congressmen were protecting the S&Ls.
Problems of fraud and excessi ve ri sk taking similar to the S&L pro blem can be
fou nd i n ma ny federally insu red programs. Among those described are programs
insuri ng workers' retirement benefits, farmers' cr ops, m ortgage loa ns, and student
loans. Similar pr oblems exist i n private-sector i nsurance programs, but these tend to
be less severe partly becau se profit-oriented insurers monit or the i nsured more carefully
and partly because private i nsurers refuse to offer insurance when the moral hazard
problem is too severe.
Moral hazard is not only a problem of markets, but exists in other kinds of
organizations as well. Air traffic controllers seeking to collect di sabi li ty benefits have
"punched out, " cau sing inci dents that seemed to indicate that they suffered j ob-related
stress and were u nable to continue their duties safely. The general partners i n oil a nd
gas drilli ng partnerships sometimes fail to com plete we lls that are profita ble for the
partnership as a whole because their own interests differ.
Various means are availa ble to control the moral hazard pro blem. O ne is explicit
monitoring, which can reduce the information problem that i s a fundamental
component of m oral hazard. A second is the use of incentive contracts that pa y for
output performance when i nputs cannot be measured. Posting a bond that i s forfeited
if the agent is caught cheating can be effective in a princi pal-agent relatio nship. This
bond ca n be implicit in the risi ng pattern of wages over a worker's career: A worker
caught cheating after several years of employment sta nds to lose the high wages paid
to more senior workers. Sometimes, the whole problem can be avoided by the " do-
 196
Motivation: it-yourself " solution, which does not rely on an agent with differing interests. Similarly,
Contracts, a firm can sometimes eliminate conflicts of interest with its suppliers by integrating
Information , and vertically, though this does not eliminate any individual differences of interests between
Incentives the formerly independent manager of the supplier and the same person who is now
an employee of the firm.
An especially important category of moral hazard is the category of inf/.uence
activities and the associated costs, known as infl.uence costs. These arise when
employees divert effort to influence organizational decisions. Even if those decisions
are not ultimately affected, the time, effort, and ingenuity devoted to attempts at
influence are unavailable for more productive activities. Influence costs are one of
the important costs of centralized control and help to explain the importance of
organizational boundaries. These costs are largely eliminated when there is no decision
maker with authority to make the decisions that employees wish to influence, and
this condition can sometimes be brought about by creating legal or other boundaries
between operating units.

• BIBLIOGRAPHIC NOTES
As with so much else in the economics of organizations, the problems of
misaligned incentives and what we now call moral hazard were noted and
understood by Adam Smith: see his discussion in The Wealth of Nations of the
incentives in joint stock companies (Book V, Chapter I, Part III, Article I) and
of the incentives for university teachers (Book V, Chapter I, Part III, Article II).
The nature of moral hazard and its importance to economic analysis was made
explicit by Mark Pauly in the context of an article on the economics of health
insurance by Kenneth Arrow. The principal-agent model, which underlies much
of the discussion in this chapter, has several early contributors, including James
Mirrlees, Michael Spence and Richard Zeckhauser, and Steven Ross. More
recent references are given in the next chapter . The explanation of the age/wage
profile and mandatory retirement in terms of bonding is due to Edward Lazear.
A useful discussion and development of Lazear's bonding model is given by Lorne
Carmichael.
Adolphe Berle and Gardner Means began the debate about whether the ownership
structure of the modern corporation has made it particularly susceptible to
managerial moral hazard. The papers published in the Journal of Law and
Economics [26 (June I 98 3)] from the Hoover Institution conference on "Corpora­
tions and Private Property" held to commemorate the fiftieth anniversary of the
publication of the Berle and Means book give some flavor of current thinking on
this issue, which in turn has been central to the debate over hostile takeovers.
The "Symposium on Takeovers" in the Jou rnal of Economic Perspectives [2( 1 988),
3-82] provides an ovef\'iew of economists' views on this debate, which we consider
in Chapter 1 5 .
The modeling of bankruptcy as a means for effecting efficiency-enhancing changes
in control is due to Phillippe Aghion and Patrick Bolton. The determinants of
firms' financing decisions are treated in Chapters 1 4 and 1 5, and further references
are given there.
The importance of the policy of selective intervention was emphasized by Oliver
Williamson. The concept of influence costs as a major element of the transactions
costs of nonmarket organization was developed by the present authors. The theory
of information provided by competing sources is developed in our Rand Journal
of Economics paper.
 1 97
Moral Hazard and
• REFERENCES
Performance
Aghion, P. , and P. Bolton. "An 'Incomplete Contract' Approach to Bankruptcy Incentives
and the Financial Structure of the Firm, " IMSSS Technical Report, No. 5 36
(Stanford, Ca: Stanford University, 1988).
Arrow, K. J. "Uncertainty and the Welfare Economics of Medical Care, " American
Economic Review, 5 3 ( 1963), 94 1-73.
Berle, A. , and G. Means. The Modern Corporation a nd Private Property (New
York: MacMillan, 1 9 32).
Carmichael, L. "Self-Enforcing Contracts, Shirking and Life Cycle Incentives, "
fournal of Economic Prespectives, 3 ( 1989), 6 5-8 3 .
Lazear, E . "Why i s There Mandatory Retirement?" fournal of Political Economy,

Milgrorn, P. , and J. Roberts. " Relying on the Information of Interested Parties, "
87 (December 1979), 1 26 1-84.

Rand fournal of Economics, 1 7 ( 1986), 1 8-32.

Milgrom, P. , and J. Roberts. "Bargaining Costs, Influence Costs, and the

J. Alt and K. Shepsle, eds. (Cambridge: Cambridge University Press, 1990).

Organization of Economic Activity, " in Perspectives on Positive Political Economy,

Mirrlees, J. "An Exploration in the Theory of Optimum Income Taxation, "

Review of Economic Studies, 38 ( 197 1 ), 1 7 5-208.
Pauly, M. "The Economics of Moral Hazard, " American Economic Review, 58
( 1968), 3 1-58.
Ross, S. "The Economic Theory of Agency: The Principal's Problem, " American
Economic Review, 63 ( 1973), 1 34-39.
Spence, A. M. , and R. Zeckhauser. "Insurance, Information and Individual
Action, " American Economic Review, 61 ( 1 97 1 ), 380-87.
Williamson, 0. The Economic Institutions of Capitalism (New York: The Free
Press, 1985).

EXERCISES

Food for Thought

1 . Widespread fraud brought down some S&Ls. Does deposit insurance itself
make fraud more attractive? How did the changes in regulation of the S&Ls contribute
to the problem of fraud?
2. As part of a plan to rescue the savings and loan industry, in late 1988 the
U. S. government encouraged private investors to purchase the assets of failing savings
and loans. They offered guarantees of principal and, in some cases, interest on some
of the properties taken over by the S&Ls when borrowers defaulted on loans. What
effect would you expect these guarantees to have on the behavior of the new owners?
3 . In automobile collision insurance and health insurance, the insurance
policy often has a provision calling for a deductible according to which the portion of
any insured loss up to some fixed limit, such as $ 500, is paid for by the insured
person; only the excess is paid for by the insurance company. In addition, health­
insurance policies often provide for copayments by the insured, according to which
the insurance company pays only some fraction, such as 80 or 90 percent, of the
medical costs in excess of the deductible, until the insured has paid some maximum
amount (such as $2, 000 in a year). What economic function do deductibles and
 198
Motivation: copayment provisions serve? Why do we see deductibles but not copayments used in
Contracts, automobile insurance policies?
I nformation, and 4. Farmland on which annual crops are grown is often rented, but orchards
I ncentives and other perennial crops are more often grown by the owners of the land. H ow can
this be explained?
5 . In a possibl y apocryphal story, a nineteenth-century English traveler in
China was shocked that the oarsmen rowing the boat in which the traveler was riding
were brutally whipped by a ferocious overseer if they slacked off on their rowing. The
traveler was even more shocked (so the story goes) to learn that the oarsmen owned
the boat and hired the overseer to beat them! H ow would you expl ain this?
6. We attribute the prevalence of li mited partnerships in oil and gas expl oration
to the tax advantages of having some cl aimants pay some of the costs and other
claimants pay other cost elements. But the tax advantages might have been achieved
by a regul ar partnership in which different partners had different forms of claims.
What are the advantages then of the limited partnership form of organization?
7. S tanford U niversity's H onor Code forbids faculty from moni toring students
during examinations. What effect would you expect this to have on student behavior?
On the rel ationships between students and fa culty?

Quantitative Problems I

1. Suppose there are two firms, Firm A and Firm B, that are consider­
ing making a j oint investment in R&D. The total payoff from the proj ect is 200 x
(VA + Va) 112 , where VA and Va are the val ues of the two investments. The two firms
expect to share this payoff equally while each absorbs the cost VA or Va, of its
investment. S how that the val ue-maximizing pl ans are those where the total investment
of the two firms is $10, 000. H ow much total val ue is created in this way?
2. l n question 1, suppose now that the firms cannot enforce a contract
specifying levels of investment for each, because they cannot observe the real val ue
of the investments that are made. Show that if Firm A expects Firm B to invest Va,
it can do no better than to invest 2, 500 - Va · H ow much, then, will be invested in
total? H ow much total val ue is created?
3. S uppose that if the firms do not sign a contract, that they can devel op their
own versions of the research project in competition with one another. A marketable
product will resul t for either firm provided it spends at least 3 5 percent as much as
its competitor. lf VA 2:: . 35 X Va, then Firm A's net profit will be 200 X (VA -
. 35 x Va) 112 - VA , and correspondingl y for B. Show that if each firm expects the
other to invest 10000/.65, then it will choose to invest an equal amount. What will
the total profits be? Woul d you expect the firms to reach a j oint venture agreement
under these circumstances?
4. (Forcing con tracts). l n principal-agent problems of efforts provision, moral
hazard may not be a problem if the structure of uncertainty all ows the pri ncipal to
infer precisel y whether the agent has failed to perform as desired. To see this, suppose
in the context of the exampl e i n the Appendix that the matrix relating the probabil ities
of various outcomes were changed so that the high level of revenues (30) was sure to
occur if the worker supplies the high level of effort (e = 2), but either level of revenue
could still happen if the low level of effort (e = 1) is provided. Thus, the first row of
Tabl e 6. 5, corresponding to e = 1, is unchanged, but the entri es corresponding to e
= 2 become O and 1 in stead of 1/3 and 2/3 . Rewrite the incentive and participation
c onstrain ts and show that it is possible to design a forcing contract that moti vates the
worker to work hard, suppl yin g e = 2, without placing any risk on him or her. H ow
 199
much is the worker paid if revenues of 1 0 are realized? How much when revenues
are 30? What arc the expected utilities of the two parties? Is there any cost in this
Moral I l azard and

case to effort not being observable, that is, could the parties do better if effort were
Performance
Incentives
observed? Would it be possible to achieve this sort of result if the low level of effort
surely resulted in revenues of 1 0, but the high level of effort could result in either
revenues of 10 (with probability 1/3) or 30 (with probability 2/3)? Why or why not?
5 . (Selling the firm to a risk-neutral agent). In many principal-agent problems
of effort provision , moral hazard is costly if the agent is risk-averse because maki ng
his or her pay reflect the full marginal impact of his or her effort choices imposes
costs on the agent that could be avoided if the risk-neutral principal absorbed the
variability in incomes . Again in the context of the example in the Appendix, show
that if the agent is risk-neutral, with utility function for income w and effort e of
u(w,e) = w - (e - 1 ), then it is possible to acheive the same expected utilities for
both parties as would result if effort were observable by having the agent bear all the
risk and the principal receive an amount that is independent of the realized level of
revenues,
 200
Motivation :

APPENDIX: A MATHEMATICAL EXAMPLE

Contracts,

OF INCENTIVE CONTRACTING*
Information, and
Incentives

The purpose of this appendix is to develop a relatively simple example of what is

conceptually required to determine an efficient contract in the presence of unobservable
effort and the consequent moral hazard problem. In the next chapter we develop these
matters much more fully, although with less visible mathematics. The example is a
principal-agent problem. As we mentioned earlier, in the standard language of
incentive theory, an agent is someone who does work on behalf of another person,
called the principal. We will think of agent as a worker and principal as an employer.
Suppose the principal is risk neutral, caring only about the expected amount of
money he or she receives, and the agent is risk averse and also averse to providing
more than a minimal amount of effort. In particular, suppose the agent evaluates
wage income and effort according to a utility function of the form U(w, e) = \lw -
(e - 1), where w is the wage and e is the effort level. According to this mathematical
formula, the marginal utility of income is 1/(2\lw), which is decreasing in the wage
level. As we will see in Chapter 7, this property corresponds to risk aversion. The
(e - 1 ) term is the cost of effort, and its form reflects the idea that providing effort is
costly only when effort exceeds 1 unit.
Suppose that two effort levels are possible: e = 1 and e = 2. The agent also
has outside job opportunities, and to get him or her to accept employment the agent
must be offered at least as much satisfaction as he or she can get working elsewhere.
We model this by imposing a requirement that the job provide the agent with a utility
level of at least some expected utility u, which we can interpret as the expected utility
value of his or her next-best alternative. To keep the arithmetic simple, let us suppose
that this minimum acceptable utility level u is 1 .
The agent's efforts are assumed to help increase the revenues of the firm.
Depending on those efforts, various possible levels of revenue might be received.
However, the outcome also depends on random factors that neither the principal nor
the agent can observe or control. Table 6. 5 gives the probabilities of the possible
outcomes for each level of effort. For example, when e = 1 , Revenue is 1 0 with
probability 2/3 and 30 with probability 1/3 .
With e = 1 , the expected revenues are (2/3) X 1 0 + ( 1/3) X 30 = 50/3,
whereas lifting the effort to e = 2 raises the expected receipts to (1/3) x 1 0 + (2/3) x
30 = 70/3 . Effort is productive in raising the probability of the good outcome and
thus the expected receipts.
If e were observable and the parties wanted it set at 2, the solution would be for
the contract to specify that e = 2 and that the agent be paid enough to get him or
her to agree to take the job when he or she provides e = 2, and to be paid nothing
if the agent picks e = 1 . Because the contract calls for a fixed wage w if the agent
provides the required effort, the agent bears no uncontrollable risk; the income will
be w for sure. The revenues, of course, remain random, but this risk is borne entirely
by the risk-neutral principal. This allocation of risk is efficient. Putting any variability
in the agent's pay would necessitate compensating him or her for bearing risk, whereas
the risk-neutral principal is indifferent about the risk. The pay needed to get the agent

· Those who are not familiar with the theory of expected utility and decisions under uncertainty
might prefer to skip this Appendix until they have studied these topics, which are developed at the start of
Chapter 7.
 20 1
Table 6. 5 Probability of Outcomes For Moral Hazard and
Differing Levels of Effort Performance
Incenti ves

Revenue

Action R = 10 R = 30
e = 1 p = 2/3 p = 1/3
e = 2 p = 1/3 p = 2/3

to agree to the contract is determined by the utility function and the mm 1 mum
available elsewhere:
Vw - (e - 1) = Vw - ( 2 - 1) � 1, or w�4
So long as the pay is at least 4, the agent will not prefer to take a j ob elsewhere.
Because the principal has no reason in this model to give the agent any more than
necessary, the principal gets an expected return net (of the pay to the agent) of
(70/3) - 4 = ( 58/ 3).
I n contrast, if the principal wants only the low level of effort, this can be
achieved at minimum cost by paying 1 in either event. This puts no risk on the agent.
l t again gives the agent an expected utility of 1 , but (47/ 3) to the principal. Thus, the
wage cost of the extra effort is 4 - 1 = 3, both to the principal and to "society, "
whereas it raises expected receipts by (20/3) > 3 . Thus it is worthwhile to require and
pay for the higher level of effort.
I n any case, if e is observable, then regard less of the desired level of e, the
effi cient contract pr otects the worker from having to bear any uncontrollable risk
while, if the high level of effort is efficient, the contract requires the agent to work at
the higher level but compensates him or her for it.
Things are different when only the level of revenues, but not the level of effort
e, is observable. ln this case, the principal cannot effectively insist that the agent take
a particular level of effort. Effort is not observable, and revenues are not fu lly
determined by effort, although they are responsive to it. Revenue levels of 1 0 and 30
are both possible no matter what the agent d oes, so a bad outcome might be attributable
to bad luck rather than shirking, and a good outcome might occur by sheer good luck
regardless of what the agent d oes.
lf a high level of effort is desired but the agent is averse to working that hard ,
the way to motivate high effort is to pay more for a good outcome than for a bad
outcome. This requires exposing the agent to some income risk.
If the principal wants e = 2, then the agent's expected utility when he or she
picks e = 2 must exceed that when he or she slacks off and picks e = 1 . Let y be
the amount the agent receives under the incentive contract when the outcome is 1 0
and z be the pay when the receipts are 30. Then the agent' s expected utility i n picking
e = 2 is
( l/ 3)(Vy - 1 ) + ( 2/ 3 )(½ - 1 )
(where we have used the probabilities that correspond to the high level of effort in
evaluating the expected wage), whereas his or her expected utility from picking e =
1 is
(2/3)( Vy - 0) + ( 1 /3)(½ - 0)
In this expression we used the probabilities correspond ing to e = l . For the agent to
 202
I\ lotivation:
I ncentive Constraint
Contracts,
----- ½ .Jz - ½
5
Information, and 1 = ',/y
Incentives 4
Participat ion Constraint
� 3
1 (',/y - 1 )
3 + 2 (.../z- 1 ) = 1
3
2

Figure 6. 1 : In this example, the point (0, 3)

satisfies both the incentive and participation
2 3 4 5 6 constraints at least cost to the principal. Note
that the graph is expressed in terms of the square
',/y roots of y and z.

be willing to pick the higher effort level, the first of these expressions must be at least
as large as the second:
( 1 /3)(\ly - 1 ) + (2/3)(Yz - I ) � (2/3)(\ly - 0) + ( l/3)(Vz - 0)
This expression is called an incentive compatibility constraint in the formal theory of
incentives. It represents a constraint on the design of a compensation scheme when
the principals want to elicit a high level of effort. Note the parallel with the incentive
compatibility constraints that arise in problems with precontractual private information
(see Chapter 5). As there, the constraint relates the utility of the agent when he or
she acts in the desired fashion to that when he or she misbehaves and it thereby limits
the arrangements that will work.
With a bit of algebra, the incentive constraint is transformed into the following
simpler form:
( l /3)Vz - I � ( l/3)Vy (6. 1 )
According to Equation 6. 1, the pay for the good outcome must sufficiently exceed
that for the bad in order to compensate the agent for providing the extra effort that
makes the good outcome more likely.
Another kind of constraint on the design of compensation is called the
participation constraint. The terms of employment, taken together, must provide the
agent with at least as much utility as that available in outside opportunities. Otherwise,
the agent will not agree to participate by accepting employment. Recall that for this
example, we assume the expected utility of outside opportunities is 1 . Then, the
participation constraint is:
( 1 /3 )(\ly - I ) + ( 2/3)(Yz - 1 ) � 1 (6. 2 )
The principal's problem is to find the values of y and z that satisfy these constraints
and give the maximum expected returns net of pay to the agent. Furthermore, both
y and z must be positive: The agent cannot (in this example) be made to pay the
principal when things go badly.
The two constraints are graphed in Figure 6. 1. The values of y and z that meet
the incentive constraint (6. 1 ) are those above the upward-sloping line, \\·hereas those
that satisfy the participation constraint (6. 2) are above the downward-sloping line. The
shaded area represents the values of y and z-the pay levels for bad and good
outcomes-that both attract the agent to take the job and motivate him or her to
perform as desired. Because the principal's net retu rn is largest \\'hen the expected pay
to the agent is smallest, the best contract from the principal's point of \·ie\\' is y = 0,
z = 9. This meets both constraints and gives the principal an expected return of Moral Hazard and
( 1/3)( 1 0 - 0) + (2/3)(30 - 9) = ( 5 2/3). Performance
The agent is no better off than he or she would be if e were observable, whereas Incen tives
the principal's payoff has fallen from (58/3) to ( 5 2/3) because the expected wage paid
the agent has risen from 4 to 6. The additional expected wage merely serves to
compensate for the risk the agent faces. Thus, the unobservability of effort and the
consequent moral hazard has an efficiency cost. ln this case, the cost arises from
having to load too much risk on the agent.
We also need to make sure that it is actually worthwhile to provide incentives
to get the agent to select the high effort level. I f instead the principal decided to settle
for e = 1 , then there would be no need to use incentive payments. Paying the agent
a constant wage of 1 -.independent of the outcome-would provide the agent with
just enough expected utility to get him or her to agree to the contract. Because it
would· put no risk on the agent, this pay scheme minimizes the costs of employing
him or her. Of course, with the agent's pay unaffected by the outcome, the agent will
minimize his or her effort provision and pick e = 1 , as planned. This yields a payoff
to the agent of 1 again, and to the principal of (2/3)( 1 0 - 1 ) + (1/3)(30 - 1 ) =
(47/3). Because this is less than the payoff when e = 2 is induced, motivating the
higher effort level is worthwhile.
This, of course, is just one example. ln other examples, trying to motivate a
high level of effort is inefficient because the costs of loading enough risk on the agent
to provide the requisite incentives exceed the gains from the higher level of effort.
Moreover, this can happen even when a high level of effort would be optimal with
full observability and no moral hazard. In the present example, taking the receipts in
the good event to be 20 rather than 30 yields such a case. lt is also easy to concoct
other examples with several possible effort levels where the solution is to give up on
trying to induce the level of effort that would be optimal without the observability
difficulties, though it is still worthwhile to provide incentives for some effort beyond
the minimum conceivable. ln these cases, the inefficiency manifests itself both in the
agent's bearing risk that he or she would rather avoid and that the principal is better
equipped to face, and in the level of effort induced being less than is desirable.
Designing real incentive contracts involves much more complex issues than this
example can reveal. Some of these are developed in the next chapter, and more are
explored in Chapters 8, 1 2, and 1 3.
 Part

IV
EFFICIENT INCENTIVES :
CONTRACTS AND OWNERSHIP

7
RisK SHARING
AND INCENTIVE CONTRACTS

8
RENTS AND EFFICIENCY

9
OwNERSHIP AND PROPERTY RIGHTS
 7
RISK SHARING
AND INCENTIVE CONTRACTS

/ w l e/ , then, says I, what's the use you learning to do right when it's
troublesome to do right a nd ain't no trouble to do wrong, a nd the wages is just the
same?
Huckleberry Finn 1

In Chapter 6, we examined how insurance of various forms can combine with

difficulties of monitoring actions or verifying information to blunt individual incentives.
We also surveyed a number of responses to such moral hazard problems. Among
these were incen tive con tracts, under which individual incentives are strengthened by
holding people at least partially responsible for the results of their actions, even though
doing so exposes them to risks that could be more easily borne by an insurance
company. In this chapter, we develop a detailed theory of the nature and form of
efficient incentive contracts in the presence of moral hazard, establishing a number
of general principles that can be used to understand, evaluate, and design such
contracts. Although we develop this theory largely in terms of employment contracting
and performance pay, the principles are broadly applicable to a wide variety of
institutional contexts.

INCENTI VE CONTRACTS AS A
RESPONSE TO MORAL HAZARD
In both theory and practice, there are more options open to society than to insure a
risk fully or not to insure it at all. Actual insurance contracts are also incentive

106 1 The Adventures of Huckleberry Finn, Mark Twain ( 1 884).

 207
con tracts: They have provisions that restrict and con dition claim paymen ts in ways Risk Sharing and
that provide better incen tives than fu ll insurance without removing the essen tial part Incentive Contracts
of the insurance coverage. The deductible clause that is common in homeown ers'
fire and theft insurance policies requires the policyholders to bear the in itial part of
an y loss they may incur while still protec tin g them again st large financial losses.
H ealth-in surance policies often require copaymen ts, acc ording to which the in surance
pays on ly a frac tion of the c osts, with the rest bein g borne by the in sured. Automobile
in surance is experience rated, so that those who arc respon sible for traffic acciden ts
pay higher rates. These features are designed to encourage the insureds to take care
and to deter their excessive use of the insurance. For example, the c opayments on
emergency room visits are set so that, rather than automatically rushing to the
emergency room, a patien t will wait to be treated in the doctor's office for illn esses
an d inj uries that are n ot extremely urgen t. In this spirit, the policy may provide n o
coverage at all for treatmen ts that are con sidered to be elective, such as c osmetic
surgery other than that which is necessary to repair damage caused by an inj ury. As
these examples make clear, insurance c on tracts are designed with profound atten tion
for the need to reduce the waste caused by moral hazard.
Similar moral hazard issues must be faced when devising compen sation c on trac ts
for employees in a firm. H ere, too, there is a balance that n eeds to be struck between
providing incen tives and insulating peopl e from risk. To provide incen tives, it is
desirable to hold employees responsible for their performance; this mean s that
employees' c ompen sation or future promotions should depen d on how well they
perform their assigned tasks. As we will see, however, holding employees respon sible
typically will in volve subjecting them to risk in their curren t or fu ture inc omes.
Because most people dislike bearing such risks and are often less well equipped to do
so than are their employers, there is a cost in providing incen tives. Efficient contracts
balance the costs of risk bearing against the incentive gains that result.

Sources of Randomness
If employees were always able to perform as required an d if it were easy to determin e
precisely whether they have behaved as they were supposed to, having pay depen d on
performance would n ot generate an y risk-bearin g c osts. An employee could choose
whether to perform appropriately or n ot. Appropriate behavior would be c ompen sated
as agreed; inappropriate behavior would go uncompensated and might be penalized.
Higher levels of required performance would be associated with higher pay to
compensate for the additional effort that the employee is called upon to expend, but
there would be n o risk in the employee' s pay because the outcome is completel y
under the employee's c on trol.
In most real situations, however, attempts to impose respon sibility on empl oyees
for their performance do expose them to risk because perfec t measures of behavior are
hardl y ever available. For example, if the employee is expected to give expert advice
on some matter, it may be impossible to determine whether the advice is based on
the best available information an d analysis an d whether the recommendation s are
actually designed to promote the employer's in terests, or whether the employee has
acted selfishly or deceptively. When care an d effort are wanted, it may equally be
im possible to determine if empl oyees are doing what they should or slackin g off In
these kinds of situations, even though the quality of effort or the accuracy of
information cannot itself be observed, something about it can frequen tly be inferred
from observed results, and c ompen sation based on results can be an effective way to
provide incen tives. Piece rates are a prime example: Rather than trying to monitor
directly the effort that the employee provides, the employer simply pays for output.
H owever, results are frequen tly affec ted by things outside the empl oyee's c on trol
 208
Efficient that have nothing to do with how intelligently, honestly, and diligently the employee
Incentives: has worked. Sales at a fast-food restaurant may be lower than expected due to the outlet
Contracts and manager's lack of creativity in devising promotional efforts or negligence in supervising
Ownership the staff, but the low level may also be caused by other factors. Road construction could
have made the location less accessible to customers. The opening of a competing
restaurant nearby could be to blame. Population growth may have been less than
forecast. In the case of a franchise, the franchisor's failure to provide attractive menus
or timely deliveries of food could be responsible. Or some combination of these and
other factors might be at work. Similarly, if an aircraft crashes, pilot error may be to
blame, or poor maintenance, or a design flaw in the craft itself, or a bolt of lightning, or
an air traffic control error, and so on. When rewards are based on results, uncontrollable
randomness in outcomes induces randomness in the employees' incomes.
A second source of randomness arises when the performance itself (rather than
the result) is measured, but the performance evaluation measures include random or
subjective elements. For example, the way an employee is evaluated may depend on
his or her supervisor's subjective perception of the employee's attitude towards the job
and behavior towards other workers. Employees may see this sort of evaluation as a
source of risk because it is based partly on elements outside the employee's control.
A worker's performance may be evaluated by sporadic monitoring, and these random
observations may not give a perfect reflection of the actual quality of the work. In
either case, the imperfect evaluation of performance induces randomness in rewards.
A third source of randomness comes from the possibility that outside events
beyond the control of the employee may affect his or her ability to perform as
contracted. Health problems may reduce the employee's strength and ability to work,
concerns about family finances may make it impossible to concentrate effectively on
the tasks at hand, or weather or traffic conditions may render meeting a regular
schedule impossible. Thus, performance itself becomes random, and so too does
performance-based compensation. Consequently, making employees responsible for
performance subjects them to risk.

Balancin g Risks and Incentives

It might be possible to insulate employees from these risks by making their compensation
absolutely risk free and unrelated to performance or outcomes. In that case, however,
the employees would have little direct incentive to perform in more than the most
perfunctory fashion, because there are no rewards for good behavior or punishments
for bad. As we will see, both here and in Chapter 1 2 (where we examine compensation
issues more specifically) effective contracts balance the gains from providing incentives
against the costs of forcing employees to bear risk.
The same considerations arise in many other business transactions. The size of
the crop produced by a sharecropper is influenced by weather and pests as well as by
the sharecropper's own skill and effort. Traditionally, landowners make part of the
sharecropper's compensation proportional to the size of the crop. This arrangement
provides helpful incentives that induce the sharecropper to plant drought- and pest­
resistant varieties, to irrigate and care for the crops, and so on. However, it also exposes
the sharecropper to the risks of a poor harvest-a risk that is at least partially outside his
or her control. Similarly, in the United States, a lawyer who sues for damages on behalf
of a client often receives a contingency fee (a percentage of the damage award or
settlement). This system provides the litigator with an incentive to work hard on behalf
of the client, but because the outcome of the lawsuit is not entirely under the litigator's
control, both the litigator's income and the client's are uncertain.
Although all of these cases share certain common features, the accuracy of the
performance assessments that can be achieved and the need for and possibility of risk
 209
sharing or in surance vary fr om case to case. Becau se of these differences, the Risk Sharing and
i nsti tu ti ons and practices that best balance risk and i ncenti ves also vary. Incentive Contracts
The conclusi on that arrangements should vary fr om case to case is too vague to
be of any use to managers or i nterest to economi sts. Fortunately, we can do better.
The pri nciples developed in thi s chapter make i t possible to reach a relati vely su btle
understandi ng of how optimal practices can be desi gned that trade off the value of
protecti ng people from risk agai nst the need to provide them wi th i ncentives.
In order to analyze how rati onal people respond to incenti ves in i nsurance-like
contracts, we must first exami ne how rati onal people behave and interact i n ri sky
situati ons. Thi s i nvolves three steps. The first i s to descri be the ri sks precisely, using
the langu age of stati stics. Then, we descri be how rati onal people, acti ng i ndividually,
can choose consi stently among ri sky choices and how varyi ng i ndi vidual atti tudes
toward ri sk taki ng can be i ncorporated into the analysis. Fi nally, we exami ne how
groups of people can share risks and form i nsurance pools, being careful to quantify
the benefits of i nsurance coverage. Gi ven thi s backgrou nd, we then exami ne how
people respond to i ncentives i n ri sky si tuati ons. Thi s then allows us to develop the
pri nci ples of efficiently designed i ncentive contracts.

DECISIONS UNDER UNCERTA INTY

AND THE EVALUATION OF FINANCIAL RI SKS
The first element we need is a theory of deci si ons u nder u ncertainty. There are, i n
fact, a number of rich theories addressi ng thi s su bj ect i n great generali ty, but for our
purposes i t is enough to consider the special case i n whi ch the ri sks are financial . The
first step is to descri be the financial ri sk. We do this usi ng two ideas familiar from
stati stical theory: the concepts of mean and variance. These terms are defined i n the
appendix to the chapter. H ere, we i llu strate thei r meani ng by computi ng the mean
and vari ance i n an example.

Computing Means and Variances

Recall that the mean or expected value of a random i ncome i s si mply the expected
amount of i ncome, computed as the weighted average of the possi ble values that
income might take on, with the weight on each value bei ng the probabi li ty of that
value occurri ng. The relevant calcu lati ons are illu strated i n Table 7. 1.
The table shows a hypothetical si tuati on i n which there is an i nvestment for
whi ch the returns are zero wi th probability one half, $ 3 , 000 wi th probabili ty one thi rd
and $6, 000 wi th probability one si xth. The mean or expected value of the return is
½($0) + ½($ 3, 000) + ¼($6, 000) = $0 + $ 1 , 000 + $ 1 , 000 = $2,000. In the table,
the calculati on works by multi plyi ng the entri es i n columns I and 2 to obtai n column
3, and then su mmi ng the column. H avi ng higher probabili ties on higher values
i ncreases the mean.
The variance of i ncome is a measure of i ts variabi lity or randomness. It is
computed i n columns 4 and 5 of the table. In column 4, we take each possi ble value,
subtract the mean (to get a measu re of how far the particular value deviates from the
expected value), and square the result (so that terms greater than the average that
result from higher-than-expected incomes do not cancel out the negative terms that
result when i ncome i s less than expected). In column 5 , these squared variati ons are
multi plied by the corresponding probability. Summing the column gives the vari ance.
In the example, the variance is ½(0 - 2 , 000) 2 + ½(3, 000 - 2 , 000)2 + ¼(6, 000 -
2,000)2 = ½(4, 000, 000) + ½( 1 , 000, 000) + t( l 6 , 000, 000) = 5 , 000, 000. (The u ni ts
are "dollars squared. ") If income is certai n, then the vari ance is zero, because the
income never devi ates from i ts expected value. Increasi ng the probabi li ty of very high
and very low values tends to i ncrease the variance.
 2 10
Effi c ient Table 7. 1 Sample Computation of Mean and Variance
Incentives:
Contracts and
Ownership 1 2 3 4 5
Pr obability Return (1) X (2) (Return Mean) 2
(1) X ( 4)
1 /2 0 0 4,000, 000 2,000,000
1/3 3 ,000 1 , 000 1 ,000, 000 333,333
1/6 6,000 1 ,000 1 6, 000,000 2, 666,667
Mean = 2, 000 Variance = 5,000,000

Certainty E q uivalents and Risk Premia

One of the main hypotheses we employ in this chapter is that most peopl e are risk
averse; that is, they would prefer receiving a certain income of / to receiving a random
income with expected value I. The amount the person would be willing to pay to
make the switch is the risk premium associated with the random income. The
magnitude of the risk premium depends on both the riskiness of the income and the
individual person's degree of risk aversion. The amount that is left after the risk
premium is paid is the certainty equivalent of the random income. It is the amount
of income, payable for certain, that the person regards as equivalent in value to the
original, random income.
One of the central results of decision theory is that the certainty equivalent can

and variance of the random variable I, and r(_J) is a parameter of the decision maker's
be estimated by a simple formula: Y - ½r(J)Va r(J), where Y and Va r(}) are the mean

personal preferences call ed the coefficient of absolute risk aversion for gamb les with
mean Y. The mean in this formula is the mean income, and the amount subtracted
from it in the formula is the risk premium; it is equal to one-half times the coefficient
of absolute risk aversion times the variance of the income. According to the formula,
the risk premium is pr oportional to the coeffi cient of absolute risk aversion: People
who are more risk averse according to this measure are willing to pay proportionately
larger risk premiums to avoid a given risk. If the coefficient of ab sol ute risk aversion
is zero, then the person is unwilling to pay any premium to avoid the risk. Such a
person is called risk neutral. A person is risk averse when the coefficient of absolute
risk aversion is positive. The amount I - ½r(J)Var(_l) that is left in expectation after

certainty equivalent of the random income I.

the risk premium is deducted is called the person's certain equivalent income or the

only value that income actually might take on is I = Y. Then, the formula yields the
If there is no uncertainty regarding the level of income, then Var(_/) = O; the

sensible result that the person is as well off with the nonrandom income I as with a
certain amount that is equal to I: The thing is as good as itself. When I does vary (so
Var(_/) is positive) and the person is risk averse (so r(_J) is also positive), the risk premium
is positive. This means that he or she would be willing to accept a lmver amount th an
I to avoid the risk. M ore precisely, the risk premium, ½r(_J)Var(_J ), is the amount that
uncertainty in I. 2
the person would pay to have the certain income Y for sure rather than face the

2 The estimate of the certa inty equivalent given i n this formula is good when the va riance is not
too large or the coefficient of risk aversion is sma ll. In terms of the example in Table 7. 1 , where the mean
income was $2, 000 and the variance was 5, 000,000, the formula becomes 2,000 - 2, 5 00, 000,(/ ). This
approximation is reasonable only for val ues of r in the range of . 00008 or less (corresponding to a risk
prem ium of 200); when r ::5 . 0008, it yields the nonsensical answer that the individual would be indifferent
 21 1
Hisk Premia and Value Maximization Risk Sharing and
Our analysis in this chapter uses the value maximization principle, which in the Incentive Contracts
context of uncertainty asserts that an arrangement is efficient if and only if it maximizes
the total certain equivalent wealth of all the parties involved. Recall from Chapter 2
that the premises needed to derive the principle are ( 1 ) that each person has enough
wealth to make whatever payments might be called for under any relevant contract
and (2) that each person has a well-defined willingness to pay for any given product
or service and the amount of this monetary valuation does not depend on his or her
income level . As discussed in Chapter 2, these are strong and often unrealistic
assumptions, but they greatly simplify the analysis and enable us to separate analytically
the effects of the level and variability of income from all other effects on the matters
of interest. In the context of uncertain income, the second assumption is reduced to
this: The risk premium that a person would pay to eliminate a given amount of
variance must not depend on the expected level of income I. In view of the risk
premium formula, this means that r(I) must not depend on I. Throughout the rest
of this chapter, we make that assumption and write r instead of r(I). With this
aswmption, the crucial formulas become:
Expected Income = I
Risk Premium = ½rVar(J)
Certain Equivalent = I - ½rVar(l)
We use these formulas to calculate the benefits of insurance and the costs of the risk
bearing that is required to provide incentives.

RISK SHARING AND I NSURANC

One of the most fundamental facts about the economics of risk is that when several
people are facing statistically independent risks, then by sharing the risks among
themselves they can greatly reduce the cost of risk bearing. Two risks are statistically
independent if knowing the realized value of one risk gives you no information about
the value that the other will achieve. For example, the amount you won or lost per
dollar invested in the state lottery today does not give you any reason to change your
estimates of the likely returns in the stock market tomorrow. In contrast, for risks that
are not independent, knowledge of one is useful in predicting the other. For example,
the prices of gold on the London and New York markets are both random , but they
tend to move together under the influence of arbitrage (buying in one market and
selling in another to make a riskless profit). Thus, knowing the price in London tells
you something useful about what the New York price is likely to be, and so the two
risks are not independent. This principle of risk sharing-that sharing independent
risks reduces the aggregate cost of bearing them-is the basis of all financial insurance
contracts.

How Insurance Reduces the Cost of Bearin g Risk

In modern economies there are many kinds of institutions to assist people in sharing
risks. One important group consists of the insurance companies. Having many
policyholders, the insurance companies can spread risks very widely, enabling the
companies to reduce individual risks greatly. If the risks are independent and the
number of policyholders quite large, the risks are effectively eliminated and insurance
works very well. For example, the risk that you will suffer an automobile accident is

between the gamble and getting a negative income for sure. In using the approximation, we thus assume
that the variance of the uncertain income is not too large relative to the individual's risk aversion.
 2 12
Efficient very nearly independent of the risk that any other particular person will do so, therefore
Incentives: automobile insurance is a feasible enterprise. Insurance companies specialize in
Contracts and evaluating individual risks and, by pooling the risk-bearing capacity of policyholders
Ownership and (sometimes) shareholders, they reduce the cost of the risk bearing to negligible
proportions. Pooling independent risks also has the additional advantage of making
the insured losses statistically predictable. An insurance company can ask each
insurance policyholder to pay a price for insurance equal to the expected amount of
the loss, plus a margin for expenses and profit, and can be reasonably sure that the
aggregate premium i ncome together with a proportionately small reserve fund will
enable it to pay for whatever losses may be suffered, even in a bad year.
Some kinds of risks, however, are so large and pervasive in their impact that
they cannot be made negligible by sharing and they cannot be managed by traditional
insurance arrangements. (Technically, the risks that people bear in this case are not
statistically independent. ) For example, an oil price increase would have such
widespread effects, reducing the effective incomes of most people in oil-consuming
countries, that no amount of risk sharing among those oil consumers can insulate
them from the loss. Risks of this general kind are shared through other markets,
especially the financial markets. By purchasing stock in companies that own oil
reserves, for example, an investor who is especially vulnerable to oil price increases
can arrange to have an offsetting profit if oil prices increase. Financial markets allocate
many other kinds of risks, as well. For our purposes, an important example is the
investment risks that are taken by firms, such as those associated with a new technology.
The risk of failure of the technology is borne by shareholders in the company that
develops it, and this capacity for risk sharing reduces the firm's cost of financing the
investment, helping to promote technical change.

Efficient Risk Sharing : A Mathematical Example

Suppose that there are two people, A and B, each of whom has some risk associated
with his or her income, where these risks are independent. Let IA and IB represent
their random incomes, with means TA and TB and variances Var(_I A) and Var(_I B), and
let rA and rB denote their coefficients of absolute risk aversion. In view of our earlier
assumption, the value maximization principle applies. Consequently, every efficient
risk-sharing contract maximizes the total certain equivalent income of all the parties,
and every such contract is an efficient one.
If the parties make no special arrangements, then the total cost they suffer on
account of risk bearing, that is, the total risk premium, is ½rA Var(_I A) + ½rB Var(_IB),
which is the sum of the two individual risk premia. Suppose that the parties instead
agree on a risk-sharing contract with party A receiving a fraction a of the income IA
and B of the income IB (and thus of the risks associated with the two uncertain
incomes. ) In addition, suppose A receives a cash transfer of 'Y for the risk-sharing
services provided. (This transfer might be positive or negative, but it is independent
of the actual, realized incomes. ) Party B receives the remaining share of each risk and
makes the cash payment -y. After this agreement, A's income will be aIA + BI B + -y
and B's will be ( 1 - a)I A + ( 1 - B)I B - -y. This is a feasible agreement because
the total income each party receives always adds up to IA + I B, the amount available.
With this agreement, the total risk premium of the two parties is:
Total Risk Premium = ½ rA Var(_aIA + BI B + -y) + ½rB Va r(_( I - a)IA + ( 1 - 13)IB - -y)
(7. 1 )
Because th e total certain equivalent income of the two parties is equal to the mean
income, TA + TB, minus the risk premium, the efficient arrangements arc those that
minimize Equation 7. 1.
 Usin g identi ties about variances (see Formula 7.18 in the appendix), Equation Risk Sharing and
7.1 is a quadratic function of a and B. The total risk premium is min imized when Incentive Contracts
a/(1 - a) = B/(1 - B) = ralrA . For exam ple, suppose rA = 2 and ra = 4. The
higher value for B's coefficient of absolute risk aversion in dicates that B finds bearin g
risk more onerous than docs A. Indeed, the risk premium that B attac hes to an y given
risk is twice the amount A would pay to avoid the risk. In these circ umstances, we
might expec t that A would bear more of the risk than would B . Evaluating the
solution, we see that a/(1 - a) = B/(1 - B) = 2, so a = B = i and (1 - a) =
(1 - B) = !: A does in fact bear most of both risks. M oreover, A bears the same
share (two thirds) of both.
To formulate the general principle that applies here, it is helpful to think in
terms of different peoples' capacity to bear risk. We measure this by introducing the
notion of risk tolerance. Someone with a c oefficient of absolute risk aversion of r wil l
be said to have risk tolerance of 1/r. N otice that in the preceding example, A's share
of each risk is equal to A's share of the total risk tolerance (i = ½/(½ + ¼)).
These calculations actually reflect a general principle that can be shown to hold
for any number of people and any number of financial risks: When risks are shared
efficiently, the share that a party bears in each risk is the same and is equal to his or
her share of the total risk tolerance of the group. M oreover, when risks are allocated
efficiently, the total risk premium comes out to be:
Total Risk Premium = ½ Var(IA + Ia)/[(1/rA ) + (lira)] ( 7 . 2)
Equation 7.2 resembles the formula for the risk premium charged by a single decision
maker. It says that when risks are shared efficiently among a group of people, the total
risk premium is the same as if the total risk were borne by a single decision maker
whose risk tolerance is the sum of the members' individual risk tolerances. I n the
preceding numerical example, (1/rA ) + (lira) = ½ + ¼ = ¾. This formula, too, is
general; it can be shown to hold for any number of people and any number of financial
risks. With efficient risk sharing, the group is less risk averse than the people comprising
it and so the c osts of bearing risks can be reduced.
When individual risks are independent, these facts imply that sharing risks can
be a very effective way to reduce the c ost of risk bearing. For example, if there are n
people, each with an inc ome with variance v and each with the same coefficient of
risk aversion r, and if each bears the inc ome ri sk separately, then the risk premium
will be ½rv per person. If the people share the income risks efficiently, then each will
have a 1/n share of the total risk. The variance of the total risk is V = nv, so the
variance of an individual 1/n share is V/n2 = v/n (see Formula 7.18 in the appendix
again). Therefore, by sharing risks, each person's risk premium is reduced from ½rv
to ½rvln. When n is large, even substantial financial losses can be reduced to ec onor.iic
insignificance by sharing them efficiently across the group.

Optimal Risk Sharing Ignoring I ncentives

For both insurance companies, with their wide base of policyholders, and public ly
traded c orporations, with their wide base of shareholders, it is reasonable to suppose
as a first approximation that the total risk tolerance of the company is infinitely larger
than the risk tolerance of any individual policyholder or employee. As we mentioned
earlier, an institution or person with infinite risk tolerance is said to be risk neutral:
The coefficient of absolute risk aversion is zero and so the risk premium for bearing
any risk is also zero. Applying our general propositions to the case where risks are to
be shared between a risk-neutral insurance c ompany and a risk-averse insurance
policyholder or between a large, risk- neutral employer and a risk-averse employee, we
find that the optimal share of the risk to be borne by the insurance buyer or employee
 214
Efficient is zero. Efficient risk sharing requires shifting all the risk onto the risk-neutral party,
Incentives: who suffers no cost in bearing the risk.
Contracts and This conclusion, however, depends on ignoring the incentive problems for
Ownership insurance and employment created by the condition of moral hazard.

PRINCIPLES OF INCENTIYE PAY

The general problem of motivating one person or organization to act on behalf of
another is known among economists as the principal-agent problem. This problem
encompasses not only the design of incentive pay but also issues in job design and
the design of institutions to gather information, protect investments, allocate decision
and ownership rights, and so on. However, we focus our discussion in this chapter
principally on the issues surrounding incentive pay, and we set our discussion of
incentives in the context of employment. The principal in this case is the employer,
who wants the employee (the agent) to act on his or her behalf.

Basing Pay on Measured Performance

As we discussed in the introduction to this chapter, there are many situations in which
providing incentives requires that employees' pay depend on their performance.
Essentially, if the employees' direct provision of effort, intelligence, honesty, and
imagination cannot be easily measured, then pay cannot be based on these and any
financial incentives must come from basing compensation on performance. Efficient
risk sharing, in contrast, requires that each person in society should bear only a tiny
share of each risk, without regard to its source. In particular, individuals should be
insulated against the randomness that would enter their pay by basing it on measured
performance. Therefore, performance-based compensation systems cause a loss from
inefficient risk sharing. The money value of the loss is equal to the risk premium
associated with the actual compensation system minus the risk premium that
would be associated with efficient risk sharing. Firms that use performance-based
compensation hope to recoup this loss (and more) by eliciting better performance
from their employees.
There are various reasons why incentives might be needed to elicit top-notch
performance. Some employees may fi nd their work distasteful and may neglect it
unless they are held responsible for achieving results. Even when employees are hard
workers who like their jobs, they may still have priorities that are different from those
of their employer. For example, without compensating incentives, managers might
be tempted to be too generous to their subordinates in granting raises and time off,
or to hire the children of relatives and friends, to spend lavishly on a pleasant work
environment or on fancy accommodations when traveling on business, to use company
resources for community projects that raise their personal status, to devote excessive
efforts to projects that advance their careers or that are especially interesting or pleasant,
and so on .
To analyze these possibilities in a model, we suppose that the employee must
exert an effort e at personal cost C(e) to serve the interests of the employer. The effort
e represents any activity that the employee undertakes on behalf of the firm, and the
cost C(e) can represent the unpleasantness of the task, foregone perquisites, lost status
in the community, or anything else that the employee gives up to serve the employer's
interests. For tasks that are pleasant, the "cost" can be zero or even negative.
The effort e is assumed to determine to the firm's profits: Profit = P(e). It is
sensible to assume that greater effort leads to higher profits. It is not necessary for the
employer actually to know the functional relationship between effort and results;
instead, the P function can be thought of as the employer's subiective estimate of the
 2 1 f>
productivity relationship. If the relationship between profi ts and effort is random, then Risk Sharing and
P(e) should be thought of as the expected value of profits when effort level e is Incentive Contracts
expended.
It may be impossible for anyone to observe an employee's direct effect on profi ts,
but it is that effect, in principle, that the employer cares about. For example, the
employee may be a sales representative whose efforts lead to no sales today but create
a good impression that brings customers back in the future. The employer may care
about the impression that is created, without actually being able to tell either how
hard and how skillfully the employee has tried to impress customers or how many
customers have actually been favorably impressed.
The general point here is that compensation can vary systematically only with
things that the employer can observe. The employer cannot pay more to sales
representatives who are particularly effective in creating a good impression if it is
impossible to .tell who they are. In addition, even some observable indicators may not
be suitable bases for compensation. It may be possible, in principle, for the manager
to photograph the faces of customers as they leave the store and pay compensation
based on how many faces were smiling. What makes this possibility seem so absurd
is its manifestly subjective nature. What is a "smiling" face? To base a compensation
formula on something that is not objectively measurable is to invite disputes and
unhappiness among employees.

A Model of I ncentive Com p ensation

For our first formal model of incentive compensation, we assume that the effort level
e that the employee chooses can be understood to be a number-for example, energy
expended or hours worked. As we have already noted, if e were directly observed,
there would be no diffi culty in providing adequate incentives; the employer could
make pay contingent on satisfactory performance without exposing the employee to
any risk. We therefore suppose that the effort e cannot be directly observed. We shall
suppose, however, that the employer can observe some imperfect indicators of e, that
is, indicators that provide some information about e but are contaminated by random
events beyond the control of the agent. For example, measured output might provide
such a signal: It is related to effort, but many influences beyond the employee's control
also affect the realized output. In addition, the employer may be able to observe other
indicators of factors, such as general economic conditions, that are not controlled by
the employee but that do affect performance.
Suppose that the indicator of effort can be written in the form z = e + x,
where x is a random variable, and that a second indicator is y, where y is not affected
by the effort e but may be statistically related to x, the noise between e and the
observed z. Note that e and x are not separately observed; only their sum, z, is
observed, and many different combinations of e and x yield the same level of observed
z. Thus, high effort might be offset by bad luck, or low effort might be masked by
good fortune.
For example, if the employee is the sales manager for some product, z might
be a measure of total sales for the product (which depends on sales effort, e, and
random events, x, such as realized demands) and y mi ght measure total industry
demand, which is correlated with the potential demand in the markets where the
employee manages sales and thus with realized sales. To keep our formulas as simple
as possible, we suppose that x and y are each adjusted to have mean zero. Then, the
expected level of sales is just the effort level. In term s of the example, instead of
making y the industry demand, we could make it the amount by which industry
demand differs from a forecast value.
The class of compensation rules that we study are those that are linear in the
 216
Efficient two observations, that is, ones that can be written in the following form, where w
Incentives: stands for wage:
w = o: + B(e + x + -yy)
Contracts and
Ownership ( 7. 3 )
Compensation thus consists of a base amount, o:, plus a portion that varies with the
observed elements, z and y. We use B to measure the intensity of the incentives
provided to the employee, so that one contract will be said to provide "stronger
incentives" than another if the first contract specifies a higher value for B. The
justification for this language is that if the employee increases his or her effort choice
e by one unit, then according to Equation 7. 3, expected compensation increases by
B dollars, so higher levels of B bring greater returns to increased effort.

variable y (as compared to z = e + x) in determining compensation. If -y is set at

The parameter -y indicates how much relative weight is given to the information

zero, then y is not used in determining compensation. Given any value for -y, the
term z + -yy gives an estimate of the unobservable e. One of the principle issues in
contract design is to determine how much, if any, weight to give to y in this estimate,
that is, to determine the level of -y.
As an example of such a contract, suppose o: is $ 1 0, 000, B is $20 and -y is 0. 5.

zero. If the employee sets e equal to 1 00, the expected pay becomes $ 1 2, 000
Then expected pay is $ 1 0, 000 + $20e, because the expected values of x and y are

( = $ 1 0, 000 + $2, 000); if e is set at 200, the expected pay is $ 1 4, 000. Unless there
and so pay will deviate randomly from its expected level. If x is more favorable than
is no real uncertainty, however, x and y will often not take on their expected values,

expected, say taking on the value 1 00, whereas y is less favorable, taking on the value
- 400, then the observed values are z = e + 1 00 and y = - 400. Now an effort
level of e = 1 00 brings pay of $ 1 0, 000 + $20( 1 00 + 1 00 + 0. 5( - 400)) = $ 1 0, 000,
and an effort level of 200 brings pay of $ 1 2, 000. Of course, if x and y take on different
values than those just specified, the compensation again will differ. For example, with
e = 1 00, x = - 1 00 and y = 1 00, pay is $ 1 1 , 000, whereas effort of 200 with these
same levels for the random factors brings an income of $ 1 3, 000. Thus, pay varies not
just with the employee's effort, but also with the random events represented by x and
y, and this randomness imposes risk on the employee (unless B is zero).

THE LOGIC OF LINEAR COMPENSATION FORMULAS The restriction to linear compensa­

tion formulas such as the one in Equation 7. 3 is not always sensible. The ideal form
of the compensation rule in any circumstance depends on the nature of the efforts
required and on the available performance measures. Linear compensation formulas
are quite popular, however, and so we take a brief diversion from our main analysis
to consider when such schemes might work especially well. The considerations that
arise in this discussion should serve as a reminder that incentive compensation issues
are very complicated ones and not all of the relevant issues are represented in our
simple mathematical models.
Linear compensation formulas are commonly observed in the form of commis­
sions paid to sales agents, contingency fees paid to attorneys, piece rates paid to tree
planters or knitters, crop shares paid to sharecropping farmers, and so on. Linear
formulas are not the only ones used, however. For example, sales agents are sometimes
paid a bonus for meeting a sales target. As compared to a system of sales commissions,
a reward for meeting a sales target has the disadvantage that the sales representative
loses any special incentive to make additional sales after the target is reached or after
a poor start leaves the target hopelessly out of reach. Commission systems apply a
uniform "incentive pressure" that makes the agent want to make additional sales
regardless of how things have gone in the past. In selling, because incremental sales
 217
are typically eq ually pro fitable for the firm after ei ther a slow or a fast start, th is Risk Shari ng and
uniform incen tive pressure is appropriate ( in fact, optimal). Incentive Contracts
Partly as a result of efforts by firms to avoid the p roblem j ust described, when
sales targets are used they are often set to co ver short period s of time, so th at the
periods dur ing which incenti ves are too low are not extended ones. This makes the
compen sation of additional sales effo rts more n early equal o ver time. The sales
represen tatives themselves can be expected to respond to time-varyin g incen tives by
advancin g or delaying the closing of sales un til the period when the co mpen sation
rate is h ighest. To the exten t that the sales representatives succeed, they h ave effec tively
arranged for all sales to be compen sated equally, that is, they have converted what is
nominally a sales target system into something closely resemblin g a system of
commissions proportional to sales.
Beyond this, of cour se, linear systems have th e ad van tage of being simple to
understand and administer. A scheme that employees cannot understand or that
cannot be ad�in istered as intended cannot provide the desired motivation.
TOTAL WEALTH UNDER A LINEAR CONTRACT An employee's ability to bear risk is
negligible compared to the employer's whenever th e employer is a large or medium
size enterprise. For this reason, it would be optimal-incen tive issues aside-for the
employer to bear all financial risks, leaving the employees fully in sured again st all
sources of fluctuation in their incomes. However, removin g all compen sation risk
also removes all the employee's direct financial incentives to increase p rofits by
providing effort. What is wan ted is an employmen t con tract that b alances the n eed
for risk sharing again st the n eed to provide incen tives.
Actual employment con tracts in volve a large number of terms, but we wish to
focus on only those few dealing directly with incen tive pay. Therefore, we will
characterize a contract by a li st of parameters (e, a, B , -y) that specify what level of
effort e the employer expects to elicit and how the employee is to be compen sated on
the basis of performance. The employee's certain equivalent wealth from such a
contract is the expected compen sation paid minus the personal co st to the employee
of supplying effort min us an y risk premium: a + B(e + x+ y) - C(e) - ½rVar
[a + B(e + x + -yy)], where x and y are the mean levels of x and y and r is the
employee' s coefficient of absolute risk aversion. Recall that, to simplify formulas, we
had assumed that both x and y are zero. Using the formulas about variances in the
appendix, we find that the employee's certain equivalen t income consists of ex pected
income minus the cost of effort and minus a risk premium for the income risk the
employee bears:
Employee's Certain E quivalen t = a + Be - C(e) - ½rB 2 Var( x + -yy). (7.4)
The employer's certain equivalen t con sists of the expected gross profits min us the
expected compen sation paid :
Employer's Certain Equivalen t = P(e) - (a + Be) ( 7 .4a)
Implicit in this is a hypothesis that the employer is approximately risk neutral.
Notice that the employee's certain equivalen t consists of a plus a function of
the other variables (8, -y, e) and the employer' s consists of - a plus another function
of those variables. That is, each p arty' s equivalen t wealth consists of a mon ey term
plus a term that depends on all the oth er aspects of the decision. By tran sferring
mon ey from one party to the other, one can rai se one party's certain equivalen t and
reduce the other's by an equal amount. This is p recisely the no wealth effec ts cond ition
that we described in Chapter 2; we can therefore apply the value maximization
principle. It follows that an y efficien t contract must specify the p arameters so that
 218
Efficient they maximize the sum of the certain equivalent incomes of the two parties. That
Incentives: sum is
Contracts and
Ownership
Total Certain Equivalent = P(e) - C(e) - ½rB 2Var(x + -yy) (7. 4b)
Equation 7. 4b specifies what is to be maximized.
INCENTl\'ES FOR EFFORT AND CONTRACT FEASIBILITY The next step is to specify which
choices of contracts are feasible. After all, it would be ideal to ask the employee to
work hard without having to provide any incentives or make the employee bear any
risk! We require, however, that the employer be realistic: The level of effort the
employer expects must be compatible with the incentives that are provided to the
employee. Although the anticipated effort level of the employee is part of the contract,
the actual effort level cannot be directly observed and is chosen later by the employee,
with his or her own interests foremost in mind. To be realistic, we (and the employer)
must therefore determine how the employee's choice of effort e will depend on the
other parameters (O'., B, 'Y) of the contract.
Equation 7. 4 provides the key to the answer. Suppose that the costs of providing
effort vary smoothly with the level provided and that the cost of effort increases at an
increasing rate or, in other words, the marginal cost of effort to the employee is rising.
Then, the level of effort that maximizes the employee's certain equivalent income in
Equation 7. 4 is the level that makes the derivative of that expression equal to zero,
that is,
B - C'(e) = 0 (7. 5 )
Equation 7. 5 is called an incen tive constraint and must be satisfied by any feasible
employment contract. It says that employees will select their effort levels in such a
way that in their marginal gains from more effort equal their marginal personal costs.
The gain is the increased pay, and a unit increase in effort brings an expected increase
in pay of B; the marginal cost is C', the rate at which the personal cost of effort
increases as the level provided increases.
An employment contract is therefore efficient if and only if the choices (e, O'.,
B, 'Y) are ones that maximize the total certain equivalent in Equation 7. 4b among all
"incentive-compatible" contracts, that is, among all contracts that are consistent with
Equation 7. 5 and thus realizable or feasible. It is useful to solve problems of this kind
in two steps. In the first step, we fix the effort e at some level and ask how the
parameters O'., B, and 'Y are optimally chosen then. This is called the implementation
problem of obtaining the specified level of effort in the most efficient fashion.
It is evident from Equation 7. 5 that fixing e also amounts to fixing B at C'(e) if
we are actually going to get the employees to provide the specified effort level. In
Figure 7. 1 , to raise the effort level that the employee will choose to provide from e
to e necessitates increasing the intensity of incen_tives from B to B. The difference in
the intensity of incentives needed can be computed as the difference in the desired
effort levels times the slope of the marginal cost-of-effort curve, C".
Also, from Equation 7. 4b, we see that O'. does not affect the total certain
equivalent at all (it determines only how the total is divided between the two parties).
Thus, putting aside any requirement that both parties be willing to agree to the
contract (which would limit the possible values of O'. to ensure that each's expected
welfare was sufficiently high), we see that the efficiency of the contract does not
depend on the choice of O'.. As for -y, it is clear that the total certain equivalent is
maximized when 'Y is chosen to make Va r(x + -yy), the variance of the estimate of
e, as small as possible because this minimizes the risk premium-the costs of imposing
risks on the employees to generate incentives.
 219
Marg i n a l Cost with Risk Shari ng and
S lope C " Incentive Contracts
�

-�
Q)

c
Q)
(.)
C:

Figure 7. 1 : Increasing effort provided from

e e e to e requires increasing B to B, where
Effort B - B = (e - e)C".

The Informativeness Princi p le

This last result-that 'Y should be chosen to minimize the variance of x + 'YY, the
estimate of e-is a special case of a more general principle.
The Informativeness Principle. I n designing compensation formulas, total
value is always increased by factoring into the determinant of pay any
performance measure that (with the appropriate weighting) allows reducing
the error with which the agent's choices are estimated and by excluding
performance measures that increase the error with which effort is estimated
(for example, because they are solely reflective of random factors outside
the agent's control).
As applied to our particular model, a measure with low error variance serves as a
better basis of performance pay than a measure with higher variance. Thus, y should
be included in the determinants of pay if and only if there is some value for 'Y that
makes Var(x + 'YY) smaller than Var(x), the estimate that results when y is ignored
and 'Y is set at zero. The optimal value for 'Y is determined by minimizing Var(x +
'YY),
Using appendix Equation 7 . 1 8, we see that Var(x + 'YY) equals Var(x) +
'Y Var(y) + 2"/Cov(x, y), where Cov(x, y), the covariance of x and y, is a statistical
2

measure of how x and y are related and vary together. Minimizing this expression
with respect to 'Y yields the result that 'Y should optimally be set at - Cov(x, y)/Var(y).
If x and y are independent, then Cov(x, y) is zero. In this case, 'Y is optimally
set equal to zero. This reflects the fact that with x and y independent, knowing y te!ls
us nothing about x and so gives us no better estimate of e: There is no point i n simply
adding noise to the performance measure. If x and y are positively related, as they
might be if x reflects the conditions in a specific market and y is a measure of general
market conditions, then Cov(x, y) is positive. Then 'Y should be negative. Good
general market conditions (positive levels of y) likely mean that conditions were also
good in the specific market (positive x). Therefore, a greater portion of any given level
of the observed performance z = x + e is likely to reflect good luck (high x) rather
than good effort (high e). Similarly, if y is low, x was also likely to be low, and a
given z signals a higher level of effort e. A negative value for 'Y takes account of these
likelihoods by increasing pay when general conditions are bad and decreasing it when
they are good. Meanwhile, if x and y tend to move in opposite directions from one
another, so that a low y is likely to correspond to a high x and vice versa, then Cov(x,
y) is negative and 'Y is optimally positive. A high y then signals that the given, observed
 220
Efficient level of z was likely obtained despite a low level of x, and therefore a high y is evidence
Incentives: suggesting a high level of e, which is rewarded through a positive value for -y.
Contracts and Also note that as the variance of y increases, the magnitude of -y optimally
Ownership decreases. Larger values of Var(y) mean more "noise"-less reliable information­
and the optimal choice of -y takes account of that by giving less weight to the signal.
Even if y is an extremely unreliable measure, it will still optimally be used, but it
will be given very little weight, affecting pay significantly only when it takes on an
extremely large or small value.

APPLICATION: COMPARATl\'E PERFORMANCE E\"ALUATION In applying the informa­

tiveness principle, consider the practice of comparative performance evaluation,
according to which the compensation of an employee (typically a manager or executive)
depends not just on his or her own performance but on the amount by which it
exceeds or falls short of someone else's performance. Debates about this practice often
revolve around the issue of controllability: As a matter of principle, it is argued, an
employee's oompensation should not depend on things outside the employee's control
because that is perceived as unfair and because it appears to make the employee bear
an unnecessary risk. So when is comparative performance evaluation a good idea?
When would it be better to base the compensation of the employee only on his or
her own performance?
To phrase this issue in the terms of our theory, suppose the measured performance
of the employee depends on the employee's efforts, on random events that affect that
employee only, and perhaps on other factors that affect all similarly situated employees.
For example, the employee's measured performance might depend on the difficulty
of the task, which is similar to that of the tasks assigned to other workers. Or, if the
employee is a manager, the profitability of his or her unit might depend on what
happens to oil prices, or interest rates, or the general level of demand in the industry.
Each of these factors could be expected to have a similar effect on the profits earned
by other similarly situated units.
To formalize all this, suppose there are two managers, A and B. Suppose the
performance measure for manager A can be written in the form z = eA + x, where
eA is the effort of manager A and x is the sum of two independent components: x =
xA + Xe - In this expression, xA is a random component that affects A's performance
only and Xe is a random component that affects both A's and B's performances. (The
subscript C stands for this "common" source of randomness. ) Similarly, B's performance
measure takes the form y = e8 + x8 + Xe, where xA, x8, and Xe are independent
sources of randomness. ls it better to compensate manager A based on the absolute
performance measure z = eA + xA + Xe or on the rela tive performance measure
z - y, which is equal to eA - e8 + xA - x8 ?
The informativeness principle directs us to the error variances attached to each
compensation scheme. The variance of the first (absolute) performance measure is
Var(xA ) + Var(xc), whereas the variance of the second (relative) is Var(x A) + Var(x8 )
(again, see the formulas in the appendix). The relative performance measure therefore
has lower variance and is to be preferred if and only if Var(x8 ) < Var(xc). In other
words, if the randomness that affects performance is predominantly due to a common
effect, such as oil price increases or the unknown difficulty of the task, and if the
variation in performance due to random events that affects particular people is smaller
than the variance of the common element, then comparative performance evaluation
is better than individual performance evaluation because it enables the employer to
elim inate the main source of randomness in evaluating performance. If the reverse
relation holds ( Va r(xc) < Var(x 8 )), however, that is, if com mon sources of randomness
that affect both employees have smaller effects than does the randomness that affects
 22 1
individual employees, then it is better to base compensation on an absolute stan dard Risk Sharing and
of performance. Incentive Contracts
Of course, in general, neither purely absolute nor purely relative performance
evaluation is most efficient. As the informativeness principle establishes, some mix of
absolute and comparative performance evaluation is generally preferred to either
extreme form. In fact the relative weights to be placed on eA + xA + Xe and on y
can be computed from the principle.
APPLICATION : DEDUCTl8LES AND COPAYMENTS IN INSURANCE In automobile insurance,
collision coverage is insurance that pays the owner of an automobile when his or her
own auto is damaged in a collision. Comprehensive damage coverage is insurance that
pays for damage to the person's automobile when it is stolen or damaged by other
means, such as by a falling tree in a storm. B oth of these kinds of coverage usually
work by specifying a deductible, which is the portion of the loss that the insured
person must pay before any payment is due from the insurance company.
Supp ose that the owner of the car can, by driving carefully, parking in a garage,
keeping the car doors locked, and so on, reduce the probability that the car will be
stolen or damaged. That is the kind of effort that the insurance company would want
to elicit. In the case of a collision or a theft, however, the owner has no control over
the size of the loss that would be suffered. In that case, the size of the loss provides
no information about the care taken by the owner. Therefore, according to the
informativeness principle, the owner's contribution toward any loss should not depend
on the size of the l oss but only on the most informative performance indicator, which
is the fact that a loss has occurred. S o, in an optimal insurance contract, the owner's
contribution should not depend on the size of the loss but rather should be a fixed
amount per accident, which is very nearly the terms of a standard auto insurance
contract. (We say "very nearly" because if the loss is smaller than the deductible, then
the amount the insured owner pays does depend on the size of the loss. )
It is helpful to contrast the practice in automobile insurance with the practice
in health insurance and health-care plans, where it is common to require copayments
from the consumer for any services used. A consumer's choices about when to visit
the doctor, whether to seek urgent care or to wait for a regular appointment, and so
on, are all choices that affect the total level of cost incurred. The total level of cost
incurred therefore provides information about how effectively the agent-in this case
the consumer-ha s conserved scarce health-provision resources . As the theory predicts,
the payments made by a health-insurance consumer therefore varies directly with the
cost incurred by the health care provider.

The Incentive-Intensity Princi ple

The next step in the general analysis of incentive contracts is to determine how intense
the incentives should be. In this step, we fix the information weighting parameter -y
at whatever level the contract specifies (whether optimal or not) and let V
= Var(x + -yy).
The Incentive Intensity Principle. The optimal intensity of incentives
depends on four factors: the incremental profits created by additional
effort, the precision with which the desired activities are assessed, the
agent' s risk tolerance, and the agent's responsiveness to incentives. The
formula for the optimal intensity is: B = P'(e)/[l + rVC"(e)] .
According to the incentive intensity principle, there are four factors that interact
to determine the appropriate intensity of incentives. The first is the profitability of
incremental effort. There is no point incurring the costs of eliciting extra effort unless
 222
Efficient the resul ts are profi table. For example, it is counterproductive to use incentives to
Incentives: enc ourage production workers to work fa ster when they are already producing so much
Contracts and that the next stage on the pr oduction line cannot use their output. Acc ording to the
Ownership incentive intensity princi ple, the optimal intensity is proportional to the profi tability
of incremental effort, provided the other three factor s remain unchan ged.
The second factor is the risk aversion of the agent. The less risk averse the agent,
the l ower the cost he or she incurs from bearing the risks that attend intense incentives.
According to the incentive intensity principle, more risk averse agen ts ought to be
pr ovided with less intense incentives.
The third factor is the precision with which performance is measured. Low
precision corresponds to high values of the variance V, which according to the formula
means that only weak incentives should be used. I t is futile to use wage incentives
when performance measurement is highly imprecise, but str ong incentives are likely
to be optimal when good performance is easy to identify.
The fi nal factor is the responsiveness of effort to incentives, which is inversely
proportional to C"(e) (see Figure 7.1). For example, an employee working on a fi xed
rate production line cannot increase his or her own output in respon se to piece rate
incentives. According to the incentive intensity principle, incentives should be most
intense when agents are able most able to respond to them. Generall y, this happens
when they have discretion about more aspects of their work, including the pace of
work, the tools and methods they use, and so on. An employee with wide discretion
facing str ong wage incentives may fi nd innovative ways to increase his or her
performance, resulting in significant increases in profi ts.

lllathematical Derivation of the Optimal

Incentive Intensi(,�
Figure 7. 2 illustrates the trade- offs that determine the optimal intensity. The
intensity, B, is measured on the horiz ontal axis and its marginal benefi ts and
costs on the vertical axis. The downward-sloping l ine rec ords the net marginal
benefi t of i ncreasin g the intensity of incentives. The net marginal benefi t of
extra effort is P'(e) - C' (e). To determine the net marginal benefi t of extra
incentives, the marginal benefi t of effort must be multiplied by the rate at which
extra effort is supplied for each extra unit of intensity. That rate, as we have
previously seen, is 1/C"(e). S ince the agent will choose e so that B = C' (e), the
net marginal benefi t is (P'(e) - C'(e))/C"(e) = (P' (e) - B)/C"(e), as shown in
the Figure. The transaction cost associated with setting effort in tensity B is the
risk premium ½rVB 2 , with associated marginal cost rVB, as plotted in the Figure.
The optimal intensity of incentives occurs at the point where the marginal
benefi t and marginal cost are equal.
To fi nd the optimal intensity by direct maximization, write the total certain
equivalent for any fi xed value of e and B as P(e) - C(e) - ½rB 2V, by Equation
7. 4b. From the incentive constraint of Equation 7. 5 , we kn ow that B = C'(e),
so the obj ective can be rewritten as:
Total Certain Equivalent = P(e) - C(e) - ½ rC '(e)2 V (7. 6 )
Equation 7. 6 gives a clear picture of the benefits enj oyed and costs incurred for
an y given level of effort. The benefi t term in this equation is j ust the profit P(e),
but the cost has two componen ts: the direct cost C(e) incurred by the agent plus
the tran saction cost ½ rC '(e) 2V of provid in g the requisite ince ntives.
The optimal level of effort e under the c ontract is foun d by differen tiating th e
 Marg i nal Net Benefit
/ = (P ' - (3)/C "
P '/C " Risk Shari ng and
Incentive Contracts

Marg i nal Transaction Cost

...
'iii
= rV/3
...CC
Q)

Figure 7. 2: The optimal intensity of

i ncentives balances the direct net marginal
(3 (3 , benefits of increasing 8 against the marginal
I ncentive I ntensity transaction cost.

tota l certain equivalent with respect to e a nd setting that deriva tive equal to
zero: 0 = P'(e) - C '(e) - rVC'(e)C"(e). U sing Equation 7. 5 a gain, we ca n
replace C'(e) by B in this expression to obtain: 0 = P'(e) - B - rVBC"(e).
S olving thi s fo r B results in the fo rmula given in the incentive intensity principle.
APPLICATION : INCENTIVES FOR JAPANESE SUBCONTRACTORS Two recent studies ha ve
been performed that compare the recommendations of the incentive i ntensity principle
with the actual contractua l practices used to compensate subcontractors who supply
parts or components for large Japa nese automobile a nd electronics firms. 3 In Japa nese
practice, the amount paid by a ma nufa cturing firm for its inputs depends on the

being a contractually fixed price. If the target level of cost is a nd the actual cost
actual costs as mea sured in the supplier company's accounting records, rather tha n
x
i ncurred is x, then the supplier is paid x + B(x - x). Tha t is, the ma nufacturing
fi rm pays the actual cost incurred plus a fraction of the difference between the target

This adj ustment is a n incentive term. If the supplier's actual cost is less tha n the
cost (which is negotiated to include a n allowa nce for profit) a nd the rea lized cost.

target, it gets to keep some of the savings. If its costs exceed the ta rget level, then the
ma nufacturing compa ny a bsorbs some of the difference. Thus, if the actual cost x is
less tha n the target , the subcontractor earns an extra profit of B(x - x); if it is more,
then B(x - x) is nega tive, which mea ns that the subcontractor pays a penalty fo r its
poor performa nce.
To a nalyze this ca se, notice that a n effort that reduces costs by 1 yen also adds
1 yen to the ma nufacturing firm's profit, so we may take P'(e) = 1. Consequently,
the theory recommends tha t B = 1/(1 + rVC"). The researchers rearra nged the terms
in this equation to obtain 1/B - 1 = rVC". Taking logarithms of both sides of the
new equation leads to an equa tion that the researchers could test using linear regression
analysi s:
log( 1/B - 1) = log(r) + log(V) + log(C") ( 7.7 )
The ideal would now be to use data on B, r, V a nd C" from different contracts
to estimate the empirica l rela tionship among these varia bles. Then one could test
statistica lly whether the empirica l relationship wa s the one predicted by the theory.

' S . Kawasaki and J. McMillan, "The Design of Contracts: Evidence from Japanese Subcontracting, "
fournal of lapanese and International Economies, l ( l 987), 1 327-49; and B. Asanuma and T. Kikutani,
"Risk Absorption in Japanese Subcontracting: A Microeconometric Study on the Automobile I ndustry,"
forthcoming in the fournal of lapanese and International Economies ( 1 99 1 ).
 224
Efficient However, the available data did not provide direct information on all these variables. 4
Incentives: In such a situation, the next best thing is to identify instruments for the theoretical
Contracts and variables of interest, which are log(r), log(V) and log(C"). An instrument for a variable
Ownership is another variable that ( 1 ) can be observed, (2) varies directly with the actual variable
of interest, and (3) is uncorrelated with the other variables of interest.
To test Equation 7. 7, the researchers first estimated 1 - B by dividing the
variation in the supplier's profits over time by the variation in their costs. These
estimates were then used to tabulate log(1/B - 1 ) for the various firms in the sample.
The risk aversion r was assumed to be inversely proportional to various measures of
the size of the firm, such as the number of the firm's employees. Size variables
therefore were used as instruments for log(r) in the equation. The variance V in costs
was estimated by determining the trend in costs and then computing the variation in
actual costs around the trend over time. In theory, C" should be inversely proportional
to the scope for performance improvement by the agent. The researchers supposed
that the scope was proportional to the firm's value added in the production process
(in the Kawasaki and McMillan analysis) or to the firm's responsibility under the
contract for supplying technology and designing parts and production processes (in
the Asanuma and Kikutani analysis). These value-added and responsibility measures
were used as instruments for C" in the actual estimation. With only these instruments
for the actual variables of interest, all that could be hoped for is that the signs of the
coefficients in the estimated equations would be the same as predicted by the theory:
The intensity of incentives B should be greater for firms with more employees, more
value added, and less variability in year-to-year performance. The empirical findings
were consistent with these predictions.
The tests we have described represent only weak evidence in support of the
theory. The equation whose coefficients were finally estimated was not the exact one
predicted by the theory, and the instruments used are not beyond criticism. Moreover,
the estimation procedure did not test whether there were other variables affecting
actual choices of B that were not predicted by the theory and, if so, how important
those other variables were for understanding incentives. Nevertheless, the evidence
obtained is consistent with the theory: Incentive contracts for Japanese suppliers do
appear to depend on the considerations identified by the theory in the general way
that the theory predicts .
T.\:X SHELTER PROGR.-\\IS Another study has
. \PPLIC\TIO!'J: I NCENTl\"ES II\ OIL .\ND G.\S
tested the incentive-intensity principle in the context of the organization of oil and
gas tax shelters in the United States in the early 1 980s. 5 At that time, many drilling
operations were financed by limited partnerships. As you recall from Chapter 6, under
the federal tax laws that then prevailed, the partners could often save on taxes if the
limited partners paid all the costs of exploring for oil (which were tax deductible when
the costs were incurred), whereas the general partner(s) paid the costs of completing
wells in which oil was found (which were "capitalized costs" for tax reporting purposes).
The general partner and the limited partners would then share any revenues enjoyed
when oil was pumped from producing wells.
A problem with this tax-reduction scheme is that it created a difference in in-

4 Kawasaki and McMillan used data reported in MITI's Census of Manufacturers (The Firm Series)
and Surve)'s of Industries. Asanuma and Kikutani limited their attention to Japanese automobile
manufacturers, from whom they could obtain somewhat more detailed information.
5 Mark Wolfson, "Empirical Evidence of Incentive Problems and Their Mitigation in Oil and Gas
Tax Shelter Programs, " Principals and Agen ts: The Structure of Business, J. Pratt and R. Zeckhauser, eds.
(Bmton: Harvard Business School Press, 1985), 101-27.
 225
terests between the general partner, who controlled the partnership's activities, and the Risk Sharing and
limited partners because each bore a different kind of expense. If a well were found Incentive Contracts
to have oil, the general partner had to bear I 00 percent of the cost of compl eting the
wel l, but typically received only 2 5 percent of the oil revenues. Suppose that after
the expl oration costs have been sunk, a well were found to have only enough oil that
the general partner woul d need to have a 50 percent share of revenues to recover the
well-completion costs. Then, it woul d not be in his or her interest to compl ete the
wel l, even though the full revenues would more than cover the compl etion costs.
S everal of the prospectuses used by the general partners to attract investors
described the problem quite candidl y. According to one:
A situation may arise in which the completion of an initial well (the
maj ority of the costs of which are capital ized costs) on a prospect would
be more advantageous to the l imited partners than to the general partners.
The situation woul d arise where a compl etion attempt on an initial wel l ,
the maj ority of the costs of which are paid by the general partners, could
apparently resul t in a marginal well which woul d return some but not all
of the compl etion cost incurred by the general partners but would return
revenue to the limited partners. 6
The conflict of interest described here is likely to be most severe when many of the
well s being drilled are "marginal" prospects. If the well that is found is a gusher, then
even the 25 percent of revenues accruing to the general partner woul d make completion
of the well highly profi table. The general partner seen as the agent of the limited
partners, therefore, is most likel y to be responsive to compl eti on incentives-to have
his or her behavior positivel y affected by expl icit incentives-when many of the wells
to be completed are marginal ones. N o expl icit incentives for completing wells are
necessary when they are very productive, and giving such incentives would not have
much effect on the general partner's behavior. Economic theory predicts that the
contracts that are actually used shoul d be responsive to this difference in completion
incentives.
To test this theory, the researcher divided drilling programs into three types:
expl oratory programs, devel opmental programs, and balanced programs. E xploratory
drilling programs were ones in which wells were drilled in new areas, where the
greatest l ikelihood was that no oil would be found but any wel ls that were found were
unlikel y to be marginal. In these programs, the conflict between the general and
limited partners' interests in compl eting wells was likel y to be small , and the general
partners' completi on decision was likel y to be littl e affected by any special contractual
i ncentives. S ome 96 percent of the money invested in these expl oratory dril l ing
programs in the sample was in contracts that were designed to minimize taxes, with
no special all owances to improve the general partners' compl etion incentives.
Developmental drilling programs were ones in which all drilli ng occurred in an area
that had been previously expl ored and where oil was known to be present, but where
no more maj or finds were expected. M any developmental wells tum out to be marginal
wells, so we shoul d expect that the general partner woul d have been quite responsive
to incentives to complete these wells. The researcher found that onl y 2 3 percent of
the money invested in these programs was in contracts that provided no compl etion
in centives. For balanced drilling programs, which contained a mix of expl oratory and
devel opmental wells, the corresponding figure was 37 percent.
This evidence provides a useful test of one aspect of the incentive-intensi ty

6
Prospectus of the Hilliard Fund ( 1982), 22, as quoted by Wolfson.
 226
Efficient principle. The impact of any given monetary incentive on the agent's behavior ,·aries
Incentives: with circumstances, and the principle predicts that incentives will be more intense
Contracts and and more often incorporated into contracts when the agent's responsiveness to them
Ownership is high. The evidence in this case generally confirms the prediction of the principal­
agent model: Incentives are provided when they are likely to make a difference.

The Monitorin g I ntensity Principle

So far, we have assumed that the measurement of performance is outside the scope
of the model; that is, the variance V with which efforts are measured has been treated
as outside the employer's control (other than through the determination of -y ) . Often,
however, it is possible for an employer to improve measurement by devoting resources
to that objective. For example, in a factory, the number of workers per supervisor
could be reduced to allow closer monitoring, or more quality-control tests could be
made. For service workers, customers could be interviewed to learn whether thev
were satisfied with the service. In a telephone ordering or service operation, call­
counting and timing equipment could be installed or supervisors could listen in on
incoming calls to see how well they are handled. All of these things are costly, but
all improve the employer's information about how employees are performing.
To investigate how much should be spent on monitoring, suppose that the
variance of the performance measure can be controlled at a cost. Let M(V) be the
minimum amount that must be spent on monitoring needed to achieve an error
variance as low as V. It is generally costly to reduce the error variance, so we suppose
that M is a decreasing function-settling for a larger V entails lower monitoring costs.
We also suppose that the marginal cost of variance reduction is a rising function, that
is, M'(V) is increasing. Rewriting Equation 7. 4b to include the cost of the resources
that are spent on measurement, we have:
Total Certain Equivalent = P(e) - C(e) - ½rVB 2 - M(V) (7. 8)
The relationship between e and B is still determined by the incentive constraint
Equation 7. 5, which is unaffected by the introduction of costly measurement. We
may therefore hold e and B fixed and choose V to maximize the expression in Equation
7. 4b. Taking the derivative of Equation 7. 8 with respect to V leads to:
- ½r8 2 - M'(V) = 0 ( 7.9)
According to this equation, the marginal cost of reducing V, which is - M'(V)-a
positive number-must be equal to ½r8 2 at the efficient solution.
The Monitoring Intensity Principle: Comparing two situations, one with
B set high and another with B set lower, we find that V is set lower and
more resources are spent on measurement when B is higher: When the
plan is to make the agent's pay very sensitive to performance, it will pay
to measure that performance carefully.
The determination of V is illustrated in Figure 7. 3. The downward sloping
curve gives the marginal cost of reducing the variance with which performance is
measured. Because the risk premium is ½ rB 2 V, the marginal cost of variance changes
is depicted in the figure by a solid line at level ½rB2 . When the incentive intensity is
reduced from B to B, the chosen level of V increases: Fewer resources are spent on
measurement.
There may appear to be some circularity in our several observations. In the
incentive-intensity principle, we claim that B should tend to be set large when V is
low. In the last paragraph, we claim that firms should try to reduce V when B is large.
 227
Risk Sharing and
Incentive Contracts

-C
::::,
� l r[3 2 1--------------
2
I
I
- - - - - t- - -
1
I Figure 7. 3 : The optimal level of measurement
equates the marginal cost and marginal benefit
V V of variance reduction. Less intense incentives
Variance lead to higher V (less measurement).

Which causes which? Do intense incentives lead firms to carefu l measurement, or

does careful measurement provide the j ustification for intense incentives?
The answer is that, in an optimally designed incentive system, the amount of
measurement and the intensity of incentives are ch osen together: Neither causes the
other. H owever, setting intense incentives and measuring performance carefully are
complementary activities in th e sense described in Chapter 4; undertaking eith er
activity tends to make the oth er more profitable.
Figure 7. 4 illustrates the situation. The two solid lines in the figure depict the
two relationships between measurement and incentive intensity j ust described. One
of these lines specifies the optimal intensity of incentives B for any particular
measurement variance; the other specifies the optimal variance for any particular
intensity of incentives. N otice that both lines slope downward. According to the
incentive-intensity principle, B falls when the variance V rises. S imilarly, according
to the monitorin g intensity principle, V falls as B ri ses; it pays to measure more
carefully (lower V) when incentives are i ntense. The point where th e two lines cross
determines the optimal combination; it is the point where V is ch osen optimally for
th e given intensity of incentives and B is selected optimally for the given measurement
error.

circumstances, i n which P' was higher or C" lower. According to the incentive­
The dotted line in Figure 7. 4 sh ows h ow B would depend on V in different

intensity principle, these changes would lead to h igh er levels of B for any fixed level
of V. That change is represented in Figure 7. 4 by the dotted line lying to the righ t

Optimal I ntensity
�
\
\
\
� V \
C
cu \
·;a V '

Figure 7. 4: An increase in P' or a fall in C"

{3 {3 '
leads to more intense incentives and more mea­
I ncent ive I ntensity surement (less variance).
 228
Efficient of the original line determining B as a function of V. Notice that the point of
Incentives: intersection of the new optimal incentive-intensity line with the optimal variance line
Contracts and is lower and to the right of the original point of intersection. The change leads to
Ownership sharper incentives and lower variance (more monitoring).

The E q ual Com pensation Princi p le

Now we enrich our conception of behavior in the firm to recognize that most
employees do more than one thing as part of their jobs. When there are several
activities being conducted, the employer will be concerned that employees allocate
their time and efforts correctly among the various things that need to be done. This
complicates the problem of providing incentives.
For example, suppose that the marketing representatives for a company making
specialty steel alloys perform several kinds of activities: They solicit business from new
customers, provide problem-solving services and advice to customers about how to
use the company's alloys, gather information about competitors' marketing activities,
and report about possible new products that might sell well. Of these several activities,
the easiest one to monitor is direct sales efforts, because it leads immediately to
measurable sales. Some of the other activities also lead to sales, with some time lag:

of time. If there is high turnover in the sales jobs, then the information about how
Keeping customers happy is likely to increase the representative's sales over a period

well customers are being served may not be available in a timely enough fashion to
use for compensating the responsible representative. Finally, some of the activities,

simple sales performance. If the firm were to compensate the marketing representatives
such as monitoring competitors' moves, are much more difficult to evaluate than is

based primarily on the accurately measured current sales figure, that might induce a
distortion in their behavior, causing them to switch efforts toward the immediate high­

customers happy and the firm well informed. If that sort of behavior led eventually
payoff activity of generating sales and away from the activities necessary to keep

to a loss of customers and declining sales, the representative could seek another job,
proudly displaying the sales performance he or she achieved in the first job. A related
problem might arise for the salespeople in a department store, who might be tempted
to maximize immediate commissions by pressuring a customer to buy a more expensive
product than necessary, leading to dissatisfied customers and lower future sales for the
store, not only of that one department's products but also of products sold in other
departments.
Alternatively, suppose that a fast-food chain wants its outlets to be profitable but
also wants them to contribute to the chain's reputation for cleanliness, fast service,
and hot, fresh food, because that reputation enhances sales at other outlets. These
profit and reputation goals can be in conflict. For example, a fast-food chain outlet
along a highway where many of the customers visit only once would suffer little loss

chain 's other stores might lose business on that account. If the chain compensates the
of profits if its hamburgers were sometimes cold and its bathrooms dirty, but the

store manager on the basis of sales alone, the manager would be unlikely to take full
account of the effects of his or her actions.
These observations lie behind our fourth principle of incentive contracting and

The Equal Compensation Principle: If an employee's allocation of time

com pensation.

or attention between two different activities cannot be monitored by the

employer, then either the marginal rate of return to the employee from
time or attention spent in each of the two activities must be equal, or the
activity with the lower marginal rate of return receives no time or attention.
 229
The equal compensation principle imposes a serious constraint on the incentive­ Risk Sharing and
compensation formulas that can be effective in practice. In particular, if an employee Incentive Contracts
is expected to devote time and effort to some activity for which performance cannot
be measured at all (V = oo), then incentive pay cannot be effectively used for any
other activities that the individual controls. The use of straight salary compensation
for managers can often be justified on these grounds.
MATHEMATICS OF THE EQUAL COMPENSATION PRINCIPLE Suppose the employee does
two different things, signified by levels of effort e 1 and e2• We will think of these levels
of effort as time devoted to two activities, and we assume that the cost incu rred by
the employee is an opportunity cost: It is time that becomes unavailable for other,
more pleasant or rewarding activities. It then makes sense to write the cost as depending
only on the total effort, not on its division between the two tasks: C(e 1 + e2 ). The
employer measures performance by observing the indicators e 1 + x1 and e2 + x2 ,
x
where x 1 and x2 have expected val ues of 1 and 2 . x
Suppose that the employer pays the employee according to a linear compensation
formula based on the two indicators: The wage paid is then w = a + B 1 (e 1 + x 1 ) +
Bz(e2 + x2 ). How should a, B i , B2 , e 1 , and e2 be chosen?
To take incentives into account in the problem, we first examine the employee's
objective given this compensation rule. A self-interested employee will choose e 1 and
e2 to maximize his or her certa in equ ivalent income:
x
Employee's Certain Equ ivalent = a + B 1 (e 1 + 1 ) + Bz(e2 + 2) - x
C(e 1 + e2 ) - ½ rVar(B 1 x 1 + B 2x2 ) (7. 10)

e i , e2 2:: 0. If e 1 is strictly positive, then at the maximizing choice for the employee,
For this problem, we suppose that the effort is restricted to a nonnegative number:

the derivative of Equation 7. 1 0 with respect to e 1 must be zero, so 8 1 = C'(e 1 + e2) .

Similarly, if e2 is strictly positive, then 82 = C'(e 1 + e2) . The analysis of the employee's
incentives alone th us establishes that 8 1 must equal 82 if each tasks is to receive some
attention.
APPLICATION : COST CENTERS AND PROFIT CENTERS As the models make clear, an
important part of the problem of design ing incentives is to determine what the
employee will be responsible for, that is, what measures will be used to evaluate
performance as a basis for compensation . As an example, consider the problem of
providing incentives to the manager of a manufactu ring facil ity. One approach might
declare that the manager is responsible only for the costs incurred in the factory, on
the theory that the manager has little control over revenues. In that case, we say that
the factory is a cost center, and the accounting systems should be set up to assess
accurately the costs attributable to the factory. Another approach declares that product
quality and speed of delivery are important to sales, so that it is a mistake to encourage
the manager to focus on cost control at the expense of these factors. Thus, sales
performance might be given some weight in determining the manager's compensation .

to which the manager might contribute are cost reduction and revenue generation . If
To represent these issues in terms of our theory, suppose that the two activities

sales revenues are subject to random variations that are outside the manager's control
and statistically independent of the randomness that affects costs, then the cost of
provid ing incentives of strength B to the manager for revenue generation is the risk
premium: ½ rB 2 Var(Revenues). The equal compensation principle implies that if the
factory manager is to be provided with sales-generation incentives at all , then it is

as those for manufactu ring cost control . If the B associated with cost control is to be
futile to do that in a half-hearted way: The incentives need to be of the same strength

large, then the B associated with revenue generation must be large as well, and
 230
Efficient therefore quite costly (in the sense of its leading to a large risk premium). Then, the
Incentives: factory is a profit center, to which revenues and costs are both attributed in determining
Contracts and performance.
Ownership Cost centers and profit centers are not the only alternatives, however, nor is
either likely to be the best alternative in the situation we have described. The firm
should actively seek ways to make production managers responsible for what they each
control without making them responsible for the performance of the sales force, which
they do not control. For example, if quality control and delays in the factory are the
chief concerns, then the firm could devise new measures of manufacturing perfor­
mance, such as the average time from order to delivery and the number of products
returned as unsatisfactory. According to the informativeness principle, these measures
are superior to measures based on dollar sales because they provide a more informative
assessment of the manufacturing manager's actual contribution to the sales effort. As
we observed earlier, the firm gains most by improving the measurement of variables
that figui:e most heavily as a basis for compensation.
The equal compensation principle suggests another possibility as well: The
manager could be paid a salary with no explicit incentive component. This would be
a plausible course of action when manufacturing quality control is important but hard
to assess accurately. Of course, the manager will still understand that promotions and
pay increases will depend on how superiors assess his or her performance, but at least
this solution avoids the distortions in allocation of time and effort and the randomness
in compensation brought about by an incentive compensation plan based on arbitrary
measures of performance.
This analysis of cost and profit centers focuses only on the issue of compensation.
Before leaving this example, however, it is helpful to recall that the actual organization
design problem is more involved than that. Managers who are given responsibility for
profits, for example, are commonly given broader decision authority than those
responsible just for costs or sales. Determining a manager's compensation amounts to
deciding what he or she is responsible for, and that decision should be made together
with decisions about the scope of the manager's authority.

APPLICATION : INCENTl\'ES FOR TEACHERS The equal compensation principle can be

applied to the recent public policy debate about whether it would be helpful to
provide cash incentives for teachers to improve elementary and secondary education.
Proponents of cash incentives argue that they would be helpful in focusing teachers
on their tasks and motivating them to be innovative in the search for effective ways
to train their students.
Opponents of the incentives for teachers, however, have a cogent response. The
measures that have been used in the past to evaluate teaching performance for
elementary school age children are tests of basic skills, and teaching these is just one
part of a teacher's job. Children are also expected to learn social skills, oral expression,
and creative thinking, and to build confidence that prepares them for the harder
challenges to be faced in later years. Teachers who are compensated based on tests of
basic skills alone would be tempted to neglect these other aspects of the job. They
might also be led to teaching the most docile students, whose performance scores are
easiest to improve, while neglecting students who have more trouble learning. In one
instance in South Carolina in 1989, a teacher was caught teaching the answers to the
actual test, a copy of which had been illicitly obtained. Compensating teachers based
011 test scores motivates teachers to help students test well, rather than to help students
learn.
According to the equal compensation principle, if it is desirable to have teachers
devote some efforts to each of several activities and if it is impossible to distinguish
efforts on the various different activities, then all these kinds of efforts must be
compensated equally. If social development, oral expression, or creative th inking
Risk Sharing and
Incentive Contracts
cannot be accurately measured, then the only realistic options are to remove the
responsibility for teaching them from the teacher or to pay the teacher a fixed wage,
with no clement of incentives pay.
It is a good idea to remember that responsibil ities and compensation should
really be determined together. In the case of teachers, for example, one proposal is
to install a system of specialist teachers who are compensated based on student test
scores but who are not responsible for other aspects of student performance. 7 This
would not, by itself, solve all the potential problems we have described, but it would
allow performance incentives and still ensure that attention is paid to developing the
very important "higher thinking skills" in young students. The general point to
remember is that by determining the job design and the compensation together, one
can sometimes solve problems that cannot be solved by compensation pol icy alone.
APPLICATION : ASSET O,�·NERSHIP The equal compensation principle also makes it
possible to give a careful treatment of some important issues in the theory of
employment and asset ownership. We represent ownership by supposing that at the
end of a period of production, the owner of the asset may take it and employ it in
other uses. For example, if the employer is the owner of a machine (the asset), he or
she can assign the job of production and the use of the asset to another worker,
whereas if the worker owns the asset then he or she can employ it on his or her own
behalf or on behalf of another employer. What kind of incentives are optimal and
who should own the asset?
Assets are notoriously hard to evaluate accurately and objectively. That is why
accountants generally report adj usted historical cost figures for asset valuations rather
than attempting to account for asset values on the basis of the asset's physical condition
(unless deterioration is obvious), its fair market value, or its productivity. The value
of a business automobile, for example, is accounted for by its purchase price less an
allowance for depreciation, even though its actual value depends on its mileage,
physical condition, and so on . Production machines are accounted for in a similar
way, even if hard use or changes in production methods has made their actual value
lower.
We represent the idea that assets are hard to value accurately in our model by
the following assumption: Although the actual value of the asset, A(e 1 ) + X i , is an
increasing function of the effort e 1 that the worker devotes to mainta ining and
improving the asset (and of random factors x 1 ), accounting measures of asset values
do not reflect those efforts and so cannot be used to provide incentives . Only the
direct output of the production process, which is e2 + x 2 , is observed by the parties
and can serve as a basis for compensation . Therefore, we may write the compensation
paid to the worker in the form a + B(e2 + x2 ).
Suppose that there is some level of total effort e that the employee is willing to
provide even in the absence of any cash incentives, although this level might be lower
than the employer would like to see provided. The efforts e 1 and e2 devoted to each
of the two activities cannot be observed, however. Should the firm induce greater
effort by setting B positive and thereby inducing more production effort ez ?
If the firm owns the asset, then the worker's certain equivalent compensation is
ex + Bez - ½B 2 rVar(x2 ) - C(e 1 + e2). If B is positive, the worker's optimal choice of
e 1 is always zero. This is just an application of the equal compensation principle:

7 Jane Hannaway, "Higher Ord<;r Skills, Job Design, and Incentives: An Analysis and Proposal, "
working paper, Stanford University , 199 1 .
 232
Efficient Because the marginal return to the agent from efforts devoted to maintaining or
Incentives: increasing the asset's value is always zero, the worker will devote no efforts to that
Contracts and act;vity (e 1 = 0) unless the returns to other activities are also zero. When the worker
Ownership is an employee and maintenance of the asset is important, we find (in this model)
that it is optimal to pay a fixed wage with no incentives for output performance
(B = 0). Then the worker will set e 1 + e2 = e and presumably will be willing to
allocate this total amount of effort as the firm directs.
The other possibility is that the worker may own the asset. In that case, the
worker's certain equivalent compensation is the sum of the asset's expected value
(which depends on e 1 ) and his or her expected compensation, less a risk premium
that reflects both the uncertainty in the asset's value and that in the worker's pay, as
well as the cost of effort: A(e 1 ) + ex + Be2 - ½rVar(x 1 + Bx2 ) - C(e 1 + e2 ) . As the
owner, the worker has a built-in incentive to care for the asset; he or she keeps any
value that is created when the asset is well cared for. In order to motivate the worker
also to pay some attention to production, it is necessary to set B > 0. Then, with
positive returns to both types of effort, the worker will choose to provide more total
e
effort than -the amount he or she would provide as an employee with no pay
incentives for working harder.
To summarize, if it is important that time and effort be devoted to both
producing and maintaining the asset, then incentive pay should always be used for
workers who bring their own tools ("independent contractors"), but it should never be
used for those who use the firm's tools ("employees"). In practice, incentives are used
more extensively for independent contractors than for individual employees, as our
analysis suggests they should be. The analysis also suggests that independent contractors
will work harder than employees, devoting more effort both to caring for the asset and
to being directly productive. They will also earn a higher average income to compensate
for the extra work they do and the greater risk they bear.
Finally, we come to the question: Who should own the asset? A detailed study
of asset ownership is contained in Chapter 9, so we are brief here. In the model just
described, if the worker owns the asset, then the worker bears risk both from the
randomness of asset returns and from the errors in performance measurement, which
add ½r(B2 Va r(x2 ) + Va r(x 1 )) to the total risk premium. Against this must be weighed
the fact that the ownership of the asset and increased incentives for the production
activity will elicit a higher level of effort. A cost-benefit calculation that balances these
considerations must be done to determine which arrangement is likely to be more
successful. Certain general principles are evident, however. Increases in the worker's
risk aversion or in the variance of asset returns or in the variance of performance
estimates in the production task all add to the risk premium that is incurred when the
employee owns the asset, making the ownership solution less valuable. If there are
many ways to improve performance, then the employee's efforts are especially likely
to be responsive to incentives (represented in our model by the assumption that C" is
small). Increases in the worker's scope for action tend to favor having the worker own
the asset. As we see later, there are a number of other considerations involved in
assigning asset ownership efficiently that are not represented in this simple conceptual

l ntertemporal I ncentives : The Hatchet Effect

model.

An especially thorny problem in real incentive systems is how to set the standard by
which performance is to be evaluated. In terms of our model, this means that the
mean of x is unknown, so that the mean measured performance corresponding to any
fixed level of effort in any performance measurement period is uncertain. If the
estimated mean value of x is x,
then the expected level of performance is e + and x
the corresponding expected pay is E = a + B(e + x). If the intended expected Risk Sharing and
compensation is E, then th e value of a that will be set is determ ined by rearranging Incentive Contracts
that equation to obtain: a = E - B(e + x). Increasing the estim ated value of x
therefore leads to changes in the fixed component of the incentive form ula. The
magnitude of the effect is proportio nal to the incentive intensity B. S etting the standard
too h igh will lead to consistently lo w levels of pay, perhaps leading to low morale and
quits. S etting it too low will lead to happier em ployees but also to co nsistently high er
levels of pay and lower net pro fits than wo uld otherwise be possible.
There are just three reasonably obj ective ways to set a performance standard.
The first, which is used frequently only for routine clerical or production tasks, is to
have engineers perform time-and-motion studies to determi ne, in detail, how a certain
operatio n is most efficiently done and how long it sho uld take. For example, an
engineer m ight use a sto pwatch to determine how long it takes a m icrofilm operator
working at norm al speed to load film into the machine, process documents thro ugh
it, and rewind, catalog, and store the film in the appro priate area. Conducting such
studies is costly, and the studies them selves can become obsolete quickly if the job is
one where workers can learn and adopt new tech niques as they accum ulate experi­
ence. The seco nd way is to use the performance of other peo ple in sim ilar jobs, that
is, to use comparative performance evaluations, which we analyzed earlier in th e
chapter. The third way is to use the past performance of th e same person in the

and rewards bad. If workers foresee this possi bility, very negative consequences can
same job. However, basing standards on past performance penalizes good performance

emerge.
The tendency for perform ance standards to increase after a period of good
performance is called the ratchet effect. The term was originally coined by students
of the Soviet economic system, who observed that m anagers of Soviet enterprises were
commonly "punished" for good performance by h avi ng h igher standards set in th e
next year's plan or, even worse, in the next quarter's plan (see Chapter 1). There are
widely known i nstances of Soviet factory m anagers who responded to newly installed
incentives with m assive gains in productivity, only to be denounced on the grounds
that their im proved performance was proof that they h ad previously been lazy or
corrupt. The ratchet phenomenon is even m uch older than th is exam ple: In a
traditio nal interpretation of the Jewish Passover story, the Egyptian Pharaoh held a
brick- producing contest among the Hebrew slaves and used the results as a standard
to set a m uch higher daily production quo ta.

to good performance is no t m erely unfair, it can be unpro ducti ve: If workers foresee
THE IMPACT OF THE RATCHET EFFECT The "ratcheting up" of standards in respo nse

the way future standards will depend on current performance, they may refuse to
cooperate wi th efforts to im prove productivity. So viet m anagers, who are well aware
of the ratchet effect, h ave often been reluctant to institute ch anges that could radically
reduce co sts, despite promised incentive payments. S im ilarly, in the U nited S tates,
the traditional animosity of labor unions toward piece rates may be explai ned by th e
concern that em ployers, once they have discovered the rates at which h ighly motivated
em ployees can work, will set a higher standard, leading to lower average pay or to the
same average pay for harder work.
There are situations in which there can be efficiency reasons for basing current
standards on past perform ance, but these arise when there are different em ployees
doing the work in different periods. According to the informativeness principle, it is
desirable to use all available informatio n that might reduce the variance in th e
measur em ent of seco nd-period performance. First-period performance will often give
useful information of this sort. The issues here are really identical to those that arise
 Efficient in the case of comparative performance evaluation, though in this case the comparison
Incentives: is across time periods in the same job rather than across workers at the same time in
Contracts and different jobs.
Ownership Using information from past performance lowers the variance with which
second-period effort is measured. According to the incentive-intensity principle, the
parties wil l therefore tend to set the incentive term, B, higher in the second period
than they otherwise would, taking advantage of the reduced variance. The theory thus
predicts that when the parties write contracts for one period at a time and when past
performance embodies useful information for evaluating future performance, incentives
will become more intense over time, as the parties utilize past experience to incorporate
more accurate performance expectations in their contracts. Because higher levels of
B induce higher levels of effort, the actual effort levels elicited from the worker will
also rise over time.
The argument using the informativeness principle is only correct, however, if
there is a new-occupant in the job in each period. If the same worker is to do the job
in each period, then the parties would be better off if they could commit initially to
a contract that el icits the same level of effort in each period. This point is subtle, and
establishing it will necessitate use of the formal model. The essential source of the
inefficiency, however, is that if the worker anticipates that standards in future
accounting periods will depend on past performance, then it becomes more difficult
or costly to provide him or her with incentives to perform well in the early accounting
periods. The example of the Soviet factory managers whose incentives were destroyed
by the fear of increased performance standards illustrates the underlying logic of this
argument. Because the problem could, in principle, be avoided if the parties could
commit themselves in advance not to rely too heavily on past performance for standard
setting, we classify it as a problem of imperfect commitment.

The Jl/athematics of the Ratchet Effect

To study the ratchet effect more closely, let us suppose that an employee works
for two periods and exerts effort e 1 in the first period and effort e2 in the second,
but that the contract that is used in the second period is the one that appears
optimal at that time, given the available information. Suppose in addition that
each of these effort levels is observed only imperfectly: The employer observes
only z 1 = e 1 + x 1 in the first period and z 2 = e2 + x 2 in the second, where
the observation errors in the two periods are assumed to have equal variances
and to have means equal to zero. We may write the employee's incentive pay
in the first period as a 1 + B(e 1 + x 1 ), that is, a constant component a 1 plus an
incentive component that is proportional to an unbiased estimate of the
employee's effort.
It is reasonable to suppose there is a positive correlation between x 1 and x 2 : A
high level of x 1 means that a high value of x 2 is likely. This would occur if some
of the same factors that contribute to a high level of measured performance in
the first period also contribute to high measured performance in the second.
Then the parties can use the observed performance in the first period to get an
estimate x 2 of x 2 , the part of second-period performance that is beyond the
employee's control. This estimate of x 2 can then be used to get a better estimate
of the employee's actual effort in the second period. That is, the parties may
define a standard x 2 = -y + 8(e 1 + x 1 ) and form an adjusted esti­
mate of the employee's second-period performance of the form £ 2 = z 2 Xz
= e 2 + X z - X z.
What makes this seem beneficial is that if & an d 'Y arc chosen we ll , then Risk Sharing and
Var(_x2 - i 2) is less than Var(_x2): The adj usted measure eliminates part of the Incentive Contracts
performance variation that is beyond the worker's contro l an d pr ovides a m ore
accurate portrayal of his or her actual performance. Then the informativeness
principle indicates that this estimate should be used in the contract. For example,
if a retail chain bases each store manager's pay partly on the level of sale s at the
store, then it seems only fair (and effi cient) that those stores in good location s
should have to meet a higher sales target. Surely, there is no more gen erally
accurate way to determine the quality of a location than by looking at the past
record of the store's sales, and that is j ust what the proposed standard achieves.
If the employee's second-period pay is given by the same sort of function as in
the first period-a constant term plus an incentive term that rewards higher
estimated effort-then the employee's total compensation over the two periods
is {a. 1 + B 1 (e 1 + x1 )} + {a. 2 + B2 [e2 + Xz - 'Y - &(e 1 + x1 )]}. Collecting
terms allows us to rewrite this as
Total Compensation = a. 1 + a. 2 + (B 1 - &B2 )(e 1 + x1 ) + Bz(e2 + x2 - -y)
(7. 1 1)
Note, however, that the coefficient of e 1 in E quation 7.11 is not the nominal
contractual amount, B i , but rather the smaller amount B 1 - &B 2 . This is the
ratchet effect at work. The direct return-in terms of first-period pay-to extra
effort in the first period is B i , but higher first-period effort increases the standard
in the second period by &. It thus reduces the pay accruin g in the second period
for any choice of second-period effort by &B2 . If the same employee occupies
the j ob in both periods, then the anticipation of first-period performan ce being
used in the second period reduces effective first-period incentives. Thus,
it makes sense to define the "effective incentives" Bt: and B�: by Bt= = B 1
- &B 2 and Bf = Bz.
In terms of the effective incentives, the total certain equ ivalent wealth can be
written as
Total Certain Equivalent = P(e 1 ) + P(e2 ) - C(e 1 ) - C(e2 )
- ½rVar<_Bt=x, + Brx2) (7. 12)
An effi cient contract maximizes this expression subj ect to the incentive con­
straints, Br = C'(e 1 ) and Br = C'(e2 ), that determine the worker's effort choices.
Using properties of the variance (see the formulas in the appendix), E quation
7. 1 2 can be rewritten as
Total Certain Equivalent = P(e 1 ) + P(e2) - C(e 1) - C(e2) - ½r[(BffVar(_x1)
+ (Br)2Var(_xz) + 2BrB�:Cov (x1 , Xz)] ( 7 . 1 3)

where Cov(xi , x2 ) is a measur e of the way the two measurement errors tend to
move together. Recall that we have assumed that Var(_x1 ) and Var(_x2) are eq ual.
Both E quation 7. 1 3 and the incentive constraints are symmetrical with respect
to time: They would be essentially un changed if we were to change each e 1 into
an ez, each ez into an e i , each Br into BL and each B�: into B\-:_ Thus, if there
is a unique total wealth maximizing contract, it must treat the two time periods
symmetrically. That is, it must have e 1 = e2 and Br = Br.
In contrast, we saw that when the parties cannot c ommit in advance to the
second-period contract terms and instead act optimally in the second period
given what has already transpired, they will set e2 > e 1 : In centives are made
more intense over time.
 236
Efficient 0YERCOl\llNG THE RATCHET EFFECT The parties would be better off if they could com­
Incentives: mit to hold the line on incentives, not using the first-period performance to adjust
Contracts and second-period performance standards. In that case, the contractual and effective incen­
Ownership tives would be the same. This policy is in fact in place at some companies. The
Lincoln Electric Company is famous for its extensive use of incentive contracts and, in
particular, piece rates. Lincoln has for decades maintained a policy that once a piece
rate has been set, it will not be changed unless the equipment is changed or new work
methods are introduced. In this case, a new time-and-motion study will be done, and
the resulting standard will remain in effect even if realized performance later suggests
that it is too low. If the standards are set too low, then the Lincoln workers may earn
a lot of extra money, but their incentives to work hard are never threatened.
Why don't more companies adopt such a system? There are many reasons, not
least of which is that it is so difficult for a firm to commit itself not to use available
information. 8 Even if the parties agree in advance not to use the information embodied
in the fii:_st-period performance and set contract terms accordingly, there will still be
efficiency gains after the fact to renegotiating the contract and using the information.
Lincoln Electric has managed to commit to its policy of maintaining standards by
applying piece rates widely throughout its organization, developing expertise at doing
time-and-motion studies, building a reputation for applying its "no revisions" policy
consistently, and tuning the rest of its policies so that they are consistent with a piece­
rate system.
Our characterization of the ratchet effect as a problem of commitment also
helps us to understand how self-employment arrangements and ownersh ip can
sometimes alleviate the problem. A self-employed person sells goods or services directly
to customers. If the industry is competitive, then the performance standards are the
comparative ones set by the marketplace, and a person's good performance does not
lead directly to higher future standards. Similarly, someone who owns an asset can
be assured of keeping whatever gains accrue from showing how to use it effectively.
Of course, the problems of risk sharing often make the ownership solution impractical,
and self-employment is infeasible in many situations.
Within companies, job rotation is another device that can be used to alleviate
the ratchet effect. By assigning people to various jobs over time and using previous
jobholders' performance to set the standard, the current jobholder is not penalized for
a job well done. Job rotation also may bring benefits in improved morale and greater
flexibility in production. Its costs, however, are in potentially reduced efficiency, as
workers have less opportunity to gain experience in any task.

MORAL HAZARD WITH RtSK-NElJfRAL AGENT

The main thrust of this chapter has been to study principal-agent problems where
motivating agents by making them bear part of the risk is costly because the agents
are risk averse. From an analytical perspective-, we can organize our study of other
aspects of the theory by assuming away the risk-sharing problem and studying principal­
agent theory with a risk-neutral agent, that is, one whose coefficient of absolute risk
aversion is equal to zero. In that case, no risk premium is ever incurred, regardless
of how the risks are shared. So, the agent can be perfectly motivated at zero cost by
setting B = 1 , that is, by making him or her bear the entire risk. For a manager­
agent running a firm, this is very much like having the manager buy the firm and

H There may be other reasons that have little or nothing to do with incentive issues directly. For
example, in a multistcp production process, there may be little value to having one worker proceeding
faster than the others, so the firm may not want to encourage every individual worker to work as fast as he
or �he individually can.
enj oy all the profi ts and suffer all the losses. In the case of automobile insurance, it Risk Sharing and
amounts to having drivers paying full cash compensation to those who they have Incentive Contracts
damaged. There are several reasons, however, why a solution of this kind would often
be unworkable, and each of these points to a factor that makes the moral hazard
problem more diffi cult to resolve.

Problems with the Risk -Neutral A gent Scenario

When is making the risk- neutral agent responsible for all financial losses not a workable
solution? First, the solution will fail whenever the agent lacks sufficient funds. A
manager may simply be unable to guarantee payments for the busi ness's expenses with
personal funds, and a dri ver may be unable to pay for damages from a serious accident.
Arrangements in the economic world are often made with these limits i n mind. Public
policy toward drivers generally requires that automobile owners have insurance or
show some other evidence of financial responsi bi lity before their cars can be licensed.
In the pri vate sector, vendors are frequently unwi lling to extend trade credit to a
company with little worki ng capital, for fear that their bills will never be paid.
A second case where making the agent bear the risk is not workable is when the
risk is a nonfinan cial one and is therefore diffi cult or i mpossi ble to transfer. There is
no way for a careless or drunken driver to undo the inj uries or death that may result
from an automobile accident simply by paying damages, or for a negligent blood bank
to bear the sufferi ng of a reci pient of AIDS-tainted blood, or for a company that
dumps toxic or radi oactive waste to eliminate the genetic damage done to vi ctims
simply by paying a cash penalty. All of these examples pose i mportant problems for
public policy, but the principles that ari se are not the kind that are most helpful for
understandi ng the institutions and practices of the business world. S o we merely note
that these problems do limit the theory and pass on.
AD \'ERSE SELECTION IN THE PRINCIPAL-AGENT PROBLEI\I A third case in which "selling
the firm to the manager" is not a workable soluti on is when the pri nci pal and the
agent cannot agree on a price, for example, because the market is disrupted by adverse
selection. This vari ati on is well typified by the case of an employee of a department
store chain who i nvents a new consumer product. Being no expert at marketing, the
employee negotiates with the chai n to market the product. What kind of arrangements
should the employee make for marketi ng the product?
The employee in this case is the princi pal who is trying to negotiate a contract
to motivate the agent (the department store chain) to market the product. U nlike our
earlier examples, this is a case in which the agent is much more tolerant of risk than
is the principal, so efficient risk sharing would dictate that the agent bear (almost) all
the risk, and effi cient incentives for marketing effort by the agent would seem to lead
to the same conclusi on. If these were the only factors involved, the effi cient solution
would be to sell the ri ghts for the pr oduct to the department store chain, which would
then bear all the market risk. Furthermore, as the owner the chain would be motivated
to work as hard as necessary to extract all the potential value from the product.
H owever, there is another element that may block this easy solution-the
element of adverse selecti on. In this case, the chain is an expert marketer who may
be much better informed than is the employee/inventor about the market potential of
the product. If the ri ghts are to be sold to the chain, how will the price be determined?
The chain will refuse the pri nci pal's offer whenever its estimate of sales is low and
will accept the offer when its estimate i s high. Therefore, the i nventor ca n only
successfully sell the rights for a price that is lower than the expected profitability of
th e product.
An alternative procedure would be for the inventor to keep the right to the
 238
Efficient product and demand royalties from the chain, that is, a payment proportional to the
lncenti,,es: number or value of the units sold. In that way, the inventor can mitigate the adverse
Contracts and selection problem because the employer could be expected to accept a royalty contract
Ownership regardless of its private information about the sales potential of the invention. It only
has to pay an amount proportional to its actual sales. There are two problems with
this option too, however. First, the inventor is made to bear too much risk, and,
second, the department store chain, which no longer receives all the prof its from
sales, will be inclined to expend too little effort promoting sales of the product.
Another way that the employee/inventor might try to avoid the problem of
distorting the chain's incentives while receiving some royalties is to base the royalties
on profits rather than on sales. In that way, the chain would be motivated to incur
the right amount of costs to maximize profits because its share of cost is the same as
its share of the revenues. The drawback of this scheme is that the accounting for
expenses is in the hands of the chain, and the employee/inventor ought to expect that
the accoupts will be manipulated to reduce royalty payments. Indeed, in the 1 980s
and in 1 990, there were well-publicized lawsuits by f ilm makers and actors against
Hollywood studios that had agreed to pay a percentage of prof its on movies or television
shows, only to claim that there was little to share. In a celebrated case, actor James
Carner sued over his rights to royalties from The Rockford Files, one of the most
successful television shows in history. According to the accounting procedures used
by the studio, however, the show never earned a profit.
The formal analysis of efficient contracting when there is both moral hazard
and adverse selection is quite complex. The inventor's best policy in the situation we
describe depends on his or her risk aversion, on the importance of motivating the
chain to promote the product aggressively and how that importance depends on
circumstances, and on the quality of the chain's sales forecasts, among other variables.
Although the theory has little to say about the details of the solution, it has much to
say about the form. In a broad range of cases, the best the inventor can do is to offer
the chain a choice between purchasing the full rights to the invention at a relatively
high price or paying a lower price plus royalties proportional to sales. The actual
prices and royalties will depend on the parties' relative bargaining power, but the
inventor should anticipate during the negotiations that the chain will want to own
the rights when it forecasts high sales and to pay a royalty when its forecasts are less
optimistic. If the chain insists that the rights are not worth much, then the inventor
should insist on receiving royalties instead of a fixed payment, even though this may
cause the chain to promote the product less effectively. By selling the invention
outright when its value is high, the inventor motivates the chain to promote sales
vigorously in that case and so increases the value of the invention.
The case of the inventor and the chain store is important because the problems
that arise and the pattern of analysis are similar to those in many other business
settings. The key characteristic of these settings is that there is one party that has
superior information about costs and that needs to be motivated to work hard. For
example, in setting procurement policy, governments (and firms) depend on suppliers
to supply appropriate information about costs and recommendations about product
design and also to work hard to build quality into the products supplied. In the
regulation of utilities, the public utility commissions rely on the regulated firms both
to supply information that will be the basis for price decisions and to work hard to
keep costs as low as possible.
Do all these variations invalidate the general principles presented earlier in the
chapter? They do not. Although such principles as the informativeness principle and
the equal compensation principle arc derived from and phrased in terms of a particular
 239
conceptual mo del (in which agents are risk averse and must be motivated to un dertake Risk Sharing and
personally costly effort), both can be rephra sed to hold over the whole ra nge of I n centive Contracts
va riations described here. Of course, the general principles cannot, by them selves,
substitute for anal ysis of particular ca ses, but they do provide a useful guide across a
wide range of applications.
 240
Efficient
Incentives:
Contracts and
Ownership
Sul\mIARY
Most people in the economy dislike bearing risk. The cost of risk bearing can often
be reduced by sharing the risk among a group of people. If the group is large and the
risks that different people contribute are statistically independent, this procedure can
virtually eliminate the cost of bearing these risks. Insurance companies exist primarily
to perform this economic function. Some kinds of risks, however, are not easily
insured, principally because they are risks that affect many people simultaneously
(such as environmental risks and energy shortages) and so would threaten the capital
of any insurance company. These kinds of risks are managed and shared through
other institutions, with the securities markets playing a prominent role when the risks
are expressible in money terms.
Principal-agent problems are situations in which one party (the principal) relies
on another (the agent) to do work or provide services on his or her behalf. When
agents' actions cannot be easily monitored and their reports easily verified, the agents
have greater scope to pursue their own interests rather than the principal's. Then, to
provide incentives for the agents to behave in the principal's interests, it is necessary
to arrange for them to bear some responsibility for the outcomes of their actions and
therefore to bear more risk than would otherwise be desirable.
Several principles govern the design of optimal incentive contracts. The
informativeness principle says that the cost of providing incentives increases with the
variance of the estimator of the employee's effort. An optimal incentive contract
should base the employee's compensation (or the insured's contribution to cover a
loss) only on the minimum variance indicator of the employee's behavior. This
principle is applied to illuminate issues of comparative performance evaluation
and the use of deductibles in automobile insurance and copayments in health
insurance.
The incentive-intensity principle says that the strength of incentives should be
an increasing function of the marginal returns to the task, the accuracy with which
performance is measured, the responsiveness of the agent's efforts to incentives, and
the agent's risk tolerance. This principle is useful for explaining variations in the
strength of incentives among Japanese subcontractors and among general partners in
certain oil and gas drilling programs.
The monitoring intensity principle states that more resources should be spent
monitoring when it is desirable to give strong incentives. This principle is the mirror
image of the observation that more accurate performance information leads to higher
optimal incentives, which is part of the incentive intensity principle. Measuring
performance carefully and providing intense incentives are complementary activities,
which should be found together.
The equal compensation principle holds that if an employee's allocation of time
and effort behveen alternative tasks cannot be monitored by the employer, then the
marginal returns earned by the employee in any tasks to which he or she actually
devotes effort must be equal. Providing strong incentives for a portion of an employee's
activities can cause the employee to cut back his or her efforts in other activities. This
principle informs comparisons of cost centers and profit centers, policy debates about
the provision of incentives for elementary and secondary school teachers, and analyses
of asset ownership patterns.
The ratchet effect refers to the practice of basing performance targets on past
performance in the same activity. Although such a practice would seem to be
 24 1
consonant with the informativeness principle because it bases performance goals on Risk Sharing and
one of the best available indicators of possible performance, it also imposes certain I ncentive Contracts
costs. If the occupant of the job does not change over time, then the effect of the
ratchet is to punish yesterday's good performance by setting higher standards today.
In modern capitalist economies, ownership and self-employment arc two principal
means by which a person who performs well can guard against being subjected to the
ratchet effect.
There are various factors not included in our models that would make the
principal-agent problem more difficult to resolve. One is that the agent may lack
sufficient capital to pay penalties for losses that he or she causes. A second is that
some losses are essentially nonfinancial losses that cannot be easily compensated using
cash. A third is that the agent may have private information that makes it more
difficult for the principal and the agent to agree on contract terms.

• BIBLIOGRAPHIC NOTES
The economic theory of decisions under uncertainty has its origins in Daniel
Bernoulli's eighteenth-century writings. This theory was put on a sound logical
foundation in the 1 940s through the collaborative efforts of the mathematician
John von Neumann and the economist Oskar Morgenstern. A modern treatment
of this theory has been presented by David Kreps. The refinements contributed
by Kenneth Arrow and John Pratt (and reviewed in the appendix) made it possible
to apply the von Neumann and Morgenstern theory to analyze risks that are
specifically financial in nature. Their work led to the formulas for risk premia
and certain equivalents that we have used. The theory of risk sharing and in­
surance was built on these foundations in the 1 960s by Karl Barch and Robert
Wilson.
The modern theory of i ncentives was begun in the 1 970s by various authors who
explored what optimal incentive compensation contracts might be like in a variety
of different applications including: insurance (Michael Spence and Richard
Zeckhauser), sharecropping (Joseph Stiglitz), tax policies (James Mirrlees), and
managerial compensation (Robert Wilson and Stephen Ross). Stephen Shavell
and Bengt Holmstrom originated the informativeness principle, which was later
generalized in the work of Sanford Grossman and Oliver Hart. Milton Harris and
Artur Raviv also made early important contributions to understanding the nature
of efficient contracts. The ratchet effect has been analyzed by Martin Weitzman,
David Baron and David Besanko, and Xavier Freixas, Roger Guesnerie and Jean
Tirole. The problem of renegotiation of principal-agent contracts was first
studied by Mathias Dewatripont; see also the contributions by Philippe Aghion,
Dewatripont, and Patrick Rey and by Drew Fudenberg and Jean Tirole . Our
discussions of optimal linear incentive contracts and the other principles of incen­
tive pay borrow heavily from the work of Bengt Holmstrom and Paul Milgrom.
There were many contributions to incentive theory in the 1 980s focusing on the
case where there is a tension between the needs to alleviate adverse selection and
moral hazard. In addition to a number of those listed above, leading contributors
to that theory were Joel Demski, Jean-Jacques Laffont, Preston McAfee, John
McMillan, Roger Myerson, Michael Riordan, and David Sappington. Their
theories were often set in the particular situation of a government regulator
(principal) trying to regulate a utility (agent), or a procurement officer (principal)
trying to negotiate a complex contract with a supplier (agent). The principles that
have emerged from these analyses, however, have wide application.
 242
• REFE HENCES
Efficient
Incentives: Aghi on, P., M. Dewatri pont, and P. Rey. " Renegotiation Desi gn under Symmetric
Contracts and Information , " mimeo, 1989.
Ownership Arrow, K. J. Essa ys in the Theory of Risk Bea ring (Chicago: Markham, 1970).
Baron, D. , and D. Besanko. " Regulation and Informati on in a Continuing
Relationshi p, " Informa tion , Economics a nd Policy, I (1984 ), 267-330.
Baron , D. , an d R. Myerson. " Regulating a M onopolist with Unknown Costs, "
Econometrica , 5 0 (July 1982), 911-30.
Bern oulli , E. "S pecimen theori ae novae de mensura sortis, " Com mentarii
Academ iae Scientiarum lmperialis Petrepolita nae," (trans) "Exposition of a New
Theory on the Measurement of Ri sk, " Econometrica , 22 (January 19 54), 23-36.
Barch, K. "Equilibri um in a Reinsurance Market," Econometrica , 30 (July 1962),
424-44.
Demski, J. , and D. Sappi ngton. "O ptimal Incentive Contracts with Multi ple
Agents, " fournal of Economic Theory, 3 3 (1984) 152-71.
Dewatripont, M . "Renegotiati on and Information Revelation over Ti me in
Opti mal Labor Contracts, " Qua rterly fournal of Economics, 104 (1989), 589-
620.
Freixas, X. , R. Gu esnerie, and J. Tirole. "Planning U nder Incomplete Informa­
ti on and the Ratchet Effect, " Review of Econom ic Studies, 52 (198 5 ), 17 3-92.
Fudenberg, D. , and J. Ti role. "M oral Hazard and Renegotiation i n Agency
Contracts, " Econometrica , 58 (N ovember 1990), 1279-1320.
Grossman, S. , and 0. Hart. "An Analysis of the Principal-Agent Problem, "
Econometrica, 51 (1983), 7-4 5.
Harris, M. , and A. Raviv. "Opti mal Incentive Contracts with Imperfect Informa­
tion, " fournal of Economic Theory, 20 (1979), 231-59.
Holmstrom, B. "M oral Hazard and Observability, " Bell fournal of Econom ics, I O
(1979), 74-91.
Holmstrom, B. "M oral Hazard in Teams, " Bell fournal of Econom ics, 13 (1982),
324-40.
H olmstrom, B., and P. Mi lgrom. "Aggregation and Linearity in the Provisi on of
lntertcmporal Incentives, Econometrica , 5 5 (M arc h 1987), 303-28.
H olmstrom, B., and P. Milgr om. "Multi-task Principal-Age nt Analysis: Incentive
Contracts, Asset Ownership and J ob Design, " SITE Worki ng Paper #6, Stanford
U niversity, 1990.
Kreps, D. Notes on the Theory of Choice (Bou lder, CO: Westview Press, 1 988).
Laffont, J. J . , an d J. Tirole. " U sing Cost Observations to Regulate Fi rms, " fournal
of Political Economy, 94 (June 1986), 614-41.
Laffont, J. J. , an d J. Tirole. "The Dynamics oflncentive Contracts, " Econometrica ,
56 (1986), 115 3-7 5 .
McAfce, R . P. , and J . McMi llan . "Competiti on for Agency Contracts, " Ra nd
fournal of Economics, 18 (1987), 396-7.
Mi rrlees, J. "An Ex plorati on in the Theory of Opti mum Income Taxati on , "
Review of Economic Studies, 3 8 (197 1 ) , 17 5-208.
Mi rrlees, J. "Notes on We lfare Economics, Information , and Uncertainty," in
Essays 011 Economic Beha vior Under U11certa inty, 1\1 . Balch, D. McFadden , S.
Wu , c<ls. (Amsterdam : North-l lolland Pu blishing Co. , l 974 ).
 Mirrlees, J . "The Optimal Structure of Incentives an d Authority within an
243
Risk Sharing and

Pratt, J. , "Risk Aversion in the S mall and in the Large, " Econometrica, 32 (1964),
Organization, " Bell fou rnal of Economics, 7 (1976), 105-31. Incentive Contracts

122-36.
Riordan, M. , and D. Sappington. "Information , I ncentives and Organ izational
Mode, " Qua rterly fou rnal of Economics, 102 (1987), 243-64.
Ross, S. "The Economic Theory of Agency: The Principal's Problem, " America n
Economic Review, 63 (1973), 134-39.
S havell, S. "Risk S haring and Incentives in the Principal and Agen t Relationship, "
Bell fournal of Economics, 10 (1979), 5 5-73.
S pence, A. M., and R. Zeckhauser. "Insurance, Information an d Individual

Stiglitz; J . "Incen tives and Risk Sharing in S harecropping, " Review of Economic
Action," American Economic Review, 61 (1971), 380-87.

Stiglitz, J. Incen tives, Risk and Information: N otes Towards a Theory of

Studies, 64 (1974), 219-56.

H ierarchy, " Bell fournal of Law, Economics and Organiza tion, 6 (1975), 5 52-
79.
von Neuman n, J. , and 0. M orgenstern, The Theory of Games and Economic
Behavior, (Princeton: Princeton U niversity Press, 1944).
Weitzman, M. "The Ratchet Principle and Performance Incentives, " Bell fou rnal
of Economics, 11 (1980), 302-8.
Wilson, R. "The Theory of Syndicates, " Econometrica , 36 (January 1968), 119-
32.
Wilson, R. "The Structure of Incentives for Decen tralization, " in La Decision
(Paris: Centre Nationale de la Recherche Scien tifique, 1969).

EXERCISES

Food for Thought

1. In the late 1960s and early 1970s, when M cDonalds (the fast-food chain)
was undergoing a period of very rapid expan sion in sales, it considered a variety of
different compen sation systems for its managers. The compan y wanted to encourage
its managers to increase sales, con trol costs, and maintain the company' s standards of
quality, service, and cleanliness. It also wanted local store managers to hire and train
people who could become managers of new outlets, which were being added to the
chain at a rapid pace. What difficulties would you expect this situation to pose for
McDonald' s man agement? What would you expect to occur if a local outlet manager' s
compensation were based primarily on sales growth? On outlet profits? What kind of
compensation plan should M cD onald' s adopt? H ow would you expect the compen sa­
tion formula to change as McDonald's moved into its next phase, with fewer new
outlets being opened in N orth America?
2. A common complaint of un iversity studen ts is that professors seem too
remote and uninterested in teaching them. H ow do university systems of compensation ,
promotion, and tenure con tribute to the problem? l s the problem likely to be more
severe for tenured or untenured faculty? Why do universities often have rules restricting
outside consulting activities?
3 . Use Figures 7. 3 an d 7. 4 to determine both how the level of monitoring
 244
Efficient and the intensity of incentives would change ( 1 ) if the total cost of monitoring were
Incentives: to fall by a fixed amount and (2) if the marginal cost of monitoring were to fall.
Contracts and 4. Unlike specialty stores, department stores sell a wide array of products to a
Ownership single group of customers, and are often especially interested in maintaining their
reputations for servicing their customers wel l . How might this consideration affect the
compensation of department store sales personnel compared to the salespeople at
specialty outlets?
5 . Suppose a Canadian subsidiary of a British company assembles a product
using inputs manufactured in Canada. U nder what conditions should the manager
of the British firm be responsible if a change in the foreign exchange rate raises the
cost in pounds sterl ing of purchasing the inputs, causing losses to result? Under what
conditions should the manager not be held responsible?
6. In 1 989 the NBC television network in the United States announced a new
way of compensating the local television stations affi liated with the network. These
independently-owned stations carry NBC programs during "prime-time" evening hours
from 8:00 to 1 1 :00 p. m . NBC earns its revenues by selling advertising time to national
advertisers whose ads appear during the network's shows. The amount NBC gets for
its advertising time depends on the number of viewers its programs attract. The
nu mber of viewers watching a particular station (and thus the network's programs)
during prime-time depends in part on the number of people who are attracted to
watch the non-network programs that the station shows before prime time. These
viewers tend to stay with the channel that they started watching on a given even ing.
NBC's innovation was to begin paying its affiliates in part on the basis of the number
of viewers that they attracted for their early-evening programming, such as local news,
rather than just on the size of the audiences that watched the prime-time network
programs on the station . Analyze this plan in terms of the principles developed in
this chapter.

I Mathematical Exercises I
1 . Suppose that a group of people A, . . . , Z share an income risk I, in
proportion to their risk tolerances. For example, individual A bears a share pA/( pA +
. . . pz) where PA = 1/rA , and so on . Show that A's risk premium in this case is
½ PA Var(I)/( pA + . . . p2) 2 and hence that the total risk premium borne by all the
members is ½Var(I)/( pA + . . . + pz)-the same as the risk premium that would be
required by a single person whose risk tolerance is PA + . . . p2 .
2. Consider the case of two people, A and B, with incomes IA and IB , and
suppose that they enter a risk sharing contract so that A's income is c:xIA + BIB + "(,
with B receiving the balance. The total risk premium for this arrangement is given
by Equation 7. 1 in the chapter. (a) Expand this equation into one expressed in terms
of Var(IA ), Var(IB ), and Cov(IA , IB ) · [Hint: Your answer should be a quadratic function
of ex and B. ] (b) To find the values of ex, B, and 'Y that minimize this expression, take
the derivatives of the expression with respect to ex and B and set the derivatives equal
to zero. Show that the solution of these three equations has ex = B = PA/( pA + PB ),
where PA = 1 /rA and PB = 1/rB are the two risk tolerances. (c) Use mathematical
induction and the results of this and the preceding problem to show that regardless
of the number of people, person l 's share of each risk should be the same as his or
her share of the total risk tolerance of the group.
3. In the text we compared the advantages of relative performance evaluation
aga inst an evaluation based solely on the employee's own performance. Here we
consider all combinations of the two as well. Thus, suppose manager A's measured
 245
performance is eA + xA + Xe and B's m easured performance is e8 + x8 + Xe , where Risk Sharing and
xA , x8 , and Xe are i ndependent sources of randomness. S uppose it is proposed to base Incentive Contracts
manager A's compensati on on hi s or her own performance mi nus 6 times some
measure of B's performance. Fi nd the value of 6 that mi nimizes the variance of the
performance measure. H ow does thi s value change with changes i n Var(xA )? Changes
i n Var(x8 )? Changes i n Var(xc )?
4. Suppose an entrepreneur can select among i nvestment proj ects that all cost
the same amount but differ i n their risk-return characteristics. The set of available
proj ects is descri bed by a curve givi ng the highest available expected net return (after
subtracti ng the initial cost of the i nvestment) corresponding to any given variance i n
the returns. Let this curve be m = 2 v - (½)v 2, 0 5 v s 2, where m i s the mean
return and v i s the variance of returns. Thus, the entrepreneur can achieve a riskless
return of m = $0, essenti ally by not i nvesti ng, while the maxi mum expected return
is attai ned by selecting a proj ect wi th v = 2, which yields an expected return of 2.
What proj ect will the entrepreneur choose if he or she must bear all the retur ns
(positi ve or negati ve) alone and he or she has a coefficient of risk aversi on of r > 0,
so that his or her preferences are given in certai n-eq uivalent form by m - (½)rv?
Now suppose that i t is possi ble to share the risk of the i nvestment with an outside
investor who has a coefficient of risk aversi on of s. What is the investment choice
that maximizes the total certai n equivalent? H ow should the risk be shared? Could
thi s be achieved by selli ng an ownership claim i n the entrepreneur's firm to the
i nvestor in such a way that the i nvestor will be willi ng to pay enough that the
entrepreneur is better off selli ng this share?
5. A ri sk-averse entrepreneur is considering selling stock i n his or her company
to the public. He or she will conti nue to manage the firm after i t "goes public. " The
entrepreneur gets utility from income, x, and from the consumpti on of on- the-j ob
x
perquisi tes, c, accordi ng to the uti li ty functi on u(x, c) = - ½var(x) + 1 00c1 , where
x is the mean of the i ncome x and var(x) is i ts variance. The uncertai n profits of the
firm are Y - c: each dollar spent on perq uisites reduces profit by a dollar. The
variance of Y, (and thus of (Y - c)), is cr2 , which we assume to be greater than 2, 500.
The entrepreneur is currently the sole owner of the firm, recei ving as i ncome the
firm's pr ofit. What level of perquisites will he or she choose?
Now, suppose the entrepreneur sells a fracti on ex of the firm to ri sk-neutral
i nvestors, retaini ng (1 - ex) for him- or herself. Thus, the entrepreneur receives as
i ncome whatever amount the i nvestors pay for this fracti on of the firm, say M(ex), and
then gets hi s or her share, ( I - ex)(Y - c), of the random profit. The variance of his
or her i ncome is thus (1 - ex)2cr2 • What is the relati onship between ex, the fracti on
of ownershi p the entrepreneur sells, and the level of c he or she wi ll subseq uently
choose? Does thi s choice maximize total value (the expected utili ty of the entrepreneur
plus that of the i nvestors)? What will be the expected pr ofit as a fu ncti on of ex?
Assume that competiti on among i nvestors leads them to pay an amount for any
given ownershi p share equal to the profits they expect to recei ve. H ow much will the
entrepreneur receive from selli ng a fracti on ex of the firm if i nvestors correctly anticipate
the level of c that the entrepreneur will choose after selli ng that fracti on of the firm?
What is the realized level of expected utility for the entrepreneur from selling a
fracti on ex of the firm if the investors have correct expectati ons? What is the best level
of ex for him or her to pick? Could the entrepreneur gain by bi ndi ng him- or herself
not to i ncrease c as ex changes?
 246
Efficient

MATHEMATICAL APPENDIX
Incentives:
Contracts and
Ownership
This appendix consists of two parts. The first is a review of statistical concepts. The
second derives the approximation reported in the main text regarding the certain
income that is equivalent, from the decision maker's point of view, to a given random
(uncertain) income.

REVIEW OF STATISTICAL CONCEPTS

The set of possible outcomes in a statistical problem is represented by a sample space
S. A random variable x is a function that associates with each element s E S a real
number x(s). For example, the elements of S may be the books on a bookshelf and
x(s) may specify the number of pages in the book s. In a coin-tossing problem, the
elements of -S may be sequences of heads and tails and x may be some statistic, such
as the one that assigns to each sequence s E S the number of heads that occur in the
first ten coin tosses. With the sample space S comes a probability mass function p
that assigns a probability p(s) to each element of s E S. The probability mass function
is used to compute such probabilities as Prob(x = 1 1 )-the probability that the random
variable x takes the value 1 1 .
Each random variable x has a mean denoted by x,
also called its expectation
and denoted E [x]. The formula for calculating expectations is:
E[x] = L s E s p(s)x(s) = x (7. 14)
Each random variable also has a variance given by:
Var(x) = E[(x - x) 2 ] = L s E sP(s)(x - x) 2 (7 . 1 5 )
Variance is one measure (among many possible measures) of the degree of randomness
of X .
Given two random variables x and y, the covariance of x and y is:
Cov(x, y) = E [(x - x)(y - y)] (7. 16)
Notice that Cov(x, x) = Var(x).
Given two random variables x and y and two real numbers a and S, we can
form a new random variable ax + By, which for each possible outcome s takes the
value ax(s) + Sy(s). Its expectation can be computed from those of x and y by the
following formula, which can be derived from Equation 7. 1 4:
E[ax + Sy] = aE [x] + SE [y] (7. 17)
Using Equations 7. 1 4-7. 1 7, we can derive the formula:
Var(a x + Sy) = a 2 Var(x) + B 2Var(y) + 2aSCov(x, y) (7. 1 8)
The two random variables x and y are (statistically) independent if for all
numbers a and S, Prob[x = a and y = S] = Prob[x = a] · Prob[y = S] . Statistical
independence represents the idea that knowing the value of one of the variables
provides no information about the value of the other. If x and y are independent,
then Cov(x, y) = 0, so by Equation 7. 1 8, Var(x+ y) = Var(x) + Var(y).

EVALUATING FINANCIAL RISKS:

CEHTA I I\ EOl l \'A LENTS AND RISK PRE� I I A
Expected utility theory establishes conditions under which a decision maker will rank
risky prospects according to their associated expected utilities. Let u be a function that
 247
assigns to each monetary outcome x a utility u(x). Then, representi ng prospects by Risk Sharing and
ran dom variables, the expected utility of prospect x i s E[u(x)]. Let us com pare this Incenti ve Contracts
prospect with a certain prospect, that is, one that yields the payment i with pro bability
1. The certain prospect will be preferred if u(x) > E[u(x)]; the risky prospect will be
preferred if the reverse i nequality holds. When the decision m aker is j ust i ndifferent
between the two prospects, then x is called the certai n equi valent of the prospect x.
The crucial form ula for our applications is the followi ng approxim ation:
Approximation . S uppose that u is three tim es conti nuously differenti able, that
x = E[x], and that u'(· ) > 0. Then, appro xim ately, the certain equivalent is:
i =x - ½r (x)Va r{x) (7. 19)
where r{x) = - u"(x)/u'(x).
Derivation . According to Taylor's theorem, for any z,
u(z) = u(x) + (z - x )u'(x) + ½ (z - x )2 u"(x ) + R(z)
.

where R(z) = u '"(z)(z - x)3/6 for some i E [x, z]. We assume that this rem ai nder
term is negligi ble. Hence we write, approxim ately,
u(z) = u(x) + (z - x )u'(x ) + ½(z - x )2 u"(x ) (7. 20)
Substituti ng x for z in Equation 7. 20 and com puting the expectation, we find,
approxim ately,
E[u(x)] = u(x) + E[x - x ]u'(x ) + ½ E[(x - x )2 ]u"(x )
But, E[x - x] = E[x] - x= x - x= 0, so approxi m ately,
E[u(x)] = u(x) + ½ E [(x - x )2 ]u"(x ) (7. 2 1 )
We expect that the certai n equi valent x will be close to x, so we approxi m ate
its utility differently, al so usi ng Taylor's theorem ,
U(X) = U(X) + (X - X)LJ (x ) + R(i)
I (7. 22)
where, R(i) = ½u"(.z)(i - x) 2 for som e i E [x, x]. Because we apply the approxim ation
only when x - x i s sm all, we agai n treat the remai nder term as negligi ble. Fo r a
certai n equi valent, we have u(x) = E[u(x)] . So, com bini ng Equations 7 . 21 and 7. 22,
we have, approximately,
(i - x)u'(x ) = ½ E[(x - x)2 ]u"(x ) (7. 2 3 )
This m ay be expressed i n the form
i - x = ½ • [u"(x)/u'(x)] · E[(x - x )2 ] = - ½ · r(x ) · Va r{x) (7. 24)
which establishes the Approximatio n (7. 1 9).
 8
RENTS AND EFFICIENCY

IR l eputation said, "If once we sever, Our chance of future Meeting is but vain;
Who parts from me, m ust look to part for ever, For Reputation lost comes not
again. "
Charles Lam b 1
i
Highly paid labour is generally effcient a nd therefore not dear labour, a fact which
though it is more full of hope for the future of the human race than any other that is
known to us, will be found to exercise a very complica ting infl.uence on the theory of
distribution .
Alfred Marshal/ 2

\rt IE!\ DtSTHIBl TIOl\ A FFECTS EFFICIENCY

Up to this point, almost all of our analysis has been founded on the value maximization
principle (Chapter 2), which holds that if the condition of no wealth effects is satisfied,
then an arrangement is efficient for the parties involved in an agreement if and only
if it maximizes their total equivalent wealth . The no wealth effects condition requires
( 1 ) that the individual parties evaluate all the benefits they receive and all the costs
and risks they must bear as being equivalent to some cash transfer, (2) that these
evaluations do not depend on the amount of wealth that the parties hold, and (3) that
people are able to make timely payments in whatever amounts may be required to
divide up the benefits of the transaction without affecting the cost or feasibility of any
other aspect of the transaction. Where the value maximization principle applies,
efficient arrangements are those that maximize total wealth, regardless of how that
total is shared.
Even though it is rare for the no wealth effects condition to be satisfied completely
and precisely, it still provides a useful starting point for many analyses. It may be
approximately correct in some circumstances, and even when it is not, it allows a

1 "Love, Death and Reputation," stanza IV.

2 Principles of Economics. 8th ed . (London: Macmillan, 1920), 423.
 249
very mcful sim plifying separatio n between the issu es of distribu tion and efficiency. Rents and
However, there are many cases in wh ich wealth effects are so important that co ncerns Efficiency
for distributio n and efficie ncy canno t be analyzed separately withou t greatly distorting
the facts. For example, in developing cou ntries, workers with higher wages migh t eat
health ier, more nu tritious diets, givi ng them more e nergy a nd higher productivity
levels. For su ch workers, paying low wages would limit th e volume or quality of the
work they can do, and so paying higher wages increases the total weal th that can be
created. This co nsideratio n is too important to ignore, especially in the poores t
countries.
Even in developed cou ntries, the distributio n of wealth migh t affect the cost or
feasibility of some ki nds of transactions. For example, according to our analysis of
i ncentive co ntracting in Chapter 7, if the agent is risk neutral then the moral hazard
problem can be completely avoided by havi ng the agent receive the full returns to his
or her actions. Wi th a risk-neu tral agent, there is no need for risk sh aring and no
gain to rem�ving any of the risk. The solu tio n is for the principal to receive a co nstant
amou nt, independent of the ou tcomes generated by the agent's actions, with the agent
bearing the full variability in income. Where the agent is a manager of a firm, this
solution involves selling him or her the firm. U nless the agent's i ncome or wealth is
sufficiently high, however, he or she may be u nable to raise the fu nds necessary to
buy the operation. The value maximization princi ple then does not apply because
the parties cannot make the necessary transfers.
A closely related difficulty arises in the questio n of who should own the tools
or machines a worker uses (see Chapter 7). Co nsider the case of a long-dis tance tru ck
driver. The driver can significantly affect the value of the truck by the care he or she
exerts in drivi ng and maintaining it. If the efficient solu tion is for the driver to own
the truck so that he or she is stro ngly mo tivated to pro tect its value, then effi ciency
cannot be achi eved u nless the driver can afford to own it. Agai n, the principle does
not apply in this case. The resulting inefficiency raises the possibili ty that the to tal
weal th created migh t be increased by allo wing the workers to recei ve a larger share of
the i ncome they generate so that they can afford to buy the truck.
Limits on a party's ability to pay cash damages are another commo n reaso n for
distributio n and efficiency to be linked. For example, financial penalties are used
much less often than simple theory would su ggest. Such theory notes that financial
penalties are especially effi cient means of providing incentives because the penalty is
only a transfer: The loss to the penalized i ndividual is offset by a gain for the o ne
receiving the penalty payment. In co ntras t, other forms of pu nish ment (j ailing,
whipping, os tracism, death) involve actual net reductio ns in total welfare-the
punished person is hurt, and no o ne has an offsetting gain. However, fines and
fi nancial penalties are of hardly any use to pu nish a borrower who has go ne broke
and defaulted o n a loan. Similarly, cash penal ties are hardly ever used in employment
co ntracts, except sometimes for employees with very large incomes relative to the
damage that their misbehavior can do. For example, professio nal athl etes' co ntracts
commo nly grant management the righ t to assess fines for breaking team rules, bu t
simil ar clauses are much rarer for factory workers. When cash penalties are limited,
other details of the transaction may have to be manipulated to provide suitable
incentives to the parties .
I nfluence activities also represent an instance where the distribu tio n of costs and
benefits affects efficiency (see Chapter 6). Costs are incurred in trying to shift
orga nizational decisions so that they favor o ne grou p or another, and no direct
aggregate benefit results.
When the value maximizatio n principle fails to apply, the way the total weal th
generated by a proj ect or venture is shared can affect other aspects of the transactio n,
 250
Efficient such as who is hired to do a job or what technology is used. In de\·eloping countries,
Incentives: for example, if workers with higher incomes are more productive because they are
Contracts and better nourished, and if the wages paid to workers are so low that nourishment is
Ownership problematic, then employers may choose to hire workers from wealthier families,
even though a worker from a poorer family may be better matched to the job in terms
of skills and work attitude. Of course, hiring wealthier workers exacerbates the problem
of income inequality and contributes to the much discussed "cycle of poverty. "
Similarly, wealthier drivers who own their trucks may earn much more than do other
drivers, further increasing the wealth differences between them. These examoles also
illustrate the important possibility that private parties may agree to arrangements that
are technically inefficient when distribution affects productivity. 3
A host of new issues and questions are raised when wealth effects are important.
One group of questions concerns how to provide effective incentives when cash
penalties for bad behavior are impossible. For example, how do lenders manage their
loans when borrowers may default and the loan may become uncollectible? How do
firms motivate workers when fines cannot be used? Other questions concern how the
wealth created by a venture will be shared. For example, how does the legal system
manage the competing claims on a firm's resources that are made in bankruptcy
proceedings when payment of all claims is impossible or when it would so damage
the firm as to lead to a reduction in total wealth? How do parties determine
compensation levels when compensation may affect efficiency? A third class of
questions concerns the sources and consequences of inequality within organizations:
How do wages and perquisites vary across jobs within a single firm? How does the
firm manage the resulting competition for the good jobs? These types of questions are
examined in this chapter.

EFFICIE!\CY WAGES FOR El\tPLOYl\lENT INCENTI\'ES

Suppose that an organization needs to rely on the good and honest behavior of some
employee. The organization may be a police department relying on its officers to
reject all attempts at bribery, an oil company relying on the captains of its tankers
to stay alert and sober on the job, an investment company relying on its analysts to
investigate alternative investment options with diligence, or a marketing division
relying on its salespeople to entertain customers and not family or friends. Suppose,
too, that (perhaps because of wealth limitations) the only effective means available to
discipline an employee suspected of cheating is to terminate his or her employment.
Our first question is: What determines whether the threat of termination of employment
is an effective deterrent against cheating? Our answer is that the employee's incentives
depend on what the employee stands to gain by cheating if undetected, how likely
cheating is to go undetected, what wage the employer offers, and what other market
opportunities the employee may have. A simple model based on one developed by
Carl Shapiro and Joseph Stiglitz will help to make the point clear.

The Shapiro-Stiglitz Model

Let w be the wage the employee is paid in his or her current job (say, $ 50,000), and
let w be the wage the employee could get if he or she looked for another job in the
market after being fired from the current job (say, $40, 000). The wage w is assumed

1 An arrangement is teehnieally ineffieient if there is another arrangement that produces more with
the same total amount of resources. In this exa mple, a tramfer of wealth from the rieh family to the poor
one, which increased the productivity of the poor worker, would increase techniea l efficieney. The i nitial
arra ngement, however, is still Pareto efficicnt; the proposed alternative is not Pareto superior because it
would leave the wea lthy family less well off.
 25 1
to be discoun ted to take into accoun t the cost of search in g for a new job, including Rents and
any period of unemploymen t durin g the search, the cost of submittin g applica tion s, Efficiency
an d so on. Let g be the amoun t that the em ployee could gain by cheating on the job,
regardless of whether he or she is detected. This gain migh t take the form of income
from bribes, increased leisure from working shorter hours than agreed, material gain
from redirecting bu siness resources to benefi t family an d frien ds, or sim ply reduced
pressures from respon din g less dil igen tly to organ izational deman ds an d crises. Let p
be th e pro bability that the cheating, whatever form it may take, is detec ted. Let N be
the mu ltiplier to express th e long- term value of th is relation ship, adjustin g for both
in terest an d the number of periods during which the worker is employed by the firm
if n ot caught cheatin g. If the employmen t relati on lasts on ly on e period and is n ever
repeated, then N = 1 . If the employee is hired from time to time an d expects to be
rehired in the future only if he or she has n ever cheated the employer, then N is
larger than l_. If there is a con tinuing employmen t relation between the worker an d
the employer, then N may be much larger than 1.
Puttin g ethics an d morality aside momentarily, it will serve the employee's
narrow personal interest to cheat if
g > p(w - w ) N ( 8. 1)
that is, if the gain from cheatin g exceeds the pro bability that the cheatin g is detected
times loss of income from being fired.4 For example, if g is $1000 an d p is . 0 5 , then
cheatin g earn s $1000 bu t incurs an expected cost of on ly .0 5($5 0,000 - 40,000) =
$500, so cheatin g is profi table. The amoun t w - w is called the rent that the
employee earn s from the job: The excess of earnings in the curren t job over
opportunities el sewhere. The possibility of earn ing this ren t makes the job an d the
prospect of being rehired in the future valuable to the employee and makes bein g
fired an outc ome to be avoided.
Despite the pure, self-in terested calculation that we have ascribed to the
employee, we do not argue that ethics an d morality do not matter. Few people are
either devils or sain ts. Most do not always cheat wh en cheating is profi table, an d most
sometimes succumb to the temptation to cheat, perhaps by con vincing themselves
that what they are doing is not really dishonest. Con sideration s of right and wrong
certain ly do en ter in to many peoples' decisions an d affect their behavior. Nevertheless,
experience suggests that the amoun t of cheatin g that goes on in an organ ization is
respon sive to the incen tives for cheating. M oreover, when cheatin g goes unpun ished,
i t breeds more of the same: Once some people are seen to be cheatin g, an "every body
does it" attitude may in fec t an increasing fraction of the organ ization's employees. In
that way, an in itial failure to provide the right incen tives may lead to an epidemic of
cheatin g spreadin g throughout the organ izati on .
Even for the most hon est people, repeated an d su bstan tial temptation s to cheat
combin ed wi th ambiguity about what is right an d wrong are likely to resu lt in
occasional cheatin g. Inciden ts like those where religious leaders are caught diverting
donation s to their own personal use should be enough to convince even skeptics that
material tem ptation s affect th e behavior of people in every kin d of position . An
organ ization can best forestall cheating by makin g it un profitabl e, that is, by settin g
its monitorin g in ten sity p an d its wage w to en sure that the expected loss of wages
from cheatin g, p(w - w)N, exceeds the gain from cheatin g, g .
Thi s an alysis shows one reason why highly pai d workers may be more diligen t

4 This formula assumes that the employee's options beyond the horizon embodied i n N are not
affceted by whether or not he or she is caught cheating, so that the gain from being with the firm is realized
solely over the period represented by the m ultiplier N.
 252
Efficient and productive than are similar workers earning a lower wage, as suggested by the
Incentives: quotation from Alfred Marshall in the beginning of the chapter: Highly paid workers
Contracts and have more to lose by cheating, so they find it in their interests to behave honestly.
Ownership
MANAGING DISTANT TRADING AGENTS The idea that high wages encourage honest
behavior is an old one. Thomas Macaulay, describing the efforts of Lord Clive to
control corruption among English civil servants employed by the East India Company
in India around 1 765, wrote: "Clive saw clearly that it was absurd to give men power
and to require them to live in penury. He justly concluded that no reform could be
effectual which should not be coupled with a plan for liberally remunerating the civil
servants of the Company. " 5 Lord Clive undertook vigorous efforts to change the
standard of behavior among civil servants but recognized that his efforts could not be
fully effective unless the civil servants found it in their individual interests to adhere
to the new standard.
A similar system of high compensation was used by the Maghribi traders in
Fustat (old Cairo) in the eleventh century. 6 These merchants, who shipped goods
across the Mediterranean to Sicily and other destinations, involved the whole merchant
group in ostracizing cheaters. This system made a good reputation a most precious
commodity that merchants would go to great lengths to preserve. Mathematically,
involving the whole group in the punishment can be seen in terms of decreasing w­
making the outside opportunities less attractive because no one else in the group will
employ a cheater-or in terms of increasing N-increasing the multiplier because
not only the opportunities to represent the cheated trader were lost, but also those
opportunities to represent other traders. Either of these makes cheating less attractive.
The extensive web of social relations among the traders, which allowed them all to
be quickly informed when an agent cheated, made this system of group sanctions
even more effective by increasing p.
Macaulay's emphasis on power corresponds to a high value of g in our model;
powerful civil servants were able to extort the locals, to sell favors, and so on.
Moreover, the remoteness of the regions in which these civil servants operated made
this behavior hard to monitor. In terms of our model, p is small. As expression 8. 1
shows, when the gain g is large and the chance p of being caught cheating is small,
then only by paying a large premium over the worker's opportunity wage can proper
incentives be provided: "it was absurd to give men power and to require them to live
in penury. "
Theoretically, the smallest wage that can deter cheating is w = w + gl(Np),
which exceeds the opportunity wage by the amount gl(Np). This wage is called an
efficiency wage; it is set higher than market-clearing wages in order to motivate the
employee to work more efficiently. By paying an efficiency wage, the company
established a financial reward for honest behavior from its employees, and so
discouraged them from cheating.
EFFICIENCY WAGES \'ERSUS INCENTI\'E PAYMENTS In view of our analy sis of incentive
payments in Chapter 6, you may wonder why explicit incentive payments are used
in some circumstances and efficiency wages in others. That question surely merits
more study, but a preliminary answer can be given now.
In our analysis, when incentive contracts are feasible, they have been calculated

5 Thomas Babington Macaulay, "Lord Clive," in Macaulay: Poetry and Prose (Cambridge, MA:
I larvard University Press, 1967), 35 5-57, as quoted by Robert Klitgaard in Con trolling Corruption (Berkeley:
University of California Press, I 988), 81.
6 Avner Greif, "Reputation and Coalitions in Medieval Trade: Evidence on the t\1aghribi Traders,"
Jo11 mal of Economic History, 49 ( 1 989), 857-82.
to be the optimal form of contract. Efficiency wage co ntracts are to be preferred whe n Rents and
explicit incentive con tracts of the type we have described in Ch apter 7 arc not feasible. Efficiency
There arc several possible reasons that in centive co ntracts may no t work. The simplest
is th at the employee may be unable to ge t by on the low levels of income that might
be ge nerated by the in centive co ntract when the signals of performance are low or to
pay the fines that the contracts sometimes req uire. This is analogous to our introductory
discussion about why ownership might be infeasible even wh en it is optimal-the
wo uld- be owner might be unable to finance hi s or her purchase. It wo uld be possible
to raise the portion of the emplo yee's pay that is not performance dependent so that
he or she can always afford the co ntractual payments and so that the performance
incentives are maintained. This may be too expensive for the emplo yer, however,
who may prefer an efficiency wage approach.
The second reaso n incentive co ntracts may not work is that an efficiency wage
scheme can �tilize subjective performance evaluatio ns in a way that is pro blematic
for a system of incentive pay. When performance evaluations are subj ective, the
employer may be tempted to underrate performance in order to reduce the required
incentive payment. In anticipatio n of that, the emplo yee must be less respo nsive to
any promised incentive payments. The problem is o ne of moral hazard o n the part
of the employer, whose assessment of employee performance cannot be obj ectively
verified. With the efficiency wage scheme, however, the employer pays the same
wage of w to any employee in the job. The employer cannot save wage payments by
dismissing an employee (assuming the employee has to be replaced), and he or she
would be reluctant to dismiss a go od employee anyway. The efficiency wage scheme
thus solves the employer's moral hazard problem.

UNEMPLOYMENT AND OUTSIDE OPPORTUNITI ES An apparent pro blem is that firms

using efficiency wages must pay high wages relative to employees' outside opportunities,
but it is impossible for all firms to pay high wages relative to o ne ano ther. O ne possible
escape from this dilemma is that there be unemployment: A worker who loses a job
is not immediately able to find another and so suffers a loss, even though o nce he or
she finds employment again, it is at the same high wage as before. S hapiro and S tiglitz
in fact accentuate this possibility. H igh wages reduce the aggregate demand for labor,
creating unemployment that allows all firms to have high wages relative to their
employees' outside opportunities. Of course, the output that could have been produced
by this "reserve army of the unemplo yed" represents a social co st.
Another po ssibility that resembles patterns in the Japanese eco nomy has been
suggested by Masahiro Okuno-Fuj iwara. At least the largest Japanese firms focus their
hiring on new graduates, and these firms offer lifetime employment to the male
workers they recruit who are recent graduates from schools or universities. There has
traditio nally been very little job changing, at least amo ng the people with permanent
jobs in major firms, and these firms have rarely hired peo ple more than a very few
years out of school. Of course, the J apanese eco nomy is generally perceived to operate
extremely efficiently, with high levels of productivity and little unemployment.
Suppose that firms focus their hiring o n new entrants into the labor force, and
suppose that o nce workers join a firm, they stay there for their whole careers. If no
one ever cheats and is fired, then there will never be mid-career workers looking for
em ployment. Because workers never observe mid-career people being hired, they
might reaso nably conclude that there is no market for such people. As a result, being
fired would mean permanent unemployment, and this prospect is so dismal that it
deters cheating completely. Thus, the high levels of effort and ho nesty associated with
effic.:iency wages could be achieved without the social cost of unemplo yment.
An y detailed co nsideratio n of whether the theoretical po ssibility identified by
 254
Efficient Okuno-Fujiwara has actual relevance to explaining the Japanese economy's perfor­
Incentives: mance would take us too far afield. However, a theoretical criticism of the possibility
Contracts and of a full-employment, no-cheating equilibrium should be noted. As Okuno-Fujiwara
Ownership himself noted, this equilibrium is very fragile. If there were, for whatever reason,
mid-career people looking for jobs, then (at least in terms of the formal model) there
is no reason for firms to discriminate against such people. They would be hired, and
this would presumably upset this expectations-based equilibrium.
Interesti ngly, there has been a widely noted increase in the number of managers
and professionals changing jobs in recent years in Japan. A major element in this
i ncreased mobility has been the entrance of international firms to the Japanese market,
especially commercial and investment banks. They have sought to hire skilled Japanese
nationals and have been especially successful in attracting those who have studied
abroad for MBAs and law degrees. It will be interesting to see what will happen if
this pattern begins to spread to Japanese firms.

A Mathematical Example :
Comparative Statics for Efficiency \V ages
In the Lord Clive example, the company's response was not simply to pay larger
premiums-it intensified monitoring of the employees as well. Lord Clive brought in
outsiders to audit and inspect local practices, being on the lookout for bribery and
favoritism. Our goal in this section is to study a mathematical representation of how
monitoring intensity and wage i ncentives are used together to provide incentives when
monitoring is costly. We then study how changes in a parameter describing the
problem affects the optimal solution.
Let M(p) be the monitoring cost incurred in every accounting period when the
probability of detection is p . If the company wants to deter cheating, it can minimize
its total cost in each period by choosing its wage and monitoring policy to solve:

p, w
Minimize M(p) + w

subject to p(w - w ) N 2:'.: g ( 8. 2 )

Equation 8. 2 is called a minimum cost implementation problem. Assuming
that the company wants to encourage (implement) a particular kind of behavior and
to deter cheating, the problem is to determine the least costly way of doing so. The
objective being minimized in expression 8. 2 is the cost that is incurred during each
period of employment, and what keeps the company from paying a low wage and
doing little monitoring is the incentive constraint expressed in the second line of the
expression. It is clear that the company will not pay a higher wage than is minimally
necessary to provide i ncentives. Therefore, at the optimal solution:

w = w + _K_
Np (8 . 3 )
Substituting that into the objective in expression 8. 2 yields:
g
Minimize M(p) + w +
p Np ( 8.4)
At the optimal solution p*, the first derivative of expression 8. 4 with respect to p must
be zero and the second derivative must be nonnegative. That is,
0 = M' (p* ) - glp* 2N and ( 8 . 5)
0 ::; M" ( p * ) + 2glp* � N ( 8. 6 )
 255
Rents and
Efficiency

p* Figure 8. 1 : The optimal probability of detec­

Probability tion, p*, decreases when N increases .

To see how the soluti on depends on the parameter N, consider Figure 8. 1. The
horiz ontal axi s measures p, the probability of detecting cheati ng. The vertical axi s
measures dollar returns. Accordi ng to Equati on 8. 5, at the optimum we must have
M'(p*) = gl(p * )2N. The right-hand side of this expressi on is the change i n the optimal
efficiency wage as p i ncreases. It is plai nly decreasi ng in p. Thi s is the curve that falls
from left to right i n the fi gure. The left-hand side is the i ncrease i n monitoring c osts
as p increases. It might be rising or falling, but expressi on 8. 6 says that if it is fa lli ng,
it does not fall as fast as the ri ght-hand side falls. In the figure we have shown the
first case. The soluti on is at p* i n the diagram, where the two curves i ntersect.
N ow suppose N, the number of peri ods the employee might work for the firm,
increases. This plaini y decreases gl(p*)2N for any value of p, which i n the fi gure
appears as a downward shift i n the curve. The new intersecti on poi nt is the soluti on
wi th the higher level of N. It corresponds to a lower value of p, as the figure shows.
As the potential length of a relationship increases, there is more trust and less direct
monitoring because the gains from the relationship a re larger and so the incentives to
cheat are less severe. 7
In an exercise at the end of the chapter, the reader is asked to demonstrate the
additional propositi on that the optimal wage w*(N) also declines with N pr ovided that
M ' and M" are posi tive. A longer horizon on the relationship reduces the surplus that
must be received in any pa rticular period to deter cheating.
The next step in the formal analysis is to check whether implementi ng the
desired behavi or is worthwhile. Let 7r(b) be the gross profi t earned if the behavi or b
is implemented, and let C(b) be the correspondi ng mi nimum cost, as determined
from the implementati on problem. We can then build on the earlier analysi s to study
the problem of maximizing 7r(b) - C(b) by choice of the behavi or b. There is li ttle
to be sai d about thi s problem in general except that, other things bei ng equal, the
higher the c ost of impleme nti ng any particular behavi or the less likely it will be that
the behavior will be optimal.
The actual pr oblem faced by any real company is much more detailed than thi s
simple problem i ndicates, i nvolvi ng choices of who to hire, what to motivate, how
to monitor, and much more. These questi ons are considered in more detail i n
Chapters I O through 13 . The formal analysis conducted here analyzes only that part
of the c ompany's problem that we have called the mi nimum cost implementati on

7 Because we employed the extra 3ssumption that p* is differentiable , we have proved only a special
case of this proposition, but it can be shown to be true quite generally.
 256
Efficient problem, which is the problem of determining how to encourage a specified behavior
lncenti\'es: at the least cost.
Contracts and One way to test the significance of the analysis would be to examine how pay
Ownership levels are affected by the kinds of variables used in the minimum cost implementation
problem. That would entail finding surrogates for g, M, and N and examining whether
increases in g and M' and decreases in N are associated statistically with larger relative
wages. A more intuitive test of the correctness and significance of the theory can be
deduced from the second step of the theory, in which the firm chooses what kind of
behavior to encourage. If the effects we describe here are important, then observation
of real organizations ought to show that the optimal choice is often to set standards
of behavior that would be inefficient if enforcement were costless but that are easier
to monitor and enforce than the seemingly more efficient behavior.
This prediction does seem to accord well with a variety of actual practices. For
example, many firms prohibit making personal telephone calls on business phones
and during business hours. In principle, it might seem better if the company policy
permitted only important personal calls to be made, but that policy would be hard to
enforce. It would be costly (and perhaps unacceptable to employees) for the firm to
monitor the content of personal calls to judge their importance. Furthermore, because
importance is such a subjective matter, judgments of importance are subject to abuse.
A policy that proscribes all personal calls (except, perhaps, when special authorization
is obtained) is far less costly to enforce. For another example, rather than trying to
limit the costs incurred on business travel to those that seem justified, many companies
allow their employees to spend any amount up to a prespecified limit, knowing full
well that the employees will eat fancier dinners and rent finer cars as a result. These
extra expenses come to be regarded as the ordinary perquisites of business travel. A
plausible explanation of that policy appears to be that it is just too costly to determine
what level of expenses actually is reasonable for any particular business trip and to
limit employees accordingly.

A l\ larxian View of Efficiency Wages

A Marxian version of this same theory has been suggested by Samuel Bowles.
According to Bowles, capitalists will tend to invest excessive amounts in monitoring
workers because increases in the probability of detection p make it possible to pay
lower wages w while still satisfying the constraint that employees are motivated not to
cheat. Although wages paid to workers represent a cost to the firm, they do not
represent wasted resources for society; they are merely a transfer from capitalists to
laborers. On the other hand, monitoring expenditures are not a mere transfer.
Monitoring consumes real resources, including supervisory time, trips to the field,
record keeping, and so on. Indeed, the factory system itself, which brings together
workers to do their jobs in a common facility when the work might be performed at
home, has sometimes been explained by Marxian historians as a wasteful system that
emerged to improve the monitoring of workers by their capitalist employers. If higher
wages were paid to workers , Bowles argued, then fewer resources could be devoted to
mon itoring and the resources saved could be used to increase the production of
valuable goods. Therefore, capitalist modes of production are inefficient, utilizing
excessive resources for the task of monitoring workers.
Bowles's argument has some interesting elements that merit further attention.
At its core, the argument is founded on the correct idea that issues of distribution and
efficiency cannot be completely separated. Nevertheless, this Marxian argument is
too i ncomplete to be really convincing. Recall that efficien cy wages provide incentives
by paying a high relative wage, that is, a wage that is high relative to other market
opportun ities. Even in a socialist economy, it is not possible for all workers to receive
a high relative wage, that is, a higher wage tha n the a verage for the economy unless, Rents and
as suggested ea rlier, there is unemployment or highly limited labor mobility. A Efficiency
uniform increase i n wages for workers in the eco nomy would transfer income from
the owners of ca pital to the workers, but it would also lea ve relative wages uncha nged
and so wo uld not reduce the need for mo nitoring. I ndeed, if socia lism were to bring
with it a greater co ncern for income eq uality among workers, leveling the wages across
industries, then it would req uire more monitoring in the old high-wage industries
than the capitalist system it re placed.
M oreover, socialism is not j ust about how the be nefi ts of productio n are divided
but also about who should own the mea ns of production. As we have alrea dy seen,
private ownersh ip of productive assets ca n sometimes ha ve enormous adva nta ges, and
these are not taken into account in the Marxia n argume nt j ust described.
Despite these weakne sses, Bowles's analysis does establish that Marx's claim that
expenditures on supervisory personnel by ca pita lists are partly motivated as a wa y of
transferring i·ncome from workers to ca pitali sts, and this has some warrant in modem
economic theory. It is also importa nt because it opens some commo n gro und for a
dialogue between modem eco nomic theory and Marxia n theory.

Additional Asp ects and A pplications of Efficiency Wage Theory

The efficie ncy wage model that we have described analyzes the situation of a particular
principal and agent, without regard to how the agent is chosen or h ow ince ntives vary
over the agent's career. In this section we offer several examples to illustrate some of
the releva nt issues.
AD..\1\1 SMITH'S ACCOl;NT OF MERCHANT HONOR Adam Smith empha sized the role that
the frequency of commerce had on a mercha nt's ince ntives. H is account matches
nicely with the compa rative statics exercise we di d earlier, focusing on how the
variations in N among countries affects ince ntives and behavior of the ir merchants:
Of all the nations in Europe, the Dutch, the most commercial, are th e
mo st faithful to their word. Th e English are more so than the Scotch, but
much inferior to th e Dutch, and in the remote parts of this country they
[are] far less so than in the commercial parts of it. This is not at all to be
imputed to national character, as some pretend; th ere is no natura l rea son
wh y an Englishma n or a S cotchma n should not be as punctua l i n
performing agreements as a Dutch ma n. It is far more reducible to self­
interest, that general princi ple which regulates the actions of every ma n,
and which lea ds men to act in a certa in ma nner from vi ews of adva ntage,
and is as deeply implanted in an English ma n as a Dutchma n. A dealer
is afraid of losing h is character, and is scru pulous in observing every
engagement. When a person ma kes perha ps 20 contracts a day, he ca nnot
ga in so much b y endea voring to impose on his neighbors, as the very
appeara nce of a cheat would make him lose. When people seldom deal
with one another, we fi nd that they are somewhat disposed to cheat,
because they ca n gain more by a smart trick tha n they ca n lose by the
inj ury wh ich it does to their character . 8
According to Smith , th e greater a merchant's volume of busi ness, the greater
the incenti ve to act honestly in orde r to protect his or her va luable business reputation.

k Adam Smith, Lectures on Justice, Police, Revenue, and Arms, Edwin Cannan, ed . ( New York:
Augustus M . Kelley, 1 964).
 258
Efficient REPUTATIONS AND THE MACHRIBI TRADERS The cost of providing incentives using
Incenti\'es: reputations also depends on the other opportunities that the parties may have. An
Contracts and agent whose outside opportunities would remain good even if caught cheating has
Ownership relatively little to lose from cheating, so motivating that person tends to be costly. On
the other hand, an agent \\'hose business opportunities lie with a small group of
principals who communicate among themselves must worry that actions that damage
his or her reputation will result in the loss of a large volume of valuable business.
These agents are more likely to resist temptations to steal or cheat, even if the wage
for the particular job is low. In other words, it is less costly to motivate the�e agents
than to motivate agents with better outside opportunities.
The example of the Maghribi traders cited earlier illustrates this point. 9 The
evolving pattern of agency arrangements among traders in the western Mediterranean
in the eleventh century can be attributed in part to the changing costs of motivating
different kinds of agents. In the ninth and tenth centuries, when the volume of trade
\\'as lower and ..the reliability of communications was poorer, merchants commonly
traveled with their own goods. Relying on a personal agent was hazardous, because
the agent "could easily have disappeared with the capital or cheated in business
conducted in far-off markets where none of his associates had any control. " 10 By the
eleventh century, however, an extensive network existed in the Muslim Mediterranean
by which merchants provided references for one another, so that the loss that any
single merchant stood to suffer from a loss of reputation grew much larger. As a result
of these changes, the merchants found it more economical to rely on o\·erseas
merchant-agents. The merchants in towns A and B would entrust their goods to one
another for sale rather than relying on a personal agent from the originating merchant's
home port. This change may be explained at least in part by the reduced cost of
ensuring the honesty of the overseas agent. An economic analysis that focuses on
resource costs alone cannot give an adequate account of these evolving trade
relationships.
CAREER PATHS IN FI RMS Career paths are another element that was missing from our
simple efficiency wage model. Our simple model gave only minimal attention to the
role of time by incorporating the number of times N that the worker expects to be
employed. In a more complete model-one in which the incentives for loyalty,
honesty, and hard work are provided by the promise of high fu t u re wages-a pattern
of pay that is rising over the worker's career provides incentives more effectively. I\1any
companies have virtually institutionalized a career pattern in \\'hich, for successful
workers, earnings rise throughout the worker's career, perhaps e\·entually reaching a
point where they exceed the worker's marginal product. Early in a worker's career,
when his or her accumulated investment in the company and in his or her career is
lo\\', the worker is placed in a job of low responsibility, where the \\·ork is relatively
more routine and easily monitored and errors can be corrected. After the worker is
better trained, has invested in learning the specific business, and when the reward for
this diligence is nearer at hand, then the worker is assigned to less routine tasks that
may also be more difficult to monitor. We saw earlier in this chapter that high pay
is a substitute for close monitoring, so the relationship between pay and responsibility
is just as expected. (Note that this analysis is complementar y to the explanation of the
pattern of rising wages with experience in terms of screening offered in Chapter 5 and
the bonding explanation given in Chapter 6. )

9
This account is based on /\\'Iler Greif, op. cit .
1 ° C. 1\ 1 . Cipolla, Before the Industrial Remh1 tio11 ( Ne\\' York: Norton, 1980), 198, as quoted by
Crcif.
l'HODl lCT HEPUT\TIONS ,\ND BHAND N.\I\IES We h ave em phasized the applications of Rents and
efficiency wage theory to the incen tives for em ployees, age nts, or business partners to Efficiency
behave honestly. The prin ciples we h ave described h ave broader application than
that, h owever. Ben j amin Klei11 an d Keith Leffler have emphasized th at th in profit
margins arc th e enem y of quality in markets for goods and services, too, because they
tem pt individual producers to cut their quality an d earn higher profits until consumers
recognize th e ch an ge. Th e firms could th en exit the m arket. If th e product is
recognizable by consumers, however, and if the product margins arc high , the value
of staying in th e business can outweigh any sh ort-ru n profits from degradin g the
product quality.
Wh ere competition keeps prices relatively low or where certain products are
replaced so frequentl y th at customer dissatisfa ction with the old pr oduct is irrelevant,
brand - names that cover a wh ole pr oduct line can provide another way to establish a
reputation fo� quality. A h omeowner may buy th is year's model of a B lack and Decker
drill if he or sh e is sati� fied th at last year's model, purchased by neighbors, was of
good quality and if the power screwdriver of th e brand th at th e h omeowner purch ased
two years ago is still working well. Th e reputation value of a brand name leads
profitable firms with established brands to work h ard to maintain quality.

REPUTATIONS AS CONTRACT ENFORCERS

Successful commerce requires that businesspeople be able to count on one another
to h onor their agreements and, as the Adam Smith quotation emphasizes, th e
reputation mechanism h as long been one of the most important mechanisms for
ensuring that contracts· are h onored. In th is section, we examine th is idea in more
detail. We conduct th is part of our analysis using game theory, wh ich makes i t possible
to examine more systematically limitation� of simple incentive systems and h ow some
of th ose limitations can be overcome by more complex and sophisticated arrangements.

The Elementary Theory : Reputations in Repeated Transactions

As di scussed in Ch apter 4, for many kinds of business transactions-th ose that take
place over an extended period of time in particular-conditions arise for wh ich no
plans h ave been made. Transactions can be classified according to h ow decisions are
made in these circumstances. In spot market transactions, there are no long-term
agreements and little likelih ood of significant ch anges in circumstances before th e
contract is executed. When th e unexpected does occur, the parties bargain among
themsel ves about h ow to respond. An alternative to spot markets is to establish a
relational contract, for exam ple, a hierarchy or authority relation i n which a manager
evaluates th e facts and alternative courses of action and th en instructs th e oth er pa rties
on what to do. Sometimes, long-term contracts allow one party to exercise discretion
with out establishing an auth ority relati on. For example, a h omeowner may delegate
som e degree of auth ority to a remodeling contractor to make decisions about h ow the
rem odeling j ob sh ould be done.
Th ese abstract descriptions of contracting practices apply to many concrete
situations. When a person becom es an employee of a firm, he or she typically agrees
to accept whatever direction is given within somewhat nebulous "customary limits"
in exch an ge for a fixed h ourly wage or for a salary. Th e em ployee' s supervisor is the
one with auth ority. Th e supervisor can exercise discretion in assigning tasks, an d the
employee must rely on the supervis or to do so fairly. In our h ome remodeling example,
even if the con tract is unusually precise, it will not specify wh at th e contractor sh ould
do if rain delays the start . of work, or if fixtures that were selected from a catalogue
are no lon ger available, or if th e dimensions on th e drawings contain an error. The
 260
Efficient
Incentives:
Contracts and
Ownership
MacRents:
Efficiency Wages in the Fast-Food Business
An em pirical implication of efficiency wage theory is that a decrease in the
intensity of supervision will result in an increase in the optimal efficiency
wages. The fa st-fo od restaurant business offers a test of this prediction.
The fast-food business in the U nited States involves a num ber of chains:
McD onald's, Burger King, Wendy' s, Jack in the B ox, Kentucky Fried
Chicken, Arby's, and Roy Rogers are am ong the most familiar. Each of these
has large numbers of outlets, some of which are owned and operated by the
parent firm, and som e of which are owned and managed by independent
businesspeople operating under franchise agreements.
The two types of outlets appear very sim ilar. They have the same menus
based on the same recipes, they are architecturally and technologically
sim il ar, they are geographically similar, they draw employees from the same
local work forces, they have similar pricing policies, and so on. H owever,
the franchise units are managed by their owners, who set the wages, fringe
benefi t levels, and work rules and conditions in their individual outlets, make
their own hiring and firing decisions, and claim the profi ts and losses of their
operations. I n contrast, the company-owned restaurants are managed by
salaried em ployee-managers, with wages and benefits being set by district
managers.
A key part of the outlet m anager's j ob is to m onitor an d train the
assistant managers and supervisors as well as the actual n onmanagerial staff
of the restauran t. S tandard incentive arguments would suggest that a
franchisee, who collects the residual profi ts, will be much more highly
motivated to monitor intensively and train effectively than will a salaried
manager. (The fact that franchises tend to be about five times m ore profi table
than c ompany outlets supports this view. ) I n turn, efficiency wage theory
then suggests that the c om pany will choose to pay higher wages to the workers
in its outlets than will franchisees in theirs. The higher wage makes the j ob
more valuable and losing it more painful. This offsets the lower probability
of being fired for cause that results from weaker m onitoring and so reduces
the am ount of surli ness, carelessness, absenteeism, shirking, theft, on-the­
j ob drunkenness, and so on from what they \rnuld otherwise be.
The technology and delivery system s of the fast-food bu siness have
been chosen to minimize the chance that a bored teenage worker can cause
any great problem s. This suggests that the effort the manager devotes to the
supervisors and assistant managers may be es pecially important and that the
efficiency wage differential for these em ployees in the company outlets should
perhaps be relatively higher than the differential for regular em ployees.
These predictions are supported by data from two surveys of fast-food
res taurants (one of \v orkers and supervisors in a sam ple of McDonald's,
Ken tucky Fried Chicken, Arby' s, and Roy Rogers outlets in 1 982, the other
of managers in a sample of McDonald's, Burger Kin g, Kentucky Fried
Ch icken, an d Wendy's restaur an ts in 1 98 5). I n the first survey, em ployees
hired in to a given j ob started at essen tially the same wages in both ty pes of
outlets, hut the wages grew much faster over time for the supervisors in the
franch ise outlets than for th ose in the com pan y- owned stores and somew hat
 26 1
Rents and
Efficiency

more rapidly for nonmanagerial workers. The result was that earnings for
supervisors were a sizeable 8. 9 percent higher i n company restaurants an d a
stati stically (if not economically) si gnificant 1. 7 percent higher for full-time
n onmanagerial workers. The second study examined onl y starting wages, but
the company restaurants were only two thirds as likely to start new em ployees
at the mi nimum wage as were the franchise outle ts. M ore over, company­
owned outlets pr ovi ded more ge ner ous fringe benefits, further in creasing the
pay di ffere ntial. The bi ggest difference in benefi ts was i n the provisi on of free
meals. Pre ve nti ng employees from helpi ng themselves to a free meal is very
diffi cult to control. If it is difficult to induce suffi cie ntly careful mon itori ng
to preve nt employees from steali ng food, i t may be best si mply to permit i t
explicitly.

Based on Alan B. Krueger, "Ownership, Agency and Wages: An Exami nation of

Franchising in the Fast Food Industry," Working Paper 3 3 34, National Bureau of Economic
Research ( 1 990).

homeowner must rel y on the contractor to react appropriately to the i nevi table
sequence of unforesee n conditions that were not specified in the original contract. In
each of these si tuati ons, at least one party (and freque ntly both) must rely on the
honesty and good wi ll of the other.
The di sputes that ari se in these si tuati ons are ofte n connected with ambiguity
about what ki nds of discreti onary behavi or are hone st or appr opriate. To start analyzing
the problem, however, let us suppose that although it is not possible to specify what
to do in any pa rticular situation, it is possible for pa rties close to the tra nsaction to
determine whether the person with authority has done "the right thing. " It is thi s two­
part assumpti on that characterizes our "elementary the ory. " The assumption implies
that we are exami ni ng a si tuation in whi ch detailed contracts cannot be wri tten,
maki ng legal enforcement difficult, but in which the parties themselves have e nough
i nformati on to evaluate each other's past behavi or, which is a basic requirement of
any system of reputati ons.
OFFERING TRUST We use the box shown below to descri be abstractly a transaction
opportunity that may occur repeatedly. At each round-that is, each time the
transaction opportunity arises-one party, called the offeror, can offer to trust a second
party, called the decision maker. The formal structure of thi s i nteraction is represented
in the box, which describes a normal form game with two players. The box shows
that each player has two strategies at each round. The offeror's strategies are to offer

Decision Maker
H onor Don't
Offeror Trust Trust
Offer Trust V, V - L, V + G
Don't Offer 0, 0 0, 0
 262
trust or not. The decision maker's strategies specify what he or she plans to do if trust

of strategies. If trust is offered, the decision maker must then decide whether to honor
Efficient
Incenti,·es: is offered. The entries in the box show what the payoffs will be for each combination
Contracts and
Ownership that trust by behaving fairly. The box describes the parties' payoffs. The first number
in each cell of the box describes the offeror's payoff, whereas the second describes the

payoff of zero. If trust is offered and honored, then both parties receive a valuable
decision maker's. If no trust is offered, no transaction occurs and both parties earn a

payoff of V. However, there is a short-run temptation for the decision maker not to
honor trust; by not honoring, he or she receives an extra payoff of C in that round.
In that case, the offeror suffers a loss of L and would have fared better if he or she
had not offered trust and had instead settled for a payoff of zero. We assume, of
course, that V, C, and L are positive numbers.
It is important to interpret the numbers here correctly. Zero is simply a value
that we have assigned arbitrarily to whatever the two parties might do if they do not
share trust-the number itself has no significance except as a baseline from which
other values are measured. That the value V is positive means that honored trust is
more valuable to both parties than whatever happens if trust is not offered at all. The
larger V is, the more the parties have to gain by offering and honoring trust. That C
is positive means that the decision maker has something to gain in the short run by
dishonoring trust. For example, if the decision maker is shamed by behaving
dishonorably, the other gains are large enough to offset this shame. That L is positive
means that offering trust entails a risk: The offeror has something to lose.
Suppose now that this normal form game were to be played only once. According
to game theory, which posits that each player is self-interested, we predict that the
offeror will not offer trust and the decision maker will plan not to honor trust in case
it is offered. The self-interested decision maker, if given the opportunity, would surely
prefer to earn V + C by dishonoring trust than to earn just V by honoring it. The
offeror, recognizing the decision maker's incentives and knowing that he or she is
self-interested, would therefore choose to avoid the loss of L by not offering trust. The
result is that each party receives a payoff of zero, missing the opportunity to obtain
payoffs of V.

REPEATED DEALINGS AND THE NASH EQLTILIBRiml Despite this unfortunate outcome,
the game-theoretic solution does seem to be a reasonable description of what might
happen if the situation we have portrayed were to arise among strangers who meet
only once, especially if the temptation C and the potential loss L are large. If the two
players are people who encounter one another in this sort of circumstance frequently,
however, they might be expected to develop some way to overcome the problem of
trust and enjoy the fruits of the honored trust outcome. For example, the decision
maker might say to the offeror: "I would like you to trust me once to test my honor,
and you would be wise to do that. After the test, base your expectations about me not

on how I prove myself in action. If you do offer trust, I promise always to honor it.
on some arbitrary, pessimistic theory about how a stranger might play this game, but

We both know that if I should ever fail to honor your trust, you would never trust
me again. For that very reason, it will actually be in my interest, as well as yours, to
honor your trust, for only by doing so can I earn your continuing trust, which I have
good reason to value. "
Is this speech believable? To check its logic, suppose that earning a payoff of X
every time the parties meet in the indefinite future has the same value for the decision
maker as docs a one-time payoff NX today. The magnitude of the number N depends
in part on how frequently the parties expect to m eet and on the interest rates that
they face. 1 1 Wha t ca n we expect to happen if the offeror believes the self-in terested
decision maker's speech a nd acts accordingly? If the decis ion maker docs not h onor
Rents and
Efficiency
the offcror's trust, the decision maker ga ins G immedia tely when he or she cheats bu t

period, he or she will earn a payoff of NV ins tead. If NV > G a nd each party expects
will never be trusted in the future. If the decis ion maker does h onor trust in each

the other party to behave as described, the n neither party could do better by devia ting
from the prescribed behavior. The offeror ca nnot ga in by wi thh olding trust: He or
she expects the trust to be h onored and therefore expects a payoff of V per period by
offering trus t, compared to O per period if the trus t is witheld. The decision maker
cannot gain by dishonoring trust: He or she expects a pa yoff of V per period, which
is valued a t NV if trust is h onored in each period, but onl y G < NV if trus t is
dish onored today, for then trus t will never again be offered. In the language of game
the ory, this situa tion in. wh ich no party ca n gain by making a unila teral devia tion
from the pre�cribed behavior is called a Nash equilibrium.
For a pa ttern of behavior to be a Nash equilibrium, it means tha t if the parties
all expect this pattern of behavior, then it is in the individua l interes ts of each to
adhere to the pattern. I n many situa tions, es pecially ones where people have little
experience, there is no reason to suppose tha t pe ople will have the same e xpecta tions
about behavior or that their expectations will be correct. If ( 1 ) all the players in the
game do ha ve the same expectations, (2) th ose expectations are correct, a nd (3) the pla y­
ers act in their individual best interes ts given their expectations, then the combina tion
of stra tegies is a Nash equilibrium.
One important difficulty with using a Nash equilibrium to predict social beha vior
is that there ma y be more than one. In the game we have portrayed, a second Nash
equilibrium strategy combination is for the offeror never to offer trus t a nd the decision
maker never to honor it. A third possible Nash equilibrium is for the decision maker
to promise to cheat only every second time that trust is offered and for the offeror to

rounds. If L < V and the rounds are frequent, neith er pla yer can ga in by changing
offer trust a t every round as long as the decision maker never chea ts in two successive

his or her stra tegy, so th is stra tegy combination is a lso a Nash equilibrium. The
hypothesis of Nash equilibrium rules out some patterns of behavior but does not
always lead to a unique prediction about what the outcome will be.
REPUTATIONS Let us return now to the first Nash equilibrium that we described, in
which the offeror continually offers trust for as long as the decision maker continues
to honor it. We studied this situa tion ass uming that the offeror a t each round is the
same pers on, but tha t is not really necessary for the a nalysis. If the decision maker
faces a series of different offerors wh o each offer trust only if the decision maker
honored trust when it was last offered, then the decision ma ker's calcula tion about
whether to h onor trus t is exactly the same as if there were a single offeror. One migh t
say that in each transaction, the decision maker h onors trus t in order to encourage
future trading partners to offer trus t or, in other words, to maintain his or her
reputation for honesty. In the world of business, a reputa tion for h onesty ca n be
valuable because it ca n attract trading partners. In addition, if it is poss ible but costly
to write detailed contracts, a good reputa tion ca n often allow the decision maker to
avoid that expense as well as the use of costly a nd error- prone lega l contract
enforcement mechanisms.

11 If the game is to be played an infinite number of times, the interest rate that players can earn on
their investments is r per period, and the players meet at intervals of t periods, then N = 1/[(l + r)'
- I ] + I. Lower interest rates (smaller r) or more frequent meetings (smaller t) lead to larger values for
N. Sec Chapter 14 for additional details of how to make present-value calculations.
 264
Efficient
Incentives:
Contracts and
Ownership
Trench Warfare in World War I:
The "Live and Let Live" System
In economic applications, we usually emphasize the financial returns that
accrue to earning others' trust. Sometimes, the most significant returns are
not financial at all. A remarkable example of valuable trust among enemies
is the "live and let live" system that developed between battalions of opposing
armies along the trenches of the Western Front in World War I. Although
the commanders of the opposing German and Allied armies wanted the
troops to shoot to kill, the soldiers often behaved differently. Trust and
reputations explain how this came about.
In the trench warfare of WWI, the same opposing battalions faced each
other for long periods of time. Deadly fire from one side could be reciprocated
by the other side. Although many of the soldiers in opposing battalions in
any war would prefer to avoid exchanges of deadly fire, they cannot trust the
other side to withhold its fire because each side has a strong interest in
destroying the enemies' fighting capabilities. In WWI, however, the daily
exchange of fire between entrenched battalions in a war of attrition offered
a rare opportunity. Daily interactions could make it a Nash equilibrium for
the two sides to fire in earnest only if fired upon. To achieve this outcome,
each side had to demonstrate to the other its ability to retaliate if the other
side initiated deadly fire, as well as ·its willingness not to use deadly fire
without provocation.
In many parts of the front, the soldiers did achieve such a truce while
still making it appear to their commanders that they were shooting at the
enemy. According to Robert Axelrod's account, "During periods of mutual
restraint, the enemy soldiers took pains to show each other that they could
indeed retaliate if necessary. For example, German snipers showed their
prowess to the British by aiming at spots on the walls of cottages and firing
until they had cut a hole. Likewise, the artillery would often demonstrate
with a few accurately aimed shots that they could do more damage if they
wished. "
Eventually, the Allied commanders uncovered these tacit truces and
took steps to end them. They initiated a series of commando raids on the
German trenches that brought retaliation from the Germans and more
retaliations by the opposing Allied soldiers, until the accumulation of fire
made the "live and let live" system untenable.

Source: Robert Axelrod, The Evolution of Cooperation (New York: Basic Books, I 984) .

. \rn biguity, Complexity, and Limits of Reputations

Although it may not be possible to specify in advance in an enforceable contract ho\\'
the decision maker should behave, it may nevertheless be possible for those involved
to decide afterward whether the parties had behaved honorably. Th is assumption is
sometimes problematic, however. In many cases, if it really were easy to decide what
constitutes honorable behavior, then it would be easy for courts to settle disputes at
low cost.
 265
In reality, perceptions of circum stances frequen tly c ome in to conflict, an d Rents and
parties differ about j ust what is " the right an d honorable thin g to do." Even when Efficiency
parties agree about the circumstances, they may disagree about what can be done or
what is likely to be effec tive. For many kinds of decisions, it is n ot possible to try out
each alternative to see what the consequences would have been .
The pr oblem we have descri bed is compounded if there is more tha n one party
offerin g trust. As difficult as it may be for a single party to know whether another has
acted fairly, it is even more difficult for an outsider to the tran saction to assess what
has happened when there has been a dispute. In a simple reputation system, that is
preci sely the task that faces a whole series of outsiders to the transaction. The c ost
an d difficulty of making those j udgments, and the need for them to be made repeatedly
by a series of outsiders undermines the effectiveness of a system of trust based on
reputations alone. Various institutions and practices have arisen to restore some of
the lost effecti ve ness.
CORPORATE CULTURE O ne way to enhance the effectiveness of a sy stem of reputations
within a group is to construct or evolve a set of workable principles and routines that
create shared expectations for group members. In a large organization, the principles
would help guide managers in making decisions as well as providing a set of clear
expectations for every one in the organization. S ometimes, routines evolve with no
far-reaching purpose in mind, but simply represent patter ns of action that have worked
well in the past. In other instances, the principles take the form of explicit rules
promulgated with a clear purpose in mind. For example, the San Francisc o Forty
Niners football team has a rule that a player in the starting lineup who bec omes
inj ured and misses some games is always entitled to return to his starting role upon
rec overy, regardless of how well the substitute plays in his absence. The rule is
designed to keep inj ured players from c oncealing inj uries. The team surely faces
temptations to make "exceptions" to the rule-especially when the substitute plays
very well-but exceptions would undermine the player s' trust in the coaches. Without
the rule, the notion of fair treatment would be far more ambiguous. What is the "fair"
thing to do if the substitute plays especially well?
At N ordstrom, the department store chain, one rule is that merchandise returned
by customers is always cheerfully accepted without requiring proof of purchase. To
teach this rule, a story is told to trainees of a salesperson accepting without question
a customer's retur n of an old, badly wor n set of automobile tires, even though
Nordstr om has never sold tires! The story helps to establish a rule that every one can
understand and makes clear to trainees the str ong commitment to customer satisfaction
that is a hallmark of N ordstr om's operations.
From this perspective, corporate culture is the set of routines for decision making
and shared expec tations that employees are taught and the stories and related devices
used to con vey those expectations. In Chapter 4 we emphasized the role of routines
and shared expectations for c oordinating action within the organization. H ere we
em phasize a different function: The culture pr ovides a set of principles and procedures
for judging right behavior and resolving inevitable disputes. Because the principles
have to be simple to be communicated effectively to newc omers to the organ ization
and because they have to be adapted to work in the typical circumstances that the
organization faces, no one culture can work well for all organizations. This helps
make some limited economic sense of the "clash of cultures" explanations that are
often cited as a reason that two organizations with very different styles or histories
ex perience difficulty when they try to merge their operation s.
REPUTATIONS AND THE LAW Robert Ellickson has taken the culture argumen t even
further, claiming that even in modern times, in stable c ommunities, the need to
 266
Efficient maintain a good reputation is more powerful than are legal sanctions in settling
Incentives: community disputes. 1 2 He reports that in Shasta County, California, a ranching and
Contracts and agricultural area, the principles actually used to compensate for damages caused by
Ownership stray livestock, for example, are not those imposed by the relevant laws, but instead
are ones that were evolved over generations in the local community. There is no legal
enforcement of the rules of neighborly relations-just the perceived need by residents
to maintain a good, trustworthy image in the local community.
In fact, most contractual disputes never reach a court of law. The parties work
out informal ways to resolve their differences, and these mechanisms are supported
at least as much by reputation considerations and by the expectations of future gains
as by the threat of legal action.
THE "END GAME" PROBLEI\I A major constraint on reputational enforcement mecha­
nisms is that the horizon over which the relationship is expected to continue must be
relatively long if the value of the reputation is to exceed the gain from cheating. We
saw this in ..extreme form in the previous game: If there is only a single interaction,
then trust will not be extended because it is sure to be violated. Thus, if there is a
last time the parties will interact, and the parties know they have reached this point,
they will fail to achieve the gains that are potentially available. Moreover, if they
know they are approaching the end of the relationship, the returns to maintaining a
reputation are seen to be small and may not be enough to prevent trust from being
violated. In these circumstances, other mechanisms may have to come into play if
inefficiency is to be avoided.
One possibility is to move to more formal incentive mechanisms. For example,
it has been suggested that concern with their reputations and their consequent career
opportunities is a major check on managers' temptation to behave opportunistically.
As a manager approaches retirement age, however, these career concerns ought to
weigh less heavily. Thus, it may be advisable to increase the use of explicit incentive
pay for managers as they approach retirement. A recent study by Robert Gibbons and
Kevin Murphy found exactly this pattern in the compensation of a large sample of
corporate chief executive officers (CEOs) in the United States. n Total compensation
of the CEOs in this sample became more responsive to their firms' stock market
performance as they neared their retirement ages.
Another kind of solution to the end game problem is the sale of businesses.
The founder of a business, nearing retirement, finds it valuable to maintain the
reputation of the business to attract a higher price from potential buyers. This can
lead the business to act as if it had a much longer horizon than its founder or its
executives.
The end game problem is also commonly experienced by companies in
bankruptcy. Bankruptcy attorneys commonly observe that the customers of bankrupt
businesses claim that the goods delivered in the last days of operation are of inferior
quality and refuse to pay full price for them. With the bilateral relationship nearing
its end, attitudes toward payment harden, and disputes to be resolved in court become
more likely.
The Advanced Theory : Reputations Aided by I nstitut ions
One of the ways that people enhance the effectiveness of a system of reputations is
by narrowing the range of people with whom they do business. Frequent transactions

12 Robert Ellickso11, "Of Coase and Cattle: Dispute Resolution Among Neighbors in Shasta Cou11ty,"
Stanford Law Rel'iew, 3 8 ( February 1986), 623-87.
1 1 Robert Cibbo11s a11cl Kevin l\lurphy, "Optimal Incentive Contracts in the Presence of Career
Concern�," mimes, University of Rochester (1990).
 267
if they arc all of similar magnitude allow trust to fl ourish. In modern societies, Rents and
however, people have specialized occupations and spec ialized skills . M any specialis ts, Efficiency
such as plumbers, furn iture re pairers, or automobile salespe ople, in teract with
in dividual customers relatively infrequen tly. Buyers may purch ase frequen tly from the
same sources, but the availability of competing suppliers still creates im pers onal
transactions that arc hardly be tter than the outside altern atives.
Business an d legal institutions of various sorts can help to fi ll the gap. Legal
institutions replace the system of reputations altogether. Parties wh o rely on the legal
system c oun t on the threat of a lawsuit, rather than that of a bad re putation, to ensure
full c ompliance with the con tract's terms. The legal system, h owever, has many
disadvan tages for con tract enforcement. Because it is a general system, it relies on
general rules that may be poorly tailored for the partic ular in dustry where the dispute
arises., Legal procedures are often cumbers ome, time consuming, and expensive.
Furthermore, legal rules based on h istorical precedents may be unres ponsive to
changing techn ologies and other chan ging realities. Judges and j uries may often lack
the expertise to evaluate industry disputes based on technical issues well en ough to
apply legal rules appropriately. In in tern ational disputes, courts sometimes lack the
auth ority to enforce their decisions. Also, through out h istory, c orrupt an d biased
j udges an d outside political infl uences h ave added to the costs and the apparen t
randomness of court rulings.
For all these reas ons, private insti tutions h ave frequently been more important
than legal ones for establishin g standards of behavior, ensuring con trac t c ompliance,
and res olving disputes. Often, the private insti tutions work by buttressin g the reputation
system itself.
FROM MEDIEVAL PRIVATE JUDGES TO MODERN RATING AGENCIES In Europe, before
the rise of the nation-state, merch an ts developed a system of private laws-the lex
mercatoria-and employed private judges to res olve disputes. The merch an t law often
differed from the generally prevailing church law on the same issues. Merch an ts at
the time depended on one an other for many services, such as security wh ile traveling
or assistance in han dlin g goods in foreign ports. G ood standing in the c ommun ity of
merchan ts was essen tial to conduc tin g business. The system of merchan t law provided
an evolving standard of behavior that was more responsive to merchan t needs than
was the ch urch law. The system of private j udges making pron ouncements about
disputes greatly simplified a th ird-party merchant's problem of in terpretin g a dispute
between two other merchan ts to reach a j udgment about their respective re putations.
The crucial tasks of deciding wh o is righ t and com municating to other merchan ts the
iden tities of offenders were handled effectively by this system. 14
Through out h istory, many devices h ave been used to inform the public when
a merchant is no longer in good stan ding. In colon ial America, the stock and pillories
were used to make a public dis play of merchan ts wh o violated commercial law. In
the eighteen th and nineteenth cen turies, when whalers from differen t c oun tries met
at sea, they would h ave a party called a "gam, " where the captains of the vessels would
exchan ge information about many th ings, including disputes with other whalers that
needed to be settled. Whalers in the Un ited S tates, concen trated as they were in a
few seaside commun ities, needed n o such prac tices to keep informed about one
an other. The whalers' system of commun ity enforcement of n orms based on repeated
in teractions was so e ffective in its early years that there are n o rec orded cases of
property disputes amon g whalers goin g to c ourt, even th ough the issues of wh o owns

1 4 See: Paul Milgrom, Douglass North, and Barry Weingast, "The Role of Institutions in the Revival
of Trade: The Medieval Law Merchant, " Econom ics and Politics, 2 (March I 990), 1-23.
 268
Effic ient a whale are quite subtle. 1 5 (Should it be the first to sink a harpoon? The first to secure
Incentives: a flag in the whale? Can a ship harvest a whale that has been killed and flagged but
Contracts and is floating free or must it allow the killing ship to come make the harvest?) Only in
Ownership the late nineteenth century, when petroleum products began to replace whale products
did the legal system become involved in whaling disputes. At that time, rents in the
whaling industry began to decline and time horizons for whalers became shorter­
effects that combined to undermine the community-based system of reputations.
In modern times, the local Better Business Bureau will mediate complaints by
customers against local businesses and report information about complaints against
local merchants and craftsmen. Similarly, credit bureaus report information about
past loans and whether they were properly repaid, and consumer agencies report
survey information about various matters, such as how satisfied consumers are with
the handling of their insurance claims at various companies. Disseminating this
information strengthens incentives for reliable behavior. Merchants are motivated to
settle complaints made through the Better Business Bureau to preserve their good
reputations. People are more eager to repay loans knowing that a faulty credit rating
will prevent them from borrowing again on favorable terms. A similar logic applies
to insurance companies.
Bon:OTIS AND E1'1BARGOES : INTERNAL DISCIPLINE The main characteristic of institu­
tions that work by enhancing the reputation system is their reliance on sanctions
delivered by individuals. If applying sanctions is costly, some individuals may refuse
to comply, undermining the effectiveness of the system.
Medieval merchant guilds in northern Europe were bodies made up of all the
foreign merchants trading in a single town. 16 A major difficulty these traders faced
was that the cities in which they traded did not always carry out their contractual
agreements, for example, to protect the traders against theft by locals. The only
significant response that the merchants had was to refuse to trade in the offending
town, but this ran into the free-rider problem noted earlier. In the major trading
centers of northern Europe, especially Bruges, the trading merchants came from many
different cities, and this made enforcing embargoes especially difficult. Moreover,
towns faced with an embargo would try to encourage traders to free ride by offering
them especially attractive deals. The guilds can be seen as an institution through
which merchants coordinated their responses to cities that reneged on trading
agreements.
In 1 280, following a dispute over Bruges's responsibility to provide physical
protection for the merchants, the guild of merchants trading in the city attempted
embargo, transferring their trade to Aardenburg. The embargo, however, was defeated
when Bruges offered special terms to merchants from various towns who found the
absence of competing traders to their liking. In response to such failures, the town
guilds united under the new, more encompassing organization-the Hansa, or
Hanseatic League. If the Hansa declared an embargo on a city and traders violated
the embargo, then their home city itself might be put under embargo or the offending
traders might be refused access to goods from other member towns. Following another
dispute, in 1 3 5 8 the Hansa endorsed another embargo of Bruges, and the city
responded again by offering special terms to Hansa member Cologne and nonmembers

1 1 Robert Ellickson , "A Hypothesis of Wealth-Maximizing Norms: Evidence from the Wha ling
Industry," Journal of Law, Economics, and Organization, 5 (Spring 1 989), 83-97.
16 These guilds should not be confused with the more familiar crafts gu ilds, which were organizations
of all the producers of a particular good in an area, for example, the weavers guild in London .
 269
such as Kampen. With its strong internal discipline, however, the embargo was Rents and
eventually successful and Bruges honored the trade charter. 17 Efficiency
There is a danger of misinterpreting this historical episode as a mere distributive
dispute, in which one party's gain is the other's loss. Actually, institutions like the
Hansa which help to enforce agreements perform an important efficiency-en hancing
function that helps to account for their emergence and survival. From the host city's
point of view, the very ability to renege on agreements is also an inability to com mit
to honest behavior. That inability can be very damaging. In 1 28 3 King Edward I of
England, reflecting on the broken promises of security for foreign merchants in his
country, observed that "many merchants are put off from coming to this land with
their merchandise to the detriment of merchants and of the whole kingdom. " 1 8 In
modern times, as in history, people can find it valuable to establish arrangements,
practices, and institutions that narrow their options or enable others to enforce
contracts against them because those very arrangements make them a more reliable
and trustworthy business partner.

RENT SEEKING, INFLUENCE COSTS,

AND EFFICIENT DECISION ROUTINE

Rents and Quasi -rents

So far, we have developed the argument in primarily nontechnical language, speaking
about the economic profits of a firm or about the high wages of the worker. In
economics journals, however, the general principles are most frequently couched in
terms of rents and quasi-rents. A rent is the portion of earnings in excess of the
minimum amount needed to attract a worker to accept a particular job or a firm to
enter a particular industry. For example, a worker who ranks j obs only on the basis
of wages and who is offered a job at wage rate w earns a rent of w - w i f w is the
highest wage he or she can earn in any alternative employment. Similarly, if a price
of {, would be just sufficient to attract a firm to produce in some market, and if the
quantity the firm sells is q , then the firm earns rents of (p - p)q , provided the actual
price p that it can charge is greater than {,. Notice that the price {, is the same as the
average total cost of operating in the industry, including the opportunity cost of capital
as well as the costs of all the other fixed and variable factors used by the firm. Rents
typically arise because of scarcity, whether natural or induced. In 1 990, for example,
baseball pitcher Roger Clemens was given a contract calling for him to be paid $2 1 . 5
million over a period of four years, in each of which he would probably not pitch in
many more than 40 games. There is surely a large element of rent in this pay.
Clemens receives this rent because his talent is in very short supply. In turn, his
employers, the Boston Red Sox, can pay him this much because they are earning
rents from their exclusive franchise to present major league baseball in New England.
A quasi-rent is the portion of earnings in excess of the minimum amount needed
to prevent a worker from quitting his or her j ob or a producer from exiting its industry.
Whereas rents are defined in terms of decisions to enter a job or an industry, quasi­
rents are defined in terms of the decision to exit. If the wage w would be just sufficient
to keep an employee at work at a particular job, taking into account such costs as

17 See Avner Greif, Paul Milgrom, and Barry Weingast, "The Merchant Guild as a Nexus of

Contracts, " unpublished working paper, Stanford University ( l 990).

18 English Historical Documents, David C. Douglas, ed. (Oxford: Oxford University Press, l 95 5),

420.
 270
Efficient those of searching for a new job and those of acquiring any needed new skills, then
Incentives: the employee's quasi-rents are w - w . The employee will exit (quit and seek a new
Contracts and job) precisely when the quasi-rents are negative. He or she will value the job and
Ownership want to keep it when the quasi-rents are positive. Similarly, in the standard theory of
markets, a firm will exit an industry only if the price does not cover its average variable
costs, which we may call p. If the actual price p is higher, then the firm earns quasi­
rents of (p - p)q on its output q.
To understand the difference between rents and quasi-rents, let us compare the
entry and exit decisions for the firm. Because the price p that is just sufficient to make
entry profitable is equal to the average total cost, whereas the price p that is sufficient
to make exit unprofitable is just the average variable cost, it must be true that p :5 p.
That is, it is possible for a firm to earn quasi-rents even when the initial decision to
enter the market leads to only normal profits and even when it yields less than normal
profits. The difference between ren ts a nd quasi-rents a rises from the presence of costs
tha t m ust be incurred to enter the market but that cannot be salvaged by an existing
fi, r m tha t chooses to exit. For this reason, quasi-rents are always at least as great as
rents.
Rents and quasi-rents are useful for analysing different kinds of decisions. In
the theories described in this chapter, it is quasi-rents-what the agent stands to lose
if forced to exit-that plays the central role. Although rents can exist only fleetingly
in a competitive economy, quasi-rents are much more common. They are created
whenever specialized (nonsalvageable) investments are made and so have the potential
to be widely useful for providing incentives.

Rent Seeking in the Public and Private Sectors

We have seen how rents can help support efficiency, but their presence can also
create incentives to expend resources attempting to reallocate the rents. This is typically
a pure cost with few or no counterbalancing benefits.
Governmental decisions to grant monopolies, determine rates for utilities, or
establish tariffs or other trade barriers can create rents or quasi-rents for firms. Firms
attempt to capture these rents for themselves, rather than having them go elsewhere.
They do so both by participating legitimately in the political and regulatory processes
and, sometimes, by paying bribes. The result is a costly public policy problem, both
because the attempts sometimes lead to distorted decisions and because so much
valuable time and energy is spent in a process that results only in transfers, not in
actual social gains. Activities that serve no social function other than to transfer rents
or quasi-rents have been called rent seeking and directly unproductive profi,t seeking
(DUP) when they occur in the public sector, and the costs of resources used and
decisions distorted are infiuence costs.
According to some analysts, inflvence costs are one of the largest costs of big
government. When there are many decisions being made and each has large
redistributive consequences, then many of the brightest and ablest people in society
will find it most remunerative to spend their time trying to influence those decisions,
either pursuing their own individual self-interests or those of their clients. This causes
many of the inefficiencies of public decision making. Resources are expended by
private parties (often through their elected representatives) solely to affect the distributive
impacts of public decisions, as when huge lobbying efforts are mounted to win a cable
television franchise, and efficiency is sacrificed for distributive or political ends, as
when a federal facility is sited in a key congressman's district rather than where it will
do the most good. Direct governmental expenditures arc just the tip of the iceberg:
The resources expended by others to influence judges and juries, legislators and
regulators, inspectors, auditors, standards boards, purchasing agents, tax authorities,
 271
and other governmen t em ployees may far exceed those expended by the government Rents and
itself on making and even on im plem en ting the decisions that are influenced. Efficiency
The reason the size of governmen t en ters into the analysis is that larger
governmen ts have larger agendas, m ore decision m akers, and m ore issues to resolve.
The opportunities for influencing governm ental decisions grow along wi th the size
and sc ope of government. This does not mean that an archy is the optimal form of
government; the presence of c osts does not deny the possible existence of still larger
benefits. M oreover, private sec tor decisi on making m ay sometimes reflect too narrow
a set of in terests, so that only the public sector can fairly and efficiently determine
some matters.
The notion that factional politics is costly is hardly a new one. James M adison
wrote in The Federalist two centuries ago, " the causes of faction cannot be rem oved,
and . . . . relief is only to be sought in the m eans of c ontrolling its effects. " 1 9 The
economic theory of organization s offers a set of principles to evaluate alter native
means. It all�ws people to decide in a discrim inating way what kinds of organizations
are best suited for what kinds of decisions, depending on the nature of the in terests
that need to be resolved, and what kinds of deci sions ought not to be m ade at all by
any central authority.
QUASI -RENTS AND INFLUENCE ACTIVITI ES IN ORGANIZATIONS Scholars have paid far
less attention to the influenc e c osts that are incurred within firm s, unions, trade
associations, and other private- sector organizations, largely because classical econom ic
theory denies the existence of any rents within those organizations that can be
appropriated by influencin g decisions.
In the classical theory, there are no good j obs and bad j obs. The wage that
workers are offered is just the sam e as they could earn in any other sim ilar j ob in that
firm or any other firm . Given any two j obs that require the same qualifications, if
one j ob were un iversally regarded as m ore pleasant or a better route to future
prom otions, then the only way to attract workers to apply for the bad j ob would be
to pay a higher wage. Thus, if being a steeplej ack is dangerous, i t will pay m ore than
some safe j ob, and if being a garbage collector i s unpleasant, i t will comm and premium
wages.
If wages were really determ ined in this way, there would be no problem of
organizational politics. People would not care much about decisions m ade by
employers because any change that affected em ployees would be exactly c om pensated
in the wage paym ents they received. I n reality, however, employer decision s do affect
the em ployees' welfare. There are quasi-rents within organization s, and organizational
politics can consume real resources as each party battles for a larger slice of the
organizational pie. We introduced this idea som ewhat inform ally in Chapter 6, but
here we give it a much fuller development.
For organizations that are already form ed, quasi-rents are created whenever
employees are call ed upon to m ake specific investmen ts in their j obs. In the efficienc y
wage theory discussed earlier, firm s may be led to pay higher- than-market wages to
motivate workers in j obs where good performance is especially im portan t and
monitoring is difficult. Firm s m ay offer higher- than-market wages to attract applications
from superior workers or to reduce turnover am ong key em ployees or experienced
workers who have received specialized training or developed special customer or
supplier relationships. Rents or quasi-rents may be created when firm s train workers

19
From essay IO by James Madison in Federalist Papers: A Collection of Essays Written in Support
of the Constitution of the United States, by Alexander Hamilton, James Madison, and John Jay (New York:
New American Library, 196 1 ).
 272
in skills that are of value to other employers because the trained workers will later be
able to demand higher wages or to move to a new employer for a higher wage. 20
Efficient
Incentives:
Contracts and One effect of all these rents is to make individual workers care about their j ob
Ownership assignments within firms. There will be good assignments and bad assignments, "dead
end" j obs and "fa st-track" j obs, positions in which pay is high, and others in which
it is much lower. S omeone will be assigned to Paris, Texas, and someone to Paris,
France. H owever, it is not only the j ob assignment decisions that affect employee
welfare.
THE VARIETI ES OF INFLL1 ENCE ACTl\'ITIES Two conditions are necessary to make
influence costs likely. First, a group of decisions or potential decisions must be made
that can influence how the benefi ts and costs in an organization are distributed and
shared. Second, the affected parties must have open channels of communication to
the decision makers during the time period when decisions are being made, as well
as the means to influence them. The first of these conditions is often unavoidable in
organizations. Who will be given the promotion to a key j ob with higher pay, perks,
and status? Will resources be allocated to one division's pr oj ects or to another's, where
access to these resources means the winner will grow, thus offering j ob security and
better opportunities for advancement for its people? Should the corporation undergo
a leveraged buyout, in which the management group borrows extensively to buy
back the company's shares, so that management owns the firm? Leveraged buyouts
greatly increase the firm' s debt-equity ratio and the value of the remaining equity but
reduce the value of any existing debt by making it riskier. Should an outside
subcontractor be used, when doing so may move good j obs outside the firm? Will an
internal reorganization be instituted that promises effi ciency gains but that will require
layoffs for some employees and the need to learn new ways and a loss of power for
others?
In such decisions, the potential winners and losers will seek to influence the
outcome, expending both their own resources and, if they can, the resources of the
organization in the process. For example, the candidates for pr omotion may take time
from their current responsibilities to improve their apparent qualifi cations for the good
j ob, or they may attempt to curry favor with the bosses. The proponents of a proj ect
may devote their efforts to building the best possible case for investing in that proj ect,
hiding the potential diffi culties and focusing on the upside, while at the same time
trying to undercut competing proposals. The potential gainers and losers in the
refi nancing may spend millions of dollars on investment bankers, consultants,
lawyers, and advertising, trying to influence the outcome. Those threatened by the
subcontracting proposal may concoct self- serving arguments about reliability, ex post
opportunism, or morale. Those fearing the reorganization will resist the change by
hiding or distorting information, withholding cooperation, and attempting to frighten
others into becoming al lies. All this is essentially a political competition for rents and
quasi-rents. As such, it parallels the phenomenon of rent seeking in the public sector.
Employees are not the only ones affected by the firm's decisions. Stockhold ers'
returns are affected by a wide range of decisions, including investments, wage levels,
dividend policy, mergers, and pricing. For lenders, the security of their loans will
depend on the riskiness of the firm's investments, the wages it pays to workers, and
the dividends it pays to shareholders. A firm' s decisions about the design and placement
of its factories can affect community housing values, traffic, and environmental
quality; its choices of suppliers can affect the distribution of wages and profi ts among

111 Many of these concl usions depend on the existence of some limitations on workers' abilities to
borrow or to post a bond .
th e poten tial suppliers, and its pricin g policy affects both com petitors an d customers. Rents and
There arc frequent disputes about which of these interests arc legitimate ones th at Efficiency
ough t to be weighed in the decision- making process. Is a small com munity entitled
to block a decision by the area's largest employer to close its local factory? Are its
down town stores and their em ployees entitled to protection when a discoun t chain
ope ns up on the edge of town? Should neighbors be allowed to demand compensation
or mitigating measures when the expansion of a com munity medical center intensifies
traffic problems in one particular neighborhood?

Organizational Design : Opti mizing I n fluence Activities

So far, we have emphasized the costs involved when interested parties participate in
decisions, but that is only half the story. Good decision making requires information
aboutthe concerns of affected parties, about what options are available, and about
the likely con sequences of each option. The first kind of information is best obtained
directly from the people who are affected. These same people are also frequently the
most highly motivated to seek out good alternatives, gather evidence, perform analyses,
and forecast consequences. Competition among interested parties with opposing
interests may offer the best chance for all the relevant facts and desirable alternatives
to be effectively advocated. Provided the decision maker is incorruptible, these flexible
processes offer the best hope for well-informed decisions, the effect of which is to
raise the total value available to be shared.
We have thus identified two main elements that must be balanced in a cost­
benefit analysis to determine the best decision-making process for particular kinds of
decisions. Opening a decision process to participation by individuals whose own
interests are at stake increases infl uence costs, and these costs are greatest when the
decision redistributes large amounts of wealth or value among the parties. These costs
must be balanced with the improved information and analysis that accompany more
participation. We examine the ways in which these costs can be mitigated and the
hypothesis that decision processes are often designed to balance these two effects in
order to make the most effective decisions at the lowest possible cost.
LII\IIT COl\11\IUNICATION Much influence activity takes the form of politicking: arguing
one's case, often strenuously and disingenuously. An obvious way to try to control
such activity is to ignore or forbid it. This is not always easy, however. A dean can
tell a department chair that he or she will not hear further complaints about needing
more faculty positions ("slots"), but the chair might make an appointment ostensibly
to discuss another matter and renew the campaign. The dean can attempt not to listen
and even order the chair to drop the subj ect, but many of the costs have already been
incurred.
One effective way to foreclose communications is to ensure that parties do not
have the information they need to politic effectively. For example, individual salary
information is often kept secret in organizations because if salaries were widely known,
lower-paid people could argue more easily for raises by comparing their own
performance with the worst-performing of the well- paid employees. M oreover, once
a salary decision is made, it is not generally subj ect to appeal. As a general principle,
to limit the time wasted in rent seeking, rent-distributing decisions should be made
once and for all. Once made, the debate should be ended and the matter closed.
The widely deplored unresponsiveness of bureaucracies and of personnel
departments in particular can be interpreted in this light. Line managers frequently
complain that Personnel is more interested in rules and salary curves than in attracting,
keeping, and motivating quality employees. Personnel managers cannot even be
reached by phone when an important matter comes up because they are always in
 274
Efficient meetings with one another, concocting new ways to keep the people who do the real
Incentives: work from doing what needs to be done. But suppose Personnel were to respond to
Contracts and every claimed need for relaxation of hiring procedures and every demand that
Ownership somebody had to receive a special raise or promotion or else be lost to the company.
In that case, employees, realizing that their supervisors can get special deals for them,
would have every reason to campaign for them. As it is, the managers can respond
that they would love to do something for the employee, but Personnel will not permit
it, and so the employee's incentives for influence activities are blunted. Requiring
people to work through bureaucratic channels has a similar effect by raising the costs
of special pleading.
Requiring openness in a democratic society means that foreclosing communica­
tion is not so easily available in the public sector, and this may explain why influence
costs are so high there. Regulatory commissions must hold public hearings, and their
decisions are subject to court review and legislative oversight. Anyone is free to appear
before these hearings and to appeal to the courts and to the political system if they
do not get their way. Furthermore, the courts, which have the power to intervene in
many private transactions, have in recent years become open to new lines of argument
that further increase the scope for influence.
Controlling influence by limiting communication is costly because useful
information is often cut out as well. Much influence activity involves mixing self­
serving information in with valuable reports. Knowing what competing firms are
paying is useful, and the only way to get such information may be from employees
who are campaigning for a raise. Refusing to listen to the entreaties means eliminating
the source of pay information as well, and, as a result, the quality of decision making
is degraded.

LIMIT THE DISTRIBUTIONAL IMPLICATIONS OF DECISIONS One clear way to limit

competition for rents is to equalize their distribution across potential competitors, or
at least to limit the possible differentials. As in the Houston-Tenneco example from
Chapter 6, this leads to a policy of narrower differences in pay and benefits than
outside market conditions or productivity considerations might indicate. For example,
it was long standard in law firms to pay all partners of a given seniority the same
amount, independent of their specialty or the revenues they generated for the firm. 2 1
This policy reduced conflict over the distribution of profits and probably encouraged
partners to devote time to activities that were in the long-run interest of the firm but
did not result immediately in easily attributable revenues. The cost of the policy is
that the informational and incentive roles of rewards are muted by closing differentials.
In law firms, the costs have been manifested in high revenue-producing partners
leaving to form their own firms or to join others that have more results-oriented pay
schemes. In recent years, the seniority pay system has been breaking down under
these pressures.
The policy of pay equity is most likely to be worthwhile when the potential for
damaging influence activities is greatest. An empirical study of pay dispersion within
1 , 80 5 university, college, and junior college academic departments provides useful
evidence of this. 22 As argued previously, increased knowledge of individual pay levels

21 Ronald Gilson and Robert Mnookin, "Sharing Among Human Capitalists: An Economic Inquiry
into the Corporate Law Firm and How Partners Split Profits," Stanford Law Review, 37 (January, 1985),
313-92.
22 Jeffrey Pfeffer and Nancy Langton, "Wage Inequality and the Organization of Work: The Case
of Academic Departments," Administrative Sciences Quarterly, 3 3 (1988) , 588-606.
 275
increases the likelihood that differen tials will lead to politicking. Departments whose Rents and
members frequently interact socially and in which people frequently collaborate rather Efficiency
than work alone should feature much closer commun ication among department
members, including communication about pay. They should then be more susceptible
to influence costs and should adopt narrower differentials than experience and
productivity (and, presumably, outside market opportunities) would generate. Public
universities often are required to make pay levels public, whereas private universities
arc not. Therefore, public institutions should, under the theory, have narrower pay
differentials. Furthermore, communication is more difficult and information is less
likely to be dispersed thoroughly in larger groups, so small departments ought to show
less disparity in pay. Finally, a participative, democratic governance structure within
the department, as opposed to a more autocratic one in which a chair makes the key
decisions, presents more opportunities for costly politicking and so might be expected
to be associated with smaller differentials. I n fact, each of these factors did show up
in the study as statistically significant: Pay dispersion was narrower in democratically
run departments, in smaller ones, in public institutions, and in departments with
extensive communication among members.
More generally, a policy of protecting individuals and groups from adverse
consequences of organizational decisions can actually promote efficiency. This may
take the form of giving individuals or groups a direct say in what is going on and
being done, although this participatory management approach risks opening up the
system to extra influence. An alternative is to require that changes will not be instituted
unless they benefit everyone. Moreover, it may be worthwhile to pass over some
value-creating opportunities if they would too adversely affect some of the organization's
members, because the apparent value may be completely lost in influence activities.
This may even extend to breaking up the organization.

DECENTRALIZE AND SEPARATE UNITS I nfluence activities are possible only when there
is a central authority with the ability to affect the distribution of costs and benefits
between individuals or units. One extreme solution is to remove the central authority.
Thus, for example, when the energy price increases of 1 97 3 made aluminum
production in Japan uneconomical, Mitsubishi Chemical, Sumitomo Chemical, and
Showa Denko all decided to contract or eliminate their aluminum production
operations. 23 The difficulty with such moves is that the affected division will try to
resist, making claims on corporate resources to keep it going or at least to slow the
pace of disinvestment. This was a special danger in the context of the Japanese system
that encourages consensus decision making and consultation. The solution was to
spin off the aluminum operations, that is, to create new companies separate from the
parent companies with no claim on the parent companies' resources. I n that way, the
opportunities for i nfluence and the attendant costs were greatly reduced.

STRllCTURE DECISION PROCESSES TO LIMIT INFLUENCE ACTI\'ITIES In some situations

it is possible to protect efficiency from distributional conflict by separating the effi ciency
aspects of decisions from their distributional consequences. For example, airline
companies do not have much reason to care which cabin attendants are assigned to
which flights; all that matters is that all the flights are covered. The attendants
themselves do care, however, and if management were to make the assignments there
would be vast opportunities for trying to influence the decisions. The solution adopted

B James Abegglen and George Stalk, Jr. , Kaisha: The Japanese Corporation (New York: Basic Books,
198 5), 24-25.
 276
Efficient is to let th e attendants themselves choose among the flights that must be covered in
Incentives: order of seniority. This separates the efficiency and distributive aspects, and, by basing
Contracts and it on seniority, puts the distributive aspect of the decision beyond management's
Ownership control and thus frees the decision from incentives to exert influence. (The same
process is used in assigning pilots, copilots, and engineers, although here it can have
an efficiency cost when it results in the least experienced pilot, copilot, and engineer
being teamed together in charge of a flight. )
More generally, setting immutable rules and establishing decision-making
procedures that must be followed can reduce influence. For example, a firm policy
of not responding to outside job offers reduces the incentives to seek offers as bargaining
tools. The cost is that valued employees may be lost in this way. Similarly, to limit
politicking, promotions may be made purely on the grounds of seniority and past
performance, even when this is largely irrelevant to j udging who would be the best
person for the job. But then people may be placed in important jobs for which they
are ill qualified, and better people passed over. The famous Peter Principle ("People
are promoted to their levels of incompetence")-a tongue-in-cheek description of a
real organizational tendency-is a natural outcome of this process.
Highly structured decision routines are very common in personnel matters like
the route assignment decision described earlier. In determining annual raises, firms
often allow the employees to fill out a self-evaluation and perhaps to meet briefly with
their supervisors to discuss their performance, but they rarely permit any more
influence than that. The main cost of measures like these that prevent rent seeking
is that they limit the information available to decision makers. However, a well­
designed decision process can often generate adequate information while incurring
relatively low influence costs.
Tenure decisions for university professors provide an example. Typically, an
academic department will have a limited number of slots for tenured faculty members. 24
U ntenured professors are reviewed after some period (typically six years) to determine
if they are qualified for tenure. Both the candidate for tenure and the other untenured
professors have a direct interest in the decision, and these interests are often conflicting
because of the limited number of slots. Because the other untenured professors
normally have little information that is not already available to the tenured faculty,
they normally play no role at all in the decision process. The candidates themselves
usually play only a small role, perhaps by providing a statement concerning their
teaching and research as well as copies of their course syllabi, teaching ratings, and
published books and papers. The tenured professors, who have less reason to be
threatened by a new tenure appointment, read these materials, solicit the opinions of
outsiders, and vote on a recommendation to the university administration.
The tenured professors do have an interest, though, in expanding the size of
their department, in order to encourage smaller classes, have more colleagues with
whom to discuss their research, and spread administrative work among more people.
Given university budget constraints, one department can grow only at another's
expense. To limit rent seeking by departments, the university administration typically
fixes the number of tenure positions in advance and has a slow and arduous procedure
for revising the number of positions allotted to each department. The process is
structured to limit opportunities for rent seeking while still acquiring the necessary
information to make a well-informed decision.
The policy long followed at Hewlett-Packard, the California-based instruments
and computer company, of allowing divisions to pursue essentially any reasonable

24 The q uestion of why there is tenure is taken up in Chapter 1 1 .

 277
research and development projects they wish, but requiring them to finance their Rents and
projects out of funds they generate themselves, is also understandable as a response Efficiency
to influence costs. The policy prevents the divisions from attempting to persuade
central management to draw resources from other divisi ons. This saves in terms of
both the costs of these efforts and the costs of other divisions' defendi ng against them,
but it limits the opportunities to undertake worthwhile projects.
In the Houston-Tenneco example, a major complaint of the Houston explora­
tion people (besides the pay level) was the frustratingly slow bureaucratic pace and
the timidity of decision making at Tenneco compared to what they had known at
Houston. Yet again this policy may have been appropriate to the large concern, where
i nfluence was a greater potential problem and so institutional checks on it were more
valuable.
Product and_ Pricing Decisions It is helpful to compare these processes used for
personnel assignments and salary decisions with those of a different kind of decision:
the determi nation of how a product should be designed and priced. Product
design and pricing decisions are crucial ones for any organization, being important
determinants of marketing success. Nevertheless, these decisions often have only
minor redistributive effects within the organization (except when there are competing
product-design teams, employees are probably unlikely to have a large personal stake
in any particular design). These decisions commonly are made using an open
committee process in which the decision makers acquire information from as many
sources as possible and are free to exercise discretion over a wide range of possibilities.
Moreover, the decision is typically open to reconsideration if new information comes
to light. Finally, unlike tenure decisions, pricing decisions are typically delegated to
middle managers with little oversight by their superiors. The comparatively unstruc­
tured product pri cing and design decisions stand in stark contrast to the highly
structured personnel decisions. The difference is attributable to the absence of any
serious di stri butive consequences of pricing and design decisions.
Many other examples of highly tailored decision processes can be found in
business practice. In selecting a supplier for some product, such as production
equipment or office computers, the firm may ask the potential suppliers to make
proposals or sales presentations, indicating how the product would work and why they
would do better than those supplied by competitors. This process may reveal issues
or information about potential applications that the firm had not contemplated and
so lead to another round of questioning. The sales representatives might be limited
to make formal presentations and to answer questi ons. After they leave, a more open­
ended discussion among insiders can take place. Once a decision is made, it is unlikely
that it could be reopened by any claim of new information by a seller, but it might
be reopened by an insider if he or she had important new information, for example,
about changing needs. The process as a whole is tailored to allow useful information
to be contributed by the interested parties who are best informed while controlling
and reducing the time wasted in excessive attempts at influence.

I nfl uence Costs and the Legal S ystem

Before legal systems throughout the world developed a set of rules to govern commercial
transactions, merchants introduced their own sets of rules and their own private courts
to render j udgments in di sputes among themselves. These rules evolved over time to
meet the changing needs of commerce. The rules of the English merchants were
eventually incorporated into the English common law, where they continued to evolve
into the modern business law used i n much of the world today. The "law and
economics" movement seeks to interpret the existing system of commercial, contract,
 278
Efficient and bankruptcy law as efficien t solutions to the problems of doing business; it also
Incentives: seeks to identify better solutions to these problems.
Contracts and
Ownership CHAPTER 1 1 BANKRUPTCY The complicated bankruptcy law of the United States
provides many examples of detailed rules that are intended to enhance efficiency.
One such rule is the distinction between liquidation and reorganization in bankruptcy.
When a company is liquidated , its assets are sold and the proceeds are divided among
the company's lenders and creditors. Liquidation is usually forced on the firm by its
worried creditors who are trying to protect what value is left in the firm. However,
liquidation is not always the best option for companies that are unable to pay their
debts or for their creditors.
In 1 990 Bud's Ice Cream of San Francisco, a national distributor of premium
ice cream, found itself in financial trouble due to management and distribution
problems. Although the company was unable to pay its debts, its production facilities,
skilled employees, and strong brand name were attractive to potential buyers, who
believed they could overcome the company's problems and restore its profitability.
When Bud's petitioned the bankruptcy court for reorganization under Chapter 1 1 of
the U. S. bankruptcy code, its creditors were prevented from seizing assets that were
promised as collateral. An orderly process was then instituted for determining how
the remaining value in the company could be maximized and how the results should
be divided among the various claimants.
Without this option, there would have been a free-rider problem among Bud's
creditors, who would each have been tempted to demand immediate payment of its
claims or to seize some of Bud's assets in the hope of being paid quickly, before there
was no cash or collateral left. This competition among the creditors, combined with
an understandable refusal by others to provide new credit, can destroy a company­
an outcome that can eliminate jobs and harm employees, suppliers, customers, and
creditors alike. By using Chapter l l bankruptcy, production in the company's facilities
can continue uninterrupted and the brand name can be preserved, avoiding the waste
that accompanies the liquidation of a company with valuable but fragile assets, such
as brand names or employee organization. In Chapter 1 1 bankruptcy, the various
creditors form a committee to make arguments to the court concerning their mutual
interests. This eliminates some of the most destructive kinds of rent-seeking competition
among the creditors and allows the group to recoup as much as possible from the
loans they had made. The resulting negotiations under court supervision make it less
likely that a single recalcitrant creditor could block a deal to sell the company to new
owners, making an efficient agreement more likely.
We can illustrate how the rules in the bankruptcy code are designed to preserve
value by considering how the priorities for the payment of the firm's debts are set.
For a company like Bud's Ice Cream, doing business requires supplies of cream,
sugar, fruits, flavorings, and other ingredients·, as well as materials for boxes and
packaging, cash for workers and shippers, power for refrigeration equipment, and so
on. Bud's must acquire these supplies and services and it must ship its goods before
it is paid by its customers. Therefore, the company relies on many of its suppliers to
help finance its operation by trade credit; that is, these suppliers don't demand payment
for the goods and services that they supply until 30 days or more after they make
delivery. If the bankruptcy rules gave priority to debts incurred before the Chapter 1 1
petition was filed, then once a bankruptcy petition was filed, suppliers might withhold
their supplies, for fear that they would never be paid for their new shipments. Refusals
of that sort would prevent the company from doing business, destroying whatever
remains of its value. Actual bankruptcy law takes that problem into account by giving
a higher priority to those who supply new credit after the bankruptcy petition is filed
 279
so that, from their perspective, it is quite likely that the trade credits they provide will
be repaid. 25 At the same time, to protect the i nterests of exi sting creditors, the cour t
Rents and
Efficiency
commonly lim its the compan y's ability to make nonr outi ne i nvestmen ts, raise wages,
or pay di vidends to owners.
CoHPOIUTE OFFICEHS' AND D IHECTOHS' LIABILITY Another area of law that is interesting
to examine in light of the influence costs theory is that regarding the liabi li ty of
corporate officers and directors. In the mid- l 980s in the U nited S tates there was a
change in the way courts i nterpreted these laws. When the directors of a compan y
resisted a takeover attempt, either thwarti ng the attem pt completely or eventually
settli ng with another company for a lower price, or when they accepted a takeover
bid that some shareholqers believed was too low, suits were filed holding the officers
an d directors of the company personally liable for the shareholders' losses. Earlier,
the courts had adhered to a business judgment rule, according to which the court
respected the · business j udgments of directors even if the results of their decisions
turned out badly, with few exceptions. Directors could be held liable only if they were
shown to be "disloyal" or " negligent" in their decision making. Thi s business j udgment
rule was eroded by an increasingly inclusive interpretation of the word " negligent."
One result of the changi ng standards was an "insurance cri sis" in which insurance
companies refused to offer coverage agai nst the growing risk of lawsuits, which in tur n
deterred many qualified people from accepting positions as directors of public
companies. Most states reacted to this crisis by changing their laws to restore protection
for offi cers and directors, for example, by changing the standard so that behavior must
be shown to be "reckless" rather than merely " negligent" before damages can be
awarded-a change that its pr oponents say will reduce wasteful rent seeking by both
shareholders and their lawyers.

Partici pato ry Management

One of the recent trends in management practice has been the growi ng emphasis on
participatory management, a style of management that encourages participation by a
wide range of employees at various levels in making all sorts of decisions: pr oduct
improvements, changes in the production line and the organization of work, training
methods, and so on. Although this practice has gained increasing attention among
American firms, their Japanese counterparts have applied it to more levels of the
organization with their emphasis on quality circles, groups of employees who make
suggestions to improve productivi ty and product quality, and other simi lar management
techniques. The openness and the attendant access that employees have to decisi on
makers raises the risk that these techniques might be undermined by individual rent
seeking. H ow, then, do these organizations avoid incurring excessive influence costs?
In J apanese firms, the system of lifetime employment with narrow pay differentials
between ranks and with both pay and responsi bility tied closel y to seniori ty makes it
difficult for employees to gain large personal advantages by any ki nd of politicking.
Indeed, the larger risk is that employees who seek their personal interests may be
branded as disloyal and lose the benefits their tenure would normally bring. Lifetime
employment has a second benefit: It assures these employees that any labor-saving
innovations that they recommend will not cost them their j ob s. The long period of
growth of Japanese companies has further contributed to the sense of security that

25 The actual rules governing the priority of various classes of creditors are quite complicated. The
class of secured creditors (those to whom collateral was promised) has a relatively high priority. Unsecured
creditors for whom the debt was incurred after the petition is filed have a lower priority, and a still lower
priority is assigned to unsecured creditors whose debt was incurred before the bankruptcy petition.
 280
Efficient their employees share. For a growing firm, jobs may be best protected by ensuring
Incentives: that the company's costs are low enough to allow the growth to continue. In
Contracts and comparison, when output in a company is shrinking, the interests of capitalists and
Ownership laborers in finding labor-saving, cost-reducing production methods may become
sharply divergent, leading to much higher influence costs.
Furthermore, for Japanese managers, frequent job rotation, lifetime employment,
limited outside opportunities, narrow pay differentials, and the relative homogeneity
of their backgrounds all contribute to a sense that what is good for the management
group as a whole is good for each individual manager. This is an attitude that limits
rent seeking and contributes to the impressive responsiveness of Japanese firms to
changing circumstances. These mutually supporting pieces form a coherent whole.
The participatory decision-making system, the pay structure, the promotion criteria,
and the lifetime employment policies are mutually complementary, and the construc­
tion of the system is a design decision where the pieces need to fit together properly.
When on_e element is disturbed, however, great costs can result.
For example, consider the situation that prevailed in the late 1980s after Sony
Corporation, the Japanese electronics firm, purchased an American record company,
CBS Records, and an American movie and television production and distribution
firm, Columbia Pictures Entertainment. Sony hoped to combine its strength in
entertainment delivery hardware (such as compact disk players, video cassette
recorders, and television receivers, including the emerging High-Definition Television
technology) with CBS Records' and Columbia's strength in entertainment software
(musical albums, new movies, television shows, plus an enormous library of older
films from Columbia) to better exploit the complementarities across these segments
of the entertainment business. To attract management talent to Columbia, however,
Sony paid Columbia's two top managers much higher salaries than the norm in Sony,
more in fact than all the parent company's senior managers put together! Top
executives at Sony soon found themselves spending hours every day answering
complaints by managers from the parent company and its "hardware" subsidiaries who
felt they should be paid comparably with the people in the entertainment subsidiaries.
In terms of costs to the organization, the time spent by top executives talking
to unhappy managers is only the tip of the iceberg. Surely, the managers themselves
spent many hours talking among themselves and thinking about how best to present
their arguments. The total cost suffered by an organization like Sony that relies on
extensive internal communications must have been substantial.
El\tPLOYEE STOCK o,rnERSHIP PLANS (ESOP) In the United States, companies often
attempt to involve their workers more deeply in decision making. One of the ways to
lessen the differences in interests between workers and capitalists is the Employee
Stock Ownership Plan, commonly known as the ESOP (pronounced "esop"). By
making employees shareholders, firms with ESOPs hope to make the workers as a
group more accepting of organizational changes, more willing to be flexible in
accommodating labor-saving arrangements, and more forthcoming with suggestions
about how to improve operations. Management publications often emphasize the role
of "listening to employees" and "making employees feel involved" in assuring the
success of ESOPs. This advice accords well with influence cost theory. The benefits
of any wealth-creating changes must be systematically shared arnong the "members"
of the organization to deter their blocking the change to protect themselves, and,
better yet, to inspire their cooperation in promoting the value-creating change.
 281
Rents and
Efficiency

SUMl\l AHY
Reputations have economic value to the extent that they make it easier for their
possessors to engage in work, trade, or other valuable activities . An activity is valuable
if it generates quasi-rents, returns in excess of what could be had in other activities.
These quasi-rents are commonly created by costs associated with changing businesses,
jobs, or careers, such as the cost of moving or of specific training. The desire to keep
one's current job or customers forms the basis of market incentives for hard, honest
work and high-quality products. This incentive to acquire and maintain a good
reputation is stronger the more frequently the reputation is useful and the longer the
horizon over which it may be used.
When there are no specific investments being made, fi rms may sometimes use
effi.ciency wages-wages higher than the worker's opportunity wage and which therefore
consist partly of a rent-in conjunction with monitoring of the worker to provide
incentives for good behavior. Incentives and monitoring are substitutes in the
implementation problem; a higher wage can be used to offset poorer monitoring
opportunities. However, this fact does not imply that improved monitoring always
leads to lower wages. Instead of substituting monitoring for wages, the firm may fi nd
it more profi table to seek a higher level of performance from the worker when
monitoring becomes easier.
The Marxian view is that capitalists waste resources on unnecessary monitoring
in order to keep wages low. This criticism is problematic, however, because incentives
depend on relative wages: Even a socialist economy cannot pay high relative wages
to every worker.
People who wish to be trusted can benefi t from establishing a reputation of
honoring trust. Reputation, operating without the assistance of specialized institutions,
is effective only when the reputed behavior is something that can be directly observed
by others who rely on the reputation. It is therefore most effective in a long-term
bilateral relationship in which the parties are aware of one another's past behavior. It
is more problematic in a large merchant community because outsiders to a transaction
can rarely know who is right in any business dispute. To overcome this problem,
merchants long ago developed practices and institutions that establish standards of
behavior, discover facts, and resolve disputes. In modern times, credit bureaus, Better
Business Bureaus, and corporate cultures can all be partially understood in these
terms, although these institutions sometimes have other functions as well.
An additional problem for the reputation mechanism is that it may not be in
the third parties' interests to participate in sanctions against disreputable trading
partners. Historically, institutions have arisen to maintain discipline in the application
of sanctions. An example is the H ansa, a coalition of merchant guilds ("kontore") in
northern Europe, that arose in the thirteenth century to enforce discipline among
merchants in applying sanctions to cities that reneged on their promises.
End games pose another problem for a reputation mechanism. People nearing
retirement or firms nearing the end of a contractual relationship have less need to
maintain the good will of their contracting partners, and they may then seek other
ways to enforce their agreements.
The existence of rents and quasi-rents tempts people to spend resources competing
for them. To the extent that these expenditures are wasteful, they are called infl.uence
costs. These costs tend to be greatest for those decisions that have the large redistributive
consequences, whereas the benefits of infl uence activities are greatest for complex
decisions that have a large impact on the organization's overall objective and that are
 282
Efficient facilitated by a free flow of information. Influence costs are created by efforts of
Incentives: affected parties to manipulate decisions. These may include employees, !;uppliers,
Contracts and customers, shareholders, communities, and so on, depending on whose interests are
Ownersh ip potentially affected.
Organizations cope with influence costs by tailoring their decision processes to
the kind of decision being made, limiting interested parties to contributing only the
most relevant information, making contentious decisions final and not subject to
review, and softening the impact of change by spreading it equitably among
organization members. For some kinds of decisions, rigid bureaucratic rules can
actually be optimal, if the total value created by improving decisions is small compared
to the total costs of influence that would be attempted. For this reason, the distribution
of a fixed pay pool is usually managed under a rigid set of procedures, whereas the
setting of other financial variables, such as product prices, is frequently more informal.
In one example, aluminum companies were separated from their parent companies
during the contraction of the Japanese aluminum industry to insulate the parent and
its subsidiaries from influence by the aluminum divisions. Generally, the extent to
which any party is allowed to participate in a decision depends on the information
and analysis the party is able to contribute and on the degree to which the decision
might redistribute resources, pay, or perquisites to that party. Influence costs can also
be reduced by rules that protect the interests of members of the organization and by
sharing the benefits of any changes with them.
The legal system provides many opportunities for influence activities, but some
legal institutions appear to be designed with the objective of reducing influence costs
in mind. Certain bankruptcy rules and corporate directors' liability rules are examples.
When worker participation in decision making is desired, a system of pay equity
and benefit sharing helps to reduce influence costs. The Japanese personnel policies
seem especially conducive to constructive participation by workers, and firms with
ESOPs in the United States find it more attractive to include a wide group of workers
in company decision making.
The problem of influence seeking is widely thought to be more costly for
decisions made in the public sector than for those of the private sector. One reason
is overlapping political jurisdictions, which allow interested parties to pursue their
interests (or require them to defend their interests) in many different forums. A second
reason is the inability of democratic governments to make final decisions that are not
subject to continuing review by affected constituents.

• BIBLIOGR\PH IC NOTES
As our opening quotation from Alfred Marshall suggests, the idea that rent
distribution can affect efficiency is by no means a new one in economic theory,
though it had not received much emphasis in modern theory before 1 980. Papers
by Carl Shapiro and by Benjamin Klein and Keith Leffler about how rent flows
provide firms with an incentive to build and maintain a reputation for product
quality, as well as the efficiency wage model by Carl Shapiro and Joseph Stiglitz
discussed in the text initiated the modern theory. Samuel Bowles and Masahiro
Okuno-Fujiwara have extended the Shapiro-Stiglitz analysis in important direc­
tions also discussed in the text.
Organization behavior scholars have emphasized the importance of corporate
stories and culture: sec the book by Joanne Martin or the volume edited by Peter
J. Frost, Larry F. Moore, Meryl Lewis, Craig Lundberg, and Martin. Among
economists, David Kreps has emphasized the role of reputations in organizations
as a substitute for contracting and especially the roles of organizational routines
 an d cultu re in es tablishing expec tations and standards of behavior. Jacques Cremer Rents and
has also contribu ted importan tly to this lin e of work. 'The analysis of repu tations Efficiency
and trust in the text makes use of game theory, a set of ideas and methods that
have had a maj or impact on economic modeling and analys is in recent years.
Recen t treatments of this subj ect that are relatively accessible to those without
extensive technical and mathematical backgrou nds are given by Avinas h Dixit
an d Barry Nalebu ff, by Robert Gibbons, and by Eric Rasmusen. The form al,
game-theoretic modeling of the dynamics of repu tation formation an d its
implications was initiated by Kreps and Robert Wils on, by the present au thors in
our 1982 paper, and by the four of us together.
Gordon Tull ock was the firs t ec onomis t to emphasize the role of rent-seeking
behavior, which has been applied to analyze the cost of big government by many
others. Many of the classic papers in the field are c ollected, with additional new
material, in the volume by James Buchanan, Robert Tollis on, and Tull ock. The
present authors introduced the first studies of influence processes in firms and
how they are optimally managed: S ee the paper by M ilgrom and our 1988 and
1990 papers.

• REFERENCES
Bowles, S. "The Production Process in a Competitive Ec onomy," American
Economic Review, 75 (March 1985), 16-36.
Buchanan, J., R. Tollison, and G. Tullock. Toward a Theory of the Rent-Seeking
Society (College S tation, TX: Texas A&M U niversity Press, 1980).
Cremer, J. "Cooperation in Ongoing Organizations," Quarterly Journal of
Economics, 101 (February 1986), 3 3-49.
Dixit, A. , and B . N alebuff. Thinking Strategically (New Y ork: W. W. N orton,
1980).
Frost, P. , Louis L. M oore, C. Lu ndberg and J. Martin, eds . Reframing
Organizational Cultures (Newbury Park, CA: Sage Pu blications , 1991).
Gibbons, R. Game Theory for Applied Economists (Pri nceton, NJ: Princeton
U nivers ity Press, 1992).
Klein, B., and K. Leffler. "The Role of M arket Forces in Assuring Contractual
Performance, " Journal of Political Economy, 89 (1981), 615-41.
Kreps, D. "Corporate Culture and Ec onomic Theory," Perspectives on Positive
Political Economy, J. Alt and K. Sheps le, eds . (Cambridge MA: Cambridge
U niversity Press, 1990), 90-143.
Kreps, D., and R. Wils on. " Repu tation and Imperfect I nformation, " Journal of
Economic Theory, 27 (August 1982), 25 3-79.
Kreps, D., P. M ilgrom, J. Roberts and R. Wils on. "Rational Cooperation in the
Finitely Repeated Pris oners' Dilemma," Journal of Economic Theory, 27 (August
1982), 245-52.
Martin, J. Cultures in Organizations: Three Perspectives (New Y ork: Oxford
U niversity Press, 1992).
Milgrom, P. "Employment Contrac ts, I nfluence Activities and Efficient Organiza­
tion Design, " Journal of Political Economy, 96 (1988), 42-60.
M ilgrom, P., and J. Roberts. "An Ec on omic Approach to Influence Activities
in Organizati ons," American Journal of Sociology, 94 (Su pplement) (1988), S154-
S l 79.
Milgrom, P. , and J. Roberts. "Bargaining and I nfluence Costs and the Organiza-
 284
Efficient tion of Economic Activity, " Perspectives on Positive Political Economy, J . Alt and
Incentives: K. Shepsle, eds. (Cambridge, MA: Cambridge University Press, 1990), 57-89.
Contracts and Milgrom, P. , and J. Roberts. "The Efficiency of Equity in Organizational
Ownership Decision Processes, " American Economic Review, 80 (May 1990), 1 54-59.
Milgrom, P. , and J. Roberts. "Predation, Reputation and Entry Deterrence, "
fournal of Economic Theory, 2 7 (August 1982), 280-3 1 2.
Okuno-Fujiwara, M. "Monitoring Cost, Agency Relationships and Equilibrium
Modes of Labor Contracts, " fournal of the fapanese a nd In ternational Economies,
1 (June 1987), 1 47-67.
Rasmusen, E. Games a nd Informa tion (Oxford: Basil Blackwell Ltd. , 1989).
Shapiro, C. "Premiums for High Quality Products as Rents to Reputation,"
Quarterly fournal of Economics, 98 ( 1 983), 659-80 .
Shapiro, C. , and J. Stiglitz. "Equilibrium Unemployment as a Worker Discipline
Device," American Economic Review, 74 (June 1 984), 43 3-44.
Tullock, G. "The Welfare Costs of Tariffs, Monopolies and Theft," Western
Economic fournal, 5 ( 1 967), 224-32.

EXERCISES

Food for Thought

1 . Firms that employ skilled workers are very conscious about the wages being
paid by other employers and where they fit in the distribution of wages. In fact, there
are thousands of surveys done each year (and sold to businesses) of the compensation
being paid in different areas and industries to different types of employees. Is this
consistent with labor being hired in a competitive market of the type usually modelled
in standard microeconomics texts, in which firms hire workers up to the point where
the wage equals the marginal product of labor? How else could you explain this
phenomenon?
2. In Chapter 5 we described the "Market for Lemons"-the possibility of
market breakdown under adverse selection that may occur when sellers are better
informed than buyers about product quality. Could repeated dealings and reputations
help alleviate this problem? How? What difficulties do you see in relying on reputations
to overcome adverse selection? What institutions might facilitate the workings of
reputations in specific contexts?
3. In many countries, corporations whose shares are publicly traded are
required to have their financial statements audited by independent accountants who
check whether the financial information being provided by management to investors
is accurate and has been prepared following accepted methods and procedures and
who then publicly attest to their findings. The auditors are generally chosen by
management (perhaps subject to nominal stockholder approval). Audit work provides
a major source of income for accounting firms. Generally, each accounting firm has
many clients that it audits: Even when it might be technically feasible for a relatively
small accounting firm to audit a large corporation's records, there seems to be a
reluctance for corporations to use audit firms when a single client would represent
too much of the accountant's business. Instead, a handful of extremely large accounting
firms typically do almost all the auditing of large corporations, with each having many
corporate clients. How do you account for this?
4. The proper level of salaries for public officials is often a matter of debate.
 285
In many countries, publi c official s supplement their incomes by performin g other Rents and
duties. In the United States, Congressmen deliver lectures for pay and former Efficiency
bureaucrats cash in on their con nection s by becoming high-paid lobbyists. Recent
U.S. pr oposals call for tyin g pay increases to limits on these activities. Discuss the
efficacy of such a plan.
5. In university decision makin g, the board s of trustees often receive reports
from students and faculty members, but these groups arc rarely represented directly
on the boards themselves. H ow can this pattern be explain ed? What difficulties would
you expect to attend such representation?
6. Describe the respective interests of the cust omers, empl oyees, suppliers,
competitors and local govern ments in a firm's decision to move operations from its
current location to a newer, more efficient plant. What would be the advantages and
disadvantages of representing each of these interests in a group respon sible for makin g
the decision?
7. In 1991 Eastern Airlin es, which had attempted a reorganization in
bankruptcy, finally filed for liquidation, stoppin g its flights and sellin g its assets.
Amon g the airline's creditors at the time were cust omers who had paid for ti ckets.
Accordin g to the value-maximizati on criterion , should customers be offered a high
priority to have their tickets refunded? Why or why n ot?
8. In 1990, Quaker Oats Co. spun off its troubled Fisher Price toy subsidiary
by distributing shares in the new company to Quaker's own shareholders. Accordin g
t o company officials, the reason fo r the chan ge was t o allow Quaker to con centrate
management attention on its basic food busin ess. (Multidivision al companies often
lavish more atten ti on and resources on some subsidiaries than on others. ) Why would
Quaker have to turn Fisher Price into a separate business to achieve that result?

f M athematical Exercises I

1. In the context of the efficien cy wage model, suppose that N i s given by

(1 - 6 1)/(1 - 6) , where 6 is a discount factor givin g the value today of a dollar to
be received a y ear from n ow and t is the number of years the worker anticipates
remain in g with the firm. Thus, N is a measur e of the horiz on over which the
employee is con cern ed, appropriately discounted. S uppose the worker expects to be
employed for 5 years, that is, t = 5, and evaluates futur e in come using 6 = 0. 9, so
that N is approximately equal to 4.1. The worker's opportunity wage (w) is $20, 000
per year, and he or she can make an immed iate gain of $ 3 , 000 by leakin g the firm' s
trade secrets to a competitor. The firm can mon itor the worker to prevent this, but
doin g so is costly: raisin g the probability p of catchin g misbehavior by an amount �p
costs $1, 500�p.
Formulate the firm's problem in selectin g the mon itorin g inten sity p and the
wage w and solve for the optimal values.
S uppose n ow that the threat of a takeover or bankruptcy endan gers the e mployee' s
future with the firm, so that the employee n ow expects to be employed only on e more
year. Will the contract you found above still d eter cheatin g? Why or why n ot?
M ore gen erally, show that in the efficiency wage model, a len gthened horiz on
allows providin g incentives with a lower wage. That is, show that the optimal wage
level w*(N) is a nonincreasing function of N provided the margin al cost of monit orin g
is positive and increasin g in p, that is, pr ovided M' (p) > 0 and M"(p) > 0 for all p.
2. Suppose there is some chan ce that a n oncheatin g employee will voluntarily
leave the firm for reason s un conn ected with the amount of the wage. What effect will
 286
Efficient this have on the probability of cheating for any given monitoring intensity and wage?
Incentives: What will be the effect on the optimal wage?
Contracts and 3 . In the efficiency wage model, suppose there is some probability q that the
Ownership worker will appear to be cheating, even though he or she has actually worked hard
and honestly. Determine the proper efficiency wage in that case.
4. In the discussion of the game of offering trust in the text, it was shown that
if the game is being played only once, then trust will not be offered and, if it should
be offered, it will not be honored. Suppose instead the two parties know they will
interact exactly twice. Would trust be offered and honored in the second, final
interaction? Why or why not? If the parties anticipate this and trust is offered in the
first round, will it be honored? Why or why not? [Hint: Will honoring trust affect
behavior at the last round?] Will trust then be offered? By this logic, what would
happen if the parties knew they would interact exactly three times? Any arbitrary, but
finite and known number of times? Do you find this to be a credible prediction about
how people actually would behave? Why or why not?
5. Suppose instead that each time the parties interact, there is a probability p
> 0 that they will interact again, and, correspondingly, a probability of ( 1 - p) that
this will be their last interaction. Then at any point, the probability that they will
interact twice more is p2, the probability of three more interactions is p 3, and so on.
Suppose the parties are interested in the expected value of the total payoffs they
receive. Thus, each seeks to maximize the sum of the payoff at the current round,
plus p times the payoff that accrues if they interact again, plus p2 times the payoff if
they should interact a third time, plus p 3 times the payoff from a possible fourth
interaction, and so on. In summation notation, the objective of each party i is to
maximize "'2:.p n ui 11 where n runs from O (the current round) to infinity and u in is the
,

payoff to party i at the corresponding round. Show that so long as the payoff V to
each party when trust is offered and honored is strictly larger than the payoff when
trust is not offered (which we set equal to zero in the text), then for any values of the
gain G to abusing trust and loss L to having trust not be honored there is a value of
p that makes it Nash equilibrium behavior to offer trust and honor it at each round.
What threats and punishments can be used in achieving this outcome? [Note: for any
number ti, 2-; = 1 p nti = pti/( 1 - p). ]
6. Alternatively, suppose there is a chance p that the party who can honor or
abuse the trust is "trustworthy, " that is, that he or she would always honor trust, even
if it were offered in a single interaction. The person who can offer trust cannot directly
tell if the other party is trustworthy, but, of course, a single instance of abusing trust
that has been offered demonstrates that the individual is not trustworthy. Show that
if L is not too large relative to p, then if the parties are going to interact only twice,
it will be equilibrium behavior for trust to be offered in the first interaction and for
it to be honored, even if the person actually is not trustworthy and will not honor
trust if it is offered at the last round. [In doing so, first calculate how large p must be
to make it worthwhile to offer trust in a single interaction. Now suppose that trust
has been offered at the first interaction and consider the options of the party who
must decide whether to honor the trust. Suppose that honoring the trust results in
the other person going into the last round still believing there is at least a chance of
p that trust, if offered, will be honored. Show that if p is large enough, trust will be
extended at the second round if it was not abused at the first. Now show it is worthwhile
to honor the trust at the first round. Finally, recognizing that trust will be honored
at the first round if p is large enough, determine whether it will be offered. ]
7 . Consider a firm employing two risk-neutral individuals, A and B , each of
whom can choose to exert unobservable effort at a personal cost of c to finding new
investment projects that increase the value of the firm. Suppose that if either individual
 287
exerts effort, there is a probability ½ that he or she will develop a project, and that Rents and
this probability is unaffected by whether the other person tries to generate a project Efficiency
and also whether he or she is successful. Thus, if only one invests, there is a probability
of ½ of a project being generated. If both invest, there is a probability of ¼ that A
succeeds and B does not, a probability of ¼ that B generates a project and A does not,
a probability of ¼ that neither will generate a proj ect and, finally, a probability of ¼
that both will generate projects. Only one project can be adopted, even if both A and
B find projects. A project results in an increase in value of $ 1 .
Determine, for each level of c, whether it is efficient to have none, one, or
both of the two people seek to develop projects.
Suppose that the actual value of c is not known to the managers. In fact, all
they know is that it is at least as profitable for both employees to try to develop projects
as for -neither to do so. Nevertheless, the managers want both employees to try to
develop projects. Design a system of pay that elicits this behavior. Recall that the
employees' effort choices cannot be observed, so that all that pay can depend on is
whether one or both do develop proj ects and, in the latter case, whose proj ect is
accepted. Assume that the total paid out in any circumstance cannot exceed the value
created by the project and that the payments must be non-negative.
Now suppose that the scheme you designed is in place but that employee A can
unobservably lower the chance that B will successfully develop a proposal from ½ to
(I - ZsA )/2 by expending a personal cost of (sA )2/2 on sabotaging B's efforts, and
similarly for B. Suppose each must decide on whether to sabotage the other before
they know if any efforts they have exerted will result in projects. Will the two
employees find such activities worthwhile, given that they have already chosen to
exert the effort necessary to develop projects? Interpret the result. Foreseeing these
incentives, will the employees be motivated by the contract to provide effort to develop
proposals?
 9
OWNERSHIP AND PROPERTY RIGHTS

lli] conom ic growth will occur if property righ ts make it worthwhile to undertake
socially productive activity.
Douglass C. North and Robert Paul Thomas •

The institution of ownership accompanied by secure property rights is the most

common and effective institution for providing people with incentives to create,
maintain, and improve assets. Examples of the effectiveness of ownership incentives
are all around us. Many of the relative inefficiencies of the communist countries of
Eastern Europe and Asia have been attributed to the lack of private property, which
dulled incentives to maintain assets, to innovate and take risks, and to create new
wealth (see Chapter 1). On a more familiar level, people tend to take better care of
their own cars than they do of cars they rent from Hertz or Avis. They drive their
own cars more carefully and protect them better fr.om theft. Similarly, since Avis was
purchased by its employees, the employees reportedly work harder and use more
ingenuity in their jobs than they did before. In fact, however, Avis's employee­
ownership plan raises many questions, most fundamentally: What does it mean for a
group of people to ''own" an asset? and What does it mean to "own" so complicated
an asset as a firm?
Throughout this chapter, our analysis assumes that there are no wealth effects.
With this assumption, the value maximization principle applies (see Chapter 2).

1 The Rise of the Western World: A New Economic History (Cambridge, l\1A: Cambridge Uni\'ersity
Pres�. I 973 ), 8.
 289
Efficient arrangements are then simply the feasible arrangements that maximize the Ownership and
total value received by the parties involved. Property Rights

Tt IE CONCEPT OF OWNEHSI I I P
Economic analyses o f ownership have concentrated on two issues : the possession of
residual decision rights and the allocation of residual returns.

Residual Control
The concept of ownership is complicated, even for simple physical assets. A person
who owns something has certain rights and obligations concerning its use. For
example, if you own a · car, you are free to drive it (provided that you have a license
and obey the traffic laws), to park it (in a legal parking space), to paint and decorate
the car (provided that you keep the windows clear and do not offend the public
morality too greatly), to choose where and how often to have it serviced (provided
that you obey the emissions control and safety laws), to lend it to others for driving
(if the borrower is a licensed driver), to transfer your rights to another party (either
permanently through gift or sale or temporarily through rental), and so on .
Similarly, if you own a company, you have the right to hire and fire its
employees (subject to legal limitations on discrimination and wrongful discharge and
the terms of any employment contracts), to determine the products, prices, and
policies of the company (subject to regulation), to transfer the profits or other resources
of the firm to your personal account (subject to the tax laws and any restrictive clauses
in the company's loan agreements or other contracts), and so on. For economic
analysis, it is often useful to interpret "owning an asset" to mean having the residual
rights of control-that is, the right to make any decisions concerning the asset's use
that are not explicitly controlled by law or assigned to another by contract.
If ownership means having residual control, then its importance must derive
from the difficulty of writing contracts that specify all the control rights. Suppose that
for some particular business relationship it were cheap and easy both to write and to
enforce complete contracts-ones that specify what everyone is to do in every relevant
eventuality at every future date and how the resulting income in each such event
should be divided. In that hypothetical situation, there would be no unforeseen
contingencies, no eventualities for which plans had not been made, no unexpected
gains or losses, and no difficulties in ensuring that the agreed actions and divisions of
income would be implemented. Residual rights would mean nothing then because
no rights would be left unspecified. Nothing would be residual. People could still
write contracts that looked like ownership and assigned ownership rights. One party's
required actions in various contingencies would not be explicitly specified but instead
would be left to that party to decide. Furthermore, that party's payments under the
contract would simply be what was left after others were paid their due. Such contracts
would have no particular advantages, however, because the parties could also write
contracts that specified everyone's actions and payoffs explicitly.
In fact, as we argued in Chapter 5, complete contracts are generally impossible
for transactions of any significant complexity that occur over a period of time longer
than a few days. Complete contracting requires freely imagining all the myriad
contingencies that might arise during the contract term, costlessly determining the
appropriate actions and division of income to take in each contingency, describing all
these verbally with enough precision that the terms of the contract are clear, arriving
at an agreement on these terms, and doing all this so that the parties to the contract
are motivated to follow its terms. Presuming all this were possible, the resulting
 290
Efficient document would, for any but the simplest arrangements, be unimaginably long. Of
Incentives: course, each of the presumed conditions is probably impossible on its own, and their
Contracts and conjunction is certainly so. It is impossible to imagine all the various possible
Ownership contingencies that might develop, let alone enumerate and describe them. Figuring
out in advance the optimal thing to do in each of these cases would require superhuman
abilities. Arriving at an agreement on the resulting contingent plans would require
lifetimes, even if there were no fundamental conflicts of interest between the parties.
Furthermore, no language could ever be sufficiently precise to specify the contract
terms unambiguously, with no scope for disagreement. Finally, the problems of
informational asymmetries can plague contracting (see Chapter 5). The unavoidable
conclusion is that contracts are necessarily incomplete. Consequently, arrangements
that leave all control rights that are not otherwise assigned to a single, distinguished
individual (eliminating the need to negotiate and reach agreement for every unplanned
event) enjoy significant cost advantages.
Although the notion of ownership as residual control may be relatively clear
and meaningful for a simple asset like an automobile, it gets much fuzzier when
applied to something complicated, like a large organization. Large organizations
bundle together many assets, and who has what decision rights may be ambiguous.
Just what rights have been contracted away and to whom? An incomplete contract
may be unclear on this point. For example, do the directors of a firm have the right
to decide to accept a takeover offer without soliciting competing bids? 2 Who has the
rights to make a decision that necessarily affects several different assets with possibly
different ownership? Can a partner in a law firm accept an unsavory client who will
bring in lots of revenues but whose association with the firm may sully the other
partners' reputations? Such decisions may be especially controversial when both the
physical capital of the fi rm and the human capital of its members are involved. We
examine this matter in more detail later in looking at the ownership of a corporation.
An additional complication is that the rights that come with ownership vary
among countries and over time. A firm's owner may have the right to hire employees
but not the right to lay them off or fire them at will. In the United States, the
traditional doctrine allowed employers the right to fire employees "at will, " but this
right has been whittled away by court decisions requiring that terminations be only
"for cause. " The allocation of such rights is important. For example, without the right
to make layoffs when demand is slack, owners will find it more expensive to hire new
workers than they otherwise would, perhaps leading to lower levels of employment
overall. At the same time, if efficient production requires that workers invest in firm­
specific skills, then changes that protect their investments, like improved grievance
procedures or employment guarantees, make them more likely to invest in acquiring
those skills.

Residual Returns
The legal notion of ownership involves more elements than just the control rights we
have described. One particular control right, the right of possession, is often emphasized
as marking ownership. Its main economic consequence is that it allows the owner to
refuse use of an asset to anyone who will not pay the price the owner demands. That
makes it possible for the owner to receive and keep the residual returns from the asset.
These may be direct, current cash flows or changes in the future flows, which then
are reflected in changes in the current value of the asset.

2 I n the Pritzger purchase of Trans Union , the board neither solicited outside bids nor sought the
opinion of an investment ban ker on the firm's value. It ended up losing a landmark stockholder lawsuit,
Smith v. Va n Corkum , 488 A . 2d 858, 876 (Del . 1985 ) .
 29 1
S uch returns are the basis of a notion of ownership used in earlier economic Ownership and
analyses of firms. Accordin g to this notion, the owner of a firm is the residual Property Rights
claimant-the one who is en titled to receive any net income that the firm produces.
That is, the owner is en titled to whatever remain s after all revenues have been collected
and all debts, expen ses, and other contractual obligations have been paid. N et income
is conceived of as the residual return-the amo unt that is left over after everyone else
has been paid.
Like residual control, the no tion of residual returns is intimately tied to
contractual incompleteness. Under complete contracting, the division of the wealth
in each eventuality wo uld be specified contractually, and there would be no returns
that could usefull y be thought of as residual.
Just as the allocation of residual control can be fuzzy in the case of firms
(because rights of control of different categories of decisio ns may be poorly specified
or may lie with various parties), the notion of residual returns is fuzzy as well. One
problem is that the recipients of the residual returns may vary with the circumstances.
When a firm is unable to pay its debts, increases in its earnings may have to be paid
to the lenders. In those circumstances, the lenders would be the residual claimants.
The success or failure of the firm may affect the market's perception of its managers'
abilities and thereby their fu ture opportunities and incomes. Thus, the managers
become residual claimants on a portion of the total returns. Firms may pay bonuses,
increase workers' pay, and promote more workers into higher-ranking, higher-paying
jobs in good times rather than in bad. None of this is contractual, so arguably at least
some of the workers then share in the residual returns of the firm.
For these reasons, ownership of something as complicated as a firm is a tenuous
concept. Our examination of firms in this book is founded mostly in more detailed
analyses of which individuals and groups make decisions, with what informatio n, and
by what decision processes. For simpler assets, however, the notions of residual control
and residual returns are a good basis for beginning the analysis.

Pairing Residual Control and Returns

Tying together residual returns and residual control is the key to the incentive effects
of ownership. These effects are very powerful because (at least in simple cases) the
decision maker bears the full financial impact of his or her choices.

and so on. If some of the parties involved receive fixed amounts of value specified by
Suppose a transaction involves several people supplying labor, physical inputs,

a contract and there is only one residual claimant, then maximizing the value received

all the parties (see Figure 9.1). If the residual claimant also has the residual control,
by the residual cl aimant is j ust the same as maximizing the total value received by

then j ust b y pursuing his or her own interests and maximizing his or her own returns,
the claimant will be led to make the efficient decisions. When it is possible for a
single individual to both have the residual control and receive the residual returns,
the residual decisions made will tend to be effi cient ones. I n contrast, if only part of
the costs or benefits of a decision accrue to the party making the decision, then that
individual will find it in his or her personal interest to ignore some of these effects,
freq uently leading to ineffi cient decisions.
The perspective of residual rights and residual returns illuminates the now­
familiar q uestion of the incentives of an owner of a car versus those of a renter. One
important attribute of the rental transaction is the extreme diffi.culty of performance
measurement-the virtual impossibility of establishing exactl y how much the car' s
value has depreci ated during any particular rental (Chapter 2). For this reason, the
rental company is unable to base its charges on its actual costs. I nstead, it b ases them
on the things it can observe (such as days and hours of the rental, miles driven, and
 292
Efficient
lncenti,·es:
Contracts and
Ownership Residual

Fixed
Obl igations Figure 9. 1 : When some obligations are fixed,
adding to the residual return or value is the
same as adding to the total, so the residual
claimant maximizes total \'alue.

obvious collisi on damage). S uch a charge is necessarily less than perfectly sensitive to
any single actual use and its effects, so careful use is not fully rewarded and rough
use is not fully charged. The one who decides on how the asset is actually used-the
renter-has residual control (for a time) but is not the residual claimant. In contrast,
the owner of a car receives both the residual c ontrol and the residual returns. If you
exercise your right not to mai ntai n your car, then you suffer the dimi ni shed services
it provides and the reduced selling price it eventually commands. For more general
assets, as long as performance measurement is imperfect, a user who does not receive
the resi dual returns is unlikely to take the value-maximizing level of care in maintaining
its value and even more unlikely to do much to add to the asset's value.
Several of the examples in Chapter I illustrate the power of assigni ng re sidual
ri ghts of control and residual returns to the same pe ople. The North West Company's
winteri ng partners had broad rights to make decisions i n the field, and as partners
they collectively claimed the profits that resulted from their actions and deci sions.
This was the key to their succeedi ng against the H udson's Bay Company despite the
latter's huge cost advantage. The bonuses at Salomon Brothers are an attempt to
replicate "high-p owered" market i ncentives inside the firm, and the experiment wi th
payi ng the bonuses partially in stock is an attempt to make Salomon's pe ople look at
the right asset-the firm as a whole-rather than j ust their own departments.
We have also seen examples where a misali gnment of residual rights and returns
have caused trouble. The residual claimant to the returns from an enterprise i n a
communi st system is the state, but the residual decisi on makers are effectively the
enterpri se manager, the workers, and the bureaucrats in the mini stry overseeing it.
None of these has any great personal stake in the value of the enterprise. Another
example comes from the U. S . savi ngs and loan i ndustry, di scussed in Chapter 6.
Those who had the right to control the S&L's investments also had the right to keep
any profits earned but were not obligated to make good on all losses. That combinati on
of ri ghts and obligations created an incentive for risk taking and fraud that was not
effectively countered by other devices during most of the 1 980s.
Properly combi ning the two aspects of ownership-residual control and residual
returns-provides strong incentives for the owner to maintain and increase an asset's
value. I ndeed, in one of the classic papers on the economics of organ ization, Armen
Alchi an and Harold Demsetz treat the essential nature of the firm as a matter of
creati ng this li nkage. They c onsi der a situati on with team producti on, i n which
ou tput is the j oint product of several workers' contri buti ons and the outputs attributable
to any i ndividual arc di fficult to define and certainly hard to observe. This creates an
incc 1 1ti\·c problem of sh irking that, were indivi dual output observable, c ould be easi ly
overcome by giving each worker title to what he or she produces. A solution in thi s
case is for one m ember of the team to undertake a speci alized function-to monitor Ownership and
the other workers an d to have the authori ty to expel members of the team who perform Property Rights
unsatisfactori ly and replace them wi th new mem bers. Alchi an and Demsctz then rai se
a crucial q uestion: "But who wil l moni tor the monitor?" Thei r sol ution is for the
moni tor to be moti vated by recei ving the res idual returns. The res ult is what economis ts
call the classical firm-an organization in whi ch a boss hires, fires, and directs workers
who are paid a fixed wage. The boss recei ves the resi dual returns. Noti ce that this
form of organization only can emerge if property rights are tradable; i t mus t be po ssible
to assign residual control and resi dual return s to the person best s uited to be boss.
Although ownership creates strong i ndivi dual i ncentives, that does no t nece ssari ly
mean that a system of pri vate owners hip is always efficient for soci ety as a whole.
Having a single decision maker bear all the risk i n an asse t's value may be incompatible
with efficient risk s haring and may, in any case, be impracticable if the amounts are
large. In other si tuati ons, increasi ng the value of a parti cular asset may be best
accomplished by undermi ni ng competi tors, setting monopoly prices, or polluti ng the
environment-all activi ties that bring about i nefficienci es.
A prominent school in eco nomi c, legal, and historical scholarshi p has argued
that these li mi ts on the advantages of busi ness contracting, al though real, are ab sent
in the great maj ority of business transacti ons. The chapter- opening q uotati on is an
example of the kind of thinking that characterizes members of thi s scho ol. They
suggest that a system of clear, enforceable, tradable property rights wi ll tend to generate
soci all y efficient outcomes. This position is important to the study of organizations
for two reasons. First, it suggests that once appropriate property rights are establis hed,
there wi ll be a natural te nde ncy for arrangements to evolve toward an efficient pattern
as pe ople rearrange their affairs to capture the mutually available gains. Thi s in turn
has important implications for interpreti ng observed patterns, especially once they
have persi sted for some ti me. Second, even if you rej ect the conclusi ons of thi s school,
the pattern of analysis does provides a useful framework for understanding busi ness
contracts.

Tt IE COASE THEOREM RECONSIDERED

According to the model of compe titive market eq uilibri um presented in Chapter 3 ,
any allocation resulti ng from the competitive market process is efficient: If a complete
set of markets exists, if behavior on these markets is competi ti ve, and if marke ts clear,
then there is no way to reallocate resources that would benefit everyone. The condi tions
of this theorem are really q ui te restri cti ve, however. Certai nly they give only weak
formal support to the posi tio n that any real-world market system wi th private-property
ri ghts is likely to pr omote economic effici ency. Ne vertheless, economists adhering to
the Coasian view would argue that the conclusion that free exchange leads to efficiency
can be extended far beyond the formal purvi ew of the compe titi ve model. Their argu­
ments utilize the Coase theorem and the effi ci ency pri nciple discussed i n Chapter 2.
Recall that the efficiency pri nci ple states that if people are able to bargain
together costlessly and can effecti vely i mplement and enforce their deci sions , then
the outcomes of economic acti vi ty wi ll tend to be effici ent (at least for the parti es to
the bargai n). The Coase theorem states that if the parti es bargai n to an effici ent
agreeme nt (for themselves) and if their preferences dis play no weal th effects, then the
value-creating acti vi ties that they wi ll agree upo n do not depend on the bargaini ng
power of the parties or on what assets each owned when the bargai ni n g began. Rather,
efficiency alone determi nes the activi ty choice. The other fa ctors can affect only
deci sions ab out how the costs and benefits are to be shared.
We saw i n Chapters 5 and 6 that the premises that people can bargain,
implement, and enforce their agreements are not automati cally valid. There can be
 294
Efficient significant transactions costs that arise from bounded rationality, private information,
Incentives: and unobservability of actions. Fully value-maximizing agreements, therefore, may
Contracts and not be reachable. At best, all we can hope for in such circumstances is some sort of
Ownership constrained efficiency that is limited by the difficulties of foreseeing, describing, and
planning, as well as by the need to provide incentives.
Another set of problems may prevent efficient agreements, even when none of

enforceable property rights that can be transferred easily. If there are not, then again,
these other factors are important. These have to do with whether there are clear,

efficiency may not be realized. If no one clearly owns a valuable asset, then no one
has an incentive to guard its value properly. If property rights are not tradable, then

use of them and so value them most. If property rights are not secure, then owners
there is little hope that assets will end up with those people who can make the best

will not invest great amounts in assets that they may lose with no compensation, or
they may sink valuable resources into protecting their claims.
These responses to insecure property rights give an efficiency cost to theft, which
at first glance might seem to be only a transfer without any direct efficiency
consequence. The reluctance to invest when property rights are insecure results in
inadequate maintenance and development of assets. Car owners in large cities who
refrain from installing radios in their vehicles for fear that they will be stolen are
foregoing a transaction that would be valuable if their rights were secure. More
importantly, there is some suggestion that the threat of government expropriation
without adequate compensation has hindered investment and development in parts
of the Third World. Meanwhile, the resources sunk into protecting assets from theft
bring no direct social returns. Security guards and burglar alarms are therefore
considered social waste.

Ill-Defined Property Rights and the Tragedy of the Commons

One of the saddest incentive problems resulting from untradable, insecure, or
unassigned property rights is known variously as the common-resource problem, the
public-goods problem, the free-rider problem, and the tragedy of the commons. The
idea is that when many people have the right to use a single shared resource, there
is an incentive for the resource to be overused and, correspondingly, when many
people share the obligation to provide some resource, it will be undersupplied. When
the residual returns to an asset are widely shared, no one person has a sufficient
interest to bear the costs of maintaining and increasing its value. In such cases,
concentrating the ownership rights can lead to increased efficiency. Fishing rights
provide a good example of this phenomenon.

THE EcoN0l\llCS OF OCEAN FISHERIES In order to maintain the stock of fish in a given
area of the oceans, a large enough proportion of each generation must survive for
breeding. Next year's population is, over the relevant range, an increasing function
of this year's. However, when large numbers of people all have the right to fish
commercially in the same area without effective limits on the amount each can take,
there is a danger of overfishing, with too many fish being taken to permit the
population to recover. Any smgle boat owner who limits his or her catch (and income)
to preserve the overall population bears all the costs of forebearance. Meanwhile, the
benefits arc spread over all the boats fishing the area that will have a larger harvest
next year, with only a small fraction of the gain going to the one who conserved.
Moreover, modern fishing methods have made it technically easier to harvest immense
numbers of fish, so overfishing is an increasing danger.
There arc a variety of ways in which assignment of ownership or property rights
might alleviate this problem, including group ownership and single ownership. (In
 295
order to make such assignmen ts, many coun tries have been claim ing con trol of the Ownership and
fisheries up to 200 miles off their coasts, even while disp uting others' rights to do the Property Rights
same.)

The Group Ownership Approach One way to combat the problem of overfishing
would be to vest a fishery association made up of the members of the local fishi ng
fleet with the exclusi ve right to control the fishery. This would include rights to
determine who may fish in the area, the total catch size, and vari ous rules affecti ng
the methods of fi shing, such as the hours or seasons when fi shing is permitted, the
mesh size of nets, and so on. Individuals could then be assigned rights to partake i n
the total allowed catch i n various ways. For example, they could share the total
realized catch equally, or each could be assigned a quota limiting the n umber of fish
he or she can take.
The relative advantages of this ki nd of arrangement are easy to see. The
i ndi vidual rights pr ovide i ncentives for each boat to gather its catch efficiently because
any extra resources used or time wasted are costs the boat's owner bears alone. As a
group, the associati on has a collective i ncenti ve to safeguard the fish populati on. In
fact, from the group's point of view, the most advan tageous fishi ng arrangements may
be those that maximize the j oi nt profits of the fishermen, taking i nto account the full
effect that this season's harvest has on future harvests. Of course, the associati on also
may have an i ncentive to limit the catch i n order to dri ve up fish prices, but i n many
regi ons that is onl y a tri vial problem because consumers have so many close substitutes
for the local fish. In that case, the catch that maximizes the present value of the fleet's
net income is the socially efficient one.
This scheme is not without i ts problems, however. First, i t may be costly to
keep out i nterlopers who would want to fish although they have no right to do so.
The fact that the state might bear most of the costs of enforci ng the association's
property right i n no way changes this.
Second, there may be severe moral hazard problems among the i ndi vidual
members of the associati on, depending on how the catch is shared. If a quota is used,
with each boat keeping the revenues from the fish i t bri ngs in and payi ng the costs of
catchi ng them, there will agai n be an i ncentive for overfishi ng by cheati ng on the
quota. H owever, these i ncenti ves will be somewhat attenuated by the fact that the
benefits of conservati on are di vided among a smaller gr oup. M oreover, enforci ng
quotas may be easier i n the limited gr oup because monitori ng fishi ng seasons and
inspecting catch sizes and types of nets is simplified when all the boats operate out of
the same few harbors. Furthermore, the possi bility of denyi ng fishi ng ri ghts to anyone
caught cheati ng provides i ncentives for obeyi ng the rules. S till, resources must be
expended to enforce the rules.
In contrast, i f each association member is given a fixed share of the total catch,
then the incentives are to cut back on effort, crew sizes, and so on, because all the
savings accrue to the i ndi vidual but the costs i n terms of a smaller catch are spread
over the whole fleet.
A third difficulty with vesti ng the fishi ng ri ghts with the fleet as a group is that
there may not be unanimity among the members about the appropriate rules to govern
the exploi tation of their fishery. Fi shi ng boats differ in their costs of fishi ng, and their
owners may vary i n their expectations about future fish prices and their outside j ob
opportuni ties, which together govern how rapi dly the y think i t appropriate to harvest
the fish populati on. They may also have differi ng ideas as to what rate of fishi ng the
population can sustain. These differences mean they will disagree about possi ble
policies. Then all the costs of political group decision making will be incurred, and
the i nefficiencies we discussed in Chap ter 5 may be incurred.
 296
Efficient Single Ownership An alternative that avoids this problem of inefficiencies is to
Incentives: assign the rights to a single individual-presumably, but not necessarily, one of the
Contracts and members of the fishing fleet. This is the "single-owner" solution that works so well
Ownership for ordinary items like automobiles, houses, and furniture. When applied to fishing
rights, it has all the same general advantages as the system that allows the community
to determine the size of the total catch and fishing rules. The single owner would
have reason to be concerned about the next year's fish harvest and to avoid the
overfishing that would destroy it. Like the fishery association, he or she would have
good reason to weigh the advantages of catching extra fish in the present season against
the cost of a reduced future harvest. With these factors in mind, the owner of the
fishing rights would choose seasons, mesh sizes, and so on, to maximize the (present
value of the) profits to be earned during the period of ownership.
The major problem with assigning all the rights to one person is deciding who
the lucky one should be. Rotating the rights among the possible candidates from year
to year would be terribly inefficient; it would encourage overfishing because the
current holder of the rights would not gain at all from preserving the population for
the next year. Another possibility would be to auction the rights to the highest bidder
and compensate the losers by distributing the proceeds of the auction among them.
This would ensure the most efficient use of the rights, especially if the winner can
then charge others for use of the rights or can hire others to help gather the catch
(and, in either case, can induce efficient behavior from them), because the winner
will be the one for whom the rights are most valuable. Because the value of the fishing
rights varies from person to person, however, perhaps in unmeasurable ways, it would
be difficult to determine appropriate individual compensation for the lost rights. This
again opens the process to the inefficiencies of bargaining under private information
and may make it very difficult to win agreement for the plan.

PETROLEl 11\I DEPOSITS AND AQL 1 IFERS The analysis of fishing rights is a very particular
one and does not apply exactly to any other situation. Still, the basic patterns in the
analysis of attention to detailed specifications of rights, to the sorts of options available,
and to the incentives involved all arise repeatedly in analyzing other problems. Two
closely related situations concern underground pools of petroleum and ground water.
An underground petroleum deposit can stretch for a great distance under the
earth's surface. If ownership rights are not established over the pool, we have a free­
rider problem closely paralleling that with the fisheries: Each party tries to draw oil
out quickly so that the others do not get it all first. This excessive pumping raises the
costs of extracting the oil and leads to a too rapid exhaustion of the resource. In fact,
the common practice is to assign mineral rights to any petroleum located under a
piece of the earth's surface to the owner of the land above. Often this means that
different parties may hold the mineral rights to different pieces of land above the same
pool. These rights give the parties title to any crude oil that they draw to the surface
through wells dug on their properties. However, a well sunk into any point in the
pool tends to draw crude from across the whole deposit as the petroleum flows, albeit
slowly, to the region of reduced pressure. This means that if different interests have
rights to draw oil from a single pool, there is still a tendency toward rapid extraction.
All this is exacerbated by the possibility of sinking a well on one piece of property but
drilling on an angle so that it hits the petroleum deposit under another's land. The
results can be disasterous-lraq's anger about Kuwait's alleged overpumping and
poaching in oil fields straddling the two nations' border was a major element leading
to the Persian Gulf War of 1990-199 1 .
Very similar problems arise with underground aquifers. These pools o f under­
ground water in porous rock formations may stretch over hundreds of miles. Aquifers
 297
are a major source of water for drin kin g and irrigation . In the western Un ited States, Ownership and
a number of aquifers are bein g drawn down much more quickly than they arc refi lled Property Rights
by natural seepage from the surface. The problem is j ust as with fisheries or crude
oil: No on e owns the aquifer, so n o one guards it pr operly. It is especially acute in
Californ ia, the on ly western state with no program of ground-water management.
Except in two small areas with regional contr ols, an yone wh o sin ks a well can take
as much water as comes out. In 1991 Californ ia farmers were expected to draw down
the amount of water in the aquifers by 4. 9 trillion gallon s. 3 The results are subsiden ce
of the ground surface-some areas in the city of S an J ose san k eigh t feet before local
con trols were in stituted-as well as pollution of the aquifers by seepage of sea water
an d pollutan ts. There is also the potential that the dried- out porous structure will
collapse, perman en tly destroyin g the aquifer.
OWNERSHI P PROBLEMS IN THE SOCIALIST COUNTR IES S ome of the reforms that took
place in China during the 1980s (before th e bloody crackdown of 1989) can be
un derstood in similar terms. Factories in Chin a h ave been mostly either state own ed
(by the n ational govern ment) or collectively own ed (by the factory workers or the
community). N o sin gle in dividual was h eld accoun table for performan ce in these
factories. Indeed, empl oymen t and wages were pr otected in Chin a, an d the auth ority
of factory managers was limited. As in man y other communist countries, losses
in curred by these factories were covered by in terest-free "loans" fr om the state, but in
practice these loans were seldom repaid. They represen t a subsidy paid (through taxes)
from profi table factories to un pr ofitable ones.
Attempts to hold Chinese factory man agers accoun table with out creating a
managerial class have led to experimen ts with rotating managers. According to our
an alysis, these were doomed to failure for th e same reason s that the rotatin g property
rights solution to the overfi sh in g problem cann ot succeed. Managers with sh ort tenure
will avoid makin g expen sive in vestmen ts in new equipmen t, R&D, worker train in g,
and so on, because the benefits of these investments are en j oyed only by others at a
later date. Chinese reformers have also raised the possibility of using long-term factory
leases patterned on the lon g-term leases on Chin ese farms. These were seen as an
ideologically safe way to tran sfer effective own ership rights to private parties and
thereby provide incen tives. Their effects on agricultural pr oductivity have been very
positive.
In 1990 a str uggle emerged in the S oviet U nion over Presiden t Gorbachev's
proposals to grant lon g-term leases to ten an t farmers in order to rein vigorate the
notoriously in efficien t farm sector of that economy. Ideological resistan ce to that
reform cen tered on the belief that long-term leases are really just a disguised form of
private own ership of the lan d. Yet until something is done to improve in dividual
accoun tability for performance, it seems unlikely that S oviet farms can become as
productive as their western coun terparts.

Untradable and Insecure Property Rights

One of the essential aspects of ownersh ip rights for ordin ary goods is that th ey can be
bought an d sold. Accordin g to the Coase theorem, assets will then ten d to be acquired
by th ose wh o can use them best. If a person own s goods or righ ts that are more
valuable to his or her n eigh bor, then there is a price at wh ich the two woul d both
fi nd it profi table to trade, and the goods move to where they are most valuable.
Ownersh ip rights are n ot always tran sferrable, h owever. Th is can interfere with

1 Lisa Lapin, "Subterranean Ocean Could Be Drying Up," San /ose Mercury News (April 7, 1 99 1 ),
A- 1 .
 298
Effi cient efficient resource use because the assets then do not get put to their best use (unless
Incentives: by chance they are assigned to those who value them most). A second problem arises
Contracts and when property rights are insecure, so that they might be restricted or lost at some
Ownership future point without fair compensation being paid. Insecure rights weaken the owner's
incentive to invest in developing and maintaining the asset because there is always
the danger that both the asset and the returns it generates will be lost. We discuss
both these problems in the specific context of water rights in California, although the
features of the analysis can be widely applied to show how patterns of ownership and
legal doctrines concerning owners' rights affect the efficiency of economic performance.
WATER USE IN CALIFORN IA The state of California is largely a desert, although this
fact is often forgotten, masked by the state's population of 30 million and its position
as the eighth-largest economy in the world with one of the world's largest and most
productive agricultural industries. In most parts of the state, rain falls only during the
short winter season, and the natural vegetation is adapted to the arid climate. Winter
snowfall in the mountains of northeastern California normally provides a huge annual
water runoff, however, most of which naturally would flow into the Pacific Ocean
via the rivers draining into the Central Valley, through the deltd of the Sacramento
River, and into San Francisco Bay. This runoff, diverted by dams, reservoirs,
aqueducts, and pipelines, provides most of the water used by farms, residents, and
industry throughout California. Although 7 5 percent of the state's water is located in
the northern part of the state, 7 5 percent of the population is in the southern part.
For many years, residents of northern California have been concerned about
the damage done to the environment by withdrawing so much water from the natural
flow patterns. Reduced water flowing into the Sacramento River delta has allowed
salt water from the ocean to reach further inland, wreacking havoc with natural
habitats and luring lost whales far inland. The water flow into the San Francisco Bay
from melting snow in the mountains cleanses the bay and keeps it from becoming
stagnant and accumulating pollutants; reduction in the water flow reduces the cleansing
action, harming the environment of the bay. Compounding the environmental
problems, droughts in northern California in the late 1 970s and 1980s have endangered
water supplies to industries and urban residents there while water continued to flow
to the southern half and agriculture. Many urban and suburban Californians look
with envy on the water rights afforded farmers.
Approximately 8 5 percent of the water actually used in California is employed
in agriculture, nearly all of it for irrigation. All other industries, as well as all residential
use, account for only the remaining 1 5 percent. Agriculture, however, represents only
about 3 percent of the state's aggregate output of goods and services. The water
delivered to farmers comes from various sources, some quite expensive, overseen by
a variety of federal, state, and local public agencies.
PROPERTY RIGHTS FOR C..\ LIFOHNL\ \�'.\TER Water rights for farmers vary considerably
throughout the state of California, depending on the source of the water. Some
farmers have inherited extremely valuable rights to have huge quantities of cheap
water delivered to them. Water for farming from the federal Bureau of Reclamation
sells for $ 1 0 to $ 1 5 per acre-foot, and the cheapest subsidized water sells for as little
as $ 3. 50 per acre-foot, even though it may cost $ 1 00 to pump this water to the
farmers. -+ At these prices, it is economical to grow cotton and rice in the desert, and
California farmers produce large amounts of both crops (and receive subsidies to do

4 An acre-foot is the amount of water necessary to cover I acre to a depth of I foot, or about 3 2 5 , 800
gallons. It is the standard unit of measurement for water in such contexts. As a basis for comparison, the
average suburba n California household uses about half an acre-foot annually.
 299
so). Mean while, households in Pal o Alto pay ab ou t $65 for the same quantity of Ownership an<l
water, an d some urban water users pay as much as $230. The mos t desperate Property Rights
n onagricultural commun ities al ong the Pacifi c coast of Cal iforn ia have gone so far as
to build desal ination plan ts to obtain potable water from the ocean at a cost of
approximately $3,000 per acre-foot.
H ow much of the cheap water is used? On e agricul tural use al one, irrigating
pastures for grazing cows an d sheep, used 5. 3 mill ion acre-feet of water in 1986. 5
This is en ough water to cover the District of Columbia to a depth of 1,250 feet! It is
also more water than was con sumed by all the households in the state for all purposes,
including filling swimming pools an d hot tubs, watering lawns, an d washing cars. Yet
the industry of raisin g cattle and sheep on irrigated pasture in Californ ia had gross
revenues for that year of less than $100 mill ion. 6 Pl ainl y, devoting so much water to
such a. l ow-value use is possible onl y becau se the water used to irrigate pastures is sold
so cheapl y.
THE LIM ITED TRADABILITY OF WATER RIGHTS Why then does the water n ot fl ow to
higher-value uses? Part of the an swer is that man y farmers' right to the water is l imited
an d n ot freel y tran sferrable. Farmers are n ot generally all owed to sell their water to
the highest bidder.
What would happen if these ran chers and farmers owned the full rights to bu y
particul ar amoun ts of cheap water an d then could resell the water (or the right to bu y
it cheapl y) to others? Suppose, as most evidence indicates, that the marginal acre-foot
of water really is more valuable to residen tial an d industrial users than to farmers and
ranchers. (Why el se would they pay $3,000 for an acre-foot of water that a farmer
gets for $3. 50?) In that case, some farmers an d ran chers would find it profi table to
cut back their pr oduction, chan ge the crops they raise to l ess water-in ten sive on es, or
switch to more effi cien t methods of irrigation. They could then sell rights to the water
they would save to other users who value it highl y. A ccordin g to the Coase theorem,
efficien t use of water is what should be expected if water rights are secure and tradable.
Water rights in Cal iforn ia are n ot assigned in that way, however. The rights are
not tran sferrable. Thus, the farmer wil l u se water un til the marginal units are worth
only the l ow price he or she pays, whil e at the same time those marginal gallon s
woul d be worth much more to someone el se.
In fact, the incen tives may be even more perverse than this. For those receiving
the cheapest water, there is sometimes a "use it or l ose it" rule: If a farmer is en titled
to use a certain number of acre-feet of water bu t uses less, perhaps by switching from
cotton to fruit orchards, then his or her cl aim to the water is l ost. To the exten t that
havin g an assured su ppl y of cheap water in the fu ture is valuabl e, the farmers might
use up their full all otmen t even when it is worth l ess to them at the margin than the
price they are required to pay.
The problem is related to the one that often arises when water is ration ed in
urban areas. In times of drought, users may all be required to cu t back equally. S o,
for exampl e, if a famil y conserves water when suppl ies are plentiful, then in a drought
year it will be asked to reduce its water con sumption from a small base. This can be
done onl y at great personal hardship. Meanwhile, an other famil y that has been
wasteful fi nds it easy to reduce its use by the required fraction or amoun t. I t is n ot
hard to see that these systems for all ocating water on the basis of past water con sumption

5 Marc Reisner, "The Next Water War: Cities Versus Agriculture," Issues in Science and Technology
(Winter 1988-89), 98-102.
6 To put this in figure in perspective, Stanford University currently receives about twice this amount
in donations each year.
 300
Efficient create incentives for wasting water. (We encounter this same sort of problem in
Incentives: Chapters 1 and 7. There, the tendency of organ ization planners to base this year's
Contracts and performance standards on last year's actual performance creates the ratchet effect that
Ownership leads to lower levels of performance-enhancing efforts . )

INSECURE WATER RIGH TS As we already mentioned, making existing rights transferra­

ble would potentially improve matters. Farmers would sell their cheap water to city
dwellers or industries who value it more, and both parties would be better off as a
result. Alternatively, the water rights could be given to the city dwellers, perhaps
collectively, city by city, and these people could sell them to the farmers if the value
of the water in raising food exceeded that in washing cars or other urban uses. The
Coase theorem says that this initial assignment of rights is immaterial to whether
efficiency will be achieved, so long as the rights are well defined, secure, and tradable .
The great problem i s to fi n d a politically palatable way t o change water rights s o that
this limited resource can be used more efficiently.
Simply assigning rights to cheap water to farmers would be problematic. As one
California legislator put it, "Taxpayers have spent billions of dollars to develop the
water resources, the canals, the dams, for these farmers. Now to suggest that the
farmers can turn around and sell that water for a profit is a major insult to the
taxpayers. "7 Many farmers in fact oppose making water rights transferrable, fearing
that any alteration in the current arrangements would ultimately result in loss of their
water rights without any compensation .

W I NDS OF C H ANC E In 1 99 1 , as California entered its fifth consecutive year of drought

and major cities struggled with severe water rationing, legal barriers to water trade
between farming and urban uses began to crack. The Los Angeles Metropolitan Water
district was particularly active in designing novel arrangements. Where possible, it
paid some farmers to let their fields lie fallow during drought years, in return for
which the district could take unused waters. It also invested in improvements in the
nearby farm counties' aqueducts, reducing water lost to evaporation , leaching, and
leakage, and receiving in return the extra water made available by these projects.
The story of Californ ia water rights told here illustrates several themes of this
book. Nontradable property rights lead to inefficient uses of resources, making
institutional change desirable. However, legislative attempts to reform the laws about
water rights also open the issue of the how extra value created by the reforms are to
be distributed . The resulting influence costs are very high, and the threat to existing
farmers has led many people to oppose any proposal for value-increasing reforms.
Furthermore, the large number of people affected and the variety of their interests
makes it more difficult to reach a mutually agreeable bargain. The costs and difficulty
of bargaining block application of the efficiency principle, allowing an inefficient
arrangement to persist for many years. Nevertheless, the increased cost of the
inefficiency during a period of drought brought new, stronger pressures for permanent
change in California, perhaps to a far more efficient system.

Bargaining Costs and the Limits of the Coase Theorem

A key premise of the Coase theorem is that the costs of arriving at and enforcing an
efficient agreement are low. From this follows the conclusion that resources will be
allocated efficiently, even in the absence of competitive markets and regardless of the
initial allocation of property rights or distribution of bargain ing power. In this section

7 Representative George M iller, as quoted in Lisa Lapin, "Perestroika for Water: Drought Inspires
Revol ution , " San Jose Mercury News, (February 3, 1 99 1 ), A- I and A- 1 4
 :10 1
we i nvestiga te reasons why there might be significa nt impedime nts or costs to reaching Ownership and
and executing efficient agreem ents a nd exam ine the na ture of these " bargaining costs. " Property Rights
This discussion relies hea vily on the material on bou nded ra tiona lity a nd priva te
informa tion as discussed in Cha pter 5.
Identifyi ng the maj or sources of bargaining costs serves three fu nctions. First,
it helps us to identify where to expect ineffi ciencies to be found. S econd, it helps to
explain a whole ra nge of practices a nd institutions, including those that arise to
minimize bargaining costs and certain government regula tions that seem designed to
achieve as much as possi ble in a n envi ronment where bargaining am ong individua l
citizens would be too diffi cult, complex, or costly to be a realistic possibility. Fina lly,
it lays a founda tion for the new theory of law a nd economics. If the alloca tion of
property rights does affect value, then one possible obj ective for the law of property
ri ghts is to assign them in a way tha t creates value.

Legal Im pediments to Trade

S ome of the m ost obvious im pediments to trade are legal ones. In our discussion of
water rights, we have seen how unassigned, ill-defi ned, nontra nsferrable, insecure, or
unenforcea ble property rights ca n stifle effi ciency. To the extent that the political a nd
legal system s are supposed to determine a nd protect property rights, the ineffi ciencies
arising from faulty property rights are traca ble to these spheres. M oreover, the political
and legal systems sometimes deliberately erect barriers to excha nges when it is believed
that allowing certain trades would be m orally undesira ble. The U.S. Constitution's
ba n on involuntary servitude not only prevents a person from selling himself or herself
into slavery but also ha s been interpreted as limiting some less drastic long-term la bor
contracts. Y ou are sim ply not perm itted to tra nsfer your property rights in your person,
even if you believe tha t the deal being offered is a good one. The limitati ons or
outright prohibitions on the sale of certain chemicals (such as alcohol, cocaine, LSD,
and heroin), books a nd videos (pornography a nd seditious material), a nd personal
services (prostitution a nd assa ssination) also have a largely m oral rationa le.
In other ca ses, various factors might interfere with consummati ng a mutually
profi table bargain to tra nsfer property rights, even when such a bargain might exist
and be ethically accepta ble. These impediments a nd the responses to them provide a
basis for understa nding som e aspects of effi cient institutions.
COSTS OF AGREEI\IENT AND ENFORCEI\IENT There are various costs of a greem ent a nd
enforcement. First, there are sim ply the costs of determining, writing, and enforcing
an agreement i n a world of bounded ra tionality, im perfect communication, priva te
information, observation a nd verifica tion diffi culties, a nd opportunism . The releva nt
parties have to be identifi ed. They m ust reach agreement on the basic phy�ical
relationships that tie actions to outcomes a nd determ ine a set of pla ns in light of the
m odel of the environment. Then the actions that each party is to take a nd the
distribution of the resulting returns must be determ ined, accepted, a nd specifi ed with
sufficient clarity that all ca n understa nd them a nd know wha t to do a nd what to
expect. Finally, some mecha nism must be put in place to enforce the agreement a nd
to resolve disputes.
In some situations these costs are small com pared to the benefi ts that result:
Transactions do get made! In others, however, they ma y be large or even overwhelm ing.
For example, it would be quite costly for the passengers on a n aircraft to negotiate
whether the flight should permit sm oking. If sm oking is not permitted, should sm okers
be compensated for being deprived? H ow should compensa tion be determined? For
exam ple, should the person who com plains loudest get the largest pa yment? Think
of the opportunities for misrepresenta tion: N onsm okers might even claim to be
 302
Efficient smokers to get compensati on! Wh o should pay the compensati on-th e airli ne or th e
Incentives: nonsmokers? If i t i s the nonsmokers, how much should each of them pay? H ow is
Contracts and thi s to be determined? Alternati vely, if smoki ng i s permitted, sh ould nonsmokers be
Ownersh ip compensated? Sh ould they be charged less by the ai rli ne because they are putting up
wi th a foul atmosphere? What prevents a smoker from payi ng the lower fare then
surreptiti ously li ghti ng up?
As another example, i t would be overwhelmingly costly for all the dri vers and
pedestri ans in even a small communi ty to trade ri ghts governi ng road use, such as
wh o can drive how fast, who has the right- of-way at intersecti ons, and so on. If there
were no costs of enforcement, a value-maxi mizi ng agreement would generally treat
different ci ti zens differently. Dri vers wi th superi or drivi ng abi li ti es and a taste for fast
cars mi gh t be permi tted to dri ve faster (and mi gh t be required to pay for that right),
whereas others mi gh t be required to drive slowly and yield the right of way to all
concerned. H owever, there are too many parties affected for them to successfully
negoti ate and enforce rules and compensati ng payments. Even determi ni ng an efficient
contract would i nvolve immense costs and would most li kely be impossi ble. Which
drivers under which conditi ons should be allowed to drive at which speeds? To which
oth er drivers and pedestri ans must they yield ri ght of way and under which conditions?
When and for whom is the benefi t of parki ng in a traffi c lane worth more than the
costs i n terms of danger and congestion that are imposed on others? H ow is all thi s
to be determi ned? M oreover, the costs of enforci ng the agreement would be immense.
Is the car approaching me bei ng driven by A, to whom I sh oul d yield at this ti me of
day at this corner, or by B, wh o should yield to me? Was that A wh o drove by at an
effi ci ent 85 miles an hour, or B, wh o i s supposed to drive at 25?
To economize on these transacti on costs, people seek other organ izati onal
solutions. Airlines make rules about smoki ng on ai rcraft, and governments about safe
dri vi ng practi ces, even though , i n fact, a dri ving speed that is safe for one dri ver may
be too fast for another. A similar analysis j ustifies the creati on of laws that specify
how property owners must mai ntai n the public si dewalks crossi ng their properties and
the damages that must be pai d if the laws are violated and an i njury results, for
example, if a passerby sli ps on the snow of an unshoveled walk. The laws governi ng
product safety can be analyzed si mi larl y. These laws mostly speci fy ri ghts that are not
tradable; a person often cannot accept a lower safety standard for a product i n exchange
for a lower pri ce. One reason for this i s to prevent private recontracting that would
i nvolve excessive costs in enforcement (borne i n part by the government through its
system of courts).

Is E\'ERYTH l'.'JC WE SEE EFFICI ENT? A common obj ecti on to the argument that these
costs are impedi ments to effici ent agreement is that the costs of negoti ating and
enforci ng contracts are, i n fact, costs, and should be treated as such. If these costs are
recognized, apparent inefficiencies then di sappear because the reason a " better"
outcome was not achieved is that the transaction costs of achievi ng i t are too high.
This line of argument has some appeal: It seems i nappropriate to ignore the costs of
running the organization, especially when the organizati onal structure is a ch oice
vari able. Furthermore, in cases like the smokin g and traffic law examples, where the
problem ari ses frequently, people may devise rule-making or law-maki ng i nsti tutions
that are intended to resolve the si tuation at low cost. In i ts most opti mistic form, the
h ypothesi s becomes not j ust that ownership pattern s ten d to be effici ent in terms of
the usual measures of costs and benefits, but that econ omic activity generally is
organ ized so that the excess of be nefi ts over total costs (that i s, the total value) is
ma xim ized.
Th is hypothesis, alth ough pro vocative and informative, is surely wrong. For
example, suppose that two firms wo uld suffer a pretax loss by merging their businesses Ownersh ip and
but would enj oy an aftertax gain. Would they merge? To do so would be inefficien t Property Rights
beeause the total profits (before tax) would fall. The reason for merging would be
sim ply to reduee the governme nt's share of the profits. Since the govern ment is not
likely to be an active partner in these negotiations, the merger m ost likely will go
through beeause the outeome is efficien t for the firms even though it is n ot effieien t
for society as a whole.
A similar problem arises whenever there is multilateral bargaining with one
bargainer who eannot realistically partieipate in the negotiati ons. It may also arise
when there are multilateral bargains when all can participate; two or more bargain ers
may gang up on the third in an inefficient way to hold him up more "effieiently. "
For example, firms may negotiate an agreement to raise prices above their competitive
levels., That outcome benefits the firms, but consumers will be damaged by the
conspiraey. Furthermore, the firms' gain is less than the consumers' loss from the
price increas�. Because of the costs to the eonsumers of participating in the agreement
or of enforeing any agreement they might negotiate, however, there is no assuranee
that the ultimate agreement will maximize total value.
Despite these difficulties, transaction costs are relatively low in many situations.
Thus, in the following ehapters, where we examine the applieation of these ideas to
a range of managerial and organizational problems, we assume that private arrange­
ments tend to be effieient for the parties involved, at least given the information- based
ineentive eonstraints. M ore importantly, we assume that when significant transaction
costs impede private eontraeting, people seek and often find innovative, sophisticated
institutional arrangements that minimize or completely avoid these costs.

Transaction Costs and the Efficient Assignment

of Ownership Claims
When transaetion costs are low, the Coase theorem holds that the initial assignment
of property rights is irrelevant to efficiency, so long as the rights are clearly assigned,
secure, and transferrable. When there are significant impediments to effieient bargain­
ing, however, it may be crucially important that these rights be assigned properly
initially. Otherwise, with little prospect of private bargaining reassigning these rights
effectively, they can end up being very badly allocated. Of course, in well-functioning
societies, the initial assignment of property rights is typically a task of government.
CREATING, ALTER ING, AND ASSIGNING PROPERTY RIGHTS Even the most ardent advocate
of laissez faire (noninterference) would likely concede a role for government in
protecting property rights. H owever, governments also create new property rights
where none were defined before and modify and transfer existing rights.
Governments create property rights, for example, through the patent, trademark,
and copyright laws. These give title to the realization and expression of ideas to their
creators. The holders of a patent can use the courts to prevent others from using their
invention without compensation; the authors of this book had to get the permission
of the eopyright holders to use the various quotations that appear in this book. These
property rights in fact result in short-term inefficieney. The use of your idea by
someone else does not wear it out, use it up, or deprive you of its use. O nce the idea
is created, it is efficient to have it employed as widely as possible because there is no
opportunity cost to its further exploitation. This solution would mean that inventors
a nd developers would receive only a tiny fraetion of the returns to investm ents,
however. This would remove much of the incentive for creative and innovative
activity, and the long-term impacts would be devastating.
Governmen ts also often alter or remove existing property rights. In the 1970s
 304
Efficient California changed the property rights of the owners of oceanfront property. These
Incentives: owners are now extremely limited in the ways they can develop their properties, and
Contracts and they have to provide access through their private land to the public beaches. In another
Ownership recent example, new buildings in increasing numbers of political jurisdictions are
now required to be accessible to people in wheelchairs. The property owners cannot
choose to develop their properties as they might prefer and would formerly have been
able to do. Similarly, people formerly had the right in the United States to smoke in
the designated sections of com mercial aircraft on domestic flights or, rather, the
airlines had the right to permit this behavior on their planes. This right has been
removed by a recent federal law: Smoking is now forbidden on domestic flights, even
if everyone on the flight is a smoker. (In the interests of efficiency, however, the flight
crew is still permitted to smoke: The lawmakers were concerned that nicotine
deprivation would adversely affect the pilots' performance. )

ASSIGNING RIGHTS EFFICIENTLY : POLLUTION At one time people essentially were free
to pollute the environment to get rid of their wastes, without charge. Although no
document gave them a clear legal right to do so, no party had a property right to the
air and water that the polluters used: No one person could go into court and sue
because a town's sewage was being dumped into the river that passed his or her house
or because the air he or she breathed was full of toxic matter. In any case, this initial
situation was not efficient because it led to serious environmental degradation: filthy
air, contaminated soil, and poisoned water; acid rain that is killing forests and lakes
far from the pollution sources; and global warming and destruction of the ozone layer
that may threaten continued life on the planet. Arguably, these costs far exceed the
gains to the polluters from dumping wastes into the environment rather than not
creating them or disposing of them in other ways.
One possible solution would be to assign a clear property right to the polluters.
The Coase theorem asserts that if bargaining were costless, this would lead to an
efficient solution. In fact, however, the results are very likely to be inefficient. Polluters
with clear rights to pollute would presumably be willing to sell their rights if the cost
of disposing of their wastes in a more environmentally friendly way were less than the
price they were offered for their rights. Those who are harmed by pollution place
some value on having a cleaner environment, and if they could bargain costlessly
with the polluters, they would presumably strike a value-maximizing deal that would
result in the efficient amount of pollution. The transaction costs of carrying out this
exchange are tremendous, however, and they make the value-maximizing bargain
unattainable.
What are the sources of the bargaining costs? First, the problem of identifying
the affected parties is very great, given the uncertainties about the long-term effects
of various sorts of pollution. When these effects cross national borders, the problems
become even more severe. Next, the free-rider problem among the victims of pollution
is overwhelming. To know whether it is worthwhile to eliminate a particular source
of pollution, we must determine the values that all the affected parties put on having
the pollution gone. If the parties will be asked to pay according to what they say it is
worth to them, they will have an incentive to free ride by underreporting, hoping that
others will foot the bill. If they will not have to pay more when they say that they
would benefit greatly from less pollution, then they will overstate their gain to ensure
that the transfer takes place. The difficulties of achieving value-maxim izing agreements
with private information about valuations, especially when there are large numbers
of people involved, suggest that this bargaining is unlikely ever to lead to transfers of
pollution rights, even when they would be efficiency enhancing. Finally, there would
be the costs of m onitoring the agreement to ensure that the polluter \Vas not cheating.
 :m5
In these circumstances, assign ing rights to polluters would not lead to an efficient Ownership and
outcome. It is better to lodge the rights to the environment with the public at large Property Rights
and to have the government enforce these rights through the legal and regulatory
systems.

The Ethics of Private Property

Thus far we have treated property rights solely as a matter of economic efficiency,
but the institution of private property has ethical signi ficance as well. On the one
hand, supporters of market systems see private property as a fundamental right, not
just an expedient mechanism for generating incentives. On the other hand, many
people worldwide believe that private property is fundamentally immoral. As the
nineteenth-century socialist P. J . Proudhon succinctly put it: "Property is theft!" This
view still ha� wide acceptance today. As well, in some cultures, the idea of private
property is quite circumscribed, and some religions forbid private property altogether.
Among many of the native peoples of North America, the idea that someone could
somehow "own" the land was unimaginable, while Hutterites practice christian
communism, living together in very successful farming communes in Manitoba,
Saskatchewan, and North Dakota, with all material goods held in common.
Even when the institution itself is not fundamentally questioned, ethical issues
arise. Pharmaceutical patents are necessary if private parties are to be willing to take
the risks and absorb the costs of developing new drugs. The holder of a drug patent
has the legal right to set the price at which the drug is offered for sale. Does anyone
have a moral right to charge immense amounts for insulin, however, without which
diabetics can die, or for AZT, which in the 1 980s and early 1 990s was the only drug
available to help those afflicted with AIDS?
Property rights are also a major point of conflict between the industrialized
world and the less developed nations. New technologies permit mining the deep ocean
bed, but only the developed countries have the resources to exploit this opportunity
widely. To encourage and protect the immense investments that are involved, it would
make economi c sense to create property rights to the sea bed. The oceans have not
been considered to be the private property of anyone, however, or the territory of any
nation-if they belong to anyone, it is to all humanity. What right do rich corporations
and rich nations have to appropriate them, especially when the poor of the world
cannot compete? Similar issues arise with communications satellites. These must be
put into geosynchronous orbit so that they are constantly above a particular point on
the earth's surface. The number of such orbits is limited. As the developed world
appropriates more and more of these by parking satellites in them, fewer are left for
the rest of the world's peoples who cannot now afford to appropriate slots for themselves.
These examples raise additional ethical issues about how anyone ever came to
own any physical property if the earth's physical resources are regarded as initially the
common heritage of all humanity. Ethicists have taken a variety of approaches to this
question, some emphasizing human rights and fairness with little regard for questions
of efficiency. A detailed comparison of these ethical approaches are beyond the scope
of this book, but students should be aware that the possibility of creating value for
everyone, provided the values are distributed in a way that is not too unfair and that
shows proper regard for inalienable human rights, is an important part of the
justification for creating the institution of private property.
The box describing the privatization process in Eastern Europe, in which
formerly state-owned companies are transferred into private ownership in the hope of
increasing total value, highlights some of the ethical tensions and practical problems
associated with this process.
 306
Efficient

Privatizing Industry
Incentives:
Contracts and
Ownership
In recent years a number of governments have "privatized" industries that
were previously state owned, transferring ownership to private investors. For
example, Japan has privatized its rail and telecommunications systems, and
the United Kingdom has privatized huge portions of its industry: airports,
water-distribution systems, seaport operations, airli nes, railroads, aircraft and
armaments manufacturing, petroleum and gas exploration, production,
refining and marketing, telecommunications, automobile manufacturing,
trucking, and more. In these cases, privatization was usually achieved by
creating shares in the companies and selling the equity to the public through
investment banks and brokers, often at prices that were much less than their
estimated market value.
In Poland, Hungary, and Czechoslovakia, where industry generally was
state owned under communism, the noncommunist governments in the early
1 990s faced special problems of how to convert industries to private ownership.
One difficulty was that often there was only a single producer of any given
product in each of these countries, and the governments were leary of creating
pri vate monopolies. A second problem was that many of these industries
were dependent on (implicit) state subsidies, and there was concern that they
would not survive as private entities and that their failure would create
economic hardships . One of the thorniest problems, however, was how to
allocate the ownership claims in the firms that were to be privatized.
One possibility was simply to create shares in the companies and give
them to their current employees. Other citizens, who were also previous
nominal owners, objected to this. Giving shares to everyone was another
solution, but this would dilute the ownership tremendously. This would not
be a problem if there were well-functioning capital markets with low
transaction costs through which people could then sell their shares, but such
markets did not exist. Moreover, it was very hard to figure out what a share
in one of these enterprises might be worth. The accounting systems in use
were totally inadequate, and in any case, the figures they did generate were
based on the largely irrelevant prices that had been set by the planners under
communism. This made it unreasonable to expect that the markets ,rnuld
be very efficient if they were created. It also made selling the shares, as was
done in the United Kingdom and Japan, problematic. Investors would be
worried about overpaying, and the governments would be fearful of a political
backlash if the shares turned out to have been grossly underpriced. A final
concern was that foreigners, who were mo-re experienced and sophisticated
businesspeople and investors, might end up owning all the good properties.
Because the importance attached to these different problems varies
among countries and industries, no single technique is always best, and
governments ought to adapt their techniques to the firm's situation. For
example, small, domestic firms might best be sold in auctions excluding
foreign investors, whereas large employers with outdated equipment might
be sold cheaply to any foreign firm willing to invest capital to save jobs.
Privatization in Eastern Europe is a continuing experiment in institutional
design.
 307
PREDICTING ASSET OWNERSH I P Ownership and
As seen previ ously and i n earlier chapters, the way ownershi p rights are assigned and Property Rights
protected can affect the efficiency of contracti ng. Incomple te contracts, bargai ni ng
costs, moral hazard, and influence costs can all be sources of i nefficiency i n business
relati onsh ips, and the assi gnment of ownership rights can affect the magni tude of
each of these problems and the possi bilities for creati ng value.
In this secti on, we seek to develop a more systematic theory of who should own
a particular asset. S uch a theory should be of value in understanding existi ng
arrangements, in predicti ng what sort of ownershi p patterns will be adopted i n different
circumstances, and i n guidi ng deci sions about setti ng ownershi p of particular assets.
Our approach is to develop the theory of value-maximizing asset arrangements based
on the attri butes of the underlyi ng transacti ons i n which the asset is used.

Asset Specificity and the Hold-Up Problem

The most important attri bute of transacti ons for studyi ng asset ownershi p is the asset­
speci ficity attri bute. We di scussed the speci ficity issue i n Chapter 5, but i ts importance
justifies recalli ng that discussion.
Assets are specific to a certai n use if the services they provide are excepti onally
valuable only i n that use. A railroad li ne that carries a mix of passenger traffic and
cargo with vari ous desti nati ons and that also carries products from a particular factory
is not speci fic to the factory. A spur li ne that carries cargo to and from an isolated
factory is speci fic to that factory, however, because i t would be worthless (except fo r
i ts scrap value) in any other use. The degree of specificity of an asset is defined t o be
the fracti on of i ts value that would be lost if i t were excluded from i ts maj or use.
When two assets are both highly speci fic to the same use, maximizing value requires
using both together i n that use. The two assets are then said to be cospecialized.
S peci ficity and cospecializati on are important because they give ri se to the hold­
u p problem. When an asset i s speci fic to a particular use, as the rai lroad spur i s
specific to servi ng the factory, the owner of the specific asset can be held up. The
factory owner mi ght i nsist on especially low rail rates, threatening to shi p the factory's
output by truck if the the railroad owner does not accede to that demand. The threat
is a powerful one because the rail li ne is hi ghly speci fic and would lose almost all i ts
value if the threat were carried out. Thus, the spur-li ne owner might i ndeed be forced
to m ake concessi ons to the factory. If the factory's product i s too large or heavy to
ship by truck, however, then the assets of the li ne and the factory are cospecialized,
i n which case each side wields a potent threat agai nst the other. Any breakdowns i n
bargai ni ng will lead to large costs o n both sides.
A hold up itself has no effect on total value. In our example, unless bargai ning
costs are high, the Coase theorem i mplies that the factory owner and the railroad
owner wi ll agree on some value-maximizing shi pping arrangement. The direct effect
of the hold up by the factory owner is simply that the railroad owner receives less of
the total value and the factory owner more; efficiency i tself i s unaffected.
Despi te this, hold ups are not i nnocuous and can lead to value-destroying
consequences. The rai lroad owner, for example, feari ng the possi bi lity of a future
hold up, might not i nvest i n the speci fic asset. The fear of hold ups can deter people
from i nvesti ng i n hi ghly speci fic assets, reduci ng the total value to be shared. This is
the real social cost of the hold- up problem.
THE OWNERSI IIP SoumoN Economic theory predicts that an efficient organizati onal
structure will protect the i nvestments of those who would otherwi se be deterred from
maki ng the m ost important, value-creati ng i nvestments. The obvi ous soluti on for the
factory and the spur li ne is for the factory to own the tracks. Two other exam ples of
 308
Efficient hold ups from Chapter 5 are mine-mouth electrical generating plants, which rely on
Incentives: an adjacent coal mine for their coal supplies, and the Fisher Body plant, which was to
Contracts and be set up as a dedicated facility to serve General Motors only. In each case, the problem
Ownership was the reluctance of one or both parties to make large specific investments for fear of
the hold-up problem. Fisher Body, for example, might have built a more expensive
plant that was not dedicated to General Motors but that could be easily switched over
to serve other customers. The extra investment might have been inefficient, but it would
have protected the company from being held up. Similarly, an electric-generating plant
located near a coal mine might be built to accept a wider array of coal grades than the
mine produces, so that the utility company would be less reliant on the output of the
mine. In each case, the danger is that unnecessary costs may be borne and value
destroyed by attempts to protect against a hold up. General Motors resolved its difference
with Fisher Body by purchasing the company, and mine-mouth electric-generating
plants in the United States usually own the coal mines that serve them. The box on
computer hardware and software contrasts the different approaches taken within the
computer industry to the problem of managing specific investments.
Generally, when an asset is specific to a particular use, the hold-up problem
for that asset can be avoided by having the user own the asset. Central to this theory
is that there will be a tendency for specific assets to be owned by those who use them
and for two cospecialized assets both to be owned by the same person or organization.
Despite its considerable importance, this "tendency" prediction can only be
tentative because it ignores other attributes of the transaction. Indeed, the prediction
itself can be ambiguous, and its two parts can even be inconsistent with one another
when there is more than one specific investment. For example, the railroad spur is
specific to both the factory it serves and to the main rail trunk line to which it
connects. To whom should this doubly cospecialized asset belong?
In this particular case, the correct answer would probably be governed by the
attributes of the related shipping transaction. If rail rates on the main trunk line are
regulated by the government and the shipper is required to accept shipments at these
predetermined rates, then there is less danger of the factory being held up by the
railroad if the factory invests in the spur line, but there may be a danger of the railroad
being held up by the factory in the reverse case. For example, shipment by truck may
be an alternative, or the factory may be only marginally profitable and considering
shutting down. Therefore, the factory should own the spur line in this case. The
hold-up analysis is central to analyzing ownership, but its application requires a more
detailed evaluation of the context, which involves studying other attributes and related
transactions.

Uncertainty and Complexity

If all relevant contingencies can be forecast and planned, then a contract specifying
what is required can be a good alternative to ownership. In our railroad and factory
case, for example, the railroad might lease the spur line to the factory for a fixed
amount per year over a term of many years. Provided the only important hold-up
possibility for the rails is tht:> rental price, this could be a fully satisfactory alternative
to ownership by the factory. For relatively simple transactions, long-term leases are
often a good substitute for ownership.
A similar example is the use of farm land for growing annual crops such as
wheat. The investments made by the farmer in any one particular growing season
such as preparing the soil, planting, fertilizing, and so on are highly specific, but the
transaction itself is simple. The landowner might want to sell the land eventually for
some kind of development, or build a house on it, or farm it alone. The owner is
unlikely to lose much if he or she changes his or her mind on one of these questions
 Ownershi p and
Property Rights

Computer Hardware and Software

Computers, the operating systems needed to make them function (such as
UNIX@, DOS@, OS/2@), and the applications programs (statistical packages,
word processors, database systems, and so on) that allow people to use the
machines are often cospecialized to one another. For example, Apple
Macintosh computers use a different operating system than do personal
computers from other manufacturers. This means that applications programs
have to be specially written to work on a Mac. Similarly, programs to be
used with Tandem Computer's machines are specialized to this hardware
and its operating system, which in turn are useless without the programs for
running · automatic teller machine systems for banks, doing transaction
processing for stockbrokers, and carrying out the other applications where the
"failsafe" characteristics of Tandem's machines are so important.
In these situations, the relationships among the producers of software,
operating systems, and hardware are very important. Apple produces its own
operating system rather than entrusting it to an outside company. It initially
also wrote many of its own applications programs, first using Apple employees
and later experimenting with a separate subsidiary, Claris Corporation, to
produce software. Tandem has alliances with a variety of software houses
that produce different programs for its machines, and it is very concerned
with their success, although it currently takes no ownership position in these
other companies.
In contrast, many manufacturers make personal computers that are
"IBM compatible, " meaning that they use the same operating system
(Microsoft Corporation's DOS) as does IBM's Personal Computer line and
that application programs that will run on an IBM will run on these machines
too. This means that the software is not specialized to any one machine, and
applications programs for IBM-compatible machines are produced by a huge
number of different fi rms. There are typically no special relationships or
arrangements between the software producers and IBM, and the hardware
manufacturers usually do not produce software for these machines. Given
that only Microsoft produces DOS, however, these machines and the
applications programs are specialized to DOS. As would be expected, IBM
and Microsoft have worked very closely together.
In 1990 the relationship between IBM and Microsoft became strained.
The point of contention was the new OS/2 operating system jointly developed
for IBM's new personal computer lines. IBM wanted to push forward with
attempting to make OS/2 the new industry standard, whereas Microsoft
preferred to stick with DOS supplemented with its Windows@ software,
which is graphics-oriented (like the system on the Macintosh) and allows
users to access several applications packages at once.
Because IBM's profits in the personal computer market depend on its
ability to introduce idiosyncratic products, whereas Microsoft can thrive as
long as IBM compatibles do well, the two firms have different interests.
IBM's management ought to have anticipated this potential for a conflict of
interests and arranged contract terms to mitigate the conflict when it negotiated
its software development deal with Microsoft.
 310
Efficient but cannot act on the decision during a single growing season, however. Therefore,
Incentives: a relatively simple contract that transfers rights to use the land for farming to a renter
Contracts and for a particular season is a good substitute for having the farmer own the land, at least
Ownership in terms of protecting the farmer from having his or her specific investments subjected
to a hold up. The terms of the contract might specify what is to be planted and make
the rent dependent on this because some crops deplete the nutrients in the soil more
rapidly than others, but this is not a major difficulty. The fact that the bilateral
relationship between the farmer and the landowner is repeated over a period of years
also provides an incentive for each to treat the other fairly, for example, for the farmer
to avoid depleting the land.
The situation with fruit orchards, vineyards, and other perennial crops is very
different. There, the farmer's investment is expected to last for a long period of time,
during which complicated decisions about crops, soil, and other uses of the land may
have to be made. Spelling out all these contingencies in a long-term contract would
be impossible. The investment and the land are still highly cospecialized-physically
separating the land from the trees or vines would be very wasteful-but this time a
long-term contract is less satisfactory. We expect to find that farmers of these kinds
of crops are much more likely to own the land they farm than are farmers of annual

Freq uency and Duration

crops. Evidence from the state of California is consistent with this distinction.

The longer the period over which two parties might interact, the more difficult it will
be to foresee and contract for all the relevant contingencies and the less likely is a
pure, arm's-length contracting solution to be satisfactory. This means that some other
governance structure including common ownership may be preferred, especially when
the fixed costs of creating the solution can be spread over more individual transactions.
However, the long horizon also means that reputations may be more effective as
control mechanisms. Which predominates is a matter of the specifics of the case.
In both the North American and Japanese automobile industries, the manufactur­
ers buy many parts and systems from outside suppliers, while making others themselves.
The largest U. S. producer, General Motors (GM), has been much more vertically
integrated than Toyota, the largest Japanese producer, with GM producing much
more in house than does its competitor. The parts, components, and subassemblies
that GM produces in house tend to be highly specific to GM. 8 For the parts it does
obtain outside, GM has usually relied on short-term, arm's-length contracting. It sets
detailed specifications and selects suppliers on the basis of competitive bidding, signing
them to simple, fixed-price contracts. Winning this year's contract is no guaranty of
getting any business next year, however. The number of firms supplying GM is in
the thousands. Toyota has only a few hundred suppliers. They often provide complex
components that they design themselves and that are quite specific to Toyota's models.
The relationships between Toyota and its subco�tractors are close, complex, and long
term. They are marked by extensive sharing of information and costs and by Toyota's
active involvement in advising its subcontractors, although Toyota often has no
ownership interest in these firms. 9
All this is consistent with the theory. As the theory predicts, the bargaining costs
on purchasing complex, specialized components in outside, arm's-length negotiations
arc too high for it to be a viable method of arranging repeated transactions, and in

s David Teece and Kirk Monteverde , "Supplier Switching Costs and Vertical Integration in the
Automobile Industry," Bell Journal of Eco11omics, 13 Spring 1 982, 206- 1 3.
9 Banri Asanuma, "Japanese Manufacturer-Supplier Relationships in International Perspective: The
Automobile Ca�e, " Working Paper No. 8, Faculty of Economics, Kyoto University (1 988).
each case an alternative was sought. At GM, the solution was one of ownership: GM Ownership and
makes a large fraction of its own components. At Toyota, the solution was to establish Property Rights
long-term relationships with a smaller number of key suppliers who would make a
wide array of components and systems and would be rewarded or penalized for their
performance when additional contracts were awarded for the next car model.
Although both systems are responses to the same problems, they are not equally
effective. By the mid- l 980s, GM had begun adapting aspects of the Japanese system
to its own operations. GM reduced the number of its suppliers, bought whole systems
(such as a seat) instead of components (such as springs, seat frames, padding, and
covers) and established closer supplier relations, to the point of allowing certain
suppliers to participate in the design of the next model.

Difficulty of Performance Measurement

Even when as.sets are not highly specific, the assignment of ownership can still matter.
There is an easy way to provide proper incentives when a nonspecific asset is used by
a single person to produce marketable output: That person can be made the owner,
both having the residual control and being the residual claimant. 1 h this way, the
person will have a proper interest in maximizing the residual value of the asset. This
solution also has disadvantages, however. Two important disadvantages are accentuated
in Chapter 7. The first is that transferring ownership also transfers the risk of
fluctuations in the value of the asset to the single individual, who may find the risk
very costly to bear. The second disadvantage arises when the person who cares for the
asset also has other responsibilities in which performance is difficult to measure. By
the equal compensation principle, these other responsibilities will be neglected unless
strong enough incentives are established for it to be worthwhile for the asset owner to
devote some effort to them. When performance in the other responsibilities is difficult
to measure, these incentives are costly and represent an added cost of transferring
asset ownership.
An alternative to the ownership incentive is the kind of incentive that we studied
in Chapter 7: the use of explicit performance-based pay. This alternative requires
attempting to measure the person's performance in caring for the asset. Compared to
transferring ownership, the disadvantage of this system is that it incurs an added cost
(for the measurement itself) and still leaves the agent facing risk-in this case the risk
associated with measurement errors. Generally, when it is relatively inexpensive to
measure performance accurately and when the risks associated with asset ownership
are relatively large, it is better to base compensation on measured performance than
to shift ownership. When care is especially difficult or costly to measure and the risks
of asset ownership are not too great, then the ownership solution should be preferred.
When the problem of motivating the person to honor his or her other responsibilities
is great, the best system may be to avoid offering any formal financial performance
incentives.
Finding the best ownership pattern is more complicated when the value of an
asset can be affected by several different people. Consider the organization of a small
shipping company that collects orders and dispatches truck drivers to carry the loads
to various destinations. Is it more efficient for the truck drivers to own the trucks or
for the shipper/dispatcher to own them? The answer depends on whose behavior most
affects the value of the trucks and how hard it is to measure performance in those
dimensions. If a single driver or team uses each truck to make cross-continental trips,
with the truck being serviced on the road by the driver or whatever mechanic he or
she can find, then the dispatcher's behavior is likely to have little effect on the value
of the truck, whereas the driver's behavior would be important. Meanwhile, the
driver's use of and care for the truck are hard to measure, so the value-maximizing
 312
Efficient allocation of ownership rights would be to make the driver the owner. to If the truck
Incentives: is used for local deliveries on several shifts, however, with several drivers using each
Contracts and truck and all service work performed by mechanics in the company garage, then the
Ownership individual trucker's incentives become less important (he or she no longer makes
maintenance decisions) and ownership becomes less effective for providing the trucker
incentives (because the cost of a poorly operating truck is shared among those who
use it). At the same time, the behavior of the shipper/dispatcher, who hires the
mechanics and drivers, provides equipment for the garage, and so on, becomes a
more important determinant of the truck's condition. Ownership should then lie with
the dispatcher. The theory therefore predicts that there will be different ownership
patterns depending on the way the truck is used and maintained. Ownership will be
assigned in each case to the person whose hard-to-measure actions have the greatest

Connectedness
effect on the truck's value.

Few real businesses conduct just one kind of transaction using just one kind of asset.
We must study ownership issues in terms of the connections among the various
transactions and assets. We have already seen an example of multiple connected assets
in our discussion of the railroad spur line and the factory. The line is specific both
to the factory and to the railroad trunk line. A complete analysis depends on assessing
both connections. When there are multiple assets, each specific to some common
uses, determining the optimal assignment of rights can become very complex.
The general principles remain the same, however. Ownership rights should be
structured with a concern to minimize the distortions in investment decisions caused
by the hold-up problem. Notice that the theory does not say that the transaction
should be structured to minimize the amounts lost by individuals in the hold ups.
These amounts are not accorded any weight themselves in the efficiency calculation
because they have no direct effect on total value. Hold ups are accounted for only
indirectly, through their effect on the investments that people are willing to make.
A second factor influences this choice. When the assets or the services they
yield are strongly complementary, so that a higher level of one significantly increases
the value of the others, there may be multiple patterns of investment in each asset
that are mutually consistent with one another, forming coherent patterns, and yet
only one of these distinct patterns actually maximizes total value. As seen in Chapter 4,
this sort of situation necessitates explicit coordination among those choosing the
amounts of the various assets or activities, even when all the decision makers are
pursuing the same objective (as they would when all the assets are under common
ownership and no moral hazard problems intervene between the owner and the
decision makers). When the decision makers have divergent interests, the coordination
problems become much more difficult because incentive problems can interact with
them. This suggests that, other things being equal, strongly complementary assets
should be brought under common ownership.
Of course, this last suggestion is subject to the qualification that has been applied
several times before: Sometimes alternative sophisticated governance structures can
be excellent substitutes for ownership, and these alternatives may avoid some of the
influence costs of ownership. Thus, the assets of Toyota's subcontractors are likely to
be complementary with Toyota's own assets. For example, Toyota's just-in-time system
is essentially infeasible unless its suppliers have adopted a system that allows them to

10 In recent years, the introduction of the tachograph, a measuring device for use in trucks to keep
a 1 1 accurate record of dri,·ing times and speeds, has significantly changed this equation. The device makes
it easier to 111eas11re performance and reduces the advantages of having drivers own their own trucks.
be ready to deliver on a m oment's no tice. Th is m eans th at close coordination is Ownership and
necessary. It is evidently achie ved between Toyota and its subcon tractors across Property Rights
organ izational boundaries, despite the lack of common ownership.

Human Capital
Finally, we turn to one of the m ost importan t kinds of asse ts in any advanced industrial
economy: workers' an d m anagers' skills and knowledge. A pers on's skills and knowledge
are assets that can only be owned by the pers on alone. In the absence of slavery,
workers are not free to transfer ownership rights permanently to som eone else.
The nontransferability of human capital is problematic when those skills are
specific to an organization or physical assets.. A team of people working together m ay
well devel op cospecial ii:ed skills and knowledge: fam il iarity with one another's habits,
a common language to describe the local goings- on, knowledge of organizational
routines, and so on. Simply allocating ownership rights in the organization or its
physical assets cannot protect the personal inves tments of all the workers. S ome of
the arrangements that do protect them are described in Chapters 10 and 11.
The imp ossibility of transferring owners hip of human capital assets als o leads to
an imp ortant question: In whose interest should firms be run?

OWNERSHIP OF COMPLEX ASSETS

We have j ust seen some elements of a theory of what would constitute an effi cient
allocation of ownership rights. We now return to the discussion of ownership itself to
reexamine the fundam ental idea and to question whether ownership is an effective
means of providing incentives in all kinds of organizations and for all kinds of assets.
We have treated ownership as possession of residual control, including the right to
dispose of the asset or appropriate all or part of its value for personal use. This notion
is fairly clear-cut in the case of m ost simple, discrete assets held in sol e proprietorships,
like the truck ownership example discussed earlier. In richer situations with m ore
complex assets, however, complications arise.

Owning Com p lex Return Streams

If the payoffs from an asset are not one-dimensional (essentially, if all returns are not
money), then there m ay be different residual elements ass ociated with each dimension.
For example, if two professors form a partnership to write a text together, then there
are a variety of rewards. During the writing, each learns different things and suffers
different costs, and in fact their families m ay suffer m ore than either author. O nce
they have finished, there are royalties from the sales. In addition, there is the effect
of the prestige of each author in the profession and within his or her department.
Presumably each will teach from the book, and this will have some effect on the
success of their courses and the status they enj oy among their colleagues and students,
as well, perhaps, as on their salaries. Finally, the success or fail ure of the book will
influence the abili ty of the authors to negotiate better contracts on future books, either
separately or together.
The royalties received on this enterprise are transferable cash payments. The
effect of the book and of the students' reception to the courses on salaries are m onetary
but very hard to measure, as are the effects on future contracting opportunities. H ow
much of each author's raise is due to the book and its impact? The other costs and
returns are n onm onetary, not eas ily transferable, and indeed, extremely hard to
id entify and quantify, even when they accrue to the authors personally rather than to
their fam ilies. M oreover, the returns are only partially resp onsive to the efforts of
either coauthor, because responsibility and credit are shared. I n effect, both partners
 314
Efficient are residual claimants, but on different, imperfectly transferable streams. Even if we
lncenth·es: decide that both authors own the book (although the publisher may well have the
Contracts and copyright), does either then have optimal incentives to work on it?
Ownership That such a simple venture gives rise to such complications is surely suggestive
that many business transactions will be at least equally complex. A manager who
carries out a particular project gains expertise that is of unknowable value in the
future. The returns from the project are not just the revenues and costs that the firm
records and that generate residuals that may be ultimately claimable by the stockholders
but also the effect on the manager's future earnings opportunities. These may never
be observable to the firm or even to the manager. Are they not residuals then? Who
then "owns" the project?

Who Owns a Public Corporation ?

The concept of ownership becomes even more problematic when we consider the
complex agglomerations of assets that make up the modern corporation. Is there any
single individual or recognizable group with effective ownership rights over the bundles
of assets that are Ford, Toyota, Fiat, Hyundai, Hudson's Bay Company, Salomon
Brothers, Sony, British Airways, IBM, Apple, or any of the other giant corporations
that are the unique characteristic of the organization of economic activity in the late
twentieth century?
STOCKHOLDERS Nominally, and by law, the stockholders own a corporation. Their
rights are in fact quite limited, however. Stockholders can vote to change the corporate
charter, they can elect the directors and (except where a "classified" board with fixed,
overlapping terms is in place) remove them by majority vote. They usually have the
right to vote on substantial "organic" changes, such as mergers in which the company
disappears or the sale of most of the corporation's assets. That is it. The stockholders
cannot set the dividends that are paid out to them, they have no role in investment
or acquisition decisions, they do not hire the managers or set their pay, and they have
no say in setting prices. In other words, they have no direct rights in deciding any of
the multitude of issues that are crucial in running the business. Of course, by electing
the directors, who are empowered to hire and fire management and to make or ratify
all major management decisions, the stockholders can indirectly affect the decisions
that are made. Furthermore, if the directors they elect do not follow their wishes, the
stockholders can replace them as the opportunities arise. Yet, this is quite removed
from the control associated with the idea of ownership of a simple asset like a truck.
Recall, however, that the formal definition of residual rights runs in terms of
decisions and returns that are not explicitly vested elsewhere by contract or law. One
could then argue that, in buying the shares in the firm, the stockholders contractually
assigned the rights to make most decisions-including the crucial decisions about the
distribution of the returns from the asset they "own"-to someone else. They are still
the owners; they just happen to have written contracts that leave them few residual
rights. This resolution is not satisfactory, however. Stockholders' rights are not residual;
instead, they are strictly delimited and enumerated.
DIRECTORS If any group could be considered to have residual control in a corporation,
perhaps it might be the board of directors. It is they who have the power to set
dividends; to hire, fire, and set the compensation of the senior executives; to decide
to enter new lines of business; to reject merger offers or instead approve and submit
them to the stockholders; and so on. In the United States, the authority of the board
of directors is buttressed by the "business judgment rule , " according to which the
courts basically will not second-guess the directors' business decisions.
If the directors have residual control, however, they certainly do not have claim
 :us
to the residual returns. If the corporation is liquidated, the receipts are distributed to Ownership and
the stockholders once debts and taxes are paid, and the directors are not free to Property Rights
appropriate the profits for their own use.
Moreover, two recent developments in the United States have rendered uncertain
just what contractual and legal obligations a U. S. corporation's directors may have
and thus what decision rights are residual. First, the rapidly changing legislative and
case law surrounding directors' obligations in the context of changes in corporate
control (mergers, acquisitions, and so on) means that directors really cannot be sure
just what rights they have. The skyrocketing cost of director's liability insurance is
evidence of how unsettled these matters are. Second, the emerging social notion of
"stakeholders" having a claim on the corporation means that politically, if not legally,
the freedom of the corporation to act as its directors might wish is circumscribed in
as yet -ill-defined ways.
MANAGERS AND EMPLOYEES Beyond this is the question of whether, even in the
context of regular operations, the directors (or the stockholders) really can have residual
control. A right that cannot be exercised is of little significance. Furthermore, for
most purposes, the directors must rely on the officers of the firm to provide them with
the information needed to make decisions. By controlling the flow of information to
the board and setting the agenda, the senior executives may have effective control of
many of the decisions that are nominally controlled by the board. Moreover, many
critics argue that the board members are effectively chosen by the senior executives
and are totally beholden to them. The shareholders may elect the board , but they
almost always simply elect the slate of candidates on the proxy statement sent out by
management, and effectively management decides who is nominated.
Yet the officers in turn are dependent on the corporation's employees to develop
the information in question and to carry out the decisions that are made. Many a
senior manager's plans have been thwarted by the resistance of lower-level employees.
Perhaps these employees are really the residual decision makers.
All this calls into question the usefulness of the concept of ownership as residual
control in the context of the large corporation. Note too that the idea of the owners
being those with residual claims is not fully satisfactory. First, if the residual claimants
are not in a position to control the decisions that affect the value of the asset, then
the incentive properties that have been claimed for ownership are certainly weakened.
Second, as we have already seen, it may be impossible to identify any individual or
group that is the unique residual claimant or, indeed, to identify the benefits and
costs accruing to any decision and so compute the residuals.
Finally, if the situation is murky in the case of a for-profit corporation, how
does the concept of ownership apply to a university, a church, or another nonpwfit
organization? In such organizations, no private parties have a right to appropriate the
residuals after debts are paid. Instead, residuals must remain in the organization, and
if the organization is dissolved, any remaining funds go to the state. If there is a
residual daimant, it would seem to be the state, but surely the state does not have
the residual control in such an organization any more than it does in a for-profit
organization.

Whose I nterests Should Count?

Proponents from both sides of an odd coupling of interests argue that companies, and
especially publicly held corporations, should not be run simply in the interests of their
stockholder/owners. The parties to the alliance are, on one hand, the managers and
employees of the corporations, often backed by the companies' suppliers and the
governments of the localities where these companies have operations, and, on the
 316
Efficient other, a variety of academics and activists who believe that the pursuit of profits is
Incentives: either socially inappropriate or immoral. Both groups are apt to characterize
Contracts and stockholders as uninvolved, absentee owners with no allegiance to the firm and no
Ownersh ip concerns but their narrow selfish interests in short-term financial gains. They are
either unworthy of having their interests predominate or incapable of realizing where
their long-term interests actually lie. Instead, the company should pursue social aims
(usually those favored by the activists) or should care for the interests of the people
who are actually involved in the organization (usually the position of the stakeholders).
Against these positions is another discomforting alliance made up of investors,
both private and institutional (mutual funds, insurance companies, pension funds,
financial intermediaries), along with the the investment-banking and stock-brokerage
industries, as well as a set of free-market economists. This alliance argues that
maximizing the value of the firms enhances economic efficiency. Entrenched
managers, self-interested employees, noncompetitive suppliers, and incompetent
governments who argue for paying attention to other concerns simply want to escape
the discipline of the market and use the owners' resources for their own ends. Having
managers pursue anything other than maximization of the value of the firms entrusted
to them is to invite calamitous self-serving moral hazard. What results is that the
managers will simply pursue their own interests, perhaps adjusted to account for the
interests of their allies.
There is surely not space enough to resolve this dispute here. However, we can
offer observations based on the ideas we have already developed, and we return to
these issues in Chapter I 5.
EXTERNALITIES AND CORPORATE SoCIAL RESPONSIBILITY In a system of complete, com­
petitive markets, the arguments for running the firm to maximize profits or the value
of the firm are largely unassailable (see Chapters 2 and 3). Similarly, if bargaining costs
are low and property rights are well established, secure, and tradable, so that the Coase
theorem would apply, there would be no basis for complaint about a firm pursuing the
interests of its stockholders. Unfortunately, we do not live in such a world.
In actuality, the decisions that are made in firms can affect many different
people in ways that the market does not adequately mediate. Externalities such as
pollution are one example. A firm that chooses to pollute may increase its value at
others' expense, and this course of action is unlikely to promote overall efficiency
when we cannot count on the mechanisms of bargaining to ensure that the firm
recognizes the full costs and benefits of its actions. Similarly, if the firm is dealing
with unsophisticated, badly informed customers, it might gain while decreasing
efficiency by selling them shoddy or even dangerous goods.
In such situations, a socially responsible course of action may be more efficient.
The question is whether asking for such behavior, even if realistic, might not bring
unintended consequences. Suppose that the firm's managers are directed to promote
social efficiency. Do they have the information to do so? How are they to learn of
the costs and benefits to other parties of various courses of action when there are
neither prices nor explicit bargaining with the affected parties to guide them? How is
all this to be monitored effectively, and by whom? Who has the information and the
incentive to tell if managers who are not pursuing profits are behaving appropriately
or are instead serving their own interests?
SPECIFIC ASSETS AND STAKEHOLDERS' RIGHTS In Chapter 8 WC discussed how quasi­
rents can arise within firms and how different business decisions can affect the
distribution of these returns among employees. Employees are not the only ones
affected by the firm's decisions, however. Obviously, stockholders' returns are affected
by a wide range of decisions, including investments, wage levels, dividend policy,
 Ownership and
Property Rights
Japanese Corporate Ownership and Responsibility
I n recent years, Japanese corporations have enjoyed immense comm ercial
success. Japan is a co untry almost without the usual natural resour ces-few
significan t mi neral an d ore deposits, no domestic petroleum reserves, very
limited farm land-and its human and physical capital stock was devastated
less than 50 years ago in World War II, when Japan became the o nly coun try
ever to suffer a nuclear attack. It now is one of the great economic powers,
and (depending how o ne co nverts yen to other currencies), it may be the
richest industrial economy per capita in the world. More than in most
developed countries, Japan's large corporatio ns seem to dominate its eco nomic
organization.
Japanese corporatio ns are structured somewhat differently than are those
in o ther developed capitalist eco nomies, and the basic pri nciples that guide
them seem to be different as well. A typical Japanese board of directors is
made up almost exclusively of full- time employees of the fi rm. According to
a 1991 study of 100 of the "best" com panies in Japan, only 19 had any
directors who were not em ployees, and of the 2, 737 directors of these
companies, only 5 5 were outsiders. Even these might be the firms' bankers
or the heads of wholely owned foreign subsidiaries. In contrast, amo ng major
U.S. firms, 7 5 percent of the directors are outsiders, and the ratio is almost
as high in the U nited Kingdom . The stock in these Japanese firms is, to a
large extent, in the "friendly hands" of affiliated companies, suppliers,
customers, or banks with long-term relatio nships with the firm. More than
80 percent of the largest firms have a majority of their stock in friendly hands.
These owners will not sell a firm's stock under any sort of normal co nditio ns,
nor are they at all likely to participate in a pro xy contest. In other countries,
stock is more often in the hands of investors who may have no other business
connection to the firm. The executives of major Japanese corporatio ns have
typically spent their entire careers in their com panies, and there is little
possibility of a senior executive being lured away from one major firm to
another. No managerial labor market enforces discipline on them.
In whose interests are J apanese corporations run? Their structure would
hardly suggest that it is the stockholders' interests alo ne that are being served,
although the profits that are generated have been handsome, and the N ikkei
Stock Index of the share prices of the top firms has climbed to remarkable
heights.
A distinguished stude nt of the Japanese econom ic system, M asahiko
Aoki, has described the Japanese corporatio n in terms of the managers'
mediating between the interests of different groups-particularly em ployees
and investors (who, under the Japanese system, may incl ude suppliers and
customers). * But what weights do managers give to these vario us groups'

Two surveys reported in a Japan ese newspaper, th e Nikkei Sangyo

interests?

Shinbun (April 23, 1990, and July 5 , 1990) are informative in this regard.
The first concerned the attitudes of Japanese middle m anagers; the seco nd
dealt with the attitudes of the presidents of major Japanese firms.
Both groups were asked to whom the firm should belo ng, that is,
presumably, in whose interests should it be run, and to whom they actually
 3 18
Efficient
Incentives:
Contracts and
Ownership
do bel ong. M ultiple answers were allowed, so that the idea of there bei ng
multipl e legitimate claimants could be expressed. Middle managers ranked
shareholders third on the first question, with only two-thirds of the respondents
naming the nominal owners as having a right to have their interests be taken
into account. The most frequently named group were employees, foll owed
by society as a whole. Further down the list were customers and management.
Regarding the issue of in whose interests they believed fi rms actuall y were
run, the middle managers again ranked employees first an d shareholders
thi rd, with management now in second place. Interestingly, fewer of the
respondents mentioned shareholders in their answer to thi s second question
than to the fi rst.
The senior executives were more oriented to the stockhol ders, men­
tioning them most frequently on the fi rst question. S till, the number
mention ing empl oyees was almost as large (80 percent versus the 87 percen t
mentioning shareholders), and most company presidents in fact indicated
that the firm shoul d bel ong to both groups, with society as a whole also
getting man y menti ons. On whose interests actual ly were served, the most
common answer was employees, with shareholders coming second. Again,
most of the presiden ts menti oned both groups, but a fu ll 20 percent indicated
that shareholders interests did not count in running the firm.

* Masahiko Aoki, The Cooperative Game Theory of the Firm (Cambridge: Cambridge
University Press, 1984).

mergers, pricing, an d many others. The security of creditors' and len ders' claims will
depen d on the riskiness of the firm' s investments, the wages it pays to workers, and
the dividends it pays to shareholders. A firm's decisions about the design and pl acement
of its factories can affect community housing values, traffi c, and environmental
quality; its choices of suppliers can affect the distribution of wages and profi ts among
the potential suppliers; and its pricing pol icy affects bot!.: competitors and customers.
There are frequent disputes about which of these interests are legitimate ones that
ought to be weighed in the decision-making process. Is a small community entitled
to bl ock a decision by the area's largest employer to close its local factory? Are its
downtown stores and their employees entitled to protection when a discount chain
opens up on the edge of town? Should residential neighbors be allowed to demand
mitigating measures or compen sation when a medical center expan ds its facilities,
increasing local traffi c with its attendant noise, parking problems, and traffic safety
issues?
With in complete markets and imperfect bargaining, the way various interests
are weighed in decisions can have con sequences for the effi ciency of the economic
system that must not be ignored. For example, in the plant closing decision ,
in vestments by workers and others in houses n ear the factory may lose much of their
value if the factory were to close. S uch investments are cospecialized with the plant.
Efficiency then requires that the homeowners' in terests be given some weight in this
decision. Sim ilarly, the workers may have in vested in firm-specifi c human capital
that is, by its very nature, cospccia lized with the plant. Clo sing the plant destroys the
va lue of these inves tmen ts. The town ship as a whol e may have in vested in roads,
 :-1 1 9
sewage-treatment facili ties, schools, a nd other assets whose value depends on the Ownership and
plant's co ntinued operatio n. Property Rights
The presence of these eospeeialized investments suggests that, at the time the
investments a re bei ng made, it would be in the mutual interest of the firm a nd of
employees, homeowners, local government, a nd other stakeholders to devise some
way to protect stakeholders a gainst opportunism by the firm, perhaps by agreeing to
restri ctio ns on plant closi ngs. Protection of thi s sort would encourage stakeholders to
i nvest i n bui ldi ng these eospecialized assets, which also benefits the firm by maki ng
a larger la bor force a vaila ble. Because there would need to be so many parties affected,
it is unli kely that bargaining amo ng them all co uld lead to an efficient solutio n, a nd
it may be better to have laws instead that settle the ri ghts of the parti es in advance.
 320
Efficient
Incentives:
Contracts and
Ownership SUl\11\L\RY
In economic terms, ownership is distinct from other forms of contract only where
contracts determining specific decision rights over an asset and assigning its returns
are imperfect and incomplete. Then, ownership may be identified with the right to
exercise residual control where the contract is silent about decision rights, or with the
right to receive any residual returns that remain after contractual obligations are
fulfilled. For simple assets, it can be advantageous to have as many decision rights as
possible vested with the person who receives the resi dual returns, for in the process
of maximizing his or her own personal returns, that person will also generally be led
to maximize total value.
When ownership rights are tradable and the conditions of the Coase theorem
apply, the initial allocation of property rights does not affect the efficiency of
arrangements because the rights will be traded as necessary to restore efficiency. When
rights are not tradable, however, they may wind up in the wrong hands. The limitations
on the trading of water rights in California provides an example. Water is frequently
directed to certain low-value uses (such as growing rice) when farmers who are entitled
to cheap water cannot trade their rights to cities who value the rights more highly.
Similarly, when rights to an asset are insecure, the owner may not find it worthwhile
to preserve and enhance the value of the asset because he or she may not be permitted
to enjoy the benefits of those efforts.
Some assets are commonly shared within a community. The result is often what
is widely known as "the tragedy of the commons, " in which the shared resource is
abused by members of the community because the costs of the abuse are widely
shared. We analyzed this problem in connection with local fishing rights and found
that similar analyses apply to aquifers and petroleum deposits. In the fishing rights
case, various ownership patterns could be tried to improve matters. The community
could own the asset, but unless the community was homogeneous such a solution
could lead to large influence costs. Also, the problem of monitoring compliance to
the community's rules must still be solved. Individual ownership provides another
alternative, but that may lead to distorted incentives if the returns to the asset cannot
be directed entirely to the owner. In socialist countries, even farms and factories are
often treated as shared resources and accordingly are not properly maintained. The
Chinese experience (especially with private agriculture) during the decade of the
1 980s demonstrates how ownership incentives can powerfully improve individual
performance.
When property rights are secure and tradable and bargaining costs are low, the
Coase theorem suggests that people will trade rights until the new pattern of ownership
is efficient. Some argue that we should presume that observed patterns are always
efficient when we account for all the relevant costs because untraded rights simply
indicate that the exchange was not worth the cost that would have to be incurred.
When the agreements being made between two (or more) parties affect others who do
not participate in the bargaining, however, it is not generally true that any trade must
be value increasing. It could instead simply be appropriating value from the
nonparticipants. In circumstances like these, there can be important advantages to
simply assigning property rights to those who are likely to value them most highly.
Property rights in the modern world continue to change, with rights to clean
air, intellectual property, sea-bed m inerals, beach access, fish ing rights, and broadcast
bands being exa mples. The rights to use new drugs serve to remind us that the
 321
assignmen t of ownership rights has ethical and distributive aspects in addition to its Ownership and
effi ciency ones. Property Rights
When bargaining costs are low, the theory predicts that ac tual own ership rights
will come to be assigned efficiently. When an asset is highly specific to a particular
use, mean ing that it is much less valuable in its next- best use, then there are gains
to assigning ownership to the final user in order to avoid the hold-up pr oblem. In the
hold-up-problem, the user threatens not to use the asset in order to extract better terms
from the owner. The reason hold ups are a problem is that they disc ourage in vestments
in highly productive specific assets. When two assets are both specific to the same
transaction, they are cospecialized, and the theory predicts they will usually both be
owned by the same party. Difficulty of measuring performance in maintaining an asset
favors transferring ownership to the asset user, but a similar difficulty in measuring
performance in other important responsibilities may make transferring asset ownership
unwi se.
Other �ttributes of the transaction also affect the value-maximizing ownership
pattern. Simple transac tions may be managed by nearly complete c ontracts, in which
case the hold- up problem is of only minor importance. A lease of the asset, for
example, may be a good substitute for ownership. Long durations of the asset and
complexity of the transaction make these alternatives less effective and favor the
ownership solution mentioned previously.
H uman capital poses a special problem because ( 1 ) it represents a large fraction
of the capital of any advanced industrial country, (2) it is generally not tradable, and
(3) it is quite commonly cospecialized with the human capital of other people.
Complex organizational arrangements are often devised to protect investments in
human capital (see Chapter 1 0).
Although the concept of ownership seems reasonably clear in many of the cases
discussed here, the concepts of residual control and residual returns that define
ownership are actually quite elusive. For large c orporations, we argued that there is
really nobody who owns both the residual returns and the residual c ontr ol necessary
for ownership to work as the simple theory describes. For nonprofit organizations,
there is nobody with a right to c ollect the residual returns and, therefore, we would
say, no owner. In the case of this book, there are various ki nds of returns, not all
fi nancial, and these c ome in different proportions to the two authors and the publisher:
There is no single residual claimant.
Despite these qualifications, ownership is clearly the most common and effective
means to motivate people to create, maintain, and improve assets, and its importance
in practical business life would be hard to overstate.

• BIBLIOGRAPHIC NOTES
As with so many other parts of the ec onomic theory of organization, Ronald
Coase made the initial modern c ontribution to the ec onomic analysis of property
rights, emphasizing that the ability of parties to trade and contract lies at the
center of the theory. The economic analysis of property rights has since attracted
the attention of a large number of scholars: The volumes by Yoram Barzel and
H arold Demsetz give an introduc tion to the work of two of the more prominent
of these. U seful discussions of current thinking about the Coase theorem are
provided by Robert Cooter and J oseph Farrell.
The importance of residual returns for the theory of the firm was developed more
fully by Armen Alchian and Demsetz; residual control as a virtual definition of
ownership was championed by Sanford Crossman and Oliver H art. Oliver
 322
Efficient Williamson, Grossman and Hart, and Hart and John Moore have contributed
I ncentives: especially importantly to the analysis of efficient ownership patterns.
Contracts and The importance of specific investments was first developed in the context of
Ownership human capital theory by Gary Becker and was elaborated in the theory of the
firm by Williamson and by Benjamin Klein, Robert Crawford, and Armen
Alchian.

• REFERENCES
Alchian, A. , and H. Demsetz. "Production, Information Costs, and Economic
Organization," American Economic Review, 62 ( 1 972), 777-97.
Barzel, Y. Economic Analysis of Property Rights (Cambridge: Cambridge Univer­
sity Press, 1 989).
Becker, G. S. Human Capital: A Theoretical and Empirical Analysis, with Special
Reference to Education (New York: Columbia University Press, 1 964).
Coase, R. "The Problem of Social Cost," fournal of Law and Economics, 1
( 1 960), 1-44.
Cooter, R. "The Coase Theorem," The New Pa/grave: A Dictionary of Economics,
J. Eatwell, M. Milgate, and P. Newman, eds. (London: The Macmillan Press,
1 989), Vol. 1 , 4 5 7-60.
Demsetz, H. Ownersh ip, Control, and the Firm (Oxford: Basil Blackwell, 1 988).
Farrell, J . "Information and the Coase Theorem, " fournal of Economic Perspec­
tives, 1 (Fall 1 987), 1 1 3-30.
Grossman, S. , and 0. Hart. "The Costs and Benefits of Ownership: A Theory
of Vertical and Lateral Integration," fournal of Political Economy, 94 (August
1 986), 69 1-7 1 9.
Hart, 0. , and J. Moore. "Property Rights and the Nature of the Firm, " fournal
of Political Economy, 98 (December 1 990), 1 1 1 9-58.
Klein, B. , R. Crawford, and A. Alchian. "Vertical Integration, Appropriable
Rents, and the Competitive Contracting Process, " fournal of Law and Economics,
2 1 (1 978), 297-326.
Williamson, 0. "Transaction-Cost Economics: The Governance of Contractual
Relations," fournal of Law and Econom ics, 22 ( 1 979), 2 3 3-6 1.

EXEHClSES

Food for Thought

1. Foster City, California, is a city that was built largely on land reclaimed
from the San Francisco Bay. In the early 1 960s, when the homes were built there,
the developer made the unusual arrangement of selling only the houses to the
purchasers while retaining ownership of the land, collecting a lease payment equal to
about 5 percent of the value of the land on a 30-year lease on each property. Between
1964 and 1 989 land values in Foster City rose by about 5, 000 percent, that is, by a
factor of about 50. In 1 989 the Foster organization announced that it planned to
increase the lease payments proportionally when the leases were renewed. In response,
the homeowners formed an organization and hired a lawyer to prevent the huge
planned rent increase. Discuss this situation in terms of the theory of cospecialized
assets. Who is being held up? What inefficie ncies, if any, would you expect to result Ownership and
from this unusual arrangement? Property Rights
2. Many fi rms re nt office space from others under contracts that run for periods
of about 5 years. Typically, the ren ter owns the furn iture an d equipmen t used in the
office and is responsible for maintaining it, but the owner of the buil di ng is responsible
for maintaining the common areas (hal lways, l obby, grounds). The building owners
own the win dows, but the renter may own the window treatments (draperies, bl inds).
The buil ding owner is respon sible for maintaining the electrical system and air
conditioning, but the renter may be responsible for decorating and painting the walls.
How can you account for this pattern of ownership rights and responsibil ities? What
kinds of companie s woul d you expect to own their office space rather than to lease
it?
3. In the late 1980s the U nited S tates and J apan became embroiled in a dispute
over the use of huge "drift nets" to catch fi sh in the Pacific Oct;an. These are nets
that may be as much as 40 miles long and are all owed to drift in the water, catching
whatever comes into them. What kinds of economic and moral issues woul d you
expect to arise in these tal ks? What kinds of enforcement issues woul d be most difficul t
to resol ve? Why?
4. Ownership patterns vary around the worl d, even within free market
economie s. In the U nited States and Canada, whe n a house or apartme nt is sold, the
kitchen cabinets are al most invariabl y sol d with it. In Europe, the opposite is often
the case, with kitchen cabinets being treated as fu rniture. What differe nce s in kitchen
design woul d you expect to find betwee n Europe and N orth America? Discuss whether
these differe nces cause or are caused by the different ownership patterns.
5. Two main systems of water rights exist in the U nited S tates. According to
one system, downstream water users are entitled to an uninterrupted flow; according
to the other, the first users are entitl ed to continue their use at historical levels, but
they forfeit their rights if they reduce their actual usage. What difficulties would you
expect in trying to modify each of these systems to make water rights tradable?
6. Publ ishers of even the smallest dail y newspapers usually own their own
presses, but even the l argest book publ ishers normally contract their printing j obs to
independe nt printers. What accounts for this diffe re nce in who owns the printing
presses?
7. When airl ine flights are overbooked in the U nited S tate s, an auction is
sometimes hel d to see which passe ngers are will ing to transfer to a later flight in return
for compe nsation. The compe nsation is determined by the l owest price needed to
induce the required number of people to give up their seats. This seems to work fairly
well. Would a bidding system work for deciding whether a flight shoul d permit
sm oking? Why or why not?
8. When Neil Armstr ong first stepped on the moon in 1969, he cl aimed it on
behalf of the U nited State s for all of humanity-the moon is not U . S. territory. The
earl ier tradition back on Earth was for Western European expl orers to claim any
country they " discovered" for their monarchs, even if there were obviousl y pe opl e
inhabiting the lands al ready. What are the economic merits and disabilities of each
pattern? What of their ethics?
9. Condominiums and housing cooperatives have become increasingl y im­
portant institutions in the U nited S tates in rece nt years. Condominiums are now
common in most states, and co-ops are a maj or feature 0f the New York housing
market.
Condominiums are mul tifamil y living units-usuall y apartme nt buildings-in
which each resident owns his or her individual living unit pl us a share of the common
areas: the grounds, hal l w�ys, exterior structure, etc. Ownership of a unit entail s
 324
Efficient membership in the condominium association, which makes policy decisions, collects
Incentives: fees to maintain and improve the common property, and contracts for maintenance
Contracts and and other services. The association operates under rules that are subject to maj ority
Ownership votes, with voting power allocated either by ownership share or equall y, one vote per
unit. Owners are generally free to sell their units (and, with them, their shares in the
common areas) to anyone, although the association may have a right to match any
offer and buy the unit. Owners typically obtain their own individual mortgage l oans
to finance purchase of their separate units.
In a housing cooperative, the members all own the whole building together,
and individuals then obtain long-term leases from the co- op for particular residential
units. The residents are, then, collectively their own landlord. Because no individual
owns any particular piece of the building, mortgages are obtained by the co-op.
Further, individuals are not free to sell their interests to anyone they please: the co­
op members must approve any new members. Decisions are again made by the
members collectively, although they typically would delegate such questions as interior
decoration to the individual tenant-members.
What do you see as the relative merits and deficiences of each of these patterns
of ownership? H ow do they compare with investor- ownership under which the tenants
rent their units?
 Part

V
EMPLOYMENT : CONTRACTS,
COMPENSATION, AND CAREERS

10
EMPLOYMENT POLICY AND
HUMAN RESOURCE MANAGEMENT

11
INTERNAL LABOR MARKETS, JOB
ASSIGNMENTS, AND PROMOTIONS

12
CoMPENSATION AND MOTI VATION

13
ExECUTIVE AND MANAGERIAL COMPENSATION
 10
EMPLOYMENT POLICY AND
HUMAN RESOURCE MANAGEMENT

[Classical economic theory's] way of viewing the employmen t con tract and the
management of labor involves a very h igh order of abstraction-such a high order, in
fact, as to leave out of account the most striking empirical facts of the situa tion as we
observe it in the real world.
Herbert Simon 1

In advanced economies, labor earnings typically account for about two thirds of
national income. In the service industries, which constitute an increasing share of
economic activity, expenditures on labor are proportionally even more important. At
the same time, the share of direct labor expenses in the total costs of modern
manufacturing firms has fallen relative to older patterns. Nevertheless, employees'
efforts, skills, knowledge, creativity, and dedication are perhaps even more crucial to
manufacturing firms today than previously. Devising and implementing effective
policies to attract and retain good people, to help them develop to their potential, to
utilize their skills and knowledge, and to inform, motivate, and reward them are thus
among the most important tasks that managers face. These human-resource policies
and the way they are managed have tremendous effects both on the success of the
organization and on the quality of life that the organization's members experience.
Our approach to human-resource and compensation issues is the one we have
taken throughout this book: We assume rational and largely self-interested behavior;
we presume that people seek efficient solutions to the problems they face; we examine
efficiency explanations for the institutions and practices we observe; and we look to
identifiable transaction costs to account for divergences from efficiency. For much of

1
"A Formal Theory of the Employment Relationship," Econometrica , 1 9 ( 1 9 5 1 ), 293-30 5 .
 ;J27
Employment
Marg i nal Revenue Prod uct Policy and Human
Resou rce
Management
.....
0
::, w
Q)

� w
C)

Figure 10. 1 : In the classical economic model,

the firm hires workers until the wage paid to the
last worker is equal to that worker's marginal
N N revenue product. Lower wages lead firms to hire
N u mber of Workers H i red more workers.

the analysis, we further simplify by assuming that there are no wealth effects and then
apply the value maximization principle.
This purely economic approach may seem seriously incomplete when applied
to human-resource issues. After all, we are discussing people and the relationships
among them. People get satisfaction from their work and from a job done well. They
often define themselves in terms of their jobs, identify with their employers, and judge
their self-worth in terms of their success at work. The relationships among people
working together are social, not just economic. Assuming a cold-blooded economic
calculus thus may seem to be a vicious caricature, and its use might seriously
misdescribe actual employment relations and misdirect policies.
All these objections may be correct but, as we will see in this and the following
chapters, an economic approach has great value in helping to organize issues and
understand phenomena and in imposing discipline on our analysis.

THE CLASSICAL THEORY OF WAGES,

EMPLOYMENT, AND HUMAN CAPITAL
The study of labor markets, employment, and wages is a major element of standard
neoclassical economics. Although this approach has serious deficiencies in explaining
a variety of actual practices, it is still worthwhile highlighting some of its central
features.

Wages and Levels of Em p loyment

In the classical economic analysis, the conditions of supply and demand at any time
determine the levels of employment and wages. Firms in such a world have no
discretion over the wage levels paid to their workers. Instead, firms take the wages of
different classes of workers as given by market conditions and decide how many
workers of each type to hire. As illustrated in Figure I 0. 1 , a firm's optimal policy is
to hire additional workers of any one type until the last worker's marginal revenue
product-the incremental i ncome that hiring the worker brings to the firm-is just
equal to the wage rate. Until this level of employment is reached, additional workers
contribute more to revenues than they do to costs, and so it is worthwhile to hire
more people.
The hiring decisions made by all the buyers of labor combine to determine a
demand curve for labor of each type. Workers' decisi ons about whether to seek
employment at the going, market-determined wage or instead to go to school, stay
home to raise a family, travel, start a new business, and so on, determine the
 328
E mployment: corresponding supply curves. The market wages for each type of worker are set by the
Contracts, interaction of supply and demand in the usual way.
Compensation , Workers in any world described by this model would be highly mobile, staying
and Careers with an employer only if that employer offered the workers a more preferred
combination of wages, hours, job location, and so on, and otherwise moving­
possibly every day-to the employment where their skills and effort are most highly
valued. Firms would raise and lower their employment levels frequently as shifting
demand for their product changed the workers' marginal revenue products. Wages
and overall employment would fluctuate with changing market conditions as well,
with the two continually adjusting to keep the marginal revenue product of each type
of worker ever equal to that type's wage.

Human Ca p ital
Workers' productivity is not solely a function of their innate strength, dexterity,
intelligence, and the amount and quality of the physical capital that they have to use,
but also of their human capital. Human capital refers to the knowledge and acquired
skills a person has that increase his or her ability to conduct activities with economic
value. Examples of human capital are a carpenter's ability to build a staircase and a
lobbyist's contacts and knowledge of the workings of government. Human capital is
most often acquired by experience or through training by others who already have the
requisite skills. It is the factor that differentiates raw labor power from skilled expertise
at some task or job and is a crucial determinant of productivity.
It is useful to distinguish between firm-specific and general-purpose (or
nonspecific) human capital. The former includes skills and knowledge that are valuable
only in the context of a particular firm, whereas the latter involves skills and knowledge
that increase the person's productivity when working for any of several different
employers. Examples of firm-specific human capital include knowledge of the
idiosyncracies of the particular firm's machinery or its accounting system, the special
terminology and decision procedures used in the firm, and the needs of the firm's
customers. Examples of general human capital include knowledge of how to operate
a type of machine or prepare accounting statements, a familiarity with general business
terminology and procedures, and general skills in sales and marketing.
In a labor market with no specific human capital, no firm would ever invest in
training workers to improve their general-purpose human capital. Such training
would only increase the workers' market-determined wages, reflecting their increased
productivity and value to other employers. There would be no payoff for the investing
firm. In contrast, workers would have reason to develop general human capital that
makes them more productive and valuable to a variety of potential employers because
that would lead to higher earnings for themselves. A farm worker, for example, might
learn to harvest fruit more efficiently, but could not expect much training from the
farmer. Similarly, a young manager might pursue an MBA degree at night, but his
or her current employer would be unlikely to help pay the tuition.
It might seem worthwhile for an employer to invest in workers' specific human
capital, that is, in knowledge and skills that are valuable only in this specific firm. There
is no market pressure to pay workers more for these skills because they do not increase
the workers' value to other employers. There is little reason for the workers to make
such investments, however, and unless they stay with the firm for an extended period,
the firm itself will gain on ly a small fraction of the potential benefits of the investment.
For example, an itinerant farm worker has little reason to invest in learning about the
peculiarities of a particular farmer's fields or equipment, and the farmer gains little by
investing in training about those things. When workers are highly mobile, the levels
of i11vcstmc11t in firm-specific human capital arc apt to be quite low.
Defects of the Classical Model Employment
Some labor markets do seem to operate in much the way the classical model describes. Policy and Human
Resource
An example is the market for itinerant farm workers, who move from employer to
Management
employer as new crops are ready for planting or harvesting. For many other
occupations-especially ones with highly trained workers-the actual markets seem
to work very differently. Pay levels fluctuate much less than would be predicted by
the classical model; pay cuts, in particular, are quite rare. Actual turnover and worker
mobility are limited relative to the predictions of the model: After an initial period
early in the average worker's career when mobility is high, workers frequently continue
to work for a single employer for a period of many years. Firms do not adjust
employment levels so frequently as the classical theory would suggest, and they often
eagerly sponsor training programs for their employees that augment both general and
specific human capital. Meanwhile, employees invest in acquiring skills and knowledge
that are valu�ble only in the firms currently employing them, and they do so expecting
that they will be rewarded. Simply put, the elementary classical model presents a very
poor description of employment relations in advanced economies.

LABOR CONTRACTS AND THE

EMPLOYMENT RELATIONSHIP
To begin understanding some of the features of actual labor markets and employment
policies, it is useful to consider the nature of the relationship between the employer
and the employee.

Em p loyment as a Relationship
The conclusion that pay equals productivity at each point in time is based on the idea
that labor is bought and sold in spot markets, where a given wage payment purchases
a given, observable amount of a particular, well-specified kind of work to be performed
at a given time and place and under given conditions. Such markets do exist. The
vast bulk of the transactions for labor services, however, are mediated by complex
contracts that cover longer, often indefinite periods of time. These contracts often
uncouple pay and productivity. They are incomplete and involve important implicit
elements, and they assign significant authority to the employer. Instead of a simple,
arm's-length market transaction between buyer and seller, most employment actually
represents a complex, long-term rela tionship between the employer and employee.
We seek in this section to understand some of the reasons for and consequences of
these relationships.

Em p loyment Contracts
The employment contract is typically quite imprecise. The employees agree that­
within limits that are rarely completely described and only partly understood-they
will use their minds and muscles to undertake the tasks that the employer directs them
to do, perhaps using the methods that the employer specifies. The employer agrees
to pay the employees. The range of actions that might be requested or required is
unclear. Future compensation and even the criteria used to determine fu ture pay and
promotions are unspecified. The mechanisms to be used in case of dispute are not
stated, nor are the penalties for most possible violations of the contract. Yet these are
among the most important contracts that any of us enter throughout our lives.
Union contracts are more explicit on all these dimensions, but in some countries,
including the United States, they cover only a fraction of the work force. This in itself
puts into even sharper focus the questions of why most employment contracts are
 330
mployment: incomplete and largely implicit and why they give such discretionary authority to the
Contracts, employer.
mpensation, The incompleteness, the implicit nature, and the shape of the employment
·------1and Careers contract are all responses to the impossibility of complete contracting. The factors
preventing complete contracting are discussed in earlier chapters. Briefly, they involve
the difficulties of foreseeing all the events that might possibly arise over time and the
appropriate actions to take, the difficulties of unambiguously describing these events
and actions even if they could be foreseen, and the costs of negotiating acceptable
explicit agreements over these many terms even if they could be described. Union
contracts attempt to specify terms more fully and explicitly. They are notoriously
costly to settle, and the rigidities that they impose are a frequently noted source of
inefficiencies. The usual employment contract instead is much more a relational
contract: It frames the relationship, specifying broad terms and objectives and putting
in place some mechanisms for decision making when unforeseen events occur.

AL1THORITY 11\J EMPLOYMENT RELATIONS The decision-making mechanism in the

employment contract is basically that the boss can order the employee to do anything
that is not explicitly forbidden by the contract's terms or by law. Strikes and formalized
grievance procedures aside, the employee's only real recourse is to quit, which he or
she is typically free to do at any time. At the same time, at least in the United States
and many other industrialized countries, the employer is basically free to fire the
worker when and if he or she wishes: Employment is "at will. "
The usual Marxian explanation for this pattern is that the employer has all the
power in the bargaining, and the contractual relationship that emerges simply reflects
this distribution of power, maximizing the employer's gains by exploiting the employee's
weakness. However, many detailed aspects of the contract can be understood better
as attempts to increase the total value of the relationship.
First, in any particular circumstance it is often efficient that one party clearly
have the residual decision-making authority. Otherwise, the costs of bargaining will
be incurred whenever a situation arises that was not covered explicitly by the contract.
These costs, in the form of delay or failure to reach agreement, could be fatal to the
success of the organization. Imagine what would happen if the employer and employees
had to bargain anew every time a new order came in, a shipment was late, or a
machine broke down. If assigning all the authority to one party presents too great a
danger of harmful opportunism in some circumstances, then the efficient contract
will provide for alternative governance procedures-either different decision-making
processes that involve more parties or constraints on the decision maker's allowable
choices. In most circumstances, however, the efficient solution is for the authority to
rest with a single party: Someone needs to be the boss.
Note that this argument applies even when there is a single employee. When
multiple employees are involved, and their tasks must be coordinated, there is a
second advantage to having one person with decision-making authority. Even if the
boss were not needed actually to coordinate the employees' actions-they might be
left to determine the plan of action by themselves-having someone who has the
power to specify and enforce a plan if the employees cannot agree among themselves
and who is responsible for the plan's success contributes to achieving the desired
coordination.
Second, it is often desirable that the same party have the residual decision­
making authority in all events not explicitly foreseen in the contract. Otherwise,
events would have to be described with sufficient precision to specify who has the
authority. It is possible that events could be described well enough to permit this,
although the contract still would not spell out the actions to be taken in these events.
In deed, the bankruptcy laws essen tially take this form: Residual deci sion-making Employment
auth ority shifts to the court-appoin ted trustee in any even t where bankru ptcy occurs. Policy and f luman
Nevertheless, there are advan tages to having a uniq ue residual decision maker at any Resource
on e poin t in time. Management
The th ird issue is: Wh o sh ould h ave th e auth ority and be the residual decision
maker? There are two angles to take on th is, each relating to a particular theory of
what a firm is.
THE FIRM AS A NEXUS OF CONTRACTS One theory views th e firm as a legal ficti on­
a contracting en tity that serves to ec onomize on the number of bilateral contrac ts th at
are needed to coordinate activity. Actual work and decisions are carried out by
individual people wh o .rely on one an other. It would be costly and wasteful, h owever,
for every pair of workers to have th eir own contract and for th ere to be c ontracts
between each worker, supplier, and customer. Instead, each party enters a single
contract with the legal fiction th at i s th e firm, economizing on informati on and
contrac t expenses. I f th e ac tions of N different individuals need to be c oordinated to
achieve potential benefits, th en th ere would have to be N(N - 1)/2 separate bilateral
contracts to link all th e parties directly. By creating a "firm," only N c ontracts must
be written, one between each individual and th e firm. Th e difference in th e number
of req uired contracts gr ows very rapidly as N increases. Wi th N = 5, the difference
is 10 - 5 = 5; with N = 50, th e difference is 1, 1 75; and with N = 500, th e difference
is 124,250. With imperfect, inc omplete contracting and c ostly communication, th ere
are th en coordi nation advantages to having th e representative of th e firm decide what
is to be done wh en unforeseen circumstances arise because th e firm is th e unique
entity th at is a party to all th e c ontracts.
Th is still leaves open the issue: Wh o sh ould appoint the boss? Th e arguments
in Chapter 9 regarding th e efficient allocation of ownership suggest th at in many
circumstances it i s appropriate that the owners of capital (or th e owners' agent) h ave
the residual decision-making power. But wh ich capital? Th e decisions th at are made
under th e employment c ontract will affect the value of both the human capi tal and
the ph ysical capital bei ng used in th e firm. If th e workers' human capital is at greater
risk, th en it would be efficient to give th em th e residual deci sion power. Th is is
effectively done in many h uman-capital intensive businesses. For example, university
professors have the ultimate decision-making power in regard to academic matters i n
thei r institutions, and many accounting, c onsulti ng, architectural, and medical­
practice firms are organized as partnersh ips so that th e owners of human capital h ave
the residual decision righ ts. In most employment relationsh ips, h owever, most of the
ri sk is arguably borne by the owners of the ph ysical capital, and it is th en efficient to
gi ve them th e decision power.
THE FIRM AS A BEARER OF REPUTATION A second, complementary theory views th e
firm as a reputati on bearer. A maj or potential motivation to treat oth ers fairly is to
build and maintain a good reputation th at will facilitate future deali ngs. As we saw
in Chapter 8, the value of a good reputation increases with th e number of times it
may be used. Th us, if one party to a relationshi p has a longer horiz on th an anoth er,
the first party, wh en put in a position of power, will have a stronger incentive to build
an d maintain a reputati on with the other parti es for using power well. Th is argues for
giving th e residual decision power to wh ichever party to a relation ship has th e longer
h oriz on.
A firm generally can conti nue in exi stence after any of th e indivi duals origi nally
involved in it are long gone. For example, th e H udson's Bay Company i s over 300
years old. Further, it seems reasonable that reputations can attach to firms rath er than
just to in dividuals. There are books listing firms that are considered to be good places
 332
Employment: to work. Firms also have reputations for making good- or bad-quality products, and
Contracts, credit reports are issued partly based on the individual firm's past financial record.
Compensation, Then a firm, with its potentially unlimited lifetime, would have much stronger
and Careers incentives to act in a way that builds and maintains a good reputation than any mortal
person. This incentive would be accentuated by the greater frequency of dealings
within any given period of time that a firm would have relative to most individuals.
A firm that behaved well would have a good reputation to which would attach a
stream of rents; one that acted badly would get a bad reputation and would be less
valuable.
Of course, the firm itself does not take the actions that lead to a reputation, nor
does it have use for the rents that accrue to a good reputation. People act in the name
of the firm, and it is they who value the resulting rents. Thus, we are apparently back
up against the regrettably short horizons of individual mortals. If, however, those who
are in a position to influence the reputation of the firm can sell their claims on the
rent stream that attaches to a good firm reputation, then they will have an incentive
to make sure the reputation remains unsullied. This is true whether they affect the
reputation directly by their own actions or indirectly by determining the incentives of
those who do act in the name of the firm. Thus, both the decision power and the
ownership of the rents should attach to those who can most easily and effectively
transfer their claims at full value.
In most contexts, this would seem to be the investors in the firm, whose claims
take the form of marketable securities, rather than, say, laborers who would need to
sell their jobs to transfer their claims. This is another reason why the boss as the
person with the power to make decisions in unforeseen eventualities should very often
be the representative of the providers of physical capital, rather than the providers of
labor.

I mplicit Contracts
Although the written employment contract is quite incomplete, it is supplemented by
unwritten, often completely implicit understandings. These implicit contractual terms
govern many of the crucial elements of the employment relationship, including pay,
work assignments, and employers' and employees' duties to one another. They are
not expected to be enforced by the courts or other third parties. Rather, they are
intended to be self-enforcing: They are structured so that the parties have incentives
to abide by them for fear of the consequences of violating the agreement. As described
in Chapter 8, the incentives in self-enforcing agreements require that each party
receive a stream of rents from the relationship, so that each will be hesitant to do
anything that endangers its continuation.
For agreements to be enforced by courts or other outsiders, these third parties
must be able to understand the agreement and verify how the contracting parties have
behaved, so that they can determine accurately whether the contract was violated.
For self-enforcement, it is necessary only that the affected parties understand their
obligations to one another, that they can observe each other's behavior, and that each
party enjoys a sufficiently large stream of rents from the contract. The weaker
informational conditions of self-enforcing contracts tend to enlarge the scope for
implicit contracts over explicit ones relying on court enforcement.
As a simple example of how an implicit contract might work, suppose a firm
and its workers would like to agree that the workers will work especially hard and that
the firm will pay them a bonus at the end of each period if they do so. The workers
can always violate the contract by not working hard, and the firm can violate it by
falsely claim ing that the workers have shirked and then refusing to pay the bonus that
has been earned. Even if the firm and the workers observe the actual effort level,
 333
court enforcement requires that the workers be able to prove that they actually worked
at the agreed level when the firm claims they shi rked. Sim ilarly, the firm must be
Employment
Pol icy and H u ma _
able to prove that the workers shirked when they claim they did not. Resou rce
For a self-enforcing, implicit contract to work, it is enough that neither party Management
be able to gain from cheating. The workers' pri ncipal threats are to quit or to refuse
to continue working hard. For these threats to deter the firm from cheating, the
arrangement must be beneficial for the firm as well; that is, the firm must be getting
rents or quasi-rents from the arrangement that it would lose if the arrangement were
terminated. On the workers' side, shi rking and not receiving the bonus must be worse
than worki ng hard and bei ng rewarded, so the bonus must at least offset the costs of
the extra effort. Moreover, the workers always have the option of quitting and going
elsewhere, so adheri ng to the contract must pay them more than (or at least as much
as) this alternative. Thus, in total there must be a strictly positive surplus from the
agreement, �nd each party must share in it.
Note that it is the prospect of future gains from maintaining the relationship
that provide the incentives under the implicit contract. This in itself gives a reason
for employment relationships to be enduring, long-term ones.
THE ROLE OF FI RM-SPEC IFIC HUMAN CAPITAL This arrangement requires the existence
of rents or quasi-rents. Rents are likely to be competed away in any long-term
relationship but quasi-rents, which represent a normal return on past investments,
can be lasti ng. The returns on investments in firm-specific human capital might be
a source of the needed quasi-rents. Once a firm has trained its workers in firm-specific
skills, the workers will be more valuable to the firm than new replacement workers
would be. Thus, as long as they are not paid the full value of the extra output of their
firm-specific human capital (and note that there is no market pressure to make them
receive this extra value), the firm would earn a quasi-rent from their continued
employment that might be enough to deter its cheating and having them quit.
Meanwhile, as long as the workers do at least as well working hard and receiving their
bonuses as they would by shirking or quitting and going elsewhere, they will have no
reason to violate the implicit contract.
Risk Sharin g in Em p loyment Relations
In the classical model of labor markets, worker incomes can fluctuate substantially
over time. Workers are constantly paid their marginal products, which change not
just with thei r personal conditions but also with the state of demand faced by firms
and the prices and productivity of other inputs. This arrangement is inefficient in the
normal case where workers are risk averse. Moral hazard and adverse selection impede
shifting these risks to those outside the employment relationship, and , to some extent,
within it as well. Nevertheless, such risk sharing can and does take place .
EMPLOH.IENT AND INCOME SECURITY The firm's owners will often be in a position to
take on some of this risk, insuring the workers' incomes against random fluctuati ons.
Indeed, one of the earliest contributors to the theory of the firm, Frank Knight, argued
that the shifting of risk and authority to the employer was the defining characteristic
of the firm. The entrepreneur takes the income risk, pays the employees a fixed wage,
and then takes the decision-making power to protect against moral hazard.
If the firm can be treated as risk neutral (perhaps because it is owned by : vesi:ors
with well-diversified portfol ios), then, other things being equal, it should seek to
insulate the workers fully agai nst income risk. This provides a value-creating service
for the employees while imposing no cost or risk premium on the risk-neutral firm .
The expected pay level can be reduced without harming the workers i f the firm takes
on some of the risk they would otherwise face, and so both sides can gain.
 If the firm is risk neutral and the workers' preferences between income and
334
Employment:
Contracts, leisure exhibit no wealth effects, then the firm should also insulate the workers against
Compensation , the effects of unemployment (because with no wealth effects, any kind of risk is
and Careers equivalent to some income risk, and we have already seen that it pays to insure income
risks completely). In this special case, the workers' utilities are made independent of
any shocks to their productivity or the fortunes of the firm. The pay-productivity link
would be almost completely broken: At the time of hiring, pay would be linked to
expected productivity, but afterwards there would be no attempt to match pay to
realized productivity.

works and there are no layoffs. If realized productivity is low, then the optimal amount
Even in this special case, full insurance does not mean that everyone always

of labor to use (defined by the equality of the marginal product of work and its
marginal disutility to the worker) is also low. There would be waste in maintaining
high levels of employment because what would be produced would not be worth the
cost. In that case, not everyone may be called in to work, and if the firm faces a
permanent reduction in its need for workers, some will be let go. However, those
who are not working because of temporary layoffs are still paid by the firm: The pay
level makes them as well-off not working as they would be if they had to work and
incur the cost of effort. Further, any workers let go will receive severance payments
that make up for the loss of their jobs. Thus, those who are working and those who
are laid off would be equally well-off under this ideal contract.

INCENTIVE PROBLE�IS AND PARTIAL INSURANCE The prescription that firms full y
insulate workers against income risk obviously is not followed in actual employment
relationships. There is some insurance, but it is only partial. It is important to
understand why this is the case.
The hypothesis that firms are approximately risk neutral is most applicable to
the kinds of random events that affect individual workers independently. For such
individual shocks, the firm can act much like an insurance company, balancing out
the losses incurred by some employees with the gains experienced by others. However,
as we argue in Chapter 7, insurance cannot work for all kinds of risks. In particular,
changes in such factors as general economic conditions, interest rates, or energy prices
have such widespread effects that the firm cannot be risk neutral with respect to those.
Moreover, if the company has limited access to financial capital, the possibility of
bankruptcy or of disruption in valuable investments makes it costly to insure workers
against the kinds of events that affect the overall prospects of the firm. Under efficient
arrangements, the workers will have to bear at least some portion of these kinds of
risks.
This fact creates a problem, however, when the firm has better information
about its financial health and prospects than do the workers or their unions. The
firm's management may be tempted to exaggerate financial difficulties in order to
justify paying lower wages to workers. To prevent this, contracts or bargaining must
be arranged so that the firm bears a cost of some sort for claiming the kind of hard

If there is a union, one possibility is that the firm might have to weather a strike
times that lead to lower wages.

to verify its claim. The strikes that have sometimes followed management demands
for wage reductions at troubled airlines in the United States are examples . Firms are
often more willing to weather a strike when anticipated demand is low because the
opportunity costs of a reduced level of operations are lower than they would be if
times really were good. Thus, such a willingness may be a credible signal of the firm's
information and assertion that times are hard. An alternative is that the firm may
have to engage in layoffs to convince workers that times are hard and wage concessions
 Employm
Policy and f
Resour
Marginal Revenue Product Management

I nsurance Figure lO. 2: The efficient level of

employment is N , but, with unemployment
N N insurance, the firm and workers can do better
Employment for themselves by choosing the lower level, N.

are needed. Once again, when anticipated demand is actually high, it may be too
expensive for the firm to reduce its work force j ust to feign fi nancial difficulty, so
layoffs can be a credible signal that anticipated demand is low.
In each of these cases, the firm provides only partial insurance to workers against
fl uctuations in thei r incomes and employment. Even when a risk- neutral government
agency provides unemployment insurance, however, full insurance may be impeded
by a moral hazard problem. As illustrated in Figure 10. 2, the firm may be able to
collude with the employees by laying them off and allowing them to collect
unemployment insurance, thereby increasing their total value, even when layoffs
would not otherwise have been the outcome.
Another possible source of incomplete insurance is the desire to provide
incentives to workers. We have seen in Chapters 6 and 7 and will see again in
Chapters 12 and 13 that providing effort incentives may require the workers to bear
some risk. This would also prevent compl etely insuring them against productivity
shocks if the effects of these shocks on measured performance are not freely
distinguishable from the effects of variations in effort. Alternatively, effort incentives
may be provided by workers receiving efficiency wages that exceed their outside
opportunities as long as they are not caught shirking and by their being fired otherwise.
If the technology for monitoring is imperfect and sometimes results in a hard worker
being incorrectly labeled a shirker and fired, then insurance is obviously incomplete.
One of the biggest problems with any attempt to insulate workers fully from
risk, however, is that they may behave opportunistically. To be fu lly insulated means
not only that the worker's wage does not fall during hard times or when the worker
is found to be of low ability, but also that it does not rise during good times or when
the worker is found to be of high ability. I n the latter case, however, there is a danger
that the workers may quit to pursue higher wages elsewhere. I n practice, it is easier
to protect workers from the risk of bad times and low ability than to collect a share
of the benefits they may receive when times are good or their abilities tum out to be
h igher than expected. This too l imits the possibil ities for employers to insure workers'
mcomes.
PARTIAL INSURANCE AND AcE-WAGE PROFILES The ideas we have j ust described have
been used to provide another explanation of the observed positive relationship between
age or experience and pay. S uppose that individual workers' abilities are initiall y
unknown both to the workers and to all their potential employers, but that high-ability
workers are more likely to produce larger output (however measured). Over the course
of the workers' careers, observations of their performance then generate information
 336
Employment: about their abilities. Each instance of good performance increases the likelihood that
Contracts, the workers are really of high ability, and bad realized results reduce this estimate of
Compensation, their talent. If the workers were always paid according to their (estimated) productivity,
and Careers their pay would vary up and down as the estimate of their abilities moved with their
realized performance. Bearing this risk is costly to the workers, and efficiency would
call for all the risk to be shifted to the firm if it is risk neutral. This would result in
the workers' pay being made independent of their estimated ability and productivity.
This complete risk-transfer solution is feasible only if both the employer and
the employees can commit themselves to it. The employer must be able to promise
not to lay the employees off or to reduce their wages if their productivity turns out to
be low, and the employees must be able to commit not to quit in favor of better­
paying jobs if their productivity turns out to be high. In practice, the firm is likely to
be better able to make such a commitment by using its reputation as a bond. The
workers would usually have more difficulty making such commitment in any credible
way because they have less use in their business dealings for a reputation as one who
does not quit. Another way to achieve commitment that is sometimes effective is to
write a legally binding contract. However, a long-term labor contract that binds the
workers to the firm and prevents them from quitting might not be enforceable in the
courts, which might treat the agreement as an unlawful slavery contract.

Predictions. Whatever the reason, if the workers cannot commit themselves, then
the complete risk-transfer solution is not possible. The most complete risk transfer
that can be achieved is for the employer to make a commitment not to lay the workers
off or to reduce their wages. If an implicit contract of this kind described actual
arrangements, what patterns would we expect to see in earnings data?
First, the workers' observed level of wages would never fall. They would however,
rise when the workers were particularly successful and the market's estimate of their
ability rose. These wage increases might be accompanied by job changes, if the highly
valued workers were bid away by other employers, but they need not be if the current
employers meet the competition. Thus, wages would tend to rise over the workers'
careers, even though their (actual but imperfectly observed) productivity remains the
same over their lifetimes.
As well, when workers do get raises or accept new jobs, their new wages will be
less than their currently estimated marginal productivity. The reason for this is subtle.
Competition among employers will mean that when a worker is hired, his or her
expected future wages over the period of employment must be equal to his or her
expected marginal product. For workers who prove to be especially productive, future
wages will increase to match. For workers who perform poorly, however, there will
be no wage reductions, so the firm will be paying more than the worker's marginal
product. Consequently, according to the model, the firm will, on average, overpay
its older, long-term workers. Because pay must equal marginal product over the
worker's career, it follows that, on average, the firm pays its younger, newer workers
less than their marginal products. In effect, the difference between estimated marginal
product and wage is an insurance premium paid by younger workers to guard against
future wage reductions.
An additional prediction is that for any given age and estimated level of ability,
wages will tend to be higher for workers with more experience and especially for those
with longer job tenure in the same firm. This happens for two reasons. One is that
the longer the worker's employment history, the less uncertainty there is about his or
her productivity and the lower the insurance premium that is deducted in determining
the worker's offered wage in any new job. The second reason, which reinforces the
first, is that a worker with longer tenure in the same firm is more likely to have had
his or her productivity overestimated since his or her last j ob change and conseq uently Employment
to be paid more than his or her cur rently estimated marginal product. Policy and Hu man
Fi nally, for any given level of experience and estimated ability or productivity, Resource
wages will tend to be higher for older workers with fewer years remaining to retirement. Management
Once again, the reason is that the implicit insurance premium is lower for older
workers because there are fewer remaining years during which the employer may have
to subsidize the wage of the worker if he or she proves to be relatively unproductive.
These predicted features have been found in the data relating to pay in the
United States and several other countries. Pay cuts are rare, although raises are not.
Wages do rise with age and with experience, even after controlling for productivity.
Finally, wages tend to vary more among workers as their experience i ncreases. 2
MANAGING AN EMPLOYMENT AND I NCOME SECURITY PLAN These analyses show that by
(even partially) insuring workers' incomes, the fi rm can reap gains in the form of
lower expected pay levels for any level of performance required from the workers.
Against this gain must be balanced the costs of providing this insurance. These may
be significant if the firm is not fully risk neutral, and they are especially painful when
business turns bad but the firm cannot reduce costs by wage cuts or layoffs. Income
insurance turns labor into a fi xed cost that cannot be avoided in economic downturns.
There is also the possibility that insuring incomes will mean divorcing pay and realized
productivity and that this will have bad incentive effects. These matters are addressed
in more detail in Chapter 12.
Despite this, many firms do seek to provide a measure of insurance. For example,
the "Big Three" U . S . automobile manufacturers-General M otors, Ford, and
Chrysler-bargained with the U nited Auto Workers union in the 1980s for changes
in work rules that enhanced flexibility and productivity. In return, the auto companies
promised unprecedented income insurance: Combining state-provided unemployment
benefits with firm-provided funds, laid- off hourly workers would receive up to 85
percent of their regular take-home pay for up to three years, unless they were called
back to work or found other j obs. B ecause even the most enj oyable j ob involves certain
costs, such as the expenses of commuting, meals away from home, wardrobe, child
care, and so on, the 85-percent income guarantee amounts to essentially full insurance
against the loss of take-home pay. In the first months of the economic downturn that
hit in mid-1990, more than 100 , 000 Big Three auto workers were put on temporary
or indefi nite layoff and began receiving these benefits.
Of course, the extent to which employees are willing to accept lower wages in
return for an income guarantee depends very much on the credibility of the promise.
This in turn depends both on the firm' s willin gness and ability to carry out the
promises. For example, in the recession at the start of the 1980s, the Big Three did
not fully meet their promises to supplement their workers' state unemployment benefi ts
because the funds they had set up for this purpose had run out and they were unwi lling
or unable to tap other sources of money for this purpose. G oing into the recession of
the early 1990s, the auto firms had accumulated much larger reserves to meet their
new contractual guarantees. H owever, there was some question of their ability to
make the payments if the recession proved to be especially deep or extended: General
Motors alone lost $1. 4 billion dollars in the last quarter of 1990, and together the B ig
Three lost another $2. 3 3 billion in the first three months of 1991. 3

2 J . L. Medoff and K. C. Abraham, "Experience, Performance and Earnings," Quarterly Journal of

Economics, 95 (1980), 703-36.
3 For more details see Gregory A. Patterson, "Hourly Auto Workers Now on Layoff Have a Sturdy
Safety Net," The Wall Street Journal (January 29, 1991), A- 1 .
 338
Employment: Some firms attempt to avoid layoffs altogether. Large companies in J apan follow
Contracts, this practice for their permanent employees (b oth blue collar and white collar workers),
Compensation , and the J apanese automobile companies have extended this policy to the hourly
and Careers workers in their U.S. production facilities. When weak demand necessitates production
cutbacks, temporary and part-time workers are let go, but the regular employees
continue, although early retirements may be encouraged. To keep the workers busy,
training and preventative maintenance activities are increased, and, if necessary, the
firm may increase its in-house production of parts that are normally purchased from

welding equipment, is an other example of a firm with a no-layoff policy. It maintained

suppliers. Lincoln Electric Company, the most successful U . S. manufacturer of

employment in the severe recession of the early 1980s, but only by assigning production
workers to such activi ties as repainting the factory. Such a demonstration of
commitment can be crucially important to the workers' believing the firm's promises.

Borrowi ng and Lending i n Em p loyment Relationshi ps

J ust as the firm can use its superior risk-bearing ability to insure workers' incomes

markets on their behalf and thereby realize a mutual gain. If the firm can borrow
profitably, so too can it use any superior access that it may have to capital and other

more, or at better rates, than can its employees, it may be efficient for the firm to
lend them money to allow them to consume in advance of income receipts. For
example, S tanford University lends money to its faculty to fi nance buying homes in
the very expensive housing market near the university. Another example comes from
pr ofessional partnershi ps, such as law firms and multiple-practitioner medical practices.
In these firms, new partners have to buy into the partnership, but it is common for
the firm to lend the necessary funds to those selected for partnership. In law firms
this is usually accomplished through a simple loan. I n medical practices, the new
partner is sometimes expected to buy his or her share of the accounts receivable over
five years, so that i n effect the other partners lend the new partner money over this
peri od. In towns dominated by a single employer, it was common for a company­
owned store to extend credit to employees against their paychecks for purchases of
food and clothing. Popular accounts of this practice often accentuate the difficult
plight of a worker heavi ly in debt to the company store. Nevertheless, because these
workers probably lacked access to other sources of credit, the loans may well have
enhanced efficiency.
I nstances of the firm investing for its employees because of better access to the
capi tal markets are less common. In many countries, however, pensions are tax­
favored forms of receiving income: No income tax is due on amounts paid by the
firm into pension fu nds, nor are the earnings of these funds taxed as received, although
withdrawals after retirement are taxed as they are made. These tax-code provisions
favor the firm's paying some part of compensation in the form of pension-fu nd
contributions, in essence saving on behalf of the employees. Deferred compensation,
under whi ch some of the money ear ned in one year is paid out in later years, also
involves an element of the firm's saving on the workers' behalf. Both of these also
have the effect of increasi ng the strength of the links between the employer and
employees.

RE< :H l ITINC. RETEl\TIOl\ . .\l\D SEP.\H.\TION

With employmen t being a complex, long- term, multifaceted relationship rather than
a sim ple marke t transaction, both sides will be particularly careful to avoid entering
the rel ation ship with an inappropriate partner, an d, on ce it is en tered, both will be
concerned with maintaining an d bui lding the value of the relationsh ip. S till, mistakes
will be made in the initial matching of employers and employees, and circumstances Employment
may change the value of the relationship for either or both sides. Thus, both must Policy and Human
also be concerned with ending the relationship when (and only when) this is Resource
appropriate. Management
From the employer's side, these concerns are the focus of human-resource
management. Managing the organization 's work force and maintaining the value of
the relationship with the employees is a central task of management-not just of those
specifically employed in the Personnel Department, but of all managers. The chapters
that follow examine two central issues in managing human resources: job assignments
and promotions (Chapter 1 1 ) and motivation and compensation (Chapters 1 2 and
1 3). Here we first address aspects of the problems of selecting and recruiting new
employees, of retaining them once they are hired, and of severing the employment
relationship when the time comes to do so.
These . are very broad issues: A full-scale treatment can absorb several full
graduate courses. Obvi9usly, we cannot attempt such a treatment here. Instead , we
focus on the difficulties imposed by the inevitable differences that occur in the
information available to (potential or actual) employers and employees. These
informational asymmetries are especially a problem in recruiting, where each side has
information about itself that the other does not have but that is important for judging
the quality of the potential match.

Recruiting
Organizations face an ongoing need to recruit new members. The decision about the
number and sort of people to recruit depends on the organization's strategy, its
technology, and its forecasts about the future.
Forecasts enter on both the supply and demand side. The demand side is fairly
obvious: A firm anticipating rapid growth will see a need for more people. Thus, the
U . S . investment banks expanded rapidly through the first half of the 1 980s, expecting
that the new products they were able to develop and market in the wake of financial­
market deregulation and the growing business opportun ities associated with corporate
takeovers and restructuring would lead to greater needs for professional staff. The
crash of the stock market in 1 987 and the decline of merger and acquisitions activity
in 1 989 cooled their hiring significantly, and in fact led many of them to reduce
employment.
Consideration of the supply side is important too. For example, some universities
began heavily recruiting faculty members in the late 1 980s and early 1 990s , even
though they had no immediate need for inore staff. They did so because they
anticipated that the impending retirement of the large number of faculty hired in the
expansion of higher education in the late 1 9 50s and early 1 960s, combined with no
apparent growth in the supplies of new faculty being trained, would yield consequent
labor-market tightness in the mid- l 990s.
The strategy and technology of the organ ization also have some fairly straightfor­
ward implications for recruiting. A dance company has little use for electrical
engineers, for example, whereas a firm that has adopted what we have called a
"modern" manufacturing strategy has a need for production workers who are able to
master multiple tasks and who can take individual responsibil ity for diagnosing
problems and responding to them .
The cost of firing or laying off employees is a significant element in the decision
about how many to hire. These costs are affected both by the firm's strategy and by
public policy (see the box entitled "Public Policy Toward Layoffs"). A firm that has
a policy of lifetime employment for its workers will be more reluctant to take on new
employees in the face of what might prove to be a temporary need than would one
 340
Employment:
Contracts,
Compensation,
Public Policy Toward Layoffs
and Careers
In various Western European countries, a firm that lets workers go is obliged
by law to pay them generous severance payments. For example, in France,
Israel, and Norway, a worker with ten years of service is entitled to a year's
pay on severance, and in Italy the figure is 20-months' pay. In Sweden, the
introduction of new legal restrictions on layoffs in the early 1 970s was
accompanied by rises in both vacancy rates and unemployment in firms.
Another study of Western Europe found that similar laws reduce employment
and labor-force participation and tend to increase unemployment. The study
also suggests that increases in the severance-pay requirements were responsible
for a significant part of the increase in unemployment that has been
experienced in a number of Western European nations during the 1970s and
1 980s. Although neither study addressed the effect of these laws and
regulations on the formation of new firms, it might be hypothesized that by
raising the costs of failure, they discourage start-ups.
There has been an active public debate in the United States concerning
the responsibilities of firms to employees in the context of mass layoffs and
plant closings such as those that followed some of the hostile takeovers of the
1 980s. There was a push for legislation at the federal level to require a firm
contemplating a plant closing to give workers six-months' advance notice.
Employers successfully resisted, and the proposal did not become law. Their
arguments were that such public announcements of strategic intentions would
be exceedingly costly because competitors would learn their plans. As well,
customers, fearing for the firm's viability, would be induced to seek other
sources of supply. Further, the firms argued that if such notice were valuable,
workers could bargain for it in the negotiations between individual firms and
their employees.

Sources: Eugenia Kazamaki , Firm Sea rch, Sector Sh ifts and Unemployment (Stockholm:
Swedish Institute for Social Research , 1 99 1 ), and Edward Lazear, "Job Security Provisions and
Employment, " Quarterly Journal of Economics, 1 0 5 (1990), 699-726 .

that has no such policy. This presumably has an effect on the hiring of permanent
employees in major Japanese firms, which are reknown for having such policies. The
rapid expansion of the Japanese economy over the last several decades, however, may
have dampened this effect by reducing concerns about the possible need to cut
employment.
At the same time, guarantees of long-term employment or protection in the
event of layoffs are attractive to risk-averse workers and thus help to attract people to
the firm. Job security was a major element in the collective bargaining between the
U. S. automobile manufacturers and the United Auto Workers union in their
negotiations during the 1 980s. The practice of such firms as IBM, Hewlett Packard,
and Lincoln Electric of avoiding layoffs has helped them in recruiting. Yet for some
occupations, a willingness to take risks is an important asset, and firms in such
businesses should be concerned about attracting people who are very risk averse. This
raises the problem of attracting the right sort of applicants.
 :34 1
ATl'IUCTINC APPLICANTS The first step in attracting applicants is making the Employment
organization and its needs known to potential members. Sometimes the problem is Policy and Human
in making them aware of the existence of the organization and the sort of opportunities Resource
it offers. For example, in the 1 980s business schools worldwide faced a continuing Management
shortage of faculty. In the North American business school faculty labor market, the
number of net openings in any given year generally was twice the number of people
entering the field from graduate schools. This gap persisted despite the schools' offering
salaries and working conditions that would seem extremely attractive. Academic-year
starting salaries for new faculty in the top U. S. business schools for 1 990 were in the
range of $7, 000 per month, with higher figures in some specialties (such as accounting
and finance). Teaching loads were low (perhaps four hours of actual classroom
teaching per week for 30 weeks per year), generous research support was provided,
and the prospects of eventually receiving tenure appeared good. It appears that the
schools did not do an adequate job of making the opportunities they offered known
to the relevant population. In other situations, however, the problem can actually be
an excess of applicants, as the box on applications to MBA programs illustrates. In
such situations, it can be useful to encourage self-selection among the applicants, so
that only those with the qualities the firm seeks will be attracted to apply.
SELF-SELECTION AND SCREENING IN RECRUITING Simply making what the organization
offers and what it expects known to the relevant group can induce some self-selection
among potential applicants. This can sometimes be a by-product of other policies.
The very clear image that IBM long projected of white shirt and dark suit with somber
tie always in place very effectively informed potential employees about the firm, so
that those who were attracted were the sort of people the company wanted. However,
policies can also be designed with the specific intent of affecting self-selection.
Recall that the basic idea i n screening is for the relatively uninformed party to
design policies or procedures that induce the informed to make choices that effectively
reveal their private information. In the present context, the privately informed parties
are the potential applicants, who know more than does the firm about their work and
risk attitudes, their abilities and i nterests, their long-term plans, and various other
personal factors that are of interest to the firm as a potential employer. 4 The choices
they make are simply whether or not to apply for a given opening. The idea is for the
firm to design aspects of its policies and structure to attract application from the sorts
of people it wants while discouraging those it does not want.
We have seen examples of this in earlier chapters. Pay that is linked tightly to
performance will be most attractive to those who know (or believe) themselves to be
unusually productive because they expect that this pay system will reward them
exceptionally well. Another example of screening is the use of seniority-based pay to
discourage applications from people who know that they are likely to leave the firm
after a relatively short time. Initially workers are not paid as well by a firm using this
strategy as they would be elsewhere, but they are compensated for this shortfall later
in their careers with pay levels that exceed their market opportunities. Only those
who intend to stay will be willing to join a firm using such a policy.
Other aspects of human-resource policy can help promote self-selection. For

4 Certainly the employer has private information about its characteristics that the applicants do not
have: what sort of people it is seeking; what job responsibilities , training and pay it will offer; what the
opportunities for promotion are; and what sort of experience it will be to work at this firm. H owever, the
applicant typically will be able to gather this information from observing the firm, talking to present and
former employees, and reading publications, and the firm's concern with its reputation can limit any
incentives it might have to conceal or misrepresent this information.
 342
Employment:
Contracts,
Compensation ,
and Careers
MBA Student Admissions
In the late 1 980s the leading graduate programs in business in the United
States each received many more applications for admission from well-qualified
people than they could possibly accommodate in their M BA classes. For
example, Harvard regularly got at least eight applications for each place in
its program, and Stanford got about a dozen applications for each place in
its class. The various schools take some pride in attracting so many exceptional
candidates, and they use increases in the number of applications they receive
as a competitive selling point. Yet, there is a sense in which a huge number
of applicants for each place is evidence that the schools are not doing a good
job recruiting students !
Handling and selecting from among all these applications is a monumen­
tal task for the schools. Each application must be logged in, checked for
completeness, read and evaluated by several people, and compared with
others. The candidates who are to be offered admission must be selected and
notified, and the others must be informed about their being rejected. These
are not the only costs: For the bulk of the applicants, the time and effort they
and their references put into their applications comes to nought.
To minimize these costs, the ideal solution would be for each school to
do such a fine job of communicating what it seeks in its applicants that the
only ones who actually apply are the ones who would be admitted from
the current pool and would accept the offer to enroll. Of course, communicat­
ing all this would be outrageously costly, especially because each school aims
to assemble the best class it can, which is not necessarily made up of the people
with the top test scores and grades. Many schools' admissions criteria involve
not only academic performance and ability but also managerial experience
and promise and the individual's contribution to the diversity of the class.
A subtle point is that even if perfect self-selection could be achieved in
this context, the schools would still need to examine all the applications. If
they did not, but instead simply counted on self-selection and accepted the
students as they applied, then the students would not be motivated to self­
select.

example, the benefit policy can be used to promote self-selection, as when a retail
clothing store offers employee discounts. The discounts tend to be most attractive to
those who are especially clothes conscious, and these people in turn are perhaps more
likely to be effective salespeople for fashion wear. The box on military service
academies offers another example.
SELECTIO;'-. CRITEHI.\ .\ND SIGN.\LING Once candidates have been identified, the
problem arises of choosing among them. Again, informational asymmetries can beset
the process. If the firm has done a good job of indicating what it seeks and what sort
of opportunities it offers, then anyone who applies wants a job. Not all applicants will
be equally desirable, however, and the employer must sort among them. Clever
interviewing and testing techn iques designed to elicit the information that the
applicants have and the firm needs may help overcome these informational difficulties
and promote good hiring decisions. Signaling may be also useful.
The example of signaling in the labor market via educational attainment 1s
 Employment
Policy and Human
Resource
Self-Selection at Military Academies Management
The three U. S. military service academies (at West Point, Annapolis, and
Colorado Springs) require their students to serve a term of active duty in the
military after graduation. A common explanation for this practice is that it
is simply a way of making the students repay the costs of the university
educations they received at government expense. If this explanation were
correct, one would expect to find that service academy graduates are paid
less than those who join the armed forces after completing a university
education on their own, which is contrary to the facts of the case.
An alternative explanation is that the military wants to attract applicants
who have a genuine interest in a military career, regardless of their abilities
to pay for a college education. The danger in offering a free education with
no obligation of military service is that some applicants may apply in order
to receive the free education, even though they have no real interest in
military careers. Students without a genuine interest in a military career are
likely to be discouraged from applying by the regimentation of the academies
and the military service requirement.

discussed in Chapter 5 . It explains why firms might select employees on the basis of
educational attainment or previous work experience, even if these are not directly
relevant to the job at hand, provided they are correlated with factors that are of
interest. For example, although the specific skills and knowledge gained in the last
year at the average high school may rarely be directly relevant on the job, many
employers will consider only graduates for employment because they perceive drop­
outs as likely to be unmotivated, undisciplined, or less able. Even such minor factors
as dress and demeanor in an interview can be signals. So long as those who have the
attributes that the firm seeks find it less costly or more beneficial to signal than do
those without the attributes, then signaling can arise and be useful to the firm.
Signaling arises from a correlation between observed choices and unobserved
characteristics. For example, high-productivity workers might signal their status by
their educational attainment. However, screening might also take place on other,
nonchoice variables if these are (believed to be) correlated with the unobservable
attributes the employer values. For example, if a prestigious school's admissions office
is thought to be especially effective in identifying talented individuals, then employers
may rationally be willing to give special consideration to those the school has admitted.
ScREENINC AND EMPLOYMENT DISCRIMINATION A related idea has been advanced to
explain discrimination in employment on the basis of attributes that are not directly
relevant to job performance: personal appearance, nationality, sex, ethnicity, and race.
A popular explanation for such discrimination is simply that the employers are
bigots who fear or despise members of the groups against whom they discriminate.
An argument against this explanation is that there would be profit opportunities in
hiring talented people from the groups suffering discrimination. These people have
more limited opportunities and so can be attracted more easily than can members of
the favored groups. Thus, employers who practice discrimination should be at a
competitive disadvantage compared to those who do not, and discrimination should
not be viable if competitive pressures are reasonably strong. A similar objection applies
to explanations based on prejudice by members of the majority work force. This kind
 344
Employment: of prejudice can explain workplace segregation, with prejudiced workers banding
Contracts, together in segregated workplaces, but it cannot explain the systematically lower wages
Compensation, paid to members of the disfavored groups.
and Careers One response to this criticism is that even if the employer is not personally
prejudiced, customers may be. Then it may be economical to serve this taste. If this
is to be the explanation of widespread discrimination, however, there must be very
widespread bias in the population. Otherwise, it would not be profitable to pander to
it so generally. Such a pattern certainly is possible but hard to reconcile with popular
support for public policies aimed at fighting employment discrimination.
An alternative explanation of discrimination is based on screening ideas. Suppose
that employers believe that some important but not freely observable determinants of
success are more likely, for example, to be held by men than women. Then the
employers may favor male applicants over female ones who have otherwise identical
observable characteristics. If the employers believe that the relative probabilities of
success are sufficiently in the males' favor, and if determining whether an applicant
actually is qualified is costly, they may even refuse to seriously consider women
applicants at all.
This pattern of behavior may be morally repugnant. It is, however, rational
behavior if the perceived correlation between sex and the likelihood of success is great
enough. The probability of finding a female applicant who is actually competitive is
thought to be so small as to not be worth the cost of considering female applicants
individually. Unfortunately, if all the relevant employers hold such beliefs, then
women never get hired for these positions. Then, even if the beliefs are completely
groundless, no disconfirming evidence ever is generated because women never get a
chance to prove the beliefs are wrong. Thus, the baseless beliefs survive, and with
them, the unjustified discrimination.
This discrimination is not just unfair; it is also inefficient because the women's
talents are not being put to their best use. This analysis points to the potential value
of equal-opportunity programs that require good-faith consideration of all applicants
and of affirmative-action programs that require special efforts be made to identify and
consider applicants from disadvantaged groups.
Although this sort of discrimination may be rational given the employers' beliefs,
the beliefs themselves may be inconsistent with the underlying facts. It might seem,
therefore, that the discrimination would not survive a little experimentation, for that
would reveal the true state of affairs. Suppose, however, potential employees need to
make investments in skill acquisition and that these investments are not freely
observable by employers. Then a pattern of discrimination can differentially affect the
incentives for investment. The members of the group discriminated against may not
expect to receive a full chance to benefit from the skills they acquir1_; and so will not
be as willing to invest as those in the favored group. Consequently, the two groups
actually do end up being differentially qualified, and the experiments rationalize the
discrimination: The bias is self-confirming.

Retent ion
The costs of recruiting are reason enough to be concerned about retention. Moreover,
oth er things being equal, experienced employees will tend to be more valuable.
I It ·,1 \ '\ C.\PIT \L .\ '\D Tt 'H'\0\'1-:R Either through explicit training or as a by-product
of being with the firm, experienced employees will have gained both general and
firm-specific human capital. General human capital is valuable in a broad set of
em ployment relationships, and so the worker must be fully compensated for the
increased productivity that it brings. Workers' general-purpose human capital thus
makes th em m ore valuable but also more costly, so that overall they are no particular Employment
bargain. I n contrast, firm-specific human capital is valuable only in the particular Policy and Human
employment relationship. I t can take many forms: knowledge of operating procedures Resource
in the firm, of local information sources, of locally specialized language usage, and Management
of customer, supplier, coworker, and machine idiosyncracies; special ski lls in tasks
peculiar to the firm; and membership in the social networks within the firm and
between the firm and its suppliers and customers. All of this increases the worker's
productivity in the firm. Yet because this capital is of little value outside the firm, it
need not be fully compensated, and workers with more firm-specific capital will tend
to be a bargain. 5 In losing these workers, the firm also loses the rents and quasi- rents
that they generate. Finally, losing a visible employee to a competitor may be taken
(both by insiders and outsiders) as a signal that the employee had some bad news
about the firm, its prospects, and its competitive position. This can make it harder to
recruit and to retain others.
One important response is simply to pay the workers for (at least some part of
the val ue of) their firm-specific human capital, even thou gh there is no direct
competitive pressure to do so. Paying workers more than their competitive opportunities
not only reduces costly turnover, it also encourages them to develop valuable firm­
specific human capital, and it can provide the basis for an efficiency-wage system that
leads employees to val ue their j obs and gives them incentives to work hard at them
(see Chapter 8).
OUTSIDE OFFERS An important issue in retention arises in deciding whether and
how to respond to outside offers that employees may receive. If the employer can tell
how valuable the outside offer is to an employee, then it can choose to match the
outside offer if the employee is suffi ciently valuable or otherwise let the employee go.
S o long as the employers know how the employee will respond, a bidding process
between competing employers would lead to a total value-maximizing solution.
This information condition, however, is not likely to be satisfied in reality. Any
j ob has certain nonpecuniary benefits and costs whose values are subj ective, known
only by the employee. Besides, people may form personal attachments to their
coworkers or communities; they may have pride or pleasure in their work and want
to see proj ects through to completion; they may value a firm' s style, prestige, or
corporate culture; they may have private estimates of the firm's prospects and how
these will infl uence their own futures; or they may simply fin d one sort of j ob more
or less pleasing to do than another. Given these uncertainties, it is very difficult for
the firm to evaluate an outside offer to one of its employees, and so it cannot easily
tell what it will take to match the offer. An incentive then arises for the employee to
shop for offers and to engage in disingenuous bargaining about how attractive these
offers are. This sort of behavior can be quite costly.
The Nonmatching Option. At the opposite theoretical extreme, when the firm
knows the employee' s productivity in the firm but has no information about the value
of outside offers, the best policy is simply to pay a fixed wage and never respond to
outside offers. The fixed wage is higher than what would be offered if there were no
threat of outside offers, with the extra amount being in the nature of an insurance
payment against having to incur the costs of losing the employee. It will be greater
the larger are the losses incurred if the employee leaves.
The great difficulty with this policy is that it requires commitment. Suppose a
valued employee comes in with an offer for more than he or she is currently being

5 Of course, the firm may have ha<l to pay for the creation of this capital at an earlier time when
the investment was being made, so the returns it gets now are not necessarily the result of exploitation.
 346
Employment: paid but much less than he or she is worth to the firm. The employee indicates that
Contracts, he or she did not seek the offer and would rather stay than go, if the firm will only
Compensation , do something about the pay. Refusing to deal in these circumstances is very hard
and Careers because there are clear gains to both the firm and the employee if he or she is given
a raise and stays. Dealing, however, deprives the no-response policy of any credibility
it may have had. One hope may lie in responding secretly. The idea would be to
explain to the employee the costs of having the no-response policy fail, tell the
employee that he or she is nevertheless too valuable to lose, and ask the person not
to tell fellow employees that he or she is staying because of receiving the raise.
Regardless of the ethical questions this raises, it is a dangerous policy to lie to
employees and to ask them to lie to one another.
A closely related strategy is to respond with valuable noncash compensation. At
many universities, for example, outside offers to professors are sometimes met by
promises of more control in departmental decisions, being named to a prestigious
research chair, better laboratory support, and so on. For individuals whose complaints
are salary alone, these responses set no unfavorable precedent, yet they can be very
effective i n helping to retain valuable personnel. At the same time, control and
research chairs, by their very nature, cannot be granted to everyone, so granting these
to one professor does not establish a clear precedent that others can follow.
Encouraging Outside Offers. Other organizations follow a diametrically opposed
strategy. They normally give minimal raises to everyone, and the only way to get a
good raise is to come in with an outside offer. This policy is used by some economics
departments and business schools where research is especially highly valued relative
to teaching.
The information conditions are, once again, the key to determining when such
a strategy may be useful. Evaluating research is notoriously difficult for everyone, but
most especially for administrators, and yet what a faculty member's research is worth
is likely to be much the same among institutions of comparable rank. The home
institution typically begins with only very imprecise information about what a faculty
member is worth . An outside offer from a close competitor reveals how highly other
comparable institutions value the professor's work. If the university's own estimate
confirms the competitor's view, then matching may be an appropriate response. At
the same time, if competitors are unwilling to make an attractive offer to the professor,
that is evidence that he or she is already overpaid and no exceptional raise is merited.
The problem with this scheme is that it forces talented people to spend time
seeking outside offers. This is costly for the university, costly for the offeror, and may
lead to the professor actually accepting the offer and leaving the university.
!\ IOBILITY CosTs A person who changes jobs incurs costs of several kinds such as
finding and evaluating another job, negotiating a new employment contract, breaking
social links, disrupting family life, and finding a new place to live and new schools
for the children. He or she may also suffer financial penalties imposed by the previous
employer. There may be golden handcuffs on the employee-large amounts of
deferred compensation and unvested pensions that will be lost by the departing
employee. Sometimes explicit bonds may be forfeited on leaving the firm. Changing
jobs will only be worthwhile for the employee if the new job is sufficiently better than
the old one to compensate for these transaction costs of moving.
[t might seem that the employer would want to institute such penalties because
they aid retention-employees can be retained at lower wages than would otherwise
be needed. The value maximization principle and the Coase theorem suggest a
different analysis, however. It would be inefficient for the employer and employee to
agree to any contract in which the employee is discouraged from moving when the
 347
Employment
Policy and Human
Resou rce
Deferred Compensation and Retention at BellSouth Management
B ellS outh is the fastest growing and most profitable of the regional telephone
companies, or "Baby B ells," which resulted from the c ourt-ord ered breakup
of the AT&T/Bell system, wh ic h had d ominated the U . S . telephone i ndustry
until the 1980s. BellS outh provides local telephone service in the southeastern
region of the U nited S tates, and it is also a fac tor in the fiber- optics and
cellular telephone ind ustries nationally.
B ellS outh reportedly employs an interesting deferred compensation
scheme to enc ourage employees to stay with the firm and ad opt a long-term
view. U nder this scheme, employees can place up to 25 percent of their
pay i nto· a special account. The company then augments the employee's
contribution. If the employee retires fr om the company, either at the normal
retirement age or by taking mutually agreed early retirement, then the account
pays off at more than twice the market rate of interest. On the other hand,
if an employee quits, the interest is c omputed at simply the market rate. The
difference can be immense, especially for more seni or people: Differences of (
hund red s of thousands of dollars are possible.
The moving costs that this scheme creates would seem to be an effective
means of limiting turnover, especially among people who have been in this
plan for long periods. N otice that the scheme provides the firm with an
opportunity to reduce the pay of more senior people relative to what it might
otherwise have needed to pay. Whether the firm exploits this opportunity or,
if not, how it commits itself not to, are i nteresting issues for study.

value created in the new j ob is greater than i n the current one. P enalties for quitting
would make the initial j ob offer less attractive to the worker and require that other
compensating benefits be paid. This will be worthwhile only when the penalties
discourage value-reducing j ob moves.

Separations
Just as a firm may need to attract new employees, it will sometimes need to reduce
employment, either permanently because of long-term shifts in its demand for labor
or temporarily because of transient shocks that affect productivity. The firm also may
have an interest in releasing particular employees who are not working out, even
when they would like to stay. At the same time, workers who find that their j obs are
a poor match for their talents and interests may want to leave. S uch separations of
employees are important for realizing labor-market efficiency, even in the c ontext of
a general pattern of long- term employment relationships.
EFFICIENT SEPARATIONS AND SURPLUS SHARING A maj or difficulty with achievi ng
efficient separations is that the party i nitiati ng the separation may not have the proper
incentives to take full acc ount of all the costs and benefits involved in the deci sion.
As we have already noted, a successful employment relationship typically requires
that both the employer and the employee be receiving at least as much fr om the
relationship as from their next- best alternatives and that overall there is a strictly
positive surplus being generated. N ow suppose now that circumstances change, and
one side finds that the relationship is no longer advantageous. For example, the
employee may find a better j ob elsewhere. Then he or she will want to quit if the
 348
Employment: excess of benefits received in the new job over those in the current employment are
Contracts, larger than are the private costs of moving. This calculation, however, ignores the
Compensation , fact that quitting would destroy the surplus that the current employer enjoys in the
and Careers relationship. Similarly, the employer's decision about firing an employee will take
account of only the employer's gains and losses, and not those that the employee
expenences.
One might think that a properly structured system of separation payments could
overcome these difficulties. The party initiating the separation would have to make a
payment to the other that offsets that party's loss of surplus, and so he or she would
internalize the full costs and benefits. There are several informational impediments
to a solution of this kind, however.
One problem would be determining the appropriate size of the payments. The
value of the relationship to each party is not likely to be freely known by the other,
and so the problems of bargaining with privately-known values will arise. Even if
these are avoided by settling for some standardized payment level s that do not depend
on individu.H valuations, a second problem arises. If different payments to be made
depend on whether the separation is a quit or a layoff, then the distinction can quickly
become blurred. An employer can often make an employee's life so miserable at work
that he or she just has to quit, or an employee can misbehave so badly that the
employer sees no choice but to fi re the offender, and yet third parties cannot tell who
is to blame. This makes a separation payment system very problematic because it will
not be clear who should pay whom.
SENIORITY AND LAYOFFS Many firms have a policy of "last-in, first-out" on layoffs:
The workers with the least seniority are let go first. This is frequently written into
union contracts, but some nonunionized firms follow this policy as well.
Firms would voluntarily adopt this policy if more experienced workers were
more productive relative to their wages than less experienced ones. As we have noted
before, however, the evidence is that the reverse is true: Wages increase faster with
experience than does productivity, so that inexperienced workers are paid less than
their marginal products whereas older ones are paid more. Thus, the direct incentives
would seem to be to lay off the older, more experienced workers fi rst.
Although these incentives are present, following them would undermine the
policy embedded in the increasing experience/wage profile. We have seen a number
of reasons for adopting this pay pattern, including providing effort incentives and
limiting turnover. Both are accomplished by paying workers less than their marginal
products early on, then making it up to them later in their careers by pay that exceeds
productivity. If workers feared that they would be laid off once they started earning
more than their marginal products, they would never enter into such (implicit)
contracts. Thus, firms have an incentive to buil d reputations for keeping on older,
more experienced workers, even when they seem overpaid relative to younger people
who are let go.
The threat to this policy is that the firm may be tempted to induce the senior
employees to quit, and others may not be able to tell that it really was the firm that
initiated the separations. Of course, a pattern of senior workers quitting and bitterly
denouncing the firm would eventually harm the firm's reputation, but this threat
might still not be enough to protect the workers fully. This may be one reason why
unions so often demand strict seniority policies, not only in layoffs but also in job
assignments and other aspects of job arrangements that the employer might be able
to manipulate to affect employees' welfare on the job.
I IUI\IAN CAPITAL IN\'ESTI\I ENTS Suppose there is an
A I\IBICL lOlJS SEPARATIONS AND
opportunity to invest in general (nonspecific) human capita l, but that employees
 :149
cannot afford to absorb the costs of this training themse lves. 6 An obvious solutio n Employment
would be for the employer to finance its employees' training i nvestments but then to Policy and Human
recoup the cost by paying a wage less than the employees' marginal product for some Resou rce
period of time. The problem is that anoth er firm might be willing and able to m ake Management
an attractive j ob offer that destroys the arran gement. A financial penalty of j ust the
right magnitude would discourage the employee from leaving just to capture the value
of the training, but would still permi t the employee to leave if he or she really were
more valuable to the other employer.
Thus, for example, some firms that underwrite their employees' study for
graduate degrees formally lend them the tuition. The company then forgives the debt
if the worker stays wi th the firm for a predetermined period after graduation. The loan
may, however, become payable in full if the employee quits before the requisite period
has expired.
This soluti on carries the risk that it may limit effi ciency-enhancing j ob changes
unless the employee has accumulated enough money to pay the penalty or the new
employer is willi ng to finance the employee's buyout from the old employment
contract. This does happen, of course, but now the new employer has to worry about
recouping its investment. A further potential problem is that the scheme may be
subj ect to moral hazard on the part of either the original employer or the employee
if it is hard for outsiders to disti nguish which party initiated a separation.
To see the difficulty, suppose that the penalty is set prior to the investment
being made, that the employee has agreed to continue working for a period after
completi ng the training at the same wage as before, and that the investment turned
out badly, so that the employee's productivity was not in fact i ncreased. There would
be an obvious problem if the employer could fire the employee in these circumstances
and collect the penalty, so suppose that the penalty is structured to become payable
only if the employee quits. S till, it may be possible for the firm to induce a quit, and
outsiders may be unable to tell who is actually responsible for the separation. Then
the firm earns nothing on the unsuccessful investment if the employee stays on but
receives the penalty if it can induce a quit. The obvious incentive is to drive the
employee out. Of course, fear of this outcome may deter the employee from investing
in the acquisiton of the human capital.
Similarly, if the employee can persuade the firm to fire him or her after the
investment worked out well, but outsiders cannot tell that the separation is really a
quit, then he or she receives the full benefits of the investment in higher wages in a
new j ob, and does not have to pay the penalty. This deters the firm from financing
the investment.

CASE STL1 DY: HUMAN - RESOURCE POL ICIES IN JAPAN

A properly constructed human-resource management policy is a system that must fit
together, with the various parts being mutually consistent and supporting. It must also
fit with the strategy and structure of the organization. Examination of some of the
characteristic features of the personnel policies employed in successful large Japanese
firms reveals such complementarities.

Hirin g and Retention

Two policies have been most frequently noted in discussions of human-resource
practices i n maj or Japanese firms: long-term employment guarantees for so-called

6 This condition is not necessarily inconsistent with the absence of wealth effects at the time the
initial employment contract is signed.
 350
Employment: permanent employees, and recruiting of permanent employees only at the early stages
Contracts, of their careers. Although these two features are actually not uniquely Japanese (as
Compensation, we will see in Chapter 1 1), they do appear to be carried further in Japan than
and Careers elsewhere. It is extremely rare for a major Japanese firm to dismiss a permanent
employee and similarly rare for one of the leading firms to hire someone at other
than one of the lowest, entry levels.
The first clear effect of these policies is in reducing turnover. A widespread
policy of hiring only at the bottom means that there are unlikely to be attractive job
opportunities in other major firms for experienced, mid-career employees. Meanwhile,
smaller firms do not usually offer as attractive compensation as do large ones. They
also give fewer advancement opportunities, have less job security, and are less
prestigious places to work. These limited opportunities for moving from major firms
both make retention easier and simultaneously reduce the costs that employees would
otherwise incur in monitoring the market for better opportunities. However, such a
system would load too much employment risk on employees if they did not have job
security: Being let go would not just mean losing this job, but also being very unlikely
ever to find a comparable new one. The permanent-employment policy mitigates this
danger, and so it supports the hiring policy: The two are complementary in the
language introduced in Chapter 4.

Protecting I nterests of Permanent Emplo yees

Still, the limited opportunities for Japanese permanent employees to move to
comparable jobs with other major employers means that much of their incomes are
quasi-rents: Large amounts could be taken away without pushing the employees' pay
below the levels available in their next-best alternatives and inducing wholesale
resignations. This would present a constant temptation in a firm that was run solely
in the interests of investors, and the employees' fear of this potential opportunism
would then make them leary about committing to a major firm. The control structure
of Japanese firms, which gives considerable power to the employees as a group, enables
them to protect their valuable employment rights.
In fact, in many ways it appears that the employees are residual claimants on
the Japanese firm's assets and have residual decision-making power at least on a par
with the investors in the firm. The typical firm's board of directors is made up almost
exclusively of senior managers who have spent their whole careers in the firm. Senior
executives see the interests of the employees as being as worthy of consideration as
those of investors, and they believe that employees' interests actually do guide policy
to a very important degree. Decision-making power is pushed down the managerial
hierarchy, often right to the shop floor, where worker groups are responsible for
determining how they will do the tasks they face, where employee suggestions are
eagerly sought and acted on, and where an individual has the right and duty to stop
the assembly line if a problem arises. Policy is formulated at the lowest possible levels
and becomes adopted and moves up through the hierarchy only by gaining concensus
support. Employees receive a large percentage of their incomes in the form of bonuses
that are tied to the overall performance of the firm. Meanwhile, only a small part of
the firm's cash flow is paid out to investors as dividends, with the bulk being reinvested
in the firm to permit its continued growth and ensure its survival.
Ttt\1'.\:li\C This system, in which employees need not fear for their jobs, expect to
share in the fortunes of the firm, do not anticipate leaving for another employer, and
have a say in the directions the firm will take, encourages them to invest in firm­
specific human capital. This is further encouraged by the policy of paying blue-collar
 :351
workers not for the particular j obs to which they are curren tly assigned, but instead Employment
for the skills they have acquired from a list established by the firm. At the same time, Pol icy and Human
low mobility allows the firm to finance the acquisition of even general-purpose human Resource
capital without fear that employees will use it to obtai n higher-paying j obs with a Management
different employer. This may explain why Japanese firms pa y significant numbers of
their employees to get graduate degrees in business and law at the firm's expense,
whereas European and N orth American firms rarely do so.
With high levels of firm-specific human capital, the decisions taken by the firm
place risks on employees' human assets that are comparable to those borne by i nvestors'
physical capital. Protecting the value of this human capital then requires that
employees' interests figure into the firm's decision making. This gives further reason
for employees' i nvolvement in decision maki ng, as well as for the policies of promoting
the survival and growth of the organization.

ASSIGNMENTS, PROMOTIONS, AND CONSENSUS An envi ronment of economic and

technological change means that a firm's employment needs will be changing too. In
this context, a guarantee of long-term employment would be tremendously expensive
if the firm could not reassign workers to new tasks as the needs arose. Thus, a system
of permanent employment differentially favors the sort of flexible work rules for which
Japanese firms are known. M oreover, to the extent that high levels of human capital
facilitate such adj ustments, investments in training enter as a supporting element in
yet another way.
Within the maj or Japanese corporati ons, promotions are very much based on
seniority. A new white-collar recruit can expect to spend at least a decade with the
firm before being considered (along with the others in the same cohort, who entered
the firm together) for a promotion. Then members of any experienced cohort are
evaluated for further promotions only after more senior people have had their chances.
Pay is tied to seniority as well, with individual merit or performance pay being rare.
Also, the differences in pay levels between ranks are typically smaller than in European
firms and much smaller than in N orth America. These systems would be infeasible
if there were an active outside labor market that could bid up the pay of star performers.
Thus, the promotion and pay policies rely on the practice of hiring only at the bottom.
The system of decision making by consensus requires that large numbers of
individuals have a role in decisions and that all affected parties be allowed to make
their vi ews, interests, and concerns known. This would leave the organization very
susceptible to extreme influence costs if pay and promotions were less a matter of
seniority and more sensitive to apparent qualifications and merit. In addition, the
policy of promoting growth, even at the possible cost of reduced i nvestor returns,
helps ensure that there will be good opportunities for promotion when the time comes,
and this too helps control influence activities and promotes effective concensus
decision making. The policy of reinvesti ng most of the cash flow also limits the
competiti on among groups and business units within the firm for the resources needed
to develop their proj ects. If the resources were more limited and internal proj ects had
to compete with paying the money out to investors, those inside the firm would
campaign much more intensely for their interests. Finally, the standard policy of
moving people around within the corporati on, rather than having them build their
careers in a single function or unit, also contributes to their taking a broad view of
their personal interests that is more i n line with overall organizational success, and
this too helps control infl uence activities.
To coordinate and manage this complex system, the personnel department must
be able to track employees effectively, ensure that their careers are developing properly,
 352
Employment:
and assign them where they are most valuable. Thi s is a crucial and complicated task
Contracts, that i nvolves managi ng an i mportant and valuable resource beari ng directly on the
Compensation, welfare of an important group of residual clai mants. Thus, the personnel functi on is
and Careers accorded great respect i n a maj or Japanese corporati on, and it is able to attract among
the best employees i n the organizati on.
We argue i n Chapter 4 that extensi ve complementari ties among the parts of a
system is one of the hallmarks of a coherent strategy. J apanese human-resource policies
are a good illustration of thi s general pri nci ple.
 Employment
Pol icy and Human
Resource
SUMMAHY Management
In classical economic theory, workers are treated as inputs to production; wages are
determined in the market, just like any other input price, and labor turnover is
frequent. Although this is a satisfactory approximation of the market for some unskilled
workers, it is not an accurate account of markets for most others. When the
development of specialized skills is important or the costs of changing jobs is high,
wages are no longer fixed by the market and can be set to help fulfill a variety of
human-resource objectives, including attracting and holding the right people, devel­
oping effective skills in the work force, motivating people, and insulating them from
excessive risk.
Employment relationships are complicated, and the contracts describing them
are highly incomplete. The details of the work that is to be done are most often
determined by a boss in whom authority is vested. Most often the boss is appointed
by the suppliers of physical capital. One possible explanation is that the physical assets
are more vulnerable to appropriation than human capital, and assigning ownership
to physical capital alleviates the hold-up threat. A second possibility is that the ways
that bosses make decisions determine the reputation of a firm, and it is important that
the builders of that reputation be able to capture some of the value they create by
selling it when they separate. Because human capital is not normally transferable, but
physical capital is, it is efficient to attach authority and reputational value to the
transferable physical-capital shares.
Human-resource policies can create value by insulating workers against the risks
associated with business fluctuations and uncertainties about their own ability . Any
such protection reduces the workers' incentives to invest in improving their own
abilities, and the firm itself may be unable to bear all the risks associated with business
fluctuations. In addition, employees may often be unable to commit not to collect
the benefits of favorable fluctuations. If the employer is committed to maintain
employment without cutting the employee's wage while paying an amount equal on
average to the employee's expected marginal product over the course of his or her
career, then ( 1 ) wages never fall but remain constant or climb over time, (2) more
experienced workers are paid more on average than are equally productive workers
with less experience, especially for experience in the same job, and (3) older workers
are paid more on average than equally productive and equally experienced younger
workers. Employers can sometimes also create value by saving or borrowing on behalf
of their employees.
Recruiting activity by firms is a function both of the firm's anticipated need3
and its estimates of future availability of appropriately trained workers. Once a firm
has decided how many of which kinds of workers it wants, it informs potential
candidates, attracts the right class of applicants, and then hires those whom it has
selected. The design of the job, including wages, responsibilities, and promotion
opportunities, all affect the numbers and kinds of workers who will apply. Once
applications are made, firms use a variety of indicators to decide whom to examine
most closely and whom to hire. A pattern of discrimination against certain groups of
workers at this stage (identified, for example, by race or sex) can discourage
disadvantaged workers from acquiring skills or applying for skilled jobs, leading to a
self-reinforcing pattern of discrimination.
Once a worker has been hired, the problem is to keep those workers who are
best suited for the job while encouraging separations (quits or layoffs) that increase
value. A difficulty in this is that if the rents or quasi-rents generated by the employment
 354
Employment: relationship are shared between the the firm and the employee, neither may have the
Contracts, proper incentives to account for the losses the other would suffer in a separation.
Compensation, Seniority rules may be useful in protecting more experienced and older workers against
and Careers opportunistic dismissals, particularly when the firm has adopted pay policies that lead
to paying senior workers more than their marginal products while paying j unior
workers less. A major complicating factor in any system of separation payments is that
it may be difficult for third parties to determine who has really initiated a given
separation. This difficulty also can interfere with the firm's financing an employee's
investment in human capital.
The distinctive human-resource management policies of major Japanese firms
have attracted much attention in Europe and North America. These policies in fact
form a coherent whole, with the principal elements mutually supporting one another
and other aspects of the firms' strategies and structures.

• BIBLIOGRAPHIC NOTES
The application of standard microeconomic theory to labor markets is discussed
in any intermediate microeconomic theory text. Gary Becker is most responsible
for developing the theory of human capital, but many authors have recognized
its significance and the concept is now a cornerstone of the economics of labor.
Herbert Simon introduced the idea of employment as a relationship into economics
and explained the crucial features of the employment relation as responses to the
necessary limitations on contracting that follow from bounded rationality. Oliver
Williamson, Michael Wachter and Jeffrey Harris developed these ideas further.
See also Williamson's books. Relational contracts in general have been explored
by Victor Goldberg. The conception of the firm as a "nexus of contracts" was put
forward by Armen Alchian and Harold Demsetz, while David Kreps developed
the treatment as the firm as a bearer of reputation.
The idea of implicit contracts was introduced by Costas Azariades, Martin Baily,
and Donald Gordon in the context of trying to explain wage rigidities for
macroeconomic purposes. The theory has been developed extensively since, with
particular attention to the possibilities for the firm to insure workers' incomes and
employment under various assumptions on the legal enforceability of contracts.

of Economics [98 ( l 983)]. A relatively nontechnical survey of the area was

Many of the key papers were published in a special issue of the Quarterly Journal

provided by Sherwin Rosen, while Oliver Hart has surveyed implicit contracts
when there are informational differences between workers and firms . Hart and
Bengt Holmstrom offer another useful survey that covers agency models and
models of incomplete contracts as well. The model of implicit partial insurance
contracts and wage dynamics is due to Milton Harris and Holmstrom.
References on screening, signaling, and self-selection are given in Chapter 5.
The analysis of discrimination based on screening was developed by Kenneth
Arrow, George Akerlof, and Michael Spence. The extension to treat human
capital investment is developed by Shelly Lundberg and Richard Startz and by
Sharon Oster and Paul Milgrom. Milgrom also showed conditions under which
it would be optimal not to respond to outside offers.
Masahiko Aoki provides an excellent discussion of Japanese firms' policies and
practices and extensive references. James Baron provides a constructive critique
of economic analyses of the employment relation from the vantage point of
research in sociology and social psychology.
• HEFEHENCES Employment
Pol icy and Human
Akerlof, G. "Discriminatory Status-Based Wages Among Tradition-Oriented,
Stochastically Trading Coconut Producers, " fournal of Political Eco_nomy, 92 Resource
Management
( 1 98 5 ), 26 5-76.
Alchian, A. , and H. Demsetz. "Production, Information Costs, and Economic
Organization, " American Economic Review, 62 ( 1 972), 77 5-95.
Aoki, M. "Toward an Economic Model of the Japanese Firm, " fournal of
Economic Literature, 28 (March 1990), 1-27.
Arrow, K. J . "Models of Job Discrimination, " Racial Discrimination in Economic
Life, A. H. Pascal, ed. (Lexington, MA: Lexington Books, 1 972).
Azsariadis, C. "Implicit Contracts and Underemployment Equilibria, " fournal of
Political Econqmy, 8 3 ( 1 975), 1 1 83-1 202.
Baily, M. "Wages and Employment Under Uncertain Demand, " Review of
Economic Studies, 4 1 ( 1 974), 3 7-50.
Baron, J. "The Employment Relation as a Social Relation, " fournal of the
fapanese and International Economies, 2 ( 1 988), 492- 5 2 5 .
Becker, G. S. Human Capital: A Theoretical and Empirical Analysis, with Special
Reference to Education (New York: Columbia University Press, 1 964).
Goldberg, V. "A Relational Exchange Perspective on the Employment Relation­
ship, " Working Paper No. 208, Department of Economics, University of
California, Davis ( 1 982).
Gordon, D. F. "A Neoclassical Theory of Keynesian U nemployment, " Economic
Inquiry, 2 1 ( 1 974), 43 1-49.
Harris, M. , and B. Holmstrom. "A Theory of Wage Dynamics, " Review of
Economic Studies, 49 ( 1 982), 3 1 5-3 3.
Hart, 0. "Optimal Labour Contracts U nder Asymmetric Information: An
Introduction, " Review of Economic Studies, 5 0 (January 1 983), 3-36.
Hart, 0. , and B. Holmstrom. "The Theory of Contracts, " Advances in Economic
Theory: Fifth World Congress, T. Bewley, ed. (Cambridge: Cambridge University
Press, 1 987).
Lundberg, S. , and R. Startz. "Private Discrimination and Social Intervention in
Competitive Labor Markets, " American Economic Review, 7 3 ( 1 983), 340-47.
Kreps, D. "Corporate Culture and Economic Theory, " Perspectives on Positive
Political Economy, J. Alt and K. Shepsle, eds. (Cambri dge: Cambridge University
Press, 1 990), 90- 1 4 3.
Milgrom, P. "Employment Contracts, Influence Activities, and Efficient Organi­
zation Design, " fournal of Political Economy, 96 ( 1 988), 42-60.
Milgrom, P. , and S. Oster. "Job Discrimination, Market Forces and the Invisibility
Hypothesis, " Quarterly fournal of Economics, 1 0 2 (August, 1 987), 4 5 3-76.
Rosen, S. "Implicit Contracts: A Survey, " Iournal of Economic Literature, 2 3
( 1 98 5), 1 1 44-75.
Simon, H. A. "A Formal Theory of the Employment Relationship, " Econometrica,
1 9 ( 1 9 5 1 ) , 2 9 3-3 0 5.
Spence, A. M. Market Signalling: Information Transfer in Hiring and Related
Processes (Cambridge, MA: Harvard University Press, 1 973).
Williamson, 0. The Economic Institutions of Capitalism: Firms, Markets,
Relational Contracting (New York: The Free Press, 1 98 5 ).
 356
Employment: Williamson, 0. Markets and Hierarchies: Analysis and Antitrust Implications
Contracts, (New York: The Free Press, 1 97 5).
Compensation, Williamson, 0. , M. Wachter, and J. Harris. " Understanding the Employment
and Careers Relation: The Analysis of Idiosyncratic Exchange, " Bell Journal of Economics, 6
( 1 97 5), 2 50-78.

EXERCISES

Food for Thought

1 . A recent study found that workers who were laid off generally received
lower wages once they found new jobs. 7 However, the gap between the old and new
wages differed systematically across various groups and by the cause of the layoff.
White-collar workers suffered relatively large declines compared to blue-collar workers,
and unionized blue-collar workers suffered the smallest declines. Also, workers who
had been in their previous jobs only a short period of time suffered relatively smaller
losses than those who had been with the employer for a longer period. Finally, those
who had lost their jobs because of plant closings experienced smaller wage reductions
than those who were laid off while their former place of employment stayed in
operation. How might you account for these patterns?
2. According to a study by a researcher at Boston University, 8 only 1 3 U . S.
firms with more than I, 000 employees have explicit policies never to institute a
general layoff: Delta Airlines, Digital Equipment, Federal Express, IBM, S. C.
Johnson, Lincoln Electric, Mazda, Motorola, National Steel, New U nited Motor
(NUMMI), Nissan, Nucor, and Xerox. Why do you think that so few firms have
adopted such a policy? Other firms, although apparently not having an official no­
layoff policy, in fact appear to follow such a policy. For example, Hewlett Packard
has consistently avoided layoffs, putting everyone on part-time work if necessary when
business was especially slow. What advantages and disadvantages might there be to
this approach?
3. Using I 980 data from 2 50 of the largest U. S. employer-provided pension
plans, Edward Lazear9 found that the present value of the pension received by the
average worker with 40 years of experience with the firm from which he or she retired
at the normal retirement age was $79, 476. If the same worker took "early retirement"
and retired I O years earlier, he or she would get pension benefits with a net present
value of almost twice as much: $ 1 5 8, 22 5. This was despite the fact that the worker
would have been with the firm a shorter time and so would have "earned" less pension
and also the fact that the employee's final pay (to which pension benefits are often
tied) would have been lower. These policies obviously encouraged early retirement.
How would you account for them? Now that employers in the U. S. can no longer
force mandatory retirement, would you expect that the extent to which they encourage
people to retire earlier would have increased or decreased?
4. German law mandates that the workers in a large corporation, although they

7 Robert Gibbons and Lawrence Katz, "Layoffs and Lemons, " National Bureau of Economic
Research working paper 2968 ( 1989).
8 Fred Foulkes, quoted in Milton Moskowitz, Robert Levering, and Michael Katz, Everybody's
Business (New York: Doubleday, 1990), 593.
9 Edward P. Lazear, "Pensions as Severance Pay," Financial Aspects of the U.S. Pension System ,
Orlcy Ashenfelter and Richard Layard, eds. (Chicago: University of Chicago Press, 1983).
 357
hold no ownership position, have the right to elect half of the firm's directors. This Employment
institution is termed "codetermination, " and its original intent was to ensure that work­ Policy and Human
ers' interest, and not just investors' concerns, were considered in corporate decision Resource
making. U nder what conditions would efficiency argue for such a system? Apparently, Management
investors' interests still seem to dominate. How would you account for this?
5 . Actuaries perform the crucial task in the i nsurance industry of estimating
the time paths and probability distributions of costs and revenues that attach to different
insurance contracts. Becoming an actuary involves passing a series of rigorous
examinations that are set by the professional actuarial society. Usually these are taken
over several years, during which time the prospective actuary is working for an
insurance company in the actuarial department. The skills and knowledge that these
tests measure are general human capital which the market rewards highly. Yet
insurance companies give large amounts of time off to their employees to study for
these examinations. Passing an exam usually results in a significant pay increase and
often i n a promotion. How do you account for these patterns?
6. Motorola, the U . S. electronics firm, reportedly has a policy that once an
employee has been with the firm for ten years, if his or her employment is to be
terminated, the president of the company must dismiss the employee in person. What
would be the point of such a policy?
7. Major Japanese firms appear to do much more pre-employment screening
and testing of applicants than firms in other countries. How would you account for
this? Japanese firms that have set up production facilities in the United States have
continued this practice. In part because the standardized aptitude and intelligence
tests that are sometimes used as part of this process are alleged to be biased in favor
of those of European or East Asian ethnic backgrounds, the extensive screening has
led to complaints of racial discrimination in employment. What should be the public
policy in this context?
 11
INTERNAL LABOR MARKETS, JOB
ASSIGNMENTS, AND PROMOTIONS

I 'I Io I

find men capable of managing business efficiently and secure to them the
positions of responsible control is perhaps the most importa nt single problem of
economic orga niza tion on the efficiency side.
F ra nk H . Knigh t 1

I n the developed economies, most people are in long-term employment

relationships with their current employers . The practi ce of nenko, or lifetime
employment, in large Japanese corporations is a fa m il iar exa mple, but stable, near­
l i fetime employment is widespread elsewhere as wel l . For example, it has been
esti mated that in the early 1 980s the typical U . S . worker was in a job that would last
about eight years, with a quarter of the work force being i n j obs that will last 20 years
or more. 2 For workers over 30 years of age, 40 percent were in jobs that would last
at least 20 years . In fact, job tenures of more than 1 5 years are apparently more
common in the U n i ted States than in Japa n , 3 and jobs lasting 20 years or more are
somewhat more common in the U n i ted Ki ngdom than in the U n i ted States. -+ Workers
in such long-term employment relationships expect to make a career with the fi rms

1 Risk, U11certai11ty and Profit (Chicago: U niversity of Chicago Press, 1 98 5) First published in
1 92 1 .
2 Robert l la l l , "The Importance of Lifetime Jobs i n the U. S. Economy," America11 Eco11omic
Review, 72 ( September 1 982), 7 1 6-24.
l Kazuo Koikc, "Japan's I ndustrial Relations: Cha racteristics and Problem�, " /apa11ese Economic
St udies, 7 (Fall 1 978), 42-90.
� Brian C. M. Main, unpublished paper cited by Hall, op cit.
now employing them, and many hope to rise through the ranks in their firms via Internal Labor
promotions. Markets, Job
Their employers share these expectations. They normally h ire new people into Assign ments, and
the firms only at a limited number of "ports of entry" that are often lower-level j obs; Promotions
they fill vacancies by reassigning or promoting members of the current work fo rce;
and th ey expect that workers will stay with the firm over the long haul, moving along
fairly well-defined career path s within the firm.

INTERNAL LABOR MARKETS

Long-term employment relationships, limited ports of entry fo r hiring, career paths

within the firm, and promotions from within are key features of internal labor markets.
An internal labor market consists of an employer and (some particular grou p of) its
lo ng-term employees. There may be more than one internal labor market within a
single firm, each involving different groups of workers. For example, the hourly
employees at each of a company's several plants might constitute a separate internal
labor market, with th e managerial employees in the firm as a wh ole making up yet
another. Each internal labor market interacts on only a limited basis with the general,
external labor market: M obility into and out of the internal labor market is limited
in practice, and the c onditions in the outside market exert only a muffled influ ence
on j ob assignments and compensation within the internal labor market. Rather
than simply and directly reflecting general market c onditi ons, an internal labor
market largely operates acc ording to its own administrative rules and shared u nder­
standings.

Labor Market Seg mentation Patterns

The concept of internal labor markets was first developed in ec onomics in th e early
1970s by Peter Doeringer and M ichael Piore. S ince then, numerou s economists,
sociologists, industrial relations experts, and organization behavior sch olars have
developed and used th e idea in studying employment patterns and policies.
N ot all workers are in internal labor markets. To understand employment
patterns and the prevalence or absence of internal labor markets in different contexts,
Doeringer and Piore di sti nguish ed primary and secondary sectors in the economy.
Internal labor markets are common but not universal in the primary sector, whereas
they are absent from th e secondary sector. Instead, the sec ondary sector is marked by
sh ort-term employment relationsh ips that h old no promise of promotion and where
wages are fully determined by market forces. The sec ondary labor market includes
unskilled, manual-labor, blue-c ollar j obs; low or u nskilled service positions (j anitors,
check-out clerks, waiters); low or unskilled wh ite-c ollar positions (office mail-room
workers, fili ng clerks); and migrant, part-time, and seasonal workers. Other j obs, such
as skilled blue-c ollar work, most wh ite-collar positions, and tech nical, managerial,
and professional employment, are in the primary sector.
Not all primary sector j obs are in internal labor markets. For example, a medical
doctor in private practice is not usefully th ought of in any internal labor market. Still,
he or sh e is certai nly not in th e sec ondary labor market that is characterized by low
skill levels, low earnings, easy entry, j ob impermanence, and low returns to education
or experience. M ost wh ite-collar workers, most skilled blue-collar workers, and most
professionals and managers, however, are in internal labor markets in the primary
sector.
Later sch olars have refined th e notion of the primary sector, subdividing it in
various ways according to differences in such factors as the level of entrance stand� rds,
promoti on and turnover rates, th e criteria for promotions, and the forms of control.
 360
Employment: For example, Richard Edwards divided the sector into "independent" and "subordinate"
Contracts, subsectors, with the latter containing most semiskilled blue- and white-collar jobs and
Compensation, the former including supervisory and managerial jobs, professional positions, and
and Careers crafts work where workers maintain a considerable degree of independence.
As noted, there may in fact be several internal labor markets within an individual
firm. Rarely do blue-collar, production workers move into white-collar jobs, except
perhaps by being promoted to supervisory positions. Movement between lower-level
office jobs and managerial positions is also rare. Yet within each group the firm can
offer extensive job ladders, with different rules and procedures for determining
compensation and promotions and different mechanisms for control and motivation.
F\IR'.':ESS A,o EFFICtE,cY Obviously , most people would prefer a job in the primary
labor market to the secondary, and a position in the independent primary submarket
would usually seem preferable to one in the subordinate submarket. Thus, the
assignment of people to different parts of the labor market can be approached as
primarily a justice or equity issue. It may just not seem fair that some jobs, such as
migrant agricultural labor, should offer so little in pay, security, and future promise
and yet involve backbreakingly hard work.
The demands made by jobs in the different sectors differ substantially, however,
and, unfortunately, not everyone is qualified for the really good jobs. People who do
not have good jobs and seem to deserve better might perform very poorly if they were
actually put into jobs for which they lack the skills and abilities. Thus, fairness is not
the only issue; efficiency matters too.
A more practical question is whether the way people are divided among
segmented labor markets reflects and promotes efficiency or, instead, retards efficiency
while also offending on fairness grounds. If the initial assignment of people to jobs is
determined by race, class, or sex, by personal influence or political connections, or
by random chance, then it is hard to believe that efficiency is being well served.
Furthermore, the limited mobility between internal labor markets then perpetuates
the misallocation. If, however, the assignment is reflective of relative abilities and

Pa y in I nternal Labor Markets

skills, then at least efficiency is being served.

A feature typically identified with internal labor markets is some insulation of

compensation decisions within the firm from external market forces. This insulation
cannot be total, of course. At a minimum, the firm must compete with other
employers at the ports of entry. More generally, workers will have outside options of
some sort throughout their careers. In addition, the employer may be willing to hire
at relatively senior levels from outside if the costs of filling a job from within are too
far out of line. For example, if a pressing need suddenly develops for employees with
expertise in a new technology, the firm may hire people with the needed knowledge
and skills from outside rather than attempting to develop the requisite expertise
internally . Or a company in distress may look outside for a new CEO to bring fresh
ideas and leadership that may improve the firm's fortunes. In such situations, the firm
will have to pay market wages to attract the new people. Nevertheless, for many
workers, wages are insulated to a large degree from outside market forces, helping to
reduce the income risks that workers bear and to make it possible to use wage policy
to achieve other objectives.
.l oR CusslFIUTIO:\S \:\D P \ Y In this context, it is often claimed that in internal
labor markets, "wages attach to jobs rather than to individuals. " The idea is that there
will be a fairly narrow range of pay specified for any job in the internal labor market.
What any particular employee is paid is then determined primarily by his or her job
 :36 1
Table I I . I Stanford University 1 990 Job Classifications and Monthly Pay Ranges I n terna l Labor
Markets, Job
Assignments, and
Range Promotions
Number Examples Min Max

B7 Office Assistant I; Medical Asst. I 1414 207 1

B8 Nursery Aid 1 485 2 1 74
B9 Secretary I; Off. Asst. I I , Sports Asst. I; Med. Asst. II 1 5 58 2285
BIO Sctry. IS; Accounting Asst. I ; Library Specialist I 1 639 240 5
BI I Sctry. 11; Off. Asst. III, Sports Asst. II; Med. Asst. II I 1717 2523
B12 Sctry.IIS; Acct. Asst. II; Lib. Spec. II 1 807 2654
B13 Editor I 2 1 97 2789
B14 Sctry. 111; Off. Asst . IV; Acct.Asst. III; Lib. Spec. III 1 98 5 2929
B1 5 Sctry. I I I (Premium) 2088 3082
B16 Administrative Assistant ("A. A. ") I ; Legal Secretary 2 1 97 323 1
B17 Graphic Composition Supervisor 2304 3393
B18 2423 3 560
B19 Patient Test Technician Specialist 2 547 3745
B20 267 3 393 1
B2 1 Medical Technologist 28 1 3 4 1 22
B22 Senior Medical Technologist 2950 4336
B23 3098 4553
B24 3252 4782
C4 Student Services Officer ("SSO ") I; Editor II; A. A. I 2 1 79 3302
C5 Assistant Librarian; Accountant I; A.A. II 2406 3766
C6 SSO II; Assoc. Libr' n; Acct 't II, Editor III; A.A. I I I 2668 422 5
C7 SSO I II; Libr'n; Acct't III; Editor IV; Manager I 3043 5080
C8 Senior Libr'n; Acct 't IV; Manager II 3 5 50 5856
C9 Manager I I I 4280 668 5
CIO Manager I V 5034 7797
NI I Manager V 6667 OPEN

assignment rather than by actual individual productivity or opportunity costs. There

is some evidence for this. Certainly, at least for lower-level jobs, many firms have
very explicit salary scales, and the only way to get a raise once an employee's pay hits
the top of the scale is to get a promotion.
Stanford University's practices on job classifications and pay are illustrative of
many large employers. Nonunion staff positions at Stanford are classified into 27
levels ranging from B7 to B24, C4 to C I O, C99, and N i l . (There is an even richer,
more complex classification of the jobs covered by collective bargaining. ) Jobs in the
B classification are "nonexempt" positions under U. S. government regulations. People
in these positions fill out time sheets and must be paid overtime for work beyond their
normal work week. Jobs in the C and N ranges are "exempt" positions that do not
carry overtime rights. These are usually professional, supervisory, or managerial
positions. Examples of some jobs in each class and the minimum and maximum
monthly wages for each are shown in Table 1 1 . 1 . 5
Jobs are classified according to a determination of the skills required and

levels: Secretary I, IS, II, IIS, III, and lll(Premium). Each involves a higher level of
responsibilities involved. Thus, for example, there are six different secretarial job

qualifications and responsibilities. Different sorts of jobs are classified as being in

the same range if they are determined to involve similar skill requirements and

5 No examples are given for the h ighest nonexempt (B) levels because no actual jobs a re currently
classified at these levels.
 Employment:, responsibil ities. For example, the jobs of Secretary I I I , Office Assistant IV, and
Contracts, Accounting Assistant I II are cons idered comparable and so they are classi fied the same
Compensation , (B- 14 ). The pay levels attached to a classification level are set at least in part in
and Careers response to market forces and are adj usted annually. H owever, differential market
demand and supply conditions faced by candidates for different jobs arc not supposed
to affect the classification of the jobs, at least in the short run . I nstead, they are to be
accounted for by adjusting the amount paid to people in different jobs \\'ith in the
speci fied range.
Once in a job, ind ividual workers may receive raises up to the maximum salary
for that job's classification . Typically, the total amount of these raises in any year is
control led by the central admin istration , but their allocation among employees is
determ ined by lower-level supervisors. Ra ises beyond the maximum salary can be
ach ieved only by being promoted to a job with a h igher classi fication or by ha\'i ng
one's current job reclassi fied . Thus, a Medical Assistant I (a B-7 j ob) m ight be
promoted to Medical Assistan t II (B-9) and then to Medical Assistant I I I (B- 1 1 ).
Similarly, a Manager I might win promotions up to Manager V . These patterns are
common to most la rge organizations with formal personnel systems, and not at all
unique to Stanford.

CHEATS .\ND QULIFICATIONS Despite all this, the idea that wages attach to jobs i n
internal labor markets must n o t be taken too rigidly. First, even with in t h e context of
a job classification scheme there can be significant differences in pay bet\\'een people
assigned to the sa me job. At Stanford, for example, the maximum salary corresponding
to a given job classi fication is on the order of 50 percent more than the minimum
for that rank, and the maxi m um at one level exceeds the min imum several grades
higher. This leaves quite a bit of room within any single job classification for merit
pay linked to producti\'ity and for pay differences that reflect seniority or other factors,
such as special individual needs or arrangements .
Second, many Japanese fi rms that h ave internal labor markets employ a "dual
hierarchy. " 6 Wages are determined not by task assignment or rank in the famil iar
reporting and authority hierarchy (the person's job), but by what skills the workers
ha\'e acquired and their performance-their places in the second , "rankings" hierarchy.
Thus, two employees with very different assignments may be paid the same, perhaps
e, ·en when one is superior to the other in the reporting hierarchy, because they ha,·e
the same positions in the ra nking hierarchy. Similarly, two people doing the same
job but at different positions in the ranking hierarchy can be paid quite differently.
Managing th is dual hierarchy is a complex task, and this helps account for the central
role, high prestige, and access to the most talented employees that the Personnel
Department has in a typical Japanese corporation. Increasingly, North American and
European firms \\·ith internal labor markets have experjmented \\·ith and adopted
similar policies, paying for skills ra ther than fo r the actual job being done. Th is
approach is especially favored in organizational systems which put a premium on
workers' being willing and able to move between multiple tasks quickly and smoothly
in response to changing conditions.

Tt I E R.\TIO:\ \LE FOR I NTERN.\ L LABOR I\ I A R K ET�

Long-term employment means lim ited job mobi lity, at least across organ izations. Y ct
the case and speed with "·hich labor mm·es to its most prod ucti ,·c use has often been

1
' For more dctai b, \Cc :\ la sahiko Aoki , Information , Incentives and Bargaining in the Japa nese

Economy (Ca 1 1 1bridgc: Cambridge l Jni,-cr�ity Press, 1 988), especially chapter �-

see n as a ma jor determ inant of economic efficie ncy and growth . What then accou nts I n ternal Labor
for the preva lence of internal la bor markets wi th their pa ttern of more-or- less pcrma1 1cnt Markets, Job
employment? And how ca n we explain their other fea tures: li mited ports of e ntry, job Assignments, and
ladders and promotion from wi thi n, a nd wages tha t arc a t least partially attached to Promotions
jobs rather tha n bei ng determ ined by indi vidual producti vi ty?

Long -Term Emplo yment

At least three distinct fa ctors contri bute to the efficiency of lo ng-term employment
relatio ns of the sort we see in i nternal la bor marke ts. These are the increased
opportuni ties to invest profita bly in firm-specific h uman capi tal, the grea ter efficacy
of efficiency wage incenti ve co ntracts in long-term rela tionships, and the enha nced
abili ty, to make an accurate assessment of an employee's co ntri butions to long-term
objecti ves by monitori ng performance over a longer period of ti me.

FIRM-SPECIFIC H UMAN CAPITAL Fi rm-specific h uman capital is knowledge, ski lls,

and interpersonal relationships that increase workers' prod ucti vi ty in thei r current
employment, bu t are useless if the workers lea ve to join other firms. Other th ings
bei ng equal, workers who have acqui red firm-specific human capital wi ll be more
efficient in their current employment than elsewhere. Th us, there wi ll be a grea ter
tendency for them to stay wi th their current employer than would otherwise be the
case. Fur thermore, long-term employment encourages the development of prod ucti ve,
firm-speci fic human capital: long-term employment and investment in firm-speci fic
human capital are complements. Unless both th e firm and i ts workers can expect the
employment rela tionship to conti nue, there is li ttle incenti ve to make costly investments
in acquiring firm-speci fic knowledge and skills. Moreover, even human capital tha t
is a free byproduct of working in the firm-such as the interpersonal relationships
that facili ta te worki ng together effecti vely-wi ll not develop unless the employment
relationship covers an extended period.

EFFICIENCY W ACE CONTRACTS The effecti veness of efficiency wages, reputation

mechanisms, and implicit, incomplete contracts increases when people take a longer­
term view, and this is encouraged by long-term employment. Consider, for example,
an efficiency-wage scheme under which workers keep their jobs and recei ve supercom­
peti ti ve pay if they perform bu t lose their jobs i f they are caught shirking. Th is requires
the possibi lity that employment rela tionships may extend over many periods. A longer
horizon also makes repu ta tion considera tions more powerfu l because it allows more
opportu nities to bene fit from a good reputation or to suffer from a bad one. This can
be important i n enforci ng implicit contracts.

ASSESSING CONTRIBUTIONS TO LONG-TERM GOALS For a team of fruit pi ckers whose

job is to pick trees clea nly, performa nce ca n be eas ily and qui ckly measured: count
the number of cleanly pi cked trees. There is no advantage to wai ti ng to assess the
picker's performance. At the opposi te extreme, i t is very difficult to evaluate a manager's
performance in selecti ng the new tech nologies in whi ch to inves t unti l some time,
perhaps a long time, has passed. It is difficult to provide incenti ves for long-term
performa nce to someone whose employment lasts for only a short time.
Th is time lag before good performance i ndicators become avai la ble can pose a
dou ble problem for some ki nds of managers-who partici pa te in making i nvestment
decisions and are also in charge of decisions that affect short-term performa nce. As
we have seen, the equal compensa tion principle requires that efforts in making short­
term decisions and long- term ones be compensa ted eq ually at the ma rgi n if a manager
is to devote some effort to both. In a short-term rela tionsh ip, offeri ng incentives for
 364
Employment: short- term performance would make the long-term incentive problem worse, because
Contracts, it would encourage managers to focus on immediate performance obj ectives while
Compensation, neglecting the long term. To m i tigate thi s problem, the firm might have to weaken
and Careers incentives for short-term performance as well. Thus, i nabi li ty to measure l ong-term
performance can degrade incentives for all ki nds of performance.

Promotion Policies
Prom otions serve two roles i n an organizati on. First, they help assi gn people to the
roles where they can best contribute to the organization's performance and success.
Second, prom oti ons serve as incentives and rewards. These conceptually di stinct roles
are sometimes in confli ct, which creates diffi cult management problem s but helps
account for many observed organi zati onal practices.
PROMOTION POLICIES AND JOB ASSIGNMENTS Hi gher- level j obs i n the hi erarchy typi cally
dem and higher levels of knowledge or ski ll and i nvolve assi gnments and responsi bilities
where successes are more significant and mi stakes or failures are more costl y. Assigni ng
people who actually have the requisi te quali ficati ons and talents to these j obs is very
im portant. Often, however, i t is difficult to tell immedi ately whi ch j obs are most
suitable for a newly hired employee, al though the best m atch is m ore easily di scerned
after the em ployee has been worki ng at the firm for some time. In that case, prom oting
employees currently in lower-level posi ti ons becomes the efficient mechani sm for
filli ng specialized and hi gher-level j obs. I n other cases, the quali ties needed in higher­
level j obs may be acqui red only through experience in the organizati on. This is
especial ly li kely where knowledge of operating procedures, technology, other em ploy­
ees' ski lls and tem peraments, and customers' and suppliers' characteri stics is im portant.
Again, prom oti on i s the effici ent m echani sm in these circum stances.
MANAGERIAL HIERARCHIES Decision making, supervision, and leadershi p at higher
levels in managerial hierarchi es have more im pact than do those at lower levels
because the j obs are desi gned that way. The lowest-level m anagers and supervisors
are given responsibi li ty for the most routi ne ki nds of tasks. They are expected to
consult wi th thei r superior managers for advice on nonroutine matters outside their
own experience, especially for any decisi ons that can have a maj or effect on the firm's
performance. Further up the hi erarchy, managers are accorded greater responsibility
and greater di screti on. Whereas the impact of a district manager's deci si ons i s largely
limi ted to the performance of that district, the deci sions of top management influence
the success of the whole firm . Thi s pattern has potenti al im plications for the assignm ent
of people of di ffering abi li ty levels to j obs. The key issue is whether i ncreasing the
effort, experience, abi li ty, and talent levels of the person assi gned to a j ob is greater
or less at di fferent levels i n the hierarchy.
Given the way managerial j obs are usually structured, it i s i n fact im portant to
have more productive people i n higher-level j obs. I n standard economic language,
the marginal returns to i ncreasi ng ei ther the effort provided by a manager or the level
of talent, abil ity, and relevant experience that he or she bri ngs to bear will be greater
the further up the manager is i n the hi erarchy. Thi s i n turn means that the firm will
want to put harder-worki ng and more talented managers in hi gher level positions,
wi th the very best people at the top. Agai n, to the extent that these characteri sti cs are
more easily di scerned am ong current employees than am ong outsi ders or are developed
primari ly through experi ence in the organizati on, prom oti on becomes the effi cient
way to fill higher- level positi ons.
PH0I\I0TI0NS . \ND Ot lTSIDE LAB0H MAH KET Col\1PETITI0N To some exte nt, outsi ders to
the firm rely on the promotion process to help them iden tify the ablest workers. Th ose
 36.5
Internal Labor
Markets, Job
Assignments, and
Bias in Promotions: Favoring Star Performers Promotions
An apparent form of unfair discrimi natio n comes in promotions when tho se
who have done well early i n their careers are favored at later ro unds. Fo r
example, they may be given better opportunities to show themselves to good
advantage, more resources to work with, or the benefi t of the doubt in
interpreting their performance. Those who have gone to the right schools
and done well there get the jobs wi th the best chances for advancement, and
then those whose early performance o n the job impresses the bosses get the
breaks o n later assignments. The same phenomenon is found in seeded
athletic to urnaments, where the highest-ranked players are given the easiest
early matches. In the job co ntext, such bias may seem to reflect unfai r
favoriti sm. However, i t also has certain advantages when the o bjective is to
identify the ablest workers.
As an illustratio n, co nsider a situation where a firm is trying to decide
which of two employees to promote into an important job. Initially each i s
as likely as the other to be the better choice. The firm has the choice o f
observing their performance either o nce o r twice, and the performance at
each stage gives information on qualifications for the job. Finally, suppose
that all the firm can observe is which employee performed better at each
round. Then if the two are treated the same at the second round of observation,
independent of which o ne did better at the first round, the firm gains no thing
from having the extra observation on thei r performance!
To see this, note first that if the firm observes the candidates' performance
only once, then it will optimally assi gn the one with the better performance
to the job. Now suppo se the seco nd o bservatio n is available and that the
seco nd round is no t biased in favor of the winner of the first ro und. Then
the firm has an optimal decisio n rule that ignores the seco nd period
information, namely, promote the winner of the first round. Thi s clearly is
the right thing to do when the same indi vidual does better at both rounds.
When o ne individual wins at the first round and the other at the second,
then there is at least as much evidence favori ng the first-round wi nner as
favoring the second, so it again pays to promote the first-round winner. The
opportunity to compare the employees agai n i n a fair co ntest, or in one
biased i n favor of the first round loser, has no i nformatio nal value.
Suppose instead that the firm makes it easier for the winner of the first
round to wi n the seco nd as well. Then, the firm's optimal decision will be
to promo te the winner of the second round, so the second round does have
informational value. If the purpo se is to identify the best candidate, then it
is optimal to bias the second round i n favor of the winner of the first round.

Based on Margaret Meyer, "Learning from Coarse Information: Biased Contests and
Career Profiles," Review of Economic Studies, 5 8 (January I 99 I ), I 5-4 1 .
 366
Employment: who are promoted, presumably, are those who have performed well in their previous
Contracts, jobs. Holding their ability fixed , promoted workers have better outside job opportunities
Compensation, than do unpromoted ones. They therefore have to be paid a higher wage. This has
and Careers two effects. First, promotions are practically always accompanied by pay increases.
Second, because a productive worker must be given a raise when he or she is promoted,
there may be some desire on the part of the firm to delay promotions and reduce the
number of job categories in order to maintain its information advantage over outsiders.
This temptation to withhold promotions may be especially great with workers who are
not already highly visible to outside employers, perhaps because they are not well
connected in the social networks of businesses outside the firm. Promoting these
people makes them more visible and increases the wage they must receive and the
chances of losing them. In contrast, there is little point in trying to hide workers who
are already quite well known outside.
PARALLEL JOB LADDERS The comparisons we have been describing, between two
successive levels of management or two kinds of secretaries or machine operators,
apply only when the successive ranks in the hierarchy entail similar kinds of tasks.
The best salesperson may not have the skills needed to be an effective sales manager,
even if his or her intimate knowledge of the product and market would be a great
benefit in the supervisory position. Similarly, it would be a strange coincidence if the
best teacher and researcher on a faculty were also the best candidate for dean. This
fact presents a special problem when the role of promotions as incentives and rewards
is recognized. Many an excellent engineer has been promoted to be a mediocre
manager, and others who would have been even worse managers but were passed over
for promotion have become disaffected or unmotivated. Tying pay to rank, as is
standard in internal labor markets, exacerbates these difficulties. In response, a number
of companies have recently been experimenting with separate career ladders for
scientists and engineers that do not require them to move into management in order
to get ahead. Examples are Analog Devices, Inc. , which makes equipment for
manufacturing semiconductors, and 3M Company, the makers of Scotch® tape, Post­
It® notes, and thousands of other products. IBM is another example. It has had some
extremely distinguished scientists make their entire careers in the company, winning
promotions and advancement without ever moving into management. Other firms,
however, believe that it is desirable for senior engineers and scientists to become
involved in management because they see this experience as helping to ensure that
those who discover and implement new technologies will also be sensitive to business
needs.
PROMOTIONS AS I NCENTJ\'ES AND RE\\:'ARDS It is clear that promotion opportunities
play a major role as an incentive device in most hierarchical organizations, either in
place of direct monetary incentives or in concert with them. This fact in turn helps
explain the common policy of normally hiring only at a limited number of ports of
entry that are usually low-ranking positions in the hierarchy. Suppose employees
constantly had to worry that none of them would get promoted the next time there
was an opening and that instead an outsider would be brought in above them. Then
their incentives to work hard to win the promotion might be severely blunted because
the chance of winning the prize would be much less.
The extensive use of promotions as incentive devices presents a puzzle, however.
It would seem that promotions are a very blunt tool, at least compared to direct
monetary incentives. Promotions are discrete, whereas monetary rewards can be varied
continuously as required. Promotions can go to a few people at most, whereas
presumably everyone can be given monetary incentives. Furthermore, promotions for
even the best people must be relatively infrequent. In contrast, monetary rewards and
 :367
punishmen ts can be meted out as frequently as desired. Promotions also create Internal Labor
competition between employees, which may be dysfunctional when cooperation an d Markets, Job
teamwork are im portant. Finally, as noted in the preceding section, there can be Assign ments, and
serious conflict between the role of promotions as incentive an d reward devices and Promotions
their role in assigning people to j obs. Promoting as a reward for good performance in
the current j ob can result in the famous "Peter Principle." This tongue-in-cheek
principle holds that people in organizations always get one promotion too many: They
keep getting promoted until they finally reach their "levels of incompetence"-j obs
they cannot do wel l-and then spend the remainder of their careers doing those jobs.
It is easy to see how this could happen under a system in which promotions are simply
a reward for good performance. Why then is there such reliance on promotions as
rewards?
Of course, if pay attaches to j obs through some routine administrative process
that is not responsive to individual skills, abilities, or performance, then the only
effective way· to award performance pay is to promote those who perform well to
higher-paying j obs. This is not an adequate answer to the question, however, because
it presupposes that pay has to be attached to j obs rather than to individuals.
TOURNAMENTS Providing incentives by promoting the best performers to higher­
paying j obs is a particular kind of comparative performance evaluation. The workers
are effectively ranked on the basis of their relative performance, and the winners get
promoted whereas the losers are passed over. Given the pay at each level, this system
creates a tou rnament. Just as in golf or tennis tour naments or the league playoffs in
various team sports, the amount a person receives for winning depends only on how
well he or she does relative to the other contestants and not on any absolute measure
of performance. The only performance information needed or used in a tournament
is the relative, ordinal information about who did better, not some absolute, cardinal
information about the amount by which an employee's performance exceeded some
absolute standard.
In many circumstances, only ordinal, relative information is available in a
timely manner and at a reasonable cost. It may be quite easy to determine and agree
about which of a gr oup of people is doing the best j ob. At the same time, it would
be extremely diffi cult to specify j ust how well each is doing in some absolute sense.
This may be especially likely when the j obs involve many different elements that are
hard to describe and to specify in advance. Yet this cardinal information may be
needed under any system that tries to reward performance by fi nely- tuned pay
adj ustments. In these circumstances, tournaments may be the only effective method
of providing incentives. Consider three examples:
• The faculty in a college may easily agree that student A in a diffi cult field of
study is doing better than is student B, who is following an easy course of
studies, even though B has higher grades overall. There is no apparent way
to specify how well each is doing absolutely, however, or even how much
better A is doing than B. Still, it is clear which student should get the award
as the outstanding graduate.
• Secretary A works for someone who requires a multitude of services to be
done quickly and accurately with skill, intelligence, and discretion. M oreover,
A regularly hel ps others in the office with their assignments. S ecretary B's
boss requires only that the phone be answered and that an occasional letter
be typed. Although both secretaries do everything that is asked of them in
exemplary fashion, A is clearly working harder than is B. There is no question
about which of the two is more deserving of a promotion. H owever, assigning
a numerical measure to their contributions in a manner that could be used
 368
Employment:
Contracts,
Compensation,
and Careers
Incentives In Tournaments: Golf
A recent empirical study has shown that effort and performance in tournaments
are sensitive to the siz e and dis tribution of the prizes. This s tudy found that
higher prizes lowered the scores in men's pr ofessional golf tournaments:
raising the total prize money by $100,000 (in 1984 d ollars) lowered each
player's score on average by I . I strokes over a 72-hole tournament. M ost of
this effect occurred i n later rounds of the tournaments when the players are
more likely to be tired and s o fi nd maintaining concentration more diffi cult.
Als o, players' fi nal- round scores res ponded to the s trength of the incentives
they faced. Because the monetary gain to fi nis hing firs t rather than second
is much greater than the gain to fi nishing second rather than third (and
similarly for all other places), there is a greater marginal return to a player's
lowering his fi nal round score by a s troke when he is among the leaders after
the firs t three rounds than when he is back in the pack. In fact, those with
a greater return to effort of this s ort did d o better in the fi nal round, even
when other factors influencing their scores are controlled for statis tically.

Based on Ronald C. Ehrenberg and Michael L. Bognanno, "Do Tournaments Have

Incentive Effects?" Journal of Political Economy, 98 (December, 1 990), 1 307-24.

to determine pay levels i n a performance-pay formula would be very diffi cult,

expensive, and subj ect to dis pute.
• Executive A has taken a faili ng division that was subj ect to severe foreign
competition and long- term labor problems and restored it to marginal
profi tabili ty while investing in new prod ucts that seem to hold real promise
for the future. Executi ve B has overseen the continued generation of cash by
a division that prod uces a high-demand product for which there is li ttle
competi tion. By all the standard fi nancial meas ures, B's divisi on outperformed
A's. It might have been possi ble to desi gn individual performance contracts
that pr ovided appropriate monetary incentives for the two executives in
their differing assi gnments, but doing s o would certainly have been hard;
determining that A has done the better j ob and should be the one promoted
is much easier .
A second reas on for usi ng tournaments to provide incentives arises when there
is a large common element to the uncertainty that affects the relationship between
people's effort choices and their measured performance. For example, the actual sales
made by i ndi vid ual sales people may all be affected in a similar fashion by the same
unobserved market factors, s uch as the actions of competitors or the market's acceptance
of a new product. We applied the informativeness pri nciple i n Chapter 7 to show that
comparative performance evaluation in general can be useful when observing how
one agent has performed. This allows a better inference to be made about how much
effort another has expended. When there is a large comm on element affecting each
agent's performance, looki ng at relati ve performance can effectively eliminate the
comm on factors and s o permi ts better inferences about who has worked hard. This
in turn allows the same incentives to be provided with less ris k being placed on thos e
being mo tivated. This possi bility may help explai n the po pulari ty of sales contests in
mo tivati ng salespeo ple.
 A third advan tage of tournamen ts arises when managemen t or owners m ay be Internal Labor
tem pted to renege on performance paymen ts that have been earned, saving mon ey Markets, Job
by claiming that performance was lower than it actuall y was and that only low Assignments, and
performance bonuses are due. For this to be a serious pro blem, it must be difficult Promotions
for employees to observe one another's performance (so that implicit con trac ts are not
effectively enforceable) an d for third parties to verify performance (so that arbitration
cannot resolve the problem). In these circumstances, it may still be possi ble to o bserve
whether the promised tourn amen t prizes have been awarded. A tournamen t, in wh ich
the prizes to be awarded are fixed in advance, then provides managemen t with a
mean s to commit to paying for performance. Without such commitmen t, the
possibilities for eliciting goo d performance by (incredi ble) promises are more limited.
A poten tial disadvantage of tourn amen ts arises when there are opportun ities for
collusion amon g the employees. I f rewards depend only on relative performance an d
not on absolu te performance levels, then the expected return s to each in dividual are
the same if everyone agrees to take thin gs easy as if they all work hard. Of course,
taking thin gs easy saves them the cost of effort an d represents a temptin g possibility
for the workers.
A secon d difficulty is that it may be as easy for a con testan t in a tournamen t to
get ahead by sabotagin g others' performance as by honest effort. This o bviously creates
inefficiency.
Even puttin g these pro blems aside, theory suggests that for tournamen ts to
provide efficien t incen tives, the prizes must be appropriately adj usted. In in tern al
labor markets the prizes are the extra pay attached to the higher-level job as well as
the opportunity for win n ers to compete for the next promotion. This mean s that pay
and promotion policies need to be determined together. Before discussing this, we
con sider some of the reasons why pay might attach to jobs.

Pay Attached to Jobs

As mentioned earlier, because a policy of promotions based on performance helps
competin g employers to iden tify the firm's most talen ted and hard-working emplo yees,
promotion s ordinarily require givin g raises, even when the n ew jobs are not more
diffi c ult. This might explain why pay tends to rise within a hierarchy, but it does not
explain why pay is not more tied to in divi dual performance or characteristics.

MEASUREMENT PROBLEMS One reason for attachin g pay to jobs arises when perfor­
mance and productivity are very difficult to measure with an y accuracy or even to
describe in sufficien t detail. This is especially likely to be the case for staff positions
in large, complex organ ization s, where it is hard to determine who has con tributed
what to the organ ization's performance. In such cases, the value of one person's
con tribution depen ds heavily on others' action s and decision s an d on factors that may
have been unforeseen at the time an y explicit performance con tracts would have had
to be written. Attemptin g to pay for performance in these circumstances may actually
amoun t to little more than a random pay policy. S imilarly, payin g for skills i s of little
value if skill levels are hard to observe and measure or, even more, if it is hard to
foresee and evaluate the skills that may be useful .
Another option would be in dividual salary bargain ing with each employee. This
could be extremely inefficien t, however, because the bargainin g costs co uld be quite
high. Moreover, unless those doin g the bargain in g for the firm are given incen tives
that make them very good represen tatives of the firm's in terests, there may be additional
pro blems. Suppl yin g the pro per incen tives to the bargainers is also problematic,
however. How is the firm to tell if its bargain ing represen tative overpaid an employee
 370
Employment: or if the representative was bargaining too hard and losing people that the firm wanted?
Contracts, How can it even inform the bargainers what behavior it wants from them?
Compensation,
DECENTRALIZED PAY DETERJ\IINATION A further possibility might be to give responsibil­
and Careers
ity to individual managers for pay determination. This alternative is attractive because
local managers are apt to be relatively well informed about their subordinates' abilities
and contributions. It might work well, provided the managers have effective incentives
that cause them to internalize the costs and benefits of their compensation decisions.
Achieving such motivation will be difficult in many circumstances, however,
especially when the manager's performance itself is hard to measure. Unless the
manager has profit-and-loss responsibility, there will be little reason not to be overly
generous with pay. Moreover, doing performance reviews is often unpleasant work,
as is allocating raises. Managers tend to avoid doing these tasks, and when they do
carry out reviews they find it difficult to make sharp distinctions between their
employees and to give low ratings to any but the worst of them. Most employees end
up being evaluated highly, and so the rankings carry little useful information. 7 The
problem is that the individual managers bear the (personal) costs of assigning low
ratings, and it is difficult to compensate them for these costs. In fact, to the extent
that the system by which the managers are evaluated rewards employee development,
the managers may in effect bear extra costs when they grade employees as poor
performers.
Moreover, to the extent that the firm has interests in employment and pay that
go beyond the purview of the ind ividual manager, further problems arise. One such
case arises when the firm wants to be able to transfer employees among jobs with
different supervisors, as is common in internal labor markets. Unless the managers'
contracts lead them to value their employees' acquiring human capital that is valuable
to the firm at large, but not to the specific function the employee is currently doing,
they are not going to encourage employees to make such investments. Furthermore,
unless managers are rewarded when their subordinates are promoted or reassigned,
they may be tempted to hide their best people. For this reason, IBM and some other
large companies evaluate their managers on the number of their employees who are
promoted or hired away by other units in the company.
Influence costs present another reason to avoid discretion in determining pay
levels. If similar jobs across the firm are paid similarly and by largely administrative
decision, there is little room for politicking about pay. If there is substantial discretion
in determining pay, however, influence costs can become very large as employees
lobby for better pay, using comparisons with those in other units. Of course, there
may still be influence exerted to affect job classifications and thereby the pay received.
This will arguably be more limited, however, because the opportunities for effecting
a change are more limited.
PA y R\I\CES The common system for pay determination in internal labor markets
can be seen as a response to these difficulties of specifying and measuring performance
and of giving full discretion on pay determination to local managers . The usual
solution is that pay ranges are established for different jobs, and then the pay awarded
to each individual employee is determined on the basis of the supervisor's evaluation
of the employee's merit. Often the managers of different units are given limited
budgets for annual raises, so the pay increase that can be awarded to any individual
employee is further constrained.
Table 1 1 . 2 illustrates these points using the 1 99 1 Simon and Schuster Salary

• Sec Chapter 12 for a more extensive discussion of performance evaluation.

Table 1 1 . 2 Simon & Schuster, Inc. Internal Labor
1991 Merit Increase Matrix Ma rkets, Job
Percentages Represent Maximum Merit Increase Guideline Assignments, and
Promotions
Location in FY '9 1 Salary Range
PERFORMANCE Below Lower Middle U pper Above
RATING Minimum Third Third Third Maximum

l : Outstanding 8.0-1 2.0% 7. 0- 1 0. 5 % 6. 0-9.0% 5 . 0-7. 0% 4. 0-6. 0%

(l 0% of total)
2: Exceeds requirements 7.0- 1 0. 5% 6. 0-9.0% 5 . 0-7. 5 % 4. 0-6. 0% 0. 0-3 . 0%
/above expectations
(30% of total)
3: Fully competent 6.0-9.0% 5 . 0-7. 5 % 4. 0-6. 0% 0. 0-3 . 0% 0.0%
(45% of total)
4: Needs improvement / 0.0% 0.0% 0.0% 0.0% 0.0%
below expectations
5: Unsatisfactory 0.0% 0. 0% 0.0% 0.0% 0.0%

Increase Guideli nes. Notice that there are target percentages for the numbers of
individuals to be assigned to each performance category. Giving managers a role in
setting individual pay allows them to use the information they have on performance.
Simultaneously, the specifi ed pay ranges attached to jobs and the limited budgets for
raises restrict the managers' discretion and so help co ntrol the possible incentive
problems that giving discretion creates.
At the same time, when the pool of mo ney available for raises is small, the
possibilities for rewarding employees differentially on their relative merit are often
quite limited. Even if the worst performers are given no raises, the funds available
may permit giving only a few dollars more than average to the best people. With so
little effectively at stake, the managers have even less reaso n to take the task of
performance appraisal very seriously and to incur the costs of sharply and accurately
differentiating amo ng emplo yees.
Internal Labor Markets as S y stems
The arguments presented so far to explain the various features of internal labor
markets-long-term emplo yment, limited ports of entry, the use of promotio ns to fill
vacancies, pay attaching largely to jobs and only seco ndarily to individual merit-are
not mutually inconsistent, and in fact many of them involve complementary elements.
Because we have made these arguments in a piecemeal way, however, you might
expect to fi nd great diversity in the practices of actual firms, depending on the severity
of the pro blems i nternal labor markets are intended to resolve.
In reality, firms operating i n a wide variety of differing circumstances have
im plemented internal labor-market systems sharing- many of the characteristic features
we have discussed. This suggests that these features are part of a coherent system
comprised of mutually supporti ng elements to solve a universal problem, and we
should seek to understand them accordingly.
U nfortunately, our understanding of internal labor markets is not yet suffi cient
to allow us to explain them in a full y systematic fashio n, accounting for all their
major features in a unifi ed way. Some fi rst steps in this directio n have been made,
however.
hn,HN.\L LABOH l\t\HKETS, St IIHKI!\:G, .\ND SELF-SELECTION One of the more success­
ful attempts to explain many of the characteristic features of internal labor markets as
 372
Employment:
Contracts,
Compensation,
and Careers
Rat Races and Career Concerns
Many professions seem to be characterized as a "rat race," with people
working much harder and longer hours to keep up with the pack (and maybe
get a little ahead) than apparently makes sense by any rational calculation of
social costs and benefits. For example, lawyers, management consultants,
and university faculty often put in 70- or 80-hour work weeks, sacrificing
their social lives, their families, and even their health for their j obs. The
rewards in these j obs are high, but it seems unlikely that the last hours these
people put in each night yield suffi cient extra output to j ustify the costs.
Similar phenomena seem to arise in the case of students stud ying "too hard"
to gai n admissi on to the prestigious universities in Japan and the grandes
ecoles in France-the students would all be better off if they competed less
intensely, and it is not clear that society would be the worse off if they relaxed
somewhat.
Why d oes this ineffi cient pattern of behavi or persi st? Why doesn't
someone-an empl oyer, for example-put a halt to the rat race? Doing so
would increase effi ciency, and presumably the employer could then capture
some of the resulting gains, perhaps by having to pay less to attract and keep
people.
One possible explanation focuses on information problems. It is rational
for employers to take high levels of past and current performance as an
ind icator that future performance levels will also be high if the performance
depend s partl y on some persistent but unobserved personal characteristic,
which we may call " ability. " Performance in any peri od al so depend s on
effort exerted and on rand om factors, but employers cannot observe whether
an instance of high performance reflects ability, effort, or luck. The result is
an extra incentive to work hard because higher performance today leads to
higher employer estimates of abi lity and thus to higher expectations about
future performance and higher future pay.
N otice that these incentives to work too hard are strongest early in
one's career, for two reasons. First, that is the time of greatest uncertainty
about actual workers' abilities and, therefore, the time when workers have
the greatest opportunity to i nfluence employers' percepti ons. That is also the
time when the length of the remaining career is longest and so when
the benefits of any percei ved increase in ability has its hi ghest value.
Later, when a person's reputation i s stabilized by repeated observations of
performance, it will be harder to alter perceptions of abi lity, and the benefit
of an improved reputation will be shorter lived.
Of course, employers understand these incentives and, in making their
inferences about ability, they should allow for the fact that younger people
will be worki ng exceptio nally hard early in their careers and less hard later.
Even though at an ind ivid ual level, hard work d oes lead to more favorable
evaluati ons and estimates of the empl oyee's ability, the extra effort does not
system atically bias employers' perceptions of employees' abilities. If some
individual dropped out of the rat race, cutting back on effort, this i ndividual
red uced effort level would not be observed. Th en, the lower level of resulting
performance would be incorrectl y interpreted as evidence of low ability, not
 Internal Labor
Markets, Job
Assign ments, and
Promotions
as the result of the person's selecting a lower, and more efficient, level of
work inte nsity. N o one in dividually finds it worth while to stop racin g, even
though the result is socially wasteful.

Based on Bengt Holmstrom, "Managerial Incentive Problems: A Dynamic Perspective , "

in Essays in Economics and Management in Honour of Lars Wahl beck (Helsinki: Swedish School
of Economics , 1 982). See also George Akerlof, "The Economics of Caste and of the Rat Race
and Other Woeful Tales, " Quarterly foumal of Econom ics, 90 (November 1 976), 5 99-6 1 7.

a unified response to a universal problem identifies two key problems faced by the
firm. The first is to motivate employees to supply effort when the actual level of
performance is observed by the firm's managers but is not verifiable by third parties.
The lack of verifiability means that it is infeasible to tie pay to performance directly
via any explicit contract that would be enforceable by the courts. The second pr oblem
arises because individual employees alone know just how difficult it is for them to
achieve any given level of performance. Efficiency requires that those who are more
able should perform at a higher level, and the problem is then to induce them to do
so.
A system with many of the features of internal labor markets can theoretically
solve these two problems simultaneously. S uppose the firm sets up the following
system. I t defines a hierarchy of regular j obs, plus an entry position. Attached to each
j ob is both a pay level and a standard of performance required of those in the j ob,
with higher-level j obs offering more pay and demanding higher performance. U nder
competitive pressure, the wage in any j ob will have to equal the extra value created
by a worker who meets the specified performance standard for the j ob. H owever,
because performance is not verifiable, the firm agrees that i t will pay the specified
wage to any worker holding a particular j ob in any time period, independent of the
employee's actual performance. This arran gement is enforceable by the courts because
whether the person was assigned to a particular j ob and whether he or she was paid
the agreed amoun t are assumed to be verifiable. If the worker does not perform to the
required level, however, he or she is fired (after getting his or her paycheck). Workers
who do perform are kept on for another period.
All hiring is into the entry-level position. Those who perform very poorly in
this position are paid and let go, but those who do better are promoted into a regular
j ob. Generally, an employee in any j ob in the ladder who performs up to a
predetermined standard (higher than that needed to keep the j ob) is pr omoted to the
next level, with its higher pay and higher demands. Even in the period in which the
employee wins a promotion, however, he or she receives only the pay guaranteed in
the current j ob.

EQUI LIBRIUM BEHA \' IOR H ow will employees behave in the system j ust described?
Provided that the requirements of a j ob are not too demanding relative to the
em ployee's abilities and that the pay is sufficiently high, the employee will choose to
work hard en ough to meet the standards set for the j ob. To do otherwise would result
in being fired, leading to a loss of a high-wage j ob. M oreover, meeting any given
sta ndard will be easier for abler people. Therefore, these people may find it worthwhile
 374
Employment:
Contracts,
Compensation,
and Careers
More Rat Races
A problem with the rat-race story as told earlier is that in many situations,
the effort that people put in is at least partly observed. In fact, young lawyers
at many prestigious firms make a point of being seen to be in their offices
night and day. In the previous story these employees should want to hide the
fact that they are working hard, so that their employers will think their
remarkable output is the result of exceptional ability, not hard work. Thus,
they should leave the office ea rly, carrying sports equipment, but with work
that they will do at home hidden in their gym bags!
However, a variation of the model can account for this. Suppose that
what employers are trying to discern is not ability as much as a sense of
responsibility and a willingness to work hard, and that this differs among
individuals but is unobservable. In that case, early in their careers, people
will have an incentive to try to build reputations for putting in long hours
because they will be perceived as willing to work harder than others in the
future as well. Staying in the office longer than would otherwise be optimal
could signal this trait, and the observed pattern would emerge. Again, of
course, if these incentives are recognized, no one is fooled, and yet no
individual can afford to relax for fear of being seen as lazy. Also, these
incentives are greatest early in a person's career, for the same reasons we
discussed in the first variation of this model.

to perform at the higher level needed to get promoted whereas less able people will
not. These are the key elements in generating the equilibrium modes of behavior
under the assumed system.
If the number of rungs in the job ladder, the pay and performance levels attached
to each job, and the promotion standards at each level are all set properly, employees
will be motivated to work for promotions until each reaches a level that is "appropriate"
in light of his or her (privately known) abilities. At this point he or she is willing to
perform well enough to keep the current job but unwilling to work hard enough to
win another promotion: The promise of higher pay but harder work once promoted
is not enough to make earning the promotion worthwhile. Thus, the employees sort
themselves on the basis of their private information, with those who find it easier to
work hard being motivated to do so.
In this theoretical system, the "Peter Principle" is subtly changed from a problem
to be avoided into an engine of sorting and motivation. In this system, people are not
promoted to their levels of incompetency. Instead, people are repeatedly promoted
out of jobs for which they are overqualified until they reach ones where the job
demands are well suited to their individual ability levels.
Moreover, if the system is properly designed there will be no shirking. Those
who are still striving for promotion will exceed the minimum performance standards,
and those who have reached the level appropriate to their abilities will find it
worthwhile to meet the standards required so as not to be fired. With no shirking,
there will also be no one fired, except those who do not make the cut in the try-out,
entry position.
PHOPJ<,HTIES OF Tl IE SYSTEI\IThe system explained by this model mimics a remarkable
nu mber of the features of actual job ladders in internal labor markets. In the model,
 375
an employee's pay depends only on his or her current j ob assignment. E ven tho se Internal Labor
wh o perform at the extraordinary levels needed for promotion are not directly paid Ma rkets, Job
anything more: Their reward is the extra pay they get after being promoted. S alary Assignments, and
and responsibilities increase as an employee moves th rough promotions from one rank Promotions
to another, and abler people end up reaching higher levels of responsibility in the
firm. All hiring is through a single port of entry at the bottom; otherwise, unqualified
people would enter at the high level, shirk, collect their pay, and leave. After an
initial tryout, employees stay with the same employer in what is effectively a long­
term relationship. Firings are rare (they occur only when someone makes a mistake,
gets promoted to a too high-level j ob, and fi nds it is not worthwhile to meet the j ob's
performance standards), as are quits (these would happen only if an earned promotion
were denied, which would be a mistake on the firm's part).
Another interesting feature is that the fi rm receives no rents on employees who
have risen as .far as they are going to go in the company: They are paid the value of
what they produce. In contrast, the firm does earn rents or quasi-rents from the
employees who are still trying for promotions because they are working to a higher
level than are other people in the same j ob but are paid no more. Thus, "rising stars"
will be treasured, and fi rms will compete to identify and attract the most promising
new j ob candidates. (The quasi- rents earned on rising stars may be viewed as a return
to these recruitment investments. )
One feature of many real j ob ladders that is missing here is that actual hierarchies
are often pyramids, narrowing towards the top, whereas nothing in the model ensures
that there will be more people at lower levels than at higher ones. Although it is true
that managerial hierarchies do tend to be pyramids, this is not a universal property of
all j ob ladders, especially when those at higher levels are not responsible for supervising
people below them. For example, higher performance might mean being responsible
for more tasks in a production process, or more unusual tasks, or serving a larger
territory, rather than supervising more people. Then there may well be more people
at the higher levels than at lower ones. For example, there are more people at the
rank of S ecretary IIS on the staff of the S tanford Business S chool than at the S ecretary
II level. The same pattern also holds in many professional partnerships, such as law
firms, where there are usually more partners than associates, and in universities,
where there will often be more full professors than assistant or associate professors.
Within this model, as in many economic models, abili ty is a one-dimensional
quality. S ome people are simply more productive than others. In reality, people
differ on a much wider array of characteristics, such as temperament, stamina,
communication skills, analytical ability, and so on, that make j ob assignments a
complex and subtle task. The next model recognizes some of these complexities.
INFLUENCE COSTS, INCENTI VES, AND JOB ASSIGNMENT
A fundamental conflict can arise when promotions are used both to provide incentives
for performance i n the current j ob and to assign the best-qualified people to key j obs.
As already noted, the best performer in one j ob may not be the person who is best
qualified for a promotion. An obvious solution would seem to be to separate promotions
and performance incentives. We have already suggested some problems with this
approach, especially those of defi ning and measuring performance and of motivating
managers to do the sort of careful evaluations that are often needed to use pay as a
mo tivator. Even putting these aside for the moment, however, this solution may not
be efficient. Influence costs can arise when it is possible for employees to divert
val uable time to affect their apparent qualifications for promotion and when employees'
time allocations are not observable. Handling these costs may lead the fi rm to use
promotions as incentives e ven when it could use performance pay.
 376
Employment:
Contracts,
Compensation ,
and Careers
Pay Differences, Hierarchies, and Tournaments
An important element of the rewards to being pr omoted in a hierarchy is
that you are positioned to try for further promotions to yet higher levels with
even greater status, pay, and perquisites. I n essence, j ob ladders in hierarchies
are elimination tournaments, where the prize to wi nning at one stage is the
immediate reward (higher salary and other direct benefi ts) plus the value of
the option to continue competing.
The higher a person rises within a fi rm, the fewer opportunities remain
for further promotions. Other things being equal, this reduces the incentives
to seek promotion because the sec ond part of the reward to promotion-the
opportunity to keep on advancing and to try for yet further prizes-is
diminished. This means that to maintain the incentives to compete for
advancement, the direct monetary gains to promotion must increase as the
person rises through the hierarchy. Consequently, the difference in pay
associated with pr omotion should increase as one rises in the hierarchy. Also,
once the person becom es CEO, there are no more prizes to win, so the j ump
in pay associated with this fi nal promoti on should be especially large.
A recent empirical study has tested these predictions. * The results
generall y support the theory. The study examined five j ob levels, from plant
manager through corporate CEO, in a sample of over 300 large U . S. firms
in a variety of different industries. The median pay (salary plus bonus) at the
plant-manager level was $21, 000 (in 1967 dollars), at the next level it was
$37,000, at the third level it was $43 , 000, at the fourth it was $60,000, and
at the top, CEO level it was $130,000. The increase associated with a one­
level promotion was thus, on average, $ 1 6,000, $6, 000, $17, 000 , and
$70,000 . This aggregate pattern is broadly consistent with the theory. M ore
significantly, examinati on of the pay differentials between ranks within each
individual company gave very strong support for both predictions: Pay gaps
widen moving up the hierarchy within a firm, and the promotion to CEO
brings an extraordinary j ump in pay relative to other promotions. These tests
are especially significant because they control for differences in the number
of ranks used in different companies and the differences between the sizes of
organizational units across companies.

* Richard A. Lambert, David F. Larcker, and Keith Weigelt, "Tournaments and the
Structure of Organizational Incentives, " draft, The Wharton School, University of Pennsylvania
( 1 989).

A Job-Assignment Problem
To illustrate this problem and to see how responding to it can influence the policies
of the organization, consider a hypothetical situati on in whic h a firm must promote
one of two workers to fill a " key" j ob, which might be a supervisory or leadership
positio n, for exam ple. Its defi ning property is that if the person holding the key j ob
were to q uit, the fi rm wo uld suffer an extraordinary cost in replacing the employee
or in having the j ob tem poraril y empty. B oth candidates for promotion are currentl y
assigned to a job in which their individual efforts are productive in th at more effort
 :n1
increases the probability of a good outcome (say, making a big sale or achievin g a Internal Labor
breakth rough in a research project). There is no difference between the two workers Markets, Job
in their abilities to produce successfully in the base-line job, but they may differ in Assignments, and
their abilities to perform in the key job. Although one candidate might be better Promotions
qualified for the key j ob (in the sense that total profits will be higher if that person is
promoted), the firm does not know initially which candidate that is.
The employees can spend time on the job in two ways. They may each devote
time to improving the chances of a successful outcome in their current assignments,
or they may devote time to developing evidence about their qualifications for the key
job. The workers, however, have no inherent like or dislike for either activity. With out
incentives to do otherwise, they are willing to allocate their efforts in any way that
the firm's management may direct.
Two observability problems complicate the situation. First, the attractiveness of
any competing offer that the person who is ultimately promoted may receive is not
observable by the firm. Second, the firm cannot observe the workers' choices in
allocating their current time and effort, nor can it tell if a worker is suppressing
unfavorable information he or she may have obtained about his or her qualifications
for promotion. U nder these circumstances, a worker who increases the share of time
on the job spent developing information about his or her qualifications and then hides
any unfavorable results will likely produce stronger "credentials. " He or she is likely
to appear better qualified for the promotion than otherwise would have been the case.
However, the diversion of attention to building credentials reduces the probability of
the worker achieving a good outcome in the current j ob.

THE IMPACT OF LIMITED OBSERVABILITY Without these observability difficulties, the

firm's optimal policy would be simple. It would trade off the value of increasing the qual­
ity of information about the candidates' qualifications for the key job against the cost
of reducing the expected output in the current j obs, and then it would direct the
workers to allocate their time between production and information development
accordingly. The workers would be paid a fixed wage for their time in th e current
job, so that pay is unrelated to performance, and the firm would promote the one
who appeared better qualified. The pay that the workers subsequently receive would
be independent of their job assignments unless the one assigned to the key job gets
an outside offer, in which case the firm would match this offer as long as that is
cheaper than losing the worker.
The firm's inability to evaluate outside offers makes the offer-matching strategy
infeasible. We saw in Chapter 1 0 that one solution is to pay a premium wage to the
worker who is promoted, as a kind of insurance against losing him or her and suffering
the resultant costs. The result of this would be to make the key job more attractive
than the other one because the promoted worker would not have to get a good outside
offer to earn more.
If the workers' allocations of effort could be perfectly observed, the pay differential
associated with being promoted would not cause any further problems. Even if the
workers can hide unfavorable information, the firm may be able to allow for this in
evaluating their credentials and still make the right promotion decisions. If the
allocation of effort cannot be directly measured, however, there is a moral hazard
problem. Without any direct incentive to devote effort to production, workers would
find it in their interest to spend all their time developing their credentials, trying to
win promotion to the better-paying job. The loss of current output occasioned by this
diversion of attention could be very costly to the firm. These costs are influence costs,
as the workers use resources to attempt to influence which of them will gain the rents
that result from being promoted.
 Organizationa l Responses
378
Employment:
Contracts, There are a number of ways that the firm can respond to the incentive difficulties
Compensation ,
that it creates when it pays more to the promoted worker. It can reduce the difference
and Careers
in pay between the key job and the other. It can introduce performance pay, paying
more if the worker's output is high than if it is low. It can reduce the importance of
credentials in the promotion decision by promoting on other criteria, such as seniority,
perhaps even not gathering the information that the workers develop regarding their
qualifications. More directly, it can make performance a criterion in the promotion
decision, even though performance in the current job carries no information about
qualifications for the key job.
OPTl !\IAL PAY AND PR0J\I0TI0N POLICIES Just which mix of these is best depends in
part on the importance of promoting the worker who actually is better qualified. At
one extreme, if the profits of the firm crucially depend on whether the better-qualified
person is assigned to the key job, it is very important that some information on
qualifications be generated and that this information be used to identify and promote
the better candidate. In this case, unless the firm wants the workers simply to ignore
their current jobs and to spend all their time generating information, some sort of
incentives must be provided to get the workers to pay attention to their current
responsibilities.
Because the gap in pay between the two jobs creates the incentives for workers
to misallocate their time, an obvious possibility would be to narrow this gap. This
alone will not help, however, unless the gap is completely closed. As long as there is
no reward for current performance, the existence of any gap means that the incentives
are to ignore productive activity entirely because it is costless to the workers to do so.
Thus, some sort of performance incentives are needed, even though employees are
(by assumption) not averse to working hard. The appropriate way to provide these in
this case is by paying a bonus for success in the current job. Promoting in part on
the basis of current performance would also provide the desired incentives, but the
extreme importance of promoting the better-qualified candidate makes this alternative
unattractive. Once performance incentives are in place, however, it is also useful to
reduce the size of the pay differential associated with promotion, thereby reducing
the amount of rents and the incentives to try to influence their ultimate distribution.
At the opposite extreme, if it really does not matter which one is promoted
because the output in the key job is largely insensitive to the qualifications of the
person holding it, then the solution is simply to promote the one who performed
better in the current job. (If their performance is indistinguishable, then some other
criteria such as seniority might be used, or a random choice could be made: The
important point is that their credentials should not be a factor in the decision. ) No
direct performance pay is needed, nor is there any need to close the pay gap between
the worker who is promoted and the one who is· passed over. This promotion policy
will induce the workers to spend all their time on productive activities, as is optimal.
Finally, when making the right assignment to the key job is of middling
importance, it may be worthwhile to provide performance incentives by a mix of
methods involving performance pay and the promotion criteria. Promoting on the
basis of a mix of performance and qualifications allows production incentives to be
maintained without having to offer high levels of direct performance pay. Thus,
promotions may be made on criteria that have nothing to do with ability or likely
success in the key job. At the same time, it can again be valuable to reduce the pay
differential associated with promotion.
More gen erally, to the extent that the firm limits the linkage between promotions
and apparent individ ual performance or qualifications, influence costs may be reduced.
Th is possibil ity gives a reason to apply a sen iority rule that only the most se nior In ternal Lahar
workers at a given level will be considered for promotion. Such a rule limits the se t Ma rkets, Job
of people wh o arc effective ly com peting for a promotion at any given time and so Assignments, and
may reduce the total resources wasted in infl ue nce activities. Promotions

Col\ll\I ITl\lENT PnoBLEl\1S A policy of promoting on seniority, performance, or other

criteria other than apparent qualifications may run into a commitment problem. If
the firm has the information that has been generated about the worker's qual ifications,
it will sometimes believe that it is promoting the less-qualified person when it promotes
on past performance. Th is would be especially galling because (at least in the current
model), past performance has nothing to do with future productivity in the key j ob. 8
Because there is no further need to provide incentives (at least in the simplified
framework considered here), there will be a temptation to renege and promote the
better-qualified person. The alternative is to throw away future profits by adhering to
the promise fo promote on performance.
If the basis of performance measurement is reasonably obj ective, so that other
workers can observe the firm reneging on its policy, then this problem may not be
too serious. Concern for its reputation will prevent management from violating its
commitments in a wholesale way. When performance measurement is subj ective,
however, the temptation to ignore performance and to promote on the basis of
perceived qualifications for the j ob may be irresistible.
Of course, if the workers believe that the firm will not adhere to its announced
promotion policy but will instead promote the person with the better credentials, then
the announced promotion scheme will be irrelevant to their decision making. In such
circumstances, the firm might prefer not to allow employees to present their credentials,
even if they are best positioned to do so and have spent resources creating a polished
presentation. By refusing to allow such presentations, they avoid the temptation to
cheat on thei r promises and hel p provide proper incentives to employees.

TENURE AND UP-OR-OUT RULES

Two aspects of the promotion policies of many organizations present especially
interesting puzzles: tenure and up- or- out rules in promotions. S omeone with formal
tenure cannot be fired except for serious cause, and then usually only through a
demanding process. An up-or-out rule means that those denied promotion must leave
the organ ization; they cannot stay on in their original j obs, even if they were doing
quite well.
Often these two practices are linked. Tenure is most familiar in universities and
colleges, which also have an up- or- out rule that a faculty member who is denied
tenure must leave. H owever, these practices are used in many professional partnerships
in law, accounting, and other fields as well. Professional employees are reviewed for
partnership at a definite point in their careers, and those who are not admitted to
partnership are let go. Meanwhile, attaining partnership has usually meant that the
person effectively cannot be fired. 9 The practices are not al ways linked, however. For

8 In fact, given the incentives in the model and the behavior they would induce, which worker had
the higher performance is purely a matter of luck. Both will devote the same time to productive activity,
so their expected outputs will be the same. The real ized differences in their outputs will reflect only random
noise.
9 In early 1 99 1 , however, the large i nternational public accounting firm of Peat Marwick announced
that it was planning to lay off 1 4 percent of its approximately 1 , 800 partners, and a number of la rge law
firms were reported to have similar plans. These were newsworthy events that caused great discussion and,
no doubt, real consternation. Apparently the partners' tenure was not as secure as outside observers (and
the partners) had believed.
 380
Employment: example, tenure (in the form of partnership) is used without an up-or-out rule at the
Contracts, investment banking firm of Goldman Sachs, where those who are unsuccessful in mak­
Compensation, ing partner in a particular year are often encouraged to stay on and try again. In
and Careers contrast, the U. S. armed forces have a repeated up-or-out rule for officers-those in
each cohort at each rank who fail to win promotion to the next rank must retire­
but no tenure system.
The puzzle regarding up-or-out rules is this: Why would the firm find it
worthwhile to keep someone employed for several years, often at a handsome pay rate
and with major job responsibilities, and then the next day refuse to keep them on for
any kind of professional work at any compensation level? One possibility is that these
lower-level positions are valued for more than just the output that their occupants
produce. They provide a training ground for new workers, an opportunity to observe
and evaluate employees' capabilities, and a way to bring fresh perspectives and ideas
into the firm, all of which are important for maintaining the firm's vitality. 1 0 However,
the very importance of turnover among employees throws into even sharper focus the
puzzle associated with tenure: Why does the organization commit itself to keeping on
senior employees whose abilities and performance may be much lower and ideas
much staler than those of their potential replacements? There is no fully satisfactory,
unified explanation available for these practices, but separate, partial explanations
based on informational asymmetries and commitment problems have been advanced.

Tenure
The nominal reason for academic tenure is to protect academic freedom: Job security
is supposed to make professors less afraid to espouse unpopular positions. More
generally, tenure may make professors willing to take risks on new and daring research
directions that are likely to fail but might lead to great breakthroughs. This explanation
does not seem particularly applicable to professional partnerships, however. In any
case, the rigidities and inefficiencies of the tenure system seem to make it an expensive
and clumsy way to protect free speech or encourage risk taking.

As ''EXPERTISE"' EXPLANATION FOR ACADEl\lIC TENL'RE One analysis that explains the
practice of tenure is based on the idea that only the current faculty in a university
department has the expertise and information sources to evaluate potential new hires,
although management (the administration) is able to recognize the quality and measure
the productivity of people once they are in the organization. Given some constraint
on the number of faculty that can be employed in the institution, the incumbent
faculty's fear is that the university will replace them with younger, better qualified,
more productive people. The only way that this could happen, however, is if such
people are identified by the incumbents and then brought into the system, whereupon
the administration can judge their quality relative to their older colleagues. This
means that unless the incumbents are unconcerned that identifying good people will
endanger their own welfare, they will have an incentive to misrepresent the
qualifications of candidates, attempting to ensure that tough competitors are weeded
out and that only weak candidates who are no threat to them are allowed to enter the
system.
One potential way to remove the incumbents' fear of being replaced would be

J O This argument and the evidence for it is developed in hrn research papers by Brendan O'Flaherty
and Aloysius Siow, "Up or Out Rules in the Market for Lawyers," Discussion Paper 90- 1 0, Economics
Research Center, National Opinion Research Center, Chicago (September 1990), and "On the Job
Screening, Up or Out Rules, and Firm Growth, " Discussion Paper 90- 1 1, Economics Research Center,
National Opinion Research Center, Chicago, (September 1990).
 :381
to give en ough severance pay to senior people who are le t go that they arc in differen t Internal Labor
be tween keepin g their j obs and losing them. De termining the necessary amount of Markets, Joh
compensation may be extremely difficult, however, and the bargaining costs that Assignments, and
would be incurred make this approach very problematic. Promotions
A tenure system presents a simple alternative solu tion. People who have met
the tenure criteria cannot be fired simply because their talents or performance levels
have deteriorated or their particular area of expertise within the field is no longer in
demand. Thus, they have no reason to fear the hiring of people who are better
qualified than they are and no reason to misrepresent the quality of candidates.
This explanation is consistent with other aspects of universities' employment
and governance procedures. Tenured professors can in fact be fired if their performance
is bad enough, but the standard that must be met is painfully low and unresponsive
to changing conditions. This is necessary to prevent firing for cause from becoming
a masquerad� for replacing less qualified professors, which would destroy incentives.
Moreover, when financial pressures necessitate a reduction in faculty size beyond
what can be accommodated by releasing untenured faculty, a common practice is to
eliminate whole departments or schools, rather than individual tenured faculty from
across several units. To do otherwise and release people based on their abilities would
encourage the senior faculty to stock their departments with mediocrities. In contrast,
closing whole units encourages the faculty to make their departments as strong as
possible to protect themselves against wholesale dismissal.
Applicability to Other Contexts. The basic idea that tenure is a system that
encourages professionals not to misrepresent their evaluations of new j ob candidates
out of the fear of losing their j obs to the newcomers also has applicability in law,
accounting, and other professions where partnerships are common. S urely the tax or
bankruptcy attorneys in a law firm are especially well positioned to evaluate potential
hires in their areas of expertise, and the firm must thus rely largely on their appraisals
of candidates. To the extent that, without tenure, the senior people woul d be
concerned about the possibility of their colleagues replacing them with superior
younger people, they would have an incentive to screen out the best candidates. This
danger would seem to be especially severe in larger firms and in firms that share
profits on a seniority basis. In these situations, the costs of having less able people are
spread widely across the organization. In smaller firms or those that divide profits
according to departmental performance, the costs of having mediocre colleagues fall
more directly on those responsible. Thus, their incentives to hire weak people are
lessened.
A LEARNING EXPLANATION A n alternative explanation for tenure applies more broadly
and seems to have more applicability to situations where acquired professional expertise
is not a maj or factor. Suppose for simplicity there are j ust two basic types of j obs, one
in which productivity is largely insensitive to innate ability and another in which
ability or talent plays a key role in determining success. Bureaucratic offi ce work or
many j obs in the secondary sector might be examples of the former, and executive
positions or artistic work could be examples of the latter type of employment. Effi ciency
requires that low-ability people should be assigned to the j obs where ability is
unimportant and high-ability people should be placed in the ability-sensitive j obs
where they are more productive. S uppose too that peoples' ability levels are not directl y
observed by anyone, including the individuals themselves. Then an effi cient all ocation
of individuals to j obs is not immediately achievable.
If an individual tries the sensitive j ob, his or her realized output in any period
generates a signal about his or her ability. The more successful the individual is in
any period, the more likely that he or she is actually well matched in the j ob, and
 382
Employment: the more periods in which he or she is successful, the more certain it becomes that
Contracts, the ability-sensitive job is the right one for this person. Eventually, after enough
Compensation , experience, the quality of the match may be almost completely learned, and the
and Careers person will either move to the other type of job or become permanently matched in
the ability-sensitive job. In this case, the person effectively has tenure in the ability­
sensitive job. Even if he or she performs very poorly in the future, the estimate of
ability will still remain sufficiently high that his or her expected productivity is still
higher in the sensitive job.
The same idea can be extended to multiple jobs with differing sensitivities to
different kinds of talent and to there being some initial information about abilities.
The argument would also work if the match between different jobs and personal tastes
differed. The pattern that will emerge is some early experimentation, followed by an
eventual settling down with effective tenure. This pattern is in fact quite common in
actual careers.

Up-Or- Out Rules

One explanation for up-or-out rules starts from the observation that tenure (or
partnership) is usually associated with a quantum jump in earnings (monetary and
otherwise), often with no major increase in responsibilities, except for a greater role
in governance. Assistant professors have essentially the same job responsibilities as do
full professors, and senior associates in law firms do much the same work as partners.
Yet the junior people are typically paid much less than senior ones. The average
assistant professor makes about half as much as the average full professor in major
research universities, and the ratio of associates' pay to partners' earnings in law firms
is probably even lower. Apparently, then, the organization collects rents on the junior
people. Their reason for staying on, despite being paid less than they are contributing,
is the chance to get promoted and then receive their share of the organization's profits
and the rents to be earned on later junior people.
In this context, promotions have little effect on the output of the organization,
but they mean that the rents and profits must be shared among more claimants. The
obvious temptation would be to tum down people who come up for partner or tenure,
so that the available rents do not have to be shared with extra people, but to offer
them a chance to stay on at competitive wages. Even more attractive would be to try
to get the unsuccessful candidates to stay on at their substandard pay with the promise
that they will be reconsidered in the near future.
Of course, junior people who recognize these incenti ves would not play this
game. They would not work hard in hopes of promotion; they would not invest in
firm-specific human capital whose value would not be reflected in their pay but only
add to the rents that the tenured people collect; and they would not work for less than
their market wage.
The up-or-out policy imposes costs on the firm when it turns someone down
for promotion: It loses the returns that the person brings, and the only way to capture
any of the returns is to award tenure or partnership. The firm will then not be tempted
to deny people promotions in hopes of exploiting them, and so the incentives that
the chance of gaining tenure provides can be effective.
The basic idea that up-or-out rules are a means of imposing costs on the decision
makers yields a complementary explanation: Tenure coupled with up-or-out rules
forces those doing evaluations to take them seriously because the decisions are made
much more important when these practices are in place. This helps overcome the
reluctance to bear the costs of doing careful reviews of subordinates that we noted
earlier.
 Internal Labor

SUMMARY
Markets, Job
Assignments, and
Promotions
A majority of the workers in developed countries becomes attached to a single employer
in a long-term relationship that commonly extends 20 years or more. Although the
classical theory of employment arrangements continuously buffeted by changing labor
market conditions provides a reasonable description of short-term employment in the
secondary sector of advanced economies, the long-term jobs in the primary sector are
insulated to a considerable degree from conditions in the external labor market.
Instead, an internal labor market develops to assign people to appropriate jobs within
the firm and to determine their pay.
One of the distinctive features of traditional internal labor markets is that, to a
considerable .degree, wages are determined by job classifications rather than by
individual qualifications. Table I l . I illustrates this proposition, but it also illustrates
that there is still latitude for individual wage differentials within jobs. More experienced
workers and workers with unusually high performance ratings tend to move toward
the tops of the categories, whereas less experienced workers tend to be nearer the
lower end. As long as job requirements and tasks remain unchanged over fairly long
periods of time, this system can be effective, but with changing technologies and
products firms may wish to encourage their workers to become flexible, paying wages
based on skills rather than on the job actually being done. This system is quite
prevalent in Japan and is found increasingly in other parts of the world as well.
Given the advantages associated with worker mobility and the relatively high
wages paid in the primary-sector jobs, there must be significant advantages to long­
term employment to offset the loss of mobility that it entails. Among these advantages
are the acquisition of firm-specific human capital, which enables a worker in the
primary sector to become especially productive in his or her job; the use of efficiency
wages and of implicit, relational contracts, in which the promises of loyalty and fair
treatment are enforced by internal mechanisms within the firm and by the desire of
each party to maintain a good reputation in its long-term relationship; and informational
advantages, especially the superior knowledge that an employer has about its employees'
abilities that permits giving more effective incentives and making more productive
assignments of workers to jobs.
Jobs within internal labor markets are arranged in hierarchies with the result
that employees doing similar work may sometimes have higher or lower rankings.
The higher-ranking employees may be more experienced and therefore better able to
handle exceptional conditions, to train or advise others about what actions are required,
and so on. Extra duties, extra responsibilities, and greater reliance of the organization
on the higher-ranking employees' judgments help account for the higher pay and the
tendency to assign these jobs to more talented, experienced people. In these kinds of
jobs, the use of promotions to reward productivity is relatively unproblematic.
In other jobs, the use of promotions for rewards are a much more nettlesome
problem because promotion involves a qualitative change in responsibilities. There is

will also be the best supervisor and teacher, or that the best researcher in an R&D
no guarantee, nor even a great likelihood, that the best or most experienced machinist

laboratory will make the best R&D manager. It may be most productive to retain the
machinist and researcher in their original jobs and promote someone whose perfor­
mance was less outstanding.
Why, then, not use cash bonuses to reward individual good performance rather
than promotions? Promotions from within have the advantage that they resemble
tournaments, which have three important advantages for providing incentives. First,
 384
Employment: tournaments require only comparative, ordinal information about who did better
Contracts, rather than the much costlier cardinal information about how much better a party
Compensation, has performed than some absolute standard. Second, even when the quality of each
and Careers performance is separately and objectively measurable, relative performance evaluation
may be a better basis for compensation if common factors affect the performance of
all the participants. Third, because the bonus pool in a tournament is set in advance,
the employer has no incentive to disparage or misrepresent the workers' performance
in order to save having to pay performance bonuses: A source of moral hazard is
eliminated. In addition to these advantages, attaching pay to jobs and using promotions
for incentives mitigates the need to bargain with individual employees over their
salaries. Such bargaining is a time-consuming, disruptive, and personally costly
process, incurring high infiuence costs.
One model that explains many of the features of internal labor markets in a
unified fashion emphasizes that workers vary in their abilities, that their efforts are
subject to moral hazard, and that their performance can be judged only by insiders
and so cannot be a basis for contracts enforceable in court. The firm creates an
internal labor market with entry only at the bottom of the hierarchy, pay in each rank
that is insensitive to performance, higher pay and performance requirements being
associated with higher-level jobs, and higher levels of performance being required to
be promoted than merely to avoid being fired. This leads workers both to select levels
of effort that are at nearly efficient levels and to sort themselves among jobs so that
the abler ones are eventually promoted to the jobs best suited to their abilities.
When promotions are based partly on forecasts of future performance rather
than on past performance alone, candidates acquire an incentive to build credentials
for good jobs and to manipulate the decision makers' perceptions, instead of simply
being productive. These activities are more examples of influence costs, and they can
degrade the performance of the organization. Organizations can reduce these costs in
various ways, including the introduction of pay for performance to create an offsetting
incentive, increasing the weight given to past performance in promotion decisions,
and limiting the input that employees have to the decision process. A decision to
establish any policy other than promoting the best qualified person, however, creates
a problem of comm itment-management will not want to carry out the policy when
the time comes to do so. If the basis of the decision is observable, then concern for
its reputation may lead management to adhere to its policy, but the problem becomes
a serious one when the proposed alternative basis of evaluation is highly subjective.
The use of up-or-out rules in promotions and the institution of permanent job
tenure, both of which are common in many professional hierarchies, including law
firms, accounting firms, and universities, are puzzles requiring explanation. Junior
accountants and associates at law firms and assistant professors at universities work
under contracts with limited job security, whereas the partners (at law and accounting
firms) or tenured professors (at universities) have an effective employment guarantee.
At some point, the young workers must either be promoted and given permanent job
security or face termination of their employment. These up-or-out and tenure rules
appear to serve a variety of functions which have been the basis for a number of
theories. One possibility is that the low-level jobs are continually turned over to allow
the organization to import fresh ideas and perspectives from outsiders or to allow the
organization to evaluate closely potential candidates for higher-ranking positions. To
the extent that only professionals are able to evaluate one another's performance,
tenure rules might be designed to encourage professionals to identify the best
candidates-candidates who are so good that their appointment might otherwise
threaten the recommender's own job. Up-or-out rules force the organization to
promote candidates or lose their services. Because junior people in organizations
 385
employing these rules arc a source of rents to the organization, the temptation would Internal Labor
be to try to hold them in the lower-level jobs, but this would destroy incentives and Markets, Job
make recruiting difficult. Assignments, and
Promotions

• BIBLIOGRAPHIC NOTES
Peter Doeringer and Michael Piore's landmark 1 97 1 study, Internal Labor
Ma rkets and Manpower Analysis, made clear the usefulness of this concept in
understanding employment policies. It remains a basic reference. Richard Edwards
suggested the subdivision of the primary sector discussed in the text. Mace Mesters'
dissertation explores theoretically the emergence of primary and secondary sectors
and the allocation of workers between sectors based on ability. Masahiko Aoki's
book (especially Chapter 3) discusses Japanese internal labor market and human
resource .management systems.
The analysis of tournaments as incentive schemes was begun by Edward Lazear
and Sherwin Rosen. Bentley MacLeod and James Malcomson developed the
model of an internal labor market as a solution to the problems of moral hazard
in effort provision and of inducing self-selection on the basis of ability. The
treatment of the job-assignment problem in the presence of influence costs is
drawn from our own research.
The "expertise in judging applicants" explanation of tenure originated with H.
Lorne Carmichael, and the "learning" explanation is based on the work of Milton
Harris and Yoram Weiss. The rationale for up-or-out rules is due to Ronald
Gilson and Robert Mnookin.

• REFERENCES
Aoki, M. Information, Incentives and Ba rgaining in the /apa nese Economy
(Cambridge: Cambridge University Press, 1 988).
Carmichael, H. L. "Incentives in Academics: Why is There Tenure?" /ou rna l of
Political Economy, 96 ( 1988), 45 3-72.
Doeringer, P. , and M. Piore. Internal Labor Markets and Ma npower Analysis
(Lexington, MA: D.C. Heath, 1 97 1 ).
Edwards, R. Contested Terrain: The Transformation of the Work Place in the
Twentieth Century (New York: Basic Books, 1 979).
Harris, M. , and Y. Weiss. "Job Matching with a Finite Horizon and Risk
Aversion," /ou rnal of Political Economy, 92 (August 1 984), 758-79.
Gilson, R. , and R. Mnookin. "Coming of Age in a Corporate Law Firm: The
Economics of Associate Career Patterns, " Stanford Law Review, 4 1 ( 1 989), 567-
95.
Lazear, E . , and S. Rosen. "Rank Order Tournaments as Optimal Labor
Contracts, " /ou rnal of Political Economy, 89 (October 1 98 1 ), 84 1 -64.
MacLeod, W. B . , and J. Malcomson. "Reputation and Hierarchy in Dynamic
Models of Employment, " /ournal of Political Economy, 96 (August, 1 988), 8 3 2-
54.
Mesters, M. Technology and Ma nagement of Human Capital, unpublished
dissertation, Stanford University Graduate School of Business ( 1 990).
Milgrom, P., and J. Roberts. "An Economic Approach to I nfluence Activities in
Organizations," America n /ournal of Sociology 94 (Supplement), 1 988a: S 1 54-
S l 79.
 386
Employment: EXERCISES
Contracts,
Compensation, Food for Thought
and Careers

1. Suppose you were put in charge of promotion and pay determination for
the faculty members in your school or department. What criteria would you use in
making your decisions? What information would you want? How would you try to
get it? Would you eliminate tenure?
2. Most "permanent" employees in large Japanese firms face mandatory.
retirement at about age 5 8 (the most senior executives are exempt). However, social
security benefits from the state do not begin until age 6 5 . Contractual, employer­
provided pensions were rare until recently, but companies typically made substantial
cash grants to retiring workers. Retired workers often seek jobs in smaller, less
prestigious firms at lower rates of pay. These firms are sometimes ones with which
the original employer has a business relationship, in which case the employer might
help the retiring employee find the job. What might be the rationale for these
practices? Why not let employees stay with their original employers until they reach
65? Why subject them to the uncertainty about how much retirement pay they will
receive? Why force them to look for another job?
3. David Jacobs 1 1 has identified two sorts of jobs depending on the relationship
between individual and organizational performance. In the first sort of job, individual
failures are extremely costly for the firm, but above-average performance has little
marginal impact on overall performance. Examples are being an airline pilot or an
inspector in a nuclear power facility: A mistake is disastrous, but doing an especially
good job has little impact. In the other sort of job, organizational performance is
largely a matter of exceptional individual performance, with individual failures being
of far less consequence. Examples are jobs in sales and scientific research. What
differences would you expect between the human resource policies designed to deal
with the two sorts of jobs? Why?
4. In the late 1 980s and early 1 990s, the airline industry in the United States
was going through a major "shake-out. " A number of lines went bankrupt, and several
of these either ceased operations entirely or reduced their scale considerably, selling
some of their aircraft, routes, and ground facilities to other lines. Meanwhile, the
successful lines were expanding rapidly: American Airlines, for example, was hiring
almost a thousand pilots a year. Many of these were experienced airline pilots who
had lost their jobs at other carriers.
Airline pilots move up through a job ladder that involves three ranks-flight
engineer, copilot, and pilot-with increasing responsibilities and that also differentiates
by the type of aircraft involved, with larger, more complex airplanes ranking higher.
However, the skills and knowledge required to fill a particular role in a particular
aircraft are essentially the same across airlines: Being a copilot on a particular type of
Boeing 727 with United Airlines involves the same human capital as it does with
Delta Airlines.
The airlines used a stnct system of hiring only at the bottom of the job ladder.
They maintained this policy, which the union supported, even when hiring very
experienced pilots from other carriers: a senior pilot who had been flying the most
advanced, wide-bodied aircraft on international routes would be required to start over

11 David Jacobs, "Toward a Theory of Mobility and Behavior in Organizations: An Inquiry into the
Consequences of Some Relationships Between Individual Performance and Organizational Success, "
American Journal of Sociology, 87 (1981), 684-707.
 ;337
a t the bottom if he or sh e chan ged airlin es, despi te all the n on-specific human capital Internal Labor
he or she had accumulated. Give n th is, a num ber of pilots instead left the industry, Markets, Job
fin ding other work. Thi� pol icy seems very wasteful. H ow would you account for it? Assignments, and
5. When la bor market su pply is tight an d it is hard to find qualified workers, Promotions
firms may be tempted to lower the standards they normally use in acceptin g applican ts
wh o would n ot be hired under n ormal circumstances. This is especially problematic
for firms with in ternal labor markets and more- or-less perman ent employment policies.
What measures would you expect such firms to take to minimize the costs of hirin g
less-qualified people?
6. Empirically, the probabi lity of a worker being fired from or quitting a
particular j ob in any unit of time is a decreasing function of the length of time he or
she has already been in the j ob. H ow would you account for this?
7. A number of management consultants in the late 1 980s argued that
companies sh oul d not have any j obs from which an employee, simply by being in
them, could not rise to become chief executive officer: If there were j obs and activities
whose requirements and focus did not connect this closely to the firms' main business,
they should be placed outside the firm and the corresponding goods or services
purchased from outside suppliers. What merits do you see in this? What are the
difficulties?
8. In a number of technology- based companies (including, for example,
Hewlett-Packard), employees move frequently across j obs and even across functions.
There are many "lateral" moves that do not involve a promotion in the average
employee's career path. What advantages do you see to this? What problems does it
create? H ow might these be �itigated?
 12
COMPENSATION AND MOTIVATION

IAI thorough understanding of internal incentives i s critical to developing a

viable theory of the firm, since they largely determine how individuals behave in
organizations.
George Baker, Michael fensen, and Kevin Murphy 1

Compensation has many roles in the organization. One of the most important
is to provide proper and effective motivation. This chapter focuses on compensation
as a means of motivation, building on the basic theory of incentive pay developed in
Chapters 7 and 8.

THE FORMS AND FUNCTIONS OF COMPENSATION

Differing Forms of Pay

Within an organization, compensation can take many forms and can depend on the
hours of work, the performance achieved, or other indicators. Production workers are
usually paid an hourly wage or, less commonly, a monthly salary. Alternatively, they
may be paid a fixed amount, or piece rate, for each unit they produce. Office workers
usually receive a monthly or semimonthly salary. If they are paid extra for overtime,
it will typically be at a higher rate than their implicit hourly wage. Salespeople may
receive a straight salary or a commission based on the number of units sold, the
amount or revenue generated, or the profit generated by the sales; or some combination
of these.

1 "Compensation and Incentives: Practice vs. Theory, " /oumal of Finance , 43 ( 1 988), 593.
 389
Employees in each of these categories may receive bo nus payments in additio n Compensation and
to their regular pay. These bonuses might be i mplicitly or explicitly tied to performance Motivation
at vari ous levels-individual, group, plant, division, or the organization as a whole­
and, i n each case, performance may be defined i n different ways. For determining
an i ndividual bonus, performance might be measured in a relatively obj ective fashion
based on skills acquired, individual production, or revenue or profit generati on, or
instead it may be evaluated subj ectively on the basis of a supervisor's determination
of the employee's merit and overall contribution. B onuses for a work group or plant
might be based on productivity improvements, reaching quality or output goals, or
meeting budgets. For a research an d development team, bonuses might accrue for
fili ng new patents or developing new products, or they might be determined as a
function of the group's direct contribution to measured profitability. At the level of
the business unit or division, bonuses might be based on meetin g plans for sales or
productivity growth, or budgets for costs and profits, or targets for return on investments.
At the level of the organization as a whole, profit sharing may be the basis fo r extra
compensation.
Rewards for good performance need not always take the form of current
compensation. It is common to reward salaried employees by giving merit raises that
increase base salary in all future periods. In this way, the rewards for current
performance are spread over the remainder of the employee's career with the firm.
M ore directly, life insurance agents may receive commission payments known as
"renewals" that continue after the first year that a policy is in fo rce, even if they have
no further contact with the client and sometimes even if the agent no longer works
for the insurance company. M any organizations also make contributions to pension
funds on their employees' behalf. These are all essentially deferred cash compensation,
with the extra feature that their receipt may be conditional on the employees' remaining
with the firm for a specific peri od of time.
Other examples do not involve direct monetary compensation, but they increase
the employees' buying power and so functi on like pay. For example, many firms give
their employees di scounts on buyi ng the employers' products or services, and
corporations often subsidize their employees' purchase of the firm's common stock.
These firms may also pay some part of total compensation in stock directly, often
putting restrictions on the employees' sales of these shares. Employee S tock Ownership
Plans are an increasingly popular vehicle for this in the U nited States, where they are
replacing contributions to regular pension funds to some extent.
Employers also provide a variety of fringe benefits that involve an element of
noncash compensation: health, dental, and medical insurance, disability and life
insurance, company cars that might substitute for employees' privately purchased
vehicles, computers to use at home, health club memberships, and so on. They also
may repay various expenses incurred by the employees, even though some (or all) of
the benefits of these expenditures may have accrued to the employees. Examples
include reimbursing educational expenses for employees (and, at many private
universities, for their children), generously defined moving expense allowances,
payment of professional association dues, license fees, and the costs of business
publications, and the provision of liberal, loosely monitored expense accounts are
examples.
Further variety and richness are found in executive compensation. Besides
receiving salaries, bonuses, and all the multi tude of i ndirect forms of compensation
that are provided to lower-level employees, executives may also have access to special
dining rooms and washrooms, company airplanes and apartments, and membership
in exclusi ve clubs and associati ons. They are especially likely to receive deferred cash
compensation, they may be awarded large amounts of company stock, and they
 390
Employment: frequently receive stock options-rights to buy the company's shares at a predetermined
Contracts, price, which may often turn out to be less than the market price of the stock.
Compensation, Executives being recruited by the firm may receive large signing bonuses that are
and Careers rarely offered to regular employees. If they lose or leave their jobs after a merger,
takeover, or other change in corporate control, they may receive large guaranteed
severance payments (golden parachu tes). Even after retirement they may be paid to
work as consultants to the company, or they may be kept on the board of directors,
receiving handsome fees for their time.
The complexity of individual compensation is compounded when we examine
how pay changes over time and how it varies within and across organizations.
Explaining this complexity and variety is a serious challenge to social scientists.
Certainly, the classical economic model of determining wages and hours worked by
supply and demand is inadequate to the task. For managers, the problem of selecting
pay policies and forms and of determining individual compensation is both important
and urgent. Their own personal success and the success of their organizations depend
crucially on these choices. In this chapter, we seek an economic explanation of the
observed diversity and complexity, in part in the hope of providing a better basis for
managerial decisions.

The Objectives of Com pensation Policy

A large part of the explanation lies in recognizing that compensation policy has a
variety of roles to play in an organization and that, even holding the total payment
constant, the form in which compensation is received can affect the extent to which
these differing goals are met. Well-designed pay policies, forms, and levels can help
attract employees to the organization and retain those that the organization wants to
keep; they may also play a role in discouraging undesirable candidates and prompting
unwanted employees to leave. As accentuated in Chapter 1 0, pay policies should help
employees deal with uncertainties in their earnings opportunities. Pay should also
signal what the organization values and what behavior and attitudes it wants to
discourage, and it should help employees decide how to allocate their time and effort
among competing ends. It should reward accomplishments and successes (and perhaps
punish failures), and it should provide motivation for behavior that contributes to
organizational success. Finally, it should do all this efficiently, meeting employees'
needs for material consumption, equity, or status at the minimal expense consistent
with realizing organizational goals.
As we have seen, these goals may not be consistent. For example, in Chapter
7 we saw that providing motivation may require basing pay on performance. However,
this may prevent realizing the gains that would come from insulating employees
against income risks. As another example, many fringe benefits are not treated as
taxable income for the employees, but the costs of providing them are deductible for
the employer. This leads firms to provide noncash compensation when, taxes aside,
the same outlay paid as cash would allow employees to allocate the funds as they
individually deem best and achieve higher levels of satisfaction. As a further example,
if employees value "fairness" (interpreted as approximate equality of pay within ranks
and "appropriately" small pay differences across ranks) or if visible pay differences give
rise to undesirable influence activities, then responding by reducing pay differences
may mute incentives and increase quits among valued employees with good alternative
job opportunities. If, at the same time, employees in higher-level positions value
status and enjoy being visibly differentiated from their colleagues, there are direct
conflicts in meeting employees' desires. Conflicts among the different objectives of
compensation policy arc widespread, and balancing the pursuit of different objectives
 :J9 1
m light of the attendant di fficulties helps account for the patterns of observed Compensation and
compensation we see. Motivation

INCENTIVES FOR INDIVIDUAL PERFORMANCE

Throughout this text, we have consistently treated the individual as fundamental. It
is individuals who deci de and act, and it is individuals who ultimately must be
motivated. Thus, although group incentives are empiri cally important, we first focus
on individual i ncentive compensation.

What to Motivate?
In our treatment of the theory of i ncentive contracti ng i n Chapter 7, the problem was
to induce the agent to provide "effort" of various sorts. There we viewed effort as a
broad metaphor, representi ng whatever the employee might find onerous that was
valuable to the employer. In analyzing and designing actual compensati on systems,
however, we need to be more specific about what it is that the employer wants from
employees and what they must be motivated to supply.
I n some jobs, the metaphor fits reasonably well because effort in the usual sense
is in fact the key to success. Productivity i n heavy manual labor, such as ditch digging,
hod carrying, or stevedoring (loadi ng or unloadi ng ships) is directly a matter of how
much physical effort is exerted throughout the work period. Even i n these jobs,
however, there are right and wrong ways to do thi ngs, and the workers' contribution
is greater when they use some wi t and skill in approaching tasks. In other jobs, what
is primarily needed is diligent applicati on of the employees' mental capabi liti es­
knowledge, attenti on, and ski lls. A bookkeeper needs to record figures accurately in
the right accounts and to ensure that the calculations are correct. Here, "effort"
becomes the care and attenti on devoted to doing the job well. In still other jobs,
physical exertion again is hardly a factor, but the employer desires higher-level
intellectual activity from the employee, such as sustained, difficult reasoning focused
on intricate and perhaps boring details from a lawyer, or ti mely creativity from an
advertising copywriter. To the extent that the employee finds delivering these activities
onerous, at least at the level that the employer desi res, any of them could be interpreted
as "effort" that must be motivated.
Nevertheless, what the employer wants from the employee is rarely representable
by some simple, one-dimensi onal construct. A rich variety of employee atti tudes and
behavior affect the organization's success and might be i nduced or deterred by the
motivati on system. Producti on workers i n a factory obviously can affect the total
volume of output by how hard they work, and they must be motivated to supply effort
in this sense. But they also affect many other important variables in di fferent ways.
They influence product quality through their attenti on to their tasks; the physical
condition of the machinery they use by the care they devote to maintaining it; raw
material costs by their economi zing on i nputs and reduci ng waste; and plant safety
by their attenti on to those matters. By suggesting improved work methods, acqui ri ng
new skills, and cooperating with the introducti on of new technologies, workers can
have a major impact on aggregate productivity, and they can improve one another's
individual productivity by helpi ng out in emergencies, paci ng themselves relative to
other workers, and trai ning less experienced employees. Motivating effort in each of
these domains should also be a concern.
Managers' jobs may be even more multifaceted. They have to supervi se and
motivate those reporting to them and win the cooperation of people in other parts of
the organi zati on. They may have to develop organizational structures, implement
plans to carry out routi ne operati ons, and devise responses to unforeseen opportunities
 392
Employment: and difficulties. They may be responsible for identifying, evaluating, and implementing
Contracts, new investments, forecasting the moves of competitors, and developing competitive
Compensation, strategies. They also may need to deal with suppliers, customers, regulators, and the
and Careers public. H ow well managers communicate, how effectively they deal with people, how
imaginatively, intelligently, and diligently they work, how efficiently they allocate
their time and attention among all the competing demands, and, ultimately, how
honestly and loyally they devote themselves to their employers' interests are all
important to organizational success.
Effective compensation and reward systems recognize that employees' contribu­
tions occur along many dimensions and motivate the employees both to exert
themselves mentally and physically and also to allocate their efforts in the ways that
serve the organization's interests.

Exp licit Incentive Pay

In some contexts, explicit incentive contracts linking pay and measures of individual
productivity or performance are used to motivate employees. Examples are piece rates
for production workers, commissions for salespeople, and stock options for executives.
Other examples from outside a strict employment relation are the commissions paid
to real estate agents as a fraction of the sales price of the property, lawyers' contingency
fees computed as a fixed share of any amount won at trial, and fees to investment
bankers in takeovers that are tied to the eventual price paid per share. Such contracts
can provide very strong incentives. However, explicit incentive contracts are not nearly
as widely used in employment relations as simple theory might lead one to expect.
We have already seen some reasons for this in Chapters 10 and 1 1 . To understand
more deeply why this is so and why other measures are substituted for such contracts,
it is important to examine some of the circumstances where such explicit individual
monetary incentives are used successfully.

Piece Rates
Piece-rate systems involve paying employees a specified amount per unit produced.
They are often applied at the individual level where, for example, an agricultural
worker might be paid a fixed amount per basket of strawberries picked, a lathe operator
might be paid a fixed amount per piece of metal shaped, a knitter might be paid a
fixed amount per sweater, or a typist might be paid by the page. They can also be
used for groups of employees. In California, for example, teams of fruit pickers are
paid based on the number of trees picked cleanly. In a factory , a team of workers
might be made responsible for producing a part, assembly component, or finished
product, with a fixed amount per unit produced being divided among the team
members.
THE ADVANTAGES OF PIECE RATES Piece rates are simple and easily understood. They
give individual financial recognition to more productive or harder-working employees,
who are thus encouraged to perform to their potential. They tie pay very directly to
performance and so can provide excellent incentives for employees to exert themselves
to produce output and generate revenues for the firm: Various studies have found
productivity increases of 1 5 to 3 5 percent when such incentive pay plans have been
introduced. 2 Compared to a system of hourly wages, piece rates differentially attract
people who are more productive or harder working. They can encourage employees
to improve their skills. They may also reduce the need for complicated performance

2 These studies are cited by Edward E. Lawler , III, Strategic Pay: Aligning Organizational Strategies
a nd Pay Systems (San Francisco: Jossey-Bass Publishers, 1990) , 57-58.
 Compensation and
Motivation

Piece Rates at Lincoln Electric Company *

The Lincoln Electric Company is renown for its long-sta nding, very extensive
use of explicit incentive pay. Since their adoption in the 1 930s, Lincoln's
systems have produced spectacular productivity for this Ohio-based manufac­
turer of arc-welding equipment and fluxes, making it both exceptionally
profitable and the market-share leader in its industry. Its labor productivity
has been estimated to be three times higher than that of other firms in
comparable manufacturing settings; it has an exceptional reputation for the
quality of its products; and it has gone al most 60 years without losing money
in any quarter.
Lincoln's factory workers are paid piece rates rather than hourly wages
or salary. Over the years management has established a credible policy by
which rates will be adjusted only when a change in work methods or
technology is instituted. The workers are also eligible for yearly bonuses that,
on average, have roughly doubled their earnings . The distribution of bonuses
is based on merit ratings of the workers on such factors as their dependability,
quality, output, ideas, a nd cooperativeness. In addition, the name of the
worker who produced a machine is stencilled onto it. If a quality problem is
discovered, the worker responsible must correct it on his or her own time.
Further, if a machine proves defective after its delivery to a customer, the
worker's annual bonus is reduced-by as much as 1 0 percent if the problem
is a particularly serious one.
Lincoln also has a permanent-employment policy. After two years with
the firm, a worker is guaranteed at least 30 hours work a week. In fact,
Lincoln has had no layoffs for over 40 years. It also has long encouraged
close communication between workers and management. The workers own
almost half the stock in the company, with much of the rest held by the
Lincoln family.
Lincoln is a model for successful use of incentive-pay systems. Each
year hundreds of visitors from leading companies and their unions come to
see how Lincoln manages its compensation and motivation schemes.

* For further information, see "The Lincoln Electric Company," Case 376-028, Harvard
Business School, Boston, 1 975, as well as Nancy J. Perry, "Here Come Richer, Riskier Pay
Plans," Fortune (December 1 9, 1 988), 50-58.

reviews and evaluations that are necessary with other systems to determine individual
merit or performance pay. Finally, pay is determined by relatively objective measures,
and so it is less subject to manipulation, politicking, or favoritism .
DISADVANTAGES OF A PI ECE-RATE SYSTEM Against these very clear advantages of piece
rates stands a variety of difficulties . First, even in the most favorable circumstances,
the basis of piece-rate compensation may not be completely uniform . In fruit picki ng,
the measure of performance is usually the number of trees picked cleanly. If some
trees are larger, more laden with fruit, or more distant from the truck where the picked
fruit must be loaded, then the effort required for picking trees can vary. That introduces
possibil ities for favoritism and manipulation of the performance measure.
Second , piece rates introduce a variety of sources of randomness into worker
 394
Employment: incomes. For example, the breakdown of a machine that the worker uses or the failure
Contracts, of suppliers or workers at earlier stages to deliver inputs on time means that the piece­
Compensation, rate worker cannot produce and will not be paid. Poor-quality inputs can further affect
and Careers the rate of production, even for workers who are exerting themselves fully.
Variability in the demand for outputs creates a special problem because some
mechanism must be put in place to reduce output in bad times. One solution is to
reduce the piece rate and with it the amount of effort exerted. This not only imposes
income risks on the workers, it also can present a dangerous temptation for opportunistic
misrepresentation on the part of the employer to try to cut rates when times are
actually relatively good. Another approach is to maintain the rate but to ration the
amount of output that will be "purchased" or, alternatively, the amount of time the
employees are allowed to work. Both of these again load risk on the worker. The first
also eliminates many of the attractive incentives of piece rates, and the second may
fail to control the actual volume of output adequately because workers may increase
their pace in an attempt to maintain their incomes. A further possible response is to
attempt to manage demand so that its variability is reduced. This may be both difficult
and costly because, for example, it may entail failing to meet customers' orders in
peak times.
Third, the incentives provided by individual piece rates are of little value in a
traditional transfer-line production system, where one employee cannot increase his
or her output independent of the others, or in other contexts where individual
variations in the pace of work are not feasible. (It might be possible, however, to use
team piece rates in such contexts. ) For individual piece rates to work, each employee
must be able to set and vary his or her own pace.
Even when allowing this individual flexibility is technically feasible, it may
require adapting other elements of the organization's policies and structure. For
example, it may be difficult to economize on inventories by adopting Just-in-Time
(JIT) methods and still allow employees to set their own pace. This problem arose
when a General Electric Company plant producing thermocouples tried to introduce
JIT methods. 3 The workers in the plant were paid piece rates, both before and after
the adoption of the new methods, which called for leveling the production schedule
to produce a constant number of units per day so that input requirements could be
more easily forecast and buffer inventories reduced. The employees resisted this
change because they feared that the leveling of production would limit their pay and
probably reduce it.
Fourth, as the equal compensation principle indicates, paying for output alone
encourages employees to ignore other valuable activities. Piece-rate workers will be
tempted to reduce quality to increase measured quantity. They will be unlikely to
help other workers if it means reducing their own measured output. Unless they are
"charged" for inputs, they will not be careful about economizing on their use.
Moreover, unless they own the tools, equipment, and machines they use, they may
fail to maintain these as well as the firm would desire because doing so takes them
away from producing output.
The quality problem can be mitigated by close quality inspection, coupled with
charging workers for flaws and failures and/or making them personally responsible for
reworking defects on their own time. The extra inspection is costly, however, and in
any case final quality may still be lower than it would be if compensation did not
focus so much on quantity. Meanwhile, holding the workers responsible for flaws and
defects again subjects them to costly risks, for it may be unclear who is responsible

3 "General Electric-Thermocouple Manufacturing," Harvard Business School Case 9-684-040,

( 1 984).
 ;-195
for the defect or whether it actually resulted from something beyond the worker's Compensation and
control. Systems may have to be changed so that each worker is responsible for a Motivation
distinguishable unit in order to permit assigning individual accountability for defects.
The difficulties i n eliciti ng cooperation among piece-rate workers and in
motivating them to maintain equipment can be reduced by maintaining stability in
the labor force and in job assignments. Having the same people work together over
long periods makes them more willing to lend one another a hand because there is
greater mutual dependence in the group and more opportunities to acquire status and
repay favors. In addition, having i ndividual workers i n the same jobs for long periods
means that if they fail to maintain the machines and equipment they use today, they
will bear part of the costs tomorrow when breakdowns occur and they cannot produce
and earn. Nevertheless, if the workers do not bear the full costs of maintaining,
repairi ng, and replacing machines and equipment, there will still be an incentive to
misuse or overuse them. This is especially problematic when machines and equipment I
must be used· by several different workers, as occurs when there are multiple shifts or
when a particular machi ne's capacity is very large compared to a single worker's usage
of it. In these circumstances there is a free-ri der problem among the workers on top
of the moral hazard problem. The solution of having the workers own the machines
(perhaps by owning the company itself) is costly and often infeasible. At a minimum,
it imposes the risk of fluctuations in asset values on the employees.

IMPLEMENTING AND MANAGING PIECE RATES Many compensation systems involve a

base salary plus piece rates. Even where individual output can be measured with
reasonable precision, both setting the base salary and the piece rates and adjusting
these as circumstances change represent major problems.
The incentive-intensity principle indicates that piece rates ought to be higher
the higher are the additional profits that accrue from extra effort, the less risk averse
are the employees, the greater the extent to which measured output accurately reflects
individual effort, and the greater the opportunities for the employees to respond to
increased incentives. Once the piece rate is determined, the base pay should then be
set to make the entire package competitive with other employers' offers. In actuality,
each of these factors must be estimated in determining the base salary and the piece
rate. As discussed in Chapter 7, a crucial step i n this is setting a standard for how
much a worker should be able to produce under normal conditions. The hi gher the
standard, the lower the worker's income from any given level of effort and production.
The workers' i nterests typically are i n having the standard set low. Thus, they
may have an incentive to work slowly when standards are being established,
misrepresenting the difficulty of the task to gain higher income in the future. This is
an instance of the ratchet effect. A second instance of the ratchet effect arises if piece
rates can be adjusted based on experience. We have already described several examples
of this, such as the Soviet chemical combine (Chapter 7) and the General Motors
panel stamping plant in Flint, Michigan (Chapter 5 ). In each of these cases, providing
stronger i ncentives led to dramatic and unexpected increases in productivity, and
authorities responded by raising the standard, declaring that the previous one was too
low. Unless management can commit itself to let workers keep the promised share of
their increased output, piece rates are more likely to cause bitterness and resentment
than permanent increases i n output. The fear that management may continually raise
the standards in a piece-rate system has been suggested as the source of organized
labor's long-standi ng opposition to the system.
A related difficulty is that the workers may have an incentive to hold down
production, even after standards are set, in hope that the standards will be lowered if
they prove difficult to meet. Edward Lawler cites an example in which a secretary
 396
Employment: was asked during a strike to take on a factory job that was paid on a piece-rate basis.
Contracts, Despite having no previous experience, she was breaking production records by 3 7 5
Compensation, percent within days. The worker who normally held the job had been in it for ten
and Careers years and regularly complained that the standard was too demanding. 4
Potential problems of setting the rates arise anew with every change in economic
conditions. The introduction of new machines and methods, for example, calls for
recomputation of standards and rates. Demand changes, which are unlikely to be
observed by the employees with the same precision as they are by the firm's
management, present the special difficulties to which we have already alluded. Theory
suggests that rates be lowered when marginal revenue falls, as it might in the face of
reduced demand. Suppose, however, that the workers are enjoying quasi-rents in the
firm, so that they will not immediately leave for other employment if their earnings
are reduced. Then management will have an incentive to misrepresent demand
conditions, claiming they are worse than they actually are, in order to justify lower
pay. In response to this problem, the usual remedy is to fix piece rates without regard
to demand conditions, but this raises the question of how levels of output are to be
controlled.
The restricted range of applicability of piece-rate systems and the difficulties of
managing them have limited their use. Apparently many companies find that there
are other, more effective methods of motivating production workers-especially direct
monitoring of effort, perhaps combined with some form of group incentives. At the
same time, these companies find that individual incentive pay is effective with
employees in other types of jobs.

Sales Commissions
Incentive pay based on explicit formulas is very widely used in compensating
salespeople, who regularly receive part or all of their pay through comm1ss1ons.
Greater sales result directly in higher pay for the salespeople, thus motivating them
to focus more energy, diligence, and imagination on their jobs. As with piece rates,
commissions also induce self-selection. Salespeople who perceive themselves to be
particularly talented, dedicated, or energetic and who are less risk averse are attracted
by the high earnings possible under a commission scheme, whereas the more risk
averse and those who see themselves as less effective salespeople will prefer a job
paying a straight salary.
As with piece rates, the first cost of a commission system is that employees'
incomes vary not just with their efforts but also with all the random factors that might
influence sales success. This means that their average income will have to be higher
to compensate for the risk as well as for the extra effort that is induced. When
combined with the self-selection effect noted earlier, this effect may explain the
difficulties that a number of major U. S. department store chains, including Sears',
Dayton Hudson, Bloomingdale's, and Macy's, have experienced in instituting or
increasing the importance of commissions for retail salespeople. Unless the new system
raises expected total income significantly, the risk-averse salespeople who were attracted
to the salaried positions will be worse off. Moreover, those who perceive that they are
not especially effective will foresee that they will have great difficulty meeting targets,
maintaining their incomes, and keeping their jobs. They will naturally oppose the
shift in the compensation system.
Commissions are in fact much more commonly used for field salespeople who
operate on their own, away from the direct monitoring of their managers, than for

4
Lawler, op. cit . , 58.
 397
retail salespeople sell ing in stores. The possibilities for monitoring employees in the Compensation and
field are quite limited, so providing incentives through direct supervision is problematic. Motivation
In contrast, store managers can more easily observe employees' direct provision of
effort. They can then provide incentives without subjecting the employees to the
i ncome risk inherent in a commission scheme. Along these same lines, com missions
in retail selling seem to be more prevalent where there is a greater opportunity for the
salespeople to influence sales by difficult-to-monitor activities, such as the quality of
advice given. Such advice is very important in selling major appliances, upscale
clothing, and consumer electronics, and commissions tend to have a relatively larger
role in compensating people selling these items. (Note that this pattern is consistent
with the recommendations of the theory of incentive contracting: If profits are more
responsive to effort, then the commission rate should be higher. )
The risk that a commission system imparts to employees' incomes is an important
determinant �:> f the desirability of using com missions at all. When the risks are
relatively small, a large fraction of the worker's total income can be based on
commissions. Selling shoes in a retail store, wholesale groceries, or standard industrial
supplies and materials are examples. But when sales are large and infrequent, such
as billion-dollar contracts to build nuclear power stations, then paying exclusively on
a commission basis would make the employee's income unacceptably variable. In
such circumstances, the options are to seek out salespeople who are especially risk
tolerant or to ignore commissions all together, by paying a salary and providing
incentives by other means.
Even when commissions are valuable, they share another difficulty with piece
rates: They may cause salespeople to slight other activities that are valuable to the
firm. For example, if the sales force is a potentially important source of information
about customers and competitors, paying commissions may cause them to focus
excessively on sales and to ignore information gathering. One logical possibility is to
reward information acquisition as well. As the equal compensation principle indicates,
if pay is to motivate gathering information, then the reward at the margin for time
devoted to this activity must be equal to that for time devoted to direct sales activity.
Performance in information gathering may be exceptionally hard to measure, however,
so basing pay on measured performance on this dimension (if it is possible at all) will
add a large measure of risk to sellers' incomes. This extra risk might well make this
approach unattractive. The solution instead may be to reduce or eliminate the sales
commission, relying more on straight salary, so that the salespeople's allocation of
their time is more in line with the firm's interests.
The same sort of problems arise when the sales force is expected to provide
customer education and after-sale service. If a salesperson serves the same customers
and territory for extended periods, his or her concern with future sales and the resulting
commissions may alleviate this problem. When there is high turnover and service is
important, however, it may be better to reduce the role of commissions.
SETTING COMMISSION RATES: THEORY AND EVIDENCE Setting commission rates in
different selling environments is a central issue in sales force management. Two
marketing professors, Rajiv Lal and Venkantaraman Srinivasan, have used the model
presented in Chapter 7 to address this issue. 5
They considered first the case in which the salesperson handles a single product.
Suppose that the sales response fu nction, which relates the expected number of units

5 Ra j
iv Lal and Venkantaraman Srinivasan, "Compensation Plans for Single- and Multi-Product
Salesforces: An Application of the Holmstrom-Milgrom Model," draft, Stanford University, Graduate
School of Business (l 99 l ).
 398
Employment: sold to the sales activity, is given by h + ka , where h is a constant representing the
Contracts, expected unit sales if the salesperson does the job making no special effort, and k
Compensation, parameterizes the responsiveness of sales to individual sales activity a . The correspond­
and Careers ing expected contribution to profits is m X (h + ka ), where m is the markup (price
minus marginal cost) on the units being sold. Finally, suppose that the personal cost
of effort born by the agent is given by a quadratic function:
Cost = b0 + b1a + ½ca 2

The theory of Chapter 7 can be applied to this problem by defining effort in a

particular way: let e = ka. Then the cost of effort C(e) is just the cost of the
corresponding level of sales activity a = elk, which is b0 + b 1 (elk) + ½c(elk)2. With
this definition of effort, the model used by Lal and Srinivasan corresponds exactly to
the model of Chapter 7, so we can use the incentive intensity principle to determine
the optimal commission rate.
According to the incentive intensity principle, the commission rate should
depend on four factors: the marginal profitability of effort, the employee's risk aversion,
the significance of random factors in determining measured performance, and the
responsiveness of effort to increased incentives. The formula for the optimal commission
rate is m/[ I + rcr 2c/k 2 ], where r is the coefficient of absolute risk aversion, cr 2 is the
variance of unit sales for any given level of effort, and c/k2 is C"(e). The commission
rate applied to unit sales should be an increasing function of the margin m and the
coefficient k of the sales response function and a decreasing function of the salesperson's
risk aversion and the magnitude of uncontrolled and random factors in determining
sales. In particular, the commission rate should not depend on the level of base sales
h . The base sales h should affect only the performance standard or, equivalently, the
salesperson's base salary.
The researchers found general empirical support for the theory using data from
studies of several different firms: The theory's qualitative recommendations tend to
match actual pay practices. Moreover, the predictions of the theory have been tested
more directly using sales-force compensation data from an unnamed large U. S. firm
selling computer-related services and hardware. 6 The company uses a variety of
combinations of salary and commissions for its various product lines. The data used
in the study were obtained from a questionnaire completed by sales managers. The
questionnaire itself was partially framed in terms of concepts and issues with which
these managers usually concern themselves, such as the importance of team selling,
advertising, product quality, firm reputation, and service, as well as some of the factors
directly figuring in the theory. To test the theory using this data, it was necessary to
interpret the managerial variables in terms of the variables in the theory, and this
creates potential difficulties in matching the empirical results to the theoretical
predictions. Nevertheless, as interpreted by the researchers, the agency model did a
good job of explaining the determinants of sales-force compensation in this firm.
When a salesperson sells several products, the salesperson's ability to affect sales
by increased effort, the gross margins, and the variability of sales may differ across
items in the line, and the salesperson must decide not just how hard to work but also
how to allocate his or her total sales effort among products. Lal and Srinivasan, using
the logic of the equal compensation principle, developed recommendations on how
the commission rate should vary across products. Other things being equal, the
com mission rate on unit sales volume should be higher for products with higher gross

6
Rajiv Lal, Donald Outland, and Richard Stael in, "Salcsforcc Compensation Plans: An Empirical
Test of the Agency 'l'heory Framework, " draft, Stanford University, Graduate School of Business (1990).
 399
margins, greater responsiveness of sales to selli ng ac tivity, and less randomness in sales Compensation and
outc omes. Again, it should be independent of the level of base sales. Motivation
These last recommendations c onflict with some conventional wisdom i n
marketing management, so it would be especially useful to test these predic tions using
em pirical evidence. U nfortunately, such tests have not yet been c onducted.

Individual Incentive Pay in Other Contexts

M ore and m ore companies are experimenting wi th fo rmal ized indi vidual performance
pay for a variety of different em ployees. For exam ple, although the bonuses paid to
Salomon B rothers investm ent bankers are norm ally determ ined by subjective evalua­
tions, The Wall Street Journal reported that in 1990 the bond-arbitrage group at
Salomon was paid by a strict form ula gi ving them 1 5 percent of the profits they
generated. 7 This group uses mathematical m odels to estimate equili brium price
relations between bonds and other securities. When the m odels indicate there are
even very small disparities i n actual relative prices, the group simultaneously buys the
underpriced securities and sells the overpriced ones. If the m odels are right, this yields
a riskless "arbitrage" profit; if they are wrong, imm ense l osses are possible. Reportedly,
the head of the group received $23 million for 1990 (som e of which probably was
paid in the form of c om pany shares that he could not access for 5 years-see Chapter
1). Others in the group reportedly got huge bonuses under the fo rmul a as well. A
similar formula linki ng pay to perform ance resulted in the $550 million that Michael
Milken, the "Junk B ond King, " was pai d by Drexel Burnham Lam bert in 1987.
Apparently he had been promised 3 5 percent of the profits generated by his group
when he began trading i n high-risk bonds in the early 1970s, and this arrangement
stayed in effect even as the pr ofits soared. 8
PAY FOR SKILLS Am ong regular employees, pay-for-skills programs are a fo rm of
incentive pay that rewards and m otivates em pl oyees' i nvestments i n skill acquisition
and devel opment rather than their direct on- the-job performance. These program s
are common i n large Japanese firm s and are being tried elsewhere in N orth America
and Europe. In these program s, an em ployee's pay depends not on his or her particular
j ob assignment but instead on the skills he or she has acquired and his or her level
of mastery. For exam ple, in a Japanese manufacturing firm, the c om pany m ight
identify several different m achines that a particular group of workers might conceivably
learn to operate. Each worker is then graded on his or her level of expertise on each
machine: Can the worker operate the machi ne at all? Can he or she operate it without
regular assi stance and close supervision? Does he or she have enough expertise to help
others in using the machine? H as he or she ac tually achieved enough expertise to be
able to teach other workers? H igher grades result in higher pay. On the one hand,
such schemes encourage human capital investment, and they facilitate valuable
flexi bility in work-force assignm ents. On the other hand, they may require paying
people for skills that they may rarely use.
PERFORMA NCE PAY FOR SCIENTISTS AND ENGINEER S Pay-for- perfo rmance schemes also
have been used with scientific and engineering personnel. For exam ple, Applied
Materials Inc. , a $500-million California-based producer of equipment fo r manufactur­
ing semiconductors, gives employees who develop successful new products a percentage
of th e resulting sales revenues. U nder this plan, the physicist who led the team that

7 Randall Smith and Michael Siconolfi , "Roaring '90s? Here Comes Salomon's $23 Million Man, "
The Wall Street fournal (January 7, 1 99 1 ), C- 1 .
8 Connie Bruck, The Predators' Ball (New York: Penguin Books, 1989), 3 1 -2.
 400
Employment: developed one especially successful product received more than $800, 000 in 1 989 in
Contracts, incentive pay. He thus ended up earning considerably more than the corporation's
Compensation, chief executive officer. Such plans directly reward and encourage creativity and
and Careers innovation, and they also help motivate researchers to be concerned with the ease
with which their products can be manufactured and sold. They are especially attractive
in the high-tech ind ustries of California's S ilicon Valley because they help hold
engineers and scientists who otherwise would be lured away to new, start-up, firms,
where they can have more independence and a significant ownership stake.9
MANAGERIAL PERFORMANCE PAY The most widespread use of formula- based incentive
pay is probably for managers. Often this takes the form of the manager receiving a
bonus that is a straightforward percentage of sales or profit (or chan ges in these) in
the manager's unit, but sometimes they can be much more complicated. McDonald's
provides an excellent example of such complex schemes.
The fast-fo od restaurant chain has used explicit perfo rmance compensation
systems for decades to reward and motivate its restaurant managers. 1 0 Over the years,
McD onald's has experimented with a variety of managerial incentive- pay formulas.
Beginning in 1967 managers' and assistant managers' base salaries were tied to their
meeting corporate standards of quality, service, and cleanliness (QSC) in their
restaurants, and quarterly bonuses were based on "controllable" profit contribution:
sales revenue less those costs that the manager was expected to be able to control.
This scheme apparently did not effectively reward cost-control efforts, however, and
it also proved unpopular. I n 1972 it was replaced by a system in which starting salary
was based on local costs of living and adj usted annually accord ing to overall
performance, and bonuses depended on a number of factors. Meeting labor expense
targets and negotiated paper and food cost targets each yielded a bonus of 5 percent
of salary. QSC performance, which was graded on a three-point scale, contributed
up to another 10 percent of salary to the bonus. Further, the manager received 2. 5
percent of any sales increase, up to another 10 percent of salary. (If extenuating market
conditions limited sales growth, the manager's supervisor could award up to a further
5 percent in place of the share of revenue growth.)
This scheme was further revised two years later. Under the 1974 plan, the bonus
depended on measured QSC performance, cost control, revenue generation, and
training of personnel, with the weights attached to each negotiated between the
manager and the manager's supervisor. The training factor was added because
McDonald's was expanding rapidly and faced a shortage of managers. S imilar factors
are still used, although the pay scheme continues to be adj usted.
These schemes are designed to motivate the managers to devote themselves to
their j obs and to allocate their attention among various tasks in ways that support
corporate objectives. They also try to recognize differential market conditions (via the
salary fi gures), the man agers' knowledge of local conditions (through having some
targets negotiated), and the extent to which different observables are reflective of
managerial effort (by treating various elements of performance separately and giving
them different weights).
Eliciting Employees' Private Information
Often employees have information that is not available to their superiors about their
own personal circumstances or interests, what performance is possible on the j ob, or

9
Valerie Rice , "Tying Pay to Sales Puts Inventor at Top , " San fose Mercury News (July 16 , 1990),
D- 1 .
Sec W. Earl Sasser and Samuel H . Pettway , "The Case of Big Mac's Pay Plans, " Han·ard Business
•
0

Review (July-August 1974), 30ff.

 401
Compensation and
H ighest Ex pected Pay
Moti vation

Figure 1 2. 1 : Which contract gives the highest

0 S' S" expected pay depends on the level of expected
Sales sales.

the returns to various activities that the firm might undertake. For example, a division
manager is apt to be much better informed about the opportunities for enhancing
performance in his or her division than is the central management staff, and an
experienced salesperson is likely to know more about the market potential of his or
her territory than is the sales manager. I n such cases, it can be useful to tap this
information in setting objectives and rewarding activity. The potential difficulty is that
unless the system is designed appropriately, the employees have incentives to withhold
the information they have or to distort their reports.
To overcome this problem and gain access to the information, the organization
must be careful not to punish honest reporting. In general, this may be difficult,
especially because i t may mean committing not to use the information fully. As an
extreme example, suppose a new airplane crashes during a test flight, but the pilot
survives. The manufacturer will be tremendously interested in learning what caused
the crash. Suppose the pilot knows that an error on his or her part was partially
responsible but that this cannot be discovered unless he or she confesses. If the
manufacturer is to get the right information, it must not treat the pilot worse if he or
she accepts responsibility than if he or she keeps quiet. Yet upon learning the pilot's
culpability, there will be a strong temptation to fire him or her.
Two common mechanisms can be interpreted as means of using employees'
private information. One is to offer a menu of contracts, a variety of different methods
of determining pay among which employees can choose. This has been done, for
example, at IBM in compensating the sales force. The second is the practice of
management by objectives, under which the employee and his or her supervisor
negotiate the criteria and standards against which the employee's performance will be
judged. Versions of this system are extremely common and are used at multiple levels
within numerous organizations.
MENUS OF CONTRACTS Different contracts in a well-designed menu offer different
combinations of salary and incentive pay-for example, commission rates-with lower
salary attached to higher commissions i n such a way that each contract offers the
highest pay over some possible range of performance. The basic idea is illustrated in
Figure 1 2. 1 with a menu that involves three contracts. In the figure, the three sloping
lines represent alternative contracts that the salesperson may choose. The slope of
each line represents the commission rate and the y-axis intercept represents the salary
paid when no sales are made. As shown in the figure, contract 1 offers the highest
salary (S 1 ) and the lowest commission rate, contract 2 offers a lower salary (S 2 ) but a
higher commission rate, and contract 3 offers a still lower salary (S,) and a still higher
 402
Employment: commission rate. Contract 1 yields the highest income if sales are less than S'; contract
Contracts, 2 is the most lucrative in the region from S' to S"; and for sales levels above S",
Compensation , contract 3 yields the greatest income.
and Careers To see how this can be useful, suppose that the sales potential of various
territories varies significantly and that the person best informed about the potential of
any particular territory is the salesperson who works there. If a simple commission
scheme is used, then each salesperson will be motivated to underestimate the sales
potential of his or her territory. In that way, he or she can make the sales target and
earn bonuses and commissions without having to work very hard. The question is
how to motivate salespeople working in promising territories to accept and try to meet
higher sales goals. The menu of contracts illustrated in Figure 12.1 provides the
answer. Faced with such a menu, the salespeople will engage in self-selection: Those
expecting low sales will select contract 1, those expecting a moderate level of sales
will take contract 2, and those expecting high sales will take contract 3 . In this way,
strong incentives can be provided for salespeople in promising territories, whereas
those in weaker territories can be protected fr om the conseq uences of low sales by
being compensated principally by salary.
MANAGEMENT BY OBJECTIVES Offering the same menu of contracts to each employee
is optimal only when the employee's report is the only source of information about
achievable performance. In general, the proper standard may be better determined in
conference between the employer and employee, enabling the parties to use all the
available information. Just as in the case of the menu of contracts, to ensure that
employees have no incentive to conceal their information, the employer needs to
offer more attractive performance rewards to those who accept higher performance
goals.

Implicit Incentive Pay

M ost often, pay is not linked to individual performance by any simple formula.
Instead, supervisors decide on pay, raises, and promotions according to some
unarticulated and perhaps subj ective or ill-defined criteria. Yet there may still be
incentive aspects to the compensation system. At a minimum, those who perform
especially badly may be fired, and in most situations this is a spur to attempting to
perform acceptably. Beyond this, people who are j udged to be doing a better j ob tend
to be treated better, and, understanding this, employees are motivated to perform in
the ways that they perceive the organization values and will reward.
THE RATIONALE FOR IMPLICIT INCENTIVE PAY Implicit incentive schemes represent
highly incomplete contracts. Their prevalence results from two factors: the diffi culty
of specifying desirable performance in advance and the difficulty of adequately
measuring performance after the fact.
Except for the most routine j obs, it is rarely possible to foresee all the various
significant contingencies that might arise and to determine in advance the responses
that would define good performance. For example, a manager has to deal continually
with all sorts of emergencies, new opportunities, and unfamiliar situations. If it were
possible to foresee all these, describe them, and specify the right behavior in each
event, there would be no need for managers as we know them: The responsibility and
discretion that mark managerial positions would be eliminated. Generally, if the
employer and employee cannot specify in advance what will constitute good
performance, they obviously cannot enter a contract that explicitly ties pay to
performance.
Yet these diffi culties in specifying in advance what constitutes good performance
do no t automatically preven t the use of explicit incentive contracts because there may
 403
still be good proxies for overall performance. Consider, for example, the chief executi ve Compensation and
officer of a firm. It would be ridiculous to try to specify i n advance what the CEO Motivation
should do i n all the si tuati ons that might ari se and on all the dimensions on which
he or she might need to act. Yet from the stockholders' poi nt of view, the change i n
the value of the company's stock may be a qui te adequate "summary stati stic" fo r the
CEO' s performance, and an explicit compensati on contract could be based on stock­
price performance.
M ore often, however, there are no such simple, obj ective i ndicators for
performance that can be easi ly observed wi thi n a reasonable time frame. What
understandable measures could adequately and unambiguously summarize the perfor­
mance of a secretary, a cost accountant, or a high school pri ncipal? Both the empl oyer
and the employee may know well what constitutes good performance, i n any given
situation, and yet descri bi ng this performance i n advance and measuri ng i t i n a
quantifiable way after the fact are both impossi ble. Generally, when no easily
quanti fiable, credible summary measures for performance exist, explicit performance
contracts will be impractical. Instead, performance will have to be evaluated after the
fact, on many dimensi ons, often subj ectively, agai nst standards and cri teria that were
not and could not have been articulated beforehand, and that may remai n ill defined.
Correspondi ngly, any i ncenti ves that are gi ven through the compensati on system will
have to be implici t.
M oreover, even when a few aspects of performance are measurable, i t may be
undesi rable to use those measures i n explici t pay form ulas. Paying for performance
on only some of the dimensi ons that matter encourages em ployees to slight the other
dimensi ons of their j obs, and the result may potentially be worse than if there were
no explicit i ncentives given at all. This observati on is j ust another appli cati on of the
eq ual compensati on principle. S ometimes, thi s problem can be overcome by mixi ng
explicit and implicit i ncentive pay. For example, Li ncol n Electric uses piece rates to
reward the direct producti on of output and the associated provisi on of effort and
a ttention, but the size of an i ndi vidual's bonus depends i n part on the firm's
evaluati on of hi s or her performance on less tangi ble dimensi ons, such as reliability,
cooperativeness, and the quality of the indivi dual's ideas. As long as the empl oyees
understand the factors that figure i nto the bonus determi nati on and believe that their
pay will depend on these, they will be m otivated to pay attenti on to these aspects of
their performance, even though there is no expli cit formula li nki ng pay to performance.

PERFORMANCE EVALUATION
At the heart of any i ncenti ve system is a system for performance evaluati on. The m ore
i ntensively the empl oyer rewards measured perfo rmance, the m ore important i t i s to
measure performance accurately. The m oni tori ng i ntensi ty pri nci ple discussed i n
Chapter 7 i s a formal versi on of thi s dictum.

Performance Evaluation with Explicit Performance Pay

When pay is to be based on an explici t i ncenti ve formula, it is especially important
tha t the right measures be used. For example, suppose a firm wants to m otivate
executives by paying for i ncreasi ng profitability. H ow should profitability be measured?
Using accounti ng profits i nvi tes mani pulati on of the accounti ng procedures, for
example, recognizi ng revenues and costs ( recordi ng them on the firm's i ncome
statement for the peri od) at times that are opportune for the executi ves. M ore crucially,
i t may cause them to emphasize i nvestments that raise revenues and mea sured profits
in the near term and to avoid even better i nvestments that do not contri bute to
immediate accounti ng returns. They may also hang on to losi ng operati ons to avoi d
 404
Employment: having to recognize losses and take write-offs that hurt reported profits. Presumably,
Contracts, stock prices are harder to manipulate and, at least in some conceptions of how the
Compensation, market works (see Chapter 1 4), these prices give appropriate weight to changes in both
and Careers the long- and short-term earnings prospects of the firm. Of course, the price of the
stock may be very insensitive to the actions of any individual manager below the
highest executive levels in the firm, and so it may not be a very effective measure
either.
Another issue is whether to measure individual performance on an absolute
basis against some given, immutable scale, or relatively, in comparison to others'
performance. The latter has the advantage that, if done right, it can help sort out
some of the variations in performance that are not due to the employee's efforts but
rather to outside influences over which he or she had no control. For example, a
given level of sales probably reflects a greater effort by a salesperson when the
salespeople in other districts generated low sales volumes than when they were very
successful. The idea is that the low general level of sales suggests that demand is weak
or that the product is overpriced, so that achieving a given level of sales probably
required more effort than would have otherwise been needed. Evaluating the
salesperson's performance relative to others' performances in such situations removes
some of the risk in the individual's pay and should thus be valuable. The informativeness
principle indicates when such comparative performance evaluation is useful and how
much weight to give to relative versus absolute performance.

may be manipulated by the group of employees who are being compared: If they all
One difficulty with comparative performance evaluation is that the evaluations

conspire to take it easy, then none looks bad in comparison. Such a conspiracy is
especially threatening when pure relative performance schemes are used. Tournaments,
where only the ranking of the contestants matters and not their absolute level of
performance, are thus especially vulnerable, particularly if the employees have some
way to share the prizes among themselves.
A second problem is that comparative performance evaluation damages the
incentives for employees to help one another because taking time to help is then
doubly costly: It reduces the helpful employees' own absolute performance and, by
improving the other person's performance, it worsens the helper's relative standing.
In fact, relative performance evaluation can even create incentives for employees to
sabotage one another. For both of these reasons, comparative performance evaluation
and tournaments tend to be used only with individuals and groups who are isolated
from one another and do not require one another's help.

Performance Evaluation in Subjective Systems

As we discussed in Chapter 1 1 , managers and supervisors seem to find subjective
performance evaluation difficult and unpleasant.· It is not easy to have to tell someone
about their failings. The task becomes even less pleasant in the ambiguous environments
that lead to implicit incentive pay because the employee can, perhaps legitimately,
dispute the supervisor's evaluation. In fact, many employees perceive that their
performance evaluations are neither appropriate nor useful, and they do not trust their
supervisors to evaluate them fairly and accurately. Further, differentiating among
employees on subjective, ambiguous, debatable grounds invites employees to try to
manipulate the perceptions and decisions of the supervisor to their benefit. These
factors may help account for some important empirical patterns.
In many companies, supervisors are expected to rate employees' performance
on a common, predetermined scale. A first guess would be that the actual ratings
should be spread more or less symmetrically across the scale. A more sophisticated
 405
Table 12. l Distribution of Compensation and
Performance Ratings of Motivation
Managers at Two
Manufacturing Firms

Percentage of Sample
Receiving Rating

COMPANY 1 :
Not Acceptable 0. 2 %
Acceptable 5. 3 %
Good 74. 3 %
Outstanding 20. 2%
COMPANY 2:
Unacceptable 0. 0%
Minimum Acceptable 0. 0%
Satisfactory 1 . 2%
Good 36. 6%
Superior 58.4%
Excellent 3. 8%

estimate would predict that scores ought to be somewhat skewed towards the high
end. If the company gets rid of employees who fall in the lower categories or helps
them improve, then performance should be improving over time. Assigning employees
to those jobs that best suit their tastes and abilities would have the same effect, as
would any learning the employees gain through experience. Then, if the standards of
performance needed to reach any rating level are not being raised over time, the
actual ratings should have a tendency to clump in the higher levels.
Yet the amount of skewness that supervisors' actual rankings show seems extreme.
For example, one study documented the performance ratings given to managers at
two (unnamed) manufacturing firms in the 1 970s. 1 1 The first company rated 4, 788
managers on a four-point scale; the second rated 2 , 84 1 managers on a six-point scale.
The results are summarized in Table 1 2. 1 . In the first company, 94. 5 percent of the
managers were rated in the top half of the scale ("Good" or "Outstanding"). In the
second company, the corresponding figure was 98. 8 percent, with essentially no one
being rated in the bottom two categories! Either these companies had an absolutely
remarkable group of managers working for them or, more plausibly, the people doing
the evaluations were reluctant to assign anyone to the lower performance levels. In
effect, the evaluators were using a two-point scale at the first firm rather than the
official four-point one and a three-poi nt scale at the second rather than a six-point
one. They thus made much less differentiation among em ployees' ratings than they
could or than they were apparently expected to make.
Also, in many systems that nominally are based on merit pay, there seems to
be very little difference in the pay actually awarded to top performers relative to those
rated lower. For example, at the first of the two companies in the study j ust cited, a
manager rated "Outstanding" was, on average, paid only 7. 8 percent more than one
whose performance was rated "Not Acceptable. " At the second company, an "Excellent"
rating corresponded to only 6. 2 percent hi gher pay than the lowest rating actually

11 James L . Medoff and Katherine C. Abraham , "Experience, Performance and Earnings," Quarterly
Journal of Economics, 95 (December, 1980), 703-36.
 406
Employment: given. 12 Thus, unless these managers are extremely sensitive to relatively small
Contracts, variations in pay, the pay scheme does not seem to give very strong performance
Compensation, incentives.
and Careers Not differentiating very much among employees in the ratings given them or
in the pay that results fits with the observations that managers find performance
evaluations difficult and unpleasant, that the employees do not have much faith in
the evaluations, and that differentiating sharply among employees on subj ective
grounds invites costly influence activities. If managers dislike having to do performance
evaluations, they may not be very conscientious about them, and if giving negative
evaluations is especially unpleasant, they will be temp ted not to give low ratings.
(N ote that it is difficult for the managers' supervisors to verify the quality or accuracy
of managers' evaluations of their subordinates. ) As we noted in Chapter 11, this
tendency may be accentuated if the managers themselves are graded on the quality
of people they have attracted to work for them and how well they get their people to
perform. In that case, a negati ve evaluation of a subordinate reflects badly on the
supervisor. Also, to the extent that there may be limited budgets for salary increases
(as happens at least in bureaucracies), a manager may be tempted to overstate the
quality of his or her subordinates to increase the share of the limited budget allocated
to his or her group for raises.
Employees who do not trust their superiors' evaluations of their performance
are perhaps especially likely to view influence activities as legitimate. After all, the
evaluations look inaccurate and even random; making one's self look better is thus
j ust correcting a misperception, not misleading anyone. Further, given that the
evaluations seem arbitrary, employees may well believe that they can be affected by
politicking. In these circumstances, using the evaluations to set large differences in
pay will simply ensure that the organization will bear large influence costs.
Furthermore, if evaluations are not in fact very informative, then the gains to
the firm from basing large salary differences on them are likely to be small. Of course,
with so little in fact riding on the evaluations, managers are hardly encouraged to take
them very seriously or to absorb much discomfort by giving negative evaluati ons.
PROMOTIONS AND SUBJECTIVE EVALUATION SYSTEMS In systems where performance
evaluation is subjective, it seems likely that much of the monetary incentives are
provided by the opportunity for promotion to a better-paying j ob in the hierarchy. As
noted in Chapter 11, promotions are rather blunt incentive devices. Also, to the
extent that good perfo rmance in one j ob does not necessarily mean that the individual
is well suited for the next level i n the hierarchy, promotion- based incentive systems
may fail to assign the best people to important j obs. S till, they have advantages in
dealing with managers' distaste for doing performance evaluations and reluctance to
give low evaluations to anyone, as well as with employees' associated incentives to
engage in influence activities.
Promotion decisions are important ones, especially in managerial hierarchies.
They determine who will make the key operating and strategic decisions instrumental
to the organization's success. This may give managers reason to do more careful
evaluations in connection with promotion decisions than they woul d if only a few
dollars in raises were at stake. Further, by making the salary change associated with
promotions large, companies can both provide incentives to employees to attemp t to
win promotion and strengthen the incentives for those making the promotion decisions
to treat them seriously.

12 George P. Baker , Kevin Murphy , and Michael Jensen, "Compensation and Incentives: Practice
vs. Theory , " fournal of Finance, 43 (May 1 988), 593-6 16.
 407
In this regard, one study found that vice presidents in large U . S. corporations Compensation and
get average annual pay raises of 1 8 . 8 percent on promotion, compared to an average Motivation
raise of 3 . 3 percent in years that they stay on in the same job. 1 3 In the same study,
vice presidents promoted to chief executive officer averaged a 42. 9 percent increase
in salary plus bonus. Finally, be.cause a promotion will often place the employee
under the supervision of a new manager (who may well be the superior of the former
supervisor), it is easier to monitor the quality of the evaluations done for promotion
decisions than those done for salary setting.
Of course, with so much riding on the promotion decision, there may be very
strong incentives to try to exert influence over the evaluations that determine the
choice. The competition for promotions can be quite fierce, and this competition
does not simply take the form of hard work. Candidates instead may spend huge
amounts of time and ingenuity trying to impress their bosses and even more time
worrying about how they are doing and whether they will succeed. However,
evaluations for promotions may be less subject to manipulation because they are done
more carefully. Moreover, once a promotion decision is made, it is largely irreversible,
so the period over which the influence costs are absorbed has a natural limit. In
contrast, salary decisions are more easily adjusted or offset over time and so the
influence activities to which they can give rise may be unending.

THE FREQUENCY OF EVALUATIONS The frequency with which employees are evaluated
seems to vary significantly across jobs. University professors receive full-blown
performance reviews only rarely during their careers-perhaps only when they are
hired and again when they come up for promotion or tenure-although they may
have less extensive reviews occasionally in connection with salary setting. Lawyers
and accountants in large professional firms are similarly subject to infrequent systematic
evaluations. In contrast, investment bankers are effectively reviewed very frequently,
receiving annual bonuses based on their measured performance. What accounts for
these differences?
Performance evaluations are costly. Even if the evaluator does not dislike doing
them, they require developing information and transmitting it, both of which take
time and resources. Thus, unless they are going to be used for a specific purpose,
they should be avoided. We have accentuated their use in performance pay and
promotion decisions. These uses, however, would not seem to be sufficient to justify
frequent reviews. Incentive payments could presumably be based on very infrequent
reviews, with correspondingly large payments. This would even have direct advantages
in contexts where the full impact of decisions is not i mmediately evident. Evaluations
for promotions should be necessary only when there is an actual promotion decision
to be made.
Another value to reviews, however, is in indicating to employees how they are
doing, so that they can change their behavior to match more closely what the employer
wants or find another job for which they are better suited. More frequent reviews thus
provide useful information to employees earlier, and they should value the increased
options that result.
The optimal frequency of review balances the largely fixed costs of the review
with the benefits of having information available sooner and being able to act on it.
Thus, reviews should be more frequent when they are less costly and when it is more
likely that the information they provide affects behavior. For example, academic
performance is hard to measure, whereas the quantity and quality of a typist's output

1 3 Kevin J. Murphy, "Corporate Performance and Managerial Remuneration: An Empirical

Analysis," fournal of Accoun ting and Economics, 7 (April 1 98 5), 1 1-42.
 408
Employment: is relatively easily determined. Thus, a professor should be evaluated less frequently
Contracts, than his or her secretary. In businesses where there is a high probability that any
Compensation , particular individual hired will not have the talent and ability needed for success,
and Careers there is increased value in employees' learning early how they are doing: If they learn
that things are going badly they can move sooner to a different job where they may
do better. Thus, evaluations should be more frequent. This may explain the pattern
in investment banking. Also, in occupations where employees develop large amounts
of firm-specific human capital, it is unlikely that, once this capital is acquired , the
employee would be better suited to a job with a different organization. Thus, there
is less likelihood that the information from a performance review will be useful, and
so reviews should be infrequent. This may explain the pattern in law and accounting
firms, where the professionals have developed significant expertise regarding their
particular firm's clients.

JOB DESIGN
Most jobs entail a variety of responsibilities. The issues of how many different projects
any one person should work on, how tasks should be allocated among individuals,
and whether one or more people should be involved in carrying out a particular task
are complex, and different organizations resolve them differently. For example,
production workers in North American factories traditionally have not been responsible
for repairing the machines they use. Instead, the firms employ specialists whose jobs
consist solely of doing repairs, and the production workers are supposed to leave the
repair work to them. In contrast, Japanese production workers often are expected to
repair breakdowns themselves. As another example, the accompanying box on the
management consulting fi rms McKinsey & Company and Bain & Co. illustrates how
job design differs between these two companies in ways that fit their differing business
strategies.
Many factors influence job design. In the example from management consulting,
the nature of the particular projects in which each company has chosen to specialize
seems to be a major factor in determining job design. In a manufacturing firm, the
decision to adopt a modern manufacturing strategy accentuating flexible equipment,
frequent and rapid new product development, high product quality, and speedy
response to customer needs has often been associated with complementary decisions
to redesign jobs. For example, product-design engineers will be expected to work in
teams with manufacturing people to ensure that new products can be produced easily
and cheaply; production workers will be encouraged to learn multiple skills so that
they can move among tasks and machines as requirements change; and salespeople
will be expected to report to the firm on customers' needs to help design a useful
product rather than just focusing on making sales. The organization's scale of
operations is another influence on job design: If the organization is small, then a
single manager may adequately oversee both sales and marketing, whereas in a larger
organization, these tasks may be split. Finally , external factors, especially demographic
changes, can be important. For example, the increasing role of women of child­
bearing age in many countries' labor forces has led some companies to design jobs
that facilitate parents' caring for their children.
Rather than attempting a full analysis of the job design problem, we concentrate
here on the interaction between job design and incentive pay. We focus on two issues:
the grouping of tasks into jobs and the determination of how much discretion
employees should be given in allocating their time and attention among tasks and, in
particular, between job responsibilities and personal concerns.
 409
Compensation and
Motivation

Competitive Strategy and fob Design

at McKinsey and Bain
McKinsey & Company, Inc. , and Bain & Co. , are major management con­
sulting firms that provide expert advice to companies on strategy and
organization. In recent years they have usually hired large numbers of new
college graduates to work for them for a couple of years before pursuing MBA
degrees. These jobs are very much in demand: They are quite well paid, the
opportunity they afford to learn about business is highly valued, and some
bel ieve that gai ning admission to competitive MBA programs is easier for
people with experience in the major management-consulting companies. At
the same time, the two firms compete directly for students.
At each firm, the new hires (called "business analysts" at McKinsey
and "associate consultants" at Bain) work in teams on projects for cl ient
companies. The teams are led by more senior consultants, who supervise
and evaluate the junior people. Although the work is quite similar between
the two firms, there are two important differences . First, at Bain, an associate
works on two to six projects simultaneously, whereas a McKinsey analyst
normally focuses on a single project at a time. Second, the average work
week at McKinsey is reputed to be about 70 to 80 hours, whereas that at
Bain is closer to 60 hours.
A single key difference in the strategies of the two firms can account for
these differences in their human-resource policies. McKinsey has traditionally
focused on projects whose final output is a set of recommendations to meet
the client's needs. Bain, in contrast, seeks projects that involve both its
formulating recommendations and staying on to help implement them.
Consequently, whereas a typical McKinsey project lasts 4 to 6 months, Bain
projects last about 1 8 months. McKinsey projects are conducted largely at
the client's site. Bain projects involve less time on the road, with more of
the work being done at the team's home office.
The difference in hours worked per week is consistent with a recognition
that the two firms' employees face different opportunity costs on their time.
In effect, extra working hours are less costly to McKinsey people, who are
usually on the road , living in hotels where they eat and get their laundry
done. Bain people are more often at home, where they face the usual time
demands of normal life and presumably also have greater opportunities for a
social life.
Regarding the difference in the number of projects being worked on ,
suppose that Bain were to adjust its team sizes to have its associates work on
only one project at a time, as at McKinsey. This might be directly inefficient
if there is an uneven time-pattern to the work and if a team needs to be of
a certain size to function effectively. But it would also mean that a new
graduate joining Bain would see at most a couple of companies up close
before returning to school , whereas the McKinsey analyst would see four to
six different firms. Given that a major reason to take these jobs is to
learn about business and businesses, Bain might then be at a competitive
disadvantage in the labor market. On the other side, if a McKinsey analyst
 410
Employment:
Contracts,
Compensation,
and Careers
were assigned to several projects at once, the extra travel costs and loss of
focus would be highly inefficient. Each firm's job design fits its business
strategy.

Based on interviews by A. C. Chidambaram and Kaare Holm with former Bain and
McKinsey employees at the Graduate School of Business, Stanford University, Fall 1990.

Job Design and Incentive Pay

Suppose there are four tasks that need to be done and each task requires enough time
be devoted to it that any two of them would constitute a full-time job. Suppose there
are no important physical complementarities between the tasks that make it efficient
to group particular ones together. This might, for example, rule out a situation where
one task is using a tool for production and another is maintaining the tool in good
condition because the ease of maintenance might depend on how the tool is used.
Ruling out physical complementarities focuses attention on how job design may
alter the effectiveness of a system of incentive pay. There are two important effects to
consider. First, a proper assignment of tasks to workers can make it easier to assess
responsibility for good or bad outcomes. For example, suppose a product may fail to
meet quality standards if either of two components is inserted incorrectly during
assembly. In this case, assigning the same person to install both components makes
it clear who to hold responsible for any product failure.
In managerial work, this task assignment principle is manifested as the principle
of unity of responsibility. When the success of a project or operation depends on the
coordinated execution of several separate tasks in a way that makes it difficult to assess
performance separately in each, then it is usually best to make a single individual
responsible for all the related tasks. For example, when the Du Pont corporation,
originally an explosives manufacturer, decided after World War I to begin manufactur­
ing fertilizers, it suffered losses for a period while a committee administered the two
businesses together. To solve the problem, Du Pont became the first firm to implement
a modern multidivisional structure, in which the two kinds of products were made in
separate factories, sold by separate sales forces, and administered by separate offices,
with a single person responsible for the performance of each separate division. Partly
for the same reason, many Japanese companies have rejected the traditional separation
between design, process, and production engineering, arranging for the engineers who
design a product to move with the product as it progresses through the start-up phase
and into the production phase, achieving a valuable unity of responsibility.
To isolate the second incentive issue, let us now further suppose that performance
in the tasks involved can be separately measured and assessed, but that the accuracy
of the individual assessments may vary. Suppose in particular the tasks differ
systematically in the difficulty of measuring performance accurately: Tasks I and 2
have a low error variance in measuring effort provision, whereas tasks 3 and 4 have
a high variance. How should the difficulty of measurement enter into the determination
of job design? Should one worker be assigned to do the two tasks where performance
is hardest to measure, with the other worker taking the other tasks? Or should those
tasks hardest to measure be split, with each worker doing one of them?
In terms of our model from Chapter 7, heterogeneity in the tasks that workers
 41 1

i ncentives. The essential i nsight comes from the equal compensati on pri nci ple. If
are assigned should be avoided because it complicates the problem of providing Compensation a nd
Motivation
i ncentives are to be provided for the worker to pay adequate attenti on to both tasks,

If an easy-to-measure and a hard-to-measure task are grouped together i n the same

then the worker's marginal return from time and effort devoted to each must be equal.

j ob, then it will not be feasi ble to provide the same i ndivi dual with strong i ncentives
for the first task, i n order to elicit great effort, and weak i ncentives for the second, i n
order to i nsulate the empl oyee from risk. Any attempt to do so wi ll only result i n the
employee devoti ng relatively too much time and effort to the task where i mproved
performance is more strongly rewarded and too little to the other task.
Croupi ng tasks according to their ease of measurement avoids this problem.
Suppose the two low-risk tasks are grouped together i n one j ob and the two high-ri sk
ones i n the other j ob. Then strong i ncenti ves can be provided for effort provi si on i n
each task i n the first j ob without imposi ng too much risk o n the employee i n thi s j ob
and without distorti ng his or her all ocati on of effort between tasks. M eanwhile,
relatively weak incentives can be applied to both tasks i n the second j ob where the
randomness i n performance measurement means that strong i ncentives are too costly
in terms of the ri sk they impose on the empl oyee' s i ncome. Proper groupi ng all ows
givi ng an appropriate level of i ncentives for each ki nd of j ob.
Joh Enrichment Programs and Com p lementarities
Between Tasks
Many firms are experimenting with programs of j ob enrichment by redesigni ng j obs
to add variety to the tasks that empl oyees are asked to perform and to gi ve them greater
responsi bility. The hope is that relievi ng the boredom associated with doing the same
thing all the time will i ncrease j ob satisfacti on and productivity. Another possi ble
source of pr oductivity gai ns is that by learni ng more about the overall operati on (and
not j ust one speci fic part of i t) employees may be able to use their i nformati on better
to suggest and implement improvements. The most obvi ous pr oblem is that reduci ng
specializati on may actually i mpede pr oductivity. The analysi s i n the precedi ng secti on
poi nts to further considerati ons.
S uppose that tasks 1 and 2 are identical to one another and similarly that 3 and
4 are the same. Thus, i n effect, there are two tasks, each of whi ch is enough to
constitute a full j ob, but it is possi ble to split the tasks with both employees doing
each task. Then, unless the accuracy of performance measurement is similar i n each

usi ng i ncentive pay. If thi s effect i s important, however, i t might be mitigated by

task, one effect of enriching the j obs by mixi ng the tasks is to create difficulties i n

devoti ng resources to i mprove performance measurement i n both tasks. Another effect

of splitting the tasks is to create an opportunity for comparative performance evaluati on.
As we have seen, thi s can offer substantial advantages, especially for systems where

If there are complementarities between tasks, so that doi ng one task i mpr oves
objective, cardinal measures of performance are hard to come by.

the empl oyee's productivity i n another, then there are direct productivity effects from
grouping certai n acti vities together. For example, if doi ng research in a subj ect
im proves a person' s ability to teach effecti ve, up-to-date courses i n the area, then
there are gai ns from assigning faculty to teach i n their areas of research and, moreover,
from havi ng people both teach and do research rather than concentrate on only one
of these tasks. Similarly, it may be desirable to have salespeople gather customer
informati on, even when sales performance is easily measured and the quality of
informati on collecti on i s hard to measure. In general, grouping tasks i nto j obs i nvolves
balanci ng several considerati ons, with the ease of providi ng incentives bei ng j ust one.
 412
Employment: Responsibilit y and Personal Business
Contracts,
Compensation , An important issue in j ob design is how much discretion employees should be allowed
and Careers in allocating their time at work between their direct j ob responsibilities and their
personal concer ns. The actual pattern we typically see in organizations is that the
more "responsible" the position, the greater discretion the employee is allowed in
pursuing his or her own interests. Clearly, from the point of view of maximizing total
value, it would sometimes be efficient for employees to spend time during the working
day taking care of personal business. This will be true whenever the marginal
productivity of time spent on the personal matters exceeds that on work. M oreover,
to the extent that worrying about their private concerns distracts empl oyees from their
job responsibil ities, it might even be in the employer's immediate interest to allow
employees to take care of their more pressing personal matters on company time.
People in "responsible" positions are allowed to do this, whereas lower-level people
are not. An executive may disappear for the afternoon with no questions being asked,
but a lower-level employee may be limited in the frequency with which he or she
may leave the assembly line to go to the restr oom.
In part, this difference seems to be a matter of whether (explicit or implicit)
incentive pay is an important element of the motivation system. S ome employees are
monitored directly on their input provision by having to record when they arrive and
leave work and by having supervisors directly observe their work. These people typically
are paid an hourly wage that does not vary in any obvious fashion with individual
productivity, although their pay may be docked for time absent from the j ob and they
may be fired if they are absent too often or if they are caught shirking. S uch workers
are usually allowed little discretion in determining whether to devote attention to their
work or other matters. In contrast, senior executives' allocation of time is apt to be
largely unmonitored, whereas their pay may be relatively sensitive to measured
performance.
This pattern is explained by a variant of the equal compensation principle.
When outside activities are allowed, employees will tend to pursue them up to the
point where time spent on them at the margin has the same impact on their welfare
as time spent on their job responsibilities. If there were really little or no connection
between the attention they devote to their j obs and their pay, they might tend to spend
all their time on other matters. Charity collections, personal phone calls, parties to
celebrate employees' birthdays, and absences to take sick children to the doctor may
absorb the entire working day. If the pay scheme puts an opportunity cost on the
employees' diverting their time, effort, and attention away from their j ob responsibili­
ties, however, then they will choose to allocate their time to other matters only when
they are suffi ciently important to offset the value of what they could have earned
devoting attention to the j ob. For example, salespeople paid largely on the basis of
commis sions or freelance j ournalists paid by articles actually published would regulate
their own allocation of time, without the need for explicit monitoring. S imilarly, an
aspiring l awyer or executive whose next promotion depends on good performance can
be allowed considerabl e latitude in deciding how to allocate his or her time.
Discretion and incentives provide another example of complementarity. For the
reasons we have j ust described, allowing discretion creates more value for workers
with intense in centives than for those with weaker incentives. Complementarity is a
symmetric relation, so we might also expect to find cases where it is the increase in
discretion that is fundamental and the intensity of incentives that is the response. For
exam ple, a salesperson who travels a wide territory making sales calls alon e must, by
the very natur e of the j ob, have great discretion over how the j ob is don e, how many
h ours to work, and how long to linger with clients who are old frien ds. The difficulty
of mon itoring tim e devoted to personal business m this case rai ses the value of Compensation and
strengthening in centives. Motivation

INCENTIVE PAY FOR GROUPS OF EMPLOYEES

Alm ost all of the formal theory emphasizes incentives for individuals on the gr ounds
that it is indivi duals who must be motivated to work. Yet the most common explicit
incentive contracts are applied across groups of employees. The performance of the
whole gr oup together determines the total incentive payment, and the total is usually
divided among individuals according to formulas or criteria that do not depend on
i ndividual performance.

Form� of Group Incentives

Tying individual pay to group performance takes many forms. S ometimes i mplicit
schemes are used, as i n the case of the bonuses pai d to Japanese workers. These
bonuses appear to be related to overall corporate profitability, but there is no formula
explicitly linking the two. Instead, the bonuses are determined by the companies'
boards of directors. In other cases, and probably more frequently in N orth America
and Western Europe, the linkage of pay and group performance is more explicit.

PROFIT SHARING PLANS The most common form of explicit gr oup incentive pay in
the U nited S tates is profit sharing, which approximately 30 percent of all firms used
to some degree i n 1988. 14 U nder profit sharing, employees are paid annual bonuses
that are supposed to vary with profitability (which may be defined at the overall
corporate level or at the level of an individual di vision) but not with their own personal
performance. A part of the bonus is sometimes deferred, being paid into a retirement
account. N ot all employees may be covered by the profit-sharing plan: For example,
senior executives almost always would participate, whereas unionized employees
would recei ve bonuses only if the contract with the union specifically called for it.
Despite the number of firms usi ng profit sharing, the amount of total employee
income received through such plans in the U nited States is small, perhaps as low as
1 percent. In contrast, J apanese workers receive on average a quarter of their pay
through annual bonuses that they perceive to depend on pr ofitabili ty. Pr ofit sharing
i n Japan typically applies across the whole company.
An extreme form of profit sharing is for the employees to own all, or at least a
substantial share, of the company. In that case, their incomes are very closely tied to
the company's performance, either through their receiving the profits directly, as in
a partnership, or through dividend payments and the behavior of the firm' s stock
price. Employee ownershi p is relatively common in certain i ndustries and countries.
Partner shi ps are the norm in a number of fields, including law, medicine, accounting,
management consulting, architecture, and, to a lesser and decreasing extent, invest­
ment banking (see Chapter 15). In addition, some taxi companies, garbage collection
firms, and plywood manufacturers in the U nited S tates are organized as employee
cooperati ves, as is the inter-ci ty bus service i n Israel, large parts of the construction
business in Italy, and the collection of diverse businesses in the city of M ondragon in
th e Basque region of S pain (see Chapter 16). These co- ops are owned collectively by
th e people who work in them. The gr owth of Employee S tock Ownership Plans in the
United S tates in the late 1980s had a maj or impact on the extent of employee
ownership (see the box entitled "Employee S tock Ownership Plans"). In the U nited

1 4 Nancy L. Perry, "Here Come Richer, Riskier Pay Plans, " Fortune (December 1 9, 1 988), 50-58.
 414
Kingdom, about 1, 950 companies have plans that give stock or stock options to
employees generally, and 4, 700 have plans aimed j ust at the executives. 15
Employment:
Contracts,
Other types of group incentives are often called gain-sharing
Compensation,
GAIN SHARING PLANS
plans. U nder these plans, when a group meets or exceeds predetermined targets, all
and Careers
the members receive bonuses tied to the extent to which goals were exceeded. Targeted
variables often include output levels, productivity gains, c ustomer service, quality, or
costs. Four examples from U. S. manufacturing firms illustrate these plans.
Dana Corporation Dana, the largest U . S. maker of truck parts and subassembl ies,
uses Scanlon Plans at a number of its plants. U nder these plans, a target is set for the
ratio of labor costs to the value of output produced, and when the actual ratio is less
then the target, the workers are paid bonuses based o n a fraction (perhaps 75 percent)
of the savings of labor cost over the standard. If the actual ratio of labor costs to value
of output exceeds the standard, no bonus is paid, and in fact the shortfall is recorded
and must be worked off before bonuses resume. 1 6
Nucor Corporation N ucor, a U. S. steelmaker, pays its workers weekly bonuses
based on the number of to ns of steel produced that meet quality standards. These
bonuses on average exceed the workers' base pay. N ucor also pays annual bonuses
based o n return on plant assets to its department managers and on corporate return
o n equity to plant managers, and it has a profit-sharing plan under which 10 percent
of pretax earnings each year are paid out to all employees below the senior executive
level. (Corporate officers get a combination of cash and stock. ) N ucor produces twice
as much steel per employee as do the large U. S. steel companies, and it attributes
half this differential to its incentive systems. 17
Carrier Carrier, part of the U nited Technologies conglomerate, is a leading
manufacturer of air-co nditio ning and heating equipment. The company uses a gain­
sharing program it calls "lmproshare. " U nder the plan, half the savings on labor costs
achieved by producing an increased number of acceptable units of output over a base
year are returned to the employees as bo nuses. Carrier credits the plan with a 24
percent increase in productivity coupled with a substantial decrease in rej ects during
the first two years of the program. 18
Analog Devices, Inc. Analog Devices, Inc. , a M assachusetts-based electro nics firm,
instituted an extremely complex group-incentive program during a period of very rapid
firm growth in the mid- l 970s. 19 First, there was a bonus system for managers, which
was used differently at the corporate versus the divisio nal or group levels. S enior
executives' bonuses were based on corporate performance o nly, whereas lower-level
managers' bonuses were based half on overall co rporate performance and half o n the
performance of their group alo ne. There was also a separate scheme for paying bonuses
to technical staff involved in developing new products.
U nder either plan, a person's bo nus was proportional to measured collective
performance and to his or her base salary. The bonus was also an increasing functio n

15 " Economics Focus: Unseen Apples and Small Carrots," The Economist (April 1 3, 1 99 1 ), 75.
1 6 "Dana Corporation: Richmond Camshaft Plant (Condensed), " Harvard Business School Case 9-

488-0 1 8, ( 1 987).
1 7 Drawn from Perry, op. cit.

18 Also drawn from Perry, op. cit.

19 See Ray Stata and Modesto Maidique, "Bonus Plan for Balanced Strategy," Harvard Business
Review (November-December 1980), 1 56-63; and "Analog Devices, Inc." Harvard Business School Case
9- 1 8 1 -002, ( 1 980).
 415
Compensation and
Motivation

Employee Stock Ownership Plans

By the late 1980s over 11, 000 U. S . fi rms had instituted Employee Stock
Ownership Plans (ESOPs). Formally, these are employee pensi on plans, an d
in many cases they have replaced more standard plan s. Under an ESOP, the
company acquires an amoun t of its own stock that i t places in trust in the
plan . Each year the individual employees are allocated some of these shares,
which the ESOP then hold s unti l they reti re or leave the firm. In the fi rst
half of 1989, U . S. corporati ons acquired over $19 bi llion of their stock to
establi sh new ESOPs, and in some cases the employees actually acquire the
whole company through the ESOP.
Among the reasons for the surging interest in ESOPs were that they
enj oyed apparent tax ad vantages over other forms of pensi on plan s and that,
by putting a large block of stock i n "friendly hands, " they could be used to
defend against attempted hostile takeovers of the corporati on. (Thi s was
clearly a maj or reason when Polaroi d Corp. set up its $300 milli on plan . )
ESOPs have also been used to sell companies to the employees when the
original owners retire and to allow the employees to save their j obs by
acqui ring divisi ons or whole firms that were going to be shut down. The
Weirton S teel Company, whi ch was a unit of Nati onal Steel before being
sold to the employees, is a well-kn own example.
ESOPs are also clai med to have good incenti ve effects. By making
employees into owners, ESOPs are supposed to tie the employees' i nterests
more closely to the firm' s success and motivate them better. For example,
Avis, the employee- owned car rental compan y, ad verti ses that its employee­
owners care about the firm's future and so wi ll gi ve customers better service.
There are examples (such as Weirton and Avi s) where employee
ownershi p has been associ ated with dramati cally improved performance. Yet
in other cases there has been no apparent motivati onal effect. The average
employee in a large firm with an ESOP hold s at most a minuscule fracti on
of the firm's stock. Econ omi c analysi s indicates that the direct incentive effect
of these plan s should be similarly tiny. At the same time, having both the
employees' retirement incomes and their current earnings be so completely
dependent on the fortunes of a single fi rm seems ri sky.

Based partly on Myron Scholes and Mark Wolfson, "Employee Stock Ownership Plans
and Corporate Restructuring: Myths a nd Realities," Arnold W. Sametz, ed . , The Ba ttle for
Corporate Control: Shareholder Rights, Stakeholder Interests, and Managerial Responsibilities
(Homewood, IL: Business One Irwin, 199 1 ).

of the employee's rank in the corporati on's managerial or sci entifi c hierarchies, with
this factor ranging from 10 to 25 percent of salary. Performance under the M anagement
Bonus Plan was defined in terms of (moving averages of) realized return on assets (a
measure of income generati on) and sales growth relati ve to the firm' s objectives and
plans. A matrix, and later an explicit formula, embodied the tradeoffs the compan y
made between these two factors and generated bon uses that grew more than
proporti onately wi th performance, up to a cap. A simi lar formula was used to
determine performance under the New Prod ucts Plan for techni cal personnel. It
embodied the dollar value of orders for new products and return on new product
 416
E mployment: investment relative to goals. It, too, had the feature that bonuses grew exponentially
Contracts, with performance, up to a cap.
Compensation, This system was very complicated. Indeed, it is unclear that it would have been
and Careers understood in situations where employees had less mathematical sophistication than
those in this high-tech firm. On the other hand, it was tied very explicitly to corporate
strategy and to the abilities of people with differing roles and ranges of responsibility
to contribute to corporate objectives.

The Effectiveness of Group Incentive Contracts

There are several reasons to suppose that incentives provided to groups of employees
may sometimes be as effective, or even more effective, than individual incentive
contracts. First, there may simply be situations where determining individual
contributions is impossible: How is the contribution of a single member of a design
team measured? Then if any direct financial incentives are to be provided, they must
be given to the work group as a whole. Second, groups of workers often have much
better information about their individual contributions than the employer is able to
gather. Group incentives then motivate the employees to monitor one another and
to encourage effort provision or other appropriate behavior. In these circumstances,
incentives might be further strengthened if the employer can arrange for the employees
to determine one another's rewards.
A third factor favoring group incentives is that individual employees may resist
their employer's directives if the employer's wishes conflict with those of the work
group. By using group incentives to change the interests of the work group as a whole,
the employer may make everyone willing to work more effectively and with a higher
1
'\ level of satisfaction. Fourth, people who work together have various ways of helping
one another, exchanging favors, covering for one another, and helping out with extra
effort when a member of the group is absent. Group incentives encourage such
cooperation, and the possibility of withholding help from slackers can be very effective
in providing incentives for members of the group to adhere to the group norms.
Finally, if the work group is empowered to change methods to improve efficiency,
then the ability of a group to be responsive to incentives may be far greater than is
possible for an individual employee, making incentives for groups more beneficial.
Several of these factors point to the advantages of making the group whose
collective performance is rewarded relatively small, so that mutual monitoring,
concern for the group, and exchanges of favors within the group can be effective.
Against this must be balanced the costs of measuring performance accurately for small
groups and the difficulties that can arise if some groups are getting bonuses whereas
others are not. Such a situation invites influence activities, with those not getting
bonuses arguing that the successful groups were just luckier, or that their own group's
contributions had not been adequately recognized and accurately measured, or that
everyone should share in the fortunes to which all contributed, or, ultimately, that
they individually should be transferred to one of the successful groups. Of course, the
winners will also be tempted to politic defensively to maintain their awards.
HISK SHARI NG WITH I N THE \VvRK GROUP If a work group is sufficiently tightly knit to
act as a single individual, and if it can share risks among its members efficiently, then
we can apply incentive theory to the whole group of people as if it were a single
person. In such cases, no new principles beyond those developed in Chapter 7 are
needed.
As we saw in Chapter 7, a group of people who share the total income received
efficiently among themselves can be regarded, for risk-bearing purposes, as a single
unit whose risk tolerance is the sum of th e risk tolerances of the individual members.
 417
Compensation and
Motivation

Group Incentives and Risk at Du Pont

An experiment with group i ncenti ves at E. I. Du Pont de Nemours and
Company attracted much attenti on, both when it was i nsti tuted i n 1988 and
when it was abandoned two years later. I t illustrates some of the difficulties
with managi ng such a scheme.
The plan covered nearly all the 20, 000 employees i n Du Pont' s fi bers
di vi si on, which had 1 989 sales of almost $6 bi llion i n departments from
i ntimate apparel to floori ngs to automobile seat coveri ngs. The plan required
employees to put some of their annual wage and salary increases i n an " at­
risk pot," with the i ntent that eventually 6 percent of their annual pay would
be at risk: The at-risk money would not count as base pay for future years.
B onuses were then to be pai d on how the di visi on did relative to its earnings
growth target of 4 percent after i nfl ati on. Just meeti ng the target would result
i n the employees getti ng back thei r 6 percent. If earni ngs grew 5 percent,
they would get their 6 percent plus an extra 6 percent as bonus, and if
earnings grew by 6 percent, they would get the maximum bonus of 12 percent
on top of the money they put i n. O n the downside, if the di visi on did not
achieve at least 80 percent of its target, they would lose the money they had
put at ri sk.
Early concerns with the plan centered on its motivati onal effects.
Would tyi ng pay to overall performance have any effect i n such a large
organizati on, where any si ngle i ndividual might have little i nfluence on
results? Would a blanket plan like this distort i ncenti ves i n work groups such
as research labs, where short-term earnings growth was not a very relevant
issue? The uni ons also expressed concern about how the management would
calculate earni ngs. H owever, because employees recei ved $ 1 9 milli on more
in bonuses the first year under the plan than they would have under the
traditi onal system, these concerns faded.
The problem came i n the busi ness slowdown of 1990, when it became
clear that the earni ngs target would not be met and that the employees would
lose at least 2 percent of the money they had risked. Complai nts mounted
and morale was endangered. In the face of thi s, management canceled the
experiment.

Based on Perry, op. cit. and Richard Koenig, "Du Pont Plan Linking Pay to Fibers Profit
Unravels, " The Wall Street Journal (October 2 5 , 1 990), B- l .

A cost of providi ng i ncentives i s the extra ri sk premi um i ncurred as a result of forci ng

em ployees to bear the risk of i ncome fluctuati ons associated with varyi ng levels of
performance. With these risks efficiently shared among the group, thi s cost i s i nversely
proporti onal to the risk tolerance of the group. This i n turn is much larger than the
risk tolerance of any i ndividual, and so the costs of risk beari ng may be quite low. 20
The issue then is: When might we expect that the allocati on of i ncome and i ncome
risks withi n a work group will be approximately effi cient?

zo The risk tolerance of the group is T = It; = I( l /r;), the sum of the individual risk tolerances,
where r, is the risk aversion index of individual i. The group risk tolerance is always greater and usually
much greater than that of even the most risk-tolerant member of the group.
 418
Employment: One such circumstance might be where the group decides how to allocate its
Contracts, total earnings among its members and does so efficiently. A possible example may
Compensation, arise in partnerships. For example, a number of not-for-profit clinics and Health
and Careers Maintenance Organizations (HMOs) that provide prepaid medical and hospital care
(including Kaiser-Permanente, the largest such organization in the U nited States)
contract with for-profit partnerships of doctors to provide the actual care. The contracts
often provide incentives for the partnerships to generate revenues and control costs.
The doctors in a partnership may be able to monitor one another and induce effort
provision without explicit incentive contracting. They then would be in a position to
share the aggregate risk to their incomes reasonably efficiently. Moreover, even when
some incentive contracting is needed within the group, there should be an opportunity
for risk sharing that reduces the costs of providing incentives.
The traditional method of dividing profits among law firm partners provides
another example. Large law firms involve specialists in many different facets of law,
and the possibilities for generating revenues differ among these areas. Traditionally,
partners divided firm profits not according to individual contribution, but rather
according to formula. The most standard one was for all partners of a given seniority
to be paid the same, with the share rising with seniority. Viewing the receipts from
different parts of the firm as separate risks, this pattern is consistent with the
recommendation of the theory of efficient risk sharing that any individual bear the
same share of each risk. Further, if all partners of a given seniority can be considered
to have approximately the same risk attitudes, whereas more senior people are less
risk averse (perhaps because their accumulated wealth is larger), then the pattern is
quite consistent with the recommendation that shares be positively related to risk
tolerances. In recent years, however, this pattern of dividing profits has begun to break
down under the demands from partners who produce more revenues that they be paid
more or else they will leave to join a competitor or start their own practices. 21

PAY EQUITY AND FAIRNESS

A major potential issue with any system of pay is its perceived fairness. It is often
claimed that people value fair treatment in and of itself, and that a pay system that is
perceived as unjust will then be dysfunctional. These sorts of preferences are not
usually part of the standard assumptions of economic analysis, but if they are actually
present, then managers cannot afford to ignore them. Moreover, economists have
recently argued that a taste for "justice" might have had evolutionary advantages
among early humans and their ancestors, so that we might have such preferences
genetically "hard wired" into us. 22 For example, people who had a sense of when they
had been mistreated and a taste for exacting revenge, even when it was costly of their
time and resources, would have developed reputations that would have prevented
further mistreatment and made them more successful in the competition to reproduce.
What fairness means in any context is a subtle matter. At a minimum, it could
mean that individuals are concerned about how they themselves are treated relative
to some reference group against whom they compare themselves, and that they feel
worse off if they perceive themselves to be mistreated. At a higher level, there could
be a sense of injustice when some group is seen as being badly treated, even when
the concerned individuals are not affected directly. Finally, there could be an extreme

21 See Ronald Gilson and Robert Mnookin, "Sharing Among Human Capitalists: An Economic
Inquiry into the Corporate Law Firm and How Partners Split Profits, " Stanford Law Review, 37 (1985),
3 1 3-92.
22 See Robert H. Frank, Passions Within Reason : The Strategic Role of the Emotions (New York:
W. W. Norton, 1 988).
 419
version that leads individuals to complai n when they believe they are being treated Compensation and
too well, and that they would prefer to be treated just the same as others are. Motivation
The last form is not unknown, but does seem limited i n i ts scope. People do
give to charities, but managers rarely hear requests from employees that their pay be
cut to permit i ncreasi ng others' wages. 23 S till, if employees are altruistic in thi s sense,
then pay differences should surely take account of this taste: Reducing i nequality will
reduce the total that needs to be paid to be competitive. The middle level (a general
concern with equity) is of little direct significance unless it leads to some manifested
desires. When it does, it becomes operati onally similar to the case in which people
are concerned about their treatment relative to others. A given level of pay may be
viewed as good or bad, acceptable or unacceptable, depending on the compensation
of others in the reference group, and as such may result i n different behavior.
When people do compare their pay, properly applied and understood incentive
pay i nevitably_ leads to i nvidi ous compari sons. U nder a well-designed scheme, equally
talented people who are moti vated to work equally hard wil l be paid differently as a
result of chance variati ons that affect their measured performance. U nderstanding
this, people might be less happy than they would be if their pay were more nearly
equalized. This is a constraint on the use of any sort of incentive pay. M oreover, if
people tend to misperceive their own value and systematically to overestimate their
contributions, as some evidence suggests, then feeli ngs of i nj ustice and mistreatment
associated with differential pay levels may be accentuated.
Of course, i n this latter case, those who end up being well paid will view this
as their rightful due. Equalizing pay then hurts them, even if they otherwi se have a
taste for equali ty, and managing this si tuati on becomes very complex. This may be a
reason to attempt to keep individual compensati on figures confidential.
Yet even when people are purely selfish, judgi ng their welfare solely on absolute
grounds and without regard to how others may be paid, differences in treatment can
be costly to the organization. Differences in realized pay point out what is possible
and invite attempts to i nfluence the distributi on of rewards by poli tical means. Resisti ng
these may be quite difficult, especially when luck rather than talent or effort plays a
part in generating the differences (as occurs under a well-functioning incentive
scheme). Avoidi ng the costs of these influence activities can be a strong constrai nt on
the use of i ncentive pay. In such a context, pay differentials may be limited, or directly
profitable i nvestment proj ects may not be accepted if they impose too great costs on
some groups or divi si ons. These policies may be interpreted even by those within the
organizati on as reflecti ng a concern for equity in treatment, whereas in fact their sole
role is to limi t costly influence acti vities.

23 It was news when, in 1 991, Los Angeles Lakers basketball star "Magic" Johnson took a voluntary
pay cut to allow the Lakers to hire another player (Terry Teagle) while staying within a league-imposed
salary limit.
 420
Employment:
Contracts,
Compensation ,
and Careers SUl\11\tAHY
The ways people are paid vary tremendously. This reflects the variety of different
obj ectives that com pensation policy must serve, as well as the different circum stances
in which organizations operate. Nevertheless, the role of com pensation in m otivating
em ployees is always a significant element in its design, and many aspects of pay
policies can be understood in term s of their incentive effects.
Incentive pay can be applied at the level of the individual em ployee or at the
group level. The purest forms of individual performance pay are piece rates and sales
com missions. Each of these has clear advantages but carry costs and disabilities. They
are strong motivators, they elicit self-selection, and they are easily understood. They
can, however, distort em ployees' allocation of their time and effort away from the
effi cient patter ns when there are concerns other than sheer volume of output or
sales-for exam ple, product quality, machine maintenance, cooperation with fellow
workers, custom er service, or information gathering. They also require a system in
which individual em ployees can work at different paces for extended periods of time
and that there be some means to allow the fi rm to limit production during periods of
lim ited dem and. Finally, pay that depends on output exposes em ployees to greater
risk than would a fi xed wage contract, with attendant costs. Nevertheless, where piece
rates and commissions are used, actual practice fi ts quite well with the recomm endations
of theory.
Individual incentive pay is used in other contexts as well, especially for managers.
(Executive and managerial incentives and com pensation are the subj ects of Chapter
13 . ) In all these contexts, there is an advantage in making use of em ployees' knowledge
about their own personal characteristics and the environm ents in which they work.
The problem is to motivate them to reveal this inform ation. Menus of contracts,
which allow em ployees to choose the schem e under which they are paid, and
management by objectives, under which em ployees help set their perform ance goals,
are two common methods of eliciting this information.
Often, however, the difficulties of defi ning and measuring good performance
and of determ ining who is responsible for results mean that it is infeasible to set up
explicit incentive plans linking pay and performance through sim ple form ulas. In this
case, there may still be implicit incen tive pay, with em ployees being rewarded after
the fact for good performance.
Either sort of incentive pay requires performance evaluations to determ ine what
pay is due. With explicit incentive schem es, the key issue is to link pay to the
appropriate measures of performance. Ideally, these should reflect the actual obj ectives
of the organization and should not be able to be manipulated by either side.
Performance evaluation in im plicit pay system s is full of ambiguity and open to
dispute. It is also apparently onerous for supervisors. Actual evaluations do not tend
to be very informative, and employees do not believe in their accuracy, fairness, or
usefulness. This helps account for the apparent low sensitivity of pay to performance
found in many organizations.
The provision of incentives can interact im portantly with job design and with
the discretion accorded em ployees in allocating their tim e between job responsibilities
and other in terests. Other th in gs being equal, incentive considerations argue for
groupin g tasks in ways that facilitate assigning individual responsibility for good or
bad performance. When groupings do not affect the quality of inform ation and
incentive pay is used, there arc ad vantages to grouping together tasks that arc sim ilar
in the ir diffi cu lty of performance evaluation. In th is way, relative ly in tense incentives
 42 1
can be used for all the well-monitored tasks without running afoul of the equal Compensation and
compensation principle, and weaker incentives can be used for the other tasks. Motivation
Employees who are given strong incentives to perform can be given broader discretion
in allocating thei r time between work and other concerns because they will take better
account of the opportunity costs of their time than will those for whom pay is only
weakly related to performance. In a complementary fashion, those whose jobs demand
more discretion should be given stronger performance incentives so that they will
more fully internalize the overall impacts of their individual decisions.
Incentives for groups of employees are very common, and a variety of different
schemes are used . These include both expl icit schemes li nking pay to productivity,
profits, or other measures of performance and implicit schemes, such as the systems
used in Japan, which award bonuses that seem related to profits but that are determined
at the discretion of the directors of the firm . Group incentives derive thei r effectiveness
from the ability of small groups of employees to monitor and enforce good behavior
among themselves. When employee groups can change production methods, the
responsiveness of groups to incentives can be greater than the responsiveness of
individuals, further contributing to the efficacy of the group incentive approach.
Fi nally, any scheme that differentiates among people in a group may raise
questions of fairness. Whether people actually have a taste for justice in pay, or care
about their relative standings, or have altruistic concerns about how others are treated,
or whether instead they are purely selfish but willing to practice influence activities
to affect the distribution of pay, well-designed policies will tend to limit the
discrepancies among rewards. Thus, an apparent concern with pay equality can arise
from a simple concern for efficiency .

• BIBLIOGRAPHIC NOTES
The basic references on the theoretical issues in performance pay are gi ven in
Chapters 5 through 8. A standard textbook treatment of compensation management
is provided by David Belcher and Thomas Atchison, while Edward Lawler gives
an up-to-date review of behavioral and managerial research in the area and its
implications. Economic discussions of a broad range of issues and puzzles in the
area of compensation and motivation are given by George Baker, Kevin M urphy,
and Michael Jensen and by Edward Lazear. The analysis of the frequency of
evaluations is drawn from Lazear. The treatments of job design and of responsibil ity
and discretion are based on work by Bengt Holmstrom and Paul M ilgrom .

• REFERENCES
Baker, G. , M. Jensen , and K. M urphy. "Compensation and Incentives: Practice
vs. Theory, " fournal of Finance, 43 ( 1 988), 593-6 1 6.
Belcher, D. , and T. Atchison . Compensation Adm inistration , 2nd edition
(Englewood Cliffs, NJ : Prentice Hall, 1987).
Holmstrom, B. , and P. Milgrom. "Multi-task Pri nciple Agent Analysis: Incentive
Contracts, Asset Ownership and Job Design, " Working paper 6, Stanford Institute
for Theoretical Economics ( 1 990).
Lawler, E. Strategic Pay: Aligning Organizational Strategies and Pay Systems
(San Francisco, CA: Jossey-Bass Publishers, 1990).
Lazear, E. "Labor Economics and the Psychology of Organizations," fournal of
Economic Perspectives, 5 (Spring 199 1 ), 89- 1 10 .
 422
Employment: EXERCISES
Contracts,
Compensation, Food for Thought
and Careers
1 . Adam Smith, in his Wealth of Nations (Book 5, Chapter 1 , Part 3, Article
2), argued that university teachers should not be paid salaries but rather that they
should have to rely on the fees they can collect from the students they teach. What
would be the advantages of this system? What difficulties do you see with this proposal
to pay piece rates to faculty?
2. In harvesting some crops, such as lettuce in the Imperial Valley of California,
the pickers are hired and paid piece rates as a team. The team then decides the
allocation of the receipts among group members. If the team members can monitor
one another, there may be no moral hazard problem within the team, and they ought
then to be able to allocate their income efficiently. At the same time, the farmers
may need to provide incentives to the team as a whole and so they use piece rates.
However, it seems that the team members in fact usually divide the receipts
equally. Moreover, in some other situations (such as celery picking i n Ventura County,
California, and asparagus harvesting in San Joaquin County), team piece rates are
used, but the farmer divides the pay among the workers, giving each an equal share.
This is not evidently consistent with efficient allocation of income risks in general.
What factors might account for this, and what might mitigate any apparent inefficiency?
3. Firms often have trouble with their performance pay systems when a
business downturn means that no incentive payments are made: Du Pont's experience
is a relatively common one. How would you account for this? What would you do
as an employer in such circumstances?
4. To avoid the problems created by supervisors' reluctance to assign low
evaluations, some firms (for example, Merck and Co. , Inc. , the very respected
pharmaceutical company) have experimented with requiring that a certain fraction of
the people being evaluated be assigned to each of the various possible grades or
performance-evaluation levels. What problems do you see with such a scheme?
5. General Motors' new Saturn automobile line was a completely new design
that was built by a specially-recruited workforce in a new factory constructed specifically
for the Saturn. To market the new car, which CM hoped would help it learn to
become competitive again on a world scale, it set up separate dealerships and
announced that the sales people in these organizations would not be paid on a
commission basis, as is common in the industry. Instead, they were to be salaried.
What might be the advantage of this pay policy?
 13
EXECUTIVE AND MANAGERIAL
COMPENSATION

n the one hand, they [executives] say that intense foreign competition
requires sacrifoce, restraint, and discipline. Yet they then turn around and
demonstrate none of those qualities by awarding themselves more personal
compensation for a year's effort than could be spent in several lifetimes.
Owen Bieber 1

It [CEO compensation] just seems to get more absurd each year. What is outrageous
one year becomes a standard for the next.
Edward E. Lawler, II/2

Make your top managers rich and they will make you rich.
Robert W. fohnson 3
This chapter deal s with special pro bl ems associated with compensati ng the
deci sion makers and senior leaders in firms. I n recent years, executive compensation
has become a focus of public concern and debate, especially in the United States,
where the compensation of CEOs in large firms has grown much faster than the gross
national product (GNP), corporate earni ngs, or the average worker' s pay. Each year,
Fortune, Forbes, Business Week, and other busi ness publications carry featured stories
reporting on the pay recei ved by the CEOs at leading firms, and even the dail y
newspapers cover these stories. Part of the general interest in executive earnings may
be pure jealousy about the spectacular sums received or curiosity about other people's
incomes, but there are substantive issues involved as well. Does senior executives' pay
motivate them to do a good job runni ng the companies entrusted to them? Or are
the huge amounts that they often receive in fact the resul t of managerial moral hazard,
with the CEOs lining thei r pockets at the expense of their firms' owners?

1 As quoted in John A. Byrne with Ronald Grover and Todd Vogel, "Is the Boss Getting Paid Too
Much?" Business Week {May l , 1 989), 49. B ieber heads the United Auto Workers' Union .
2 As quoted in John Byrne, "The Flap over Executive Pay," Business Week (May 6, 1 99 1 ), 90.
Lawler is an expert on compensation.
3 As quoted in Money Talks, Robert W. Kent ed. {New York: Facts on File Publications, 1 98 5),
320. Johnson was the CEO of Johnson and Johnson Corporation. 423
 424
Employment: PATTERNS AND TRENDS IN EXEClITI VE COMPENSATION
Contracts, Data on the pay of top-level corporate executives in the United States are publicly
Compensation, available through required reports to shareholders. These list the compensation of the
and Careers five most highly paid executives in each firm. The actual details of the contracts under
which this pay is generated are not public, however, and systematic information on
executives' pay in other countries and at lower levels in U . S. corporations is also not
so easily obtained. This differential availability of data has helped focus researchers'
attention on senior executive compensation, but the size and behavior of the amounts
involved are responsible for much of the public's interest.

CEO Com pensation in Large U.S. Firms

Each year, Business Week publishes a survey of the pay of the two top executives in
each of 365 of the largest publicly-held corporations in the United States. According
to the 1 99 1 survey (covering 1 990 earnings), the total compensation of the CEOs of
these firms grew 2 1 2 percent during the decade of the 1 980s. This was four times the
growth in pay of the average factory worker (whose pay rose 5 3 percent over the period)
and three times the income growth of the average engineer. In the same period,
earnings per share-the returns to the firms' owners-grew 78 percent. 4 The average
salary and bonus of the CEOs in the Business Week survey reached $ 1 . 2 million per
year in 1 990, and when long-term compensation through stock options and other
plans is included, average total compensation was $ 1 . 9 5 million. It would take the
average factory worker 8 5 years at 1 990 pay rates to earn this amount, and the average
engineer 4 5 years.
Other studies reached similar conclusions. In Forbes' survey of 1 990 compensa­
tion of the CEOs of 800 large American firms, average total compensation was $ 1 . 6 3 5
million, of which 43 percent came from various long-term incentive elements and
the rest from salary and annual bonuses. 5 In another study of 200 companies the
average CEO total compensation for 1 990 was $ 2. 8 million. 6
Hidden in these average compensation figures are some truly exceptional sums
of money. According to Forbes, Stephen M. Wolf received $ 1 8. 3 million in 1 990 as
CEO of UAL, the parent company of United Airlines. Of this figure, $ 1 . 1 2 million
was salary and current bonus, with the rest coming through realized gains on stock­
based, long-term incentive plans. John Sculley of Apple Computer earned $ 1 6. 7
million in 1 990, $2. 2 million of which was salary and annual bonus. Paul Fireman
of Reebok, the athletic shoe company, received $ 14. 8 million, all of it in current
salary and bonus. These figures are dwarfed, however, by the $78 million paid by
Time Warner to Stephen J. Ross, its chairman and co-CEO. Most of this was a $7 5-
million bonus awarded in connection with the merger of the publishing company
Time Inc. with Warner Com munications, the entertainment company Mr. Ross
headed, to create Time Warner. Even more spectacular was the $ 1 86 million received
by Donald A. Pels when LIN Broadcasting, the company he headed, was merged into
McCaw Cellular Com munications and he exercised his stock options. Overall, the 2 5
best-paid executives on the Forbes list averaged over $ 1 2 million a piece in 1 990.
Of course, with such huge sums included in the calculations, the (arithmetic)
average pay figure is larger than the median-the amount received by the CEO half­
way down the list. Moreover, the reported long-term incentive component of pay

4 Joh n A. Byrne, "The Flap Over Executive Pay, " Business Week (May 6, 199 1 ), 90-96.
5 "What 800 Compan ies Paid Their Bosses, " Forbes (May 27, 1 99 1 ), 237-89.
6 Testimony by Graef Crystal to the United States Senate Subcommittee on Oversight and
Co\'ernment Managem ent, May 1 5, 1 99 1 , as reported in Kevin C. Salwen , "Executive Pay l\lay Be Subject
to New Scrutiny," The Wall Street Journal (May 16, 1 99 1 ), A-3.
 42,5
Executive and
Managerial
Compensation
Forms of Senior Executive Compensation
Salary: An amount paid over the course of the year and fixed in advance.
Salary may be adjusted regularly based on length of service, competitive
conditions, the cost of living, performance, or other considerations.
Bonus: A nominally variable amount, often paid as a lump sum, at the
discretion of the firm's di rectors . May be tied to performance, either implicitly
or through an explicit formula . If so, performance is usually measured on a
short-term basis, such as the previous year's accounting profits or earnings
growth, or the extent to which these exceed targets.
Stock Options: Rights given to the executive to buy stock in the firm at a
prespecified price during a specific period of time. The price is usually at or
above the current price of the stock when the options are awarded , and the
time period is usually several years. No actual cash is received until the
executive actually buys the stock. Then the compensation is the difference
between the market price of the stock and the exercise price (the amount
paid by the executive).
(Restricted) Stock Awards: Shares in the firm given to the executive or sold
to him or her at a deep discount. Often the ability of the executive to sell
this stock is restricted, at least until certain conditions are met (for example,
meeting certain growth or profit goals or the executive's retirement). Number
of shares awarded may depend on past performance (a "performance share
plan").
Phantom Stock Plans: Units that correspond to common stock but carry no
ownership claims. These entitle the executive to receive the share price
appreciation and dividends that would have been received on actual stock.
Stock Appreciation Righ ts: Rights to collect the amount of any share price
appreciation on a specified amount of common stock over time.

tends to be highly variable because the entire amount is reported in the year in which
the gains are realized, even though they may be the result of several years' performance.
For example, Michael D. Eisner, the head of Walt Disney Company, received $ 3 2 . 6
million in 1 988 from stock options that became valuable because the company had
become successful under his leadership in the preceding three years . In 1 989 and
1 990 he received no long-term compensation. His average annual pay was j ust short
of $20 million per year over the period , but over $40 mill ion was recorded in one
year. Because the news stories tend to focus on the highest-paid executives in any
year, this method of recording long-term compensation tends to bias the apparent
income levels upward relative to the corresponding steady flows. Still, there is no
question that these CEOs receive significant amounts of money. Table 1 3 . 1 lists the
ten best-paid CEOs in the United States from 1 986 to 1 990.

Patterns and Comparisons

The pay patterns and levels for CEOs in the largest U . S . corporations differ radically
from those of the CEOs of comparable firms in other countries and the heads of
smaller (but still substantial) U . S . firms .
l ,TEHNATIONAL DIFFERENCES IN CEO COl\1PENSATION The average CEO of a very
large Japanese firm (the equivalent of $30 billion in sales) earns about 1 7 times what
 426
Employment: Table 1 3. 1 The Ten Best-Paid U . S . CEOs: 1986-1990
Contracts,
Executive and Com pany
Compensation,
and Careers
5-Y ear Compensation
Steven J. Ross, Time Warner $ 1 37, 2 54, 000
Charles Lazarus, Toys 'R' Us 9 1 , 09 5 , 000
Michael D. E isner, Walt Disney 7 1 , 069, 000
Paul B. Fireman, Reebok International 69, 3 56, 000
Lee A. Iacocca , Ch rysler 49, 1 97, 000
Martin S. Davis, Paramount Commun ications 37, 380, 000
Dean L. Buntrock, Waste Management 3 3 , 00 3 , 000
John Sculley, Apple Computer 32 , 3 1 0, 000
Saul P. Steinberg, Reliance Group 32, 1 8 3 , 000
Richard A. Manoogian , Masco 27, 802, 000

Source: Forbes (May 27, 1 99 1 ), 2 1 8.

the average Japanese worker does. For com parable firms in France and Germany, the
figure is about 24 tim es. I n the U nited States, it is 1 09. 7 An extrem e exam ple comes
from the oil i ndus try. I n 1987 Exxon and Royal Dutch/S hell had about equal sales
and equal profits. B oth fi rms are am ong the leading internati onal petroleum com panies
and are i nvolved in a similar range of activities. The head of U . S .-based Exxon was
paid $5. 5 milli on; the CEO of its European com petitor received $500,000. Another
exam ple comes from the Sony Corporation. As with m ost J apanese firms, S ony's
board of directors is made up almost exclusively of executives i n the firm. I n S ony's
case, 34 of the board mem bers and statutory auditors in 1989 were full-time em pl oyees
of S ony or its subsidiaries. As a group, they were pai d $8. 2 million in salary and
bonuses (calculated at 1989 exchange rates), an average of about $242, 000 a piece.
Had the entire $8. 2 mi llion instead been paid to the S ony CEO, he still would not
have been am ong the ten top-paid American executives.
CEO COMPENSATION IN SMALLER FI RMS In smaller U.S. firms, CEOs are still paid
m ore than their Japanese or European counterparts, but the differences are much less
extrem e. According to surveys by Towers, Perrin, Forster & Crosby, a com pensation
consulting firm, the CEO of the average U. S. firm with sales of $250 milli on receives
about $600, 000 a year (including an es timated value of various benefits and perquisites).
The head of a comparably sized Japanese or European firm receives about half to two
thirds as much. Another Towers Perrin study i ndicated that the average Canadian
CEO receives about 65 percent of the pay of a typical U. S . chief executive.
THE ROLE OF LONG-TERM I NCENTl\'E PLANS The chief difference between the
com pensati on of the heads of maj or U.S. firms and thei r counterparts i n large forei gn
firms and in smaller U.S. firms is the element of long- term incentive com pensation
tied to stock prices. Such plans became alm ost universal in large U.S. firms in the
1980s but are rare abroad and i n smaller com panies. 8 Executives in smaller U . S.
firms us ually receive a salary, plus perhaps a bonus that may be based on annual
accounti ng meas ures such as current profits or return on inves tm ent.9 The sam e is

7 C stal, op. cit.

ry
8 The chief exceptions are entrepreneurial firms, where the founder is still CEO and retains a
significant ownership share.
9 New firms, especially in high-technology areas, such as elect ronics and biotechnology, are often
an exception. In such "start-up" firms, executives typically receive ownership claims that will become
extremely valuable if the company succeeds and goes public, selling its stock to outside investors.
 427
true in continental Europe and Japan (where the executives get bonuses in the same Executive and
spirit as all other company employees); the various long-term incentive-pay schemes Managerial
afforded to heads of large U. S. corpora tions are essentiall y absent. This is especia lly Compensation
the case in Fra nce, where a number of the largest firms are sta te- owned and their
executives are (a t least by U.S. standard s) paid quite modest amounts. In the U nited
Kingdom, however, there has been more of a move towards long-term incentive pay,
a nd large performance- ba sed incentives are common in H ong Kong. 1 0
Beca use it often seems that large U. S. firms add various long-term incentive
pay components to their CEOs' compensation packages without obviously decrea sing
\ a ny of the other elements, knowledgea ble observers of executive compensation consider
these long-term programs the source of the rela tive jump of CEO pay in the 1980s.
In 1980 the rela tive pay of CEOs, engineers, and factory workers were all in a bout
the same relationship to one another a s they had been 20 years before: All three
figures had in_crea sed a bout three and a half times over the two decades. In the 1980s
the pay of the engineers a nd factory workers grew at essentially the same ra te, whereas
that of the CEOs grew four times as fast. 1 1 Direct performa nce pay now accounts for
the bulk of compensation for CEOs of large U . S . companies, with annual bonuses
contributing 25 percent a nd long-term incentives accounting for 36 percent of total
compensation. 1 2

Middle- Level Executives

Much less systematic information exists a bout the pay of middle-level executives. All
receive salary, and some form of incentive pay is common in the U nited S ta tes,
whether based on individual, divisional, or corporate performance. During the 1980s
salary increa ses in U . S. firms generally seemed to lag behind inflation, but increased
use of incentive pay helped total managerial compensa tion keep up with the cost of
living. Increasingly, executives are rewarded for measured performance in the units
for which they are responsible. S till, the importance of incentive pay in total
com pensation seems to be quite a bit lower for middle-level managers tha n for those
at the top of the hierarchy.
DOWN-SIZING AND THE DETERMINANTS OF BASE PAY A recent key factor in managerial
compensation in the U nited S ta tes has been a trend to redefine the criteria for
determining base pa y. This ha s accompa nied the reduction that ma ny firms have
sought in the number of layers of management in their organizational hierarchies a nd
the size of their white-collar staffs.
Given the difficulties of measuring performance, a significant portion of pay for
middle-level managers is often tied to the breadth a nd significa nce of their responsibili­
ties. These in turn are often measured in part by the number of people reporting to
the manager. This system leads to perverse incentives. The only way for a manager
to get a substantial raise is to increase his or her responsibilities. One obvious way to
do this (apart from winning a promotion to a higher level) is to increase the number
of subordina tes, particularly the number of higher-level ones. This creates incentives
to add staff.
These incentives might be nullified by the resistance of higher-level executives
who have profit-a nd-loss responsibility and are paid for orga nizational performance.
These people have a hard time telling whether extra staff in some function or offi ce

10 See Shawn Tully, "American Bosses Are Overpaid. . . ," Fortune (November 7, 1 988), 1 2 1-36.
11 In after-tax terms, the disparity may have been even larger, because the U. S. personal income
tax rate structure became less progressive over the decade.
12 The Wall Street Journal (April 1 7, 1 99 1 ), R-5 (Special Report on Executive Compensation).
 428
Employment: is actually worthwhile, however, especially when output is hard to measure and
Contracts, everyone is clearly working hard. Here a second perverse effect enters: Each extra staff
Compensation, person creates work that justifies hiring still others! A manager whose staff is
and Careers overwhelmed by the demands of writing memos, preparing reports, replying to
proposals, and so on, may win approval to hire another staff person to help. This
person in tum will generate more memos, reports, and proposals. Then other offices
can no longer keep up, and they need to add staff, who in tum generate even more
work. Thus, massive staffs come into being, with everyone working very hard for long
hours, busy keeping one another busy.
To counter these tendencies toward bureaucratic expansion, many firms are
attempting to divorce managers' base-pay levels from the sizes of staffs reporting to
them. They are also seeking to increase the role of incentive pay tied to measured
performance, for example through management-by-objectives plans, where control of
staff costs are explicit goals figuring into the determination of the bonus.
PROMOTION TOURNAMENTS AND THE ROLE OF INCENTIVE PAY Perhaps the major
incentive for most middle-level managers, however, still comes through the possibility
of promotion. Firms' removal of layers of management has reduced the frequency
with which promotions can be gained and the number of promotions that can be
won. As we noted in Chapter 1 1 , some of the incentives provided at each stage in
elimination tournaments come from the opportunity to continue to be considered for
further promotions. The leveling of organizational hierarchies has also probably
increased the number of people at the lower rounds and, correspondingly, decreased
the chances of winning promotion at these rounds. Both of these effects might well
have adversely affected the incentives for middle-level managers. Any such negative
incentive effects may have been offset by the impact of the recent increases in CEO
compensation, however. Higher pay at the top of the hierarchy has increased the
prize attached to winning in the last rounds and becoming an extremely well-paid
senior executive, and thus may have increased middle-level managers' incentives to
keep competing for promotions.
At the top of the hierarchy, there are no more promotions to win. This means
that if there are to be any financial incentives for CEOs, they must come from direct
performance pay. Thus, the theory also suggests that explicit incentive pay should
become more important at the top of the hierarchy, as is observed in practice. This
conclusion is also consistent with the recommendations of the theory of performance
contracting, which indicates that the intensity of incentives should be higher when
the marginal productivity of effort is higher and when the potential responsiveness to
incentives is greater. The quality of the job done by the top executives typically has
a greater impact on overall organizational performance than do the efforts of people
in lower-level positions. At the same time, the broader discretion and responsibilities
accorded the top-level people mean that they have more ways in which they can be
responsive to increased incentives. Both of these factors thus suggest that direct
performance incentives should be stronger and performance pay a more important
part of total compensation at higher levels in the hierarchy than for middle-level
executives.
THE EFFICACY OF INCENTl\'E PAY FOR MIDDLE MANAGERS Although performance pay
is (and should be) less significant for middle-level managers than for senior executives,
its importance is increasing. This raises the issue of whether it is effective in improving
these managers' performance.
The publicly available evidence on the efficacy of incentive plans aimed at middle
managers is very limited, because employees' individual performance evaluations are
extremely confidential. What evidence does exist, however, indicates that incentives
 429
do work at these levels. For example, a study of 92 middle- and upper-level managers Executive and
in an un iden tified mid western U. S. man ufacturin g firm found that managers' hon uses Managerial
were sen sitive to their performance, with the sensitivity being higher for those at Compensation
higher levels in the hierarchy, tho se working in the head office rather than a plan t,
and those with shorter job ten ure. More to the poin t, greater sen sitivity of the bon us
to performan ce tended to increase future performance to a statistically sign ifican t
exten t. This effect remained even after con trollin g for current performance and,
indirectly thereby, for such factors as differin g ability. 1 3

MOTIVATING RISK TAKING

Large public corporation s, with their large n umbers of shareholders, can often spread
the risks they face so widely that they ought to be virtually risk n eutral. 1 4 Then ,
assuming they can raise an y n eeded in vestmen t fund s, they should be willin g to make
large in vestmen ts and risk large lo sses if the expected returns on the in vestmen t, taking
into accoun t the po ssible gain s and losses, are positive.
Some compan ies do in fact take immen se risks. Together, oil firms in the late
1970s and early 1980s paid billions of dollars to the U . S . federal govern men t to
purchase drilling rights in the Baltimore Can yon, an area beneath the Atlan tic Ocean
off the eastern coast of the U nited S tates. They then spen t hundreds of millions more
dollars for the actual drillin g in the ocean floor. No commercializable deposits of
hydrocarbon s were ever found, however, and even tually the oil compan ies let their
rights expire without developin g any of the property. Risks of this magn itude n eed to
be und ertaken if the oil producers are to make gian t finds like those in the North S ea
or Alaska's Prudhoe Bay.
Very large in vestmen ts were also made in the early 1980s in biotechnology. By
the end of the 1980s, these in vestmen ts had yielded few pro fitable prod ucts, leadin g
many of the early start- up compan ies to go out o f busin ess and others to reduce their
scale of o peration s. In con trast, a small number of the biotech firms had become
remarkably successful. In 1990, for example, Hoffman n-La Roche, the gian t S witzer­
land-based pharmaceutical firm, paid $2.1 billion for a 60- percen t share of Genen tech,
which had sales of $383 million and profits of $44 million. The price ind icates the
buyer's confidence in Genen tech's future growth. More recen tly, the Lockheed
Corporation and its allies, the Boein g Company, Gen eral Dynamics Corporation ,
and Un ited Technologies Corporation, spen t approximately $1 billion of their own
funds developin g the pro to type of the F22 fighter j et that, in 1991, won a U .S.
Departmen t of Defen se con test that could lead to between $60 and $100 billion in
sales over the comin g decade. Another group of firms, led by McDonn ell-Douglas,
spen t a similar amoun t develo pin g the losin g altern ative design , the F23 , which will
likely never be produced . As a final example, H aloid Corporation , a small pho tographic
pa per man ufacturer, spen t over a dozen years tryin g to develop paten ts that were
already a decade old when it started before it finally laun ched the plain- paper co pier
in 1959 (and changed its corporate n ame to Xerox).

The Puzzle
As these examples indicate, some compan ies have managed to en courage risk takin g
on a major scale. Nevertheless, observers of the business world often commen t

n Lawrence M. Kahn and Peter D. Sherer, "Contingent Pay and Managerial Performance,"
Industrial and Labor Rela tions Review, 43 (February 1990), 107S- 1 20S.
1 4 Strictly speaking, this applies only to "unsystematic risks" that are idiosyncratic to the firm itself
and can be effectively diversified away by stockholders adjusting their portfolios of investments. (See Chapter
14 for a more detailed treatment of portfolio choice and securities pricing. ) Companies need not be risk
 430
Employment: criticall y on the reluctance of managers in some companies to undertake profi table
Contracts, projects where large losses are possible. Assuming that this is true, the issue is: Why
Compensation, should managers exhibit risk aversion in the investme nt decisions they take on behalf
and Careers of their firms? After all, it is the stockholders' money that is being put at risk, not
their own, and the stockho lders ought to be approximately risk neutral. Thus, they

might occur if things work out well or badly-without regard to the risk involved. If
should want decisions to be based solely on expected returns-the mean of what

managers shy away from profitable but risky investments, then the interests of the
stockholders are not being served and economic efficiency is not being realized.
One reason managers might be risk averse in their investme nt decisions is that
their pay might be closely tied through ince ntive contracts to the results of the
investme nts they make. They would then face the investment risks themselves, and
unlike the stockholders they would not easily be able to diversify this risk away through
their portfolio decisions. Yet, empirically, this does no t seem to be the right answer.
Relatively few managers have such contracts; further, managerial risk aversio n and
timidity seem more prevalent in large, bureaucratic organizations, where explicit
incentive co ntracts are especially rare. Even in the absence of explicit linking of pay
and investment performance, however, the managers' fu ture incomes might still
effectively be tied to the outcomes of the investments they undertake. This in turn
can induce risk aversio n in their decisio ns about corporate investments.
�lanagerial I nvestment Decisions and Human Ca p ital Risk
In western economies, successful managers are able to move from fi rm to fi rm,
commanding a salary on the basis of their past performance. Good past performance
that suggests that a manager is tale nted, hard working, or lucky therefore has a positive
market value, which is part of the manager's human capital. For many of these
people, their human capital is by far the most valuable asset they own. It is also an
asset that is esse ntiall y no ntransferable, so there is little hope of diversifying away risks
that attach to it.
When managers undertake a risky investment, they put their human capital at
risk. The success or failure of the investme nt is likely to reflect on the managers who
proposed and oversaw it, affecting (current and po tential) employers' estimates of their
ability and thus their value in identifying and implementing fu ture investments. A
good outcome may help their reputatio ns and increase their future earning potential,

lack thereof) rather than entirely to random bad luck. If the project performs badly,
but a bad outcome will al so be attributed at least partly to the managers' tale nts (or

fu ture salaries, jobs, an d promotions may be e ndangered. Thus, even if pay is not
directly and explicitly l inked to the performance of the investments for which a
manager is responsible, in fact the market mechanism may create such a linkage.
Without some incentive for risk taking, managers may se nsibly be reluctant to
put the value of their human capital at risk: Career concerns will make managers risk
averse about corporate investments. The key pro blem is that the returns to the
investments are no t solely the financial ones accruing to the risk- neutral mvners of
the firm but also the effects on the risk-averse managers' human capital. The n it may
be necessary to devise special ince ntive plans to induce desirable risk taki ng.
TH E l \tPOHTA:\C E OF THE .H-\HKET FO H E.\ECUTl \ ' E TALE:\T As our analysis in Chapter

man agers' human capital. If the managers' future pay were fully insulated from the
1 1 suggests, the ideal solutio n would be for the fi rm to insure the value of the

neutral in their evaluation of the risk of economic downturns, oil price increases, or other risks that ha,·e
an economy-wide character .
 43 1
success or fail ure of the investment projects they recommend and manage, they would Executive and
have no reason to be concerned about risk. Managerial
I n Japan, where successful managers typically spend their entire careers employed Compensation
by a single company and where salaries vary much less from job to job than is common
in North America or Europe, the risk to a manager's human-capital value from
recommending or undertaking a risky project is relatively lower. In essence, the pay
promised by the firm is the only relevant consideration, and the firm can insulate the
manager from the risk inherent in the investments that are undertaken. Moreover,
the practice of obtaining consensus on major decisions spreads responsibility and
reduces the risk to individual managers. For such firms, no special incentive plan
may be required to encourage risky undertakings.
Western firms operate in a different managerial and market environment that
makes the Japanese solutions less effective. The active executive recruitment market
means that managers whose investments have performed well and who are thus seen
as talented m ust be compensated accordingly or else they will be hired away. This
means that firms operating in this environment cannot provide the full insurance that
would otherwise be optimal and that Japanese firms are better able to offer.

Inducing Risk Taking

Successful firms have a variety of devices for encouraging risk taking in this context.
Generally, they follow the pattern suggested by our analysis in Chapter 1 1 of optimal
policy when labor mobility makes full income insurance infeasible: Give as much
insurance as possible, which basically means reward people for success and try not to
have them bear the costs of failures. Note that charging them for failures would
accentuate the risk they already face and make them even more reluctant to recommend
projects.
Giving stock options to managers has this effect because successes are rewarded
and there is no direct cost to failures. The asymmetry in financial payoffs helps to
offset the greater weight that risk-averse managers give to the decrease in the value of
their human capital that follows a failure than they give to the increase that follows
success and that thus leads them not to want to take risks.
Similarly, many firms in the oil industry give shares in the income generated
by successful wells to the petroleum engineers and geologists responsible for their
development without charging them for unsuccessful ones. This financial incentive
offsets the human-capital risk involved in proposing and developing a project. These
companies also consciously aim to attract "high rollers"-people who are less risk
averse than the average manager. Many also use a system of approval for risky
investrµents. This has the same effect as does the consensus system in Japan. Decisions
to drill in difficult terrain, such as on the ocean floor or in the Arctic, are recommended
by teams of geologists and explorationists and then must be approved by a management
team, relieving the individual experts at each level of some responsibility for the
decision. Further, the budget for bidding on oil-bearing lands is typically proposed
by management, but it must be approved by the board of directors.
Similar systems of shared blame or responsibility are used in other industries,
where major expenditures must be approved by higher-level managers than those who
identify and manage the projects and where truly large risks must be approved by the
board of directors.

Paying for Investment Pro posals

If these review processes are effective, that is, if they reveal a large part of the basis
of the expert's recommendation, then one measure of the employee's performance is
 432
Employment: the number of his or her recommendations that are accepted. How many attractive
Contracts, prospects for drilling did the geologist find? How many attractive biotechnology projects
Compensation, did the scientist propose and successfully defend?
and Careers According to the informativeness principle, if the final profitability of an
investment is affected by random factors that could not be known when the investment
decision was being made, then compensation ought not to be based solely on the
outcomes of the investment. Instead, employees ought to be rewarded just for making
and successfully defending suggestions for investment projects. Moreover, rewarding
employees for getting projects adopted (rather than on the basis of the projects' success
or failure) offsets the tendency to conservatism by providing a return to counter the
risk to a person's reputation that is necessarily involved when recommending a project.
Of course, paying for accepted projects and giving employees a share of the
"upside" on the projects in which they are involved, with no punishment for failures,
provide an incentive for successfully promoting projects. This may lead to employees'
recommending dubious projects, being overly-optimistic in evaluating prospects, and
withholding negative information. For these reasons, a company that adopts such
policies will need to set higher standards for project approval than would otherwise
be necessary and will need to use more independent auditing of information and other
such devices to protect itself. Having different people or groups within the firm
responsible for proposing projects and for approving and monitoring them facilitates
this.

DEFERRED COMPENSATION
We have already seen instances of deferred compensation that might apply to any of
the employees of a firm. For example, the explanations of the positive relationship
between pay and age or experience in terms of incentives for effort and discouraging
turnover imply that pay is deferred from early in an employee's tenure with a firm
until later. Deferred compensation should be especially important, however, in
rewarding managers.
The simple compensation formulas involved in piece rates and linear commission
schemes are appropriate for motivating steady performance in routine work, where
the variations in outcomes that might occur as a result of any single decision or action
by the employee are small and performance information on which pay can be based
is available quickly. At the highest levels of the firm, where decisions must be made
about massive, indivisible, and nonrecurring issues such as acquisitions and divestitures,
major investments, and corporate reorganizations, the arguments favoring pay ac­
cording to simple linear compensation formulas do not apply. Moreover, reasonable
measures of performance emerge only over a period of many years for these kinds of
decisions, when the profitability of major investments and strategic changes becomes
clearer. These information lags make it difficult to compensate executives for taking
a long-term view and make it necessary to base compensation on the long-term
performance of the firm.
Deferred compensation formulas, when properly designed, can help to achieve
that objective. For example, deferred payments that are tied to the firm's future
performance will help encourage managers to take a long-term view rather than to
concentrate excessively on short-term results. Payment in stock that cannot be sold
immediately or in options that cannot be exercised before a certain date would be
examples. This is the essence of the plan adopted at Salomon Brothers which was
described in Chapter 1.
Many actual compensation programs cannot be described in such simple terms,
however. Often, managers will have moved on to other jobs long before the full effects
of their decisio ns become known, and their compensation in the new assignments is Executive and
largely or completely u naffected by the unfolding results of their previous efforts. Managerial
Deferred compensation schemes also have roles other than promoting a lo ng­ Compensation
term view. Compensatio n is deferred to pro vide incentives through bonding or to
discourage turnover. U nder a system of progressive income taxatio n, deferred
compensatio n may also serve to postpo ne income not needed today into a period of
time after retirement, when income and taxes may be lower. Other tax co nsideratio ns
are important in compensation design, but becau se the tax laws vary from cou ntry to
cou ntry and even within cou ntries over time, a proper treatment of this su bj ect would
be quite lengthy. 15

Commitment Problems
A com pl ication i n using deferred compensatio n for incentive purposes is that the
practice invites renegotiatio n. Su ppose an executive has taken an action whose results
will not be evident for some time and that the incentive system deferred part of the
executive's compensatio n u ntil the results of the actio n become known. O nce the
action has been taken, however, there is no further reaso n to su bj ect the executive to
the income risk inherent in the incentive contract, especially if the execu tive is retiring
or leaving the firm. In fact, by renegotiating to eliminate the risk, the parties can
increase total value by the amount of the risk premium the executive assigns to the
income risk. Of course, if this incentive is foreseen and the contract is expected to be
renegotiated, then the original co ntract will be nothing but the starting point for the
renegotiation. In this context it may affect the spl it between the executive and the
firm of whatever surplus is available, but it cannot normally be expected to provide
appropriate incentives. Thu s, providing incentives through deferred compensation
requires the ability to make commitments and stick to them.

PERFORMANCE PAv FOR CEOs?

The key issue in managerial compensation is the pay of senior executives, especially
CEOs. Some CEOs are paid immense amou nts, but even so, their earnings are
usually a trivial fractio n of the earnings of the firms they head. From the perspective
of total value creatio n, the important question is not how much they are paid; that
amount is just a transfer and, co nsequently, without significance for effi ciency. Rather,
the more important issue is whether CEOs' pay provides them with the appro priate
incentives to maximize value. The sheer magnitude of pay might have some effect
here. For example, various observers have asserted that the growing gap between the
earnings of senior executives and the pay of average working peo pl e in the U nited
States is destructive of employee morale and of the trust that is needed to make
businesses run effi ciently. If this is true, the highly pu blicized exampl es of CEOs' pay
skyrocketing while the companies they head flou nder, losing mo ney and laying off
people, certainl y are destructive of value. Yet the overwhelming evidence is that, o n
average, CEOs' pay and wealth are responsive to company performance. The issue is
whether they are appropriatel y so.

Setting CEO Pay

The compensation of the senior executive officers of a corporation is set by the firm' s
board of directors. S hareholders have no direct say in the matter, and the U . S .

1 5 Readers interested i n a study that treats tax and incentive issues together are referred to Myron
Scholes and Mark Wolfson , Taxes a nd Business Stra tegy: A Global Planning Approach (Englewood Cliffs,
NJ: Prentice Hall, 199 1 ).
 434
Employment: Securities and Exchange Commission's policies have generally allowed companies to
Contracts, keep compensation issues fr om being the subj ect of shareholder votes. The directors
Compensation, usually appoint a compensation committee charged with recommending executive
and Careers compensation to the board as a whole. Often, but not always, the compensation
committee is composed exclusively of outside directors, who are not offi cers of the
corporation. Nevertheless, the CEO is usually a member of the committee ex officio
(by virtue of his or her office). Corporate officers are not supposed to have a role in
the compensation committee's nor the board's deliberations and decisions regarding
their own pay. Instead, the CEO makes recommendations to the compensation
committee regarding the other officers' compensation and then is supposed to leave
the room while the committee decides on these and on what the CEO should be paid
as well.
Frequently, compensation consultants are hired (often by the CEO) to advise
the board on compensation. These consultants are often firms-such as Towers
Perrin-that specialize i n this line of business, although some general consulting
firms also have a compensation practice. The consultants have access to information
on compensation forms and levels at other firms, and they often have statistical models
relating pay to performance, firm size, and other possibly relevant variables. The
information they provide is used in setti ng pay and in designing compensation
packages.

The Debate on Executive Com pensation

Many knowledgeable critics of executive compensation see this mechanism as having
been captured by the executives themselves for their own benefits. The board of
directors is supposed to represent the i nterests of the stockholders, but there are real
questions about whether and how well they can do this when the shareholders' interests
are in conflict with the CEO's.
Even the outside directors (members of the board who are not employees of the
corporation) are likely to have close ties with the officers. They are effectively
nominated by the CEO, they must rely on the executives for most of the information
they receive, and they need good relationships with the officers if they are to function
well in guiding corporate policy. Often, directors share similar backgrounds and
interests with the firm's executives. Frequently, they themselves are senior executives
in other firms. M oreover, outside directors who are not CEOs of other firms may
well derive a significant portion of their incomes from their directorships.
Accordi ng to K orn/Ferry International, an executive recruiting firm, in 1990
the average outside director of a maj or U . S. firm was paid over $ 32,000 as a retainer
and for attending board meetings, and many firms were even more generous. For
example, Pepsico paid each of its outside directors $78,000 in 1990. 16 M ost firms paid
extra for serving on board committees, and many firms made stock grants to their
di rectors, gave them insurance and retirement benefits, and even gave them free
samples of the company's product, for example, a new car every six months at General
M otors. The results could be substantial: The dean of Northwestern U niversity's
Kellogg S chool of Management was reported to earn at least a third more in direct
compensation from his service on several boards than he did from his regular j ob. 17
In this context, critics wonder how much weight compensation committees, even

16 Judith Dobrzynski , "Directors' Pay Is Becoming An Issue Too, " Business Week (May 6, 1 99 1 ),
94.
17 Joann S. Lublin, "While Outside Directors' Pay Increases, Independence From Managers May
Fade, " The Wall Street foumal (April 22, 1 99 1 ), B- 1 .
 435
when composed of outsiders, gi ve to stockholders' i nterests compared to those of the Executive and
executives whose pay they are setti ng. Managerial
O n the other side of the argument are some who claim that CEO s "are worth Compensation
every nickel they get. " 1 8 These observers claim that CEOs can have a tremend ous
effect on the performance of thei r firms and that they i n fact collect a very small part
of the gai ns that are generated when they make the firms perform well for stockholders. 19
These observers worry that corporate CEOs are in fact not adequately compensated,
both i n absolute terms (compared with alternati ve opportunities i n such fi elds as
i nvestment banki ng or entrepreneurshi p) and, more i mportantly, i n terms of the
explicit i ncentives they recei ve to improve corporate performance. 20
Research results i ndicate that the stock markets respond very positively to the
adopti on of both short-term and long-term i ncentive programs for seni or executi ves. 21
Those who d oubt that CEO i ncentives are adequate point to the fact that sophisticated,
knowledgeable i nvestors bid up the stock of firms that strengthen i ncenti ves as showi ng
that the seni or executi ves' i ncentives are important for the performance of the firm.
Thi s posi ti on i s rei nforced by the fact that, when ownershi p becomes concentrated i n
the hands of professi onal owners ru nning leveraged-buyout firms, the new owners ti e
executive compensati on very closely to results by gi ving management si gnificant
ownership shares. The observers then argue that management i ncentives actually
provided i n most large corporati ons are too muted to be optimal. E valuating these
opposi ng posi tions requires consideri ng what sort of behavi or needs to be i nduced
from seni or executives and what sort of temptati ons they might be su bj ect to, as well
as exami ning the evidence on the actual strength of i ncenti ves and thei r i mpact.

The Tasks and Tem p tations Facin g Senior Executives

Seni or executives have remarkably broad responsi bilities and great lati tude and
discreti on i n determi ni ng their behavi or and the objectives and policies their firms
will pursue. A system that seeks to provide i ncentives to senior executives therefore
may need to be concerned not just with any single dimension of thei r behavi or, such
as how hard they work, but also with how they allocate their ti me and attenti on
among different concerns. What will executive deci sions accentuate? M arket share?
Growth of sales, assets, or employment? Stability of employment and earni ngs?
Accounti ng rates of return on assets or stockholders' equity? Long-term stock price
appreciation? Or any of a multitude of other considerati ons? What trade- offs will be
made among these? Will concerns with the environment, with the communities i n

18
Kevin J. Murphy, "Top Executives Are Worth Every Nickel They Get," Harvard Business Review
(March-April 1 986), 1 25-32.
19 On average, even the 25 highest-paid executives in Forbes' survey of 1 990 pay took away only
$2. 26 for every $ 1 ,000 in revenues their companies received, and $ 30 . 20 on every $ 1 , 000 in profits. Recall
that these were the highest-paid executives for the year and that the compensation of the highest-paid
individuals in any year tends to be inflated by recognition of the gains taken on long-term compensation .
20 See, for example, the comments of Michael C. Jensen in "A Roundtable Discussion of
Management Compensation, " Midla nd Corporate Finance Journal 2 (Winter 1985), 32; and Michael C.
Jensen and Kevin J. Murphy, "CEO Incentives-It's Not How Much You Pay, But How," Ha rvard
Business Review (May-June 1 990), 1 38-5 3 .
2 1 Hassan Tehranian and James Waegelein ("Market Reaction t o Short-Term Executive Compensa­
tion Plan Adoptions," Jou rnal of Accoun ting and Economics, 7 (April 1 985), 1 3 1 -44] found that stock
prices jump 1 1 percent on announcement of the company's first executive compensation plan tying pay to
accounting returns, and James Brickley, Sanjai Bhagat, and Ronald C. Lease ("The Impact of Long-Term
Managerial Compensation Plans on Shareholder Wealth, " Jou rnal ofAccoun ting a nd Economics, 7 (April
1985), 1 5 1-74] found that adoption of a long-term incentive plan gave stockholders a direct 2 percent
return.
 436
Employment: which the firm operates, or with national policy absorb the executives' attention and
Contracts, influence corporate policies? The appropriate choices on any of these matters are not
Compensation, immediately obvious, even if one believes that the firm ought to be run in the interests
and Careers of the stockholders.
Even when the desired behavior may be clear, there are issues about whether
it will be pursued. Will hiring and promotion decisions be driven by efficiency, or
will favoritism rule? Will excessive costs and unprofitable activities be eliminated,
even though doing so may require tough decisions that make the CEO appear hard
or even heartless? Will risks be taken when the rewards to the corporation make them
worthwhile, and only then? Will the executives deny themselves perks that cannot be
justified on economic grounds? Will they forego making investments and acquisitions
that increase their empires at the expense of profitability? Will they meet value­
maximizing takeover attempts with honest efforts to get the best terms possible, even
though the change of control might cost them their jobs, or will they oppose the
change out of self-interest, seeking to entrench themselves in their jobs?
With so many relevant dimensions, there is an obvious difficulty in attempting
to shape executives' behavior: There simply would not seem to be enough different
instruments available for the task.

Value Maximization and I ncentives

The principle of value maximization actually makes the task of motivating senior
executives appear relatively simple: Executives should be guided to maximize value.
Furthermore, under a frequently made assumption, determining whether they are
doing this is also straightforward.
Suppose that the capital markets efficiently aggregate information in the sense
that securities prices fully reflect everyone's information to the extent that it is relevant
to forecasting future returns. (This presumption is called the strong form efficient
market hypothesis and is discussed more fully in Chapter 14. ) Weaker forms of this
hypothesis, which posit that all public information is reflected in prices, have
considerable formal empirical support: The prices at which shares trade are determined
primarily by the best estimates of skilled, knowledgeable investors who specialize in
evaluating firms' plans and prospects and making recommendations to investors.
Under this hypothesis, the market's reaction to various moves that the firm makes is
the best available measure of their effect in creating economic value: If the market
sees a new policy or strategy as increasing total value, then it will bid up the market
value of the firm correspondingly. Under these conditions, executive incentives should
be structured to encourage maximizing the market value of the firm.
Assuming such strongly efficient markets and assuming that the conditions of
the principle of value maximization hold, the prescription to maximize the market
value of the firm has obvious appeal . No matter in whose interests you believe the
firm should be run, the appropriate thing to do is to make the total pie to be divided
among the claimants as large as possible, and if the market freely signals whether the
total pie has been augmented or reduced, then follow the market's dictates.
But suppose that transfers are difficult, or that bargaining is constrained by
informational asymmetries and other transactions costs, or that executive decisions
are based in large part on private in formation that the market cannot assess. What
then? Clearly, market-value maximization may not be efficient if the market ignores
some of the managers' private information. Market-value maximization may also be
inefficient if bargaining costs prevent the value from representing all the interests
affected by management decisions.
Of course, the issue is much clearer if the firm should be run in the interests
of its owners. Then, still assuming that the market value of the firm at any point
 437
Executive and
Managerial
Compensation
Golden Parach utes and the Stock Market

Golden parachutes are contracts that give large severance payments to senior
executives who lose or leave their jobs following a merger or takeover. They
became very wi despread in the United States during the l 980s, especially as
the perceived threat of successful hostile takeovers loomed larger as the decade
progressed. The largest golden parachutes generated huge sums for thei r
possessors: F. Ross Johnson, the former CEO of RJR Nabisco, received $ 5 3. 8
million followi ng the leveraged buyout of the firm by Kohlberg, Kravis and
Roberts, and the vice-chai rman received another $45 . 7 mi llion.
Golden parachutes have often been attacked as another example of
executive greed being catered to by compliant boards of directors. They have
also been seen as weakening managerial incentives by reducing the threat
posed by the prospect that an ill-run company will be taken over and the
nonperformi ng managers fired. In contrast, they have been rationalized as
being useful in aligning executives' and stockholders' i nterests. The severance
pay helps deter executives from resisting takeover attempts that would be
costly for them personally but advantageous for stockholders.
One possible way to decide between these competing views is to consider
the reaction of the stock market to adopti on of a golden parachute provi sion.
A study focusing on the period from 1975 to 1982 found that the share prices
of firms announcing such plans in fact rose around the announcement dates
relative to what would otherwise be expected. * Thus, at least in this early
period, the market responded favorably to the adoption of golden parachute
plans. A comparable study covering later periods has not been done.
The favorable market reaction does not necessari ly imply that golden
parachutes are actually in stockholders' i nterests, however. The difficulty is
that the adoption of a golden parachute might be taken by the market to
mean that the firm's executives and board had received private information
that the company was a li kely target of a takeover attempt. Gi ven that
shareholders in firms that were acquired in hostile takeovers in the early
1980s realized average gains on their stock holdings of 40 to 50 percent, any
indication that a takeover was more likely ought to have driven up stock
prices, even if the parachute provi sion itself were contrary to shareholder
interests.

*Richard A. Lambert and David F. Larcker, "Golden Parachutes, Executive Decision­

Making, and Shareholder Wealth," Journal of Accounting and Economics, 7 (April 1 98 5), 1 79-
204.

reflects the best estimate of the future cash flows that the firm will generate, market­
value maximi zation is again the proper test of performance. Tying the CEO's pay to
the value of the firm would direct attention exactly where it belongs.

The Evidence on Performance and Pay

The presumptions that executives should seek to maximize the market value of the
firm and that incenti ve systems should be structured to encourage this behavior i n a
cost-efficient manner underlie much of the systematic empirical work on executi ve
performance and pay.
 438
Employment: PAY AND ORGANIZATIONAL SIZE The best-established empirical regularity concerning
Contracts, executive compensation is that pay rises with the size of the unit or organization the
Compensation, executive heads. ln study after study the same result appears, and the magnitude of
and Careers the effect is remarkably stable across firms, industries, countries, and time periods: A
1 0 percent increase in sales results in about a 2 to 3 percent increase in salary plus
bonus. 22 Larger firms pay more, and when a firm grows, its pay rises to the level
corresponding to its new size. ln fact, after surveying the evidence from different
studies, one observer has gone so far as to suggest that a new "universal constant" has
been found.
This size-pay effect has been interpreted by some as giving managers an incentive
to focus on sales volume and growth rather than value, on the presumption that larger
size will mean more pay. Causation need not How in this direction at all, however.
ln particular, if the marginal value of having more able people increases with
organizational size (because their talents can influence more people and decisions),
and if the allocation of executives is efficient, then larger organizations will pay more
because they have more able people who have a larger economic impact. Thus, the
size-pay correlation is hardly conclusive evidence that pay is not tied to performance.
PAY AND PERFORMANCE ln fact, it is statistically well established that pay for senior
executives is sensitive to firm performance. lt is certainly easy to find individual
examples where there is little apparent connection between the two, particularly in
the numerous stories in the business press on executive compensation. A CEO's
compensation may rise, year after year, often from an already high level, whereas the
firm's profits erode and its stock price falls from sight. In contrast, examples of very
explicit linkage between pay and performance can also be found. The box on Michael
Eisner of Walt Disney presents one example. In any case, the general pattern is that
pay responds to performance.
When performance is measured by accounting rates of return (accounting profits
divided by the accounting value of total assets, expressed as a percentage), the estimates
are that top executives' pay increases between 1 and 1 . 2 5 percent when the accounting
rate of return rises 1 percentage point. An alternative measure of performance is
shareholder rate of return (dividends plus any change in the stock price over the year,
divided by the price of the stock at the beginning of the year, again expressed as a
percentage). With this measure, the magnitude of the effect is lower by a factor of
1 0, but it is still statistically significant and, again, remarkably stable across data sets.
An increase in shareholder returns of 1 percentage point raises the CEO's salary plus
bonus by 0. 1 to 0. 1 6 percent. 23
THE INTENSITY OF CEO PERFORMANCE INCENTI\'ES For a CEO earning $ 1 million in
salary and bonus in a firm that has been yielding 10 percent market returns to its
shareholders, increasing the return to 1 1 percent is thus statistically associated with
an increase in pay of between $ 1 ,000 and $ 1 , 600. If the value of the company's stock
was, say, $ 1 billion, then the 1 percent gain in returns translates into an increased

22 Sherwin Rosen, "Contracts and the Market for Executives," National Bureau of Economic
Research working paper 3 542 ( 1 990).
23 Rosen, op. cit. Of course, the shareholder return might be dominated by changes in the share
price that are not so much attributable to anything specific to the company but rather to the overall behavior
of the stock market. This suggests that relative performance evaluation should be used in setting executive
compensation , so as to help filter out the elements of stock returns that are not attributable to the individual
firm's performance. The evidence on whether boards of directors in fact do this is mixed, although there
is some suggestion that it is becoming more prevalent. See Robert Gibbons and Kevin J. Murphy, "Relative
Performance Pay for Chief Executive Officers," Industrial and Labor Relations Review, 4 3 (February 1990),
30S- 5 1 S .
 439
Executive and
Managerial
Compensation
A CEO Performance Contract That Performed

In the early 1 980s the Walt Disney Company was in disarray. Managed by
family members of its brilliant but deceased founder, it had few worthwhile
movie production projects, its profits and stock price were substandard, an d
it seemed to be wastin g its resources. I t even paid $3 1 million in greenmail
to a hostile raider, payin g him above-market prices for the Disney shares he
had accumulated in return for a promise not to attack again.
In 1 984 Disney hired M ichael D. Eisner to become its CEO. M r.
Eisner had formerly been at Paramount, an other entertainment industry
giant. Mr. Eisner's compen sation contract with Disney gave him a base salary
of $750,"000 per year, a very modest amount by H ollywood standards.
H owever, it also entitled him to an annual bonus of 2 percent of all profits
in excess of a 9 percent return on equity (ROE). The 9 percent figure was
above what the company was earning when he j oined. H e also received
options on 2 million shares of Disney with an exercise price of $ 1 4.
Eisner rein vigorated the fa iling compan y. Under his leadership, the
return on equity j umped to 25 percen t, and the stock price climbed to $84.
In 1 988 he realized $ 32. 6 million on his stock options.
In 1 989 he entered a new contract with Disney. This left his salary at
$750,000 for ten years, but con tinued to give him 2 percen t of the profits
over an elevated new stan dard of an 1 1 percent ROE. He also got new stock
options: I . 5 million at a price of $69, the market price of Disney stock at
the date of the new contract, and an other 500, 000 at $79.
M r. Eisner received approximately $7. 5 million in bonus paymen ts in
1 989 and an other $ 1 0. 5 million in 1 990. The company enj oyed $820 million
in profits in 1 990 on revenues of $6 billion.

Based on Graef Crystal, "CEO Compensation: The Case of Michael Eisner," Fred K.
Foulkes, ed. , Executive Compensation: A Strategic Guide for the 1 990s (Boston: Harvard Business
School Press, 1 99 1 ), as well as data from the Business Week and Forbes surveys.

shareholder wealth of $ 1 0 million. The CEO's salary plus bonus rise by about 1 0 to
1 6 cents for every extra $ 1 , 000 created for the shareholders. These may n ot seem like
very inten se incentives. M oreover, a recen t noteworthy study by Michael Jensen and
Kevin M urphy has argued that the incen tives fa cing CEOs in large firms are even
more muted. 24
Rather than estimating the respon siveness in percen tage terms of CEO compensa­
tion to stockholders' returns, Jensen and M urphy estimated the change in salary plus
wages as a linear function of the change in shareholder wealth, which is essen tially
the sum of dividends paid during the year plus the change in the market value of the
firm's common stock. Their estimate was that an extra $ 1 , 000 of shareholder wealth
created was associated with onl y I . 3 5 cen ts more pay to the CEO. Comparing this
estimate with the fi gures in the preceding paragraph, we see that it is an order of

24 Michael C . J ensen and Kevin J . Murphy, "Performance Pay and Top-Management Incentives,"
fournal of Political Economy, 98 (April 1 990), 22 5-64.
 440
Employment: magnitude smaller than those found using the regression of the change in pay on the
Contracts, rate of return.
Compensation, Of course, there are many other elements of personal financial concern to a
and Careers CEO than just the current salary and bonus. To take account of these, Jensen and
Murphy estimated the effect of changes in total stockholder wealth (as defined
previously) on the change in CEO wealth (defined to include the salary and bonus
plus the change in the value of stock options), as well as the value of any of the firm's
stock held by the CEO and an estimate of any lasting effects that increased current
compensation might have on future pay and retirement benefits. Even allowing for
the impact of performance on the probability of being fired, the largest figure they
could generate was that CEO wealth rose just $3. 2 5 for each $ 1 ,000 increase in
shareholder wealth. Moreover, the sensitivity was much lower for the larger firms in
the sample. Averaged over the firms in the top half of the sample (ordered by market
value), the sensitivity was only $ 1. 85 per $ 1 , 000, whereas it was $8. 0 5 for the firms
in the bottom half of the size distribution. Most of the sensitivity actually came
through the CEOs' holdings of stock in their own companies, whose value is of course
perfectly correlated with total shareholder wealth. This accounted for $2. 50 of the
total figure of $ 3. 2 5 , and the fact that the CEOs of the smaller firms tended to hold
a much larger fraction of their firms' stock accounts for most of the difference in
sensitivity between large and small firms. In the sample as a whole, the sensitivity of
the factors that the board of directors control-salary and bonus, stock options, and
likelihood of performance-related dismissals-amounted to only 75 cents over a
lifetime per $ 1 , 000.
ARE CEOs' INCENTI\'ES INADEQUATE? Jensen and Murphy argue that these sensitivity
figures are too small to provide adequate performance incentives. For example,
consider a CEO contemplating whether to use $ 1 million of the firm's resources on
some pet project-perhaps an endowed chair to support research and teaching at the
university the CEO attended. Ignoring risk aversion, the median CEO in the sample­
the one relative to whom half get stronger incentives and half get weaker-would find
it directly worthwhile to give away the shareholders' $ 1 million if the project brought
$3,250 in personal pleasure to the executive. If the CEO had no substantial ownership
stake in the firm, the breakeven occurs at $750. This is less than a single morning's
pay for the median executive in the sample. Ethical considerations aside, it might be
quite tempting to spend the shareholders' money this way. Of course, the same sort
of calculations apply even to less worthy diversions of funds.
Jensen and Murphy also consider a variety of factors that theory would suggest
would tend to limit the intensity of incentives provided to CEOs. Risk aversion should
lower the intensity of incentives, and CEOs' limited wealth also restricts the intensity
of incentives that can be given. The imperfections that arise in performance
measurement because stock prices are influenced by much more than the efforts of
the C EO are another reason to deemphasize shareholder wealth. Other possibilities
are that nonfinancial incentives are adequate or that CEOs' concerns with their market
value as managers or with the threat of hostile takeovers are sufficient to induce them
to pursue the stockholders' interests diligently. Jensen and Murphy do not find any
of these compelling, however. They then argue for tying CEOs' wealth much more
to stockholders' welfare to improve corporate performance, essentially by greatly
increased use of restricted stock awards. 25

25 Michael Jensen and Kevin Murphy, "CEO Incentives-It's Not How Much You Pay, But How, "
Harvard Business Review (May-June , 1 990), I 38-53.
 441
Tt lE COUNTER AHGUMENTS J ensen and Murphy's analysis and recommendations have Executive and
attracted attention and criticism from both academics and practitioners. The academics Man agerial
have tended to concentrate on econometric and theoretical issues. Cogent arguments Compensation
suggest that at least the first of the Jensen-M urphy estimations-the one relati ng the
change in salary plus bonus to shareholder returns-may have assumed an i nappropriate
functi onal form relating the two variables, and that this accounts for the order- of­
magnitude difference between the estimate they obtai ned and those generated in
earlier studies. 26 Perhaps more telli ngly, however, are the arguments that the
multidimensional nature of the CEO's j ob may mean that the incentives tied to stock­
price movements optimally should not be too strong.
Tying pay to stock performance certainly discourages excessi ve perks and
unprofitable empi re bui lding. There is, however, more to worry about and motivate
than these "effort" decisions. Even assuming efficient capital markets, it appears that
short- term returns i nfluence market percepti ons of long- term prospects and thus stock
prices.27 I n that case, tying pay to share prices may invite executives to favor investments
that are too short-term oriented. (Of course, if markets are not efficient, then the
attractiveness of share prices as measures of performance is even more questionable.)
Tying pay too closely to share performance also can have harmful effects on executives'
willi ngness to undertake risky but profitable investments. The use of stock option plans
mi tigates this somewhat (because the executive does not bear the financi al losses that
result when investments do not work out but does share in the successes), but these
plans may encourage taking excessive risks. Further, accounting measures can be
useful in controlling a tendency to take too many perks, and so they should also be
used in determining compensation, rather than j ust stock prices alone. As the equal
compensati on pri nci ple indicates, providing strong incentives for some acti vities risks
distorting executi ves' choices about others that may be as important. It may be that a
low level of sensitivity to stock prices is appropriate in this context.28
The other cri ticism of the J ensen-M urphy positi on comes from those who are
already concerned that executive pay is often excessive and ineffective. It starts from
the apparent tendency of boards to add new i tems to CEOs' compensation packages
without adj usting the size of the existing ones. Given thi s, a recommendation to
increase the linkage between CEO and shareholder wealth through stock awards
amounts to an invitation to bestow large amounts of stock on the CEOs, giving away
even more of the shareholders' money to already well-paid executives.

Does CEO Pay Affect Performance ?

Of course, giving stock to executi ves could still be worthwhile if it would i ncrease
performance enough to offset the diluti on of share values. This raises the fundamental
issue: Do executive incentives actually work? Pay may be sensitive to performance,
but thi s is not the real issue. The i mportant questi on is whether providing incentives
to CEOs improves corporate performance. Thi s turns out to be a very subtle issue to
decide empirically. Relatively few studies have addressed this question, and the results
appear mixed.
STATISTICAL ANALYSES Jensen and M urphy have presented some limited evidence
that the stock-market performance of the 25 companies in their sample which provided

26 Rosen, op. cit.

27 See Chapter 1 4.
28 Bengt Holmstrom , "Comment on S . Rosen: 'Contracts and the Market for Executives'," draft,

Yale University, School of Organization and Management (1990).

 442
Employment: the strongest incentives exceeded the performance of the firms that gave the weakest
Contracts, incentives by a substanti al, statistically significant margin. 29 Working with the entire
Compensation, Jensen-Murphy data set, however, Graef Crystal found that differences in the strength
and Careers of incentives had little or no power in explaining the variation in performance among
companies. 30
These results are not necessarily inconsistent. Pay may affect performance both
in fact and statistically (that is, in the sense that the two are more closely linked in
the data than is plausibl y the result of pure chance), and yet the influence of other
variables on performance may totally dominate that of pay. Then, increasing the
sensitivity of pay to performance will improve performance in a probabilistic sense,
but the effects may quite possibly be drowned out by other factors. Nevertheless, when
only the extreme cases are considered, the relationship might show through.
M oreover, there are also subtle issues in interpreting the response of stock market
prices to pay systems. For example, if the market believes the adoption of an innovation
in CEO pay will improve future performance, then the stock price will immediately
be bid up to reflect the expected magnitude of this effect. Subsequently, share
performance should not be exceptional because all the gains were realized at the
outset when the stock price adj usted to make the anticipated return on holding the
firm's stock j ust competitive with other investment opportunities. Thus, it may be
hard to see any difference i n stock price performance between firms using high­
powered incentives and those that do not unless the data cover the period when the
plans were instituted.
M ore fu ndamentally, however, neither the Jensen-Murphy calculations nor
Graef s regression analyses are really the best way to test whether i ncentives affect
performance. A better test would be to look at changes in the intensity of incentives
and the changes in subseq uent performance. This has been done in a recent study
by J ohn Abowd of some 1 6 thousand executives in 2 50 large U . S. corporations in the
peri od from 198 1 to 1 986. 31 H is study indicated that increasing the sensitivity of
changes in salary and bonuses to current corporate performance increases future
performance in a statistically significant fashion: increases in the sensitivity of either
the executives' annual raises or their bonuses in a given year to current shareholder
rate of return were associated with an increase in the shareholder rate of return for
the next year. Abowd's study did not, however, take account of long-term incentive
plans or examine their impact. Doing so would be desirable, but would involve a
massive effort to assemble the data.
INTERNATIONAL COMPAR ISONS An entirely different approach to the question of
whether CEO incentives affect performance and are worth their cost arises from the
observation that German and Japanese CEOs receive few or no long- term performance
incentives of the kind common in the United States and yet large companies in
these countries are generally seen as performing much better than their American
counterparts. For example, W. Edwards Demming, the American statistician whose
quality-control methods are a key feature of Japanese manufacturing, criticizes most
incentive plans as being dysfu nctional. Demming sees them as focusing executives'

29 Michael C. Jensen and Kevin J. Murphy, "Letter to the Editor, " Harvard Business Review (July­
August, 1990) , 190-1.
30 Graef C rystal, "The Great CEO Sweepstakes", Fortune (June 1 6, 1990), 94-1 02; and The Crystal
Report on Executive Compensation , 2 (May/June 1990). (Mr. C rystal is the former head of a major
compensation consulting firm who now teaches at the University of California, Berkeley. The Crystal
Report is a newsletter he publishes in Napa, California. )
31 John M. Abowd, "Does Performance-Based Managerial Compensation Affect Corporate Perfor­
mance?" Industrial and Labor Relations Review, 43 (February 1990), 52S-73S.
 443
attention on narrow numerical goals rather than on the long-term health of the Executive and
company. Various senior Japanese executives have also been highly critical of
American executive compensation practices. 32
Managerial
Compensation
There is certainly a nice irony, as well as an important puzzle, in the apparent
negative correlation internationally between the prevalence of performance incentives
and the level of realized performance. Nevertheless, moving to a Japanese pay system
does not seem obviously desirable in the N orth American context. First, the j ob of
the CEO in a Japanese firm is markedly different fr om that of a U . S . company's chief
executive. J apanese firms traditionally push decision-making power and responsibility
down the hierarchy. They rely on consensus, with plans and proposals originating at
low levels and working their way up to the executive offices. They do not go outside
to hire hot-shot executives to tum the company around. The CEO in this system is
supposed to represent the company and its values, not run it in the American sense.
Second, Jap�nese firms are not run in the interests of their shareholders. I ndeed,
Japanese CEOs do not even profess to believe that they should be: The interests of
the employees (in j ob security, opportunities for advancement, and so on) are seen as
being of at least equal importance. This means that the "long-term health of the
company" about which Demming worries is a prime consideration in Japan. Yet it is
not obvious that it is the appropriate measure of performance to be pursued, at least
in the N orth American context. Third, the period when American industry got into
its worst trouble and lost so much of its international competitive position was before
the now-popular long-term incentive plans were adopted. At that time, U . S . executives'
pay was much less closely aligned with shareholder concerns. Is there any reason to
believe that going back to the practices of that era would be an impr ovement?

Implications and Conclusions

It is very hard to decide whether CEO pay provides appropriate and adequate
incentives. Is $ 3 . 25 per $1, 000 an adequate incentive? Or is the real issue the millions
of dollars that top executives are paid to sit in fancy offi ces, ride in corporate j ets, and
decide the fates of hard-working people they do not even know exist? I ndeed, it is
even hard to figure out how to decide this issue.
Even if high pay is well explained as a motivator in average cases, one still
might be concerned about the individual cases where it clearly is not. The general
pattern may be that executives' pay is responsive to performance and that performance
in turn is positively affected by the incentives that corporate leaders receive. It may
even be the case that executive compensation systems on average are quite appropriately
designed and calibrated. Nevertheless, there are certainly plenty of examples in which
whatever performance incentives were in place did not motivate the chief executives
in question to pursue much of anything other than their own narr ow interests.
M oreover, the boards of directors have often sat by as these CEOs led their companies
to decline and even ruin, all the while giving them handsome raises.
The questi on that arises is whether there are other mechanisms that might
ensure that such examples are rare and transient. Of course, if the conclusion is that
presently prevailing pay schemes do not give CEOs adequate motivation but instead
just pay them exorbitant amounts, then the question of the possible existence of other
discipline devices is even more pressing.

32 Dana Weschler Linden with Vicki Contavespi, "lncentivize Me, Please," Forbes (May 27, 1 99 1 ),
208- 1 2.
 444
Employment:
Contracts,
Compensation,
and Careers SUMMARY
Executive compensation involves a number of specific issues beyond the general ones
addressed in the preceding chapter on pay and motivation. Many of these relate to
the form and level of CEO compensation in large U. S. firms.
During the 1 980s CEO compensation in the largest U. S. corporations grew four
times as much in percentage terms as did the pay of the average employee. Much of
this seems to be attributable to the proliferation of long-term incentive provisions that
link the executives' pay to the stock-market performance of the shares in their firms.
Such schemes are rare in smaller companies (except entrepreneurial ones where the
founder is CEO and still retains a substantial ownership share). They are also rare in
Europe and Asia. In any case, the absolute level of total CEO pay in large American
corporations is approaching $2 million per year, and the ratio of the CEO's total
compensation to the pay of an average worker is several times larger than the
comparable figure in other leading industrial countries and in smaller U . S. firms.
Pay for middle managers has not risen nearly as fast as for their bosses.
Performance-based incentive pay seems to be playing an increasingly important role
for these managers, but it is still much less a factor than at higher levels in the
corporation. This relative pattern is in line with the predictions of both the tournament
theory of promotions as incentive mechanisms and the theory of incentive contracting.
The former suggests that direct incentives need to replace promotion incentives as the
employee reaches the top of the hierarchy. The latter suggests that incentives should
be more intense where the employee's efforts have a bigger marginal impact on
profitability and where the employee has greater opportunity to respond to increased
incentives. Both these conditions seem more likely to hold at higher levels in the
firm. Another major trend in middle-management compensation has been to redefine
the basis for determining salary to avoid some of the perverse incentives to expand
staff, which are occasioned by basing managers' pay on the number of people they
directly supervise.
Deferred compensation can be especially important in providing proper motiva­
tion for executives because the results of many of the major decisions and actions
they take do not become evident for extended periods. In fact, however, there is some
question about the extent that managers are held responsible for these choices: Often
it seems they have moved on long before the results become known, and they are
never held accountable. This tendency may actually reflect a commitment problem.
Once the decisions are made and the actions taken, but before the results become
known, an efficiency gain can be realized through insuring the manager by removing
the dependence of pay on results.
Motivating executives to take appropriate risks, especially in investment decisions,
is complicated by the fact that such decisions put not only the financial capital of the
firm's owners at risk but also the human capital of the managers associated with the
project. Again, success reflects well on the managers' talent and dedication, and
failures hurt their market value. A possible response is to use explicit contracts that
pay the managers for successful investments but do not punish them for failures. This
asymmetric payoff pattern can help offset the managers' risk aversion about the value
of their human capital and motivate them to take on the risks inherent in investing.
Stock options have these essential features, as do the contracts used by oil companies
to reward professio nals involved in developing petroleum deposits. In this context, it
may also be useful to pay simply for having proposals adopted. Both of these measures
may, however, give excessive incentives to try to get projects funded, and this may
 445
necessitate close scrutiny of investment proposals and the use of more deman ding Executive and
criteria for undertaking investments than would otherwise be employed. Managerial
A maj or debate concerns the pay received by the CEOs of large U. S. firms. Some Compensation
observers see the high levels of pay as unjustified: They see them as the outcome of
managerial greed unfettered and even abetted by compliant boards of directors. Others
worry that executives are not paid enough to attract the best talent to corporations
rather than to such fields as investment banking, consulting, and entrepreneurship. An
additional concern is that CEO pay may not provide adequate financial incentives for
these executives to pursue stockholders' interests with sufficient diligence. Evaluating
the arguments in this debate involves dealing with some subtle issues. The evidence
currently available suggests that, on average, CEO pay responds to firm performance
and that the intensity of CEO performance pay positively affects firm performance (at
least in the context of U. S. fi rms). Whether the sensitivity of pay to performance is
great enough, and whether the improved performance is worth the cost is less clear.

• BIBLIOGRAPHIC NOTES
Sherwin Rosen gives an excellent (although somewhat technical) survey of the
economic issues in executive compensation. The annual feature stories that
appear in Business Week, Forbes, and Fortune during the late spring and early
summer provide up-to-date information on executive compensation, as well as
the criticisms of pay excesses. The explanation of how increasing staff size in a
bureaucracy increases the work load is due to Jane Hannaway. The analysis of
managerial risk aversion in investments is based on the work of Bengt Holmstrom
and Holmstrom and Joan Ricart i Costa. The special February 1 990 issue of the
Industrial and Labor Relations Review contains a number of excellent papers on
executive compensation and its connection to firm performance. The paper by
Baker, Jensen, and Murphy and the book by Lawler referenced in Chapter 1 2
also have discussions of executive compensation, and the volume edited by Fred
Foulkes contains a number of useful papers on the subject.

• REFERENCES
Foulkes, F. , ed. Executive Compensation: A Strategic Guide for the 1 990s (Boston:
Harvard Business School Press, 1 99 1 ).
Hannaway, J. , "Supply Creates Demands: An Organizational Process View of
Administrative Expansion, " fournal of Policy Analysis and Management, 7 ( 1 987),
1 1 8- 1 34.
Holmstrom , B. "Managerial Incentive Problems: A Dynamic Perspective, " in
Essays in Economics and Management in Honor of Lars Wah/beck ( Helsinki,
Finland: Swedish School of Economics, 1 982).
Holmstrom , B. , and J . Ricart i Costa. "Managerial Incentives and Capital
Management, " Quarterly fournal of Economics, 1 0 1 ( 1 986), 83 5-60.
Rosen, S. " Contracts and the Market for Executives, " National Bureau of
Economic Research working paper 3 542 ( 1 990).

I
EXERCISES
Food for Thought

1 . Senior executives often appear to have effective partial control over the
form of their compensation in that they can opt to receive a greater amount in stock
 446
Employment: and other performance pay and less in straight salary (although they probably cannot
Contracts, so easily switch risky income into safe receipts). Why might allowing the sort of
Compensation, freedom that executives seem to have be advantageous to stockholders?
and Careers 2. The study cited in the text of middle managers' pay indicated that the
sensitivity of pay to performance tended to be higher for those assigned to the corporate
head office than for those working in plants away from head office. How would you
account for this pattern?
3. When a company undertakes a massive risky investment whose failure
would endanger the company's survival, it puts at risk not only the value of the stock
held by its shareholders, but also the earnings of its employees, the profits of its
suppliers and of the customers who count on it, and the general welfare of the
communities in which it operates. Should this affect investment decisions? If so, how?
4. Consider two CEOs who head similar companies. The two executives are
similarly situated except that one has a large holding of the stock in the firm he or
she heads and the other does not. How should the compensation contracts of the two
differ in theory? What problems do you see with implementing such a recommendation?
5. We accentuated the negative role that career concerns can have by inducing
managers to be risk averse in the investment decisions they make for firms. In contrast,
concerns with their market value might also provide positive incentives for managers
to try to make sure that their organizations do well because the success of their firms
will reflect well on them and improve their market opportunities. What implications
does this have for the optimal strength of explicit performance contracts at various
stages in a manager's career? How would you test this?
6. Suppose you have decided that executive and CEO performance incentives
are important. You then note that managers in the not-for-profit sector (for example,
university administrators) typically receive straight salaries. Although their raises might
possibly depend on performance, they get no bonuses and certainly no long-term
incentive pay. Is this a problem for thes.: organizations? If so, how would you seek to
overcome it?

Mathematical Exercises I

1. Suppose that the change in stock market value X of a firm with annual
sales of S is equal to f(S) + S 1 -ae + Se, where e is the "effort" expended by the
CEO, e is earnings variations associated with uncontrolled random influences, and
f(S) is some function of the firm's size. An important part of the dispute over proper
types and levels of executive compensation is the argument about how much effect
an executive can have on the performance of a large organization. In our model, this
effect is represented by the elasticity parameter a. For a = 0, the impact of executive
efforts is proportional to the size of the organization; for a = I , the impact is fixed
independently of the size of the organization. For intermediate values of a, the
executive's ability to affect profits grows by about a% every time the organization's
size grow's by 1 % .
Use the analysis of Chapter 7 to determine the optimal rate at which the
executive's earnings should increase with the company earnings. Assume that the
CEO's cost of effort is C(e) = ½e 2. (Hint: Use the earnings X to construct an appropriate
statistical estimate of e. Compute the incentive intensity B using this estimate and
then restate the answer as a function of X. ) For S large, how does the answer depend
upon a?
 Part

VI
FINANCE : INVESTMENTS,
CAPITAL STRUCTURE, AND
CORPORATE CONTROL

14
THE CLASSICAL THEORY OF
INVESTMENTS ANO FI NANCE

15
FINANCIAL STRUCTURE,
OWNERSH IP, ANO CORPORATE
CONTROL
 14
THE CLASSICAL THEO RY OF
INVESTMENTS AND FINANCE

IQ l ctober. This is one of the particularly dangerous mon ths to speculate in

stocks in. The others a re f uly, lanuary, September, April, November, May, March,
lune, Decem ber, August and Februa ry.
Pudd'nhead Wilson 1

The previous four chapters have investigated the relationships between a firm and its
employees. This chapter and the next deal with the relationships with another crucial
constituency: the providers of capital. We introduce the subject in this chapter by
focusing on the basic theory of classical fi nancial economics, obtaining its central
results. At the heart of this theory is a view of the firm as a simple Aow of financial
streams from investments and an assumption that the way that a firm finances its
investments does not affect the investment returns themselv es. The resulting theory,
which is probably used more i n actual business than any other part of economics, is
valuable for several reasons. First, it leads to a practical and useful account of how
i nvestments ought to be evaluated: The theory of net present value. The Fisher
separation theorem clarifies a common confusion about when an owner's personal
preferences might matter for investment decisions. The celebrated Modiglian i-Miller
theorems eliminate another confusion, with their surprising assertion that if capital
structure (the way i n which the firm is financed) and dividend decisions affect neither
the firm's underlying cash Aows nor investors' perceptions of them , then they cannot
affect the firm's value. The theory of portfolio choice and the resulting capital asset
pricing model not on ly pro\'ide explanations of the equilibrium pricing of securities,

448 1
Mark Twa in, Pudd'nhead Wilson (189-t).
 449
they al so indicate how to calculate the eost of capital to the firm. Finally, augm ented The Classical
with the perfect ( or efficient) markets hypothesis, the theory provides a basis for Theory of
evaluating managerial performance. Investments and
As valuable as this theory is, it leaves important questions unanswered. We need Finance
a richer theory if we are to understand why and how capital structure can affect the
firm's value. The classical anal ysis views financial instruments as streams of returns
promised to lenders and investors. These returns may depend on the stream of earnings
of the firm, but, in the classical analysis, financing decisions can affect neither the
firm's earn ings themsel ves nor investors' expectations about them. M ore modern
analyses recognize both possibilities: Financial structure can alter management's
incentive to work hard on behal f of the firm's owners, and it can also affect investors'
beliefs about likely future profits. As well, it can affect the possibilities for changing
ownership and c ontrol of the firm. All of these relationships affect the firm's market
value. Allowi�g for these possibilities is a key element of Chapter 1 5 .

THE CLASSICAL ECONOMICS OF

INVESTMENT DECISIONS

The classical economics of finance is characterized by the assumption that managers

and investors have no conflicting interests, or at least that the financial decisions of
the firm play no role in any such conflict. Even if you rec ognize the potential for
conflict, the classical assumptions remain useful to the extent that any managerial
incentives that can be provided using financial structure might be equally well provided
by explicit incentive compensation contracts or by rules limiting managerial di scretion
that are unrelated to the firm's financing. The modern parts of the theory are still
unsettled, but there is little disagreement that the classical theory provides important
theoretical and practical insights that are well worth learning.
In classical economic theory, the firm is conceptualized as a set of possible
production plans from which the most profitable one is chosen. Classical finance is
rooted in this theory. It treats a firm as a set of potential investments from which
some are chosen. The questions to be studied concern which investments ought to
be undertaken, how the funds needed to pay for the i nvestments ought to be raised,
and how investors determine what prices to pay for shares of stock and other securities.

The Fisher Sep aration Theorem

Suppose that the owner of a shop is considering opening a second outlet in a nearby
suburb. H e or she must rent a building, hire staff, acquire an inventory, and invest
in initial advertising and promotions to inform local consumers that the shop is there.
Over a period of perhaps ten years, the owner expects to reap rewards from this
investment in the form of additional sales and profits. After accounting for all the
costs, the owner expects to earn profits of P 1 in the first year, P2 in the second, . . . ,
and P 1 0 in the tenth year. Suppose that these numbers are certain, or that the investor
is risk neutral and that these are the expected returns. Is the investment worthwhile?
If the owner must finance the shop out of personal savi ngs because there are
no developed financial institutions to lend the needed eash, then the answer to this
question depends on what else he or she could do with the money. If our entrepreneur
had been eagerly looking forward to buying a new sailboat with the money, he or she
might be discouraged from investing by the prospect of having to delay that purchase.
When individuals finance an investment entirely with personal funds, the decision
depends on personal preferences about the timing of consumption. This conclusion
changes when the investment can be financed with borrowed money, however.
 450
Finance: IN\'ESUIENTS FINANCED B, BORRO\nNC Suppose the owner can borrow the money
Investments, needed to open the shop at an interest rate of r per year. Suppose also that the loan
Capital Structure, agreement stipulates that the store's entire cash flow-its sales revenue after payment
and Corporate of operating expenses-will be used to pay the interest due and to reduce the balance
Control on the loan until it is fully repaid. Further, if the loan is not repaid by the end of ten
years, or if the store closes before then, the owner will be responsible for paying
whatever loan balance remains from personal funds.
In evaluating the investment when it is financed by borrowed money, the owner
will conclude that the investment is worthwhile if the loan can be repaid using the
store's cash flow, with something left over. This is because the owner profits by keeping
what is left over after the loan is repaid. If the loan cannot be repaid out of the store's
earnings, then the owner would do better not to open the new store because he or
she would have to turn over all the store's earnings, plus some private savings, to the
lender.
Although useful, this preliminary account of how the owner should make this
investment decision is thin in several respects. One important omission is the cost of
the owner's time in setting up the store and contributing to its operation. This is not
a serious conceptual problem, however. To account for the owner's time, we simply
ask: Will the revenues be sufficient both to pay the opportunity cost of the owner's
time and to repay the loan, with something left over? The investment is worthwhile
if, and only if, the answer is yes.
Notice the important contrast between this conclusion and the case we first
analyzed in which the owner financed the investment using personal savings. When
the markets for borrowing and saving operate perfectly, there is no necessary connection
between the timing of income from the investment and the timing of the owner's
personal consumption. In that case, the question of whether an investment should
be undertaken is transformed. It is no longer a decision based on personal preferences;
it becomes instead an objective business decision based only on projected earnings
and interest rates.
I N\' EST l'.\IC '' THEN THERE ARE PERFECT C.\PIT.\L 1\1\ RKETS A more realistic case is the
one in which the owner can decide whether to finance the investment using any
personal savings, to finance by borrowing, or to use some combination of these two
means. Suppose the interest rate at which the owner can borrow is the same as the
rate he or she could receive by lending to others: This is one aspect of a condition
known as perfect capital markets. 2 In this case, the owner's opportunity cost of using
savings is just the same as the cost of borrowing, so it is correct to analyze this
investment as if it could be financed only by borrowing. The conclusion of this logic
is summarized by the following:
Fisher Separation Theorem: When there are perfect capital markets, the
owner's decision about whether to invest will depend only on the returns
he or she forecasts from the investment and on the interest rate. It will
not depend on his or her preferences regarding personal consumption or
its timing.
This theorem is named for economist Irving Fisher, who developed this logic in the
first decade of the 20th century. The term separation in the name of the theorem
reflects the conclusion that an owner can separate the decision about whether to
invest from the decision about how to arrange consumption over time. The first

2 More generally, perfect capital markets obtain when all parties, whether they are buying or sel ling,
transact at the same terms and these terms are unaffected by the amounts any one party transacts.
 45 1
The Classical
Theory of
Cash Accumulation Present Investments and

Year Flow Factors Value

Finance

t ct A t = ( 1 + r)l C/A t

0 1. 000
1 99 1. 1 00 90. 00
2 121 1.210 1 00. 00
3 1 00 1. 3 3 1 75. 1 3
Total Present Value 265. 1 3
Initial I nvestment 220. 00
Net Present Value 45. 1 3

decision depends on earnings forecasts and interest rates; the second depends on
interest rates, wealth, and personal preferences.
SIGNIFICANCE OF THE SEPARATION THEOREM This separation of investment decisions
from personal preferences is very important for the analysis of firms. For example, in
a joint business venture, if the partners agree on their forecasts of investment returns,
and if they both know what rates of interest are offered by the bank for its loans, then
they should agree about whether to undertake an investment, even if their personal
circumstances are quite different. To take a particular case, if one partner is near
retirement and wants to start spending some of the wealth he or she has accumulated
over a lifetime, that is no reason not to invest, provided the investment can be financed
with loan proceeds. Without perfect capital markets, this difference of interests between
the partners could be a source of disagreement and conflict.
The conclusion about separation is important because it helps to justify the
common practice of thinking about firms, especially large firms, as being separate
entities from their owners . For small firms, the conclusion may be different because
the perfect capital markets conditions is not a good approximation of reality: The
owner cannot borrow and lend at (nearly) the same rate of interest. In that case, it
may be optimal for a retiring owner to forego making new investments in the business.

Net Present Values

To complete our description of the owner's optimal investment policy, we now show
how to calculate whether the loan used to finance the investment can be repaid from
the forecasted sequence of cash flows, after deducting the cost of the owner's time
and all other relevant costs. The way to do that is to calculate year by year how much
of the loan can be repaid using just the earnings generated in that year and then to
add these figures to find the largest loan that can be repaid from the earnings of all
the years together.
The size of the loan that can be repaid from any year's cash flow is called the
present value of the cash flow. This concept is very important in practical financial
calculations. Table 14. 1 illustrates how present value is computed , using both numbers
and formulas. To keep matters simple, we assume that all payments are made at the
ends of years, though similar calculations could be done using monthly or quarterly
payments.
Let us focus first on the column labeled "Accumulation Factor. " A debt of 1
(dollar, or thousand yen, or whatever) at an annual interest rate of 1 0 percent grows,
or accumulates, after one year, to a debt of 1 . 1 0 . If unpaid, that debt grows by 1 0
 452
Finance: percent again in the next year; that is, it is multiplied again by 1 . 1 to become a debt
Investments, of 1 . 1 X 1 . 1 = 1. 2 1 . The numbers in the accumulation column show how much
Capital Structure, is owed at the end of each year if the $ 1 loan is allowed to accumulate over the years,
and Corporate with interest compounding annually at 1 0 percent. The number for each year is
Control computed by multiplying the value for the previous year by 1 . 1 , starting with the
number 1 . 000 in the row for year 0.
Interest rates are commonly quoted in percentage terms; that is, we speak of
rates of 3 percent, 1 0 percent, or 3 5 percent. For writing formulas and doing
calculations, it is more convenient to express interest rates as the decimal fractions:
. 0 3, . 1 0, or . 3 5. In general, if the interest rate is r, the accumulation factor A1
corresponding to year t is the result of multiplying by the number ( 1 + r) by itself t
times, or (1 + r)I, as indicated in the column heading. This sort of calculation is
easy to program on a personal computer using any of the popular spreadsheet programs.
It is also preprogrammed into many hand-held calculators, especially business and
financial calculators.
At the end of year 1 , the second column of the table shows the owner's forecast
that the cash available for servicing debt (paying interest and principal) will be 99,
after first paying the necessary operating expenses of the business. That sum is just
enough to repay a loan of 90 made at the beginning of the year, if the annual interest
rate is 1 0 percent. The reason is that, using the accumulation factors, a loan of 90
accumulates in a year to 90 X 1 . 1 00 = 99. The proper procedure for computing
the figure to be entered into the present value column is to perform this operation in
reverse. In this case, the maximum loan that can be repaid by 99 is 99 divided by
the accumulation factor of 1. 1 00. Similarly, the 1 00 present value for year 2 is
calculated by dividing the forecasted cash flow of 1 2 1 by the accumulation factor of
1 . 2 1 0.
In general, a loan of L accumulates with interest at rate r after t years to a
balance of L( 1 + r)l. If the available funds or cash Aow Ct are just sufficient to pay
the loan, then Ct = L( l + r)I ; therefore, L = C/( 1 + r)l. In terms of the table, this
means that the present value of the cash flow can be computed simply by dividing
the cash flow in the second column by the accumulation factor in the third column.
The result is the present value reported in the fourth column.
After having computed the present value of the available cash in each year, the
total present value of the stream of cash payments is calculated by adding up the entries
in the fourth column. This is justified because the reduction to present values expresses
all of the flows in the common unit of money today: We are not adding today's dollars
or marks or yen to tomorrow's, which are different things with different value. In our
sample calculation, this total present value is 26 5. 1 3. The initial expenditure required
for this project is 220, and the net presen t value of the whole project is 45. I 3 ( = 265. 1 3
- 220). That is, the project is expected to generate enough cash to justify loans of
265. 1 3, but only 220 must be borrowed today to-finance the project, so there will be
surplus cash generated by the project beyond what will be needed to repay the loans.
lf the only options are either to undertake a project or to reject it, then the project
should be undertaken only if the net present value is positive. Generally, when deciding
a mong several mutually exclusive alternatives, the best one is the one with the highest
net present value, because that is the one that leaves the most additional funds beyond
those necessary to repay the lenders who finance the project.

TI IE COST OF C.\PIT\L IJ\.i TI IE Ct..\SSl<:.\L FR\\IEWOHK The interest rate r used in this
calculation is called the cost of capital. It represents the lowest rate a business m ust
pay to raise the necessary financing for the project. One of the important practical
problems businesspe ople face in trying to use this theory is to determ ine what the The Classical
actual cost of capital is. The tax deductibility of interest paymen ts, the restrictive Theory of
clauses in loan agreem en ts, the use of in terest rates that vary according to the security Investments and
of the collateral, and the use of mixed sources of fi nancing involvin g different terms Fi nance
and interest rates for different parts of the same project all affect the real cost of
borrowing. Add to this the possibility of taki ng in partners for equi ty fi nancing, and
i t becom es very difficult to pin down the number that corresponds to the theoretical
cost of capital. In addition, the returns that i nvestors require from any particular
investment depend on the risks they incur in fi nanci ng it, so it would be incorrect
simply to use the average cost of capital in a company to evaluate the attractiveness
of each investment the company makes. Finance textbooks contain lengthy treatments
about how to compute the cost of capital. 3
APPLICATION : ''SHORTSIGHTED" MANAGEMENT AND THE Cmff OF CAPITAL The cost of
capital is one · of the principal ingredients in determining whether to undertake an
in vestment. When it is high, investments become less attractive. For example, consider
an investment of I 00 today that is certain to return a cash flow of C in two years but
nothing until then . If the interest rate is 6 percent (r = . 06), then C must be at least
112. 36 for the investment to be pr ofi table because a loan of I 00 at that interest rate
would accumulate after two years to a debt of 100 X l . 06 2 = 112. 36, and thi s debt
must be repaid from the cash flow. If the interest rate is I O percent (r = . I 0), then
the cash flow must be at least 121 ( = I 00 x I . I 0 2 ) for the investment to be pr ofitable.
The cash flow after two years on an i nvestment of 100 needs to be 7. 7 percent
( = (121 - 112. 36)/112. 36) hi gher when the interest rate is I O percent instead of 6
percent, and the returns net of the initial investment must be 69. 9 percent more
[(21 - 12. 36)/12. 36].
The difference in required returns gr ows dramatically as the period of investment
grows longer. The cash flow required after ten years to repay a loan of I 00 today is
179. 08 if the cost of capital is 6 percent, but the corresponding figure is 259. 37 if the
cost is I O percent-a difference of almost 45 percent. If the period of investment is
20 years, the corresponding figures are 320. 71 for the 6 percent cost of capi tal and
672. 75 for the I O percent cost of capital. In order to be profitable at a cost of capital
of I O percent, a 20-year investment must generate more than double the cash flow
than would be needed if the cost of capital were only 6 percent. Expressed another
way, the difference i n required cash flow is more than 3 . 5 times the original investment
of 100!
One of the popular explanations of the relatively poor performance of N orth
American manufacturing industries duri ng the 1970s and 1980s was the "shortsighted­
ness" of their managers. Accordi ng to critics, U.S. managers refused to make
investments for the long term at a time when inter national competitors i n Europe
a nd Asia (especiall y in Japan) were maki ng heavy investments in new equipment and
new technologies that would guarantee their competitiveness over long periods of
years.
A look at the data on the cost of capi tal (adj usted to account for the different
ta x polici es of different countries) helps to explai n thi s phen omenon. The cost of
ca pital in the U nited S tates was arguably higher throughout this period than was the
correspondi ng cost in countries like Germany, and much higher than i n Japan.
Indeed, accordi ng to one study, when the after-tax cost of fi nancing a new plant in

� See, for example, James C. Van Horne, Financial Management and Policy (Englewood Cliffs,
NJ: Prentice Hall, 1989).
 454
Finance: Japan by issuing equity was 6 percent in 1 988, the corresponding figure for an
Investments, investment in the United States was about 1 0 percent,4 this was the basis for the
Capital Structure, particular numbers illustrated earlier. One reason for this large difference was the
and Corporate much lower savings rate in the United States compared to Japan; another was the
Control different tax rules that applied in the different countries.
TECHNOLOGICAL SHORTSIGHTEDNESS It would be incomplete to cite these differences
in the costs of capital as a full explanation of the different patterns of investments in
the different countries. Many scholars believe that differences in management practices
also account for part of the difference.
In practice, analysts within firms compute net present values using estimates of
the quantifiable cash flows generated by an investment. Benefits that are difficult to
quantify are likely to be ignored. However, many of the benefits of investments in
new technology are of this very kind: They maintain and improve the company's
standing on the frontiers of advanced technology. A technologically advanced company
benefits by being better able to incorporate advanced features into its products, to take
advantage of new materials, or to duplicate product or process innovations initiated
by its rivals. Companies that are run by managers trained in financial analysis but
untrained in the relevant technologies may place too little value on these benefits
because they may seem too difficult to quantify and to justify later if the investment
does not turn out well. They may also fail to distinguish technologies that offer these
important benefits from ones that do not. Although evidence on this point is scanty,
some observers believe that a major reason that many U. S. firms have made poor
decisions about investments in new technologies is that their managers have weak
technical training and place too much emphasis on "objective" but incomplete
financial analyses.

Strategic Investments as Desi gn Decisions

More broadly, there are whole categories of investment decisions which are complicated
by the fact that some of their most important benefits are difficult to quantify. The
most important of these is probably the strategic investments category. Strategic
investments are those whose benefits are likely to accrue to many parts of the
organization, including especially units not involved in making the investment. The
problem with strategic investments is that middle-level managers, compensated on
the basis of their own unit's performance, will be inclined to invest too little in projects
whose benefits accrue largely to other units. 5
DESIC'.'IJ ATTRIBllTEs An important example of a strategic investment decision is a
firm's decision about whether to build a new factory that will use a new technology
to manufacture an existing product. In principle, this investment should be undertaken
if the net present value of the cash flows that the firm expects to receive would be
higher than if it continued to use the old technology. In practice, however, managers
who must defend their estimates are likely to focus on a narrow set of "hard" benefits.
For example, the manager might use an official sales forecast to determine how many
units of the product will be made and how much it will cost to make that number of

� B. Douglas Bernheim and John B . Shaven, "Comparison of the Cost of Capital in the U . S . a nd
Japan: The Role of Risk and Taxes, " Center for Economic Pol icy Research Publication No. 1 79, Sta nford
U niversity ( September 1 989).
1 The idea of stra tegic effects used here corresponds to the idea of positive external effects in market
economics. The �tandard analysis in the latter case is that individuals and firms will be inclined to invest
too little effort in activities that benefi t others but that cannot be sold to the beneficiaries.
 4f>5
un its in a new or existin g factory using either the n ew or old techn ology. This l im ited The Classical
approach can give the wrong an swer if the new techn ol ogy has other importan t Theory of
benefits, such as (1) in creasin g the flexibility of the factory an d thereby in creasin g the Investments and
ran ge of products that can be profitably made in the future, (2) demon strating new Finance
production techn iques, the best of which can be selectively adopted at other plan ts in
the company, (3 ) developin g new skills an d capabilities amon g the compan y's man agers
and workers, which can be effectivel y used in plan ned n ew products, (4) enablin g the
firm to in crease the special ization of its other plan ts, l eadin g to greater econ om ies of
scale, or (5 ) discouraging aggressive growth by competitors, who m ight seize on any
delay or in action by the compan y to expan d their own capacities an d their shares of
the product market. The monetary value of these strategic aspects of the in vestmen t
depends on the firm's plan s: What kin ds of n ew products does the firm plan to
in troduce? What core competencies does the firm plan to develop? What are the key
sources of advan tages that the firm plan s to achieve over its principal competitors?
When the firm's decision about whether to in vest in the new techn ology is
tightly linked to other parts of the compan y strategy, it is on e with design attributes.
The sign ifican ce of these strategic, design attributes is that the ben efits of the in vestmen t
cannot be evaluated in isolation, usin g on ly the day- to-day kn owledge of the product
and the factory available to lower- level man agers. The firm's l on g-term plan s abou t
competition, techn ology, an d new products all mu st en ter as premises in to the
evaluation . As we explain ed in Chapter 4, the optimal way to take in to account all
of the relevan t information in this sort of con text is often for the top man agemen t,
after suitable information gatherin g, to specify the "design premises" or "design
parameters" for the decision. These might in clude the capacity of the new plan t, a
timetable for the proj ect, an d objectives for the degree of flexibility of the plan t.
Subj ect to these premises, local managers may be permitted to make decision s amon g
various alternatives on the basis of techn ical feasibility an d n et presen t value
companson s.

Co!\FLICTING MAN..\GERI..\ L INTERESTS In orgamzmg an y kin d of decision , it is

importan t to recogn ize that the in terests of the lower-level decision makers may n ot
all coincide with those of top man agemen t. For example, if the factory man ager is
compen sated on the basis of factory costs, he or she may n ot be eager to advocate a
new techn ology that leads to higher costs this year in order to reduce the costs at other
factories in future years. The effect could also be in the opposite direction. Division al
managers, who common ly fin d their salaries tied to the size of the division un its they
manage, are likely to spend more time thin kin g abou t the advan tages of large
in vestmen ts in their own un its than about the associated risks. S imil arly, techn ology
managers may seek to ju stify their own j obs by fin din g reason s to adopt new
techn ol ogies. When the choice is between alternative techn ologies an d does n ot affect
the level of total in vestmen t in the division, then these influence costs associated with
the decision are likely to be small. When the future size and importance of the
division is at stake, however, the top man agemen t will often wan t to gather information
from its own in depen den t sources to reduce the information al bias it faces.
An effective system of capital al location, then, mu st meet a diverse set of
chall en ges: It must en sure that maj or in vestmen t decision s are con sonan t with the
com pan y strategy, that they are fin an cially well-ju stified, that the evaluation of
in vestmen t proj ects are n ot excessively tain ted by the personal an d career in terests of
the managers in volved, an d that the process taps the kn owledge of those who are best
informed. These individuals are often the very same people whose personal in terests
are most affected by the decision .
 --156
Finance: CLASSICAL ANALYSES OF
Investments, FII\Al\CIAL STRL'CTL1 RE DECISION
Capital Structure, So far our analysis has focused on how managers make investment decisions inside
and Corporate
the firm. The other part of financial economics concerns how managers can acquire
Control
the funds they need to make investments and how much it will cost to raise those
funds. This cost of capital, expressed as an interest rate, is then used in the net present
value calculation.
Suppose, for example, that management plans to undertake a project that
requires an initial investment of 1 00. Suppose, too, that the returns on the investment
are uncertain and that the whole of the 1 00 must be raised from outside investors.
Management will wonder: If we borrow 50, how much will investors be willing to
pay for a share of the earnings that remain (assuming that first claim on any returns
must be paid to the lenders)? Could the cost of raising the required funds be reduced
if the amount borrowed were greater than 50, with the balance raised by selling shares
in the financial results of the venture? To serve the interests of its existing owners,
the decision should be made in a way that maximizes the amount of money received
for whatever claims on the project returns are assigned to investors. Here \\"e explore
this decision using a conceptual model that is richer in two ways than the one
considered earlier in the chapter: We allow that investments may be risky and recognize
that there are multiple kinds of financing. Also, we focus on companies whose stock
shares are publicly traded.

The Modig liani-Miller Anal yses

For most of our analysis, we suppose that the firm relies on only two kinds of financing:
borrowing from a bank (or from other investors by issuing bonds) and selling shares
of stock in the company. We suppose further that the chief difference between debt
(loans) and equity (shares of stock) is that the firm's lenders must be paid in full before
the shareholders can be paid anything. After the debt has been paid, the owners of
shares in the firm divide any remaining earnings among themselves in proportion to
the numbers of shares that they own. (The box on corporate liabilities discusses some
of the other sorts of financial claims and securities that firms actually use. )
To make things even simpler, let us focus on the case where any loans made
to the firm are perfectly secure. That is, although the earnings of the firm are uncertain
and subject to fluctuation, let us suppose the financial resources are always at least
adequate to repay any outstanding loans, even if very little is left over for the
shareholders. If the total earnings of the firm is some random amount X, and the
amount the firm has borrowed is B, then the shareholders will receive the residual
return X - B( 1 + r), that is, whatever earnings are left after the lenders have been
paid B( l + r)-the principal and interest on their loans. The amount P that the share­
holders will pay to acquire this claim on earnings is denoted by P[X - B( l + r)] to
indicate its dependence on the portion of the cash flow that the shareholders acquire.
The total amount that will be paid by the lenders and shareholders to the firm in
exchange for the firm's promise to pay them each their appropriate parts of the earnings
X is B + P[X - B( 1 + r)] -this is called the firm's value. What we want to know
is how the firm's choice of B affects its value.
To understand the answer to this question, it is crucial to recognize that the
lenders and the shareholders need not be different people. On the contrary, investors
frequently diversify their holdings, spreading their investments among the debt and
equity of various companies as well as government bonds, real estate, and other assets.
To the extent that investors are rational, they do not evaluate investments in isolation,
but as part of their larger portfolio of investments.
 The Classical
Theory of
Investments and
Corporate Liabilities and Financing Finance
Firms have a variety of ways of financing their operations, an d corporations
in particular may issue a rich array of financial claim s again st themselves. I t
is common to speak of debt an d eq uity, but these term s en compass m ore
than j ust simple loans and shares of stock, and actual financial claim s may
mix elem en ts of each.
A broad range of fi nancial claims are labell ed as debt. All share the
property that they are obligations to pay specifi c am ounts or the firm can be
forced into bankruptcy. The simplest are accounts payabl e, bill s from suppliers
that have not yet been paid. M ortgage and eq uipment loan s are secured by
a particula r real asset that can be seized for n onpayment. Simi larly, a firm's
inventory m ight be pledged to secure a loan of operating capital to fi nance
production and distribution in advance of sales receipts. In other cases, an
asset may be pledged as security for a loan even though it has no direct
connection with how the loan proceeds are to be used. For example, the
bank that fi nanced Japanese tiremaker Bri dgestone's purchase of Firestone
Tire & Rubber put a lien on one of Bridgestone' s Japanese factories, giving
it the right to seize the factory if the loan were not paid. S ome companies
also issue u nsecured debentures, which are bonds protected only by the
general creditworthiness and cash flow of the company. Debentures m ay
differ in the priority they are assigned when the company is unable to meet
all its obligations: f unior or subordinated debentures rank behind m ore sen ior
debt in their claims on the firm's assets. Lenders whose loans are secured by
particular assets have a priority in m aking a claim against those assets in the
event of default or bankruptcy.
Equity is distingui shed by the fact that it does not have to be repaid
and so is j unior to debt. It too comes in several varieties. H olders of comm on
stock are the firm's ultimate residual claimants: They are entitled to get what
is left after everyone else is paid. H olders of preferred stock receive divi dends
before any are paid to comm on stockholders, but preferred shares do not
carry voti ng rights. Convertible preferred stock can be traded for comm on
stock at a specified rate.
S om e kinds of debt have eq uity-like features. For example, convertible
bonds can be exchan ged at the lender's option for common stock at term s
specifi ed in the contract. They offer the protection of debt but the op tion to
share in the "upside" if the company does especially well. Because the option
to convert is valuable, these bond s usually carry lower interest rates. Other
eq uity-like securi ties include warrants, which are rights issued by the firm
(often in connection with bonds) to purchase shares of its stock on specifi ed
terms. They differ from call options in that the latter are issued by outsiders
rather th an by the firm itself. (Put options are ri ghts to sell at a specifi ed
price. )
 458
Finance: THE INVESTORS' PERSPECTI V E To gain perspective on how the problem looks from
Investments, the investor's side, suppose there are two similar firms: One issues shares of stock but
Capital Structure, does not borrow at all, whereas the other both issues shares and borr ows some positive
and Corporate amount B. A person who wants to receive, say, the return . 05 · X can accomplish
Control this by buying 5 percent of the shares of the first firm. Alternatively, he or she can
buy 5 percent of the shares of the second firm for .05 · P[X - B(l + r)] and 5
percent of the debt of that firm for . 0 5 · B. The alternative strategy will yield an
income of . 0 5 · B(l + r) from the bonds plus . 0 5 · [X - B(l + r)] from the stock,
for a total income of . 0 5 · X. Because the two strategies would lead to the same cash
income for the individual investor, if anybody is to buy shares of the first firm, the
first alternative strategy must not be more expensive than the second. This implies
P[X] $ B + P[X - B(l + r)]
N ow consider an investor who wishes to receive 10 percent of the returns X - B
(I + r). The investor's first alternative is to purchase 10 percent of the shares of the
second firm. Alternatively, if the investor can use his or her shares as collateral to
borrow on the same terms as the firm, then he or she can buy 10 percent of the shares
of the first firm and then borrow 10 percent of the amount B at interest rate r, leading
to net receipts of 10 percent of X - B(l + r), which is just the same as the first
alternative. If anybody is to buy the shares of the second firm, the alternative portfolio
with the same cash return must not be cheaper:
P[X - B(l + r)] $ P[X] - B
Together, these two inequalities imply that P[X] = B + P[X - B(l + r)] . Evidently,
the value of the firm is the same in both cases; it does not depend on how the firm
finances itself.
The crux of this argument is that the firm's choice of B has no effect at all on
the return streams that investors could acquire by manipulating their portfolios.
C onsequently, it cannot affect either the returns that investors eventually acquire in
constructing their portfolio of investments or investors' willingness to pay for those
returns. The conclusion of this line of reasoning, which was first developed by Franco
M odigliani and Merton Miller, is one of the most celebrated propositions in the
theory of financial markets:
Modigliani-Miller Theorem # 1 . Suppose that the total return X available
for distribution by a firm is unaffected by the firm's financial decisions
and that investors can borrow on the same terms as the firm by pledging
the firm's stock as security. Then the firm's fi nancial structure decisions
cannot affect its value.
I n the classical representation of markets, a firm's financial policy can never,
by itself, affect the firm's value. The firm can borrow massively from banks or issue
j unk bonds, and there will be no consequences in terms of the value of the firm.
Something else, besides the simple workings of classical markets, must account for
the effect that fi nancial stru cture seems to have on what investors are willing to pay.
T \XES, BANKRllPTCY, AND CAPITAL STRUCTUR E The M odigliani-Miller (MM) theorem
itself directly suggests several possibilities. In some countries, debt and equity are
treated differently for tax purposes. M ost notably, interest paid by firms to lenders in
the U nited S tates is treated as an expense that is deductible for the firms in determining
the base for their corporate in come tax, but dividends paid to individual shareholders
are n ot. This gives a very substan tial advantage to debt financing in the U nited States,
but it can not explain the use of debt financing in the era before it had any tax
 459
advantages or in other countries where the tax laws treat the two forms of financing The Classical
symmetrically. In the U. S. context, where the corporate income tax rate hovered Theory of
around 50 percent for half a century until it was cut to 34 percent in 1 986, the tax Investments and
advantages of debt led some to wonder why firms issue any shares at all. One possible Finance
answer is that firms that have too much debt incur an increased risk of bankruptcy,
and that bankruptcy destroys value. However, we shall find that the richer conception
of financial structures given in the next chapter provides a more convincing account
of the purposes of debt and equity.
DI VIDEND POLICY IN THE CLASSICAL MODEL The ideas used in the Modigliani-Miller
analysis can be applied to other financial decisions besides the choice between debt
and equity. An argument very like the one presented earl ier can be applied to all the
diverse and complicated financial instruments that are used in modern financial
markets, including options, warrants, callable bonds, convertible bonds, and so on .
In each case, the argument is that the investor can put back together the parts of the
return that the firm has separated . The investor does that by purchasi ng some of each
kind of financial instrument that the firm issues. 6 Such arbitrage arguments (ones
based on the requirement that equilibrium securi ty prices do not leave room for surely
profitable trades that take advantage of pricing discrepancies among assets) form the
crucial logic in financial economics.
The same ideas can also be applied to analyze a firm's dividend policy, provided
we continue to assume that the firm's investment decisions are given and thus too its
(random) cash flow. In this case, the funds for the dividend must come essentially
from issuing new securities. Suppose then that a firm pays a dividend D that it finances
by issuing shares. Shareholders who prefer not to consume this dividend can undo
the firm's decision by using the dividend to buy additional shares. If the firm finances
the dividend by borrowing, then the investors can use the dividend to purchase debt
of the firm, effectively neutralizing the impact of the firm's decision on their own
financial returns. Just as in the case of the firm's debt-equity decision, we find again
that the investor can undo or neutralize the effect of the fi rm's dividend decisions on
his or her own financial returns . Because the investor's opportunities do not depend
on the firm's dividend policy, the amount he or she is willing to pay for his or her
investments also does not depend on the firm's policy. This conclusion about dividend
policy is regarded as so important that it is the basis of a second major MM proposition:
Modigliani-Miller Theorem #2. Suppose that the total return X available
for distribution by a firm is unaffected by financial decisions and that
investors can buy and sell securities on the same terms as the firm. Then
the firm's policy for paying dividends can have no effect on its value.
The MM theorems provide a useful point of departure for thinking about
financing and dividend policies. Their essential message is that a firm can no more
change its value by divvying up its earnings differently than increase the weight of a
cake by cutting the pieces up in different sizes and shapes. Only by changing the
ingredients, that is, by changing the total return X that is available for distribution,
can the firm create value.

The Allocation of Investment Capital by Markets

The preceding analysis leaves us with something of a puzzle. Champions of free
markets sometimes claim that the capital markets funnel limited investment funds to

" Alternatively, if the firm has failed to take the pieces of return apart by issuing a variety of financial
obligations, then investors can do that for themselves, by issuing derivative securities, as real investors
 460
Finance: the most pr oductive companies by maki ng it easy for those compani es to raise capi tal
I nvestments, while penalizing companies that perform poorly. The puzzle is: How can capi tal
Capital Structure, markets direct capi tal at all if the firms' analysi s of i nvestments is separate from the
and Corporate analysi s of fi nanci ng?
Control In actual i ty, there is nothi ng i n the classical theory of markets to j ustify these
claims for capi tal markets. In the classical theory and i n life as well, so long as the
securi ty offered for a loan is adequate, the lender has no in terest i n ensuring that the
funds themselves are well i nvested. For example, the willi ngness of a bank to lend might
depend on the quali ty of the collateral offered or on the overall prospects of the borrower.
If the collateral and overall prospects are strong, then the bank's deci si on would not
depend at all on its i ndependent evaluati on of the planned new i nvestment project.
A variati on of the same argument begi ns from the premi se that firms that have
performed well and that have the prospect of conti nui ng to perform well wi ll have
higher share prices. 7 From this premi se, the adv ocates claim that firms wi th profi table
i nvestments fi nd it cheaper to rai se funds, but thi s conclusi on is wi th out any warrant
i n l ogi c or reali ty. Share pri ces for a firm depend on the average profi tabili ty of i ts
i nvestments because the shares are cl aims on the whole cash fl ow of the firm, not the
returns from any one of i ts i nvestments. H owever, the questi on of how new capi tal
should be allocated depends entirely on the returns to new, incremental i nvestments. 8
When correctly i nterpreted, the classical theory of capi tal markets holds that
i nvestment funds are allocated effici entl y throughout the economy precisely because
all the profi t-maximizing firms face exactly the same cost of capita l for any i nvestment
project. It is for this reason, accordi ng to the theory, that worthwhile proj ects are
undertaken regardless of the investor's iden tity and proj ects that are not worthwhile
are rej ected. In the classical the ory, the firms that i nvest most are not those wi th the
lowest cost of capi tal but those wi th the largest number of attractive (posi tive net
present value) investments to make.

IM'ESTl\ l ENT RISK AND THE COST OF CAPITA

In pri nci ple, the proper i nterest rate to use for evaluati ng an i nvestment is the actual
cost of rai sing money to fi nance that particular i nvestment. By an extensi on of the
M odigliani-Mi ller analysi s, a firm cannot create value by undertaking a project that
i nvestors would be unwilli ng to fi nance on i ts own because the total market value of
the firm is j ust the same as if a separate company were formed to undertake each of
the firm's proj ects. In the classical fi nancial model, the returns on a proposed
investment are high enough only if they would be high enough to be attractive to
"the market," that is, to i nvestors who buy and sell shares of the vari ous fi nancial
assets exchanged i n securi ti es markets. Determi ni ng the rate of return that the market
will demand requires taki ng a closer look at the way sophisticated investors evaluate
ri sky i nvestments.

Risk and Return

An i ndivi dual i nvestor may face many possible investments: stocks, bonds, real estate,
mutual funds, bank accounts of vari ous ki nds, fi ne art, tradi ng cards, and so on.

sometimes do. The description of these, however, would take us too far afield. For more details, see, for
example, William Sharpe and Gordon Alexander, Investments (Englewood Cliffs, NJ: Prentiee Hall, 1 990).
7 Though reasonable, the premise is hardly indisputable. See the following discussion of the efficient
market hypothesis.
8 Th is same concern of investors also leads them to gather information about average returns on
existing projects, rather than about returns on new projecb, to the detriment of the capital allocation
system.
 46 1
Some of th ese are quite safe but offer return s that have, at least historically, been The Classical
much lower than those available from riskier i nvestments. Other kin ds of i nvestments Theory of
have had higher average rates of return in the past but have sometimes incurred huge Investments and
losses for i nvestors. Within any single class of risky investments, there can be enorm ous Finance
variation regardi ng the actual returns. For example, in the class of high-technology
small company stocks, the value of shares in a small computer company may be rising
at j ust the same time that the value of shares in a software systems company is falli ng.
Given these risks, what should the investor do?
It would be a mistake for the investor to pose the question as: Which investments
are most attractive? This phrasing of the question is too limiting. By combining
investments appropriately, the investor can create a portfolio (collection of assets) that
can achieve the same level of expected return as any si ngle asset but with a less
variable pattern of returns. Instead, the investor should ask: What portion of my total
portfolio should be invested in each of the various kinds of assets?
l\lEAN-V AHIANCE ANALYSIS The traditional analysi s of thi s question is based on three
main assumptions. First, the investor dislikes risk and likes high expected levels of
return, where the risk is measured by the variance of the investment returns or their
standard deviati on (the square root of the variance). These are j ust the sort of
preferences assumed in the analysis of incenti ve contracti ng in Chapter 7. S econd,
there exists a ri skless bond payi ng a safe rate of return r. Finally, capital markets are
perfect; that is, an i nvestor can buy or sell as much of any security as he or she likes
without affecting the security's price and without incurring signifi cant buying or selling
costs.
The basic data of the investor's problem are taken to be a set of / assets of which
shares are traded. A typi cal asset i yields an uncertain rate of return R;, These returns
are usuall y expressed i n percentage terms, like interest rates.
In the standard mathematical representation, a portfoli o is defined as a list
(vector) of numbers (a 1 , . . • , a1 ) indicating what fraction of the total fu nds available
is to be invested in each asset. These fractions must, of course, add up to 1 .
(L� = 1 a; = 1). In real financial markets, it is possible (within limits) to promise future
delivery of assets that are not owned today but will be bought later to fu lfill the
contract. This practice is called short selling; it amounts to having some of the a/s
be negative. With or without short sales, the investor' s realized return from the
portfoli o i s equal to L� = 1 a1R;, I t is the mean and variance of this return that the
investor cares about.
Fi gure 14. 1 depicts the set of combinations of mean returns and standard

E A
--
-­
- ____....,.-
--
--
::::, A
Q)
a: r _.,,,,,.--
-
B
I
C I
I
� I
I
I
I
I
Figure 14. 1: Points on the line through M, r,
SA and C show the highest possible mean return for
Standard Deviation any level of risk.
 462
Finance: deviations that the investor can attain by choice of some portfolio. The shaded set
Investments, shows the combinations that can be attained by portfolios of the risky assets alone.
Capital Structure, Each point in this region represents the return and risk from some single asset available
and Corporate in the market or from a portfolio of assets. The point on the vertical axis at height r
Control is the risk-return pattern given by investing only in the riskless security. The figure is
drawn assuming that there are investments that have higher expected returns than r,
although these returns are risky.
By investing a fraction 1 - S of his or her wealth in the riskless bond and the
remaining fraction S in the risky portfolio A with mean A and standard deviation SA ,
the investor creates a new portfolio with mean ( 1 - S)r + SA and standard deviation
( 1 - S) · 0 + SSA = SSA. 9 For example, the risk-return combination B can be
attained by investing one third in portfolio A and two thirds in the riskless bond. If
the investor can borrow at interest rate r, then S can take on values larger than 1. For
example, the point C can be attained by borrowing at rate r and investing all the
proceeds in portfolio M.
How should investors combine risky and riskless assets to assemble a portfolio?
As the figure illustrates, for any level of standard deviation, the highest mean return
obtainable by any portfolio lies on the solid line connecting M to the riskless rate r.
This rate of return is achieved by investing a fraction of the available funds in the
riskless bond and the remaining fraction in the portfolio M. This leads to a key
conclusion of portfolio theory:

The One-Fund Portfolio Theorem. For any investor who cares only about
the mean and variance (or standard deviation) of returns, the optimal
portfolio consists of some mix of the riskless bond and a portfolio of risky
assets that has the expected return and standard deviation associated with
the point M in Figure 1 4. 1 .

The one-fund portfolio theorem gets its name from the fact that if there were a
mutual fund whose portfolio had the risk and return profile corresponding to point
M in the figure, then no investor need ever invest in any asset except that mutual
fund and the riskless bond. In the analysis that follows, we use RM to denote the
(uncertain) returns on the one fund.

. \ P R lctl\C FoR \ l l 'L \ An important implication of the theorem just described is a

pricing formula indicating how the returns from buying any of the assets available in
the market must be related to RM. Let R denote the uncertain returns on any asset.
Then its excess return, R - r, which is the expected return on the asset in excess of
the riskless rate of interest, is given by the following equation:

( 14. l )

Intuitively, the expected return on any investment in financial securities can be

decomposed into a riskless rate of return plus an excess return, which is to be
understood as compensation for the risk that the investment entails. Using the expected
rate of return suitable for the asset, the appropriate asset price can be determined by

9
To verify this mathematically, if the return on the portfolio is R, note that E[BR + ( l - 13),] =
( 1 - 13), + 13.E[R] and Var[( l - 13), + BR] = 132Var(R) = 132S}.
computi ng the presen t value of expected cash flows. I n a one-peri od mode l, today's The Classical
price is equal to tomorrow's expected return di vided by (1 + R ). For example, if the Theory of
expected returns tomorrow are $ 1 32 and R is 1 0 percent, then today's appropriate Investments and
asset price is $ 1 20 (= $132/ 1 . 1 0). Because the rate of return R determines the asset Finance
price, Equati on 14. 1 is called a pricing formula.
The prici ng formula may i nitially seem surprisin g. Given our other assumptions,
you might have expected that the risk for an individual i nvestment would be measured
by somethi ng like i ts variance, Var(R). Thi s is not the case, however: The excess
return to compensate for risk does not depend on Var(R) at all. I nstead, the
compensati on is proporti onal to Cov(R, R M ), the covariance of the return to the
security with that of the fund; 10 thi s is the only term on the right-hand side of the
equation that depends on the i nvestment R.
The mystery is resolved when you remember that the i nvestor is concern ed not
with the ri sk associ ated wi th i ndividual i nvestments, but wi th the risk associated with
the portfolio a� a whole. For example, if an asset has a negative covari ance, i ts returns
tend to be hi ghest j ust when the returns on the other assets i n the portfoli o are lowest.
When added to the portfoli o M, such an asset tends to reduce the variabili ty of
portfoli o returns. Ri sk-averse investors will regard this reduced variability as something
valuable and will be willing to accept an even lower return on the asset than they
would demand from the riskless bond, regarding the difference as a kind of i nsurance
premi um paid for reduced variabi li ty of the returns to the overall portfoli o. Similarly,
when Cov(R, RM ) is large, the asset is one whose returns fluctuate signi ficantly and
tend to be hi gh when the rest of the portfolio is enj oying hi gh returns and to be low
when the rest is suffering low returns. H oldi ng such an asset tends to amplify swings
in portfoli o returns, maki ng the whole portfolio ri skier. I nvestors will purchase such
an asset only if i ts expected returns are high enough to compensate for these swi ngs.
N otice how different these conclusi ons are from what you would obtain if you
had made the easy mi stake of evaluati ng the assets i ndividually. From such a
perspective, assets whose returns have the same variance would appear equally risky,
and it would appear that the i nvestor should d emand a correspondingly high excess
return. Evaluated as part of a portfoli o, the two assets are quite different, and the
excess return on the asset with the negative covariance should actually be negative.
The portfolio perspective is a necessary part of any sound investment theory.

Mathematical Proof of the Pricing Formula

Gi ven the uncertain return R wi th mean return R , defi ne B to be the number
tµat solves BR M + (1 - B)r = R . Fi ndi ng B amounts to finding the combi nati on
of the fund and the riskless bond that has the same expected return as R. (I f R
exceeds R M , as it might if the asset has hi gher risk than the fund and so
corresponds to a point in the shaded regi on above and to the right of M, then
B must exceed 1. Thi s means that the coefficient on r is negati ve and that the
desired combi nati on entails selli ng short the riskless securi ty. ) The mean and
standard devi ati on of this new portfoli o li e on the solid line through M i n Fi gure
14.1. B y inspection of the fi gure, i t follows that this portfoli o has the lowest
possible variance for the gi ven level of return, R .

1 0 Recall that the covariance of two random quantities, x and y , is the expected value of the product
(x - xXy - y) and is a measure of how the two move together (see the appendix to Chapter 7).
 464
Finance: Consider now a composite portfolio consisting of a fraction 'Y invested in
Investments, asset R and a fraction 1 - 'Y invested in the fund BRM + ( 1 - B)r. Because
Capital Structure, each component of this composite portfolio has expected return R , its expected
and Corporate return is -yR + ( 1 - -y)R = R . Consequently, its variance must be at least as
Control high for all values of 'Y as the variance of the portfolio BRM + ( 1 - B)r. That
is, the variance of the returns on the -y-weighted portfolio must be minimized
when 'Y = 0. The variance is:
Var['YR + ( 1 - -y)(BR"1 + ( 1 - B)r) ] ( 14. 2)
= -y 2Var(R) + ( 1 - -y) 2 B2 Var(RM ) + 2-y( l - -y)BCov(R, RM )
The minimum is calculated by taking the derivative with respect to 'Y and
requiring that it be equal to 0 when 'Y equals 0. When the resulting expression
is solved for B, it yields:
Cov(R, RM )
B= (14. 3 )
Var(RM )
However, B was initially defined so that R - r = B(R M - r). Hence,
- _ = Cov(R, RM ) - _
R r (R r) ( 1 4. 4)
Var (RM ) M
which is just the pricing formula that was asserted earlier.

The Capital Asset Pricin g Model

The significance of the pricing formula just derived is that it holds the promise of
resolving the question about how risk should be accounted for in business investment
decisions. The pricing formula expresses the expected rate of return R on a new
project in terms of r and R M. This is the cost of capital to use in computing the
present value of the new investment, rather than, say, some rate of return corresponding
to the overall collection of assets in the firm. For this formula to be useful, however,
you must be able to identify the portfolio corresponding to M. The first and most
prominent attempt to identify M is the Sharpe-Lintner capital asset pricing model
(CAPM, pronounced "cap-em").
Suppose all investors were to invest in a way consistent with the one-fund
portfolio theorem. Because the investors together own all the assets, if each owns a
(possibly different) fraction of the same fund, then each must own a fraction of the
portfolio of all assets, so the point M corresponds to the market portfolio, or the port­
folio consisting of all the financial assets. Even when investors hold different portfolios
of risky assets, all consistent with the point M, it is still true that the market portfolio
is among those corresponding to M, so it can be used for the calculations in the
pricing formula.
The number B, computed according to Equation 14. 3 using the return on the
market portfolio for RM, is known as the security's beta among financial economists
and sophisticated investors. According to the CAPM, the excess return on a security
should be proportional to its beta computed in this way. This predicted relationship,
known as the security market line, is depicted in Figure 1 4. 2. The intercept is at r,
the risk-free interest rate, because the beta of a riskless bond is zero. The line is
upward sloping because higher values of B mean that the security adds risk to the
portfolio and so must bring returns above r to be attractive enough to hold. The actual
slope of the line is determined by the total risk tolerance of investors and the overall
variance of market returns.
 465
Expected
Retu rn The Classical
Theory of
Investments and
Finance

Figure 14. 2: The security market line. The ex­

pected return depends linearly on beta and not
at all on the variance of the security's own
0 Beta (�) return.

THE FIRM'S EVALUATION OF RISKY I NVESTMENTS For investment decisions by individ­

ual firms, the signi ficance of the CAPM is that it identifies a theoretically correct way
to account for the riskiness of its projects in computing present values. According to
the CAPM, the only measure of risk that matters is the beta. The parts of the
randomness or uncertainty in the returns on a project that are idiosyncratic or
unsystematic, that is, that are unrelated to variations in returns for the market as a
whole (the systematic risk), are irrelevant according to the CAPM for determining
the interest rate used to evaluate investments. They will thus not need to be paid for
it, nor should it be considered when evaluating investments because investors will not
actually bear it.
There are two important caveats to add to the theoretical proposition that
idiosyncratic risk does not matter. Both involve recognizing that even though such
risks may not matter directly to outside investors, they can have important consequences
within the firm. One way this can happen is when the firm uses performance incentives
to motivate its managers. Like any other source of risk, idiosyncratic risk in an
investment project is a part of the project return that is outside the manager's control.
All such risks make it harder to evaluate a manager's performance and more difficult
and costly to motivate superior performance. Worse, unlike systematic risks that affect
all similar projects, idiosyncratic risks cannot be mitigated by using comparative
performance evaluations. It is proper for firms to account for these costs in evaluating
potential investments.
Second, if the project can be abandoned or scaled up as uncertainty about the
investment resolves, then uncertainty can actually increase value. For example, doing
research into synthetic fuel (synfuel) technology may be uneconomical if energy prices
follow their expected pattern over time. If there were no uncertainty about energy
prices, this research should not be undertaken. In reality, however, there is uncertainty
about future energy prices, and that may make the synfuel research investment
worthwhile. With increased uncertainty, investments that increase the decision maker's
options may have great value, even if the individual options are valueless under the
most likely scenario.
The same effect is present for individual investors who own warrants or other
rights to purchase shares of stock at a preestablished price. These rights might be
valueless if the shares do not fluctuate much in their value but could become quite
valuable if share prices are highly uncertain because this uncertainty increases the
chance that the options can be exercised at a large profit.
CRITICISl\ts AND EXTE.NSIONS OF THE CAPM The CAPM was the first sophisticated
theory of how asset risks affect security returns, but it is hardly the last word on
 466
Finance: the subj ect. The theory has been subj ected to a variety of criticisms that h ave pro­
Investments, vided th e impetus for variations of the th eory and for the emergence of competing
Capital Structure, theories.
and Corporate One maj or criticism is th at because the set of all traded assets includes some
Control wh ose returns are not readily observed, like h ousing and other real estate, the pricing
formula using the market portfolio for RM is not testable using available data. This is
a cogent objection to th e theory, mainly because the needed data are unlikely ever to
be available.
Oth er criticisms of the CAPM focus on the assumptions of th e theory. What,
in reality, plays the role of the riskless bond? Even a government bond that carries
no chance of default is risky if it is denominated in nominal (money) terms-as is
almost always the case-because unpredictable changes in the price level make the
real value of the returns uncertain. What h appens if some investors consider their
human capital as a risky asset in evaluating their portfolio? This is certainly a rational
and reasonable thing for many people to do because, for many, the h uman capital
acquired through years of education and experience is the most valuable asset they
own. What if some investors do not behave rationally, in the form assumed by the
theory? The argument th at RM corresponds to the market portfolio assumes that all
investors invest rationally.
Economic researchers h ave continued to explore all these issues and h ow they
affect the conclusions of theories, but no new consensus h as yet emerged.

Ex pectations, Asset Pricing, and Efficiency

The CAPM and most other pricing theories place great reliance on the mathematical
notion of expected rates of return. When we try to match the model with the world,
h owever, a serious issue arises: Wh ose expectations sh ould be used in applying the
model? Expectations about h ow firms will perform are h ardly uniform among investors.
Managers of firms h ave their own expectations about h ow their decisions will turn
out, and these are frequently more optimistic than th ose of outside investors.
To give the theory empirical content, we use the metaph or of the market as a
single investor wh ose expectations are a kind of weigh ted average of the expectations
of real investors, with the weights depending on the amounts invested by different
individuals. Empirical tests of the theory pr oceed by investigating whether the market's
expectations provide a reasonable and accurate estimate of the fu ture returns of the
firm.
Th ough the wh ole issue of the market's expectations may seem abstract, it
actually h as an immedi ate, practical importance for both business and public policy.
If share prices are truly determined according to the pricing formula where the market' s
expectations are based on sound analysis of everything that investors and managers
may know, then the cost of raising capital for various investment proj ects will be
similarly well founded and firms will be encouraged to undertake the righ t investments.
Managers will not fear that good investments made today in the h opes of a long-term
return will depress the market price, h arming current sh areh olders and making the
company more vulnerable to takeovers. Also, as we discussed in Chapter 1 3 , when
stock prices are an accurate indicator of value, stock price changes can be an effective
basis for evaluating executive performance and determining executive compensation.
Conversely, if the expectations reflected in market prices are not well founded, then
even the best-managed companies may be among th ose most subj ect to takeover
attem pts, an d compensation based on stock market prices may encourage executives
to man ipulate in vestor expectations rather than to create value. For all these reasons,
 467
it is of great importance to determin e j ust how well share prices mirror the fu ndamental The Classical
values of firms. Theory of
Investments and
II\FOH!\IATION AND Tl I E PR ICES OF FINANCIAL ASSET,' Finance
What an investor expects depends on what the investor knows. For example, a drug
company employee who kn ows that the company is about to an nounce the discovery
of a new AIDS vaccine, or that next quarter's income statement will show sales and
profi ts larger than anyone had forecast, or that the company has been approached
about a merger wi th a larger company at very attractive terms, would normally expect
a higher return to holding the company's shares than would other investors who lacked
this information. The question of how well share prices mirror fundamental values
can thus be decomposed into two parts: What information does the market use in
valuing firms, and how accurately does it use that information?

Forms of the Efficient Market Hypothesis

The most optimistic answers to these questions are the various forms of the efficient
markets hypothesis. All of these affirm that securities' pricing depends on investor
expectations and that investors make good use of available information in forming
these expectations. They differ about what information, at a minimum, is used in
determining those expectations. The weak form efficient markets hypothesis holds that
the relevant expectations are based at least on information about past and current
stock market prices. In other words, an investor who knew only the pattern of past
prices for shares of stock in various corporations could not, on that basis alone, pick
stocks that would on average have higher excess returns than predicted by the pricing
formula. The ability to make such choices would be evidence against the theory; it
would mean that market prices do not accurately reflect the best predictions of value,
using all the information contained in the present and past prices. If the weak form
efficient markets hypothesis is correct, then those who advocate technical analysis­
using charts based on past price movements to predict future price movements-are
mere charlatans whose predictions should not be trusted. There should be no patterns
to price changes over time, so that the best estimate of future prices is just today's
pn ces.
Two stronger forms of the hypothesis have also been proposed. According to
the semistrong form efficient markets hypothesis, the market's expectations are those
based on a set of information that includes, in addition to the past prices, everything
else that is public information, such as fi nancial statement information and press
reports about business events (earnings announcements, new products, takeovers,
changes in the executive suite, and so on). The semistr ong form of the hypothesis
does not, however, require that information that is not generally known be refl ected
in prices: In the example of the AIDS vaccine, the price of the firm's stock woul d not
necessarily move when the discovery was made, but only when it became publicly
kn own. At this point, however, prices shoul d adj ust rapidly to incorporate the new
information. According to this form of the hypothesis, it is impossible for an investor
who tries to pick stocks using onl y past prices plus earnings reports and similar public
information to form a portfolio that will on average perform better than predicted by
the pricing formula. In this case, a strategy of buying and holding a diversified, market
portfolio is optimal .
Finally, the strong form efficient markets hypothesis posits that market prices
reflect even information that is available only to executives and managers inside the
firm but that has not yet been publicly announced. This might happen , for example,
 468
Finance:
Investments,
Capital Structure,
and Corporate
Insider Trading and Market Efljciency
Control
Insi der trading is co nsidered unlawful in the United States, and i n recent
years the Securities and Exchange Commi ssion (SEC) and federal prosecutors
have bro ught criminal (rather than j ust ci vil) cases against alleged insider
tradi ng. U nfortunately, just what constitutes i nsi der tradi ng has never been
specified i n the law or i n government regulations. The basic idea, however,
is that those whose positions give them access to private information relevant
to forecasti ng stock retur ns and prices sho uld not be allowed to trade on this
informatio n. For exampl e, a corporate officer should not be allowed to buy
the firm's stock after seei ng unexpectedly good earnings reports that have not
yet been released or, perhaps worse yet, tell shareholders that earni ngs are
do wn when they are actually good and then take advantage of the temporary
fall in the stock price to buy shares. (It was j ust such an i nstance in 1942
that led to the SEC's adoption of its Rule I 0b-5, o n which insider tradi ng
cases are based.)
Most of the disc ussio n in busi ness and government about insider trading
has to do with exactly what behavior by whom sho uld be forbidden. However,
there is also an active debate among legal, fi nancial, and economic scholars
abo ut whether any restrictions should be placed on i nsider tradi ng at all.
This debate was initiated by Henry Manne in the mid- 1 960s and continues
to thi s day.
There are basically three arguments agai nst prohibiting insider tradi ng.
First, allowing insider trading should mean that inform ation will be embodied
i n market prices more quickly, and so security prices would become better
reflectors of fundamental value. Seco nd, if firms want to compensate their
employees by letti ng them trade on i nside information, why not allow them
to write these contracts? The logic here is that the information belo ngs to
the firm, which should be able to determ ine how it i s used. Third, insider
trading is apparently very widespread anyway, the prohibition is unenforceable,
and actual enforcement is capricious and unfa ir. The evi dence for thi s last
point is the frequency with which stock prices start movi ng before information
is publicly released but after it becomes available to insiders. This pattern is
so common that scholars doing "event studies," measuring the market reaction
to events and announcements, automatically include several days before the
information is offici ally made public i n the studies: Otherwise, they wou ld
miss much of the reaction.
The counterarguments are often couched in fairness terms, but i nsi der
trading can be criticized on efficiency grounds as well. First, the most that
allowing insider tradi ng on the basis of insider informatio n might accompli sh
is to make prices reflect informatio n a few days earlier, which has only tri\'ial
consequences for the efficiency of investment decisio ns. More significant is
the efficiency-destroyi ng adverse selection problem that insider tradi ng creates.
If ordi nary traders i n the market are deali ng with i nformed insi ders, they face
a no-win situation: The i nsiders will trade only when their private information
i ndicates that being on the other side of the trade is a losing proposition.
This cou ld deter others from maki ng trades, leading to le ss trade and more
difficu lty in executing trades quickly. Efficiency wou ld hardly be enhanced
by that.
 469
if informed insiders-like a drug compan y employee who kn ows about the AIDS The Classical
vaccine-trade on the information themselves, or if the in side information is leaked Theory of
to an alysts or to friends an d fam ily members of compan y in siders. Their buyin g o r Investments and
sellin g based on this information will tend to move stock prices to reflect the Finance
information an d its effects on fu ture return s. The box on in sider tradin g explores the
soc ial-efficiency implications of this kind of activity.

I low PRICES Col\JE TO INCORPORATE INFOHl\lATION The theoretical argumen ts in favor

behavior. If the expectation s man ifested in the prices did n ot accurately reflect available
of the efficien t markets hypotheses are largely based on a simple theory of in vestor

information about future return s, then in vestors who used the available information
would be led to buy the stocks with the highest expected return s relative to their
prices. This extra deman d would drive the stock price up an d hence drive down the
rate of return. S imilarly, in vestors would sell over- priced stocks, drivin g their prices
down an d their expected rates of return up. Provided there are en ough sophisticated
in vestors who pay atten tion to the relevan t information an d act upon it, prices can
never be too far out of line, or so the theory goes.
A careful study of portfolio theory provides on ly partial support for this argumen t.
It is true that other thin gs (such as the beta) being equal, investors will ten d to buy
more shares of those stocks on which they expect a higher rate of return . Deman d by
sophisticated in vestors will ten d to push prices toward levels con sisten t with their
expectation s. H owever, the extra deman d from sophisticated in vestors will only be
proportion al to the expected extra return . The pricin g error that remain s will depen d
on the fraction of investors who are kn owledgeable an d sophisticated an d on the risks
they incur in reducin g the diversification of their portfolios to take advan tage of the
pricin g error. The market's expectation s remain a weighted average of the expec tation s
of all in vestors, weighin g the expectation s of sophisticated investors (only) in proportion

If the vast bulk of market trades are made by sophisticated in vestors, then prices
to the magn itude of their in vestmen ts.

would reflect their well- informed expectation s. S uppose, however, a sign ifican t fraction
of the tradin g volume is by market participan ts who buy an d sell n ot because their
ration al forecasts in dicate that asset prices are misaligned but because they are followin g
the advice of some television or n ewspaper in vestmen t advisor who tells them when
and how to trade based on "fundamen tals, " or because they are followin g (or runnin g
counter to) the herd of other in vestors, or perhaps because they are tryin g to behave
like the "smart mon ey" but simply have the models wron g. Then , the weighted average
may n ot reflec t the best available information an d may be more volatile than is
j ustified by chan ges in un derlyin g econ omic con dition s. Fads, booms, an d sudden
crashes can be the result. To distin guish between these c ompetin g accounts of how
prices are determin ed, we must turn to an assessmen t of the evidence.

Evidence on the Efficient Markets Hypotheses

In the first years after the efficien t market hypothesis was in troduced, the custom was
to test it by focusing on short- term price changes. The return to a stock earned by an
investor over an y period of time con sists of the dividen ds paid in that period plus the
increase in the stock price. This is con verted in to a rate of return , an d then a statistical

rate in the precedin g periods. If the weak form efficien t markets hypothesis holds,
analysis is performed to see if the rate of return in an y given period depen ds on the

there shoul d be no such depen dence: The c han ge in a stock price today should be
uncorrelated with its chan ge yesterday. When it was in troduced, this was regarded as
a bold predic tion, but the repeated fin din gs of a series of studies were con sisten t with
 470
it. 1 1 Similarly, studies of the semistrong form effi cient markets hypothesis failed to
fi nd any public information that was useful for predicting short-term price movements. 1 2
Finance:
Investments,
Capital Structure, Evidence that market prices reflected more than j ust public information (as
and Corporate predicted by the strong form efficient markets hypothesis) was found by using a
Control technique called event studies. (These involve using CAPM or some other asset pricing
model to estimate how a stock's price would have moved over some period in the
absence of new informati on and then comparing this to its actual behavior during a
period in which some event, such as an information release, occurred.) For example,
if the earnings reported by a company were less than had been forecast by stock
analysts or by the company itself, then you might expect the news to cause the price
of the company stock to fall relative to the market as a whole. Evidence, however,
shows that although low earnings do lead to lower prices, much of the price decline
occurs in the few days before the earnings are publicly an nounced. This is typically
the period between the ti me that the first draft of the earnings report is prepared by
company employees and the time of public announcement. 13 The run-up of stock
prices before merger announcements is also suggestive of inside information being
reflected in stock prices.
As a result of these studies, by the late 1970s, there was substantial consensus
among fi nance scholars on the general validity of at least the weak form efficient
markets hypothesis, a belief that one of the semi-strong form migh t also hold, and
some willingness to consi der the strong form.

NEC.\Tl\'E EYIDENCE Two papers published in 1981 shattered the consensus and
raised serious questions about the consistency of the effi cient markets hypothesis with
the observed behavior of aggregate stock prices. 1 4 These papers appeared to show that
variations in stock prices were much too large to be explained as responses to changing
expectations about future dividends. Later statistical analysis of longer-term price
movements sh owed large negative correlations of prices over periods of th ree to five
years that were inconsistent with the elementary versions of the theory that had been
used in earlier empi rical studies. 1 5 The most recent econometric studies tend to
support the view that the Weak Form Efficient Markets Hypoth esis is not fully
consistent with the evidence, but that the deviations from pricing effi ciency are not
so great as to contradict the hypothesis "grossly. " 1 6
The worldwide stock market crash of October 19, 1987 gives perh aps an even
more cogent reason to question whether the hypothesis is valid. It is still hard to see

1 1 See Eugene Fama, "Efficient Capital Markets: A Review of Theory and Empirical Work, " Journal

of Finance, 25 (May 1970), 383-417.

12 This research is surveyed in Burton Malkiel, "Efficient Markets Hypothesis," The New Pa/grave:

A Dictionary of Economics, J. Eatwell, M. Milgate, and P. Newman, eds. (London: The Macmillan Press,
1987), Volume 2, 120-23, and, more completely but also mcire technically, in Stephen LeRoy, "Efficient
Capital Markets and Martingales , " Journal of Economic Literature, 28 (December 1989), 1583-162 1 .
1 3 See George Foster, Financial Statement Analysis (Englewood Cliffs, N J : Prentice Hall, 1986),
380. A confounding element in this is that the information may appear to become publicly available before
it appears in The Wall Street fournal, but researchers often take the date of the Journal story carrying the
news as the release date for their studies.
1 4 Stephen F. Leroy and Richard D. Porter, "Stock Price Volatility: Tests Based on Implied Variance
Bounds, " Econometrica, 49 (198 1 ), 5 5 5-74; and Robert J. Shiller, "Do Stock Prices Move Too Much to
be J ustified by Subsequent Changes in Dividends?" American Economic Review, 71 (June 1981), 421-36.
1 1 Eugene Fama and Kenneth French, "Permanent and Temporary Components of Stock Prices,"
fournal of Political Economy , 96 (April 1988) , 246-73.
16 N. Gregory Mankiw, David Romer, and Matthew Shapiro, "Stock Market Forecastability and
Volatility: A Statistical Appraisal, " Review of Economic: Studies, 58 (May, 1991), 4 5 5-78.
 47 1
what news of the day could have justified a downward revision by almost a third in The Classical
expected future returns and thus in stock prices between Friday and Monday. Theory of
I nvcstmcnts and
Shortsig hted Markets and Shortsi ghted Management Finance
The first kind of criticism of the efficient markets hypothesis was a criticism of its
empirical validity. There is evidence to challenge the assertion that the semi-strong
form or even the weak form efficient markets hypothesis is true. The second kind of
criticism applies regardless of the evidence on these matters: Unless the strong form
efficient markets hypothesis is also true, maximizing stock market value may not be
a proper objective for the firm's managers. In general terms, when the strong form of
the hypothesis is not true, the firm's own executives have better information about
the value of various alternatives than do investors, and it would contribute to efficiency
if they could be induced to use that information in selecting among activities for the
firm, regardless of how those activities are accounted for in the market price. This is
a special problem for the many firms that explicitly compensate top executives on the
basis of share-price performance, but it is also a problem for other firms where
executives may believe that their job security depends on good share price performance.
Generally, investors find it easier to gather and evaluate information about some
determinants of the firm's fundamental value, such as short-term earnings performance
or the sales of existing products, than about others, such as the market potential of
new products or the stage of development of a secret new product. Market prices
naturally wili be more sensitive to the sources of value that the market can see , such
as the levels of quarterly sales and earnings, than to things like investments in
promising new technologies, which investors find harder to evaluate. Consequently,
an executive who is compensated based on share-price performance will be tempted
to put too much emphasis on activities that boost short-term performance compared
to those whose benefits will be hidden from investors for a long period of time.
Worse, this problem can be self-reinforcing. Short-term investors are inclined
to gather the information that is most valuable for forecasting short-term price changes .
Realizing that short-term performance is the prime determinant of prices, these
investors will then be inclined to devote even more resources to getting early and
accurate information about ea rnings, neglecting the difficult, costly, and time­
consum ing task of assessing long-term prospects .
A FoRM.\L MODEL Most of these points can be illustrated using a principal-agent
model of the kind introduced in Chapter 7 that also incorporates a model of how
share prices are determined. In the model , the value of the firm depends on top
managers' allocation of effort to creating current and future earnings. Management's
information advantage over investors is that it alone knows how much effort it has
devoted to short- and long-term activities. Investors attempt to infer the current and
future earnings by gathering information about each , and the value they put on the
firm accurately reflects their estimates of these. Thus, stock prices accurately reflect
all public information , and yet it turns out that management will devote too much
effort to improving short-term returns and not enough to long-term earnings . The key
to the analysis is that management's allocation of effort is governed by the equal
compensation principle.

The Mathematical A11alysis of the Model

Suppose there is a firm whose fundamental value V is the sum of the values of
the current and future earnings of the firm . Current earnings are themselves
the sum of two components. One of these, C(tc), depends on the time and
 472
Finance: effort tc applied by the firm's top managers to boosting these earnings. The
Investments, second, Vc, represents all the other unknown factors that affect near-term
Capital Structure, earnings, whether due to history, environment, or chance. Future earnings
and Corporate similarly consist of the sum of a term, F(tF), that depends on the time and effort
Control management devotes, and a second term, VF, that represents all the other factors
affecting value. Thus, the total value is:
V = C(tc) + VC + F( t F) + VF
Investors are unable to observe these components of value directly, but
they are able to gather imperfect information about them. In our model, we
express the investors' initial uncertainty about the two components of value by
treating Vc and VF as random variables that, for illustrative purposes, we take
to be normally distributed with means Vc and VF and variances equal to 1 /pc
and 1/p F. (The amounts Pc and PF are the precisions of the investors' initial
estimates of the components of value. ) The information investors observe about
these prospects is represented by C( tc) + Vc + Ee for current earnings and
F( t1-.) + VF + E F for long-term ones. The terms Ee and EF are random errors
in the information the investors obtain. They provide a way to represent the
idea that some information is gathered but that the information is imperfect.
We assume that these terms have mean zero and variances 1/,rc and
1/,r F · Finally, we assume that investors expect managers to allocate time Tc and
TF to the two kinds of activities.
A statistical calculation shows that after observing the relevant information,
the investors' estimates of the current and future components of value are,
respectively:
\Ic= [pc(C( Ic) + Vc) + 1rc(C(tc) + Vc + Ec)]/(pc + ,re) and
VF = [ p F(F(tF) + VF) + ,r F(F(tF) + VF + E F)]/(p F + 1T F)
In the expression for Vc, the term C(Ic) + Vc represents the investors' prior
expectations based on their original beliefs about both how managers will behave
and what the underlying values are. Similarly, the term C(tc) + Vc + Ee is
the estimate they would form based solely on the newly observed, imperfect
information. Their actual estimate is a weighted sum of the two, with weights

applies to the expression for VF ·

determined by the precision of each information source. A similar interpretation

Under the CAPM, for any fixed beta, the price of the stock is proportional
to the investors' expectation about total returns, Vc + \IF · Other models of
stock prices yield this same conclusion. Thus, the total stock price is proportional
to:
_ c(Vc + C(1c)) + 1Tc( Vc + C( tc) + Ee)
A
VC + v"F - P
P c + 1Tc
PF(VF + F(1F)) + ,rF(VF + F(tF) + EF)
+

If the top management receives a bonus that is proportional to the share price
P F + 1TF

change, then its marginal reward for efforts (tc) to boost current earnings is
proportional to: C' (tc)1rc/(pc + ,re) and for future-oriented efforts (t F) to
F' (t1•.)1r i:l(p i: + 1r,.,). In each case, the expression is proportional to the marginal
contribution of efforts to the component of value multiplied by the weight
accorded by investors to the affected information. That weight is the relative
prec 1 s1on of the new information compared to whatever information existed
before.
 47;3
According to the eq ual com pensation principle, if the top managers are The Cla ssical
to be motivated to devote some tim e and effort to both activities, then their Theory of
returns to both kinds of efforts must be equal: Investments and
Finance
7Tc C ' (tc) = '1Ti,· F'(tF)
Pc + 7Tc PF + 'TT,..
Effi ciency, however, requires that the marginal value created by effort in each
activity be equal: C'(tc) = F'(t1.).
We have hypothesized that m ore weight will be placed on the current
perform� nce measure. The statistical condition that ensures this is:
7Tc > 'TTF
Pc + 7Tc PF + 'TT F
This ineq uality means that there is less err or in forecasts of current perform ance
than in more distant future earnings (each measured relative to the error of prior
inform ation).
The inequality, com bined with the imm ediately preceding equation,
im plies that F'(tF) > C'(tc): The effort allocation is ineffi cient. The marginal
increase in value resulting from additional efforts to boost long- term earnings is
greater than that for current earnings. Fundamental values could be increased
if the top managers could be induced to ignore the market valuations and tum
som e of their efforts away from activities oriented to the near term and toward
m ore future- oriented activities.
LESSONS FRO!\I THE MODEL The mathematical formulation clarifies three points.
First, the responsiveness ot prices to information about current and future returns

what was known about returns before the information was received. If long-term
depends not only on the precision of the inform ation but also on the precision of

perform ance is h ighly uncertain, it is not clear that even relatively poor information
about those prospects will not elicit a substantial response. S econd, the problem is
not caused by any mistaken special concern by investors for prom oting short-term
performance. Rather, the whole matter hinges on the fact th at managers have better

value. If investors could observe the managers' actions and their impacts-C(tc ) and
information than investors do about how their efforts are being directed to increase

F(tF) in the m odel-directly, then they would correctly reflect the m anagem ent' s
performance in the share price. U nderstanding the problem is the first step in
constructing better solutions, as we see in the next chapter. Finally, the argument
applies regardless of the correctness of the investors' expectations about fundam ental

model, these quantities do not even enter into the incentive calculation. If investors
values (Vc and V F) or about managerial efforts (Ic and "fi.). Indeed, in the form al

use new information in the way the m odel describes, then they will place m ore weight
on short-term performance measures than on long-term ones. That, according to the
m odel, is what encourages managers to place m ore em phasis on short-term performance
an d less on long-term performance than is desirable for m aximizing the fundam ental
value.

Implications of the New Theories for Organizations

The new theories discussed here challenge both the various form s of the effi cient
markets hypothesis and the business and policy conclusions based on the hypothesis.
First, according to these theories, all forms of the effi cient markets hypothesis must
be false because there are many unsoph isticated investors whose ill-inform ed beliefs
an d volatile expectations create inaccurate and excessively volatile prices. That the
hypothesis is false does not mean that none of its implications are true, however. The
 474
Finance: presence of some sophisticated investors still implies limits on how severe the mispricing
Investments, among various kinds of assets can be, and these limits may explain the evidence that
Capital Structure, it is difficult to earn excess risk-adjusted returns over short periods of time using just
and Corporate public information. That appears consistent with the efficient markets hypothesis.
Control The failure of the hypothesis also does not imply that firms cannot benefit from
taking public actions to enhance their long-term profitability. Indeed, there is evidence
that stock prices do rise in response to companies' announcements of increased R&D
expenditures and other long-term investments, even when these might be expected to
depress cash flow in the short run. 17 Sophisticated investors, and even naive ones,
may react to these investments by raising their expectations about the firm's total
value, thereby raising the firm's price.
Investors may also raise their expectations on the news that companies have
improved their incentive compensation plans for top managers, for example by
increasing the weight of long-term earnings in compensation. Applying this conclusion
requires making a careful examination of the changes. Increasing the intensity of top
management's incentives transfers risk to the affected managers, raising their risk
premium, and requiring higher pay. A proper incentive plan needs to balance the
demands of short- and long-term performance incentives and to keep the risks
transferred to a bearable level.
The theories of shortsighted management indicate that even if the semi-strong
form efficient markets hypothesis is true, there may still be validity both to the
complaints that managers give too much weight to the short term and the retort that
stock market pressure makes them behave that way. Managers who attempt to maximize
the market value of the firm's shares will not be led to make socially efficient decisions.
Instead, to the extent that their actions are unobserved or based on private information,
they will be tempted to focus their efforts on attempts to boost current earnings while
devoting too little effort to the business of managing for the long term.
This theoretical prediction raises important issues about whether alternative
institutions might perform better. For example, might economic systems perform
better when stocks are held primarily by long-term investors who will not sell on the
first bad earnings report? This pattern of ownership has been the norm in Japan,
where a large fraction (70 percent) of the shares of most major companies are held
by other firms that will not normally sell them and by financial institutions that have
traditionally held stocks for long periods rather than actively trading them. Firms in
Germany obtain about 90 percent of their financing from banks, who in turn are
represented on the firms' supervisory boards. The relationships between banks and the
firms they finance are long-term ones.
There are signs that a similar pattern may be emerging in the United States. A
large fraction of all U. S. stocks, especially those of the larger firms, are held by
instituti onal investors, such as insurance companies, mutual funds, and pension
plans. It has often been claimed that institutional investors are especially fickle and
impatient, and their rise has been blamed for the short-term performance pressures
that managers claim to feel. By the early 1 990s, however, many of these investors
were finding that it was unprofitable to manage their portfolios actively, buying and
selling in response to market developments, not only because they found that they
did not have enough inside information to create profitable trading strategies but also
because their holdings were so large that their transactions had too great an impact

,- Office of the Chief Economist, United States Securities and Exchange Com mission, "Institutional
Ownership, Tender Offers, and Long Term Investment, " 1 98 5 , and John J. McConnell and Chris J.
1\1 u�carclla, "Capital Expenditure Decisions and Market Value of the Firm," Journal of Fi11a11cial Eco11omics,
1 4 ( 1 985), 399-422.
 475
began to adopt buy-and-hold strategies and to attempt to put pressure on managers in The Classical
the firms they owned to manage more efficiently. Theory of
The move in the United States in the 1 980s toward companies going private, Investments and
with most financing from debt and with ownership concentrated among managers Finance
and a small group of outsiders, had the same effect of insulating managers from the
temptations to worry about short-run stock market pressures and to neglect long-run
returns. It also strengthened their i ncentives to maximize the firm's fundamental
value.

ought to be managed to maximize the value of currently traded shares. It may better
The new theories also challenge the traditional economic notion that firms

serve the i nterests of efficiency to serve the more basic objective of maximizing
fundamental value. The problem with this latter objective is detecting whether
managers actually are adopting it rather than following their own agendas. Even if
the market is. not a perfect indicator of managerial performance, without an alternative
way to evaluate and motivate management, it may be better to motivate them to

Another issue is the i nterpretation and evaluation of takeovers. If market prices

follow the dictates of share prices than to give them no motivation at all.

takeover can be profitable only if it can make the firm more valuable. If it is possible
accurately reflect the value of the firm in the control of its current managers, then a

for corporate raiders to gather better information than other investors, the issues
become more complex. The raider may simply be buying an underpriced company
and evicting a competent management, destroying value i n the process. Alternatively,
it may be that raiders are the investors most highly motivated to assess the long-term
value of the company, and their activities may help make share prices more accurate
estimates of the fundamental values of companies. Takeovers raise a number of other
issues as well, which we treat in more detail in the next chapter.
 -176
Finance:
ln\'estments,

SUMl\lARY
Capital Structure,
and Corporate
Control The classical economics of finance is based on a model of the firm as a flow of income
and expenses. Managers must decide which flows represent good investments, whether
to finance those investments by borrowing or issuing stock, and what portion of the
firm's earnings should be paid to shareholders in the form of dividends. Investors must
decide how much of which companies to buy. Their combined actions determine the
demand for various securities and consequently the relationship among the returns
on different kinds of investments. Finally, in the classical theory, the whole system
can be evaluated to determine the efficiency with which it allocates capital among
potential investments.
We examined a number of major results from the classical theory. First, in the
idealized situation where the firm can borrow and save at the same interest rate, the
firm's investment decisions can be evaluated separately from any consumption decisions
made by the owners and without reference to the owners' preferences. This first result
is called the F isher separation theorem. Second, an investment is profitable and should
be undertaken exactly when the present value of the cash flows it generates, which is
defined to be the largest loan that could be repaid using these cash flows, exceeds the
amount of the initial required investment. An equivalent statement is that the net
present value of the cash flows ("net" of the initial investment expenditure) must be
positive. Third, for stra tegic investment decisions, a correct single-project net present
value calculation using only local information about the direct cash flows from the
project lead to the wrong answer because a large part of the value of the project may
come from its indirect effect on other units or projects in the organization.
The fourth conclusion of the classical theory is that if investors can borrow and
save on the same terms as firms, and the firms' financing decisions do not affect their
total cash flow, then the firms' choices between debt and equity and their dividend
decisions have no effect on their total market value. These propositions about capital
structure and dividends are known as the Modiglia ni-Miller (MM) theorems and are
often captured by the words: "Capital structure and dividend policies cannot create
value unless they affect the total returns. "
Fifth, when the Fisher separation theorem applies, investment throughout the
economy is efficient because all firms face the same cost of capital on their investments
and the only investments that are undertaken are those that can repay the cost of
capital while still leaving something for the owner.
Several additional conclusions are obtained when it is assumed that investors
care only about the mean and variance of the returns on their portfolios of investments
and there is a riskless bond available. There then exists a single portfolio of assets such
that any risk-averse investor would optimally invest a portion of his or her wealth in
the riskless asset and the remaining fraction in the identified portfolio. The pricing
formula then asserts that the expected rate of return on any investment is equal to the
riskless rate plus an amount that is proportional to the beta of the asset, which is the
covariance of the asset returns with the returns on the identified portfolio divided by
the variance of the return on that portfolio. The capital asset pricing model (CAPM)
adds the assumption that all investors optimize in this way and concludes that the
identified portfolio coincides with the market portfolio, the portfolio of all assets
available for investment.
The efficien t markets hypothesis has three variants. The weak form holds that
past prices contain no information that can be used to predict future price changes.
The se111 istro11g form holds that published public information also cannot be used to
 477
predict price changes. Finally, the strong form holds that even insider information The Classical
cannot be used to predict price changes. The weak and scmistrong forms of the Theory of
hypothesis formerly had wide acceptance, but even the weak form has been challenged Investments and
in recent empirical and theoretical work. There is now signi ficant doubt about its Finance
general validity.
An important implication of the strong form of the hypothesis is that maximizing
the fundamental va lue of the firm to its shareholders is equivalent to maximizing the
value of the firm's shares. The reason is that the market's reaction to strategies and
events will embody the best possible estimates of any value they created or destroyed
that can be made from currently available information . Some newer theoretical work
emphasizes that these same conclusions cannot be derived from the weak and
semistrong forms . In particular, even if markets use all publ ic information perfectly
to forecast efficiently, that information may sti ll reflect more fully efforts that boost
short-term earnings than ones that increase long-term va lues. Consequently, executi ves
whose pay or job security depends on share prices may be led to neglect long-term
concerns in favor of increasing short-term performance measures .

• BIBLIOGRAPHIC NOTES
The subject of finance has evolved into a tremendously active field of research
begi nning in the late 1 960s. The field now supports several specialized journals
devoted exclusively to the topic, and our reporti ng in this chapter has been even
more selective than in the other chapters. For the selection of results reported
here, a few names stand out. Irving Fisher's 1 907 book laid the foundations for
present value analysis by treating interest rates as prices and cash flows as amounts
that could be priced. His later book is, however, better known and somewhat
clea rer.
Franco Modigliani and Merton Miller's main work is described in the text. An
excellent discussion of the MM results and thei r impact over the 30 years since
they were developed is given by Miller's paper in the Journal of Economic
Perspectives and the comments that follow it by Modigliani , Sud ipto Bhattacha rya ,
Stephen Ross, and Joseph Stigl itz.
The capital asset pricing model (CAPM) is the creation of Will iam Sharpe and
John Lintner, building on the portfolio theory analyses of Harry Markowitz and
James Tobin. Jan Mossin is also an important early developer of this line of work.
Douglas Breeden developed a major extension of CAPM which responds to some
of the criticisms levelled against it, and Stephen Ross has developed a major
competing alternative model of asset pricing. The asset pricing analysis of Robert
Lucas has also been very influential . Bengt Holmstrom emphasized that incentives
create a reason for firms to care about idiosyncratic risk, even when the CAPM
appl ies.
Paul Samuelson provided the theoretical underpi nni ngs for the efficient markets
hypothesis. An excellent discussion of it and its implications is provided by Burton
Malkiel's classic, A Random Walk Down Wall Street. Markowitz, Miller,
Modigliani, Samuelson , Sharpe, and Tobin have all been awarded Nobel
Memorial Prizes in Econom ic Science.
The challenge to the efficient ma rkets hypotheses began with empirical papers by
Robert Schiller, and by Stephen Leroy and Richard Porter. The papers in the
Journal of Economic Perspectives' 1 990 "Symposium on Bubbles" address the
possibility and some of the impl ications of systematic deviations from ma rket
efficiency. Especially usefu l is the paper by Andrei Shleifcr and Lawrence
 478
Finance: S ummers. A survey of the efficie nt markets hypothesis is pr ovided by LeR oy. Our
Investments, mathematical model of short- sighted management and short-sighted markets is
Capital Structure, based on the work of Timothy Bresnahan, Paul Milgrom, and J onathan Paul,
and Corporate who also develop the basis for our discussion of the role of financial markets in
Control allocating capital for investments. Related models of myopia have been developed
by Jeremy Stein and by Richard Zeckhauser and J ohn Pound. Henry Manne
inaugurated the debate on insider trading.
One of the most important eleme nts of the classical the ory of finance that we
have not treated is the pricing of derivative securities, such as options and futures,
that are derived from other, more fundamental securities, such as stocks. The
breakthrough in this area came with the work of Fisher B lack and Myron Scholes
on options pricing. A review of this material is pr ovided by Mark Rubinstein.
The accompanying paper by H al Varian examines the importance in financial
economics of the arbitrage arguments originated by M iller and Modigliani.
Excellent textbooks in the aspects of finance discussed here include those by
Richard Brealey and S te wart Myers, James Van H orne, and S harpe and Gordon
Alexander.

• REFERENCES
B hattacharya, S. "Corporate Finance and the Legacy of Miller and M odigliani, "
Tournal of Economic Perspectives, 2 (Fall 1988), 13 5-48.
Black, F. , and M . Scholes. "The Pricing of Options and Corporate Liabilitie s, "
fournal of Political Economy, 81 (M ay 1973), 637-54.
Brealey, R., and S . M yers. Principles of Corporate Finance (New York: McGraw­
Hill, 1991).
Breeden, D. "An l ntertemporal Asset Pricing M odel with Stochastic Consumption
and Investment Opportunities," fournal of Financial Econom ics, 7 (September
1979), 265-96.
Bresnahan, T., P. Milgrom, and J. Paul. "The Real Output of the S tock
Exchange," in Output Measurement in the Services Sector, Zvi Griliches, ed.
(Chicago: U niversity of Chicago Press, 1991).
Fisher, 1. The Rate of Interest (Ne w York: Macmillan, 1907).
Fisher, I. The Theory of lnterest (New York: Macmillan, 19 30).
LeRoy, S. "Effi cient Capital Markets and Martingales," Tournal of Econom ic
Literature, 27 (December 1989), 15 8 3-1621 .
LeRoy, S . , and R. Porter. "S tock Price Volatility: Tests Based on Implied Variance
B ounds, " Econometrica, 49 (May 1981), 55 5-74.
Lintner, J . "The Valuation of Risky Assets and the Selection of Risky I nvestments
in Stock Portfolios and Capital Budgets," Review of Economics and Statistics, 47
(February 1965), 13-37.
Lucas, R. "Asset Prices in an Exchange Economy, " Econometrica, 46 (N ovember
1978), 1429-45 .
Malkiel, B. A Random Walk Down Wall Street (New York: W. W. Norton,
1991).
Manne, H. Insider Trading and the Stock Market (New Y ork: The Free Press,
1966).
Markowitz, H. "Portfoli o Selection," fournal of Finance, 7 (March 1952), 77-
91.
 479
Miller, M. "The Modigliani-Miller Propositions After Thirty Y cars, " Journal of The Classical
Economic Perspectives, 2 (Fall 1 988), 99- 1 20. Theory of
Miller, M. , and F. Mod igliani. "Dividend Policy, Growth and the Valuation of Investments and
Shares, " Journal of Business, 34 (October 1 96 1 ), 4 1 1-3 3 . Finance
Modigliani, F . "MM-Past, Present, Future, " Journal of Economic Perspectives,
2 (Fall 1 988), 149-58.
Modigliani, F. , and M . Miller. "The Cost of Capital, Corporate Finance and
the Theory of Investment, " American Economic Review, 48 (June 1 9 58), 26 1-
97.
Mossin, J. "Equilibrium in a Capital Asset Market, " Econometrica, 34 (October
1 966), 768-8 3 .
Ross, S. "An Arbitrage Theory of Capital Asset Pricing, " Journal of Econom ic
Theory, 1 3 (December 1 976), 341-60.
Ross, S. "Comment on the Modigliani-M iller Propositions, " Journal of Economic
Perspectives, 2 (Fall 1 988), 1 27-34.
Rubinstein, M. "Derivative Assets Analysis, " Journal of Economic Perspectives, 1
(Fall 1 987), 73-94.
Samuelson, P. "Proof that Properly Anticipated Prices Fluctuate Randomly, "
Industrial Management Review, 6 (Spring 1 965), 4 1 -49.
Schiller, R. "Do Stock Prices Move Too Much to be Justified by Subsequent
Changes in Dividends?" American Economic Review, 71 (June 1 98 1 ), 42 1-36.
Sharpe, W. "Capital Asset Prices: A Theory of Market Equilibrium under
Conditions of Risk, " Journal of Finance, 19 (September 1 964), 42 5-42.
Sharpe, W. , and G. Alexander. Investments (Englewood Cliffs, NJ: Prentice
Hall, 1 990).
Shleifer, A. , and L. Summers. "The Noise Trader Approach to Finance, " Journal
of Economic Perspectives, 4 (Spring 1 990), 1 9-34.
Stein, J. "Efficient Capital Markets, Inefficient Firms: A Model of Myopic
Corporate Behavior, " Quarterly Journal of Economics, 1 04 (November 1 989),
65 5-70.
Stiglitz, J. "Why Financial Structure Matters," Journal of Economic Perspectives,
2 (Fall 1 988), 1 2 1-26.
Tobin, J. "Liquidity Preference as Behaviour toward Risk, " Review of Econom ic
Studies, 2 5 (February 1 9 5 8), 65-86.
Varian, H. "The Arbitrage Principle in Financial Economics," Journal of
Econom ic Perspectives, 1 (Fall 1 987), 5 5-72.
Zeckhauser, R. , and P. Pound. "Are Large Shareholders Effective Monitors? An
Investigation of Share Ownership and Corporate Performance," Asym metric
Information, Corporate Finance, and Investment, Hubbard, R. G. , ed . (Chicago:
University of Chicago Press, 1 990), 1 49-80.

EXEHCISES

I Food or 111ought

1 . A stock mutual fund is a company that invests in shares of the stock of

other companies. These funds exist to make it possible for small investors to pool their
 480
Finance: Table 1 4. 2 Calculating Present Values in an Economy with Inflation
lm·estments,
Capital Structure,
and Corporate Nominal Inflation Real Accumulation Present
Control Year Cash Flow Factor Cash Flow Factors Value
t ct lt = ( 1 + i)t C/I t A t = ( 1 + r/ C/A1

0 1 . 000 1 . 000
1 1 24 1 . 250 99 1 . 100 90. 00
2 ? 1 . 563 121 1 . 2 10 1 00.00
3 195 ? ? 1. 331 ?
Total Present Val ue ?
Initial Investment 220. 00
Net Present Value ?

money in order to hold a widely diversified portfolio of investments. In an open-end

fund, shareholders can redeem their shares and demand to be paid the corresponding
fraction of the value of the fund's shareholdings. In a closed-end fund, shareholders
can obtain cash for their shares only by selling them to others. It is common for
closed end funds to sell for prices lower than the actual value of company's holdings.
How can one account for that? Under what circumstances would a closed-end fund
be preferred to an open-end fund? See The New York Times, MONEY, (Sunday, July
1 5 , 1990): "Close Ties at a Closed-End Fund, " F l 3 .
2 . Suppose the preferences of individual investors depend on more than just
the mean and variance of asset returns. Which aspects of the discussion of individual
investments would you expect to be most affected? How would you expect them to
change?
3. What kind of information advantage would you expect insiders to have over
ordinary investors? Is it reasonable to suppose that they know more about the prospects
of long-term investments? About imminent earnings reports? How does each kind of
information advantage damage the efficient functioning of a system of finance based
on securities markets? Which effect is most important?

Quantitative Problems I

1. Suppose a graduating student from a U . S. community college expects to

earn $20, 000 per year for 5 years and $60, 000 per year for the subsequent 30 years
until retirement. Assume that the income arrives at the end of each year. What is the
present value of the student's future incomes, using a 5 % rate of interest? How does
your answer change if the income arrives instead .at the beginning of each year? The
middle?
2. Show (using calculus) that if a firm produces positive profits in every future
year, then the present value of the profit stream decreases when the interest rate used
in the calculation rises. To see that the conclusion is different when the stream has
both positive and negative elements, compute the present value of the stream ( - 50,
+ 190, - 1 00) at interest rates of 3%, 5 %, and 7%.
3. The accompanying table (Table 1 4. 2) shows a present value calculation for
a country that is suffering inflation of 2 5% per year when the company's real cost of
capital is I 0%. Fill in the missing figures. Show that the resulting present value is
the sa me as if the calculation followed the pattern of Table 14. 1 , but using the
nom inal cost of capital of 37. 5 %. [The nominal cost of capital n is determined from
 481
Table 14. 3 The Classical
Theory of
Value of Firm' s B ondholders' S tockholders' Investments and
Payoff Rece ipts Receipts
Finance
Bonds Issued

(A) 50 1 00 50 50
300 50 ?
Expected 50 ?
(B) 1 00 100 ? ?
300 ? ?
Expected ? ?
1 00 ? ?
300 ? ?
Expected ? ?

the formula 1 + n = ( 1 + r)( 1 + i) where r is the real cost an d i is the rate of

inflation . ]
4. A firm is about to be sold by its founder. The firm' s future earn in gs fr om
its operation s are uncertain . They may be either 1 00 or 300, each with probability
one-half. Investors are risk neutral, so they would be willing to buy the firm for its
expected return of 200. The foun der, however, wan ts to con sider whether a higher
total price could be obtained by sellin g the claim to the firm's earn in gs as some
combination of bon ds and stocks, in which the bon dholders have first claim to the
firm' s earnin gs up to the amoun t of the bon ds. The bond amounts con sidered are 50,
1 00, an d 1 5 0. In Table 1 4. 3, fill in the amoun ts that the stockholders and bondholders
receive in each con tin gency an d de termine the price that will be paid in each case.
Does the M odiglian i-M iller con clusion hold for this e xample?
5 . Suppose a particular in vestor can allocate his or her in vestments amon g
three assets on ly. The first is a riskless asset that ge nerates a certain net re turn of 5
for every 1 00 in vested. The secon d is a risky asset that generates an expected net
return of 9 for every 1 00 in vested but where the varian ce of that re turn is 64. The
third is an other risky asset that generates an expected return of 7 per 1 00 in vested with
a varian ce of 100. Suppose the return s on the two assets are statistically indepen dent:
(a) Compute the mean an d varian ce of a portfolio in vested 80% in the riskless bon d
and 20% in the first risky asset; (b) Compute the mean an d variance of a portfolio
in vested 75% in the riskless bond, 1 5 % in the first risky asset, and 1 0 % in the secon d
risky asset; (c) Why isn' t the third asset, with its lower expected return an d greater risk
al ways worthless to the in vestor?; (d) U se calculus to determine the fraction of the
One Fun d that should be in vested in each of the two risky assets. (Observe from
Figure 14. 1 that the One Fun d is the portfol io of risky assets for which the ratio of
expected return to the stan dard deviation of return is maximized. )
 15
FINANCIAL STRUCTURE,
OWNERSHIP, AND CORPORATE
CONTROL

I 'I I
I

he Raiders' Creed: "Give me your undervalued assets, your plan ts, your
expenditures for technology, research and development, the hopes and aspira tions of
your people, your stake with your customers, your pension funds, a nd I will enhance
myself and the dealmakers. "
Robert Mercer 1

As we discussed in the preceding chapter, the classical theory of finance is

extremely useful, but al so seriously incomplete. The classical theory regards financial
securities simply as claims on streams of net receipts whose magnitude and variability
are ex ogenously given. For example, a share in a firm simply entitles its owner to a
fraction of the revenues remaining after the firm pays its contractual obligations. The
theory allows that these receipts may be uncertain, but their likely magnitude (and
investors' beliefs about these) are assumed to be unaffected by the firm's financial
structure. There are at least three significant elements that make the classical theory
mi sleadi ngly simple.
First, the forms and patterns of financing can actually affect the returns generated
by the firm by affecting the incentives and behavior of various parties and the
probability of incurring the costs of bankruptcy. Second, because a firm's managers
often know more than investors do about the firm's likely prospects, some investors
will scrutinize management's financial decisions for indications about its business
forecasts. For example, a reduction in dividends may convince investors that
man agement expects the firm to be sh ort on cash. By affecting investor beliefs,

1
As quoted in Donald B. Thompson , "A Hollow Victory for Bob Mercer, " Industry Week (February
23, 1 987 ) , 48. Mr. Mercer was CEO of Goodyear Tire & Rubber Company from 1 983 to 1 989.
financial decisions can affect stock prices. Finally, fin ancial securities arc not just Fi nancial
claims on return streams: They also give their hol ders certain rights concerning Structure,
decision maki ng and con trol. As we saw in Ch apter 9, a pro per join t allocation of Ownership, an<l
decision rights an d returns can have im portan t co nsequences on economic performance Corporate Control
and consequen tly on the firm 's share prices.
In th is chapter we develop these ideas and use them to examine the differences
in the patterns of fin ancial structure between coun tries, the ch anges in financing that
took pl ace during the 1 980s, and th e dramatic contes ts for corporate co ntrol th at arose
in that decade.

CJ li\NCES IN CORPORATE CONTROL:

PKITERNS AND CONTRO\'EHSIES
Although businesspeo ple h ave long wo ndered abou t the bes t ways to finan ce th eir
operations, the seemingly arcane su bject of corporate finan ce took on a new urgency
in the 1 980s. Dramatic ch anges took place in corporate fi nancial structure, ownership,
and con trol, and th ese ch anges sparked great pu blic debate and led to a new awareness
of the in tern ational differen ces in financial practices.

Corporate Control Changes in the 1 980s

Th e decade of the 1 980s was m arked by a wave of ch anges in corporate ownership
and control in North America and the U nited Kingdom that involved som e of th e
world's largest an d most prominent firms. M any of th ese changes were precipitated
by hostile takeover attem pts, th at is, attem pts to buy co ntrol of a com pany despite
active resistance by th e top managem ent of th e target firm . Som e were management
buyouts (MBOs), in wh ich the firm's m anagers purchased its ou tstanding sh ares
themselves i n order to gain ownersh ip co ntrol. In many cases, these fi nancing ch anges
were accompanied by other structural changes, parti cularly the breaking up of l arge
firms in to sm aller, independen t operating u nits.
MERGERS AND ACQUISITIONS The scale of the takeovers in th e 1 980s was unprecedented
and in volved many well -known corporations. For exam ple, Ch evron Oil bought Gulf
Oil for $ 1 3. 3 billio n in what was th en the largest-ever com bination of two industrial
firms. Marathon Oil was bough t by U . S . Steel, and Mo bil Oil merged with retailer
Montgomery Ward. Four years later, Mo ntgom ery Ward reemerged as an inde pendent
but privately held corporation, th at is, one whose ownership shares are not pu blicly
traded. Olym pia & York Develo pm ent, th e giant in tern ational real estate firm, bough t
Gulf Oil of Can ada and H iram Walker, a dis till ery and resource fi rm , bu t let th e
liquor business go to Allied-Lyons, th e British food, brewing, and liquor con cer n.
General Electric Com pany bough t RCA, acqu iring the N ational Broadcasting
Company (NBC) television network in th e process. R. J . Reynolds, the to bacco
company, bought Nabisco Brands, th e bakery and food concern th at had recently
been formed in th e merger of Nabisco and Standard Brands. General Motors bough t
both Electronic Data Services (EDS) and H ugh es Aircraft. Bridgestone, the Japan ese
tire an d ru bber com pany, acqu ired U . S . tire maker Firestone, and Fren ch tire maker
Michelin bought Un iroyal Goodrich . Sony bough t CBS Records an d Colum bia
Pictures, and Matsush ita then bough t entertainment giant MCA. Thomson Corpora­
tion , owner of th e l argest num ber of U.S. and Can adian newspapers an d British
regional papers, became the larges t Canadian retailer when it acqu ired the venerable
Hudson's Bay Com pany.
The list goes on and on . In 1986, at the heigh t of activity, there were 3,336
announcements in the U nited States of mergers and acquisitio ns involving transfers
of ownersh ip of at leas t l O percen t of a company's assets or equity and with a price
 Finance: of more than $ 500,000. The dollar value of these transactions was $ 1 7 3 billion, up
Investments, from $44 billion at the start of the decade. By 1 988 the number of transactions had
Capital Structure, fallen, but their total value reached a quarter of a trillion dollars. 2
and Corporate
Control H osTILE T.-\KEO\"ERS The 1980s were also the era of the hostile takeover. Many of
the offers to buy the shares of corporations in the face of opposition from management
and their boards of directors came from corporate raiders. T. Boone Pickens, Jr. of
(relatively) tiny Mesa Petroleum went after Unocal, the giant oil company. Sir James
Coldsmith, who earlier had acquired Crown Zellerbach and had bid for Goodyear
Tire and Rubber, 3 launched the biggest takeover attempt in the United Kingdom
when he went after B. A. T. Industries. Ronald Perelman gained control of Revlon
and attacked Salomon Brothers. The Walt Disney Company paid Saul Steinberg $ 3 1
million more than the market value of his stockholdings (greenmail) to leave the
company alone. Robert Campeau acquired both the largest and the fifth-largest
department store groups in the United States, giving him control of some 250 stores
in 3 3 chains, including such famous names as Brooks Brothers, Bloomingdales,
Bonwit Teller, Donaldson's, Filene's, I. Magnin, and Rich's. Alan Bond acquired
breweries, hotels, publishing, broadcasting, mining, and real estate operations on five
different continents. Carl Icahn bought Trans World Airlines.
Yet other hostile offers were launched by corporations. For example, AT&T's
offer to buy NCR, a computer manufacturer, in a deal that was realized in early 1991
was initially strongly opposed by NCR's management. When Mobil Oil attempted a
takeover of Marathon Oil, U. S. Steel played the role of a white knight, rescuing it
by making a bid that Marathon's management favored and finally accepted.

1\1.-\N-\CE�IENT BurntiTs Often the fear of corporate raiders led firms' managers to
institute management buyouts, in which they and their allies acquired their companies'
outstanding stock, thereby ending public trading of the shares. For example, the
management of Safeway, the world's largest food retailer, fought off an attempted
takeover by the Haft family by executing a management buyout with the assistance
of Kohlberg, Kravis, Roberts & Company (KKR), a buyout special ist.
There were other motives for MBOs as well. For instance, initially the
management of RJR Nabisco proposed a buyout because they professed to believe that
the stock was undervalued and that a buyout would allow stockholders to realize the
underlying value. (The senior executives themselves stood to gain substantially as
well. ) Eventually, the firm was restructured, but not by management: They lost out
to a bid from KKR.

DECO'.\JGLmlER..\TIO'.\J Many of these transactions (mergers, acquISihons, takeovers,

and buyouts alike) in the 1980s were followed by bust-ups, in which a firm's businesses
and divisions were sold to other companies, or spin offs, where units were separated
as independent companies and shareholders received stock in the new companies in
proportion to their holdings in the original concern. In a large number of bust-ups,
the businesses were acquired by firms already having interests in the same or related
industries. 4 The result was a large-scale undoing of the conglomerate mergers of the
1960s and a refocusing of corporations on a narrower, better-defined range of

2 Data from W.T. Grimm & Company, Mergerstat Review, Chicago.

; Thereby provoking the quotation with which this chapter opened.
4 Sanjai Bhagat, Andrei Shleifer, and Robert W. Vishny, "Hostile Takeovers in the 1980s: The
Return to Corporate Specialization," Brookings Papers: Microeconomics 1 990, I (Washington: Brookings
Institution, 1990).
 485
businesses. (See the box on focus, productivity, and corporate control for more on Financial
this phenomenon. ) Structu re,
Ownership, and
Corporate Control

The Rise of Debt

Many of these control changes in the United States were accompanied by vast changes
in the financial structures of the firms involved. The source of financing for the
American corporate raiders and many of the management-led buyouts alike after mid­
decade were junk bonds-bonds deemed by a major rating agency to have a high risk
of default. 5 These bonds were used to i ncrease the firm's financial leverage-the ratio
of the firm's total value to the underlying amount of equity capital invested by
stockholders (the rest of the capital coming from debt financing). This pattern became
so common_ that the going-private transactions came to be called leveraged buyouts
(LBOs). Other firms also greatly increased thei r leverage by taking on debt to finance
acquisitions or to buy their own shares to put into Employee Stock Ownership Plans
(ESOPs) and by simply buying back their shares in "leveraged recapitalizations. " For
example, Polaroid bought 22 percent of its shares and put them i nto a new ESOP
when threatened by a raider, and GenCorp (the former General Rubber and Tire)
issued debt to pay for acquiri ng half its outstanding stock. The $2 5 billion leveraged
buyout of RJR Nabisco by KKR resulted in RJR's having $23 of debt for each $ 1 of
equity. It had to pay out $ 3 . 34 billion to service its debt i n 1 989 alone. Between 1 980
and 1 989 the ratio of long-term debt to total corporate assets rose by almost half, from
16. 6 percent to 24. 2 percent. 6
Correspondingly, as firms were acquired or merged out of existence, repurchased
their shares, or went completely private, stockholders' equity declined. Between 1 980
and 1 989 the ratio of stockholders' equity to total assets on the balance sheets of U . S.
manufacturing firms fell from 49. 6 percent to 40. 5 percent, and the ratio of long­
term debt to equity went from 34 per cent to 59 per cent. These balance sheet trends
were mirrored in the markets as well. In 1 984 and 1 98 5 alone nearly I O percent of
the market value of U . S. shares outstanding were retired. 7
By 1 990 the trend to increasing corporate debt i n the United States had stopped,
at least temporarily. In part under government pressure that forced the savings and
loan associations to sell the junk bonds they had bought and in part under the impact
of the conviction of Michael Milken on securities charges and the bankruptcy of
Drexel Burnham Lambert, the market for new junk bonds had largely disappeared.
At the same time, the takeover and LBO waves collapsed. The volume of LBOs
fell from $ 54 billion in 1 989 to one tenth as much in 1 990, and the dollar volume
of takeovers was almost cut in half. Many of the firms that had taken on exceptionally
high levels of debt in buyouts began reducing thei r debt loads by selling assets or

5 These bonds were backed by the target company's cash flow, but because so m uch of the firm's
total financing was from debt, there was a significant chance that there would not be enough money to
meet required interest payments.
6 The data on debt and equity relative to assets come from the U . S . Department of Commerce
Quarterly Financial Report, 1 980 through 1989. These are so-called "book values," the numbers appearing
on companies' balance sheets. Balance sheets list the book value of assets (often their historical cost less
accumulated depreciation) against liabilities (short-term plus long-term debt), with the difference being
stockholders' equity. Thus, the figures need not be in any close relation to the market value of the assets
or of the firms' shares.
7 Henry Kaufman of Salomon Brothers Inc. , quoted in J. Fred Weston, Kwang S. Chung, and
Susan E. Hoag, Mergers, Restructuring, and Corporate Control (Englewood Cliffs, NJ: Prentice-Hall,
1 990), p. 1 2 3 .
 -186
Finance:
Investments,
Capital Structure,
and Corporate
Focus, Productivity, and Corporate Control
Control The merger wave of the 1980s was only the latest of several that have occurred
over the last century in the United States. The preceding one, in the 1 960s,
featured the "conglomerate merger, " under which unrelated businesses were
brought under common ownership. Although the most obvious result of this
was the emergence of the conglomerate firm exemplified by Tenneco, ITT,
and United Technologies, in fact most large companies expanded the scope
of their activities very broadly. The rationale offered for these mergers was
twofold. First, management was thought to involve a generic set of talents
and skills such as budgeting, financing, procurement, personnel management,
and so on, so that industry expertise was unnecessary and firms would be
most efficient by having them run by the especially good managers that the
conglomerates had or could attract. Second, it was argued that professional
managers were better than investors at the tasks of directing capital to its best
uses and diversifying risks among businesses.
Although the stock market at the time generally reacted favorably to
conglomerate mergers and acquisitions, in fact both arguments now appear
flawed. As was noted at the time, stockholders can achieve diversification for
themselves by adjusting their investment portfolios, so they hardly need it to
be done for them by firms. Moreover, despite their better information,
managers have other objectives and face other pressures than do shareholders
who are investing their own money. Since the 1 960s it has also become
evident that there are real advantages in managers being more than casually
acquainted with the businesses they run, and there is accumulating evidence
that more focused, less diversified firms perform better. For example, Birger
Wernerfelt and Cynthia Montgomery found that more focused firms had
higher values of "Tobin's q" (defined as the ratio of the market value of the
firm to the replacement value of its assets and a measure of the value being
created in the firm) than did more diversified firms. Frank Lichtenberg found
that plant productivity fell significantly with increases in the number of
different lines of business in which the firm owning the plant was engaged.
By the 1 980s many diversified firms' stock-market values were less than their
apparent bust-up values-the total for which the separate pieces could likely
be sold.
The control changes in the 1 980s in part undid the diversification of
the I 960s. Companies spun off unrelated lines of business as separate
organizations or, in the relaxed antitrust environment of the period, sold
them to firms with related interests. Some· of this was done by existing
management teams under conscious strategies of refocusing, but often it
followed bust-up takeovers designed to take advantage of the difference in the
value of the firm and its pieces. LBOs also contributed to the reduced
diversification, as businesses were sold to generate cash to pay down debt.
The result was a significant decrease in overall diversification. Lichten­
berg found that firms that ceased to exist in the second half of the decade
tended to be more diversified than average, new firms that emerged in the
period were much more focused, and the firms in his sample that continued
through the period tended to reduce sign ificantly the number of different
industries in wh ich they were active. Overall, the mean number of industries
 487
Financial
Structure,
Ownership, and

i n which the firms operated fell 14 percent, whereas the fraction of highly
Corporate Control

diversified fi rms (defined as operating in more than 20 different industries)

in his sample fell 37 percent and the fracti on of highly focused firms operating
i n a single industry rose 54 percent.
Combi ning the increase in focus with the connection betwee n focus
and pr oducti vity, we should expect that the changes in control should have
increased productivity. There is some direct evidence for this: Lichtenberg
and Donald S iegal found that pr oducti vi ty, measured at the plant level, rose
relative to the industry mean in the followi ng year in manufacturing plants
involved i n ownershi p changes in the 1970s and also in plants i nvolved i n
LBOs in the 1980 s. Further, i n the 1980s stocks of firms outperformed the
market in years that the firms became more focused, and did worse than the
market i n years i n which they became more diversi fied.

Based on John G . Matsusaka, "Takeover Motives During the Conglomerate Merger

Wave," mimeo , University of Chicago, 1991; Cynthia Montgomery and Birger Wernerfelt,
"Tobin's q and the Importance of Focus in Firm Performance," American Economic Review 78
(March 1988), 246-50; Frank Lichtenberg, "Industrial De-Diversification and Its Consequences
for Productivity," National Bureau of Economic Research (NBER) Working Paper 3231 (January,
1990); Lichtenberg and Donald Siegal, "Productivity and Changes in Ownership of Manufacturing
Plants," Brookings Papers on Economic Activity 3 (1987), 643-73, "The Effects of Takeovers on
the Employment and Wages of Central-Office and Other Personnel," NBER Working Paper
2895 (September 1989), and "The Effects of Leveraged Buyouts on Productivity and Other
Aspects of Firm Behavior," NBER Working Paper 3022 (June 1989); and Gregg Jarrell and
Robert Comment, "Corporate Focus and Stock Returns," mimeo, University of Rochester,
199 1 .

issui ng new stock, and more firms were issui ng new shares of stock to rai se money
for new investments. B etween the fourth quarter of 1989 and the fourth quarter of
1990 the rati o of long-term debt to assets in man ufacturing fell sli ghtly to 23 . 9 percent,
although the rati o of equity to assets also conti nued i ts decline.

The Debate
All thi s acti vity caught the public attention and sparked much debate. The need to
generate cash to service the new debt led highly leveraged firms to trim staff and di vest
or close various parts of their businesses, often thr owing long-time employees out of
work. Critics worried that firms burdened with heavy debt would be forced to curtail
i nvestments in new products, new equipment, and new technology and would then
be condemned to a future of obsolescence and decline. They also worried about the
increased likelihood of bankruptcies in a recessi on, a concern that proved to have
some vali dity in the recession that began i n the summer of 1990. They saw corporate
managers as bei ng totally absorbed by financial dealmaki ng and by defendi ng against
irresponsible raiders bent on bust-up takeovers. They feared that foreign competitors,
with no such di stractions, would seize total market dominance, leavi ng those Americans
who were not investment bankers with nothing but badly- paid j obs with no future­
"flippi ng burgers at McDonalds. "
N ot everyone shared this assessment, however. Advocates cheered the corporate
 488
Finance:
Investments,
Capital Structure,
and Corporate
Michael Milken, Drexel Burnham Lambert,
Control and Junk Bonds
High-yield junk bonds played an important role in the LBOs and takeovers
of the 1 980s. They played an even more important role in the public
perceptions of the financial activity of the period-both the LBOs and
takeovers and the S&L disaster. Michael Milken of Drexel Burnham Lambert
was personally the creator and center of the junk-bond market.
Corporate bonds with high risk and high promised returns are not new:
There have always been "fallen angels, " bonds issued by major firms that
were initially regarded as safe but that became risky when the firms ran into
trouble. The key innovation, developed by Milken in the late 1 970s, was for
Drexel to underwrite and make a market in debt issued by firms which
previously would not have been able to sell bonds to the public because they
would not have been rated as sufficiently safe investments by the rating
agencies. These deals were immensely profitable for Drexel, which charged
fees of four or five times what other investment banks charged for underwriting
more standard debt issues. Drexel came to dominate the junk market, which
grew immensely in the early 1 980s as more established firms began to issue
junior, subordinated debt in it, and, because his operation generated a huge
share of the firm's profits, Milken dominated Drexel.
Junk bonds became very popular investments for many S&Ls attracted
by their yield-risk mix. When the public began to become aware of the
impending S&L debacle in the late 1 980s, many were quick to blame junk
bonds and Milken for the industry's problems.
In 1982 Drexel began to help finance MBOs with junk bonds, and
shortly thereafter, partially as a means to break into the booming business in
mergers and acquisitions, it began using its junk bond operation to finance
hostile takeover attempts by such raiders as Carl Icahn, Ronald Perelman,
Boone Pickens, and Saul Steinberg. Drexel and Milken raised hundreds of
millions, and even billions, of dollars at a time to finance takeovers, much
of it in bonds to be backed by the target companies' cash flows in specific
deals but some for "war chests, " funds to bankroll future raids.
In 1981, new issues of junk totalled $ 1 . 3 billion; by 1 986 they were
$ 3 2. 4 billion, and the total junk market involved $ 1 2 5 billion. Despite the
attempts of other investment banks to enter the business, Drexel-through
M ilken-still dominated it, and for the first time Drexel was attracting "blue­
chip" companies as investment banking clients. Drexel made immense profits
on its junk-bond business. Milken also m·ade millions for himself: $ 5 5 0
million from Drexel i n 1 987 alone, plus hundreds o f millions more on his
own account.
Although Milken �nd Drexel made money, they also made enemies of
those who saw the junk-bond, bust-up takeover as a fundamental threat to
American business and probably to America's future. Milken's pronounce­
ments that he and his associates were rescuing American business from overly
conservati\·c, incompetent, self-serving managers did not add to his popularity
in these circles. At the same time, others in business, the press, and Congress
saw them as heroes .
The fall of Milken and Drexel began in 1 986, when arbitrageur Ivan
 489
Financial
Structure,
Ownership, a nd
Corporate Control
Boesky, a Milken client, pleaded guilty to insider trading and began
cooperating with the government in investigating others who had violated
the securities laws . It became clear that Drexel and Milken were targets in
these investigations, and this limited the firm 's opportunities and weakened
it abilities to raise funds and force takeovers . Still, it did back Perelman in
his attempted takeover of Salomon Brothers in 1 987, and it was the sole
manager on the junk offeri ng for the gigantic KKR takeover of RJ R Nabisco.
Finally, in 1 988, the government filed criminal and securities law
charges against Drexel and Milken for insider trading, manipulating stock
prices, making false filings with the SEC, falsifying records and documents,
and defrauding clients. Drexel pleaded guilty to felony charges, paid $650
million in fines, and agreed to cooperate in the government's case against
Milken . Facing potentially huge liabilities in civil suits and without the
source of its revenue and power, Drexel soon went bankrupt.
In 1 990 Milken entered a plea bargain, admitting guilt in six "narrow"
felonies. He was, however, given an unprecedentedly severe sentence: 1 0
years i n jail, three years required community service, and a $600 million
fine. Many viewed this, for good or ill, as a "verdict on a decade of greed . "

Based largely on Connie Bruck, The Predator's Ball (New York: American Lawyer: Simon
and Schuster, l 988), as well as news reports. The quotations are from Chris Welles and Michele
Galen, "Commentary: Milken is Taking the Fall for a 'Decade of Greed'," B usiness Week
(December I O, 1 990), 30.

raiders, arguing that the target firms had become bloated and complacent. Why else
were savvy investors willing to pay an average of 40 to 50 percent more than the
market price of shares for the opportun ity to replace existing management and boards?
Acqui red firms were usually thoroughly reorganized to create units more focused
around single busi nesses, with the top management of each business having a larger
ownership share and with the providers of capital having greater power to remove top
management if it performed poorly. The raiders and leveraged buyouts were shaking
business from the self-satisfied lethargy into which it had fallen after World War II
in the easy environment of limited international competition , and the new debt levels
just reproduced financing patterns that were common in other industrialized countries.

I nternational Patterns of Financing and Ownership8

The heavy reliance on equity financing that marked U . S . firms before the rise of debt
in the 1 980s differed considerably from the traditional patterns of financial structure
in other industrial ized economies, which themselves were also changing in this
decade. In France in 1 980, for example, bank loans provided 71 percent of companies'
external financing, with the rest spl it between shares and corporate bonds in
approximately a 3 to 1 ratio. Debt slightly exceeded two-thirds of the value of assets,
and it peaked at 72 percent in 1 982. Then equity financing became much more
attractive after the large banks were nationalized (making bank loans a less attractive

8 This section draws heavily from Bill Emmott, "The Ebb Tide: A Survey of International Finance,"
The Economist (April 27, 1 99 1 ), which is also the source for most of the data quoted.
 490
Finance: means of financing ) and the stock market's rules were liberalized. The volume of new
Investments, issues of stock rose sixfold between 1 984 and 1 990, reaching the equivalent of $42
Capital Structure, billion, and about 30 percent of firms' funding came to be through shares. Banks
and Corporate became significant purchasers of these shares, as did mutual funds, most of which
Control were bank subsidiaries. By 1 988, bank borrowing accounted for only 5 3 percent of
the financing of French nonfinancial firms, and debt had fallen to 63 percent of the
value of assets.
Japan experienced a similar, but stronger, move away from bank loans and
toward equity and other sources of financing in the 1 980s. At the start of the decade,
the average debt-equity ratio among major, publicly-traded firms was 2. 7 5 to 1 , and
64 percent of their external financing was from bank loans. Then Japanese firms
began to seek financing elsewhere. First they raised funds in the international bond
markets, then on the booming Tokyo Stock Exchange, and later still through
convertible bonds (debt that could be converted into common stock) and bonds with
warrants (rights to buy the firm's stock at a prespecified price) attached. By 1 990 the
overall debt-equity ratio was almost 1 to 1 , interest-bearing debt was only about 7 5
percent of the value of equity (whereas i t had been more than twice the value of
equity in the mid- l 970s), and the ratio of debt to assets was less than 30 percent.
Firms retired bank loans and built up hordes of cash. Toyota Motors, for example,
was jokingly called "Toyota Bank," because in 1 990 it held over ¥ 1 . 36 trillion (about
$ 1 0 billion) in cash and cash-equivalents, and another ¥. 84 trill ion in short-term
investments.
In Germany, bank loans provide over 90 percent of firms' external financing.
This fraction changed relatively little between 1 970 and 1990, despite some increase
in the importance of equity. ln part, the new equity replaced corporate bonds, which
have ceased to be a noticeable source of financing. Few German firms' shares are
publicly traded, however. Of the approximately 360, 000 German firms whose
shareholders have the limited liability for the firm's debts that marks the U. S.
corporation, only 2,300 are establ ished as Aktiengesellschaften, whose shares could
potentially be publicly traded, and of these only 6 1 9 actually are traded on the different
German stock exchanges. 9 Moreover, most new issues of stock are sold through
negotiated private placements rather than general offerings to the investing public.
The banks are close to the firms they finance, and they normally have places on the
firms' supervisory boards-the second board that German firms are required to have
by law and that also must include workers' representatives. Often a banker chairs the
board and can wield great power.
The hostile takeovers that were so prominent in the English-speaking countries
in the 1 980s were rare in France and essentially absent in Japan and Germany. In
part, the patterns of financing and ownership explain this. ln France, there are
different types of shares with different voting rights, and Germany carries this even
further by allowing companies to limit the share of voting rights that any single
stockholder can ha\·e. In both countries, banks' stock holdings would not normally
be for sale to a raider. These holdings can be significant: Deutsche Bank, for example,

9 The Aktiengesellschaften are identifiable by the letters "AC" after the company name, while the
private corporations (Cesselschaften mil beschriinkter Haftung) are denoted by "CmbH." Examples of the
former are the chemical firm Bayer AC and the electronics firm Siemens AC. The private firms tend to
be smaller and less well known internationally, but Sony's wholly-o\\'ned European subsidiary is such a
firm . The parallel designations in the United Kingdom are the letters "pie," for "public limited company,"
and "l td, " indicating a privately owned firm. Again, the large firms-for example, British Petroleum pie
and the international co11glomcratc I lanson pie-tend to be organized as public limited companies, \\'hile
�mailer ones arc private. Notable private companies arc such publishers as the l\lacmillan Press, Ltd, and
111any British securities firms.
 49 1
holds more than 2 5 percent of the shares in five different major German companies, Financial
including auto-maker Daimler-Benz. It reputedly organ ized resistance to the attempt Structure,
by Italian tire maker Perclli to take over Continental, a German tire firm . These Ownership, and
factors make win ning a takeover bid difficult, and the fact that most companies in Corporate Control
Germany are private removes them completely from the threat of raiders inviting
stockholders to tender their shares. Of course, the key distinctive feature of ownership
patterns in Japan is the cross-ownership of shares by other firms sharing business
relations. Some 24 percent of the shares of major Japanese industrial and financial
corporations are held by other corporations, and the figure reaches a full 70 percent
when the stockholdings of non-corporate financial institutions (such as mutual
insurance companies and investment funds) are considered. 10 The shares held by
other corporations would not be offered to a raider because doing so risks breaking
the economic alliances they help cement, and the financial institutions have also
traditionally been friendly to management. Nevertheless, there have been very active
mergers and ·acquisitions markets in Europe (and to a lesser extent in Japan) throughout
the 1 980s and into the 1 990s, with control of businesses changing hands in friendly
transactions.
The events of the 1 980s, as well as the international differences and trends,
have puzzled many observers. Were the takeovers necessary to revitalize the target
organ izations? Does it real ly matter whether firms finance their investments by issuing
stock or by borrowing? What determines the firm's choice between these two options?
Why do the patterns differ so much internationally? What effect does the choice have
on the efficiency of the firm and of the economic system? What are the implications
for business in the 1 990s and beyond?

FINANCIAL STRUCTURE AND INCENTIVE

The Modigliani-Miller-MM-analysis of the irrelevance of financing and dividend
decisions begins with the premise that financing decisions do not affect the cash flow
stream itself. We have already observed that debt financing can reduce the firm's tax
obligations and, in that way, it can transfer value from the government to the
shareholders. The effects that the diverse tax treatments of dividends, capital gains,
and interest payments have on firms' decisions can be detected statistically by studying
the differences in firm's financing decisions in different jurisdictions with different tax
rules. Although the way firms respond to incentives created by tax rules is an important
subject of study for both public policy makers and businesspeople, it is too lengthy
and involved for a detailed treatment here. The interaction of business decision
making and taxation merits has received its own book-length treatment. 1 1 The main
focus of our attention here is on how capital structure can affect value by changing
the incentives of managers, by affecting the conflicts of interest that can arise between
stockholders and creditors, by influencing the probability that the costs of bankruptcy
will be incurred, and by providing incentives for equity investors and lenders to
monitor management and limit its excesses .

Conflicting Interests : Managers Versus Owners

Capital structure can affect two broad sets of conflicting interests: managers versus
equity owners, where the issues are basically the familiar problems of managerial

10 Jack McDonald, "The Mochai Effect: Japanese Corporate Cross-Holdings, " f ournal of Portfolio
Management, 16 (Fall 1989), 90-5.
1 1 Myron Scholes and Mark Wolfson, Taxation and Business Strategy: A Global Planning Approach

(Englewood Cliffs, NJ: Prentice-Hall, 1 992).

 492
Finance: moral hazard in controlling costs and maximizing value, and existing creditors versus
Investments, other suppliers of capital, where the issues revolve around the choice of investments.
Capital Structure, Consider the case of a small, growing firm whose founder wants to "go public,"
and Corporate that is, to sell shares to outside investors to raise money for additional investments.
Control The founder must decide how many shares to sell to outside investors and what share
of ownership to keep for himself or herself. This decision is not one that outside
investors can neutralize through financial market transactions as in the MM
propositions. The fraction of the firm that the founder keeps is determined fully by
the number of shares sold versus the number retained.
As we have seen in several earlier chapters, ownership affects incentives. In the
present case, selling part of the firm to outside stockholders can adversely affect the
owner-manager's incentives for maximizing profits. If the founder retains a specific
percentage share in his or her firm, then he or she enjoys that share of the financial
benefits and incurs that share of the financial costs of his or her decisions. This alone
causes no distortions because the decisions and actions that maximize a fixed fraction
of the benefits less that fraction of the costs also maximize the difference between the
total benefits and total costs. To the extent that the decisions have nonfinancial costs
and benefits that are not easily observed and contracted over, however, the owner­
manager is likely to bear these directly. In that case, maximizing the net benefits that
accrue to the owner-manager is not the same as maximizing total net benefits.
For example, an owner-manager who holds a 50 percent stake in the firm and
who decides to fly first class, to stay in expensive hotels, to have a luxurious company
car, and to invest the firm's resources in pet projects that are of dubious value bears
half the financial costs of these decisions, but gets all the benefits. The temptation to
indulge in this type of behavior may be limited by a 50 percent ownership stake, but
the attractiveness of such decisions is much greater if the manager retained only 5
percent of the firm because 9 5 percent of the costs are paid by the outside stockholders.
Small shareholdings by the top managers of the firm also reduce their incentives to
be diligent in monitoring the performance of others within the firm, controlling costs
and increasing revenues. The box on executive behavior and ownership stakes illustrates
two recent examples of such conflict between management's and shareholders' interests.
Generally, when management owns only a small fraction of the outstanding
shares in a firm, several sources of conflict may exist. First, managers will be more
interested than shareholders in the growth and longevity of the company. Growing
companies provide more opportunities for promotion, and top managers in larger
companies tend to earn higher salaries. Meanwhile, when companies close, become
bankrupt, or are sold or taken over, lucrative jobs are often lost, especially in the
management group. Second, management will be tempted to spend too much on
perks for themselves and, possibly, for their subordinates or other employees because
they enjoy these perks but bear only part of the cost. Finally, management
prefers independence from outside interference, . both for its own sake and because
independence contributes to job security and allows managers to set higher rates of
pay in their own ranks. All of these problems become worse as the managers' equity
stake in the firm becomes smaller.
These tendencies may be exacerbated when the firm has a large free cash flow.
Cash How amounts to revenues less direct outlays, or, equivalently, profits plus
depreciation allowances. Free cash How is the cash How in excess of the amount that
can profitably be reinvested in the firm. For efficiency, these funds should be returned
to stockholders, for example by increasing dividends or repurchasing shares of stock.
The stockholders then can put the money to whatever use they think best in
consumption or other investments. Managers and their boards may be severely tempted
 Financial
Structure,
Ownership, and
Executive Behavior and Ownership Stakes: Corporate Control
Armand Hammer and F. Ross Johnson
Occidental Petroleum and RJ R Nabisco provide two recent extreme examples
of CEOs with few shares being apparently less-than-frugal with corporate
resources (and of the firms' boards failing to restrain them). Occidental
Petroleum was built and led by Armand Hammer, who served as CEO and
Chairman of the Board until his death in 1 990 at age 92. He pursued a
variety of acquisitions and business strategies that were widely criticized,
including money-losing businesses in the USSR and China (Hammer had
been one of the first western businessmen to meet with Lenin, and he kept
in contact with the succeeding generations of communist leaders) and meat­
packing and seed-research operations. He also kept the company's dividend
high and stable at $2. 50 per share whereas in 1990 earnings were only $ 1 . 03
and the company had to borrow to pay the d ividend. During the last decade
of his leadership, the firm's share price fell by a third wh ile oil company
shares on average tripled in value. Discontent with Hammer's strategies and
the firm's performance led most institutional investors to abandon the stock.
Over his lifetime, Hammer had acquired a large personal art collection
(some of it allegedly with corporate funds), and he decided it should be seen
by the publ ic. He therefore had the corporation undertake building the
Armand Hammer Museum of Art and Cultural Center in Los Angeles to
house his collection . Suits attempting to stop this diversion of corporate funds
were settled without going to trial, and the museu m was built at an eventual
cost to the corporation's shareholders of as much as $ 1 20 million .
Hammer held less than 0 . 5 percent of Occidental Petroleum's stock,
although he had once owned nearly half the company. The Occidental board
incl uded his grandson and two of his lawyers. After his death they renounced
many of Hammer's strategies that outsiders had criticized.
F. Ross Johnson was CEO of RJR Nabisco until the KKR buyout in
1988. As the head of Standard Brands and then of Nabisco Brands, Johnson
had already built a reputation for spending corporate money lavishly. He
doubled executives' salaries at Standard Brands and provided them (and
himself) with company apartments, a private box at Madison Square Garden ,
and multiple country cl ub memberships. He also put a variety of former
athletes on the payrol l. He was quoted as saying "Give me a guy who can
spend creatively, not one who is trying to squeeze the last nickel out of the
budget, " and he expressed admiration for the executive he put in charge of
entertainment and "extravaganzas" for being "the only man who can take an
uni imited budget and exceed it. "
These patterns conti nued at RJR Nabisco. He also handed out $ 1 , 500
watches, hired former U. S . president Gerald Ford to play in a golf tournament
the company sponsored and Frank Sinatra and Bob Hope to entertain guests,
expanded the company's stable of athletes (a number of whom apparently
did nothing for the retainers of up to $ 1 million a year they received),
authorized extremely luxurious office furnishings, company-provided Cadil­
lacs, Mercedes, and even Rolls Royces with chauffeurs, and hired 36 pilots
for the company's ten corporate jets. The planes of the "RJ R Air Force" were
 494
Finance:
Investments,
Capital Structure,
and Corporate
Control
available for executives' use under any pretext of a business reason, and
sometimes allegedly even when there was none.
Johnson held 60, 000 of RJR Nabisco's 247 million shares, or 0. 02
percent. Until the board accepted KKR's bid over his, he seems to have had
its full support. He was quoted as saying "One of the most important jobs a
CEO has is the care and feeding of the directors. "

Based on "Wife's Heir Sues Armand Hammer," The New York Times National Edition,
(July 1 5, 1 990), 8Y; "Occidental Petroleum: Taking a Sickle to Hammer's Empire," The
Economist, (January 1 9, 1 991), 64-65; Nell l\1inow, "Shareholders, Stakeholders, and Boards
of Directors," mimeo, Institutional Shareholder Services, Inc. ; and Bryan Burrough and John
H elyar, Barbarians at the Gate: The Fall of RJR Nabisco (New York: Harper & Row, 1 990).
Burrough and Helyar are the source of the material in quotation marks; the direct quotations of
Johnson are from pages 2 5 and 26 of their 1 991 Harper Perennial edition.

to use these resources within the firm, however, carrying out new investments that
(by definition) are not profitable and not in the shareholders' interests. They may also
be more inclined to indulge themselves in excessive perks and to share the wealth
with the employees.
Despite the problems that arise from small managerial shareholdings, we cannot
conclude that the net effect is beneficial when the managers retain a large share of
ownership in the firm. Sometimes the problem of the top management owning a
minor share of the equity is unavoidable. For example, few managers have the
personal wealth to purchase a large share of a major automobile company, and it is
not obviously in the shareholders' interest simply to give large shareholdings to repeated
generations of managers. Even if the company in question is small enough and the
owner wealthy enough to avoid this first problem, there may be a second one: A
manager whose wealth is entirely tied up in a single firm bears too much of the firm's
risk personally. Risk-averse managers may then be too cautious about risking their
entire personal fortunes in aggressive company investments that would be desirable if
the risk could be optimally shared. Some diversification of their investments would
both reduce the risk premium that is incurred and make the managers more willing
to take appropriate risks.
Of course, it may be possible to provide incentives for managers through explicit
and implicit compensation contracts rather than through shareholdings directly, and
concern with their professional reputations and consequent market opportunities may
also discipline their choices. Yet neither of these mechanisms is perfect, so the
ownership position of managers may still play a role.

Conflicting l n tt'rests : Current Lenders versus

Otllt'r Ca p ital Su pp lirrs
If the needed funds are not raised by issuing shares, then the primary alternative is to
finance company investments by borrowing. However, this can give rise to conflicts
between the interests of shareholders (perhaps including management) and those of
the firm's creditors. In general, unless those who select the firm's investments own
equal shares of the firm's debt and equity, they will not be financially motivated to
select the investments that maximize the total value of the firm.
 495
Financial
Structure,
Ownersh ip, and
CEO Shareholdings Corporate Control
Michael Je nsen and Kevin M ur phy have assembled data on U . S . CEOs'
holdin gs of their companies' stock. * The data arc from the 1987 Forbes
survey of executive compensation and concern 746 executives.
The mean value of the CEOs' holdings across the whole sample was
$41 .0 million, which represented 2.42 percent of the outstanding shares.
The distribution was quite skewed, however, because some CEOs (often the
companies' founders or their heirs) had very large holdings, whereas 80
percent of the CEOs held less than 1 . 38 percent of the stock in their firms.
The median holdings were only $ 3 . 5 million, representing 0. 25 percent of
the shares outstanding. The total value of the shares of the median firm was
$ 1 . 2 billion.
In firms below the median size in the sample, the CEOs held an
average of 3 . 0 5 percent of the shares, representing a value of $ 1 9. 3 million,
but again the medians were much lower: 0. 49 percent or $2. 6 million. In
the larger half of the firms, the CEOs' percentage holdings were smaller, but
the values were bigger. The averages were 1 . 79 percent of the shares, worth
$62. 6 million, whereas the medians were 0. 1 4 percent of the total shares,
with a value of $4. 7 million.

• "Performance Pay and Top-Management Incentives," fournal of Political Economy,

98 (April 1 990), 225-64.

EXCESSIVE RISK-TAKING O ne problem between shareholders and creditors is that of

asset substitution, which we emphasized in Chapter 5 in our discussion of the savings
and loan crisis. When a firm has too much debt compared to its equity, the owners
may be too ready to undertake risky investments. This is because the owners of shares
enj oy virtually al l the benefits if returns on the risky investments turn out to be high,
but the lenders suffer a maj or portion of the losses if the returns turn out to be low. For
a given expected net present value to the investments, the parties' interests are in direct
conflict: Any increase in risk that reduces the ex pected return to creditors results in an
equal increase in the expected return to shareholders. As we saw in the savings and loan
example, shareholders may even prefer risky investments with negative net present
val ues to safer ones with positive total returns. This ineffi ciency is a cost of debt.
Lenders ought to forecast that the firm's owners will have these incentives to
make inefficient investment choices (or to direct or motivate managers to make them).
This means that the amount they are willing to lend will be less or the return they
will demand wil l be increased. Consequently, the expected costs of the inefficient
investment choices will fall on the shareholders. This in turn will give them incentives
to try to find ways to commit themselves not to undertake risky investments. For
example, they may agree to pl ace covenants in the debt contract that restrict their
ability to make risky investments, or they may use short-term bank debt, allowing the
lender to withdraw the loan in case of excessive risk taking.
DEBT 0\'ERI JANG AND UNDERIN\ ' ESTI\IENT A second problem arises particularly when
the firm currently does not have enough cash to pay its current debt obligations but
does have profitable investment opportunities. As an extreme example, suppose the
firm has an outstanding debt of $ 1 0 million more than the value of its assets and that
 496
Finance: it then obtai ns an opportunity to make an i nvestment of $5 mi llion yieldi ng a sure gross
Investments, return of $12 mi lli on for a gu aranteed net profit of $7 mi lli on. If the debt covenants
Capital Structure, give the current debt priority for repayment, then no new lender or i nvestor will be
and Corporate wi lli ng to fi nance the i nvestment because the first $ 1 0 milli on accrues to h olders of its
Control exi sting debt, leavi ng only $2 mi lli on in returns for the $5 milli on of new i nvestment.
As a result, th e profitable investment may not be undertaken and value may be lost.
Th is i s a parti cular case of debt overhang, the inability of a company with
profitable i nvestment opportu nities to fi nance them because it has excessive levels of
debt relative to its assets. Debt overh ang has been a newsworth y issue i n the context
of Third World economic development. Like firms, countries that are deep i n debt
may be unable to borrow money to fi nance pr omisi ng new i ndustri al development
pr oj ects because any pr ofits may have to be pai d first to exi sti ng lenders. (The pr oblems
may be even more severe for these countri es than for firms because their very
sovereignty may prevent nations from committing themselves using loan covenants
that restri ct their activiti es.)
The fact that value is lost from the underinvestment caused by debt overhang
means that the outcome is i neffici ent and there is some other arrangement th at
everyone would prefer. In our example, the fai lure of the firm to make th e i nvestment
will lead to losses of $10 mi lli on for the lender. If the firm and its lender can agree
to reduce the debt by $ 5 mi lli on, then the investment can be undertaken profitably,
the reduced loan can be repaid, and the new i nvestors can earn profits of $2 milli on.
The debt reducti on reduces the loan losses suffered by holders of exi sti ng debt from
$10 m i lli on to $5 mi lli on.
In vi ew of the advantages of bei ng able to renegotiate debt contracts in case
earni ngs are low, it might seem that the parti es would want to ensure th at renegotiati on
is possi ble. For example, concentrating the debt in the hands of a few lenders would
make it easi er to renegoti ate the debt, creati ng additional value, if the firm experie nced
hard ti mes. The err or in thi s logic is th at the very act of protecti ng owner-managers
agai nst the consequences of bad outc omes reduces their i ncentives to act dili gently to
ensure good outcomes and rei nforces their already excessive i ncentive to engage i n
risky activities.
When the h olders of debt are many in number and h oldi ngs are diffuse, the
threat of debt overhang with out the possi bi lity of debt renegoti ati on favors li miti ng
the le\· el of debt. When debth oldi ngs are concentrated i n the hands of a few parties,
so that debt renegotiati on is more likely, incentives are created for managerial
mi sbehavi or. In either case, high levels of debt carry a si gnificant risk of destroyi ng
part of the firm's value.
FREE C.\SH FLmr .\ :\D DEBT Ft:\ANCt:\/G Balanci ng these negative effects of debt are
direct, positive effects on managers' incentives not to waste free cash flow. Replaci ng
equity by debt commits and compels managers to pay out cash to meet debt service
requirements or ri sk losi ng control of the company i n bankruptcy. Equity fi nancing
bri ngs no such direct, unavoi dable pressure, and unless the board is very responsive
to stockh older i nterests and well positioned to ensure that perqui sites are kept in check
and empire bu ildi ng is arni ded, there may be little effecti\ · e pressure of any sort to
turn over free cash A o\\' to investors. A mi ssed divi dend may anger and upset
shareh olders, but they m ay be powerless to do much about it. In language someti mes
heard am ong fi nanciers: "Equity is soft; debt is hard. "

Owrwrs· l 1 w<'1 1 t i ves for \ lo 1 1 i t or i 1 1�

The board of directors i s supposed to e nsure that the i nterests of sh areh olders, and
perhaps of other stakeholders, arc represented i n the firm's maj or decisi ons. Ho\\'ever,
 497
the creation of a board of directors j ust drives the issue on e level deeper: Wh at Financial
disci plines the beh avior of the members of th e board? Who monitors the mon itors? Structure,
U ltimately, som eone with an ownership interest is needed to ensure th at th e Ownership, and
management does not squander the shareholders' in vestments. Y ct in dividual investors Corporate Control
with relati vely small holdings have little incentive to bear th e costs of monitoring
man agement's performan ce or the board's di ligence.
For example, su ppose you held 1, 000 shares of IBM in June 1991. At th e stock
prices then prevailing, that holding would have been be worth over $98, 000 dollars.
Because th ere were over 592 mi llion shares of IBM stock outstanding, however, your
ownersh ip claim would have been o nly 0.000168 percent of the total. Th us, if you
exerted enough effort in mon itoring to increase the profitability of IBM by $ l , you
would have gotten back only . 000168 cents. You would clearly have been unwilli ng
to do this un less you are moti vated by very different co nsideratio ns than we h ave
assu� ed. Only investors with a relatively large stake will be i nclined to do signi ficant
amounts of monitori ng. Thus, the concentration of share ownership can affect a firm's
value.
Co:-.;CENTR.\TED SHAREHOLDING Amo ng the 500 largest U . S. industrial concerns i n
1980 (the "Fortune 500"), i n only 15 cases was there no i nvestor holdi ng at least 3
percent of the sh ares of the company. On average, the largest i nvestor owned 15. 4
percent of the sh ares and the five largest owned 28. 8 percent, which are significant
fractions of the total. 12 I n smaller companies, the co ncentration of ownership is even
greater.
Large shareho lders who have large sums at stake can play an important role i n
discipli ni ng management. These shareholders have little personal i nterest i n u nprofit­
able growth for growth's sake because the new career opportu nities and higher salaries
for top management that accompany growth do not accru e to them. The very takeo vers
that threaten managerial jobs hold th e promise of lucrative sales of shares for the large
shareho lders. Sometimes, these shareho lders may be willi ng to i ncur the expense of
fili ng lawsuits to protect their i nterests. If they are represented o n the company boards
of di rectors, they may be able to fire managers who are performi ng badly, replaci ng
th em with more effecti ve managers. Th e threat of bei ng fired from a lucrati ve job
(o ne that provi des rents to the manager) can be a very effecti ve incentive for good
performance.
In Japan, the other co mpanies that hold a corporatio n's stock to cement busi ness
relations between th em can also serve as mo nitors. They have a very direct and
pressing interest i n the success of the firm, and th e close relatio nshi ps among them
may facilitate their taki ng an active role in mo nitoring. The Japanese practice of
executives from major firms becomi ng CEOs of smaller, related firms may h elp i n
this.
An important issue has been the extent to which large institutio nal in vestors­
pension funds, mutual funds, and insurance companies-have been willing and able
to play an acti ve mo nitori ng role. Pension fu nds alone control 30 percent of th e shares
of major publicly traded firms in the U nited States and the U nited Kingdom, and i n
at least 20 percent of the cases noted previously, the co ncentrated ownership comes
from i nstitutio ns' holdings. The professio nal money managers hi red to ru n the
institutions' portfolios are not necessari ly especi ally able monitors of man agement.
Moreover, i n th e past, insti tutio nal investors' policies were usually to avoid "interferi ng"
in the compani es' management and simply to sell their holdi ngs if they were dissatisfied

12
These data were reported by Andrei Shleifer and Robert Vishny, "Large Shareholders and
Corporate Control ," Journal of Political Economy, 94 ( 1986), 46 1-88.
 �98
Finance:
Investments,
Capital Structure,
and Corporate
The Efficiency of Ownership Patterns
Control The extent to which ownership of a corporatio n's shares is co ncentrated varies
widely acro ss American business. For example, in a sample of 511 large
corporatio ns from 1980, the fractio n of the total stock held by the five largest
shareholders ranged from 1. 27 percent to 87 .14 percent, with a mean of
24.81. What accounts for the variation?
Harold Demsetz and Kenneth Lehn suggest that efficiency provides
one answer.* They argue that ownership sho uld be more co ncentrated where
the net returns to co ncentratio n are higher. Large firms, they argue, should
have more diffuse ownership for three reasons. First, it is more expensive to
increase one's percentage of ownershi p in a larger firm. Second, a given
fractional shareholding wi ll give more infl uence in a larger firm than in a
smaller one. Third, risk aversion amo ng investors im plies that hi gher
co ncentratio n impo ses increasing risk prem ia as the firm becomes larger.
Size, however, is not the only variable that affects net returns to increased
concentration. Greater firm-specific uncertai nty increases the opportunity to
add val ue by careful assessm ent of the firm's performance. Greater government
regulation limits the firm's range of actio n (thereby redu cing bo th manage­
ment' s discretion and owners' optio ns), reducing the returns to concentratio n.
Finally, they suggest that in some types of firms there may be co nsumptio n
value in determining the nature of the firm's output. The two examples they
highl ight are professio nal sports, wh ere there may be an extra thri ll to seeing
your team win when it really is "your team," and the media, where the
benefi t is in having your opinions broadcast and infl uencing public opinion.
They tested these predictio ns by regressing measures of ownership
co ncentratio n o n variables representing the elements listed previo usly. Each
turned out to have the predicted sign and to be statistically significant.
Moreover, the extra co ncentratio n for media companies came almost entirely
from individual and family holdings rather than i nstitutional investors,
cons istent with the idea that it is the perso nal consumptio n value of co ntro l
that accounts for the effect.

" Harold Dernsetz and Kenneth Lehn, "The Structure of Corporate Ownership: Causes
and Consequences," fournal of Political Economy 93 (December 1985), 1 1 5 5-77.

with performance. As noted in Chapter 14, howe\·er, there are indications that this
pattern may be changing. We will return to this point later.
An additio nal pro blem comes from the fact that bei ng a large shareholder in a
large firm necessitates ho lding a very unbalanced portfolio unless th e investor is
extremely weal thy. What additio nal returns accrue to the large investor to offset the
costs of being poorly diversified? Perhaps the fear of adversely affecti ng th e stock price
deters large stockholders from selling their holdi ngs (except to others \\·ho are will ing
to take a comparably large positio n), bu t this is sue is as yet no t well understood.
A third question involves the infl uence th at a large stockhol der actual ly can
exert in the face of ma nagemen t's resista nce. A necdotes suggest that thi s resistance
can be very effective. For example, Ross Perot became General l\ lotors' largest
sharehol der a nd a mem ber of its board after lie sold EDS to the au tomoti\·e giant.
 499
He was in constant conflict with Board Chai rman Roger Smith over policy and was Financial
extremely critical of what he saw as GM's institutionalized bureaucratic lethargy. Yet Structure,
he seemed unable to make any difference, and eventually he agreed to be bought out. Ownership, and
In the late 1 980s Carl Icahn was by far the largest shareholder in U SX (the former Corporate Control
U. S. Steel), with a 1 3 . 3 percent stake. He sought to have the firm divide its steel and
oil businesses into separate entities, but management resisted. Icahn sold his shares
in 1 99 1 after only limited success. That same year T. Boone Pickens gave up trying
to win a seat on the board and redirect the strategy of Koito Manufacturing, an auto
lighting supplier to Toyota. Pickens had acquired 26 percent of the shares in this
Japanese concern, making him the largest shareholder, but he was systematically
rebuffed in attempting to influence its policies. Indeed, he was even denied access to
the company's accounting books, and when he won the legal right to examine them,
he co,uld not find a Japanese accounting firm willing to assist in this task.
There is also some limited systematic evidence on the efficacy of large
shareholders· as monitors, which suggests a more positive inference. 13 We might
hypothesize that an outsider would be a more effective monitor in firms with relatively
little firm- and manager-specific capital than firms where capital is very largely firm­
or manager-specific, because in the latter case it will be harder both to analyze
investment decisions and to change them usefully if performance seems lacking. For
example, it might be more difficult for an outsider usefully to redirect an R&D­
intensive computer company than an oil or steel company. In fact, across 1 1 industries
where capital is generally unspecific (as measured by a low ratio of R&D to sales),
the presence of a single entity owning 1 5 percent or more of the stock had a very
significant positive effect on the ratio of stock price to current earnings, implying a
higher than normal expected growth in future earnings. No such effect showed up
for firms in industries with high ratios of R&D to sales.
LBO ASSOCIATIONS AND ACTIVE I NVESTORS The problem of inadequate monitoring
by owners is a free-rider problem; it arises because any individual bears the full cost
of any monitoring effort but shares the benefits with all the other owners. If
concentrating the ownership in one single entity is difficult because the value of the
firm's equity is high, one obvious solution is to replace much of the equity by debt
and then to have the debtholders organized to monitor management intensively to
avoid the problems described earlier. This pattern is the essence of what Michael
Jensen has labelled the LBO association. 14
Typically, LBOs are financed with 80 to 90 percent debt. Much of the financing
may be arranged by a buyout firm such as KKR or Forstmann, Little & Company,
or by the LBO fund of an investment bank. They provide the bulk of the equity for
the buyout and arrange for the loans (and, in very large deals, any needed equity
partners) to finance the rest. They then have very strong incentives to make sure that
the firm performs. They do this by careful monitoring, as well as by giving management
very strong incentive compensation (including a share of the equity). The other
members of the association are the institutional investors who invest in the LBO firm
or fund, providing the resources to buy the equity in the LBOs, and the banks and
other investors who make the needed loaQS and purchase the debt. Their moni toring

1 3 Richard J. Zeckhauser and John Pound, "Are Large Shareholders Effective Monitors? An
Investigation of Share Ownership and Corporate Performance, " R. Glenn Hubbard, ed. , Asymmetric
Information, Corporate Finance, and Investment (Chicago: University of Chicago Press, 1990), pp. 149-
80.
1 4 Michael Jensen, "The Eclipse of the Public Corporation, " Harvard Business Review (September­
October 1989), 6 1 -74.
 500
Finance: of the buyout firm, and its incentives to maintain a good reputation so that it can do
Investments, further deals, are additional support for performance.
Capital Structure, More generally, concentrating equity holdings gives the usual strong incentives
and Corporate of ownership. The common pattern of success among such active investors as the
Control Hanson Group in the United Kingdom and Clayton & Dubilier and Berkshire
Hathaway in the United States, despite their very different styles of dealing with the
companies in which they hold major stakes, is evidence of this. 1 5
VENTURE CAPITAL For start-up companies in some areas and industries, venture
capitalists play much the same role that large shareholders do for established firms.
These investors provide the financing that start-up companies need and also monitor
their activities closely to ensure that the funds they have supplied are well used.
One of the main problems that venture capitalists face in financing start-ups is
that the customary kinds of financial data are of little use for evaluating the firm's
performance. In the early years of a new firm's existence, it may have to invest
enormous sums in new products and systems before there are any sales or production
data that can be compared with competitors' results or other standards to give a
reasonably objective assessment of the new firm's performance. For example, Alza
Corporation, a California-based drug company, took seven years of testing of its
products to obtain approval from the U . S . Food and Drug Administration before its
first product could be marketed .
Without the benefit of objective measures, those who provide financing to new
firms must stay close to the business, keeping track of the kind of detailed, subjective
information that only an insider could know. Venture capital ists, who install their
own employees in high positions in the firm (often as chair of the board of directors),
continue to finance the start-up firm until it reaches a stage where there is enough
objective performance information to convince public investors to buy its shares.

Monitoring Incentives for Lenders

Venture capitalists, like the large shareholders discussed earlier, are essentially equity
investors in the firm . 16 It is not only equity investors, however, who can be usefully
enrolled as monitors of the firm. There can also be incentives for lenders to monitor
firms in a way that improves the borrower's economic performance .
BANKS Most firms that borrow obtain their money from a bank, which monitors the
firm's financial health and only lends funds if it judges that repayment is likely. Short­
term loans for working capital based on anticipated revenues are the major business
of commercial banks. For example, a bank might make a seasonal loan to a farmer
to hire workers for a harvest, expecting to be repaid when the harvest is sold. Similarly,
the bank might lend to a toy manufacturer to finance a build-up of inventories in
anticipation of Christmas · sales. In the months before Christmas, the manufacturer
needs money to pay wages, to acquire supplies, to pay rent on the factory building,
and so on. From the banker's point of view, the chances that the firm can repay the
loan after the Christmas season may appear to be good, especially if the firm has
received advance orders from retailers and if it has successfully marketed its toys in
the past.
Longer-term bank loans are an important feature of business in other countries,

1 ' See "Punters or Proprietors: Survey of Capitalism," The Economist (May 5, 1 990).
16 In fact, in return for their investments venture capitalists typically receive preferred stock that
can , at their option, be converted to regular common stock. This conversion is usually made when the
firm sells its stock to the public. Until then, the venture capitalists' preferred stock gives them the protection
of having a prior cla im on the firm , ahead of the holders of common equity.
 501
where they often represent the bulk of external financing, even for large firms. I n Financial
these cases, each firm typically has an ongoing relationship with a "main bank. " This Structure,
bank may have a representative on the firm's board of directors and will have close, Ownership, and
ongoing contacts with management. It may also hold some stock in the firm. It is Corporate Control
thus in a good position to mon itor the firm's health and the safety of its loans.
BONDHOLDERS When lenders are simply the purchasers of the firm's bonds in the
debt markets, they have less direct access to information and less direct say in operating
decisions. In this case, they may insist on covenants in the bond contracts which limit
the actions that the firm can take. Such covenants are designed to protect the value
of the bonds against opportunistic behavior by managers (on their own behalf or in
the interests of shareholders). For example covenants may limit the firm's right to
issue new debt of equal or greater seniority in order to prevent undercutting the value
of the existing bonds by givi ng other borrowers higher priority on the firm's cash flow.
Other covenants may limit investments in totally new lines of business or might
restrict the sale or pledging of major assets to ensure that they are available to secure
repayment of the bonds.
The bondholders in RJR Nabisco, for example, suffered very large losses­
estimated at perhaps $ 1 bi llion-on the market prices of their bonds when the firm
went through its LBO because the huge debt load it took on made their claims much
more uncertain: Essentially, the highly rated bonds became junk. The assets the
company then sold to meet immediate debt service requi rements arguably turned this
risk into reality and became the basis of a lawsuit by the Metropol itan Life Insurance
Company, which held RJR Nabisco debt with a face (nominal) value of a quarter of
a billion dollars. These events illustrate the value to lenders of including appropriate
covenants in bond agreements .
LONG-TERM LENDERS Long-term loans are sometimes granted by banks and other
lenders (such as insurance companies) to finance the purchase of specific major assets,
which are then used as secu rity for the loan. Mortgages on real estate are a familiar
example, but other assets can also be financed in this way . For example, an airli ne
might borrow to purchase an airplane, with the plane being subject to seizure by the
lender if the airline fails to make its payments of interest and principal. New airplanes
are especially popular security for loans because there is an active secondary market
for planes and the planes are easily relocated to the airline where they are most
valuable.
A popular variation of th is practice is the use of a long-term lease, in which the
lender buys an asset from a company and then leases it back to the compa ny to use
in its business. Once again, airplanes are a popular example. The lease agreement
may also stipulate that after a fixed period of time, the a irline company has the right
to purchase the plane for some bargain price. Such a lease is almost the same as a
loan because the airline is committed to pay the amounts in the contract over a period
of years and stands to lose the plane if it defaults . Among the main differences are
the tax treatment and the treatment in bankruptcy: Leased airplanes do not belong to
the airline. In 199 1 Trans World Airlines faced the prospect that its planes might be
seized just as its peak summer season approached because it had fa iled to meet
scheduled payments on a number of its ai rcraft and jet engines .

Default and Bank ru p tcy Costs

Sometimes firms are unable or unwilling to meet their loan payments on schedule,
and the lenders then must choose between extending the time allowed the borrower
to repay the loan or forcing the borrower into bankruptcy to collect what they can .
Both options are costly. If the lenders extend the loan, the firm may sink deeper into
 502
Finance: financial problems and become still less able to pay. There may also be adverse
Investments, reputation effects with other borrowers. Forced bankruptcy usually means that the
Capital Structure, firm is liquidated. The fi rm is then nothing more than a collection of used physical
and Corporate assets that may have to be sold at a deep discount. In contrast, if it stays in operation,
Control the firm has much more: As a going concern it has a team of knowledgeable managers
and employees, trademarks and brand names, business systems, and established
supplier and customer relationships that are all needed if the assets are to generate
any income. Some of these values may be lost in bankruptcy, and the employees may
be thrown out of work, at least temporarily. In the interim, the firm's managers are
distracted from running what is left of the business.
Bankruptcy proceedings also create legal and administrative costs that may, if
the lawyers are effective maximizers, absorb a huge fraction of the net value left in
the company. Further, the conflict between management and lenders is intensified
in bankruptcy proceedings, leading to additional losses of value. Influence costs are
incurred as the claimants to the firm's resources--creditors of various sorts, workers,
governments (if taxes are owed), shareholders, and the lawyers representing each
group-struggle to claim as large a share as possible.
Even the prospect of bankruptcy can be costly. If employees and managers are
encouraged to develop firm-specific human capital by the promise that they will share
in the returns it generates, then the fear that bankruptcy will deprive them of the
promised quasi-rents may lead to inadequate investment. Also, when the firm is in
financial difficulties, management may be led to take unwise risks. If these turn out
well, the employees' and managers' jobs will be saved, whereas the chances are that
they are lost if the risks are not taken. Moreover, suppliers of a firm that is in financial
trouble will be reluctant to extend credit and, estimating that their relationships with
the firm are less valuable because it may fail, they will be less willing to invest in
maintaining these relationships and may cut quality, be slow with deliveries, and
so on.
The larger the amount of debt relative to equity financing, the greater is the
probability of the firm's being unable to meet required payments, and the greater is
the expected value of the costs that are incurred in bankruptcy.
CHAPTER 1 1 BANKRUPTCIES These costs of forced bankruptcy and liquidation help
explain the existence of the Chapter 1 1 provision in U. S. bankruptcy law and similar
provisions in other countries. Firms that declare Chapter 1 1 bankruptcy are given
protection from their creditors (that is, the creditors cannot seize assets or force
immediate liquidation for nonpayment), and current management is allowed to try
to reorganize the firms' activities to make them profitable again. This means that a
plan is put forward to pay the creditors only part of their claims, with any debts
incurred after declaring Chapter 1 1 getting first priority. The creditors ultimately must
approve the planned reorganization for it to be accepted by the courts, and if they
cannot or do not agree, then forced liquidation follows. Typically, however,
management has at least six months at least to come up with a plan, and in fact the
firm may continue in Chapter 1 1 almost indefinitely.
Without this institution, firms that were basically healthy but suffering from
short-term cash-flow problems would be at the mercies of their creditors. A single
creditor among many, fearing for its loan or refusing to renegotiate terms when other
lenders are willing, could force liquidation of the firm. The result might be that good
firms would be liquidated, with all the attendant costs, when they optimally should
have continued in business.
!:\FORMAL "WoRKotrrs" Although Chapter 1 1 reorganizations in bankruptcy may
save on the costs of forced liquidations, they too have their costs. In particular, lawyers'
 50:J
fees can be immense, eating up more than ever flows through to claimants. A Financial
particularly disturbing example i nvolves the U. S. asbestos manufacturers that declared Structure,
Chapter 1 1 bankruptcy when the immense claims against them as a result of incidents Ownership, and
of asbestos-related diseases began to emerge. In the.;e cases, legal costs were $ 1 . 70 for Corporate Control
every $ 1 that the plaintiffs received. 17
In response to these costs of bankruptcy, informal debt workouts have emerged.
Under these, the firm and its creditors bargain over rescheduling debt payments and
over the amounts that ultimately are to be repaid. We have already discussed how,
in the presence of debt overhang, it can be beneficial for a lender to reduce its claim,
and the same principles apply when bankruptcy is threatened. Although the courts
are not directly involved in these negotiations, the possibility of Chapter 1 1 bankruptcy
lurks in the background and influences the bargaining, as does the ultimate threat of
forced liquidation.
This bargaining results in shifts of gains and losses among claimants relative to
their contractual claims. Under forced liquidation, the holders of senior debt have
absolute, contractual priority on the available fu nds (after the bankruptcy lawyers, the
taxing authorities, and wage earners). Junior or subordinated debt comes next in line,
then preferred shares, and, finally, common stock. The courts typically enforce these
priorities. However, the costs of bankruptcy mean that it may be advantageous to give
up some claims i n Chapter 1 1 negotiations to save the dissipation of value that would
result from pushing for forced liquidation. Meanwhile, the rules under Chapter 1 1
give junior debt, preferred stock, and common equity some say i n accepting or
rejecting a proposed reorganization. These rules in turn give them bargaining power
in the negotiations i n workouts. I n each case, it may be better to have a slice of a
larger pie than all of a smaller one.
Over the course of 4 1 Chapter 1 1 bankruptcies examined in a recent study,
junior claimants managed to extract $878 million that contractually was owed to
senior debtholders. 18 Of this $878 million, junior debtholders gained more than half
and the owners of common stock gained a third, despite their arguably having no
valid claim at all. In 47 workouts examined in the same study, senior debt gave up
almost $ 1 . 4 billion to which it was contractually entitled. Most of this went to
common stock: Although it had no legal claim, it had the power to declare Chapter
1 1 bankruptcy.

STRATEGIC ASSET DESTRUCTION The equity holders and managers of a firm in

financial distress may actually face i ncentives to destroy the value of the firm's assets
to improve their bargaining position in negotiating with the firm's creditors. To see
how such perverse i ncentives could arise, consider a firm with assets and liabilities of
$ 1 0 million, so that there is just enough value in the firm to satisfy its creditors' claims
arising from the resources they have provided, leaving nothing for the shareholders.
If management destroys part of the firm's value, the entire cost falls on the creditors.
By threatening such asset destruction, management might be ·able to win concessions
from the creditors for itself and the shareholders. For example, if management is i n
a position to destroy $2 million in value, the creditors are better o ff abandoning up
to $2 million of their claims in return for not having the assets destroyed.
Of course, the same sort of incentives exist if the firm's assets are smaller than

1 7 James S. Kakalik, et.al. , "Costs of Asbestos Litigation , " (Santa Monica , CA: Institute for Civil
Justice, Rand Corporation, Publication R-3042- I CJ , 1 983).
18 Julian Franks and Walter Torous, "How Firms Fare in Workouts and Chapter 1 1 Reorganizations,"
London Business School working paper (May 1 991) , as reported in "Economics Focus: The Kindness of
Chapter 1 1 ," The Economist (May 25, 1 991), 83.
 504
Finance: its liabilities, although the bankruptcy court may be able to prmide creditors with
Investments, some protection. :\ lore strikingly, there can be incentives for asset destruction even
Capital Structure, when the firm is solvent. For example, suppose that the firm has liabilities of S 1 0
and Corporate million but assets worth S 1 0. 1 million and that it is still possible to destroy S 2 million
Control in value. If management now destroys S 1 00 thousand in value and threatens to destroy
the remaining S L 9 million, the creditors find themselves in much the same siruation
as in the previous paragraph. It is distinctly possible that they then ,,ill be willing to
make enough concessions to management and the equity holders that these groups
are better off than when the assets were intact. Of course, if all sides recognize these
incentives, it may be possible to exact concessions without e,·er destroying the assets'
,·alue. On the other hand, some asset destruction might be needed to make the threat
credible.
The obvious way to amid this problem is to ensure that the firm's assets are so
much larger than its debt obligations that it will ne\'er be worthwhile to destroy value
as a bargaining ploy. This means that there must be some minimal share of equity
in the fi rm's financial strucrure.

Financial Strncture. Incentives. and Yalue

The incenti,·e and rights aspects of financial strucrure prm·ide ample reason to
challenge the :\ 1odigliani-:\ 1iller theorems' conclusions that financial decisions cannot
affect \'alue. A firm's capital strucrure affects the incenti,·es and behavior of its
managers, lenders, and equity holders, as well as the likelihood of incurring the costs
of bankruptcv. Changing the financing of the firm changes incenti,·es, and the
resulting real changes in beha,·ior affect the rerums that are generated. These in rum
determine what the firm is worth to outside investors, and consequently what they
will be willing to pay.
Howe,·er, it is not just simple financial indicators like the debt-equity ratio that
determine these incenti\'es. The number of shares held by the management team,
the concentration of ownership among large outside in\'estors, the composition and
powers of the board of directors, the restricti,·e clauses in loan agreements and bond
contracts, and the anticipated sharing of a\'ailable resources in the e\'ent the firm gets
into financial trouble all affect incentives and behavior.
A first pass at a theory of the optimal financial strucrure of the firm would
recognize the complex, interacting incenti,·e effects of different financing decisions
and the tradeoffs that must be made. It would then recommend that strucrure that
balances these tradeoffs in a manner that is value-maximizing in the context of the
particular firm in question. For example, other things being equal, highly profitable
firms in markets that ha,·e little grm,th potential should carry more debt and have
less equity financing in order to deal "ith the incenti,·e problems of free cash flow.
A prime example here might be the tobacco business in the United States because
the industry generates immense cash flow but its market is shrinking. The oil business
in the late 1970s and early 1980s may be another because, at least at that time, the
industry was highly profitable but overexpanded. As another example of the trade­
offs, debt should be lower in firms with largely intangible assets that may be particularly
susceptible to being opporrunistically destroyed by management acting in its mm
interests or those of the equity holders during debt renegotiations. In this regard, it is
notable that physical-capital intensive industries, like public utilities, airlines, and
steel producers, are hea,·ily debt financed, although arniding the costs of bankruptcy
would argue for low debt in rnlatile industries like the latter two. In contrast, industries
where assets are less tangible, such as high-tech firms and ad,·ertising agencies, are
primarily financed by equity.
 505
Although the incentive and rights aspects of financial structure are surely Financial
important, there is even more to be considered. Structure,
Ownership, and
SIGNALING AND FINANCIAL DECISIONS Corporate Control

It is well established empi rically that increases in the amount of equity the firm issues
lower the firm's share price (and decreases in the number of shares increase their
price), whereas increases in debt tend to increase the share price. Further, in
transactions involving more complex securities (such as preferred stock or convertible
debt) or simultaneous buying and selling of different securi ties (such as borrowing to
finance a share repurchase), the more "equity-like" is the security being issued, the
more negative is the effect on share prices. 1 9 A variety of signaling models will explain
these patterns.
Regardless of how diligent investors may be in monitoring their investments,
the management of the firm wi ll usually be much better informed about the firm's
prospects than will most outside investors: Managers and investors are asymmetrically
informed. As we have discussed in earlier chapters, conditions of asymmetric
information create incentives for the relatively uninformed parties to draw inferences
from the choices made by the better-informed parties. The i nformed parties, if they
recognize that their actions are being interpreted as signals, may attempt to manipulate
the signals to convey a particularly favorable message. Financial decisions can serve
as just this kind of signal.

Debt and Equity

Suppose managers gain when the stock price of the firm is higher and that they lose
personally if the firm goes bankrupt. The former effect might arise from their having
personal stock holdings or incentive contracts. Management could lose in bankruptcy
because they enjoy quasirents in their current employment in the form of high pay,
perks, status, and independence in their decision making about the firm's activities or
because their professional reputations are harmed by failure. Debt financing is costly
for these managers because the higher the level of debt, the higher the probability of
having to suffer the costs of bankruptcy. However, managers whose firms have higher
expected cash flows face a lower probability of bankruptcy than do those with otherwise
similar earnings distributions. Finally, assume that the market values hi gher earnings
distributions but cannot observe them directly, but that the managers can.
A signaling equilibrium in this context involves managers with higher distribu­
tions of earnings adopting higher debt levels. A higher debt level then signals a better
distribution of returns to the market, which responds by assigning a higher value to
the firm. This higher value is sufficient to compensate managers with good return
distributions for the increased risk of facing bankruptcy that comes with increased
debt. It is not enough to induce managers whose returns are poor to take on the same
level of debt, however, because the same amount of debt is more likely to lead them
into bankruptcy.
According to this theory, when a firm's earnings prospects improve significantly,
it will be led to change its financial structure, increasing the role of debt. The market
will then read this as signaling the underlying change in the managers' private
information about returns and will bid up the value of the firm. In contrast, worsened

19 See J. Fred Weston, Kwang S. Chung, and Susan E. Hoag, Mergers, Restructuring, and Corporate
Con trol (Englewood Cliffs, NJ: Prentice-Hall, 1 990), p. 1 2 5 , for a more precise statement and Milton
Harris and Arthur Raviv, "The Theory of Capital Structure, " foumal of Finance, 46 (March 1 99 1 ), 297-
3 5 5 for a variety of references to the original studies that found these patterns.
 506
Finance: prospects will lead managers to want to issue more equity because the risk of bankruptcy
Investments, is now too high. New equity issues will signal bad news to the market, and it will
Capital Structure, respond by marking down the value of the firm.
and Corporate A related model uses managerial risk aversion as the source of the differences
Control in the costs of signaling across differently informed managers that is needed to support
a signaling equilibrium. More leverage allows insiders to keep more of the equity for
themselves, which is most beneficial when management expects high earnings. Risk
aversion means that holding risky equity is costly, but the cost of risk bearing is the
same regardless of the level of expected returns. Thus, management that expects
higher returns is inclined to issue more debt and less equity, so sophisticated investors
would interpret a debt issue as a signal of high anticipated earnings. This implication
for stock price reactions to changes in financing is again consistent with the observed
pattern.
REAL EFFECTS The investor reactions described in this section represent a problem
for firms trying to raise equity capital. The very act of selling shares drives down the
share price, to the detriment of the firm's founders. The signaling analyses suggest
that the problem is worst for young firms, when the uncertainty about the firm's
quality and prospects is greatest, because it is then that the insiders' information
advantage is likely to be greatest. This perspective helps to explain the reliance of
young firms on bankers and venture capitalists, who invest resources to acquire
information about the firm, its prospects, and its operational needs, and use that
information to form their own judgments without needing to rely extensively on
management's signals. All of this is expensive, of course, and adds to the firm's cost
of raising capital. Once the firm has an established record of performance, reducing
the uncertainties facing individual investors, these losses can be significantly reduced.

Dividends, Monitoring, and Sign aling

Just as incentive and signaling arguments can provide a basis for understanding the
choice of financial structure and some of the empirical regularities regarding actual
capital structure patterns, so too can they give insight into dividend policy. Dividends
present a puzzle that goes beyond the Modigliani-Miller irrelevance proposition. In
many countries, capital gains (wealth changes arising from increases in the value of
assets) are taxed much less heavily than are dividend receipts. For example, some
countries have no tax on capital gains at all, whereas dividends are taxed as ordinary
income. In the United States, capital gains are now taxed at the same rate as regular
income, but only when they are realized by selling the asset. This allows the taxpayer
to defer the tax liability. Further, given the peculiarities of U. S. tax laws, it may be
possible for people to avoid the tax altogether by passing the appreciated assets on to
their heirs. This suggests that firms would increase (private, if not social) value by
returning earnings to investors in other forms than dividend payments, such as repaying
debt or buying back some outstanding shares. Yet not only do firms pay dividends,
but the market on average reacts favorably to increases in dividend payments.
An immediate objection to the suggestion of not paying dividends is that some
shareholders look to them for current income. However, this need could be met by
selling a portion of the appreciating portfolio. The firm could even offer to repurchase
the shares. Unless the transaction costs of selling shares are very high, the puzzle
rcmams.
A variety of signaling models can help rationalize the practice of paying
dividends, with dividends again signaling management's private information about
future cash flows. One, mentioned earlier, is that a firm that anticipates a future cash
flow problem may try to conserve its current cash resources by cutting its dividend.
 507
Similarly, a firm that an ticipates an improving cash flow may be willin g to increase Financial
i ts dividend. Investors, reading these changes as signals about insiders' well-informed Structure,
expectations, may respond by bidding up the price of a firm that raise s its dividend Ownership, and
and bidding down the price of a firm that cuts its dividend. Corporate Control
Dividends may also serve incenti ve roles. Paying them removes resources from
the firm and so helps overcome some of the difficulties of free cash flow: Shareholders
who might have a difficult time monitoring management's i nvestment deci sions and
expenditures can easily tell if they are getting the dividends that they demand. This
incentive effect argues for dividends being higher in slow-growth industries than ones
with good investment opportunities. Dividends also force firms to go to the capital
markets more frequently than otherwise would be necessary. This in turn may
faci litate investors' monitoring management's performance and putting checks on bad
investments that would otherwise be diffi cult to prevent if retained earni ngs could be
used for financing.

Objectives in Selecting Financial Structure

Given these theories and the evidence that capital and ownership structures affect
value, the next step i s to ask how they will be chosen. Will the capital structure be
chosen to maximize the market value of the firm, that is, the total amount that the
founders of the firm can collect for all the claims to the firm's assets? The strongest
theoretical argument favoring this conclusion is that a firm's founders will want to
extract as much value as possible when they sell the firm, and consequently would
choose the capital structure that maximizes the firm's market value. Any other choice
means that the amount the founders can realize from the firm is less.
M ore generally, however, if the firm's founders expect to participate actively in
the firm's management even after the financing is completed, they might value their
independence as managers to make the choices they prefer as well as the cash they
receive from selling claims to the firm's earni ngs. If the investors in the firm buy
shares and make loans in competitive markets, then they can expect to gain nothing
from the transaction because this investment opportunity wi ll be priced to be equally
as good as their next best opportunity. B ecause capital structure decisi ons will not
affect the welfare of new investors, an efficient choice of capital structure entails the
founder-managers' most preferred combinati on of personal control and net proceeds
received from loans and the sale of shares.
After the departure of the founder, control of the firm will typically pass to
professional managers operating under the oversight of a board of directors. Will these
parties be motivated to continue to make the capital structure choices that are value­
maximizing as circumstances change? M ore generally, will the choices they make on
the whole variety of issues confronting the firm be the ones that shareholders would
want? What mechanisms ensure that firms that are not being run well are redirected,
with managers who are not performing and with boards that do not provide the proper
i ncenti ves bei ng replaced by ones who will?

CORPORATE CONTROL
Financial securities are not j ust claims on return streams. They also give decision and
control rights. For example, stockholders have the right to elect boards of directors to
represent them in monitoring, motivating, and disciplining managers and in overseeing
strategy. They can also vote to replace directors who are not performing these tasks
to their satisfaction. Furthermore, in man y countries the corporation cannot sell off
the bulk of its assets or merge itself out of existence without a shareholder vote.
Creditors can typically force a firm that is in default to liquidate assets, and in the
 508
Finance: United States they have the right to approve or reject any reorganization plans under
Investments, Chapter 1 1 bankruptcy. They can also sue to enforce restrictions on the debtor's
Capital Structure, behavior which are written into the debt contracts. These powers, even when not
and Corporate actually exercised, can affect economic performance.
Control The transferability of securities (especially equity shares) and the rights they give
are also the basis for the market for corporate control. The functioning of this market
can be another important determinant of managerial and corporate performance.
Assets that would be more valuable under different management can be purchased,
and control of them can be vested with those who can use them best. In particular,
corporations with poorly performing management teams whose boards have failed to
discipline them are subject to having a controlling block of their shares acquired by
investors who will replace the incumbents with new managers who will realize the
firms' potential and new directors who will ensue that they do this.
This market for corporate control was especially active in the 1 980s, with mergers
and acquisitions, takeovers, and leveraged buyouts. During that period, managers and
boards discovered many ways to protect their companies from irresponsible raiders
and themselves from market discipline. The market has quieted since then, although
it remains a force. The protective measures remain, however. The diminished
disciplinary role of the market for control coupled with the increased power of
incumbents to entrench themselves has led a number of institutional investors to look
for new methods to increase their abilities to monitor and discipline management.

The Mechanics of Control

Two options are open to dissatisfied shareholders: "exit"-sell their shares-and
"voice"-vote against the current board. 20 Neither of these is a perfect solution. Selling
means suffering the low price that attaches to the shares of a poorly performing firm.
Casting a vote against management may be a futile gesture. This is especially the case
because management and the board typically control the processes for nominations
in board elections and for putting resolutions to a shareholder ballot. There may
simply be no alternative to vote for. Tender offers and proxy contests are responses to
these problems.
TENDER OFFERS In a tender offer, the person or entity making the offer invites
shareholders to sell (tender) their stocks at an announced price. The offer may be for
some or all the stock, and it may or may not be conditional on a certain number of
shares being tendered for sale. It may also be a "multitiered" offering, at least in the
United States, with a certain fraction of the shares being offered one price and the
remainder being paid lower prices. (The United Kingdom's Takeover Code bans
paying less to minority shareholders once 30 percent of the shares have been
acquired. ) Firms themselves might make tender offers to buy back their own stock in
recapitalizations that decrease retained cash or increase debt while lowering equity.
This is how management buyouts and other going-private transactions are accom­
plished. The other use of a tender offer is for outsiders to gain control of the firm in
a takeover, whether friendly or hostile.
The idea in a hostile tender-offer takeover is that the raider, having obtained
the firm's stock, will vote it to replace the existing board. The new board will
then order management to implement a new strategy designed to generate better
performance. Often, the current management will be replaced and new management

20 These terms are due to Albert Hirschmann, who has applied the concepts much more broadly.
See his Exit, Voice and Loyalty (Cambridge, MA: Harvard University Press, 1 970).
 509
brought i n to implement the strategy, or the firm may be merged into one already
held by the buyer. 2 1 The promise of the better performance means that the raider can
Financial
Structure,
bid more than the current price of the shares. Ownership, and
PROXY CONTESTS AND SHAREHOLDER RESOLlITIONS Few shareholders i n a large, Corporate Control

broadly held firm find it worthwhile to attend the annual shareholders' meeti ng at
whi ch directors are elected and other shareholder votes are taken. Instead, they gi ve
a proxy to someone else to vote on their behalf. M anagement includes a form soliciti ng
proxies for itsel f when it sends notice of the annual meeti ng and normally recei ves
the vast bul k of the proxies. In a proxy contest, another enti ty seeks to obtai n
shareholders' proxi es i n order to vote them agai nst management and the current board.
Thi s al lows changi ng control without the expense of havi ng to acquire a controlling
bl ock of shares from the current owners. H owever, proxy contests are expensi ve and
difficult to organize and wi n, and the gai ns from reorganizati on are broadly spread
among all the stockholders. Thus they may suffer from a free-ri der problem.
A shareholder resoluti on is a measure requesti ng or instructi ng the board and
management to follow particular policies. Examples might be to renounce a previously
adopted poison pill 22 or to avoid usi ng ani mal s i n research. They represent stockholder
attempts to direct the affairs of the corporati on wi thout repl aci ng the board or
management. Shareholder resoluti ons are restricted by law i n many countri es. For
example, under U . S. S ecuri ties and Exchange Commi ssi on rules i n effect through
1990, shareholders had no ri ght to vote on executi ve compensati on issues, whi ch
were deemed to be an ordi nary busi ness deci sion that is reserved for the board of
directors. This parti cular rule was, however, revamped i n 1991 i n li ght of the popul ar
concern and complai nts about excessi ve CEO pay.
BANKRUPTCY AND CONTROL In bankruptcy, the control structure of a corporati on
shifts. In an i nvoluntary bankruptcy (a Chapter 7 bankruptcy i n U . S . law), the court
appoi nts a recei ver or trustee to wi nd up the busi ness and li quidate its assets.
M anagement and the board, as representati ves of the most j uni or claimants, the
stockholders, lose control, and to the extent that any busi ness decisi ons are made,
they are supposed to be taken i n the i nterests of the more seni or cl aimants. U nder
Chapter 11 voluntary reorganizati on bankruptcies, current management retai ns some
control: It is all owed time to develop a plan to reorganize the firm, pay some fraction
of its debts as settlement of the claims agai nst it, and repositi on it to be profitable.
For a plan to be accepted, two-thirds of the creditors. of the preferred stockholders,
and of the shareholders must each vote for it. If no agreement is ultimately reached,
then the credi tors can force the firm i nto Chapter 7 bankruptcy and li quidati on.
Meanwhile the firm continues to operate. If the credi tors believe that management is
wasti ng the exi sti ng value of the firm and further depleti ng the resources avai labl e to
meet their claims, they may seek to have the court appoi nt a trustee to take over
operati ons. Thi s is unlikely to happen, however, until management has had reasonabl e
time to devel op a plan and have it wi n appr oval unl ess the depleti on of assets is too
egregi ous. For example, the credi tors of now-defu nct Eastern Airlines complai ned
bitterly but with li ttl e effect about the addi ti onal losses it incurred during its extended
period of operati ng in bankruptcy.

21 Even if only a majority of the shares are purchased, then the acquired firm may still be merged
into another by vote of the new board. The stockholders in the acquired firm whose stock was not purchased
originally must be compensated for their holdings when the firm ceases to exist, but they are largely unable
to prevent the merger.
22 These are described in Chapt er 6.
 510
Finance: Takeovers and Restructurings in the United States in the 1980s
Investments,
Although the market for corporate control was especially active throughout the 1 980s,
Capital Structure,
and Corporate
there were some real differences in the sorts of transactions pursued over the course
Control of the decade. In the early period much of the activity involved oil companies' buying
one another, acquiring other firms in a continuing pattern of diversification, and
being subject to hostile attacks themselves-sometimes from other oil companies.
Food, lumber, banking and finance, and insurance firms were also heavily involved
in the activity in this period. Later, there was much less industry specificity, and
towards the end of the decade it seemed that any firm might be a target. However, it
does seem that targets tended to be firms with low market valuations relative to the
replacement costs of their assets.
Junk bonds were not a factor until the middle of the 1980s, when Michael
Milken showed their potential and created a large secondary market in them. Earlier,
financing was by cash, stock, or private placement of subordinated debt. LBOs also
were not a major factor in the early part of the decade, but they became so in the
second half. The nature of the LBOs changed as well. Early on they tended to be
relatively small and to involve divisions of corporations or firms in a single business.
Until 1 98 5 the organizations involved in buyouts by KKR averaged only 6, 3 5 1
employees and $834 million in revenues. 23 Then came the Beatrice and Safeway
deals. With the continuing flow of smaller buyouts, these raised the averages through
early 1 988 to 3 7, 000 employees and $4. 8 billion in revenues. However, Safeway
alone had more revenues and almost as many employees as all of KKR's previous
buyouts together. The RJR Nabisco deal of course dwarfed them all. This pattern was
not unique to KKR. There were still some buyouts of small, single product firms, but
the later part of the decade also saw transactions involving huge, diversified companies.
This activity continued, at a much slower pace, into the 1 990s.
Throughout the period, despite all the attention focused on hostile takeovers,
they were actually a small part of the activity in the corporate control market. For
example, in 1986, there were only 40 hostile tender offers out of the 3 , 3 36 transactions
listed in the W. T. Grimm Mergerstat Review series. There were another 1 1 0 tender
offers unopposed by management, and the rest were negotiated transactions agreed to
by management (although perhaps with the threat of a hostile tender offer in the
background). 24

TAKEOVER PREMIA In the average control change, the existing shareholders received
a price that was 30 to 50 percent higher than the price at which their shares had
previously been trading, and sometimes this takeover premium was over 1 00 percent.
For example, shares of RJR Nabisco stock had been trading in the mid-forties before
the bidding war that began after management proposed a buyout. The final price paid
by KKR was $ 1 08 a share: $80 in cash, with the rest in preferred stock and debentures.
Similarly, when Bridgestone Tire launched its bid to take over Firestone, the total
market value of Firestone's shares was about $ 1 billion. The final price paid was $2. 6
billion. Firestone's shareholders thus enjoyed a takeover premium of about $ 1 . 6
billion, or 1 60 percent. There is some evidence that the magnitude o f premia rose
over the decade, and they were much higher than in the 1 960s. Over the ten-year
period up to 1986, the premia paid to selling-firm shareholders totaled $346 billion

23 William F. Long and David J. Ravenscraft, "The Record of LBO Performance," Arnold W.
Samctz, ed. , The Battle for Corporate Con trol (Homewood, IL: Business One Irwin, 1 99 1 ).
24 Michael C. Jensen, "Takeovers: Their Causes and Consequences, " Journal of Economic
Perspectives, 2 (Winter 1 988), 2 1 -48.
 511
(in 1986 dollars), which was equal to almost 45 percent of the value of corporate Financial
dividend paymen ts over the same period. 25 Structure,
In contrast, the shares of acqui ring fi rms on average did not change in value or Ownership, and
fell sligh tly during these transacti ons. 26 This average, however, masks some significant Corporate Control
variability. For example, the shareh olders of Occidental Petroleum l ost $ 3 6 5 million
when their fi rm acq uired Midcon as a white knight, and Marri ott's stockh olders lost
$ 1 62 mi llion when their fi rm acquired Saga in a hostile takeover that paid S aga's
shareh olders a premium of $148 million.2 7 In the 29 h ostile takeover attempts in the
United S tates from 1 984 through 1 986 with values over $ 50 million where the change
in bidding fi rms' stockh older wealth could be computed, there were losses in 1 6 cases
and gains in 1 3 . The losses totalled $1. 25 billi on; the gains totalled $ 508 mi llion . On
average, the buyers' stockh olders lost $ 1 5 milli on, but thi s i s n egligible n ext to the
average transaction value of $ 1 . 74 billion. 28
The stockholders at the time of the transacti ons thus did very well in total, with
most of the · gains apparently accruing to the stockholders in the acq ui red fi rm. But
do their gains represen t value creati on and increased efficiency or something el se?
What accounts for the takeover premia?
POSSIBLE SoURCES OF THE GAINS

Overpaying. One obvi ous possi bility is that the buyers paid too much. This might
have occurred as a result of "hubri s, " with the acquirers' overestimating their ability
to add value, but it would have been especially likely when the management of th e
acqui rin g fi rms were intent on building empi res, diversifying, or pursuing other
strategies th at were n ot in stockh olders' interests. Thi s certainly cannot account fo r
the wh ole of the premia, h owever, because if it were we sh ould h ave expected to see
bidders' stock prices fall approximately one-for- on e wi th the rise in the targets' prices
when they attempted acquisiti on s. Nevertheless, it may h ave been a factor, at least
in some cases: There is evidence that bidders' stock prices did tend to decline when
the acquisi ti on s appeared to be motivated by managerial considerati on s. 29
Stock Market Mispricing. Some critics claim that the premia were simply evidence
that stock market valuati on s are highly inaccurate and that the buying fi rms were
acquiring underpriced fi rms wh ose long-term strategies were n ot properly understood
and valued by the market. Thi s i s why they could afford to pay the premia. The
accuracy of stock market valuations remains a h otly disputed topic, and i t i s certainly

25 Ibid.
26 Michael Bradley, Anand Desai and E. Han Kim, " Synergistic Gains from Corporate Acquisitions
and their Division Between the Stockholders of Target and Acquiring Firms, " foumal of Financial
Economics, 2 1 ( 1 988), 3-40.
27 Sanjai Bhagat, Andrei Shleifer, and Robert W. Vishny, "Hostile Takeovers in the 1 980s: The
Return to Corporate Specialization," Brookings Papers: Microeconomics 1 990 (Washington: B rookings
Institution, I 990). These figures are based on estimating (via CAPM methods) what the price of the stock
in question would have been on the date the transaction was completed versus what it actually was. Such
losses were not unique to North American takeovers. A similar calculation done by Evan Davis and Graham
Bannock ["The Nestle Takeover of Rowntree," David Hume Institute Paper 31 ( 1 99 1 )) indicates that the
shareholders in the Swiss firm Nestle lost £500 million in stock market value when their fi rm acquired
Rowntree, the British cand'y company, for £2.6 billion, or about 2. 7 times what its market value had been
ten months earlier.
28 Bhagat, Shleifer, and Vishny, op. cit. These acquisitions were by publicly traded companies; the
returns to raiders operating through private companies could not be computed.
29 Randall Morck, Andrei Shleifer, and Robert W. Vishny, "Characteristics of Hostile and Friendly
Takeover Targets," Alan J. Auerbach, ed. , Corpora te Takeovers: Causes and Consequences (Chicago:
University of Chicago Press, 1 988), pp. 1 0 1 -29.
 512
Finance: possible that this argument is correct. However, the available evidence indicates that
Investments, LBOs and takeovers were concentrated on firms and industries that were not R&D­
Capital Structure, intensive and that firms with high R&D expenditures were not more vulnerable to
and Corporate takeovers, as this undervalued stock theory would suggest they should have been. 30
Control The takeover targets in the U nited States between 1 977 and 1 987 had accounted for
less than two percent of total U. S. R&D spending even before the takeovers occurred. 3 1
A related argument is that the raiders diverted firms from a proper long-term
focus to an emphasis on short-term performance, which the market is asserted to
value relatively too highly. However, except in the oil business (which was arguably
overexpanded anyway), there is little evidence of cutbacks in R&D or other investments
following mergers or takeovers as would be expected if the buyers were intent on short­
term gains. There is, on the other hand, evidence of cutbacks in R&D and other
investments after management buyouts. 32 This is hardly support, however, for an
argument that management had previously been producing unrecognized value by
aiming for the long term, for when managements' ownership stake increased, it
reduced investment.
A separate line of evidence regarding the possible mispricing of target shares
involves the fact that when takeover attempts were defeated, the share price of the former
targets tended to drop back to essentially their levels before the attempt. Given the close
attention paid to firms when they are "in play, " if there were really significant informa­
tion that had not previously been incorporated in the share price, one would expect it
to come out and for the stock price to have adjusted permanently to reflect it.
A stronger version of the underpriced-assets theory claims that the raiders were
not even paying the full value of the firms they acquired. Thus, management had a
duty to resist. The buyers in these takeovers did not obviously enjoy especially high
returns after the takeovers, however, as might be expected if their purchases were a
bargain (although performance does seem to have improved in many LBOs). In fact,
as competition in the market for control intensified and court rulings forced companies
"in play" in effect to conduct auctions, allowing and even encouraging competing
bids, we might expect to see some overpaying for firms because of the "winner's curse"
(see the box on the winner's curse for an explanation of this phenomenon. )
Transfers and Breach of Trust. Other critics argue that part of the premium
represented not an increase in total value but transfers of value away from others with
claims of the firms' cash fl o ws. Control changes sometimes resulted in reductions in
employment and wages. For instance, the reduction in total wages when Carl Icahn
took over 1WA in 1 98 5 was about $200 million, which in itself was enough to justify
the takeover premium. In Youngstown, Ohio, the layoffs that followed the 1 979
takeover of Youngstown Sheet and Tube were followed within a year by a reduction
in the median home sale price of 23 percent. 33 The holders of the targets' bonds or

30 Abbie Smith, "Corporate Ownership Structure and Performance: The Case of Management
Buyouts," draft, U niversity of Chicago Graduate School of Business ( 1 989); Bronwyn H . Hall , "The Effect
of Takeover Activity on Corporate Research and Development, " Alan J. Auerbach, ed. , Corporate Takeovers:
Causes and Consequences (Chicago: University of Chicago Press, 1 988); and U. S. Securities and Exchange
Commission, Office of the Chief Economist, "I nstitution Ownership, Tender Offers, and Long-Term
Investment" ( 1 98 5 ) .
3 1 Bronwyn Hall, op . cit.
32 Stephen N. Kaplan, "The Effect of Management Buyouts on Operating Performance and Value, "
foumal of Financial Economics, 24 (October 1 989), 2 1 7-54.
33 These last two calculations are reported by Andrei Shleifer and Lawrence S ummers in their paper,
"Breach of Trust in Hostile Takeovers," A. Auerbach, ed . , Corporate.Takeovers: Causes and Consequences
(Chicago: University of Chicago Press, 1 988).
 5 13
Financial
Structure,
Ownership, and
The Winner's Curse Corporate Control
Auctions are often used for the sale of properties of unknown value.
U ncertainty is present whether the obj ect of the biddi ng is drilli ng rights on
an unexplored tract of land (Is there oil? H ow much?), timber rights to an
area of forest (H ow many board feet of each type of wood is there?), or shares
of a company (What is the profit potential of the widget division?). B idders
do their best to deal with that uncertainty by gathering information and
making estimates of the underlying values. Because the factors underlyi ng
these estimates are themselves subj ective and uncertain, there can be
considerable variation among the bidders' estimates of the true underlying
val ue. .
This variation in estimates has an interesting consequence that has
come to be called the winner's curse. A bidder is more likely to win when
its estimate is too optimistic and it bids high than when its estimate is too
pessimistic and it bids l ow. O n average, then, the winning bidder tends to
be one who has overestimated the value of the object being sold. In a similar
fashion, contractors who bid for j obs may find that their bids are most likely
to win when they have been too optimistic about the cost of doing the j ob.
A bidder ignores these tendencies at its peril: N aive bidders may pay too
much for the items they buy at auction or ask too l ow a price for the services
they sell.
The winner' s curse of overbidding was first noted by professionals in
the oil industry. It shows up repeatedly in experimental settings. Experimental
subj ects are asked to bid for an envel ope containing an amount of money
after each has been given an unbiased estimate of the amount actually in the
envelope. Predictably, the winner bids more than is actually in the envel ope
because his or her estimate was higher than the true amount and he or she
failed to account for the fact that he or she would tend to win only when
this was the case.
In the context of takeovers where several bidders are competing to
acquire a firm, the winner's curse would mean that the successful bidder
would frequently be one that overestimates what the target firm is worth and
consequentl y pays too much for it.

preferred shares al so might have l ost if the value of their assets was reduced in the
takeovers, with the gains being captured by target stockholders. Certainly, the value
of RJR N abi sco's bonds were adversely affected in its takeover. Other takeovers resulted
in reducti ons in taxes paid to the government because the increased debt reduced
corporate tax liabilities, the upward revaluation of assets after the takeovers permitted
greater deductions for depreciati on and did not require payment of corporate capital
gains taxes, and previous tax l osses in one firm could be applied against the profits of
the other after a merger.
Under the hypothesis of no wealth effects, simple transfers do not affect efficiency
directl y. Nevertheless, if the transfers are the returns from reneging on implicit
contracts, there can be long-term costs. Empl oyees, for example, may be much less
willi ng to invest in firm specific assets and to remain l oyal to the firm if they fear that
the implicit promise to reward these sacrifices will be broken after some future change
in control.
 514
Finance: Acc ording to recen t studies, however, these transfers can account for only a
Investments, fraction of the increase in the market value of firms. It is difficult to get any systematic
Capital Structure, evidence on wage changes, particularly relative to industry trends, but the one study
and Corporate to examine this issue found that wage cuts would explain a trivial perce ntage of the
Control premium. 34 Bhagat, Shleifer, and Vishny found that layoffs occurred in only a fraction
of all hostile take overs an d that they tended to be concentrated in the white-collar
staff, particularly at corporate headquarte rs. Furthermore, layoffs were on average
three times as large a fraction of the work force whe n the target successfully resisted
as when the bidder was successful, and two- thirds larger when target management
found a white knight to acquire the firm rather than its falling to the hostile bidder.
In some cases the savings from labor are a significant part of the premium, but those
cases are a small fraction of the total. 35 In any case, if the organization were inefficiently
overstaffed, the reallocation of labor might increase efficiency. S imilarly, transfers
from bondholders and holders of preferred shares may have been substantial in some
cases where there were not covenants in place, but it seems that bondholders on
average did not lose from buyouts (at least in the early part of the peri od)36 or hostile
take overs. Finally, tax considerations were clearly a factor in some control changes,
especially where leverage increased significantly, but again they are not enough to
explain more than a fraction of the premia on average. 37
Value Creation. Acc ording to Bhagat, Vishny and Shleifer, the maj ority of the
increase in the case of hostile takeovers was accounted for by the high prices paid by
the final buyers of the divisions being sold, either in the original transaction or in the
asset sales (bust-ups) that often followed. In most cases, these buyers were other firms
in the same industry who used the purchases to expand their own operations. Advocates
of take overs argue that these buyers, firms that are specialists in the industry where
they are buying, were better able to manage the divisions they acquired than the
conglome rate firms had been, and that improved manageme nt accounted for the
increase in value. 38
These factors may also have been central in other control changes, but there
are at least three other possible sources of value. First is simply the possibility that
inc ompete nt or self-serving managers were replaced. Second is the increased intensity
of incentives that the LBOs typically provided: Managers came to hold a much higher
fraction of the equity in the firm, were subjected to the performance pressures from
the high debt levels, and-where a buyout firm was involved-were made subject to
intense monitoring. Finally, in the oil and tobacco businesses and perhaps some
others, the premia may have been simply a matter of returning free cash H ow to the
shareholders, with value being increased by managers' being unable to waste the
corresponding resources on economically unj ustified investments and diversification.

H Joshua Rosett, "Do Breaches Explain Takeover Premiums: The Evidence on Union Wages"
(1990), as cited in Sanjai Bhagat, Andrei Shleifer, and Robert Vishny, " Hostile Takeovers in the 1980s:
The Return to Corporate Specialization, " Brookings Papers: Microeconomics 1 990, 1 (Washington: Brookings
Institution, 1990).
3 5 Bhagat, Shleifer and Vishuy, op. cit.
36 Kenneth Lehn and Annette B . Poulsen, "Sources of Value in Leveraged Buyouts, " M. Wei­
denbaum, ed . , Public Policy Toward Corporate Takeovers (New Brunswick, NJ: Transaction Publishers,
1 988).
37 Bhagat, Shleifer and Vishny, op. cit.
38 Critics sometimes contend that even this portion of the increase in value is partly a transfer,
resulting from increased market power that enables the larger firm to charge higher prices rather than from
any real social gain, but they offer no evidence to support this thesis .
 515
Takeover Defenses Financial
Various practices were spawned in the 1980s as the boards and ma nagement of Structure,
potential target firms sought ways to defend their firms a nd themselves against Ownership, and
attempted hostile takeovers. S ome of these seem justified as devices to defend legitimate Corporate Control

interests in an efficient wa y against threats by corporate ra id ers or to strengthe n

' ma nagement's barga ining position to allow it to get the best possible price for the
shareholders when the firm is eventually sold. Others, however, raise disturbing ethical
and public policy questions.
INADEQUATE AND COERCIVE OFFERS At first glance, it ma y be hard to see why there
would be any need for takeover defenses at all, other tha n to protect entrenched
ma nagers a nd their allies on the board. If a takeover bid is bel ow the underlying worth
of the compa ny, shareholders presuma bly will rej ect it. H owever, matters a re actually
somewhat � ore complicated.
First, it may be difficult for shareholders to know how much the firm is worth.
They may thus be tempted to accept a n offer when they might actually be a ble to get
a better one. If ma nagement and the board of the target are working in their
shareholders' interests, they should try to ensure that the firm is' sold for the highest
possible price. H owever, finding out how much ca n be obtained, even from the
current bidder, may take time. For example, it might be necessary to find a nother
bidder, so that competition could generate the highest possible price. Takeover defenses
that allow ma nagement and the board to delay the process ca n thus provide value for
shareholders.
Also, offers may actually be coercive in the sense that individ ual shareholders
might feel pressure to sell even when they know that as a group they would be better
off rej ecting the bid. I t is ea siest to see how this ca n ha ppen by considering a
conditional offer to buy only a controlling stake in the firm. (S imilar coercion could
be felt by small stockholders in other forms of takeover bids.) Suppose that a raider
makes a tender offer for 5 1 percent of a firm's shares at some price P 1 , with the offer
conditional on at least 51 percent of the shares being tendered. If the target is acquired,
the buyer ca n vote to merge the target into a nother firm it owns, paying some lower
amount p 2 for any outstanding shares that were not acquired initially.
Consider now the incentives facing a small shareholder-one small enough to
ignore the impact of his or her tender decision on the likelihood the 5 1 percent level
of tendering is reached. If the fraction of shares tendered is less tha n 5 1 percent, then
it does not matter whether the individual tenders because no transactions will take
place a nyway. If 51 percent or more are being tend ered, then the individ ual gets p 2
by not tendering but ha s some cha nce of receiving p 1 if he or she tenders. Thus,
tendering is a "no-lose proposition": only by tendering ca n the shareholder have a
chance of receiving the high price, p 1 . . Yet, there is no guara nty that P 1 , let alone p 2 ,
has to be a n attractive price. I n the U nited States, state la ws would ensure that p 2
could not be too low relative to the pre-merger price of the target's stock, but both p 1
and Pz might be much less tha n the stockholders thought the firm wa s worth, and
they would still feel compelled to tender their shares! 39

39
Strikingly, if the bidder simply attempts to buy a controlling interest and cannot prevent
shareholders who do not tender from enjoying the full value of their share of the firm after the takeover,
then the best strategy is for the small shareholders not to tender unless they are offered the full post-takeover
value of their shares. Again, if no control change is going to occur, it does not matter what any individual
does. If a change will occur even if the individual does not tender, then by tendering he or she gets the
offered price, p. By not tendering, he or she gets v, the value under the new ownership. As long as
p < v, it is better never to tender. Thus, takeover bids would succeed only when p = v, which would
 516
Finance: Although such coercion is theoretically possible, note that it depends on there
Investments, being only a single bidder. Thus, its empirical significance and the justification it
Capital Structure, gives for takeover defenses is questionable. In particular, a management-led buyout
and Corporate would seem to eliminate the threat of coercion (unless the original bid is from
Control management itself!).

TYPES OF DEFENSES The variety of defenses developed is truly remarkable. 40 Poison

pills (discussed in Chapter 6) raise the cost of a takeover by giving current shareholders
extra claims if a control change occurs or even if a single entity acquires a threatening
large holding of stock. Differential voting rights can give extra votes to shares that
have been held for extended periods, so that a recent buyer of half the shares has only
a small fraction of the votes. Scorched earth policies of various sorts deliberately
reduce the firm's value to the bidder, even if it reduces value to shareholders in the
process. For example, in resisting Mobil's takeover attempt, Marathon Oil gave USX
an option to acquire its "crown jewel, " the Yates oil field, at a bargain price if some
other company gained control of Marathon.
A related but less-extreme technique involves restructuring the firm to make it
harder for the raider to finance the debt taken on in the merger. For example, the
firm might spin off attractive divisions that the raider might have counted on selling
after the merger to raise funds for debt repayment. It might also reduce corporate cash
holdings that also would facilitate servicing the post-takeover debt. Share repurchases
are one possible way to accomplish this; another is to buy other companies at possibly
inflated prices. Classified or staggered boards, where only a fraction of the members
are up for election each year, and supermajority rules, which require as much as 90
percent of the votes to effect a control change, make it more difficult to gain control
after buying what would otherwise be a controlling interest. These can be supplemented
by putting a large block of shares in an Employee Stock Ownership Plan, on the
assumption that the shares will be voted for management because employees will be
worried about losing their jobs in a takeover. Antitakeover statutes passed by the state
government where the firm is incorporated, such as those passed by Minnesota to
protect Dayton Hudson, are also a possibility, and the firm can also move its state of
incorporation to one with protective rules. This list just scratches the surface.
Each of these would be just a means for entrenching managers but for the fact
that, for many of them, their application can be waived at the option of the board of
directors. This means that they can be used to strengthen the board's bargaining
position as well as to thwart an unwelcome offer. Of course, they are waived when
the board and management is positively disposed to the offer. To get a flavor of the
effects of these different sorts of measures, we consider three in more detail.

GoLDEN PARACHUTES As discussed in Chapter 1 3, a golden parachute is a clause in

a compensation contract providing for very attractive benefits in the event that a
manager leaves after a control change. Among the most important arguments that
have been advanced in favor of golden parachutes is that career executives have a
legitimate right to expect pn.,tection of the· rewards that have been earned by years of
hard, skillful work. Moreover, without adequate protection, executives would mount
fierce resistance against any attempt to undermine those rights, making worthwhile

leave no return for the bidder (even to cover its costs) unless it were able to obtain some block of shares
before making the tender offer.
40 Sec J. Fred Weston , Kwang S. Chung , and Susan B. Hoag , Mergers, Restructuring and Corporate
Con trol (Englewood Cliffs , NJ: Prentice Hall , 1 990) , Chapter 20 for a much more complete discussion.
 517
Fi nancial
Structure,
Ownership, and
Paramount and Time Warner: In Whose Interests? Corporate Control
Time Inc. , the largest U . S. magazine publisher (Time, Fortune, People, Sports
Illustrated) acquired Warner Brothers, the movie, television production, and
record company (Warner/Reprise and Atlantic) in early 1 990, but most
observers regarded the transaction as a Warner takeover of Time. Time and
Warner were in merger negotiations, a transaction that the shareholders
would have had to approve, when Paramount Communications (movies and
sports plus publishing-it owns Prentice Hall) made a $200 a share, all-cash
offer for all of Time's stock. Backed by a landmark court ruling that a board
had a right to "just say 'no' " to a takeover offer without consulting
shareholders, Time bought Warner, paying what knowledgeable observers
regarded as a hefty premium. Carrying out the transaction this way meant it
did not have to go to a vote of Time's shareholders, who might have preferred
to sell their shares to Paramount.
Time's management argued that its intrinsic worth in combination with
Warner was about $250 a share, so that the Paramount offer of $200 was
totally inadequate. However, the market price of Time Warner never got
above $ 1 2 5, and in June 199 1 , Time Warner's efforts to eliminate much of
the debt acquired in the merger sent its price plummeting to less than $ 8 5
per share.

For further information and analysis, see Gregg A. Jarrell, "The Paramount Import of
Becoming Time-Warner: A Present-Value Lesson for the Lawyers," The Wall Street Journal
(July 1 3 , 1989), A- 1 4.

takeovers less likely to succeed, blocking smooth transitions, and diverting attention
from value-creating activities.
The two arguments are readily construed in terms of efficiency: Executives may
be unwilling to invest in their careers without some assurances that the return on
their investments will not be easily appropriated by outsiders. Moreover, unless their
investments in their careers are protected, managements' efforts to resist takeovers will
divert their attention from the business, lead them and the raiders to incur huge legal
and professional fees, and generally use up resources in a way that is wasteful because
it produces no increase in output. These latter costs are influence costs; they arise
from a process in which individuals fight to defend their unprotected rents.
In outline, these arguments in favor of golden parachutes are just the same as
the other arguments we have made about the advantages of secure ownership rights.
It is difficult to make a general evaluation of arguments like this because the premises
of the argument need to be verified in each case. Generally, if we suppose that
executives have made a valuable specific investment, it is curious that raiders would
so often be eager to dismiss them. It seems more reasonable to suppose that raiders
dismiss managers who are performing badly and who will vigorously resist the takeover
because there is no other firm willing to offer comparable employment terms with so
little evidence of good performance.
In any case, golden parachutes are typically given only to the most senior
executives, who have already made most of the investments in firm-specific capital
that they might ever undertake. The incentive arguments in terms of encouraging
'
investments, at least by these individuals, thus seem irrelevant.
 518
Finance: The first objection to golden parachutes is that they defend entrenched managers,
Investments, not the firm, and that they are costly for stockholders. Moreover, if they are too large,
Capital Structure, managers may be too ready to encourage a control change and may not defend the
and Corporate stockholders' interests well. Further, they might encourage valuable managers to leave
Control the firm after a control change, and this would make desired takeovers less likely.

GREENMAIL The main problem with defensive measures undertaken to fend off
corporate raiders is that they can also be used to defend an incompetent management
or to defend management's interests even when they are in conflict with the interests
of shareholders and others. One of the most ethically troubling examples of this is
the case of greenmail, which is essentially (though not legally) a bribe paid to raiders
by a firm's managers, using the shareholders' money, to induce them to break off the
raid.
An example of greenmail is the case of Walt Disney Productions. ln early 1 984
the price of Disney stock was in the range of $ 5 0 to $ 5 5 per share. Raider Saul
Steinberg began buying shares and prices began to rise. On June 4 of that year, an
article in Forbes magazine estimated that the company's assets, if sold separately,
would fetch a total price of $ 1 1 0 per Disney share. On June 8 Steinberg offered to
buy 37 percent of Disney's shares at $67. 5 0 per share. After three days of negotiations,
Disney Productions agreed to buy Steinberg's shares from him at $70. 83 per share,
plus $28 million for his "investment expenses. " The price of Disney shares promptly
plummeted to less than $ 50 per share.
From a small shareholder's point of view, greenmail is company money that
has been paid to a raider by company management to leave things just as they were
before the raid was begun. Someone in this tale is a villain, but who? Should Disney's
management be condemned for spending the shareholders' money to protect its own
position of control? Or should they be congratulated for keeping a productive
organization intact despite the reckless machinations of greedy financiers? Those who
believe that a company's share price represents the best available estimate of its value
(including most mainstream financial economists) will believe that Disney management
destroyed value by its actions, as evidenced by the plunge in price after the greenmail
was paid. Those who distrust market evaluations, thinking them too much subject to
fads and manipulations, may think that the company was undervalued all along and
that Disney management was actually acting in the interests of its long-term
shareholders. The evidence on the accuracy of share prices as an estimate of value is
mixed, and the debate rages on.

VOLUNTARY RESTRUCTURING Many executives were convinced by the successes of

the 1 980s takeover wave and the associated increase in share prices that the
organizational strategies of the corporate raiders-tighter focus, higher debt, and
increased incentive pay-actually did add value. Some firms implemented programs
of voluntary restructuring that were intended to duplicate the benefits of leveraged
buyouts while avoiding their costs. ln these reorganizations, stock was repurchased
with debt financing, and sometimes the changes went deeper, with management
teams in operating divisions being given a larger stake in the divisional returns. In
effect, corporate headquarters became the lender to the divisions, and the divisions
became increasingly leveraged, even though the firm itself retained a more traditional
debt-equity ratio that allowed it to weather economic downturns without cutting back
too severely on its long-term investments.
Another aspect of restructuring was the sale of unrelated divisions. Conglomerate
firms come under intense pressure from insiders to provide subsidies to declining
divisions in order to reduce the hardships suffered by employees. Companies have
 519
long sought ways to isolate unprofitable divisions to mitigate this pressure . One Financial
example we cited earlier was in 1 974, when increased oil prices made the Japanese Structure,
aluminum industry unprofitable. Mitsubishi Chemical, Sumitomo Chemical, and Owne rship, and
Showa Denko separated the unprofitable subsidiaries "to isolate the problem and the Corporate Control
losses. "4 1 In general, spinoffs have been of underperforming units. This makes
economic sense if others could manage them better or if keeping them in the firm
leads to large influence costs. One of the major aims of restructuring is to create
focused units, with fewer opportunities for managers to gain by competing for shares
of a fixed pool of resources .
Preempting a raider is also an excellent defense . A problem is that many of the
steps involved in a genuine restructuring are difficult for shareholders to distinguish
from scorched earth policies designed to thwart the raider while maintaining control
for the current incumbents. There is also a question of why, if the restructuring is a
good thing for shareholders, management waited until threatened by a takeover to
implement it. Influence costs provide a possible explanation . Restructuring reallocates
rents within the firm and would normally be strongly resisted. A raider threatens to
take some of the rents for itself and pass some to shareholders. This would severely
diminish the amounts that would be available to current claimants. In this context,
resistance to management's restructuring, which may keep more rents in the firm
than the raider would leave, may be lessened.

The Aftermath
The U . S. takeover boom went bust in 1 990. 1WA defaulted and faced repossession
of a large share of its fleet; Drexel Burnham Lambert went bankrupt; and M ichael
Milken was sentenced to ten years in a federal penitentiary for securities law violations.
Corporate bankruptcies exploded , the volume of merger and acquisition activity was
cut in half, and LBOs all but disappeared. Dozens of troubled companies that had
taken on high debt in the 1 980s sought to negotiate away their high interest obligations,
replacing junk bonds with equity or lower-yield debt. Other, more successful, leveraged
companies reduced their debt loads and even successfully issued new equity.
Meanwhile, the Federal Deposit Insurance Corporation and the Resolution Trust
Corporation, charged with cleaning up the Savings and Loan debacle, sued Milken
and seven pages worth of other defendants, charging that the junk bond market had
been a fraud that had duped the S&Ls into disaster.
Many people clearly hoped that the threats, costs, and discipline of the market
for corporate control were gone for good . Others, however, were worried about exactly
the same possibility.

ACTIVE INVESTORS AND CORPORATE GoVERNANCE REFORMS The takeover boom clearly
produced excesses and inefficiencies . Yet the threats of takeovers also made manage­
ment much more conscious of shareholders' interests . The collapse of the market for
corporate control in the recession of 1 990, coupled with the strong takeover defenses
that had been erected through the decade, led some investors to seek alternative means
of protecting their interests.
A number of institutional investors, and especially the public employee pension
plans of states such as Wisconsin and California, were leaders in this . They reasoned
that attempting an active investment strategy designed to beat the market was futile
both because they had no special access to information that was not already available
to other sophisticated investors and because their shareholdings were so large that

41 James C. Abegglen and George Stalk, Jr. , Kaisha, The fapanese Corporation (New York: Basic
Books, 1 985), p. 2 5 .
 520
Finance: their purchases and sales of companies' shares affected prices. Thus, the only way
Investments, they could really improve the performance of their portfolios was to improve the
Capital Structure, performance of industry overall. If they could not rely on the pressure of takeover
and Corporate threats to keep management focused on shareholder value, they would have to become
Control more active monitors of management.
With this objective, they began introducing shareholder resolutions to dismantle
various defenses corporations had put in place and to shift their states of incorporation
away from ones that had enacted especially restrictive antitakeover measures. Although
many of these failed to win shareholder approval when actually put to votes, others
were accepted by management and the corporate boards in negotiations with the
institutional investors. They also used the leverage they possessed in the control
contests that did occur to win concessions from management. For example, Lockheed
Corporation's management agreed to nominate a number of outsi de directors suggested
by the institutional investors to its board and to meet regularly with large shareholders
in order to win their support in a control battle it faced with investor Harold Simmons.
Various proposals were also put forward to facilitate proxy fights and successful
shareholder resolutions, and the institutional investors lobbied the Securities and
Exchange Commission and Congress to adopt these.
A major element of these reform efforts was to redefine the functioning of the
board to make it more effective as the agent of the shareholders. One interesting
proposal was that the institutional investors should identify and develop a cadre of
professional board members. Each of these would be elected to serve on several
boards, making a full-time job out of these responsibilities, and would be evaluated
by a group representing the large shareholders at reelection time. 42 CalPERS, the
California state public employees retirement system, was actively investigating this
proposal in 1 99 1 with the support of some of the corporations of which it held shares.

ALTERl\ATl\ ' ES TO THE PCBLICL Y HELD CORPORA TIOl\

The problems of corporate control stem from many of the same features that are
recognized as particular strengths of the corporate form. This fact raises the question
of whether some alternative organizational form might be more efficient.
The free transferability of ownership shares permits extensive risksharing and
facilitates raising capital. This feature also encourages homogeneity of interests among
the owners of the firm, who will generally tend to favor increasing its market value,
and thus avoids the political costs of disagreements over policies. The fact that the
firm is not obli gated to repurchase the shares it issued to raise capital and, as a legal
entity separate from its owners, can in principle continue in existence long after its
founders have died facilitates its taking on long-term projects and building valuable
reputations with other parties. Limited liability for shareholders frees them from having
to risk more than they have invested, and so they do not need to monitor the firm's
operations nearly as closely as would be prudent if their entire personal assets could
be claimed to settle the firm's debts. This further facilitates efficient diversification
and permits investors to leave the day-to-day operations of the firm to full-time
professionals. Further, lenders do not have to be concerned with the personal
credi tworthiness of the firm's owners because they have no claim against the owners'
other assets. The practice of having the board oversee management, with shareholders

42 This idea was put forward in Ronald Gilson and Reinier Kraakman, "Reinventing the Outside
Director, " draft, Stanford University Law School ( l 990), and was endorsed editorially by The Economist
["Redirecting Directors" (November 17, 1 990), 19-20].
 52 1
Financial
Structure,
Ownership, and
Time Warner's Rights Offering Corporate Control
and the Shareholder Movement
In June, 1 99 1 Time Warner announced a novel plan to obtain additional
equity that would permit it to retire some of its debt. Current owners of the
firm's 57. 8 million shares were to be given the right to purchase up to 34. 5
million new shares of the firm's stock. Such "rights offerings" are common,
at least in the United Kingdom, but the unorthodox element in Time
Warner's plan was that the price that investors would have to pay would
depend on the fraction of them that accepted the offer. If only 60 percent of
them accepted, then they would pay only $63 a share, about half the price
at which Time Warner was trading at the time of the announcement. On
the other hand, if there was 1 00 percent participation , the price would be
$ 1 0 5 . The offering was to have been the largest equity issue in Wall Street
history. Also unusual was that the investment banks behind it, notably Merrill
Lynch, would collect up to $ 14 5 million in fees although they would not
take the normal risk of underwriting the issue.
Stockholders reacted very negatively. They objected to not knowing
what the securities were going to cost them and to the investment banks'
fees, and they decried the plan as "coercive. " An "avalanche of protest" from
institutional and individual investors followed, and the stock price fell by
over 2 5 percent in a matter of days. Eventually, Time Warner's executives
gave in to the pressure and did not appeal an SEC rul ing against the plan.
In its place they announced a fixed-price offering at $80 that would be
underwritten, largely by Salomon . The stock price quickly fell toward this
new level .
Observers saw the episode as potentially a watershed event for the
shareholder-power movement: A company that two years before had gone to
court to prevent its owners from considering the Paramount offer had been
forced to back down by shareholder pressure. Moreover, a month later it
decided to scrap its antitakeover poison pill rather than face an embarrassing
vote at its annual meeting. Nevertheless, it retained another poison pill that
would come into effect in case of a successful proxy fight.

For more details, see Judith H. Dobrzynski, Dean Foust, and Blanca Reimer, "Time
Warner Feels the Force of Shareholder Power, " Business Week (July 29, 199 1 ), 58-59.

having only limited direct involvement in decision maki ng (for example, in voting on
sales of all or most of the firm's assets), further reduces the need for investors to expend
resources to stay well informed, as would be necessary if they were called on to play
a more active role. Yet these strengths also contribute to ownership bei ng widely
dispersed among small shareholders who may have little reason to care about anything
but the stock price and dividends and few incentives to monitor the performance of
management, to the ability of an entrenched management to pursue its own objectives
with the shareholders' money, and to boards that may identify with and defer to the
managers they are supposed to monitor and discipline.
 522
Finance: We have already seen some alternatives to the publicly held corporation at
Investments, various points throughout this text. Privately held firms (including LBO associations)
Capital Structure, and employee ownership are two, and we will examine another-cooperatives-in
and Corporate the next chapter. Here we consider two others: partnerships, and the not-for-profi t
form. Each of these forms is widespread in certain industries.

Partnerships
Control

The dominant mode of organization in professional services industries (apart from

sole proprietorship) is partnership. The biggest international accounting firms, most
major law practices, many consulting and architectural firms, the best-known medical
clinics, and some investment banks are organized as partnerships, rather than as
corporations. ·H Moreover, some of these partnerships are huge: the largest accounting
firms have more than a thousand partners, and the biggest investment banking
partnerships carry out billions of dollars in transactions. What explains this choice of
organizational form in these industries?
Partnerships involve an association of two or more persons to carry on a business,
sharing gains and losses. They are not generally legally distinct from the individuals
comprising them, and the partners jointly own the partnership. They are free
to determine their membership, internal decision procedures, and organizational
structures, and rather than having perpetual life, they exist only so long as the partners
maintain the relationship. Partners typically have the right to remove their assets from
the partnership at any time but cannot sell or otherwise transfer their ownership claims
except with permission of the other partners. If a partner withdraws or dies, the
partnership ceases to exist (unless a clause in the partnership agreement provides
otherwise, as it normally does in large partnerships). Most crucially, the partners are
each individually responsible for the entire liabilities of the partnership, and each of
the partners individually, so long as they are acting within their authority as partners
(either as set by law or defined in the partnership agreement), can take actions that
expose all the partners to unrestricted liability. This liability extends to the partners'
personal assets, including bank accounts, stocks and bonds, houses, and cars.
The risk of ownership in a corporation is divisible and transferrable; that in a
partnership is not. This limits the possibilities of diversification of risk. Thus the
partnership form is at a clear disadvantage relative to the corporation in regards to risk
sharing and the consequent ability to raise capital. It might seem that involving third
party investors as passive partners would solve this problem, but that raises the problem
of unlimited liability of partners.
UNLIM ITED LUBILITY Without limited liability, investors have to be concerned that
they would lose most or all of their wealth if the firm cannot pay its bills. This leads
them to expend extra resources on monitoring the firm and its managers and it might
well deter some from investing at all (especially investors who would own only a small
share and have little to gain from the firm's· successes). This factor constrains
partnerships from expanding the number of partners and reducing the amount im·ested
by each to permit better diversification. Similarly, it makes it expensive for partnerships
to expand into different lines of business or into new geographical markets because
such moves increase the cost of monitoring.
The potential advantage of unlimited liability is that it represents the posting of
a bond, with the individuals' private wealth serving as collateral against the debts of
the organization. Bonding can be useful in situations with either moral hazard or

H In �orne ca�es these organizations are legally "professional service corporations." For our purposes,
however, tlris form is largely indistinguishable from the partnership.
adverse selection. The cost of posting a bond is lower for businesspeople whose private Financial
information about their honesty and competence and about the quality of the firm's Structure,
earnings opportunities indicates that they arc unlikely to have to forfeit the bond. The Ownership, and
danger of having to forfeit a bond also improves incentives for working hard at Corporate Control
controlling costs and enhancing revenues. This same danger encourages the partners
to monitor one another to prevent one of them from incurring losses for the partnership
that the others will have to make good out of their own private wealth.
The predominance of the corporate form for organizing all but the smallest
businesses suggests that the relative disadvantages of the partnership form normally
outweigh the advantages. Special features of the businesses in which partnership is
common reverse the ranking in these particular cases.

AD\'A)'ffAGES OF TI IE PARTNE RSHIP FORI\I First, except perhaps for the investment
banks (which have largely switched to the corporate form in any case), the most
important specialized input in partnerships is typically the knowledge and abilities of
the workers, that is, their human capital. Human capital is not easily tradable, and
if the residual returns on that capital belong to the humans who embody it, then the
usual arguments about ownership rights suggests that the residual control should be
assigned to them, too. Moreover, the need for physical and financial capital in large
professional partnerships is often small, reducing the value of outside financing to
share the asset value risk.
Perhaps even more important is the nature of the work done by professional
partnerships, which makes effective, timely monitoring by outsiders very difficult. In
particular, outside directors and professional managers are poor monitors of the quality
and quantity of the crucial inputs. Is that doctor doing too many hysterectomies, or
is the number a result of the patient mix? Are the contracts the lawyer has drafted
air-tight? Are that architect's ideas breathtakingly original or naively impractical? Other
professionals are in a much better position to judge these matters than is anyone else,
and even they find the judgments difficult.
In addition to the problem of monitoring, there is a problem of bonding: What
assets does the firm make available as a bond to guarantee its good performance? As
we have already seen, a professional partnership requires few physical or financial
assets to do business, so the direct partnership assets against which a dissatisfied client
could make a claim are meager. However, the damages that a professional can cause
by his or her poor performance can be very large, indeed: The patient dies after minor
cosmetic surgery, the building falls down, investors are defrauded out of millions, or
an innocent person is convicted.
Professional partnerships help to solve both the monitoring problem and the
bonding problem. The partners personally are legally responsible for one another's
work, and this provides powerful incentives for mutual monitoring. Their unlimited
liability also provides the necessary large bond even though the actual capital required
to run the firm is small. Unlike shares of stock, partnership interests are not tradable
because the wrong pattern of ownership could destroy incentives and reduce the bond
and because outside investors would fear adverse selection in purchasing profit-shares
from partners who have much better information about the state of the firm.
Finally, with regard to the choice of the partnership form, some of the decision­
making costs that would otherwise be incurred are reduced by the relative homogeneity
of the partners. All are doing similar sorts of work, and they share common educational
backgrounds and a common professional orientation. This does not mean there will
not be conflicts and disputes, but these are likely to be less than would be found in
organizations where the work is more varied and the workers less homogeneous in
their outlooks and interests.
 524
Finance: Charitable Activities and Not-for- Profit Organizations
Investments, The same themes that govern the analysis of control of for-profit organizations arise
Capital Structure, in the not-for-profit sector. For example, following the earthquake in northern
and Corporate California in 1989, the Red Cross appeal for donations to its relief fund was so effective
Control
that the agency made an unannounced decision to divert some of those funds to its
other programs. When donors discovered how their contributions had been diverted,
the organization was pressured to reverse its decision. More serious scandals involving
misappropriation of funds for personal use have also occurred, even in organizations
headed by religious leaders. Jim Bakker, charismatic leader of a Christian evangelist
organization known as the PTL, Father Bruce Ritter, founder and inspirational leader
of Covenant House, a New York home for homeless and runaway teenagers, and the
Indian guru Bhagwan Shree Rajneesh, whose Oregon commune promoted love and
spiritual awareness, were all embroiled in financial scandals in the 1980s. In 1 99 1
Stanford University, a large not-for-profit institution, was raked by allegations that
U. S. federal government research funds had been used for such nonresearch purposes
as maintaining a large yacht and purchasing an Italian fruitwood commode for the
university president's house. Whenever large sums of money are being managed,
inattention to financial controls creates temptations and the risk of large financial
losses.
Not-for-profit organizations in the U nited States provide food and social services
to the poor, bring art into the community, support or undertake research into
alleviating killer diseases, produce and distribute noncommercial radio and television
programming, provide a large fraction of education at all levels from kindergarten
through graduate school, and generally provide a wide variety of goods and services
at prices that do not cover the full costs of operation. It is not unimaginable that all of
these services could be provided by profit-making organizations. Most drug research is
a for-profit activity; popular music, films, television, and videos are provided predomi­
nantly by for-profit firms; and housing for the poor is often built, with the help of
government subsidies, by private, profit-making firms. Yet food giveaways, social ser­
vices, art museums, and many other activities are most commonly conducted on a not­
for-profit basis. Why are not-for-profit organizations so important in these fields?

TI I E ROLE OF VmXNT.-\ RY DoN..\TIONS One major part of the answer has to do with
the fact that, in the United States, these activities are funded in significant part by
voluntary donations from private individuals. A chief characteristic of charitable
activities is that those who provide money to support the activities find it difficult to
ascertain if their money has been well spent. Either a large portion of the revenues
of the organization are not provided by those who receive the services (as when
donations go to help feed the poor) or the marginal impact of an individual's donations
is nearly impossible to determine (as when the money supports a public television
station). This fact means that the monitoring that occurs naturally in a normal market
(where buyers refuse to pay unless the supplier performs acceptably) cannot be nearly
as effective in disciplining the providers of charitable services. This problem is
compounded by the fact that, like most service activities, performance in the charitable
services "business" is often very difficult to measure. It is therefore hard for the donors,
and especially small donors, to check that services they are paying for are being
properly delivered.
Organizing as a not-for-profit provides some assurance to donors. ++ The key fact

+1 Of course, under current tax laws, not-for-profit 5tatus often also gi\'es the donors tax deductibility
for their donations. I lowe\'er, this ca nnot be the dominant motivation for the form, which existed e\'en
before the adoption of the i nco me tax .
 ,5 25
about a n ot-for-profit is th at no i ndividual has a legal claim to the resid ual return s: Financial
There is no owner in this sense. If revenues exceed costs, the surplus must remai n in Structure,
th e organizati on. N or does anyone have the right to se ll the assets of a not-for- pr ofit Ownership, and
an d pocket th e pr oceeds. This eliminates one maj or source of potential confl ict Corporate Control
between th ose wh o finance the organizati on (donors) and th ose wh o run it. I t explains
why organizati ons that rely on voluntary donati ons would so often be organized as
not-for-profits. 45
l\lOH.\L 1 1.\Z.\HD tN NoT-FOH-PtmFITS Adopti ng a not-for-profit form does not solve
all the problems associated wi th the revenue providers and the service receivers bei ng
different people. The danger is that, wi th the ownersh ip incentives for cost-contr ol
missing, the deci si on makers and employees of the organization may still appropriate
the donati ons for their own use, alth ough probably not by pocketi ng residual returns
di rectly. I n scandals like th ose cited earlier, th e top managers of the not-for-pr ofits
paid themselves high salaries, lent large sums at low rates of interest to fri ends or
fami ly members, enj oyed expensi ve residences and cars supplied by th e organizati on,
as well as plush offi ces and surroundings, busi ness tri ps to exotic locales, sh ort h ours,
and frequent vacati ons.
These moral hazard problems are not unique to not-for-pr ofits, of course.
H owever, the competitive pressures from the output and input markets and fr om the
market for corporate contr ol that help control them in for- profit organizati ons are
weaker in not-for-profits. Th e recipients of the goods and services are not paying the
bi lls and so cannot easily turn to more efficient providers; the donors wh o provide
revenues are unable to monitor their use cheaply; and the impossi bility of claiming
the returns that would result from increased effi ciency removes the incenti ves to take
over and impr ove the performance of inefficient not-for-pr ofits. Donors' recogni ti on
of th ese pr oblems would endanger the A ow of donati ons, and so it becomes important
(and in the interest of th ose involved in the not-for- profit) to establi sh other mech anisms
to control them.
One of these mechani sms i s th at managers of not-for-profits are held to stricter
legal standards i n their operati on of the firm than are the hired managers of pri vate,
for-pr ofit firms. A second wi dely used mechani sm is to recruit employees wh o h ave
a strong, manifested interest in advancing the obj ectives of the organizati on and so
are less likely to misuse its fu nds. Volunteers are clearly of this sort, and the extensive
practice in many charities of relying on volunteer labor and favoring former volunteers
when hiring for paid staff posi tions can be understood in this ligh t. A third is to vest
responsi bility for oversigh t of the organizati ons' operati ons in a board wh ose members
are not legally permi tted to gain personally fr om their service and wh o mi gh t be
expected to want the interests of the donors to be pursued effi ciently. Th us,
large donors are in themselves excellent board candidates, both because th ey have a
demonstrated personal interest in seei ng that the fu nds are well spent and because the
size of their donati ons indicates that they can afford to di ver t time from earni ng money
to serve in a voluntary capacity. Other good candidates are well-known public figures
wh ose names small donors will recognize, wh ose past accomplishments give indicati on
of their ability to provide effecti ve control, and wh o can be expected both to be free of
any personal interest in pr ofiti ng from the activity and to be concerned with mai ntai ning
their own personal reputati ons by preventi ng mi sbehavi or duri ng their watch.
This analysis suggests a reason wh y uni versity boards of trustees hi storically did

-1; It also explains why governments might prefer to fund provision of services through not-for-profits,
as is common i n the Netherlands, rather than profit-making fi rms. Why governments use not-for-profits
to provide the services rather than doing it directly themselves is another issue, to which we will return .
 526
Finance: not include faculty and current students among their members and even now do not
Investments, have more than token representation from these groups, despite the fact that the
Capital Structure, faculty and students are among those most affected by the board's decisions. The
and Corporate board's role is to guard the long-run viability of the university by protecting its
Control reputation and revenue sources. To do so, it must protect the outside contributors
(whether private donors or taxpayers) from having their money used to promote private
benefits rather than the public weal that they presumably wanted to support. The
parties who have presented the greatest threat of misappropriating these contributions
have historically been the faculty and students because until quite recently they were
the only signif icant interest groups within the institution. An effective board must not
be led or too greatly influenced by those who are to be monitored, and so student
and faculty participation has been severely limited .
In recent years, however, a number of universities have developed large cadres
of professional managers in addition to the academic support staff. This former group
in particular represents an independent threat to misappropriate contributed funds.
They too do not typically have board representation, but because most of the
information that the board must use for its decisions is developed and presented by
the professional administrators, there is the real possibility of their influencing board
decisions to their private benefit. Of course, to the extent that the board must rely on
some group within the university to provide this information, the possibility of attempts
at influence is unavoidable, although collecting information from different interested
groups might allow the board to sort out the facts better.

Go rnRN�IENT Fl !NDINC In most countries, private donations (except to religious

organizations) are rare, and the government funds social, charitable and cultural
activities. At least in some of these situations, however, the government-fi nanced
services are provided through not-for-profits. This is especially common in the
Netherlands. Even in the United States, many not-for-profits are financed jointly by
government funds and private donations. For example, a large amount of basic
research is paid for by the federal government through grants and contracts to private,
not-for-profit universities whose employees produce the actual research. The arguments
already given explain why governments might prefer to supply these services through
not-for-prof its rather than prof it-oriented firms, but why might they choose to have
activities they finance be provided through not-for-profits rather than by government
organizations?
Several reasons for this pattern exist, although none is fully satisfactory. One is
that the government may be able to avoid various constraints by using this mechanism.
For example, government wages may be higher and work rules more restrictive than
in the private sector, and in this case a not-for-profit may enjoy lower costs than
would the government. Similarly, use of the not-for-profits may allow paying higher
wages than are politically acceptable for public servants but which are necessary to
attract the sort of people needed. What is unsatisfying about this is that it leaves the
existence of the crucial constraints unexplained.
A second possibility is that this pattern helps insulate the provision of the services
from direct political interference; for example, the government cannot simply dismiss
the directors of an independent foundation. To the extent that this interference is
aimed at distributional considerations (serving the elected officials' supporters rather
than some other group) but has efficiency-reducing side-effects, insulation enhances
efficiency. The problem here is to explain why influence costs should be less when
a client organization is used rather than a branch of government. For example, an
effective civil service organization might well accomplish the same objective.
A final possibility is that this pattern accommodates differences in tastes and
 527
encourages inn ovation better than a government-ru n pr ogram, wh ich will tend to
produce uniformity. This is of course j ust a version of the sta ndard argu ments fo r
Fi nancial
Structure,
federalism: The different jurisdi ctions can pattern what they do to local situations, Ownership, and
and successes (or failures) in one ca n be a lesson to others. This too has a ring of Corporate Control
truth. For example, i n the province of Ontari o taxpayers can ind icate wh ether they
want their sch ool taxes to go to support the nondenominational pu blic sch ools or
instead th e sch ools run by th e Roma n Cath olic ch urch: The point is to provide a
level of funding while still permitting some responsiveness to differences in preferences.
H owever, again the crucial question arises: Why can't government provide th e desired
diversity wh ile producing the services itself?
This last question is closely related to the question of why firms ever rely on
independ ent suppliers to provide them with inputs. Th ere is not yet a consensus
answer to that question, but some tentative a nswers are supplied in the next chapter.
 528
Finance:

8 UJ\ IJ\ IAR Y

Investments,
Capital Structure,
and Corporate
Control In this chapter, we enrich the classical theory of finance by adding three ideas. First,
managers, lenders, and shareholders may all have different interests, different from
those of the firm's lenders or shareholders, and financial arrangements may affect how
different those interests are and what decisions management will make. Second,
managers may be better informed than investors about the firm's prospects, so the
financial decisions they make may affect investors' beliefs and therefore the price of
the firm's shares. Third, financial securities are not just claims to part of a firm's net
receipts; they also give the security holder certain rights. Just as in the theory of
ownership, a careful matching of rights and returns can create incentives that increase
total value. The chapter closes by considering the control problems in partnerships
and not-for-profits, where some of the same issues of motivation and control arise.
The financing patterns of firms vary significantly among developed countries,
with more equity financing having been used traditionally in the United States,
Canada, and the United Kingdom than in many of the other industrial countries.
The huge wave of takeovers in these countries in the 1 980s often involved refinancing
firms using large amounts of debt and concentrating equity ownership in the hands
of fewer large investors, who could exercise more effective control. Because these
transactions increase the firm's ratio of assets to equity or leverage, these deals are
called leveraged buyouts or LBOs. Especially noteworthy were the management
buyouts (MBOs), in which ownership was concentrated in the hands of the firm's
managers and, often, the shares of the firm ceased to be publicly traded, and hostile
takeovers, in which indivi duals and firms attempted to acquire control of companies
against the wishes of the firm's management. When the target firms were conglomerates
with many different busi nesses, the changes in control were frequently followed by
bust-ups, in which the various businesses were separated and sold to other investors,
leading to a signi ficant deglomeration of the industries in the affected countries. All
this activity casts doubt on the conclusion of the Modigliani-Miller analysis and leads
us to ask anew: How can capital structure matter?
Given the relative payment priorities, increasing the riskiness of cash flows tends
to transfer value away from the holders of debt to the holders of equity. This occurs
because when favorable outcomes occur, the equity holders enjoy all the extra returns,
whereas debtholders suffer part of the loss when the most unfavorable outcomes occur.
Consequently, when equity holders control management, high levels of debt encourage
excessive risk taking.
A second kind of cost arises when high levels of debt combine with a run of
operating losses. This can lead to debt overhang, in which a firm with profitable
investments is unable to finance them because . the suppliers of new financing fear
that part of their investment will be diverted to satisfy the demands of existing lenders
and debtholders. Debt overhang thus leads to underinvestment. The problem of debt
overhang can sometimes be overcome by debt renegotiation, in which existing lenders
agree to reduce their claims in order to encourage others to supply new capital, but
renegotiations are difficult when there are many debtholders who must agree to the
deal. Moreover, even the option of renegotiation may damage incentives because it
protects equity holders and managers from some of the losses from risky investments.
When renegotiations arc unsuccessful, a third cost of debt is that default may
lead to ba nkruptcy, even if the current operations of the firm arc fundamentally
sound. Even if bankruptcy leads only to a reorganization of the company's affairs,
rather tha n to a n actual liqu idation of its business, the bankruptcy event can be
 529
profoundly disruptive to the firm's operations . Suppliers of goods, fearing that they Financial
may not be repaid, may no longer be willing to ship supplies on credit. Suppliers of Structure,
capital engage in legal battles among themselves and between them and management, Ownership, and
which divert the attention of management from the pressing need to reinvigorate the Corporate Control
firm. Partly to avoid these costs, suppliers of capital may engage in informal workouts,
agreements made without the assistance of the court to avoid the costs of bankruptcy.
Against these three disadvantages, debt has the important advantage that it
compels management to return earnings to investors, rather than retaining them for
its own purposes within the firm. This prevents managers from using the firm 's free
cash /low-the amount of cash generated by the firm in excess of what can be profitably
reinvested, to enlarge their units (unprofitably), decorate their offices, and so on.
Moreover, given that managers have limited resources, high levels of debt make it
possible to concentrate a higher fraction of the equity ownership in the hands of the
management than would otherwise be possible. This leads to better incentives for
managers to ·increase value in the event that the firm does not default on its loans.
The LBO association is a form of organization that attempts to enjoy these
advantages of high levels of debt while avoiding its worst disadvantages. Using capital
provided by outside investors, the association sponsors LBOs to improve management
incentives, concentrates financial control in a few hands, actively monitors the firm's
management, and stands prepared to replace its m anagers or to inject new financing
as circumstances demand.
The LBO association illustrates the benefits of concentrated ownership. There
is considerable concentration of ownership in many companies, even without this
form of organization. Individuals and funds that supply venture capital to new
companies perform a similar function, monitoring the performance of management
during the start-up of the new venture. In many countries, where relatively high
leverage is the norm, the principal lender to a company serves a similar monitoring
function, often having a representative on the company's board of directors . Bondhold­
ers and long-term lenders attempt to protect their investments by other means, such
as by requiring collateral for loans or writing loan covenants that restrict the
management's ability to sell assets or encumber the security for existing loans.
These considerations indicate how financial structure can affect the parties'
incentives in a way that alters the total value of the firm .
The second class of theories we considered are the signaling theories, according
to which financial decisions can affect what investors believe about the firm's prospects.
One of the most robust empirical conclusions is that attempts by a firm to refinance
in the direction of increased equity reduces the total market value of the firm, whereas
refinancing in the other direction increases it. Similarly, reductions in dividends lead
to a decreased value, whereas increased dividends have the opposite effect. According
to the signaling theories, these simply represent optimal responses by investors who
are interpreting the actions of the firm's managers, who are better informed than the
investors . Managers with an ownership stake in the firm, acting in their own interests,
will want to sell equity when they think it is overvalued and to buy equity when they
think it is undervalued. They wi11 want to replace debt by equity when they expect
the firm to have too little cash to meet its future obligations. Increases in dividends
and stock repurchases indicate the managers' confidence that the firm's future cash
Aow will be adequate to meet its needs. Reacting to these indicators, it is rational for
investors to bid up the price of the firm's share when the firm announces its intention
to repurchase its stock, whether the repurchases are to be financed by debt or retained
earnings, and when the firm announces a dividend increase, and to do the opposite
when the firm announces a new stock issue or a dividend reduction.
Financial structure can also affect value by the way it aHocates rights. Stockholders
 530
Finance: have the right to elect corporate directors and to vote on certain major changes in the
ln\'estments, organization's charter, but not to interfere with the discretion of the board of directors
Capital Structure, in ordinary business decisions. These voting rights are useful to those who wish to
and Corporate effect a change in control of the company. They may engage in a proxy figh t, soliciting
Control a proxy to vote on behalf of other shareholders to place their own representatives on
the board of directors. They may also purchase shares to be used together with any
proxies and the shares of their allies to elect directors.
In the United States in the 1 980s these takeovers led to huge increases in the
market values of the target firms, but essentially no change in the value of the
acquiring firms. On average the takeover prem ium-the amount by which the takeover
price exceeds the earlier market price-was 30 to 50 percent, but it sometimes
exceeded 1 00 percent. It appears that most of this value was created by selling or
reorganizing divisions of the target firm to create more focused firms operating fewer
businesses. Recognizing the value of these reorganizations, many other companies
undertook volu n tary restructurings, in which the corporation itself became the lender
to its divisions, which could be highly leveraged and in which the division managers
could own substantial shares.
Management devised a number of means to defend themselves against the
corporate raiders. Golden parachutes, which provided generous payments to managers
displaced in control changes, were justified as means to reduce managerial resistance
to takeover attempts and protect their investments in firm-specific capital. Green mail,
in which a firm repurchases a raider's shares at a special price, fends off the raider
with the shareholder's money, leading to no improvements in operational efficiency.
This tactic has been widely condemned, but commentators disagree about whether it
is the raider or the management that is the proper object of the condemnation.
In 1990 takeover activity was substantially diminished as the market for new
bonds below investment grade, known as junk bonds, had virtually disappeared. Some
large public investors began to press companies to adopt the reforms that had increased
value during the takeover wave and to eliminate some of the takeover defenses, to
provide discipline over themselves, and some companies complied with these requests.
In the first half of 1 99 1 takeover activity began to increase again, though on a much
smaller scale than found during the peak of takeover activity in 1 986 through 1 988.
The issue of who controls a large organization and monitors its activities is not
limited to public corporations. Groups of professionals, such as doctors, lawyers,
accountants, architects, and others are usually organized as partnerships. The major
assets of these businesses are the skills and reputations of the partners, which are
human-capital assets, and a matching of residual returns and residual control would
suggest the desirability of a partnership form, in which the professionals own the firm.
Moreover, because professional work is so hard to evaluate, outside monitoring is
likely to be ineffective compared to a system in which the professionals monitor one
another. The heads of these organizations are virtually ahvays other professionals
because nobody else would be qualified to evaluate and criticize the professional's
work.
Charitable and not-for-profit organizations often handle significant sums of
money. There have been many scandals in which donations were used to finance
fancy cars and residences, pay high salaries, make loans to directors and officers, and
so on. In addition, the priorities of those who run the organizations may not coincide
\\'ith those of the donors, so moral hazard remains an issue. Organizations that receive
voluntary donations are usually organized as not-for-profits to counter the ob\'ious
problem that an owner might neglect providing services to increase profits. This is
most problematic when there arc many small donors or when services are especially
hard to measure. Boards of directors consisting of volunteers, who arc also often
principal donors, helps to resol ve the monitoring problem and to ensure that directors Financial
do not have financial interests that conflict with the organization's goal s. Stru cture,
O ne of the puzzles of not-for-profit organization arises when fu nding for the Ownership, a nd
organization comes from a government. The question is: Why doesn't the government Corporate Con trol
pro vide the services itself, rather than relying on an independent organization?
Although a number of answers suggest themselves, none treats the fundamental issue
of why the government can't organize within itself a unit that duplicates, in all
essential respects, the attributes of a not-for-profit. This is the same issue that arises
in connection with vertical integration in the private sector, and some answers are
offered in the next chapter.

• BIBLIOGRAPHIC NOTES
Michael J ensen and William Meckling published the first systematic economic
analysis of how inadequate managerial equity holdings might reduce managerial
incentives for effort and innovativeness and how debt might encourage excessive
risk-taking. S tewart Myers first analyzed the debt overhang problem, suggesting
that excessive debt can lead a firm to bypass profitable investments. Michael
Jensen devel oped the free cash flow hypothesis and accentuated the role of the
debt and LBO associations in dealing with free cash flow problems. Andrei
S hleifer and Robert Vishny, H arold Demsetz and Kenneth Lehn, and Richard
Zeckhauser and J ohn Pound all present both theory and evidence for the
proposition that large shareholders add value by monitoring their firms.
Signaling theories in financial economics were pioneered in 1977 by H ayne
Leland and David Pyle (who hypothesized that high levels of share ownership by
management signaled high value per share) and S tephen Ross (who argued that
a high debt-equity ratio signaled val uable equity shares). Stewart Myers and
Nicholas Maj l uf further advanced this theory. S udipto Bhattacharya developed
the first model s of dividend signal s, which were later refined by Kose J ohn and
J oseph Wil li ams. Frank Easterbrook championed the role of dividends in providing
incentives. The danger of strategic asset destruction and its implications for
financial structure were pointed out by Yaacov B ergman and Jeffrey Cal len.
Milton Harris and Artur Raviv provide an excellent survey of research on financial
structure emphasizing incentive considerations.
Henry Manne first accentuated the importance of the market for corporate
control in increasing effi ciency. Jensen has been a prominent proponent of this
interpretation of takeovers. Richard Roll proposed the "hubris theory" to explain
takeover premia, and S hl eifer and Lawrence S ummers accentuated breach of
trust and transfers. The formal analysis of coercive offers was first developed by
Lucian Bebchuck, whil e S anford Grossman and Ol iver H art pointed out the free­
rider probl em in tender offers. The Symposium on Takeovers in the fournal of
Economic Perspectives ( wi th an introduction by H al Varian and papers by Shl eifer
and Vishny; Jensen; Gregg Jarrell, James Brickl ey, and Jeffry Netter; and F. M.
Scherer) provides a very accessible introduction to economic research on this
question. A textbook treatment of these matters is given by J . Fred Weston,
Kwang S. Chung, and S usan H oag.
The choice of organizational form has been analyzed by Eugene Fama and
Michael Jensen and by Henry H ansmann, among others. A survey of the
economics of the not-for-profit firm is given by Estell e James and Susan Rose­
Ackerman.
 532
Finance: • REFERE�CES
Investments,
Capital Structure, Bebchuck, L. A. "The Pressure to Tender: An Analysis and a Proposed Remedy,"
and Corporate in J. Coffee, J r. , L. Lowenstein, and S. Rose-Ackerman, eds. , Knights, Raiders
Control and Targets (New York: Oxford University Press, 1988), 3 71-97.
Bergman, Y. , and J . Callen. "Opportunistic U nderinvestment in Debt Renegotia­
tion and Capital Structure," Working Paper 90-24, Brown U niversity Department
of Economics (1990).
Bhattacharya, S. "Imperfect Information, Dividend Pol icy, and 'The Bird in the
H and' Fallacy," Bell Toumal of Economics, 10 (Spring 1979), 259-70.
Demsetz, H., and K. Lehn. "The Structure of Corporate Ownership: Causes and
Consequences," Toumal of Political Economy, 93 (December 198 5), 115 5-77.
Easterbrook, F. "Two Agency-Cost Explanations of Dividends," American Eco­
nomic Review, 74 ( 1 984), 6 50- 59.
Fama, E. , and M . Jensen. "Agency Pro blems and Residual Claims," Toumal of
Law and Economics, 26 (198 3), 327-49.
Fama, E. , and M . Jensen. "Organizational Forms and Investment Decisions,"
Touma/ of Financial Economics, 14 (198 5), 101-19.
Grossman, S . , and 0. Hart. "Takeover Bids, the Free-Rider Pro blem, and the
Theory of the Corporation," Bell Touma/ of Economics, 11 (S pring 1980), 42-
64.
Hansmann, H . "The Ownership of the Firm," Touma/ of Law, Economics, and
Organization, 4 (Fall 1988), 267-304.
Hansmann, H . "When Does Worker Ownership Work? ESOPs, Law Firms,
Codetermination, and Economic Democracy," Yale Law Touma[, 99 (J une 1990),
17 51-1816.
H arris, M. , and A. Raviv. "The Theory of Capital Structure," Toumal of Finance,
46 (March 199 1 ), 297-3 5 5 .
i
James, E. , and S. Rose-Ackerman. The Nonproft Enterprise in a Market Economy
(Chur, Switzerland: Harwood Academic Publishers, 1986).
Jarrell, C. , J. Brickley, and J. Netter. "The Market for Corporate Control: The
Empirical Evidence Since 1980," Touma/ of Economic Perspectives, 2 (Winter
1988), 49-68.
Jensen, M. "The Eclipse of the Public Corporation," Harvard Business Review
(September-October 1 989), 61-74.
Jensen, M. "Agency Costs of Free Cash Flow, Corporate Finance and Takeovers,"
American Economic Review, Papers and Proceedings, 76 (May 1986), 323-29.
Jensen, M. "Takeovers: Folklore and S cience," Harvard Business Review,
(November-December 1984), 109-121.
Jensen, M. "Takeovers: Their Causes and Consequences," Touma/ of Economic
Perspectives, 2 (Winter 1988), 21-48 .
Jensen, M. and W. Meckling. "Theory of the Firm: Managerial Behavior, Agency
Costs, and Capital Structure," Touma/ of Financial Economics, 3 (1976), 3 0 5-
60.
Joh n, K. , and J. Williams. "Dividends, Dilution, and Taxes: A S ignal ing
Eq uil ibrium,'' Touma/ of Finance, 40 (198 5), 10 5 3-70.
Leland, H . , and D. Pyle. "Informat ion Asymmetries, Financial Structure, and
Financial I ntermediatio n," Tournal of Finance, 32 (1977), 371-88.
 Manne, H . "Mergers and the Market for Corporate Control, " /ournal of Political Fi nancial
Economy, 73 ( 1 965), 1 1 0-20. Structure,
Myers, S. "The Determinants of Corporate Borrowing, " /ournal of Financial Ownersh ip, and
Econom ics, 5 (1 977), 147-7 5 . Corporate Control
Myers, S. , and N. Majluf. "Corporate Financing and Investment Decisions
When Firms Have Information that Investors Do Not Have, " /ournal of Financial
Economics, 1 3 (June 1 984), 1 87-22 1 .
Roll, R . "The Hubris Hypothesis of Corporate Takeovers, " /ournal of Business,
59 (April 1986), 197-2 1 6.
Ross, S. "The Determination of Financial Structure: The Incentive Signalling
Approach , " Bell /ournal of Economics, 8 (Spring 1977), 2 3-40.
Scherer, F. "Corporate Takeovers: The Efficiency Arguments, " /ournal of
Economic Perspectives, 2 (Winter 1 988), 69-82.
Shleifer, A. and L. Summers, "Breach of Trust in Hostile Takeovers, " Corporate
Takeovers: Causes and Consequences, A. Auerbach, ed . (Chicago: Un iversity of
Chicago Press, 1 988).
Shleifer, A. , and R. Vishny. "Value Maximization and the Acquisition Process , "
/ournal of Econom ic Perspectives, 2 (Winter 1988), 7-20.
Shleifer, A. , and R. Vishny. "Large Shareholders and Corporate Control , " /ournal
of Political Economy, 94 (June 1986), 46 1-88.
Va rian, H. "Symposium on Takeovers, " /ournal of Econom ic Perspectives, 2
(Winter 1 988), 3-6.
Weston, J. F. , K. Chung, and S. Hoag. Mergers, Restructuring, and Corporate
Control (Englewood Cliffs, NJ: Prentice Hall, 1990).
Zeckhauser, R. , and J. Pound. "Are Large Shareholders Effective Monitors? An
Investigation of Share Ownership and Corporate Performance, " in R. G. Hubbard,
ed. Asym metric Information, Corporate Finance, and Investment, (Chicago:
University of Chicago Press, 1990), 1 49-80.

EXERCISES

Food for Thought

I . A stock mutual fund is a company that invests in shares of the stock of

other compan ies. These funds exist to make it possible for small investors to pool their
money in order to hold a widely diversified portfolio of investments. In an open-end
fund, shareholders can sell their shares and demand to be paid their pro rata share of
the value of the fund's shareholdings. In a closed-end fund, shareholders can obtain
cash for their shares only by selling them to others. It is common for closed-end funds
to sell for prices lower than the actual value of company's holdings. How can one
account for that? Under what circumstances would a closed-end fund be preferred to
an open-end fund?
2. Firms frequently establish long-term relationships with a single bank. If the
firm is large, the single bank may then enlist other banks to meet any especially large
loan needs for their customer. What are the advantages of such an arrangement? For
example, why do individuals not make small loans to firms di rectly, rather than
depositing their money in a bank and having the bank lend the money? Why does
the firm not deal with the individual banks separately? Would it be better if, in the
 ,334
Finance: United States, there were fewer, larger banks to make loans to large companies rather
Investments, than the multitude of small banks that actually exist?
Capital Structure, 3. What inferences might investors draw when a company announces an
and Corporate increase in its quarterly dividend? What would you expect to happen to share prices
Control following such an event? What are the associated problems of commitment? How
might those problems be resolved?
4. In 1 990, American Telephone and Telegraph, the giant telecommunications
company with sales of $36 billion a year from long-distance telephone service (60
percent) and sales of telephone equipment, launched an attempted takeover of NCR
Corporation, a major U . S. computer manufacturer. AT&T had long tried to establish
itself in the computer business but had been unable to produce and market successfully
competitive machines. It was estimated to have spent $2 billion on this effort in six
years and still to be losing $200 million a year on its computer business. It had also
attempted to buy its way into the market by acquiring equity stakes in other computer
firms, but had failed. AT&T offered to acquire NCR (formerly called the National
Cash Register Company) for $90 a share. This was about 1 5-times NCR's earnings
per share, which were falling, and it put a value of $6. 1 billion on the target firm.
Before the AT&T offer, NCR shares sold at $48.
NCR's board and managers rejected the offer . AT&T persisted, making the offer
hostile, and NCR resisted strenuously, adopting a tougher poison pill, establishing an
ESOP, and in\'oking a state law that forbid buying more than 1 0 percent of a firm
without its directors' approval. The senior managers, whom AT&T wanted to retain,
also threatened to quit if AT&T gained control. AT&T then launched a proxy fight.
In response, NCR changed its by-laws to make it extremely difficult to remove a
majority of the board in such a fight. All this was done despite strong shareholder
pressure to accept the offer, which AT&T had increased. The NCR chairman claimed
he was willing to sell the firm, but demanded $ 1 2 5 a share. E\'entually, after almost
six months from the time AT&T made its initial offer to NCR, NCR agreed to an
offer of $ 1 1 0, with the managers of NCR taking over AT&Ts computer operations.
How would you judge if NCR was a worthwhile investment for AT&T? How
would you judge whether NCR's leaders were pursuing their own interests or those
of their stockholders? What further information would you want?
5. In 1 99 1 , General Motors announced its i ntention to raise additional equity
capital by issuing a new form of hybrid securities it called "preference shares" or
"perks. " The preference shares were to be issued at the same price as GM's common
stock, but were to offer an expected dividend yield of twice what the firm's common
stock had been returning. Although the perks could fall in value, just like the common
stock, their dividends would have to be paid before those on the common stock. They
would automatically convert after three years to common stock on a 1 to 1 basis .
l\ leanwhile, Gl\ l was free to redeem them at any time at a predetermined price. This
limited their possible appreciation. Why would anyone buy these rather than regular
stock? Why would GM want to sell them?
6. In recent years, the large accounting firms have expanded into the field of
management consulting, bring non-accountants into the firms for this purpose. l\ lany
of these consulting businesses have grown \'Cry fast and are now apparently \'ery
profitable. What problems would you expect this to create in these partnerships?
7. The venture capital industry is predominantly organized through limited
partnerships. Venture capitalists establish limited partnerships to invest in new
businesses, with themselves as the general partners . Investors supply the bulk of the
money for investment by buying limited partnerships in the funds. The general
partners put up about one percent of the funding, but typically receive 20 percent of
the disbursements. They choose the actual investments, wh ile the management
com panies that they own make recommendati ons to the partn ersh ip on possible Financial
i nvestmen ts an d charge the partnersh i p for these services. H ow do you account for Structure,
th ese practices? What possi ble problems do you see? Ownership, and
8. I n the U nited States there are both for-profit and not-for-profit hospitals. Corporate Control
What di fferences in policies and practices would you expect to see between th e two
forms?
9. I n the U nited States i n recent years there has been a great increase i n the
number of malpractice suits agai nst professi onals (doctors, lawyers, etc.) and i n th e
size of the awards. M alpractice i nsurance, which pays legal fees and the costs of any
awards to plai ntiffs, dulls the i ncentives of unlimi ted liabili ty, but may be " necessary"
to get people i nto these i ndustries. What form should it take?

Quanti tati ve Exerci se I

1 . S uppose firms differ i n thei r possi ble cash flows. Hi gh-cash-flow firms have
cash flows which might take on any value i n the i nterval [O, H] wi th equal probabi lity,
while the cash flows of low-cash-flow fi rms are uniformly di stri buted over the i nterval
[O, L], where L < H. Each manager knows what kind of firm he or she has, but
i nvestors do not have thi s i nformati on: All they know is that some fracti on of the firms
have high cash flow and the rest have low. They do, however, observe the amount
D of debt that the fi rm issues.
Suppose each manager chooses a value of D and is paid in such a way that he
or she seeks to maximize a weighted average of the current value of the firm Y0 and
its expected value i n the next period (at which poi nt the cash flow is realized and
becomes publi cl y known) less a bankruptcy penalty, P, that is i ncurred by the manager
(not the firm's owners) if earni ngs turn out to be less than D. Thi s occurs with
probability D/H for high-cash- flow firms and D/L for low ones. The expected value
of the firm next peri od is j ust its expected cash flow, L/2 or H/2. The current value
may depend on D through i nvestors' i nferences. Thus a manager of a fi rm of type t,
t = H or L, who selects a debt of D receives a payoff (proporti onal to)
(1 - -y)Y0(D) + -y(t/2 - PD/t),
where -y is the weight.
U nder the assumpti on that managers i n the two types of firms pick different
levels of D, i nvestors will infer firm's values from the choi ces. Thus, if a firm has
debt D L it wi ll be thought to be worth L/2 i n the first peri od, and simi larly for D H ,
no matter what its true type. Thus, Y0 is determi ned: Y0(D H ) = H/2, and
Y0(DL) = L/2. H owever, the firm's value next peri od will be determi ned by its actual
type.
Show (usi ng the appropriate i ncentive constrai nts) that DH must exceed D L :
M ore profi table firms issue more debt. Show that if D L is set at zero and D H is set as
low as possible to meet the i ncentive constrai nts, then D H decreases as P i ncreases:
Higher bankruptcy costs decrease the amount of debt and the probability of bankruptcy.
 Part

VII
THE DESIGN AND DYNAMICS OF
ORGANIZATIONS

16
T HE BOUNDARIES AND
STRUCTURE OF THE FIRM

17
T HE EVOLUTION OF BUSINESS
AND ECONOMIC SYSTEMS
 16
THE BOUNDARIES AND
STRUCTURE OF THE FIRM

I 'I I
i
I

he make-or-buy decision tha t occasioned so m uch deba te in mass production

f rms struck Ohno and others at Toyota as la rgely irreleva nt, as they bega n to
consider obta ining componen ts for cars and trucks. The rea l question was how the
assembler and supplier could work together smoothly to reduce costs and improve
quality, wha tever formal, legal relationship they m ight have.
lames Womack, Da niel l ones and Dan iel Roos 1

This chapter is concerned with the positive ("what is") and normative ("what
should be") analysis of the design of the firm. How do firms determine their vertical
boundaries and arrange their relationships with suppliers and customers? When should
a firm use the market to acquire the goods and services it n eeds? When should it
undertake to provide these for itself? And when should it seek to develop some hybrid
organizational arrangements to conduct a transaction? How do firms decide on their
horizontal scope-what activities to undertake and what businesses to be in? How
should they structure their internal mechanisms for coordination and motivation,
allocating decision-making power and responsibility for different activities, and what
is the relationship between scope and the firm's internal structure? How have the
choices that firms actually make on these dimensions changed, and what has motivated
these adaptations? What principles ought to guide these decisions?
These questions are centrally important to the economics and management of
organizations, and entrepreneurs and managers have been exceptionally innovative
over the last century in devising new answers. Systematic understanding of the
strengths and weaknesses of the alternative organizational forms has not always kept

1 The Machine That Changed the World (New York: Rawson Associates, 1 990 ) , p. 58.
up, altho ugh study of these issues is one of the most active and promising areas of The Boundaries
research in economics and mana gement. and Structure of
In our analysis, we con tinue to regard organizational forms as being chosen by the Firm
people who enter in to relationsh ips that arc efficient for themselves. The first ste p in
determ ining what the efficien t forms of organiza tion arc must be to identify the feasible
organizationa l alternatives. These have changed over time, but we largely limit
attention here to modern practices. The alternati ves arc m ultiface ted, and part of the
pro blem is to review the com plemen tarities among the parts; that is, how the pieces
of the organ ization fit together in to a coherent design. The second step is to identify
the qualita tive costs and benefits of alterna ti ve forms. Wha t is gained and what
resources are used by each form of organ iza tion? What causes these cos ts and bene fits
to vary over time? Finally, differences in costs across segmen ts of the economy, across
countries, and over time are used to analyze differences in the choices that have been
made.

TI I E C t IANCINC NATURE OF THE FI RM

An y serious a ttempt to explain actual business organiza tion must gra pple with the
historical fact that the na ture of the firm has changed tremendously over the last
cen tury and a half.

Emergence of the Industrial Enterp rise

Before 18 50 hierarchical structures were virtually nonexistent outside the episcopal
churches and the m ilitary. The H udson's Bay Company had a prim itive managerial
hierarchy, with the hea ds of each trading post directing the local em ployees wh ile
ta king orders from London , and the East India Com pany had a well-developed
bureaucracy. Both com panies enjoyed the quiet life of the monopolist, however, and
when the HBC was faced with com petition, it came close to failing. Manufacturing
was generally con ducted on a small scale. One person, or perhaps a small group of
people, hired all the em ployees and directed all the acti vities of the bus iness, and
thus each business was li m ited to a scale that an entrepreneur could personally
supervise. The smal l scale of operations generally fit the markets of the time. Bankers
and traders had operated over long dis tances for centuries , and the industrial revol ution
in Grea t Britain had seen the development of some na tional and even in ternational
markets in industrial products, but most markets remained local, and business was
adapted to this pattern.
Three crucial technological developmen ts changed all this. The steamship
permitted scheduled, predictable ocean shipping. The railroad allowed speedy transpor­
tation of people, goods, and informa tion over continental distances. The telegra ph
initiated near-instantaneous communica tion. Together, these developmen ts made it
possible to conceive of doing business on a scale never before possible. By the end of
the nineteen th cen tury, larger en terprises had emerged to sell their produc ts on a
nationwide scale and even in in ternational markets. No longer limited by the size of
their local markets, firms began to use new methods of production , design ing
specialized tools and in troducing mass-production methods. These developments
allowed firms to produce larger quantities of higher-q uality products a t lower cos t.
These strategic changes were accom panied by equally revolutionary changes in
organization . As we sa w in Chapters 3 and 4, the price system , which can be so
effective for coordinatin g trade in standardized commodities, is much less useful for
coordinating activities tha t exploit economies of scale. The growing scope and
com plexi ty of business overwhelmed the ca pacity of owners to manage their opera tions,
 540
The Design and requiring them to hire supervisors and middle managers who would oversee parts of
D y namics of the business, increasing the span of the firm's activities and the number of people
Organizations who could be monitored and motivated.
With the owner no longer personally and directly overseeing all the crucial
activities, it became necessary to create information and reporting systems to guide,
control, and evaluate the managers who carried out these tasks. Traditional financial
accounting systems were refined to provide investors and lenders with information to
help them assess the firm's creditworthiness and prospects and, eventually, the
performance of its hired managers. In addition, cost accounting systems were designed
to provide information about the costs of making specific products in particular
factories. The information generated proved valuable in setting prices, deciding which
products to make, identifying sources of any cost increases over time, and pinpointing
the activities where cost reductions would be most valuable.
Along with the new information systems, new financing arrangements emerged.
In the United States, the demands of railroads and other large enterprises exceeded
the financial capabilities of local banks, leading to the development of bank syndicates
and bond markets and to the great expansion of stock markets and contributing to the
emergence of an insurance industry. Other countries resolved the large-scale financing
problem in other ways. In Canada, Germany, and Japan, for example, giant banks
played a dominant role in financing growing firms.
As industrialization proceeded, the profitable monopolies of local producers
were challenged by competition from national and international firms. Businesses
responded in part by being innovative in their collusive organizations. Trusts and
cartels of national producers conspired to maintain high prices. In the United States,
most major manufactured goods were brought under price-fixing arrangements. Then,
when enforcement of the Sherman Antitrust Act ( 1 890) made cartels unattractive, a
wave of mergers created many of the giant firms that dominated U. S. industry for the
next century. U S Steel, for example, was formed in 1 90 1 through the amalgamation
of 1 1 firms in what remains (in inflation-adjusted terms) the second largest merger in
U. S. history. Such familiar names as General Electric and Eastman Kodak also
appeared in this period. Some of these new organizations were centrally managed and
organized on functional lines, but others were simply holding companies, giving
unified ownership with little central direction. The business institutions of the early
twentieth century were radically different from those in place just 50 years earlier.

The Development of the Multidivisional Form

Change in the organization of firms has continued unabated throughout the twentieth
century. An especially important development, however, occurred in the years
immediately following World War I, when the multidivisional form of organization
was introduced. In this form managers who control individual divisions and are held
responsible for their performance report to higher-level managers, who evaluate them,
coordinate their activities, and plan the firm's strategy. This multidivisional form
became the template for big business in North America and much of the world for
most of the twentieth century. Four U. S. firms-General Motors, Du Pont, Sears
Roebuck, and Standard Oil c,f New Jersey-were the innovators in developing the
multidivisional form. Given the importance of this innovation, it is worth considering
its history in greater detail. 2
The history of General Motors in the early 1 920s with which this book began

2 The story of the emergence of the divisional ized firm is told hy Alfred Chandler, J r. , Strategy and
Structure: Chapters i11 the / listory of the America11 Industrial En terprise ( Cambridge, !\IA: l\tlT Press,
I %2). We discussed aspects of this story briefl y in Chapter 3 .
 54 1
is a story of the sim ultaneous creation of a new strategy and a new business structure. The Boundaries
The strategy en tailed a diversification in CM's existing product line, introducing high­ and Structure of
price cars with many standard features and low-price ones with few featmcs. Th e the Firm
structure was the creation of separate divisions matched with an expanded central
office that would monitor divisional performance and coordi nate business strategics.
Du Pont, which in the nineteenth and early twentieth centuries had been an
explosives manufacturer, had expanded to meet the surge in demand during World
War I. Its capacity to make smokeless powders to propel shells exploded from 8. 4
mi llion pounds per year in 1 9 1 4 to 45 5 million pounds in 1 9 1 7. Looking for ways to
use its excess manufacturing capabilities after the war, Du Pont diversified into the
manufacture of chemical fertilizers, which it hoped would exploit its special abilities
in nitrogen-based chemistry. However, the production and marketi ng of fertilizers to
farmers was a very different business from making and selling smokeless powder to
gover�ments at war. Unlike governments purchasing war supplies, farmers had to be
educated individually about Du Pont's products and convinced to buy them. Different
farmers' needs varied with the crops they grew and the soil and weather conditions
they faced. New distribution channels were also needed. Product developers and
salespeople thus needed to be close to their customers and well informed about the
products and about farmers' needs and concerns, and the people who supervised and
managed them needed similar knowledge. To accommodate the demands of the
fertilizer business, Du Pont eventually established a separate division, with its own
independent management running its own manufacturing and sales departments,
without detailed direction from the central office.
Before World War I, Sears Roebuck had been a highly centralized mail-order
catalog business dealing in "hard goods, " such as tools. After the war, under the
leadership of former army general Robert Wood, it introduced retail stores on main
streets in U . S. towns and cities and expanded its product line to include "soft goods, "
such as clothi ng. Sears initially tried to maintain its centralized organization.
Centralizi ng procurement and offering a standardized product line in all its retail
outlets allowed the company to tempt suppliers with large orders, enabling it to
negotiate especially favorable prices. A strategy of standardized product offerings,
however, was incompatible with the product-line expansion and was doomed to failure
because customer demands varied widely over the nation. For example, during the
winter, warm coats would sell well in the cold northern states but not in Florida or
Texas. I f the organization was to be responsive to local conditions, managers in the
different regional markets needed to be given more authority to run their local
operations. U ltimately, Sears reorganized its business by establishing divisions on a
regional basis so that its managers could exploit some scale economies while still being
responsive to regional and local needs.
The fourth firm to develop the multidivisional form was Standard Oil of New
Jersey (now called Exxon). When the U . S. Supreme Court dismembered the old
Standard Oil Company in 1 9 1 2 for antitrust violations, gasoline had only recently
surpassed kerosene as the major product of the oil industry. I n the 1 3 years that
followed, buffeted by war and recession, growing demands for gasoline, and the
discovery of huge new oil fields in Texas and Oklahoma, Standard Oil of New Jersey
sought to integrate vertically all the operations from oil exploration to gasoline
marketing in a single huge firm. From its strong oil refining base, it integrated
upstream (or backward) to supply its refineries with oil, establishing its own extensive
oil field exploration and development operations, pumping its own crude oil and
transporting it through its own pipelines. It also expanded downstream (or forward),
distributing and selling refined products to retail customers. I n addition to these
vertical expansions, the company spread horizontally from its original kerosen e refining
 542
The Design and base, expanding its production of gasoline and lubricating oil and introducing other
Dynamics of petroleum-based products. The company reorganized its management several times
Organizations in attempts to cope with the increasing complexity of its operations. Some of these
changes were intended to centralize authority in order to improve coordination within
the firm in marketing or manufacturing petroleum products. These changes typically
placed individual plant managers under the control of a central office committee.
Other changes were intended to reduce the reliance on committees in order to fix
responsibility for performance on individual managers and executives. Finally, in
1 92 5 the company began a process of divisionalization. Within two years the company
had arrived at a multidivisional structure with individual managers, rather than central
office committees, in charge of each unit.

The Multiproduct Firm

As new management methods and systems evolved and improvements in transportation
and telecommunications continued, it became possible to include more and more
activities within the domain of a single firm, rather than organizing the same
transactions through the arm's-length purchasing that characterized traditional market
arrangements. At the same time, rising consumer incomes generated demand for
larger quantities of a broader array of products and services, and technological and
commercial innovations met these demands.
The giant firms of the early twentieth century were narrowly focused: U. S. Steel
made steel, Ford made cars (in only one model and color), and Gillette made razors
and blades. At Du Pont, General Motors, Sears, and Standard Oil of New Jersey, the
years after adoption of the multidivisional form were accompanied by a great expansion
in the firms' scope of activities. Exploiting its corporate strengths, Du Pont introduced
new fibers like nylon and other chemistry-based products. GM expanded vertically,
making most of its own systems and parts and even some of its own basic materials,
like steel. It also expanded horizontally, making trucks, refrigerators, air conditioners,
railroad locomotives, and other products that exploited its design and manufacturing
expe::tise, as well as into commercial lending to car buyers. Sears, whose special
competence was retail marketing, introduced its own brands of durable goods,
integrated vertically to a limited degree into manufacturing, and later expanded its
horizontal range by introducing insurance and other retail financial products. Standard
Oil also continued to expand both vertically and horizontally into all aspects of
petroleum-related businesses, becoming the largest corporation in the world.
During the course of the century, firms worldwide took increasing advantage of
the multidivisional form to integrate forward and backward, bringing within the firm
transactions with suppliers and customers that previously might have occurred in
markets. They also greatly expanded the range of unrelated activities that they
encompassed. This was accomplished both by entering new businesses themselves and
by buying existing firms. This process peaked in the 1 960s with the emergence of the
giant conglomerates, any one of which might involve businesses from baking through
hotels and car rentals to industrial components manufacturing and telecommunications
(as ITT did). The extent to which firms could profitably take over the operations of
their suppliers and industrial customers and also expand into new activities is
highlighted by the following description of the growth of South Korea's giant Lucky­
Goldstar chaebol or industrial group:
My father and I started a cosmetic cream factory in the late 1940s. At the
time, no company could supply us with plastic caps of adequate quality
for cream jars, so we had to start a plastic business. Plastic caps alone
were not sufficient to run the plastic-molding plant, so we added combs,
 toothbrus hes, and s oap boxes. This plas tic business als o led us to The Boundaries
manufacture electric fan blades and tele phone cases, which in tur n led us and Structure of
to manufacture electrical an d electronic products an d telecommunication the Firm
equi pmen t. The plastics business als o took us into oil refining which
needed a tanker-s hipping company. The oi l-refini ng company alone was
payi ng an insurance premium amounting to more than half the total
revenue of the then largest insurance company in Korea. Thus, an
insurance company was started. This natural step-by-step evoluti on through
related businesses resulted in the Lucky-C oldstar group as we see it today.
For the future, we will base our gr owth primarily on chemicals, energy,
and electronics. Our chemical business will conti nue to expand toward
fine chemicals and genetic engineering while the electr onics business
will grow in the directi on of semiconductor manufacturing, fiber optic
· telecommunicati ons, and eventually, satellite telecommunications. 3

Drivers of Change: Complementarities and Momentu m

Organizations change when their envir onments and the technologies they use change ,
and as they accumulate informati on and experience about what ki nds of organizati ons
work best for particular tasks . The key envir onmental change in the U nited States in
the nineteenth century was the development of technologies (the railroad and telegraph)
permitti ng emergence of unified national markets. As we have discussed, this
change favored large-scale producti on technologies, which in turn favored large-scale
organizati ons like those of the railr oads and the steel and oil companies, and led to
the development of instituti ons for large-scale finance. The firms involved in this first
round of changes impr oved an d refined their methods, which further changed the
environment of traditi onal industries by maki ng it less costly for them to expand their
scales, to find managers trai ned and experienced in the new methods, and to finance
their own large-scale operations.
A system of complementary changes, once put in moti on, develops a natural
momentum as the range of applicati on expands and as firms improve their methods
in the course of using them, making the methods even more effective and valuable.
(Recall that two activi ties are complementary if the profit or value created by doi ng
both is greater than the s um of the individual profits from doing just one or the other . )
The complementarities among changes in technology, demand, and the structure and
scope of enterprise continued to generate positive feedbacks on each other through
the twentieth century.
The 1980s in particular s aw a refocus ing, with firms limiting the number of
different businesses they tried to encompass and manage. Simultaneous ly, many firms
expanded the number of products they pr oduced in any one line, increased their
abilities to develop and market new products, expanded their activities internati onally,
and searched for new ways of organizing their internal structures to facilitate these
strategic changes . They als o sought to create new sorts of relati onships with other,
independe nt companies, cooperating in s ome ways while remai ning separate in others.
We will return to exami ne these more recent developme nts later. First we must
examine the factors that figure i nto the choice of organizational structure and the
scope of a single firm's activities.

3 Koo Cha-Kyung, son of the Lucky-Goldstar founder, as quoted in Francis Aguilar and Dong­

Sung Cho, "Gold Star Co. Ltd . , " Harvard Business School Case 9-38 5-264, reprinted in Management
Behind Industrialization: Readings in Korean Business, Dong-Ki Kim and Linsu Kim, eds. (Korea U niversity
Press, 1989), p. 426.
 544
The Design and Tl IE INTERN .\L STH.UCTL 1 HE OF THE F I R�I
Dynamics of The multidivisional form is the dominant model for organizing business. What are
Organizations the sources of its strengths, and what are the characteristic difficulties entailed m
designing and managing a divisionalized operation?

Advantages of the Multidivisional Form

The multidivisional firm emerged in an era when there were only two major
alternatives: highly centralized organization, such as that which had previously existed
at Sears and Du Pont, and organizations with almost no central control, such as the
form that existed at General Motors before the reforms introduced by Alfred Sloan.
Compared to those alternatives, the multidivisional form had decisive advantages.
In centralized organizations, top management and its staff try to stay informed
enough to control the operating decisions of their many units directly. In the Standard
Oil case, for example, there were oil fields, pipelines and shipping, refineries, marketing
operations, research laboratories, financial operations, natural gas operations, and
so on. In addition to these domestic operations, there were various foreign operations
that further increased the complexity of the business. Even with head-office specialists
in finance, marketing, production, and the like overseeing the various activities and
with staff assistants to gather and cull information, the operations were far too varied
and extensive for any single group of executives to manage closely. The members of
the central office committees could never be well enough informed about even a
small fraction of these operations to participate effectively in making good operating
decisions.
The problem was similar to the one the Hudson's Bay Company had faced more
than a century before: The people empowered to make decisions were too far removed
from the action and could never have the relevant information in a timely fashion.
However, an extra dimension had been added by the vastly greater volume and scope
of Standard Oil's business compared to the HBC's. Even if all the relevant information
could have been quickly and accurately conveyed to head office, its volume and
complexity would have overwhelmed the central decision makers.
Comprehensive decision making in a large organization must involve consider­
able delegation of authority to lower levels of the organization. Even if the organization
chart calls for a single individual or committee to run everything, the sheer numbers
of decisions to be made would overwhelm the process and as a result, a great number
of decisions are made at lower levels or not at all. The design of the multidivisional
form recognized this reality and sought to put mechanisms in place to ensure that the
decisions made at lower levels by those with the relevant local information would be
well coordinated and would be guided by proper incentives-the goals that the North
West Company's partnership arrangement had realized long before.
I\IPROYED h:FOR\I\TIO" .\:\D 1:---;cE:\Tt\"E::- In a multidivisional firm, the division
manager is empowered to make operating decisions in his or her unit. That manager
resides not in the central office, where he or she is forced to rely on second-hand
reports, but near the actual production or markets that define the division, visiting
the most critical operations to gain first-hand knowledge of how they work, consulting
with those on the scene, hearing suggestions, viewing options, specializing in the
knowledge that is relevant for running that particular division. There is no attempt to
convey all this information to the central office. By consulting with the central office
when that seems advisable or necessary, the executive can still take advantage of the
experience and perspectives of the headquarters staff, but the headquarters does not
exercise direct day-to-day control. Decisions can be made quickly by people on the
scene with the relevant knowledge.
 With deci sion-makin g authority goes responsibi li ty as well. Managers arc held The Boundaries
accoun table for the performance of th ei r di vision s an d rewarded accordingly. Th i s and Structure of
obviou sly requires that the central office have information by wh ich to judge the Fi rm
performan ce, but at an operatin g level this can be largely su mm ary data and financi al
measures that can be si mply transmitted and relatively easi ly evaluated.
The divisi on manager's intimate knowledge of the operation and its suppliers,
cu stomers, and competitors also means that he or she i s often also uniquely well
placed to recognize new strategi c options an d evaluate at least their local costs an d
benefits. I nvolving the division m anagers i n the formulation of plans an d strategies
then permits their knowledge to be used i n these processes. More detailed i nformation
may be needed for planning m aj or strategic i nvestments, but th is information need
not be tran smitted conti nuously.
By separati ng di fferent parts of the bu si ness into disti nct divi sion s with well­
speci fied areas of responsibili ty an d auth ority, the problems of determi ning who is
respon sible for good or bad performance are lessened and the li nkage between the
efforts of individuals or sm all groups and measured performance of their unit are
ti ghtened. This allows an d encourages stronger i ncentives to be provided, becau se
performance can be more accurately measu red. These increased in ce ntives are
provided first at the level of the divi si on managers, who are judged on di vi sio nal
performance, but pay can also be li nked to divi sional performance for other divisi onal
employees.

ENHANCED COORDINATION AND CONTROL Some of the organizati ons that predated the
multidivi sional form, such as the N orth West Company's, di d h ave decentrali zed
decision-maki ng authority. However, the broadened range of activities that was
undertaken within single firms in the era after World War I brought increased demands
for coordination. B efore Sloan's reorganization of General M otors, the m anufacturing
man agers of Buick, Cadi llac, Chevrolet, Oakland, and Olds operated with suffi cient
indepe ndence but not with sufficient coordination among themselves an d with the
sales operations. The central office di d not have the i nformati on it needed to evaluate
an d coordi nate the plans an d deci si ons of indepen dent managers. The separately
chosen product strategies of the units led to more competition among th em selves than
with Ford. Fai lure to coordinate on desi gn standards also prevented the divisions from
taking full advantage of the potential economies of scale in maki ng or purchasin g
common components, li ke sparkplugs and beari n gs. Failure to coordinate sales and
production or to charge i nve ntory costs to the divi si on led to production levels that
someti mes substanti ally outpaced sales.
The multi divi sional form could resolve these problems becau se in additio n to
its decentrali zed divi sion s it also h ad a central offi ce staff to plan strategy, coordi nate
divisional activities, and assess the performance of the divisions and their man agers.
At Ge neral Motors, the ce ntral office was respon sible for assi gni ng target market
segments to the i ndivi dual division s, creati ng more comprehensive an d useful divi sional
performance measures, and spo nsori ng group meetin gs to explore ways to achieve
economies of scale in the combined operations. Its success in achi evi ng these obj ectives
led others to mi mi c GM's structure and strategy.
The head office also took respo nsibility for raisi ng capital and allocati n g it among
the divi sion s. Centralizing the dealings with the capital market economized on the
skills needed in th is fu ncti on and, in an era before the development of very sophi sticated
financi al markets, may have led to a lower cost of capital than the i ndividual divi sions
would have en joyed as separate companies. The allocation of capital amon g divisio ns
also m ay h ave been more efficient than the m arket could h ave accompli shed because
the central executives in the firm had better access to information.
 5-46
The Design and l'.':CHE.\SED ABI LIT) TO l\hf\..\GE Ot\"EHSE BUSINESSES Adopting the multidivisional
Dynamics of form is complementary to engaging in a wide range of business activities. This can
Organizations be seen either by thinking about how the form affects the value of widely ranging
business activities or about how the widely ranging activities affect the form. In one
direction, adopting the form is most valuable when the businesses being managed are
already diverse, as we saw in the histories of the four companies that innovated the
structure. In the other direction, adopting a multidivisional form makes additional
expansions into new activities easier to manage and therefore more valuable.

Problems of Managing a Divisionalized Firm

The challenges in efficiently designing and managing a divisionalized firm involve
the same dimensions on which the form finds its strengths. First, divisions must be
delineated, reporting relationships structured, and activities allocated among the
divisions and between them and the head office so that coordination is facilitated.
Second, information, decision, evaluation, and reward systems need to be structured
to encourage the appropriate behavior. Third, the scope of activities that the firm is
going to undertake must be chosen in light of the costs and benefits.
DEFI :\ l'.\C O1 \'ISI0'.\S .\'.\D REP0RTl'.\G REL.\TI0'.\SHI PS Often there will be a variety of
ways in which divisions can be defined: geographically, by the technology employed ,
by the products produced or the particular market segments being served, and so on.
Moreover, there is an issue of how many divisions to establish and where to draw the
lines between them. The choices made on these dimensions may be very important
to the way the firm functions and its success in coordinating its activities.
No matter how the cuts are made to divide up the company's business, there
will inevitably be mismatches between the way problems present themselves and the
way the company is structured. If the divisions are defined geographically with, for
example, a domestic and a foreign operation, then coordination problems will arise
with customers or suppliers who operate in both markets, have not structured
themselves this way, and want to deal with one entity, not two. This has been an
issue for IBM in dealing with international businesses that use its computers and \Vant
to have uniform systems and service across their operations.
Defining divisions technologically facilitates dealing with R&D, design, and
especially manufacturing issues, because these are apt to be heavily determined by
the technology. Having only a single unit dealing with each technology then means
that coordination across divisions is not necessary on that dimension. However, the
technology of production may have little to do with customer needs: Du Pont' s attempt
to operate their explosives and fertilizer business as one unit based on the nitrogen­
chemistry underlying both products illustrates that kind of difficulty.
A product-based definition becomes problematic when opportunities are present
to make different divisions' products in the same production facilities or to sell the
products of different divisions to the same custome�s. The customers may want to buy
a whole variety of different products without having to deal with different divisions.
Developing new products in this context can lead to special problems. For example,
at one point two Hewlett Packard divisions-one that made hand calculators and one
that made minicomputers-were developing personal computers, with the two designs
being incompatible. Defining divisions by market segments means that the same
problems with suppliers and production will arise across divisions and have to be
addressed repeatedly.
Ideally, to avoid the need for coordinating across division boundaries, divisions
should be self-contained units. The problem is that being managcably small and also
self-contained may be incompatible. Defining the divisions very broadly minimizes
 !547
The Bou nda ries
and Structure of
the Firm
Coordinating the Computer Businesses
of 1-/ewlett Packard
William Hewlett an d David Packard founded the l kwlctt Packard Com pan y
(HP) in a Pal o Alto garage, where they made scien tific an d industrial ins tru­
men ts. The com pan y achieved immense success, a superb reputation fo r
high-q uality produc ts, and an exceptionally loyal clien tele.
HP has a very strong corporate culture. One of its elements is to
enc ourage responsibility and innovativeness by m aintaining extreme divisional
independence. Its policy has been to keep divisions as small, manageable
e 11 tities with which their mem bers can identify and wh ose success depen ds
on their efforts. Cons equently, whenever a succe ssful division grew too large
($20 million in sales or 1500 em ployees was the usual point in the 1980s),
it was spl it up or a new division established.
This degree of independence created coordination problems, eve n in
the instrument business, which is characterized by relatively discrete products.
H owever, with H P's entry into the com puter business, these difficulties
in tensified. The traditional divisions were individually too small to design
and produce whole com puter system s on their own. Thus, work h ad to be
broken up am ong divisions, with one producing hard drives, another the
input- output system s, a third handling display terminals, and so on, but
without a single individual with responsibility for the project's success and
auth ority to match . Th is division of responsibilitie s led to serious coordination
difficul ties for a system wh ose parts have to fit and work together. The
coordination task was further com plicated by the rapid pace of change in the
com puter business, which put a premium on speedy product development
and quick response to changing technological, dem and, and com petitive
conditions.
To overcome the resulting conflicts, HP experimented th rough the
1980s with increasing cen tralization, assigning single executives more and
m ore responsibility over com puter operations, and sh ifting the structure of
reporting rel ationships to encourage better coordination. The centralizing
moves represented wrench ing changes for H P and presented a serious
challenges to H P's vision of itself.

Based in part on Richard T. Pascale, Managing on the Edge, (New York: Simon and
Schuster, 1990).

the interactions am ong them and the need to coordinate these. H owever, divisions
that arc too big run into all the problems of firm s that are too big: It becomes
impossible for the senior divisional managers to kee p informed and to find time to
deal with all the problem s, and measuring individual contributions for performance
evaluation s becomes too difficult.
A modern solution to these problem s is to keep divisions relatively small but to
group toge ther th ose th at arc likel y to need coordination am ong them . The individual
division managers in the group then all report to one senior executive, wh o is
responsible for the group and rewarded for its performance. The largest firms may
even have a n umber of groups, with th eir presiden ts reporting to another level of
 548
Execut ive Offices
The Design and
Dynamics of
Organizations Consumer Products

G roup #1 G roup #2 Division F

Division A Division D

D ivision B Division E

Figure 16. 1 : The structure of

a divisionalized firm with divi­
Division C
sion groupings.

executive below the top levels of the firm . Figure 1 6. 1 illustrates a hypothetical
divisionalized firm with a group structure. Divisions A, B, and C are in Group 1 and
report to a group president. Divisions D and E are in Group 2 under its president.
These two group presidents report through a single executive to headquarters. In the
figure, these are groups of consumer products divisions. Division F, wh ich m ight, for
example, be a heavy equipment division with little need to share information or plans
with consumer products divisions, reports directly to headquarters.
In this structure, disputes between divisions are ultimately to be settled at the
lo\\'eSt level of the hierarchy to which both divisions report. Thus, a con flict between
divisions A and B over which will get a given customer's business would be settled by
the group president, but conflicts between C and D over the transfer price that one
will pay the other for its output would be settled at the level to which the presidents
of the t\\'o different groups report. A conflict between E and F would go all the way
to the top of the firm .
I n general, the problem i s to design this structure to reduce coordination
problems and to handle those that do arise at the lowest possible level . The higher
the le\·el to which decisions must be pushed, the slower is decision making, the greater
are the costs of information transmittal , and the greater are the possible influence
costs that may be incurred as successive levels of executives campaign for their units'
interests .
.-\SSIC:\l:\C ACTl\'ITIES A:\'D RESPONSIBILITI" TO LEYELS Adapting well to changing local
circumstances, using local information we11 , saving on the costs of information
transfer, and making effective use of scarce central management time and attention
all argue for pushing decision-making pm\·er and responsibility as far down in the
organization as possible . Economies of scale of various sorts may mean that not all
activities should be pushed down to the division level , however.
For example, almost all firms maintain the finance function at the corporate
b·el . Research and development may be done at quite low levels, where the scientists
and engineers are close to the constituencies they serve, or they may be centralized.
Hewlett Packard has tended to follow the first model, whereas AT&T, before its court­
ordered break-up, centralized at least the more basic R&D for the Bell System in Bell
Laboratories. Japanese firms, in which personnel are frequently shifted between
divisions and firm-financed i 1l\'estments in human capital are especially important,
tend to centralize the h uman-resource function to facilitate tracking individual
employees' progress and managing their career development. In many North American
 549
firms, in con trast, personnel man agement is more decentral ized. By separating the The Boundaries
sales and distribution function s from the product divisions, many firms seek to gain and Structure of
the econ omics of scale in distribu tion and sales, for exam ple, by having one sales the Firm
represe ntative call on customers or one truck deliver a range of differen t produ cts. A
problem is that the information that salespeople may gather from customers needs to
be tran smitted to many different units.
11\iCENTl\'E AND CONTHOL l sst rEs Decentralizing authority facilitates making good use
of local information, but the very fact that decisions are based on information that is
not available to headquarters exacerbates the moral hazard problem. Consequently,
incen tive systems must be put in place, incurring the kinds of costs discussed in
Chapters 7 and 8. Granting decision-making authority and assigning financial
responsibility for outcomes to the same person are complementary actions: Each is
more' valuable when done with the other. As we have seen, providing incentives
effectively requires creating good performance measures, which is another cost of
divisionalization. The spread of divisionalization was promoted in part by the
development of accounting and management methods that allowed improved incentives
for division managers.
The point that a manager with broader authority should be given stronger
incentives is an application of the incentive intensity principle. For example, a division
manager who controls sales and marketing as well as production has more ways
available in which to improve unit performance and can be profi tably given more
intense incentives. The relationship, however, is one of complementarity, so the
reverse implication also holds: The greater the manager's financial incentives, the
greater the proper scope of the manager's authority.
THE TtUNSFER PRtCINC PROBLEI\I The need for divisions to transact with one another
provides a second way in which incentives can affect the appropriate allocation of
activities among divisions. As we discussed in Chapter 3 , interdivisional transactions
are quite common. They are unavoidable in a vertically integrated company in which
one division supplies inputs to others, as in the example of the integrated oil company
where the refi nery buys crude oil from the pipeline division that in turn buys the oil
from the production division. If a company sells systems as well as individual pr oducts
such as personal computer systems that consist of the actual computer, a display, a
keyboard and mouse, a modem for communicating with other computers, and a
printer, each of which could also be sold separately, then each of these might sensibly
be made in a different division. When sold together they will have to be transferred
to the unit selling the system. Because transfer prices determi ne the revenues received
in internal transactions, they are of immense concern to division managers who are
paid for division performance.
We saw in Chapter 3 how, if managers have discretion to decide how much
they want to buy and sell in interdivisional transactions, wrongly set transfer prices
can hurt corporate profi tability. Yet allowing managers such freedom gives them
greater opportunities to respond fully to incentives and provides competitive pressure
for internal suppliers to maintain cost and quality control. This means that the setting
of transfer prices is an important issue.
We also saw in Chapter 3 that if there is a competitive market for the product
or service being transferred, then the transfer price should equal the market price
(perhaps adj usted for any cost adva ntages of internal suppl y). H owever, when the
exter nal price reflects monopoly distortions or when the product simply is unavailable
outside, then this soluti on is inappropriate or unavailable.
An alternative is to let the managers in volved bargain over the transfer price
and the quantities to be transferred. Because each is likely to be privately informed
The Design and about the costs and benefits of the transaction to his or her department, howe,·er,
Dynamics of there is the possibility that the sort of bargaining problems discussed in Chapter 5 "·ill
Organizations create inefficiencies. The supplying di\'ision will be inclined to overstate the costs of
production in order to get a higher price, and the purchasing di\'ision will be inclined
to understate the marginal value of the transferred good in order to reduce the price
it pays. The result could be that too little is transferred and final output is too low.
The same result is likely to occur with other systems as well.

Double .llargi11alizatio11: The .llathematics

of Trrmsfer Pricing
The nature of the difficulty can be seen by assuming that the system in place
allows the supplier or upstream di\'ision to state the price at which it will supply
an input to the final producer or downstream division , and then the downstream
di\'ision decides how much to buy. Suppose the downstream di\'ision markets
the final product directly to customers outside the firm at a constant cost cF per
unit. It faces a demand curve that is downward sloping. Production of a unit of
final product requires one unit of the intermediate good, which can be produced
at a cost of c1 per unit. Thus, the total cost of the final good to the firm as a
whole is Ct· + c1 , and the firm would want the quantity sold of the final good
and thus the amount transferred of the intermediate good to be determi11ed by
the condition that the marginal re\'enue on final sales be equal to cF + c1 .
For example, if the final demand is gi\'en h P = 1 0 - Q/ 1 6, where Q
is the amount sold at a price of P, then total re,·enue is I OQ - Q 2/ 1 6, and
marginal re,·enue is MR = 10 - Q/8 (this can be quickly obtained by
differentiating the total revenue function). If the \'alue of cF is 2 and that of c1
is I , then the marginal cost to the firm as a whole is 3. Equating J\IR to marginal
cost gi,·es IO - Q/8 = 3, or Q = 56, which means the final price should be
P = 6. 50. These calculations are shown in the first column of Table 1 6. 1 .
To achie,·e this in a dh·isionalized firm under the gi,·en procedure, the
transfer price should be T = 1 , or more generally, T = c1 • Facing th is price,
the manager of the downstream di\'ision will see a marginal cost to his or her
di\'ision of 2 + T = 2 + I = 3, the actual cost to the firm. He or she will
then select Q = 56, (where J\IR = 2 + n and order this many units of the
intermediate good from the upstream di\'ision. If any other value ofT is selected ,
then the dmrnstream manager will buy and produce an amount that does not
maximize corporate profits. If T is set at 4. for example, then the manager
downstream will face a marginal cost of 2 + 4 = 6 and will select Q = 32,

Table 16. 1 An Example of Double l\larginalization in Transfer Pricing

Integrated Downstream Upstream

Fim1 Division Division

Demand P = 1 0 - Q/1 6 I' = 1 0 - Q/1 6 T = S - Q/8

Total Re,·cnue I OQ - Q'l 1 6 IOQ - Q'/16 SQ - Q'l8
\ 1arginal Rc,u1uc 10 - Q/8 1 0 - Q/8 8 - QI-I
Total Va riable Cost 3Q (2 + T)Q Q
\ 1arginal Cost 3 2 + T I
Quantit\' 56 8(8 - T) 28
Price 6. 50 8. 2 5 T = 4. 50
 ;;,; 1
where l\lR equals the marginal cost he or she faces. Thi, \\ ould lead to le" The Bou11daric1
profit for the firm. Entering different va lue, for T in the second column of the and Structure of
table gives the quantities that result. the Firm
At a transfer price equal to its constant marginal cost, however, the
revenues of the upstream division arc just equal to its total variable co,t,. • It,
manager, being evaluated on the division's profits, would do better with a higher
transfer price, even though this hurts corporate profitability. For example, he
or she shows a net d ivision profit of (4 - I )32 = 96 if the transfer price is sci
at 4.
!\lore generally, the relationship MR = cf + T defines how much the
downstream division will want to sell 011 the final market and thus how much
the upstream division can sell internally at any transfer price 'J'. Equating the
marginal cost of Cf· + T to the marginal revenue of IO - Q/8 leads to Q =
8(8 - n. This formula indicates the amount the downstream division will want
to sell as a function of the price it pays for the input and thus the amount of
input it will buy. Th is is effectively a demand curve for the upstream d ivision
and is entered accordingly in the first row of the third column. For the upstream
division to maximize its profits, its manager will equate the marginal revenue
facing the upstream division to its marginal cost, c1 • ll1c fact that the upstream
di,·ision considers the marginal revenue of the firm as its (inverse) demand cun·e
and looks to the cun·e that is marginal to it in computing the transfer price is
the source for the term double marginalization.
The inverse demand corresponding to a demand of Q = 8(8 - T) is
T = 8 - Q/8. This gives a marginal revenue of 8 - Q/4. With a marginal
cost of c1 = I, this yields Q = 28, T = 4. 50 as the optimal choice for the
upstream di,·ision (see the third column of the table). This gives the upstream
di,·ision profits of 28(4. 50 - I) = 98. With only 28 units rather than 56 being
transferred, however, final goods' output is half what it takes to maximize
corporate profits. Corporate profits, which would have been 56(6. 50 - 3) =
196, are now only 28 (8. 25 - 3) = 1 47, where 8. 2 5 is the price the downstream
di,·ision charges in the final market when it faces a transfer price of 4. 50 and
correspondingly chooses to sell 28 units. (We will see this sort of double
marginalization problem again later in the chapter . )

Approaches t o the Problem. It is probably unreasonable t o expect that the upstream

manager would be allowed to set the monopoly price for the intermediate good. Still,
there will be a tendency for the upstream manager to select too high a price for
corporate profit maximization. Similarly, if the price is left to the d0\'11Stream
manager, with the upstream manager deciding how much to supply, then the
downstream manager will have an incentive to take advantage of his or her monopsony
position and set too low a transfer price. This results again in loo little being transferred.
One solution to this problem migh t be to order the transfer to occur at the
upstream di,·ision's marginal cost. However, if the head office does not know the
actual costs in the upstream division , then the manager of this division may have an
incenti,·e to overstate his or her division's marginal costs to get a higher transfer price.
This would again result in more re,·enues and recorded contribution in the upstream
division, but lower corporate profits overall. Another might be to make the upstream
division a cost center, with transfers at marginal cost and the manager being rewarded
for keeping costs down. 111is will not be easily implemented if the upstream division

,. \\'ilh constant marginal costs, a\·erage variable cost equals marginal cost.
 552
The Design and Raw
Materials
Dynamics of
Organizations +
Parts
+
Systems

Final
Asse mbly

Distri bution

Customers Figure 16. 2: A hypothetical single product firm.

also has markets outside the firm, however. A third possibility is to merge the two
divisions, but this merely converts a problem between divisions in the firm into one
between departments in the division, unless it is accompanied by some other real
change in responsibilities and incentives.
OECIDINC ON THE SCOPE OF THE FIRM The multidivisional firm is able to take on and
manage a much wider set of activities than would be possible under a less decentralized
form of organization. As Coase accentuated, the costs of administering transactions
internally versus the transaction costs of conducting them through markets determine
the boundaries of the firm. The innovation of the multidivisional form lowered the
costs of internal administration. This means that the range of transactions optimally
brought within the firm should increase. Which activities should the expanded firm
undertake? That is the subject of the next two sections.

VERTICAL 80L l\DARIES Al\D REL\TIO!\S

We begin our study by examining a highly simplified version of the problem of
producing and distributing one single kind of good. Figure 1 6. 2 illustrates the usual
conception of this process. Production begins upstream, for example, with the
extraction of ores to be processed into the raw materials from which the good will be
made. Using hired labor and machinery of various kinds, the firm and its suppliers
convert these raw materials through a sequence of steps into parts and systems before
assembling the various systems into a final product. At the last stage, the product is
distributed to customers. For example, among the materials used in the production
of an automobile are steel, aluminum, rubber, plastics, foam, and so on, which are
made into parts like frames, padding, and covers for seats, as well as chassis, dashboards,
and other parts. These parts are in turn assembled into systems. For example, seat
systems may incorporate frames, levers, springs, padding, covers, and so on. Finally
the several systems are assembled into the product, which must be distributed
downstream to the customer.
Figure 1 6. 2 highlights the many steps involved in the creation of a product and
its delivery to the final consumer and hints that there may be many ways to divide
responsibility for the steps. At one theoretical extreme, each activity might be carried
out by a separate firm, which buys inputs from the firm one step upstream and sells
its product to the next firm downstream. At the other extreme, a fully vertically
integrated automobile company would own mines from which to obtain raw materials,
smelters and mills to create usable rolls of steel, and aluminum mills, plastics factories,
rubber plantations, tire plants, die makers, and so on, providing the capabilities for
operations from the di gging of the raw materials to the final assembly of the product.
It would also be integrated further into downstream operations, owning trucks, rail The Boundaries
cars, and cargo ships with which to distribute its cars to the dealerships, which it and Structure of
would also own . Its own employees would also provide the various staff services it the Firm
needs. Its law department would handle contracts and litigation . Its accounting,
engineering, and advertising departments could provide those services without relying
on outside suppliers. Its finance specialists would handle its relations with the capital
market, and its senior executives would plot strategy and its implementation on their
own, without relying on outside consultants.
A strategy based on extreme vertical integration was tried by Ford Motor
Company before World War II, but was later abandoned. In modern times, few firms
even approach the pattern of complete vertical integration . The question is: Why not?
What alternatives does the firm have for organizing its activities? Why does it rely on
independent suppliers for some services and its own divisions for others? What
determines which services are (or should be) purchased from outside suppliers and
which the fir� provides for itself?

Advantages of Sim p le Market Procurement

For many of the inputs that a company uses, actual market conditions pretty well
approximate the idealized competitive market model of Chapter 3. As long as the
firm's scale of purchases is too small to make it profitable either to design highly
specific inputs for its own particular needs or to make standardized products at an
efficient scale, firms tend to rely on independent suppliers. For example, even the
largest, most highly integrated firms rarely make the paper clips, ballpoint pens, and
coffee pots used in their offices, nor do they make the furniture, file cabinets, and
office equipment. Most do not maintain construction divisions to build their buildings
and factories or maintain fleets of aircraft to transport employees. The reason is that
these are standard goods and services for which there are many suppliers, so the firm
buying inputs can en joy the many advantages of the competitive market system without
fear of a hold-up problem .
ECONOMIES OF SCALE One significant advantage of using independent suppliers arises
when scale economies are important in producing the input the firm needs but its
own level of usage of the input would not allow the firm to achieve the minimum
efficient scale. For example, airlines are most efficiently run at a large enough scale
to enable regular flight scheduling. Even the largest industrial companies benefit by
buying air transportation services provided as part of a system serving many customers
rather than providing the services for themselves.
Companies do sometimes maintain their own small planes for executives whose
time is so valuable that it is efficient to schedule special flights to meet their needs.
They may also own planes for highly special ized transportation needs. For example,
a company exploring for oil in the Arctic might provide its own specially equipped
aircraft and prepare its own landing sites. Even in this case, however, if other
companies need a similar service and there are economies of scale or scope in providing
it, the service might be provided by an independent firm . Alternatively, the firm
might set up an operation to serve its own needs and also market the service to others .
For example, the Kimberly-Clark paper company, which has production facilities in
a number of small towns in Wisconsin, started an air service for its personnel that
developed into Midwest Airlines. 5

5 Milton Moskowitz, Robert Levering , and Michael Katz, Everybody's Business (New York: Double­
day , 1 990) , p. 501.
 55-1
The Design and Ec O:\Ol\ I I ES OF SCOPE Even when the scale of the supplier's production of any
Dynamics of particular product or service is small, the supplier may enjoy economies of scope by
Organizations engaging in activities that are unconnected with the buying firm's business. An
example comes from gasoline retailing, where it is often profitable to sell convenience
foods or to provide car washing or repair service at a gas station. In this business, oil
companies frequently purchase retailing services from independently owned gas
stations. They also sometimes handle retailing themselves through company-owned
outlets managed by employees. Theory suggests that the choice between these two
patterns of organization should depend on whether the outlet provides services other
than gasoline sales.
Even though the provision of these other services does not directly affect oil
company profits from gasoline sales, these extra activities can affect the relationship
between the oil company and the retail gas station manager. A business with multiple
activities has an effort allocation problem, requiring that incentives be devised so that
the station manager will devote the proper amount of time and effort to each aspect
of station operations. As we saw in Chapter 7, it is more expensive to provide
contractual incentives to agents engaged in multiple activities. The problem becomes
more severe when measuring performance in some of the activities is relatively hard
or when accurate performance assessments can be made only over long periods of
time. In these cases, it is advantageous to deal with an independent owner rather than
an employee.
In automotive repair, performance is hard to measure m·er short periods of time
because shoddy service may increase short-run profits while eroding sales and profits
(both in repair work and in gasoline sales) only over a much longer period of time.
This implies that the managers of stations providing repair services should tend to be
residual claimants rather than oil company employees so that they will have an owners'
incentives to care for the long-term value of the station. The relative ease of hiding
and misappropriating revenues from repair business works in the same direction. The
evidence supports this prediction. One recent study found that among stations offering
automotive repair services, 96 percent of outlets in the sample were not owned by the
affiliated oil company but were instead operated as franchises . 6 Another study used a
different sample in which 83 percent of all outlets offered automotive repair services.
Only 2 5 percent of the outlets owned by the oil companies offered such services,
however. 7
C ORE Cmt PETE '.\ CIES We have previously explained that core competencies are a
kind of economy of scope connecting products made at different points in time. For
example, the costs that a firm incurs in developing an ability to design, manufacture,
or market a certain group of products may be recouped from sales of all the products
in the group. A firm may choose to rely on an independent supplier even when there
are no economies of scale in manufacturing the current generation of the product
and no economies of scope among existing products because the supplier has de\·eloped
a special competence to improve the product or to incorporate quickly the latest
manufacturing innovations to keep production costs low. The box on EDS and
Continental Airlines illustrates such a decision: EDS has a core competence in

6 J . A. Brickley and F. H. Dark, "The Choice of Organizational Form: the Case of Franchising, "
/ournal of Financial Economics, 4 ( 1 987), 40 1-20. In a franchise arrangement, the franchisee (the sen·ice­
station owner) makes contractual payments to the franchisor (the oil company) for use of its name, but
remains the residual claima nt. A fuller discussion of franchising is given later.
' Andrea Shepard, "Contractual Forni, Reta il Price and Asset Cha racteristics, " working paper, l\1 1T,
( 1 990).
 55!'>
The Bounda ries
and Structure of
the Firm
EDS and Continen ta l Airlines
System One is a subsidiary of the U . S. carrier, Continental Airlines. The
subsidiary owns the airline's computer reservations system, which provided
services to nearly 20 percent of the airline reservation agents in the United
States. Until l 99 1 the subsidiary also provided various data and computation
services, such as air traffic scheduling, to about 1 70 different airlines.
In 1 99 I System One signed a ten-year contract with Electronic Data
Services (EDS), a subsidiary of General Motors Corporation, to perform the
airline services function. EDS, with 1990 revenues of $6. 1 billion, was
already a leading supplier of information services to the energy and banking
industries. It contracted to provide the information service at a lower cost
than Continental could achieve and agreed to accept 1 , 860 System One
employees as part of the bargain.
There are economies of scale in providing information services to
airlines, which arise from writing the programs and maintaining the data
bases. Many of the services EDS will provide, such as route scheduling, are
standard services that require no specific investments. These facts clarify
why smaller airlines do not provide the information services function for
themselves, but not whether Continental Airlines or EDS would be the most
efficient provider. EDS has developed special skills useful for running its
information service businesses, but it has limited knowledge of the operation
of airlines. By hiring System One employees as a group, it may be purchasing
the necessary airline expertise.

Source: Agis Sapulkas, "G. M. Unit Will Run Airline Data Service, " The New York
Times, (April 24, 1 99 1 ), C4.

computerized information services and that Continental cannot match . Alternatively,

a firm might undertake production of a good itself even when outside procurement
would be directly c heaper because the firm wants to build or maintain a competency.
The magnitude of economies of scope and the nature of core competencies are
particularly important for decisions about the horizontal extent of the firm, so we
postpone most of our discussion of them , including our examples, to that section of
this chapter.

INDEPENDENT, COi\lPETITl \'E SUPPLI ERS As a buyer in a competitive market for a

standard input, a firm need not show favoritism to any particular supplier. It can
purchase the input from the suppl ier will ing to meet its needs at the lowest price.
When technical innovations occur, even if they are adopted by only some of the
suppl iers, the buyer typically enjoys some of the extra value that is created, in the
form of either higher quality or a lower price.
If the buyer had its own supply division that was less innovative than the
competition, the buyer might face pressure to continue to purchase suppl ies from its
own unit to maintain employment and employee morale. Even it the firm successfully
resisted the influence attempt, it would suffer infl uence costs as employees spent their
energies trying to protect their jobs. It is also possible that these pressures would
succeed in forcing the company to incur extra costs to maintain an inefficient unit.
 556
The Design and Compounding these problems is the fact that if the supply division anticipates this
Dynamics of protection, its employees will feel less pressure to be innovative.
Organizations Generally, for standardized inputs that can be supplied by competitive firms
earning only normal profits, there is little that the firm can gain by making the product
itself. Owner-operated suppliers in a competitive market are driven to be innovative
by the desire to realize even temporary cost advantages and earn an increased profit.
A similar force drives them to be responsive to customers' other needs. Accumulating
experience as they run their own businesses, owner-managers have the knowledge,
motivation, and authority to make decisions that maximize value. The buyer cannot
do better producing on its own from scratch. The alternative of buying out the
independent supplier and bringing it into the firm as division exposes its manager (the
former owner) to being held up, with the new boss reneging on promises.
The same qualitative advantages are present even in firms that are not owner
operated, though usually to a smaller degree. The top executives of an independent
firm usually capture a larger portion of the value they create by making good business
decisions than do divisional managers within a larger firm. Indeed, large organizations
tend to emphasize distributional equity to avoid organizational politics and the resulting
influence costs. The greater claim by top executives on their business profits strengthens
incentives and contributes to improved performance.

CoMPETITl \'E BIDDING Even when the needed supplies are specialized and are not
available through simple competitive market transactions, many of the advantages of
competitive markets can sometimes be enjoyed by soliciting competitive bids. For
example, an office building may have to be built at a particular time and place and
according to a particular design, but the service of constructing the building may still
be a general skill for which there are many qualified suppliers. By relying on an
independent supplier, the firm takes advantage of the best available resources and,
provided there is adequate competition, pays only the real economic cost of the
service. From the perspective of the economy as a whole, hiring the contractor and
subcontractors with available resources avoids leaving crews idle and creates value,
some of which is captured by the purchasing firm.

Advantages of Veriical I ntegration

As we have seen, the advantages of simple market procurement are greatest when
particular circumstances prevail. These include the use of standard inputs, the presence
of several competing suppliers, economies of scale in the supply firms that are too
large to be duplicated by the buyer, economies of scope that would force the vertically
integrated firm into unrelated businesses, and the absence of specific investments on
the part of either the buyer or the seller. When these conditions fail, vertical integration
can enjoy significant advantages over simple market procurement.

II\IPRO\'ED COORDINATION AND BETTER PROTECTION OF I N\'EST!\IENTS In modern econo­

mies, firms frequently use components and services that are highly specialized. In the
area of marketing services, for example, a company selling a new medical imaging
system would need to train the sales force in the uses of the product so that they can
explain how the product can be used and why it is advantageous to use it. This
knowledge would then also permit the salespeople to report back effectively on
customer reactions, needs, and suggestions. Using an independent distributor would
require close coordination in the development of training materials and a routine for
transferring information from customers back through their firm to the producer. The
training itself would constitute a specific investment, and some arrangements would
have to be made to protect that investment. The box on the Rolin Corporation reports
 557
The Boundaries
and Structure of
the Firm
Strategic Business Decisions at the Rolm Corporation

The Rolm Corporation was founded in 1 969 as a manufacturer of computers

made to military specifications. By 1 97 3 the company was profitable on sales
of about $ 3 million annually. Looking for additional ways to apply technology,
the company's founders saw an opportunity to market the first computer­
controlled private branch exchange (PBX) systems. PBXs are the systems used
in private businesses to switch incoming telephone calls among employees,
to manage internal office telephone communications, and to direct outgoing
calls to an available "trunk" connected to the publ ic telephone system.
The product that Rolm proposed to make had several advantages over
existing PBX units. l t introduced new features, like call transfer, multiparty
conference calling, speed dialing, restrictions on toll calls from particular
phones, accounting records for calls from individual phones, and so on, all
of which were beyond the capability of the older electromechanical switching
equipment. In addition, using the computer to route calls in the cheapest
way could reduce phone bills. Despite these advantages, Rolm's decision to
enter the PBX business was regarded skeptically by industry observers. Rolm's
own consultants had argued that computer technology would be too expensive
for use in telephone switching and that it was foolhardy for such a small
company to use technology to compete with AT&T and its world-famous
Bell Laboratories research subsidiary. Spurning this advice, Rolm decided to
proceed with the project.
The decision to develop and market this new product required that
Rolm make a number of choices . First, should the system be designed to use
existing telephones or new ones better suited to a computer-controlled system?
Using existing telephones would reduce the technical capabilities of the
system. However, making special telephones would either stretch the com­
pany's engineering and production resources or force it to coordinate with a
special ized telephone suppl ier, which might lead to delays or other problems.
Rolm decided to use standard telephones for its first system .
Second, should the company build its own sales force to market the
new system to businesses? Or, should it instead rely on some existing sales
force? At the time, AT&T held a 78 percen t share of the national market,
with General Telephone and Electronics (GTE) and United Technologies
as its nearest competitors, but there were also some smaller compan ies
providing installation and service of imported PBX systems.
Building its own sales force to sell to a national market would be time
consuming and costly and would not make the best use of the company's
strengths, which lay in its engineering capabil ities. Accordingly, Rolm decided
to rely on eight existing regional PBX sales and service organizations. To
enable installers trained in the older electromechanical technology to install
its computer-controlled systems, Rolm compromised its design so that the
external connections on the new PBX resembled those on the older systems.
To protect the sel ling organizations' specific investments in training, Rolm
offered them exclusive territorial franchises .
The first PBX units were shipped in 1 97 5 , leading to sales of $ 1 . 3
million for that year. These grew to $ 1 0 million in 1 976, and $20 million
in 1 977. In 1 978 GTE signed up to become a Rol m distributor and Rolm
 558
The Design and
Dynamics of
Organizations
began to internalize part of the marketing fu nction by adding three marketing
companies of its own. By 1 980 annual sales were $200 million, with half
coming from Rolm's own sales force.

Source: Lecture by Rolm cofounder Robert Maxfi eld, Stanford University , May 2 1 ,
1991.

how one high-technology company made and managed its decision to use independent
distributors.
Buying inputs for manufacturing raises similar difficulties. In simple market
transacting, the downstream firms with knowledge of customer demand decide what
to make and sell and then buy the materials, parts, systems, or services they need
from upstream firms. Taki ng into account quoted prices plus shipping costs and other
terms, they seek to acquire the needed inputs as inexpensively as possible. They may
solicit competitive bi ds or negotiate with suppliers over prices and features, especially
for nonstandard inputs. Soliciting bids to provide a good at the best possible price
requires that the characteristics of the desired product be specified in great detail, with
drawings and performance specifications prepared before the bidding can begin.
Otherwise, the bidders can only guess at the costs involved and may refuse to make
any price commitments. This problem is especial ly severe for new products and for
companies that engage in frequent product redesign. M oreover, if producing the
desired input effi ciently requires that the supplier make highly speci fic investments,
the necessary investments may be blocked or diminished by the threat of a subsequent
hold-up.
Vertical integration alleviates all of these pr oblems. In the integrated organiza­
tion, planning entails consultation between those who sell the product, those who
make it, and those who supply parts or systems for it. Together they forecast capacity
needs and identify product improvements and investments in specialized equipment
that promise higher qual ity or lower production costs. If the investment is hi ghly
specific, vertical integration alleviates the hold-up problem by eliminating the
opportunity to negotiate over the price paid to the owner of the newly created asset.
It would be a gross exaggeration, however, to suppose that verti cal integration
completely eliminates these incentive problems. As we have already seen, even in an
integrated organization, individual managers have different interests. The manager of
the sparkplug unit may still haggle over the transfer price of sparkplugs because this
number affects the unit's recorded revenues and the manager's measured performance.
U nlike the owner of an independent firm, however, the inside manager cannot
threaten to withhold supply and bring downstream production to a halt if he or she
is unhappy with the established transfer price. This ability to continue performing
disputed services while deferring any settling up until later is more limited in ordinary
market procurement.
Bl·:nl Cl:'\C Tl IE NEED FoH STHO;'l;C PEHFOH\I \'.\CE l'.\C :E:\Tl n:s An independent supplier
makes its own decisions about how to all ocate its time and effort among various
activities, including suppl ying profitable services to other customers. According to the
equal compensation principle, to induce this supplier to pay adequate attention to the
fi rm's needs, appropriate financial incentives are needed. The more difficult it is to
 559
Loss of Val ue Caused The Boundaries
by Monopoly Pri c i ng and Structure of
the Firm

Marg i nal Reve n u e Figure 16. 3: Monopoly losses.

measure performance and hence to provide fi nancial incentives, the more costly it is
to use an independent supplier for the service.
The importance of this factor was confi rmed by a recent study of the choices
that electronic components manufacturers face about whether to use their own direct
(employee) sales agents or to rely on independent manufacturers' representatives
who would also represent other firms. 8 Examining 1 59 U . S. sales districts of 1 3
manufacturers, the study found that the two principal determinants of vertical
integration into sales were the difficulty of evaluating performance and the importance
of nonselling activities. When performance is difficult to measure, providing strong
incentives is costly, whether they are given to employees or independents. Weaker
incentives can be used for employee agents, however, without the fear that they will
emphasize selling the products of others, as an independent representative would if
faced with low commissions. Similarly, because nonselling activities by sales agents
in the field are especially hard to evaluate, the equal compensation principle suggests
that the commissions for sales by these agents should be set low to avoid inducing an
overemphasis on selling versus nonselling activities. Again, using weak incentives with
independent representatives risks their shifting effort to other fi rms' products.
A \'OIDINC MONOPOLY DISTORTIONS A principal cost of using an independent supplier
when the input market is not competitive is that the supplier may try to exercise
monopoly power, driving its price up above marginal cost. Figure 1 6. 3 recaps the
standard economic analysis, in which the higher price may lead to an inefficiently
low use of the input and a real loss of value. This outcome is not assured because
the parties may negotiate a price-quantity agreement between themselves to avoid
some of the inefficiency. In general, however, bargaining will be imperfect and some
value losses will remain.
Simply integrating into the supply business to eliminate the monopoly distortion
is hardly a costless solution. As indicated earlier, independent firms enjoy a number
of significant advantages. The buying firm could try to offset these advantages by
increasing its own scale and selling to others (as Lucky Goldstar did). Alternatively,
it could acquire the supplier and maintain its scale of operations. It could then use
an incentive contract to motivate the formerly independent manager, but this involves
incurring costs of the kind we analyzed in Chapters 7 and 8.

8 Erin Anderson, "The Salesperson as Outside Agent or Employee: A Transaction Cost Analysis, "
Managemen t Science, 4 (Summer 1985), 234-54.
 560
The Design and If both a firm and its independent supplier have some monopoly power, then
Dynamics of the tendency of each is to add a monopoly profit margin to its costs. This can lead
Organizations to an excessively high price for the buyer's output, a price even higher than the level
that would maximize the two firms' total profits. A vertically integrated firm that
sought to maximize its profits would charge a lower price, resulting in both higher
profits and happier consumers. This double marginalization is the same phenomenon
that we studied in transfer pricing, and we can illustrate it with the same example.

A J.Vumerical Emmple of Double

1lfargi11alizatio11
The data and calculations are essentially those summarized in Table 1 6. l ,
except that now the first column should be interpreted as the choices that would
be made in an integrated firm that maximized overall profits, whereas the second
and third columns refer, respectively, to the downstream and upstream firms
when there is no vertical integration. The (inverse) demand for the final product
is P = 1 0 - Q/ 1 6, where again Q is the number of units that will be purchased
if the price is P. The upstream firm produces the component needed by the
downstream firm at a marginal cost of $ 1 per unit, and it charges a price T for
it. Besides the price that it pays to the upstream firm, the downstream firm
incurs additional costs of $2 per unit that it makes. Thus, as in the transfer­
pricing discussion, the actual cost of making the product is $3 per unit, but the
downstream firm's margi nal cost is (2 + T).
If the two stages of production are combined under the control of a single
firm (without affecting the cost of production), the total variable cost of
production is 3Q, with marginal cost 3. The integrated monopolist's marginal
revenue is 1 0 - Q/8. Equating the marginal revenue to the marginal cost of 3
leads, as before, to the conclusion that the optimal quantity is Q = 56. Then,
from the demand equation, P = 6. 50.
Suppose the upstream sets a price of T for its component if the two stages
of production are not integrated. Then the downstream firm's cost per unit is 2
+ T, which is also its marginal cost. Equating this to the marginal revenue of
l 0 - Q/8 leads, as before, to Q = 8(8 - T) as the demand facing the upstream
firm. Solving for T, the inverse demand facing the supplier is T = 8 - Q/8.
Because the upstream firm's marginal cost is l , this leads to the conclusion that
T will be set so that Q equals 28-only half as much as the output of an
integrated monopolist. Substituting into the consumer demand function, this
quantity implies that the consumer price must be $8. 2 5, even higher than the
price charged by an integrated, profit-maximizing monopolist!
Note, however, that these calculations assume that the integrated firm is
a profit maximizer that somehow solves the transfer pricing problem. This is
not an innocuous assumption. Without it, the problem is not really solved by
integration: It is merely converted into a problem of transfer pricing.

C. \ PTL HI \C St 'Pl'LI EBs' RENTS There are various reasons that suppliers could earn
above-average profits in providing inputs to the buying firm. We discussed one
important reason in Chapter 8: It may not be possible to induce a supplier to sustain
the quality of its production without giving the supplier something to lose. That
something generally takes the form of a stream of extra earnings, or rents, from the
supply relationship. These rents imply that the price the firm pays for its inputs must
exceed the supplier's fu ll econom ic cost of production, including a normal return on
the su pplier's capital . Vertical integration pr ovides a wa y for the firm to ca pture some The Boundaries
or all of these re nts. and Structure of
For any of the reasons described earlier, it may not be possible for the firm to the Firm
match its su ppl ier's performa nce. The vertically integrated firm will need to provide
ince ntives to managers of the su pplying divisio n, who may then earn re nts as managers
rather than as owners. The firm could al leviate that by su bstituting mo nitoring for
rents, reducing total value but increa sing the firm's profits. Extra mo nitor ing is
sometimes easier to im pose on em ployees, over whom the em plo yer ha s discretio nary
authority, tha n on an indepe ndent su pplier. Even if the firm ca nnot qu ite match the
technica l efficiency of its formerl y independent su ppl ier, however, it may still be a ble
to pr oduce for itself at a lower c ost tha n the price it form erly paid. Note, of c ourse,
that there will be incentives to avoid value-reduc ing vertical integratio n that is a imed
solely .at capturing re nts. It may still happen, however, if contracting problems between
separate entities prevent costless reallocation of rents.
ENTHY DETERRENCE Sometimes an advantage of vertical integration from the narrow
point of view of the integrating firms is that it creates a barrier against fu ture e ntry by
firms that may wa nt to c om pete in the sam e market. This gain exists, however, only
provided that the entry would reduce the com bined profits of the two integrating
firms. For exam ple, if the vertical integration elim inates a n im porta nt potential source
of supply to a downstream c om petitor, then any new com petitor would ha ve to arra nge
an alternative source of su pply. There is no guara ntee that such a source would exist.
Even if it did, the problem of bringing the necessary production capacity on stream
in a pair of new enterprises is more difficult tha n entry at just one stage of the vertical
chain. That extra difficulty might deter the c om petitor from even trying.
The pr oviso that entry must reduce the com bined profits of the two inte grating
firm s is im porta nt, for otherwise the arra ngement would not be total value ma ximizing.
For example, suppose the potential entra nt is a com petitor for the downstream
compa ny. The expa nsion of the market that would likely follow entry would increase
demand for the input supplied by the u pstream firm a nd raise it s profi ts. If this market
expansion is expected to increa se the upstream firm's profits by m ore tha n what the
downstream firm loses, then the existing downstream firm will be u nable to pay a
high enough price to attract the u pstream fi rm to merge with it. 9

Alternative Veriical Relations

S im ple market procurem ent and vertical integration do not exhaust the im porta nt
options open to firms in a vertical su ppl y relationship. M odern firms co ntinue to forge
innovative orga nizational arra ngem ents in their attem pts to enjoy the incentive
advantages of independent firm s whil e still facilitating close pla nning where necessary,
pr otecting investments from hold-u ps, and avoiding monopol y inefficie ncies like that
of dou ble marginaliza tion. As firms cont inue to innovate in this arena, the set of
options will continue to e xpa nd.
We explore three possible options that represe nt orga nizational innovations of
different eras. The co-op form of orga nizat ion was a n outgr owth of the utopia n
experimental c ommu nities of the nineteenth century. Franchise reta iling bega n to
emerge as a n important form in the first half of this century. The last example is
the su ppl ier organization used by Japanese automo bile com pa nies, which 1 s an
organizational innovation and success story of the second half of this century.

9 The situation becomes more complicated if you consider the further possibility that the combined
firm� might negotiate with the entrant to buy its technology or its production services, but we omit that
possibility here.
 562
The Design and CooPERATl\'ES 10 Cooperatives (co- ops) are business organizations that are owned by
Dynamics of the indivi duals who transact with them-usually workers or customers, but occasionally
Organizations suppliers-and that are organized under specific legal rules. Indivi dual members'
voting power in co-op decisi ons is not determined by the amounts of capital they have
provided ( as it is in a corporati on with a one- share/one-vote rule) but instead on a
one-member/one-vote basis or, less often, in pr oporti on to the amount of business
done with the co- op. S imilarly, profits are distributed in the form of reduced prices
for whatever services the co- op supplies. Often, only members can transact with
the c o-op. (In the U nited States, to qualify legally as a c ooperative, at least half the
organization' s business must be sales to members.) Memberships are sold by the
organization only: There is no secondary market for them, and a member who quits
may receive nothing or may receive a return of his or her initial capital contributi on.
Co- ops play a significant role in some segments of modern ec onomies. Prominent
examples of worker co-ops include the diverse set of businesse s around the city of
Mondragon in Spain, large portions of the c onstruction industry in Italy, and the
maj or bus company in Israel. Many of the new businesses emergi ng in the S oviet
Union were, for legal reasons, structured as co-ops. In the U nited States, co-op book
stores serving college and university campuses supply about 10 percent of the overall
book market. In home hardware, retailer- owned supply cooperatives like True Value
and Ace account for half the U . S. market. Suppl y co-ops handle al most one third of
wholesali ng of groceries to U. S. retailers (excludi ng firms that are vertically integrated
between wholesale and retail). Associated Press, which is owned by the ne wspapers,
is one of two maj or international news services in the U nited States. In agriculture,
farm supply co-ops serve over 25 percent of the U . S. market and are especially
important i n manufacturing and distributi ng petroleum products, feed, fertilizers, and
chemicals. Rural electric power is also often supplied by co- ops. Cooperatives are also
active in marketing farm products. Among the best-known brands in the U nited States
are Land O'Lakes (dairy products), Blue Di amond (almonds), and Sunkist (citrus).
S imilarly, co-op grain elevators are a common sight across Western Canada. Fi nally,
consumer co- ops provide a variety of services, especially retailing food and groceries.
Why Are There Cooperatives? In the early days co- ops had an explicitly social and
ideological basis. More recently, gover nme nt policies to pr omote cooperatives by
provi ding cheap credit or tax advantages and the development of supporting institutions
are partial explanati ons of the popularity of this form of organization and of regi onal
differences in how it is used. They do not explai n why the form is more common in
some industries than in others, however. M onopoly power seems to provide such an
explanati on.
I n many cases of thriving supply cooperatives, ec onomies of scale make it
difficult for there to be more than one or two suppliers. Co- ops then sen·e to al leviate
problems of monopoly pricing. For example, college students buying textbooks have
a highly inelastic demand, making them especially vulnerable to high mark-ups at
local bookstores. Economies of scale in handl ing inventories of the many books
pr ofessors assign limit the number of competitors. In this case, a cooperative spons ored
by a group of activist stude nt<; can help to cap retail textbook prices, even if the not­
for-profit cooperative is not efficient enough to replace its for-profit competitors
completely. The coexiste nce of vertically integrated firms and supply c ooperatives
for small firms i n the grocery and home hardware industries suggests that both forms
arc ai med at solving the same problem-that of monopoly supply. Consumer-

JO Based in la rge part on l lenry Hansmann, "The Ownership of the Firm , " fournal of Law,
Economics, and Organization , 4 (Fall 1 988), 267- 304.
co-op grocery stores that compete with for-profit supermarkets have a similar The Boundaries
explanation. and Structure of
Besides alleviating monopoly power, some cooperatives have been formed to the Firm
create power for their members over consumers . For example, agricultural shipping
and marketing cooperatives in California have sometimes set quotas for members,
which have the effect of creating monopoly prices.
Generic Problems of Cooperatives. If the membership of the co-op is not too large
or heterogeneous, the ability of members to monitor the quality of co-op services and
their general agreement about what should be done can make this form of organization
quite effective. Nevertheless, cooperative organizations, especially large ones, do suffer
from a number of predictable organization problems.
With limited residual claimant status and no concentration of ownership,
individual co-op members may have little incentive to keep tabs on the co-op
management. If some members expect to participate in the co-op only in the short
term, they may not be inclined to vote for investments that benefit the co-op over a
longer horizon. Worse, they may promote projects that have short-term benefits but
tend to run down the co-op's capital. If the co-op's charter permits it to engage in a
wide range of activities, then diverse objectives among its members can lead to
damaging influence activities.
An instructive example is the Berkeley Co-op, a consumer cooperative in
Berkeley, California, which mainly operates grocery stores but has also sold gasoline,
auto supplies, prescription drugs, and clothing. Its membership in the 1 980s was quite
heterogeneous, reflecting a population that included college students and poor families
as well as very affluent individuals and families. This organization saw its very existence
threatened by a series of struggles over cross-subsidies among its activities. Should
there be low prices for basic necessities, to subsidize poor households? What about
upscale products to serve the desires of the wealthier clientele? Free in-store child
care? More locations for better accessibility in lower-income areas? These battles tied
up the organization and its management in political maneuvering while the stores
suffered large losses.
Similar problems would arise if the co-op approach to electric power distribution
that is successful in rural America were attempted in cities. The diversity of the set
of customers would cause problems in agreeing on the policies to be followed. How
should costs be allocated among industrial, commercial, or residential users? Should
rates climb rapidly with usage to encourage conservation or more slowly, so that those
stuck at home can afford to heat and cool their homes? Should rates be higher at
peak times for everyone or just for those identified as putting the extra demands on
the system? Rather than crippling decision making in the urban electric co-op by
fights over these questions, a pattern of investor ownership is used, with public utility
regulation to control the monopoly inefficiencies.
FRANCHISE RETA ILING In the United States, automobile dealerships, gas stations,
convenience stores, clothing stores, hotels, restaurants, tax preparation services, car
rentals, and banking are among the many businesses that are frequently operated on
the franchise system. By the mid- l 980s over 300, 000 establishments in the United
States operated as franchises, with an average of about half a million dollars in annual
sales each, 1 1 and North American firms were expanding their overseas franchising
rapidly: There are McDonald's outlets on the Champs-Elysees in Paris and Red Square
in Moscow, for example.

1 1 James
A. Brickley, Frederick H . Dark , and Michael S. Weisbach, "An Agency Perspective on
Franchising, " Financial Management, 20 (Spring 1991), 27.
I
 564
The Design and The franchisee owns and runs a retail business using the franchisor's brand
Dynamics of name, often buying inputs or goods for resale from the franchisor. The franchisor
Organizations collects fees and royalties from the franchisees for the use of the brand name and
commonly also pro\'ides training, advertising, and other services. Franchisors also
generally maintain rights to set and enforce standards on the franchisee. For example,
a clothing franchisor may require that the retail store operator maintain a specified
number of varieties or a minimum level of inventories. An oil company may require
a station operator to remain open for certain hours and days, even when the additional
hours are not profitable for the franchisee. Franchise restaurants must typically submit
to inspections, with penalties for failing to maintain cleanliness and food quality
standards. They may even have their franchise revoked if they refuse to acquiesce to
the franchisor's demands.
This kind of arrangement takes advantage of owner-operators' incentives to keep
costs low, attract customers, and care for the premises. In this respect, franchising
arrangements share the advantages of ordinary market arrangements. At the same
time, the control exerted by the franchisor adds value by overcoming a variety of
problems arising from specific assets, free riding by franchisees, and economies of
scale in marketing and perhaps in purchasing. When these additional sources of value
are important, franchise contracts can have important advantages over spot market
contracting.

Specific Assets and Incentives. Consider the case of McDonald's Corporation,

which serves customers in approximately 6, 900 restaurants in the United States and
hundreds more in other countries. Approximately 5 , 300 :t\1cDonald's restaurants in
the United States are run by franchisees who mrn and operate the individual
restaurants. In these arrangements, one specific asset is the restaurant building itself,
which is typically specially designed and equipped to be a McDonald's restaurant and
would be less valuable in other uses. In addition, the owner-manager makes a specific
investment in learning the company systems. At McDonald's, this is done partly by
working for a period without pay as an apprentice at another restaurant and partly by
attending Hamburger University, as the company calls its restaurant manager training
center. From McDonald 's point of view, the franchisee's expenditure of time and
money signals his or her commitment to the business. From the franchisee's
perspective, it is an investment that is vulnerable to hold-up.
The unusual architecture of the buildings in which McDonald's restaurants are
located would cause a potential hold-up problem if the franchisees were to own the
buildings. Thus, unlike many other franchisors, McDonald's owns the individual
structures. More generally, the potential for hold-ups by the franchisor is a serious
matter in many franchise businesses. A common complaint by franchisees is that,
after they have invested in developing a market, the franchisor holds them up by
establishing another (competing) franchise in the same market area. The franchisor
may claim that the market has now grown to be too large to be most profitably served
by a single franchise. Automobile franchisees have been particularly acti\·e in seeking
legislation to prevent such an encroachment on their markets, and franchisee
organizations often work with franchisors to establish mutually agreed standards for
how many franchises an area can support, in order to reduce this potential for conflict.
Like other franchisors, McDonald's values its brand name reputation, which in
this case promises customers clean premises and surroundings and hot, fresh food,
served quickly and cheerfully, as well as certain specific menu items, like the Big
�lac. The overall reputation is a valuable asset that a local franchisee could damage
by fa iling to li\·c up to the franch ise standards. In fact, the individual franchisee has
an incentive to free-ride on the general brand reputation, skimping on service and
 .565
quality, because all the cost savings accrue directly to th e in dividual franchisee, but The Bounda ries
the damage is spread am on g all the outlets and the franchiser. The rights of the and Structure of
fran chisor to in sist upon the franchisees' following certain procedures, to inspect to the Firm
en sure that these req uirem en ts are being m et, an d to exact pun ishmen ts (includin g
termin ation of the franchise) if they are n ot are importan t for man aging this incen tive
problem an d protecting the valuable asset.
Franchising has been less comm on in Europe than in the U nited S tates, and
the European Econ om ic Commun ity, as part of its efforts to establish un iform business
standards, has been hostile to franchisors' control of franch isee operations. In I 989,
however, M cDonald's won a rulin g that protected some of its rights to regulate
franchisees: "While M cDonald's Corporation m ay still tell its European fran chisees
how big a dollop of ketchup to put on each Big Mac, it can't make them sell Coca
Cola" in preference to other bran ds. 12 This rulin g reflects the European Commun ity's
attempt to balan ce the goals of effi cien t contractin g an d local control of European
businesses.
Coordinati on Issues. An other kin d of problem that franchise system s n eed to
an ticipate is the con flict that frequently accom pan ies chan ges in system s or in the
product lin e. In m an y franchise situation s, un iformity in what is offered at differen t
outlets is importan t fo r m arketin g: People value kn owin g that they can get what they
expect. If M cDonald's and its fran chisee group seek to reposition their image ar oun d
healthier foods, or Exxon an d its franchisees decide as a group to in crease their
advertisin g expen ditures or to build a reputation for clean bathrooms, there n eeds to
be a way to enforce these stan dards across all the outlets without in curring excessive
bargain in g costs. There can also be econ omies in havin g uniform operatin g system s.
For e xample, if Benn etton in troduces a new computerized inven tory trackin g system,
then the fixed costs of developing an d running the system an d the econ om ies of scale
achievable in implemen tin g it m ay imply that total value is m aximized when every
outlet participates in the chan ge. Such un iversal participation would be practically
im possible if it were necessary to n egotiate with the franchi sees in dividually about
their costs and benefits of impl emen tin g the system in order to subsidize those stores
for which conversion is especially costly. When these kin ds of chan ges are in troduced,
there is usually a need for close coordination an d a mechan ism to enforce decision s
on in dividual franchisees. These coordin ation problems arisin g out of the econ omies
of scale create an additional reason for giving the fran chisor the authority to in terven e
in the fran chisees' operation s.
Authority and Its Limits. Of course, fran chisees seek to protect them selves against
excessive deman ds or penalties imposed by the fran chisor, for example, by forming
association s of fran chisees to bargain with the franchisor. Because the parties cannot
foresee the circumstan ces an d proposals that will arise over time in these relation ships,
the processes by which decision s are to be made become the focus of the con tract.
Accordingly, franchise agreemen ts specify some of the rights an d respon sibilities of
the various parties in the con duct of their busin ess. Routines developed in the course
of doin g busin ess also provide guidan ce regardin g what is n ormal an d expected
behavior. Such arran gemen ts are fun damen tally differen t from the cash-for- product
con tracts that characterize sim ple m arket tran saction s.
SUPPLI EH RELATIONS IN TI IE JAPANESE AITTOMOBILE INDUSTRY In ordin ary marke ts,
the firms that are most successful in meeting customer needs at low cost are rewarded

12 Philip Revzin, "European Bureaucrats Are Writing the Rules Americans Will Live By, " The Wall
Street /ournal (May 1 7, 1989), l .
 566
The Design and
Dynam ics of
Organizations
Franchises or Company-Owned Outlets?
Typically, franchisors themselves own about 20 percent of the outlets, hiring
managers to run them. What factors should enter the decision between
establishing a particular unit as a franchise or keeping it company owned?
First, agency considerations suggest that the extent of franchising should
increase and company ownership of units decrease, where monitoring
managers is more difficult. Second, franchi ses should be more common
when repeat business at a particular outlet is greater because the free-rider
problem should be less acute. Third, franchising should be less common
where the capital requirements of the bu siness are higher because of the
increased costs of ineffi cient risk sharing. Fourth, as new laws and regulations
have made it more diffi cult for franchisors to terminate franchise agreements,
there should be a tendency in recent years to reduce the extent of franchising
relative to company ownership.
We have already cited evidence regarding the first of these in connection
with the greater frequency of franchising among gas stations that provide
repair services versus those that do not. M ore generally, recent empi rical
work tends to bear out all three of these predictions.
One puzzle appears in these studies, however. We would expect from
free- rider considerations that units located along freeways would be more
likely to be company owned because they should get less repeat business, but
this does not appear to be the case. We know of no satisfactory explanation
for this.

See James A. Brickley, Frederick H . Dark, and Michael S. Weisbach, "An Agency
Perspective on Franchising," Financial Managemen t, 20 (Spring 1991), 27- 3 5, and the
references given there for additional details.

with increased sales. The North American automobile manufacturers have traditionally
tried to take advantage of these market incentives by selecting suppliers by competitive
bidding. Where specific assets were involved or where few quali fied bidders could be
found, the usual solution was to arrange for the specific assets to be owned by the
automobile company, often through a pr ocess of vertical integration.
How U.S. and Japanese Practices Differ. Japanese automobile companies have
been innovative in creating ways to achieve the advantages of market incentives even
where specific investments are required for efficient production. As we described in
Chapter 9, the Japanese auto firms have been much less vertically integrated than the
American ones and have instead purchased fully assembled systems from a relatively
small number of suppliers. For example, General Motors used some 35,000 different
suppliers in 1986, whereas i n th at same year Toyota made virtually all of i ts purchases
from its core group of 224 suppli ers, a rati o of more than 150 to I . J' l The Japanese
and U . S. figures are not precisely comparable, however, because of differences in the
defi nitio n of "supplier" used in data sources. Excludi ng GM's smallest suppliers closes

n Tltesc data and those that follow in this section arc drawn from Banri Asanuma, "Japanese
Manufacturer-Supplier Relationships in International Perspective: The Automobile Case, " Working Paper
No. 8, Faculty of Economics, Kyoto University ( 1 988).
the gap somewhat but still leaves a remarkable difference between the companies: The Boundaries
General Motors' 5, 500 largest suppliers accounted for only about 80 percent of that and Structure of
company's purchases. By any reasonable calculation, Toyota relied on a far smaller the Firm
number of suppliers than did CM. Toyota's practices set the pattern for other Japanese
automobile compan ies .
Japanese management of supplier relations differed from U. S. practices in other
ways, as well. The Japanese made much less extensive use of competitive bidding to
obtain low prices from their suppl iers. Instead, suppliers were evaluated according to
how well they had performed on earlier contracts, with the top performers being
rewarded with additional orders. This reward system leads to long-term relationships,
wh ich in fact characterized the Japanese pattern . Systems based on competitive bidd ing
typically require that the bidders know in advance exactly what is to . be supplied, so
contracts can be bid only after detailed drawings of the desired parts are prepared.
Because they did not rely on competitive bidding, Japanese companies could select
suppliers bef�re the parts specifications were made final , making it possible to exploit
the suppliers' expertise in design engineering, to design parts to fit the capabilities of
the suppliers' existing equipment, and to allow suppliers more time to plan and prepare
for production . The most reliable and technically sophisticated suppliers could even
be trusted to design their own parts completely, subject to general design specifications
provided by the automobile manufacturer.

The Performance Appraisal Problem. A system like this would appear to have two
important potential disadvantages . The first concerns performance appraisal. Without
a competitive bidding stage, how does the automobile maker know how much the
input should cost to produce? Were the supplier's costs really as low as they should
be? Was its quality as high? Could another supplier have produced the same component
or system with better qual ity and at a lower cost? The second problem is one of
specific assets. Once the firm and its supplier have committed themselves to a contract
in which, say, headlamps for Toyota Corollas are to be suppl ied, how is the price to

sales of the car are disappointing? If the company's own work force is not fully
be determined? Who determines the delivery schedule? Can orders be canceled if

employed, can it choose to take over production of headlamps to ensure that its
employees are kept busy?
To manage the first problem , Toyota has a two supplier policy: Except for

for each category of component. If supplier A makes headlamps for model X, then a
products produced exclusively in house, there must always be at least two suppl iers

different suppl ier B will make the headlamps for some other model Y. Having just
one supplier for the headlamps of a single model allows the supplier to take full
advantage of economies of scale in making the lamp. Guaranteeing the supplier that
it will continue to make the part throughout the life of the model helps to protect
any specific investment the suppl ier makes. Having a different supplier for the other
model allows Toyota to use comparative performance evaluation to assess how well
each suppl ier is performing. The reward for good performance in making headlamps

on two different models (provided the two-supplier policy is not violated). If the
may be contracts to make headlamps and taillamps on the next model , or headlamps

supplier demonstrates technical prowess, the reward may be an upgrading that qualifies
the suppl ier to make more complex parts.
Because suppl iers make complete systems, it is practical to hold them responsible
when a system fails, without endless arguments about which supplier is responsible.
The system maker must accept responsibility for the cost of any necessary repairs.
Th is practice also encourages suppliers and the core firm to cooperate in identifying
problems and fixing them.

L
 The Hold-Up Problem.
568
The Design and The second kind of problem of determining prices and
Dynamics of avoiding hold-ups is less easily resolved. The long-term, mutually beneficial nature
Organizations of the relationship does help to mitigate this problem by providing both sides with an
incentive to find mutually acceptable terms. Even though they do foster goodwill,
long-term relationships cannot themselves determine prices or resolve real differences
of opinion. Initial prices can be set in light of the planned cost of production, based
on the company's experience with other similar parts. Profits are then based on the
supplier's ability to produce at a lower cost than previous suppliers, by seeking
improvements in their production systems. The supplier earns additional points in
the supplier rating system for codifying methods so that they can be used by other
suppliers, leading to lower costs for the core firm. The objective is continuous
improvement in productivity by all of the core firm's suppliers.
A complex relationship such as this holds opportunities for hold-ups besides
price renegotiation. For example, a supplier may fear not getting proper credit for its
innovations, which are transferred to competing suppliers. Resolving the ever-changing
issues in the relationship between the firm and its suppliers appears to be one of the

an association for parts suppliers (Kyohokai) and for suppliers of tools, equipment, and
roles of the supplier associations. In this regard, it is illuminating that Toyota maintains

construction services (Eihokai), but no associations for the supply of basic materials
like oil or steel, where specific investments are much less important. A similar pattern
is found at other Japanese automobile companies.
Advantages. This system of managed supplier organization has formidable advan­
tages over the system of competitive bidding among independent suppliers. It allows
the core firm to assess supplier performance accurately, to ensure that the suppliers
with the best performance records get most of the contracts for the next car models,
to take advantage of suppliers' design capabilities and the peculiarities of their
equipment, and to allow longer lead times for design and planning. Contracts
extending over the life of a car model promote cost-reducing specific investments. The
system encourages suppliers to be innovative but also promotes rapid communication of
valuable information among competing suppliers-an achievement that would not be
possible with more traditional supplier relationships.
The more relevant comparison, however, is not with simple market transactions,
which are unlikely to be viable when there are the kinds of advantages that exist in
the auto industry to making specific investments. Rather, the issue is how the Japanese
system compares with a system of vertical integration used in North America. Both
systems allow for creation and protection of specific assets, with the Japanese relying
on reputations and repeated dealings rather than ownership. 14 One significant
advantage of the Japanese system is that it may be easier to break relations with a
supplier who fails to perform than it is to stop obtaining supplies from a badly
performing division. The influence costs that would be involved in closing a division
and going to outside suppliers, or even in replacing the managers of the division,
m ight be substantial. The two-supplier system potentially sacrifices some economies
of scale, but this is not a great drawback if the minimum efficient scale is not too
large relative to demand. In return, it allows comparative performance evaluation and
creates competition that keeps prices low and quality high. It would be possible, in

H There are occasional complaints that the major Japanese firms hold up their suppl iers, however.
American firms that have obtained supply contracts have alleged this: See, for example, Dana l\1ilbank,
'"Culture Clash: Making Honda Parts, Ohio Company Finds, Can Be Road to Ruin," The \Vall Street
Journal (October 5, I 990), A I . T. Boone Pickens has claimed that Toyota forces unfairly low prices on
Koito, an auto-light suppl ier in the Toyota group of which Pickens was, for a while, the major shareholder:
Sec T. Boone Pickens, "Pickens to Toyota: I Give Up, " San Jose l\tercury News (April 30, 1 99 1 ), 7B.
 569
theory, to have competing internal supply divisions, but this seems difficult i 1 1 practice,
again for influence cost reasons. Also, because at least the independent suppliers
The Boundaries
and Structure of
(those that are not members of the suppl iers' associations of any of the major auto the Firm
manufacturers) can work for any of the large firms, and even members of Nissan's
group may supply parts to Toyota, there is flexibility in the system to respond to
shifting demands. It would likely be more difficult for CM to act as a suppl ier to Ford
than it is for a separate company to supply either or both as needs dictate.

HORIZONTAL SCOPE AND STRUCTURE

The meaning of the term vertical structure is clear in the case of the process depicted
in Figure 1 6. 2, where a single, stable product is processed through a fixed series of
steps on its way to the final customer. The issue is how the transactions between
successive stages are managed: through spot market transactions, through administrative
di rection after integrating both stages within a si ngle firm, or through a more complex
institutionai arrangement with a specially crafted governance structure for making
decisions and resolving disputes? The choices offered by this vision of the firm are
useful, but they are far too simple and narrow to describe the scope of modern busi ness
organizations.

Competitive Strategy and Organizational Innovation

Recall the quotation describing the development of Lucky Coldstar. U nable to find
caps of adequate quality for its jars of cosmetic cream, Lucky (as it was then called)
decided to make them for itself. In our simple concepti on of the firm, the decision
was one of vertical integration , and it was arguably the right choice. Presumably
Lucky might have been able to arrange for some other Korean firm to develop the
expertise and make the physical investments to meet its needs. However, with only
Lucky demanding such high-quality plastics, the investments would be specific, at
least unti l a larger market developed. There would thus be a bilateral monopoly supply
relation between Lucky and its supplier, and this favored vertical integration . On the
negative side, however, the scale of the molding operation would be too small to
enable production at an efficient scale; it might be cheaper to rely on a foreign firm
that could produce at an efficient scale, provided the South Korean government would
allow such an import.
Some of the problem of scale could be overcome by making other molded
plastic products: toothbrushes, combs , and soap boxes, for example. Because Lucky
toothpaste had a virtual monopoly in the toothpaste market, the firm already had a
distribution system that could handle these new products. The resulting expansion
into plastic molding was thus both vertical-into a new stage of cosmetic cream
busi ness-and horizontal-into a new set of consumer products. 1 5
I n plastics, Lucky sought to expand its market to achieve the advantages of scale.
It grew by making electric fan blades and telephone cases . It increased the demand
for those products by enteri ng the electrical and telecommunications equipment
busi nesses as well . This might be thought of as a vertical move into another stage of
the production of fans and phones, but it is probably more useful to think of it as
another horizontal extension.
The field of corporate strategy or competitive strategy focuses on one of the
questions that the Lucky Coldstar story raises: What busi nesses should a company be

15 Our use of the term horizontal conforms to the use in business writi ng, where it refers to activities
that are not vertically related . In the literature of antitrust law and economics, the term horizontal merger
is used to mean something different, specifically, merger between two firms in the same market.

l
 570
The Design and in? To put the same question another way, what should be the horizontal boundaries
Dynamics of and scope of the firm? To address this question broadly would take us much too far
Organizations afield. However, one aspect of it is central to our concerns: How does the extent of
activities that a firm undertakes affect the way it structures itself, and how do
the mechanisms it puts in place for information gathering, decision making,
implementation, and evaluation affect the range of activities that it can successfully
undertake?

Directions of Divisional Expansion

As several of the examples in this chapter suggest, the direction of business expansion
has often been determined by considerations of scale, scope, and core competencies.
The Lucky Goldstar story is one of (somewhat opportunistic and even haphazard)
expansion through effective exploitation of complementarities and economies of scale
and scope. Firms often avoid expansion even into related businesses where these kinds
of economies are absent. For example, Sears, with no special competence in
manufacturing, relied primarily on independent suppliers to make its store brands. As
argued previously, there are real advantages in relying on independent suppliers,
especially when there are several competitive ones and the product being purchased
is a standard one, like hammers or tires. More recently Sears has expanded into retail
financial services, including insurance, real estate, and credit card businesses. It
gambled that it could use its retailing experience to sell those services like other
consumer products, but it has so far found only meager success.
The takeover by Electronic Data Service (EDS) of Continental Airlines' airline
information systems business (see earlier box) illustrates a horizontal expansion by
EDS. Here, EDS's bet is that the competencies it has developed by managing
computerized data bases in energy and banking give it an advantage managing a
system for airlines as well. Another interesting example is the Sony Corporation's
purchase of CBS Records and Columbia Pictures. Sony's videocassettes in the
techr.ologically superior Beta format lost the competition to Matsushita's VHS format
partly because rental movies were initially less available in the Beta format. Sony's
digital audio tape (DAT) format for sound recording and reproduction, introduced in
1 987, was opposed by the record companies and achieved only limited market
penetration. With its new purchases, Sony hoped to ensure coordination between any
new entertainment hardware technology that it may introduce and the development
of entertainment software in a corresponding format. For example, the former CBS
Records was expected to offer music in Sony's Mini Disc (MD) format as soon as the
format became commercially available, and Columbia was to produce feature
entertainment using Sony's High Definition Video Systems.

CORE Col\1PETENCIES AND BUSINESS STRATEGY Generally, firms seeking profitable

expansion are most likely to be successful by identifying the areas of their own special
competencies, investing to build those competencies, and introducing products where
the competencies give them a cost advantage or a marketing advantage. Often, this
involves a measure of guesswork. Sears guessed that many of its department store
customers would want to buy financial services in the same retail stores where they
bought pajamas and bicycles; Continental Airlines initially guessed that its status as a
major airline with a leading computer reservations system gave it a competency in
providing information services for airlines; GM guessed that the marketing and
manufacturing skills it honed in making and selling automobiles would be valuable
in making other consumer durables, like refrigerators and air conditioners; Walt
Disney guessed that the characters it successfully created for movies could be the basis
for a successful set of theme parks.
 fi7 1
Evid ently, not all such guesses work out well . Sometim es, even if the con cept The Boundaries
is correct, a strategic expansion may fail because the timing is wrong. For example, and Structure of
si nce the early 1980s, it has been widely believed that the marriage of data tran sfer the Firm
and data-processi ng technologies is inevita ble. The extensi ve use of networki ng among
individual com puter work stations ill ustrates the close con nections between these ki nds
of technologi es. Some of the largest com puter and telecommunications com panies
have in vested heavi ly i n this vi sio n of the future, suffering large losses, and many
conti nue to make new investments pursuing thi s vision. For example, IBM entered
telecomm uni cations in the 1970s by in vesti ng in Satellite Busi nes s Systems, purchased
16 percent ownershi p in MCI in the early 1980s, and then entered a joint ve nture
with Canada's Mitel Corporation. In 1984 IBM acquired Rolm Corporation , a ra pidly
growi ng and innovative maker of telephone switchi ng equi pment, for $1. 5 bi llion.
After pumpi ng nearly another $1 billion into Rohn to cover losses and pay for new
inve�tments, IBM sold Rolm's manufacturing operation and half of its marketi ng
operation tb Si emens just four years later. IBM is not alone in these strategic
i nvestm ent attempts. Japan's NEC has invested sim ilarly and, in 1991, the U . S.
telecomm uni cations giant AT&T purchased NCR Corporation, a computer ma ker,
to build in a competency in this technology that AT&T sees as key to future competiti ve
success.

Disadvantages of Horizontal Expansion

As several of these examples illustrate, some mergers and horizonta l extensions do
fail. What are the disad vantages that attend integration? Gi ven the multidivisiona l
form, it mi ght seem that a large organization can d o anything a sma ll firm co uld do
by just duplicati ng the style, incenti ves, and authority structure that are present i n
the small firm withi n a small division. I t mi ght even d o better because the small
division can call upon the resources of a large corporation i n times of growth­
resources that may be unavaila ble to a small, i ndepend ent company. The debate
about what costs are incurred by horizontal integration is one of the most acti ve in
modern eco nom ic theory.
PROBLEMS OF INFORl\tATION AND COORDINATION One i mportant effect is that an
organization that gets too large tends to suffer from poorer coord i nation and decision
making. As the size of the firm increases, either each divi sion becomes bi gger or the
num ber of di visions is increased. In the latter ca se, the amount of i nformation comi ng
in from the divisions to the head office increases. Top managers become overloaded.
They may respond by adding staff to support them or by further decentralizi ng,
pushing more decisions to lower levels and perha ps adding levels of management
between the top executives and the di visions, but both these responses bri ng costs.
Providing staff with the proper i ncentives i s especia lly hard because the ir performance
is very difficult to mea sure. Decentrali zing more decisions means sacrificing control
and coordi nation across these decisions, and the add itiona l layers of ma nagement
slow decision maki ng, act as filters in information transmi ssion, and are themselves
d irectly costly. If the growth comes with a gi ven di visional structure, the individ ua l
di vi sions start to become too bi g to manage. Then they i n turn must be decentra lized,
again creating extra layers of management and the attendant probl ems.
Expansion i nto unrelated fields creates additional pro blems because central
decision ma kers cannot become or stay adequately i nformed a bout the peculiariti es
of the technologi es of the i ndivid ual businesses and the markets i n whi ch they operate.
Thi s means that they ha ve more difficulty evaluati ng the performa nce of different
divisions and the proposals they make for new i nvestments. More and more reliance
is necessarily placed on fina ncial data in making decisions, and i nadequate attention
 572
The Design and
Dynamics of
Organ izations
The Failed Merger of Rolm and IBM
Rolm Corporation epitomized the small, fast-growing companies of Silicon
Valley, projecting a relaxed corporate culture but maintaining a sense of
mission. There were no dress codes or fixed working hours, but the parking
lot would often be full at 6:00 on Fridays and on weekends. The company
had a gym and swimming pool that employees could use in the middle of
the day, and lunchtime barbecues were common. Stock options for key
employees allowed some to get rich as the value of Rolm stock rose, and a
2 percent rate of commission on top of a $24,000 salary in the mid- 1 980s
allowed typical salespeople to earn $60, 000 per year.
To maintain the small company atmosphere, even as it grew, Rolm
subdivided its operations into many divisions, with separate divisions selling
to specific markets like banking, education, government, health care, hotel/
motel, and retail. Another division handled distribution to phone companies
and interconnect companies. The manufacturing operations were also
subdivided. A set of four planning groups were set up to ease coordination
among the groups in the areas of strategic planning, product maintenance,
product development, international sales, and procurement.
Rolm's principal competitors were the telecommunications giants
AT&T and Northern Telecom, which were able to devote more resources
to product development than Rolm could generate. It therefore encouraged
IBM to invest in its operations, and eventually to become full owner of the
company. IBM declared it would exercise a hands-off policy in managing
the company in order to allow it to maintain its entrepreneurial spirit.
Problems soon surfaced, however.
Organized along regional lines, Rolm's sales force would sometimes
offer different deals to the same corporate customer in different regions of
the country. IBM, which sold using uniform terms nationally, found itself
subjected to pressures from its customers to offer uniform terms in its
communications products, too. Succumbing to the pressure, it tried to enlist
Rolm salespeople to participate in national marketing, which created pressure
on the compensation scheme. Were Rolm salespeople to be paid a higher
commission than IBM people even when selling to the same clients?
Moreover, national sales are more complex and take more time to close than
do sales to smaller regional clients, threatening to reduce the incomes of
Rolm salespeople in the short term. IBM responded by raising base salaries
and reducing the rate of commission. The lower commission rates had the
largest effect on the best salespeople, leading some to quit.
It was not only in sales that IBM dampened Rolm's incentives. Rolm
engineers, previously motivated by stock options in Rolm Corporation, were
now given stock options in IBM instead. Although their successes in
developing a new produci could make a substantial difference in Rolm's
fortunes and the price of its stock, nothing they could do would have much
effect on the value of IBM stock, so performance incenti\'es were muted.
By January 1 986 Rohn co-founder Kenneth Oshman had quit, triggering
the departure of other key executives. With the passage of time, IBl\I
executives gained increasing control of Rolm's operations, trying to stem the
tide of losses the company was incurring. Finally, in 1 988, IBM sold Rolm's
 The Boundaries
and Structure of
the Firm
manufacturing divisions and a 50 percent interest in the marketing divisions
to Siemens.
Even this did not end the problems. Rolm Systems (the manufacturing
arm) and Rolm Company (the marketing arm) battled over whether the
salespeople should be allowed to be full-service suppliers, carrying the products
of other companies where Rolm Systems product line was incomplete. The
manufacturing company wanted the marketing company to keep it informed
of product developments in advance, so it could design and make products
to meet all its customers' demands . The marketing company believed it could
perform best if it could serve its customers' complete needs, even if that
sometimes meant selling a Rolm competitor's product.

is given to important but nonquantifiable information. Excessive reliance on accounting

numbers is often cited as a problem at General Motors.
Another problem is that performance measurement may become more costly as
the scope of the organization grows. Consider two separate firms of comparable size,
each big enough to have access to the equity markets. Investors will follow both firms,
and the prices they put on the firms' shares will reflect their estimates of the earnings
that will be generated by their managers and the strategies they adopt. The performance
information in the stock price is free to the firms' owners. If the two companies merge,
then there is only one share price, and the separate information about performance
of the constituent parts is lost. Moreover, there will now be only one set of financial
accounting reports rather than two, so again less i nformation is available for judging
performance.
INFLUENCE COSTS Influence costs arise from attempts to reallocate and protect rents
and quasi-rents within the organization. Presumably, the larger the organization, the
larger are the rents in total and the more people competing for them. If the bigger
prizes and greater numbers of contestants result in greater competition for these rents,
a larger firm will suffer more than proportionally higher influence costs. The following
example is too simplified to be realistic, but it does show how these effects can come
about.

A Mathematical Example of Competition for

Rents
Suppose the rents that can be reallocated through influence are a fu nction P(n)
of firm size n, where we measure size by the number of employees. Consider
for simplicity a winner-take-all contest, in which the successful person gets the
whole prize P(n). This might represent the political aspects of the competition
to become CEO, with P(n) representing the extra income, perks, and status that
accrue to being the head of a company of size n. Empirically, it appears that
all of these are increasing in n. Individual contestant i can expend resources s
at a personal cost cs on influence, and if the different contestants i = I , 2,
. . . , n each expend si , then the chance of any one of them winning is
s/S, where S is the resources expended by everyone in total . To decide how
hard to try to win, individual i considers the extra costs of increasing s i , which
 574
The Design and are c, and equates this to the increase in the benefits from increasing the chance
Dynamics of of winning the prize. This can be computed to be {(S - sYS 2}P(n). 16 If each
Organizations contestant forecasts correctly how hard the others will try (so that each correctly
forecasts the value of S - sJ and chooses his or her expenditure accordingly,
then they will all choose a common level s(n) which satisfies the equation
(n - l )s(n)
P(n) = c
n 2(s(n))2
or, {(n - 1 )/n}P(n) = ncs(n). The right-hand side of this expression is the total
cost of the influence activities to the contestants. The left-hand side approaches
P(n) as n becomes large. Thus, even if the prize does not grow as the size of
the firm does, the total influence costs will. Moreover, if P(n) also grows with
n, then the costs increase more than proportionally with the size of the prize.
In the limit, the entire rents are absorbed in influence costs. Of course, if the
contestants can use firm resources rather than their own in the contest, they
may well value them at much less than they are worth to the organization. In
this case, the personal cost cs may be much less than the actual cost, and the
resources spent may actually exceed the value of the prize!

SPECIAL PROBLEMS OF ACQUI SITIONS AND MERCERS A firm that has decided to expand
its scope has two choices: to develop the business itself from scratch, hiring new
people, investing in new capacity, and developing new suppliers and new distribution
channels; or to acquire another company already in the business, along with the
expertise its employees already have and the capacity and relationships it has already
developed. The latter may seem appealing, but there are special costs incurred when
an independent organization becomes integrated into a larger firm that arise to a
smaller degree, or not at all, in transactions between the independent entities. The
costs have been described in various ways by different observers. There may be conflicts
of corporate cultures, political battles leading to influence costs, or misbehavior by the
central office of the acquiring firm, which reneges on promises made to managers and
employees of the formerly independent subsidiary. Because all these occur together
in most of the cited examples, we shall not try to distinguish sharply among them.
These problems can arise when similar-sized organizations are brought together.
Shearson American Express experienced great difficulties integrating the investment
bankers from Lehman Brothers Kuhn Loeb into its operations, which were primarily
in the retail stock brokerage business. As well, computer maker U nisys has experienced
major organizational difficulties since its formation in a merger of Burroughs and
Sperry. Part of the trouble is attributed to differences in styles between the organizations.
Burroughs' formal, no-nonsense style contrasted with Sperry's relaxed one, while
the cost-conscious retail brokers at Shearson were often appalled by the lavish style of
the investment bankers of Kuhn Loeb. In each case, many employees had chosen the
work environment they preferred and resisted attempts at change.
The problems of conflicting cultures seem most intense, however, when a large
company takes over a smaller one in hopes of taking advantage of complementarities
in their technological capabilities and the smaller firm's entrepreneurial flair. We saw
examples of this in earlier chapters, particularly Tenneco's acquisition of Houston Oil
and Minerals in Chapter 6, and in the boxed accounts in this chapter of IBM's
purchase of Rolin and GM's purchase of EDS. In each case, the acquiring firm

16 The calculation is done as follows. Let S _ 1 be the resources spent by e\'eryone except indi\'idual
i. Then i's payoff is s,P(n)l(s, + S _ ,). The expression in the text is the derivative of this with respect to s,.
 575
promised to keep the smaller one independent, with its original manageme nt, The Boundaries
procedures, and style. In each case, this proved impossible. and Structure of
The smaller, e ntrepreneurial firms were organizations that emphasized and the Fi rm
rewarded initiative with large bonuses. H igh performance led to high levels of pay,
and poor performers dropped by the wayside. These firms did not specialize in activities
that called for reliable, routine service, and they did not reward reliable, routine
performers. Their key employees were risk takers, gambling on new technologies, new
industries to service, and new oil fields.
Once integrated into a larger organization, the compensation and perquisites
accorded employees of the entrepreneurial units were regarded as special deals and
became a source of j ealousies. GM managers were angry about the bonuses paid to
EDS employees for cost reducti ons. "If you paid me that much and allowed me that
much di scretion and staff resources to use in cutting costs, I'd find you even greater
savings." IBM employees were jealous of the commissions and bonuses paid to their
Rolm counterparts. At Tenneco, the pressure to maintain pay equity throughout the
company forced the firm to integrate H ouston' s operations with its own, destroying
the unique identity of the H ouston Oil unit. S ony Corporation executives report
feeling similar pressures over the salaries and bonuses paid to executives in their
entertainment business subsidiaries, but are determined to resist them.
As long as the central office of the firm maintains some control over its divisions,
the political pressures within the organization to equalize pay and opportunities will
be large. In purely economic terms, unequal treatment leads employees to divert their
attention from productive work, instead conniving ways to get the good j obs for
themselves. Of course, there are other costs-not all economic-to having to deal
with angry and j ealous employees.
A different set of issues arose in the recent conflict within Sears over its j oint
venture with IBM, the Prodigy system. Prodigy is a network system for home computer
users, which incorporates electronic games, mail, news, shopping services, and more.
U sers connect their home computers to the Prodigy system using ordinary telephone
lines. Among the services offered by Prodigy is a discount brokerage service called
Personal Control, in which investors manage their own portfolios, transmitting their
own buy-and- sell orders electronically to Personal Control, which is managed by
D onaldson, Lufkin and Jenrette. A problem for S ears is that the commissions charged
by Personal Control are approximately 75 percent below those charged by Sears's own
Dean Witter Reynolds brokerage unit. Dean Witter is a full- service brokerage house,
and its fees must cover its cost for research, sales, and order entry-costs that are
avoided by the Personal Control system. If Sears had no control over Prodigy, Dean
Witter employees could hardly complain about its fee structure and would spend their
energies competing as effectively as possible. Because Prodigy is partially owned by
S ears, however, considerable anger and energy are channeled within the S ears
organization, and the trust between S ears management and its subsidiary' s employees
is undermined.
Effective legal boundaries between firms can avoid attempts at influence within
the firm, limit top management's ability to renege on agreements, and isolate the
conflicting cultures of different units. These can be important advantages of independ­
ent firms.
BUSIN ESS A LLIANCES

In our discussion of vertical structure, we have already n oted that there are many
more alternatives than j ust simple market procurement and full vertical integration.
The nature of links between firms can be subtle and complex, and the way the
transaction is managed depends on all the key dimensions of the transaction (see
 576
The Design and
Dynamics of
Organizations
General Motors and EDS
When General M otors purchased Ross Perot's Electronic Data S ervices (EDS)
in June 1984 for $2. 5 5 billion in cash and stock, expectations were high on
both sides. EDS was an entreprene urial company that had thrived by taking
over clients' computer operations and promising to offer be tter service at
lower cost than the clients could do for themselves. Because EDS' s own
profi ts depended on its employees' ability to gather business and cut costs, it
rewarded them directly for doing that, paying bonuses i n the form of EDS
stock for outstandi ng performance. Paying bonuses in stock also gave the top
performers a vested interest in the success of the company.
The giant automobile manufacturer under Roger Smith had dedicated
itself to the introduction of high technology in its operations, i ntending to
lead the world wi th its computer-integrated manufacturing and electronic
order processing technologies. What be tter way to begin than by purchasing
EDS?
CM was committed not to tamper with EDS's style of doing business.
Within CM, EDS was to be an agent of change, introducing new computer
technology where it saw fit and collecting a fraction of the cost savings as
EDS earn ings. EDS employees would still be rewarded with EDS stock, but
now that EDS stock would be listed as a special " class E" General M otors
stock. At the same time, EDS founder Ross Perot would become a maj or
CM shareholder and a member of the board of directors, bringing hi s valuable
computer experience to CM's inner circle.
In 1985, largely as a result of its work with CM, EDS revenues surged
by 400 percent. As CM made more money available for computer services,
EDS employees took credit for the benefits of the improved systems, to the
chagrin of CM's managers. Disputes arose over how much of the savings
that EDS employees claimed they achieved were really attri butable to them.
Meanwhile, bitter arguments arose between Ross Perot and Roger Smith,
who Perot believed was reneging on his commi tments to EDS' s people. By
December of the following year, CM purchased Ross Perot's shares for $750
million and Perot parted company with CM .

Source: Doron Levin, Irreconcilable Differences: Ross Perot versus General l\fotors (New
York: Penguin Books, 1 989).

Chapter 2). Various special arrangements, especially nonexclusi ve ones, between

separately owned companies are called business alliances. The arrangements between
IBM and Mitsubishi and between Rolm and GTE discussed in boxes in this chapter
are examples, as is the alliance between IBM and Apple discussed in Chapter 9.
The use of business alliances varies greatly among firms. General M otors, for
example, has used alliances to transform its busine ss. In the l 960s CM had a vertically
integrated organization that tried to be self-reliant. By the late 1980s it had developed
an exte nsive network of relati onships with automobile companies and suppliers in
Europe and Asia as well as North America. Some alliances were to improve foreign
marketing; others involved sharing tech nologies for such things as robotics to be used
in automob ile ma nufacturing; still others were to add models to the weakest parts of
 !'i77
The Boundaries
and Structure of
the Firm
Mitsubishi Electric Selling IBM Computers
IBM Corpora tion ha s tradi tionally reli ed on i ts own sales force for all i ts sales
of large business computers. H aving a full li ne of computers from persona l
computers thro ugh large mainframe computers and networks to hook them
all together, IBM's strategy was to provide products to serve compa nies of all
sizes relyi ng exclusively on i ts own products. Sales representatives from
ma nufacturers who do not make large mainframe computers are put at a
di sadvantage by this strategy, because they cannot gene rate as much revenue
per visit as a n IBM representative. So i t was news when i n 199 1 IBM agreed
to allow Mi tsubi shi Electri c Corporati on to sell an IBM mai nframe computer
under its own name. The machines bei ng sold cost between $1 million a nd
$5 million, and Mi tsubi shi hoped to sell 900 of them during the three years
after the agreement.
Accordi ng to theory, a deal li ke this should occur only if the two
compani es ca n crea te more total value combi ned tha n they could achieve
separately, tha t i s, if the two can supply complementary resources. What
mi ght these be?
IBM , the world's largest computer maker, has been tryi ng hard to
i ncrea se i ts sales i n Ja pan. Its 1990 Japanese revenues were 1. 3 tri lli on yen,
or about $9. 6 bi llion. It faced resi stance i n expa ndi ng i ts sales i nto some
segments of Ja panese i ndustry, however. Mi tsubishi Electri c, a core member
of the Mi tsubi shi keiretsu (group of firms), is the fifth largest computer vendor
i n Japan, and lacks the scale necessary to make all its own machi nes at low
cost. It was i n danger of falli ng behi nd its larger Japa nese competitors because
of its i nability to supply a complete li ne of computers. The M i tsubi shi keiretsu
is the largest i ndustrial group i n Japa n, which allows Mi tsubishi Electric
salespeople to open doors that had been clo sed to IBM. If this a llia nce i s
moderately successfu l, it could benefit both IBM a nd Mi tsubishi Electric. If
it is very successful, i t may help Mitsubishi Electri c to bui ld i ts own
mai nframes, improvi ng i ts ability to compete directly with IBM in the future.

Source: Andrew Pollack, "IBM Model to Be Sold by Mitsubishi," The New York Times
(April 29, 1 99 1 ), 1 7.

GM's product li ne. A notable example is GM's joint venture with Toyota, establi shi ng
New U ni ted Motors Manufacturi ng, Incorporated (N UMMI). In this venture, a n old
GM automobil e pla nt i n Fremo nt, California became a showcase for Ja panese
automobile ma nufacturi ng methods a pplied i n a North America n environment (see
the box entitled "New U ni ted Motors Ma nufacturi ng, Incorporated"). U nli ke its
i ntegration deci sions i n which GM purchased a n ownershi p share i n suppliers or other
automobile compa nies, this joint venture was explicitly a temporary allia nce, during
which GM hoped to learn about Toyota's production methods a nd To yota hoped to
establi sh a production foothold i n its North America n market a nd learn about doi ng
busi ness i n the U nited States.
Busi ness allia nces often arise when each partici pa ti ng compa ny ha s some special
resource or competency that the other lacks. We have already seen several examples
of two-party allia nces of thi s kmd, but multi party allia nces are also possi ble. For
 578
The Design and
Dynamics of
Organizations
New United Motors Manufacturing,
Incorporated (NUMMI)
In 1984 To yota and General Motors established NUMMI to build Chevrolet
Novas and Toyota Corollas in a closed GM plant in Fremo nt, Califo rnia.
Built in the 1960s, the Fremo nt plant had been a money loser for GM and
had been plagued by l abor pro blems. Absenteeism rates of 22 percent and
drug and alcohol abuse in the factory had created a nightm are for manageme nt.
Em plo yees were equally unhappy, with some 2, 000 form al grievances still
unresolved whe n GM had finally closed the factory.
For GM, the new agreeme nt promised a way to learn about Japanese
m anagement practices from Toyota, a m aster and principal innovator of the
system . The top management of the new plant would be Toyota executives,
but GM co uld place up to 1 6 managers of its own in the plant where they
would learn abo ut Toyota's methods in order to transfer what they had learned
to other plants in North America and around the world. But what could
To yota hope to gain?
Toyo ta had been the last of the Japanese automo bile companies to
build facilities in th e U nited States. With increasing pressure from trade
negotiators to reduce Japan's enormous trade surplus with the U nited S tates,
Honda and Nissan had res ponded q uickly by building U . S . plants. One thing
that Toyota needed was a way to begin productio n in the United S tates
quickly, and GM's existing Fremo nt plant offered a quick solutio n. As it
prepared for its expansio n, Toyo ta wanted to begin buildi ng a network of
North American U. S . suppliers and to gain experience dealing wi th North
American workers and their unio n, the U nited Automo bile Wo rkers (UAW).
One statis tic stands out as a special indicator of its success: Although 8 5
perce nt of its work force came from former GM Fremont plant workers, the
absentee rate at Toyota was only 2 percent.
The arrangement was, by agreement, a temporary one. I ndeed, the
U .S. anti trust authori ties insisted that the agreement expire by 1 996 in order
to thwart the possibility of an intern atio nal automobile cartel. The labor
arrangemen ts set a pattern that deeply affected th e 1987 nego tiatio ns between
GM and the UAW, in which GM obtained greater flexibility in assignments
in exchange for m uch greater job security for workers. CM also patterned
many of the labor and manufacturing practices at its new Saturn subsidiary
on its NUMMI experience. Toyo ta took similar advantage of what it had
learned about American workers, who dem anded more auto nomy and more
informatio n than Japanese workers a nd who did not always participate in
"voluntary" company programs. Based on the confidence and experience it
had gained, Toyota built large new automo bile plants in Canada and the
U nited S tates.
 579
example, in 1 990 Hewlett Packard and EDS each purchased a porti on of Ask C omp uter The Boundaries
S ystems to help i t finance a purchase of Ingres, a data base software company. The a nd Structure of
four companies plan to team up to develop an d sell factory automati on systems. the Firm
In m an y alli an ces, the greatest danger is that one of the parti ci pati ng companies
may learn enough ab out the other's operati ons to dupli cate i ts routines an d special
competencies an d become an effective competi tor. For example, B oeing, in order to
participate in the developmen t of a Japanese fighter aircraft, agreed to conduct the
project as a j oint ven ture wi th Japanese partners, but i t excludes th ose partners from
i ts man ufacturing facili ties to pr otect i ts trade secrets an d speci al competen cies in
aircraft manufacturin g an d technology. Lear ni ng methods an d techn ology can also
be the reason for an alli an ce, as in the NUMMI example .
Economies of sc;ale can b e an importan t stim ulus to j oint projects for the same
general reason that they can fa vor the developmen t of cooperati ves. For example, the
In tern ati on al Telecommunicati on s Satelli te Pr oj ect (INTELSAT) i s a j oi n t project
wh ose members are governments wh o contribute capi tal and own shares i n proporti on
to their inten ded usage of the system. Similarly, U . S. semi con ductor firm s h ave formed
a re search con sorti um, Sem atech, to prom ote large- scale research in sem i conductor
m an ufacturi n g techn ologies. I t i s estim ated that the cost of a state- of-the-art, efficien t­
scale semicon ductor m an ufacturing fa cili ty m ay reach $1 billi on to $2 billi on by the
year 2000. This pri ce tag is far too h i gh for the firm s each to mai n tain separate
facili ties, an d the firms fear having to deal wi th a m onop oly supplier an d especi ally
one wedded to a foreign competi tor.
Econ omies of scope in m arketi ng can be an other reason for an allian ce. For
example, the Lenscrafters chain whi ch sells eyeglasses h as arran gemen ts wi th a variety
of optometrists, wh o run a completely separate eye examinati on business in the same
stores. The operati on s of the two parts of the store are qui te di stinct an d i ndepen den t;
the purp ose of the alli ance is to provi de m ore con veni en t service to the sh ared
customers.

Keiretsu
An especially interesting ki nd of aili an ce an d one of great imp ortan ce i n the m odern
world is the Japanese keiretsu, or gr oups of related firm s. These groups, whi ch
dominate the Japanese econ om y, con sist of indepen dent firms wi th close links an d
often a shared name. The biggest of these i s the Mi tsubi shi group, wi th 1990 aggregate
sales of approximately $175 billi on . Wi th 28 core companies and h undreds of other
affiliates, the members of the group produce an d deli ver th ousan ds of different pr oducts
an d servi ces.
Members of· the keiretsu are linked fin ancially i n two maj or ways. First,
com panies i n the group common ly own shares in the other m embers. On average,
ab out 24 percen t of the shares of the largest J apanese companies are own ed by oth er,
related companies. Nevertheless, because it is rare for any one compan y in a group
to own m ore than 10 percen t of an other, the companies do n ot autom ati cally di rect
purchases to related companies un less these other compan ies offer the best econ omi c
deal.
A di stincti ve role of the group of companies i s as an informati on network an d
source of j oin t ven ture partners. The presiden ts of member companies meet together
regularly to hear presen tati on s, share b usiness ideas, an d cement personal relati on ships.
At Mi tsubi shi, the meetings occur twi ce a mon th . To prom ote inform ati on exchan ge
and business dealings among em ployees below the presidenti al level, there are clubs
an d restauran ts reserved for employees of keiretsu members. The cen tral offices of the
member compani es are all located in close pr oximi ty to one an other in one area of
Tokyo to make atten dance an d i n teracti on s easy. These practices combine wi th an
 580
The Design and expectation that the member companies will support one another in projects that
Dynamics of make economic sense to create a powerful ability to move into promising new
Organizations businesses with an instant market for one's product. For example, all 28 of Mitsubishi's
core companies participated in establishing Space Communication Corporation, a
satellite communications company. The clear expectation is that the members would
become early customers of the new company, helping it become established and
profitable quickly. In a world where business alliances to exploit special competencies
are frequent, having a regular group of partners who encompass all the important
competencies and who are ready to deal can be an important advantage.
Although the keiretsu do not have the same central structure as a multidivisional
firm, there remain some important similarities. Just as the GM central office had to
direct its divisions not to compete among themselves in the product market, the
trading companies may play a similar role in the keiretsu. For example, Mitsubishi
Heavy Industries and Mitsubishi Electric both make industrial air-conditioning
equipment. Mitsubishi Corporation, the trading company often considered to be the
central Mitsubishi firm, distributes its products to export markets and works to prevent
these two separate companies from bidding against one another. In addition, capital
for new investments beyond that generated within the division is often allocated by
the central office in multidivisional firms, which reviews competing requests for funds.
In the keiretsu, the main group bank plays a related role, as well as serving as a
central, well-informed agent that can make opportunities for cooperation within the
group known to the member firms.
 !381
The Bounda ries
and Structure of
the Firm
SU!\IMAHY
Business organization has changed enormously over the past 1 5 0 years and continues
to change in important ways. From the small, fami ly-owned firms of the early
nineteenth century, the development of national and international markets linked by
improved transportation and communication led to larger-scale firms, which required
complementary innovations in management and finance. These innovations in turn
favored the development of similarly large firms in other industries. Experimentation
and complementary innovations give a momentum to change and stamp a consistent
pattern on it.
An especially important development was the creation of the multidivisional
form of organization, which was pioneered by Du Pont, General Motors, Sears, and
Standard Oil of New Jersey in the years following World War I. This form balanced
the needs of decentralized decision making, in which authority for decisions was
placed on those with local knowledge about operations, with the need for joint
planning to exploit economies of scale. As the form evolved, businesses found they
could handle larger numbers of divisions by organizing them into groups of related
divisions. Before the introduction of this form, most firms were either controlled from
a central office or operated as a holding company, in which the role of the central
office was little more than to provide financing and receive profits. Compared to
centralized organizations, the multidivisional firm had the advantages of decentralized
decision making. The manager in command was better informed about operations
than a central office committee could be and, being held responsible for performance,
could be motivated more effectively. The multidivisional firm was also better able to
expand into new businesses without overextending the few top executives of the central
office than were more centralized organizations. Compared to the holding company,
the m ultidivisional form could adopt a more coherent strategy and could sponsor joint
proj ects that enhanced the overall competitiveness of the divisions.
Implementing a multidivisional form requires deciding on how activities should
be grouped into divisions and resolving certain problems that are particular to this
form. Generally, the economic principles governing the divisions and thei r groupings
are that activities that require close coordination should be grouped together closely
to facilitate coordination, that division managers should be given wide latitude to
allow them to be fully responsive to performance incentives, and that divisions should
be small enough that those in control can have meaningful individual effects on the
division's performance.
The multidivisional form also gives rise to the transfer pricing problem as the
problem of determining the price at which trades between divisions will occur. If a
division makes a good for which there are competitive suppliers, then the competitively
determined price leads, theoretically, to efficient decisions. Otherwise, the double
marginalization issue arises, and the upstream will be inclined to set too high a price
leading not only to a transfer of earnings from the downstream firm but to a loss of
total profits as well.
The success of the multidivisional form leads to the question: What limits the
size of the firm? We address this question in two parts, focusing first on the vertical
and then the horizonta l structure of the firm .
An early and still common abstract conception of the firm i s a s a maker of a
single product, occupying some part of the vertical chain in which raw materials are
processed into parts and systems and finally assembled into products for distribution
to consumers. According to Coasian transaction costs economics, the chain will tend
 582
The Design and to be divided into firms in the way that minimizes the total costs of managing the
Dynamics of transactions between successive stages. Two successive stages will be vertically
Organizations integrated if the advantages of doing so outweigh the advantages of simple market
contracting. The latter include the benefits of competition among suppliers and
ownership incentives for them and any economies of scale and scope (including core
competencies) that may be unavailable to a small integrated firm. Compared to simple
market procurement, vertical integration allows for better coordination and better
protection of specific investments, permits the firm to capture any rents its supplier is
earning, avoids distortions like double marginalization that are associated with
monopoly supply, and may help to deter entry by potential competitors.
Vertical integration and simple market procurement, however, are not the only
available alternatives. New forms are continually developed among which we have
reviewed three. Cooperatives allow the members to reap the advantages of scale
economies while avoiding monopoly distortions in their purchase of supplies (buyers'
cooperatives) and marketing services (sellers' cooperatives). Lacking a large owner,
however, these cooperatives may suffer from insufficient monitoring of management,
conflicting objectives of members, and political infighting. Franchise retailing is
another form that maintains the powerful incentives of ownership for the franchisees
while allowing the franchisor to protect its brand-name reputation and to force
cooperation when close coordination is needed. The danger is that a powerful
franchisor may exploit franchisees, the mere threat of which discourages specific
investments. However, franchisee organizations to negotiate with the franchisor can
mitigate that problem.
A third new arrangement is that used by Japanese automobile manufacturers to
manage their suppliers. Using comparative performance evaluation instead of competi­
tive bi dding to pick suppliers, the system rewards efficient suppliers by awarding them
additional supply contracts. This promotes good performance and allows suppliers to
participate in product design, encouraging designs that are easy to manufacture. To
encourage innovation and the rapid diffusion of new ideas, the automobile maker
gives points to its suppliers in the internal rating system for any cost-reducing
innovations they share. To achieve economies of specialization and scale, the supply
contracts extend over the whole period for which the part or system is made. This
practice also protects any specific investment the supplier may make. The supplier
organizations contribute an additional layer of protection.
The identification of a firm with the product it makes is a tenuous notion even
in a static context, and it becomes quite untenable in an environment characterized
by constant product change. An alternative conception of the firm is as an organization
that musters and organizes the resources it controls to exploit appropriate business
opportunities. The way a firm operates its existing businesses helps it to acquire certain
core competencies. These competencies as well as opportunities to exploit other
economies of scale and scope are a major factor in determining the most profitable
directions of business expansion.
Multidivisional firms may also incur costs that the same divisions would not
incur if operated as separate companies. A clash of corporate cultures, based on
differing routines and expectations in different divisions, can cause friction as divisions
attempt to impose their standards on other parts of the organization. High levels of
incentive pay in small entrepreneurial organizations have often caused jealousies in
larger, more traditional organizations, leading to destructive influence activities such
as lobbying efforts for "pay equity. "
Alliances between independent business firms can arise for several reasons. A
common one is that the participating fi rms have different competencies or special
resources that can be usefully combined for some new business venture. Alliances
can also be c, reated to facilitate learning or to exploit economies of scale in research The Boundaries
or supply operations. and Structure of
Japanese business is dominated by a unique organization form: the keiretsu, or the Firm
group of firms. The independent companies that arc the members of these groups arc
bound together by mutual ownersh ip arrangements, frequent meetings of the company
presidents, physical proximity, common financing and direction from a main bank
and trading company, and a complex web of social relations among their employees.
The exchange of business information among group members and the ready availability
of qualified partners and financiers for economically sensible joint ventures give these
groups an unusual ability to identify opportun ities and to respond quickly and flexibly
to them .

• BIBLIOGRAPHIC NOTES
Most of the theoretical ideas used in this chapter are developed at more length
in other chapters, with contributors cited there. The other major infl uence on
this chapter is the writing of business historian Alfred Chandler, who has
emphasized the changing character of business organization, researched and told
the stories of the companies that invented the multidivisional form , emphasized
that changing business strategy drives organizational change and that new
organization structures affect the business's subsequent strategy, and insisted that
the relentless pursuit of economies of scale and scope were historically the key to
busi ness success.

• REFERENCES
A. O. Chandler, Jr. Strategy and Structure (Cambridge, MA: MIT Press, 1962).
A. O. Chandler, Jr. The Visible Hand: The Managerial Revolution in American
Business (Cambridge, MA: The Belknap Press, 1 977).
A. O. Chandler, Jr. Scale and Scope (Cambridge, MA: Belknap Press, 1990).

EXERCI SES

Food for Thought

1 . In 1 99 1 , Wal-Mart Stores overtook Sears as the leadi ng retailer in the

United States, and its rapid growth appeared to be conti nuing. Wal-Mart's computer­
ized inventory tracking systems allow it to keep close track of its stocks and alert its
suppliers to coming orders. Concerned about its own reputation among customers,
Wal-Mart pays close attention to its suppliers' quality and prices and engages some of
its suppliers in long-term relationships making unique brands, such as its Liberty
brand children's wear. What kinds of problems would you expect Wal-Mart to face
in coordinating with its suppl iers? How would this affect the kinds of contacts between
the companies? Compare the costs and benefits of th is system with those of an arms'
length supply relationship or a system of vertical integration in which Wal-Mart would
manufacture its own store brands .
2 . In recent years, the leading Japanese producers of consumer electronics
have purchased leading U . S . producers of entertainment "software" like records and
movies in an effort to improve coordination in new product introduction and
marketing. For example, Sony purchased the former CBS Records and Columbia
 584
The Design and Pictures Entertainment and Matsushita purchased MCA. What are the likely
Dynamics of advantages of this sort of horizontal integration? Could these consumer electronics
Organizations companies have achieved substantially the same results by entering into long-term
relationships with the software suppliers? Could arrangements for new product
introductions be worked out equally well on a product-by-product basis?
3 . Prudhoe Bay near the Arctic Circle in Alaska is one of the largest oil fields
in North America. Shippin g access to the Bay is blocked by ice during large parts of
the year, so a pipeline was built to transport oil across Alaska to the Port of Valdez,
where tankers can be loaded year-round. The eight large oil companies with major
holdings in the area formed a joint venture, Alyeska, to manage the operation of the
pipeline. Each oil company has a transportation subsidiary that owns a fraction of the
capacity of the pipeline and charges a fixed amount per barrel for shipping oil.
Expenses for operating the pipeline are allocated among the companies according to
a formula that Alyeska employs based primarily on volume. What problems would
you expect from a system of this kind? What might happen if a company's share of
ownership differed significantly from its share of the oil being pumped? Identify
alternative ways to organize oil transportation and compare their advantages and disad­
vantages for shipping Alaskan crude oil.
4. Honda Motor Company is one of three companies that the public generally
knows as Honda. The Motor company engages in sales and manufacturing of
automobiles, but there are also independent Honda companies for engineering and
for research. What problems or concerns might have prompted Honda to adopt this
unusual form of organization?
5 . In the U . S. airline industry, companies often form alliances to expand and
coordinate the network of cities they serve and to expand the schedules they offer to
customers. For example, the regional air carriers affiliated with American Airlines are
called American Eagle carriers, and they schedule their flights to connect with
American's longer distance flights, carrying connecting passengers to cities throughout
a regional market. In addition, American Airlines maintains relationships with some
national and transnational carriers, like Trans World Airlines and British Airways,
giving credit in their frequent Ayer program for mileage flown on these lines. At the
same time, the experience of smaller U. S. national air carriers like Eastern Airlines
and Continental Airlines suggests that size really is important for success. Remaining
mindful of the possibilities for coordination among independent airlines, discuss the
role of scale and scope in organizing airline operations.
 17
THE EVOLUTION OF BUSINESS
AND ECONOMIC SYSTEMS

ml . . . acmg a change more extenszve,

� usmess organzzatzons are 1C . more 1'-ar-reach mg
'
in its implications, and more fundamental in its transforming quality than anything
since the "modern" industrial system took shape in the years roughly between 1 890
and 1 920. These changes . . . come from several sources: the labor force, patterns of
world trade, technology, and political sensibilities. [T]he changes are profound,
and they are occurring together.
Rosabeth Moss Kanter 1

We opened this book by recounting historical episodes-stories of competi tion

among Ford, General Motors, and Toyota, and between the Hudson Bay Company
and the North West Company, of compensation reform at Salomon Brothers, and
the failures of communism in Eastern Europe and the Soviet Union. The episodes
exemplify many of the themes developed in the intervening chapters, but none more
clearly than the ongoing drama of organizational change.
As the previous chapter stressed, the scale, scope, and organization of firms
have changed massively over the decades as managers have experimented with
organizational innovations to solve coordination and motivation problems and to
respond to ever-changing markets and technologies. Within the firm, new pay systems
and new bases for compensation have emerged. Job definitions and career paths have
changed. Financing and ownership patterns have been radically altered. Relationships
between suppliers and customers have been restructured, the scope of activities carried
out within the firm has shifted, and the patterns of information flows, authority, and
reporting within and between organizations have been redrawn.
Recent decades have also seen massive changes in organization at the level of
the economy as a whol�. In the capitalist world, governments have deregulated,

1 The Change Masters: Innovation and Entrepreneurship in the American Corporation (New York:
Simon and Schuster, 1 983), 37-38.
 586
The Design and reducing the extent and intensity of regulatory oversight, and privatized, sel ling or
Dynamics of transferring firms that were previously government owned to private im·estors. ::\Iany
Organ izations examples of these phenomena can be found in Western Europe, North and South
America, and Japan. At the same time, the Soviet Union and the formerl y communist
nations of Eastern Europe have undertaken a total restructuring of their economic
and political institutions, introoucing new systems of democratic political decision
making and market-oriented economic-resource al location. In this final chapter, we
highlight and analyze some of these changes and the major forces that still drive
them.
One theme emerges repeatedly in our analysis: When a group of activities is
complementary, changes that increase the effectiveness of some activities in the group
promote adoption and impro\'ement of the other acfaities. Within a group of
complements, causation does not work in just one direction. The most successful
organizational innovations are often mimicked and copied by competitors and adapted
by firms in other industries. Those firms will be led by the logic of complementarities
to develop additional technologies and practices that enhance the effecti\'eness of their
newl y adopted organizational arrangements. As knowledge of these changes spreads,
the new organization becomes e\'en more advantageous, leading to e\·en more
widespread adoption and additional improvements. Changes that feed on themselves
in this way can generate a snowbal ling momentum that leads to sustained trends and
major shifts in technology and organization.

THE PRESEi\T .\i\D FL 1Tl'RE OF THE BL'SL\ESS FI R.\ !

In the late twentieth century, as throughout the preceding 200 years, business
organization continues to evoke as companies adapt to new challenges and opportuni­
ties. In particul ar, the rapid pace of technological change combined with falling
international barriers to trade have created new strategic and organizational opportuni­
ties to which many business firms have responded creatively. These responses ha\·e,
in turn, encouraged the spread and further development of the new technologies, as
well as further reduction in international bar:-iers.

Technological and Organ izational Chan ge i n l\ lan ufactu rin g

The frontiers of technological advance in the late twentieth century have been
remarkably broad, impinging on organizations from many directions. One example
is the changing technology for making semiconductors, where the growing importance
of economies of scale in both research and manufacturing has changed the face of
the industry. The cost of a semiconductor manufacturing facility capable of producing
modern memory chips is expected to rise by a factor of about ten from about $200
million to about $2 billion between 1 990 and the year 2000 , promoting further
concentration in the already concentrated chipmaking business. In response, chip
users and makers in the United States have initiated joint \·entures in sem iconductor
research and manufacturing (for example, the Sematech R&D venture and the
unsuccessful U. S. � lemories manufacturing effort). Mcamd1ile, the world's largest
chip maker, IBM, has tried to increase its own scale of production by dropping its
long-standing pol icy of producing semiconductors only for its mrn internal use and
has entered into a series of cooperative efforts with various firms to develop and
manufacture new sem iconductor-based products. \Vorld\\·ide, strategic al liances of
various sorts ai med at sharing the costs and risks of research and development in the
electronics a nd other industries ha\'C become more and more important. The rise of
these alliances has blurred organizational boundaries and increased the need to
encourage coordination among organizations.
Col\tl\ll lNICATION AND THANSPOHTATION TECI INOLOCIES Scale econo mics arc only one The Evolution of
way that techn ology affects organ ization. A sec ond modern example is the fal ling Business and
costs of telec omm unications and rapid air cargo systems, which have reduced the Economic Systems
need for many firms to hold high levels of finished go ods inventories in local shops
an d wareh ouses near their customers. Instead, these firms emphas ize gathering an d
quickly pr ocessin g information about customer deman ds, adj ustin g production to
match current demand, and sh ippin g products to arrive only wh en they are actually
needed. For example, Benett on-the Italian clothing man ufactur er-c ollects n ightly
sales data fr om retail stores and tries to sh ip j ust what is n eeded from its sin gle central
wareh ouse near Venice. Th is style of operation reduces the need for high levels of
clothing in vent ories in the stores an d econ omizes on overall inventories by shipping
goods on ly wh en an d where they are most valuable. This style of operations would
have been impossibly expensive with the costly data c ommunications, slow an d
unrel iable computing, or slow transportation systems of earlier eras. The style req uires
stan dardization of the systems used by the individual stores and close coordination
among them on changes to the system, which in turn req uire a sh ift in responsibility
for systems design from the stores to the system planner. In this case, strategy an d
organization are tightly linked, and the feasibility of the strategy is a c onsequence of
new techn ologies.

FLEXIBLE l\lA.NUFACTURINC TECHNOLOGIES An other change of the late twentieth

century is the increasin g use of flexible man ufacturin g techn ologies, which allow the
same equipment to be used efficiently to make a variety of different pr oducts. For
example, leadin g Japanese auto makers are approaching the point where th ey will be
able to serve th eir local markets on essentially a make-to-order basis, buildin g each
in dividual car to the specifications of an individual cust omer's order. Th e contrast
with Henry Ford's assembly line, which could pr oduce an yth ing the cust omer wanted
as long as it was a black model T, is remarkable.
The use of these flexi ble man ufacturing techn ologies is complementary with
the use of n ew telecommun ications and transportation capabilities. The case of Allen
Bradley, a U . S . manufacturer of electronic contr ols, illustrates the complementarity.
Makin g h undreds of different models of contr ols in its Milwaukee factory, the c ompan y
programs its flexible pr oduction eq uipment to fill each electron ically transmitted
in dividual order separately, h olding n o finished goods inventory, and sh ips the products
by air express within a day of receivin g the order. With out the ability to process orders
quickly an d to ship them expeditiously, the ability to sh ift production from one model
to an other in seconds would be much less valuable.
Flexible operations require very different organization al policies and procedures
than have been traditional i n manufacturing. For example, traditional cost-acc ountin g
procedures, devised in an era of mass production systems i n wh ich specialized tools
were used for each type of product and the important controllable fact ory costs were
th ose of labor time, h ave proved less useful for informing decisions about products
made in these new ways. When machines are put to multiple uses, th e acc ounting
system needs to provide information for deciding among th ose uses. Companies, such
as Hewlett Packard, h ave pioneered n ew systems to provide more accurate assessments
of th e marginal c osts of makin g each pr oduct, for example by design in g devices to
record th e time used on each type of machin e in a manufacturing process an d figurin g
the cost of each product acc ordin g to the n umber of machine h ours used. This c ost
information helps n ot only in pricin g decisions, but als o in establish ing pr oduct
designs that can be man ufactur ed at low cost an d in creating mean ingful measures to
evaluate the performance of factory managers.
Flexible production eq uipment als o encourages more freq uent product chan ge,
 588
The Design and which changes the demands on engineering, marketing, and supply, as well as on the
Dynamics of assembly plant. For example, frequent product change has forced a reorganization of
Organizations the product design and engineering functions of firms. In traditional mass production,
product designers were aloof from actual factory operations. A product would be
designed, then the design would go to engineers who would figure out how to turn
it into a physical reality, and then the engineers would pass it on to manufacturing
to figure out how to produce the product economically in volume. In an earlier era,
this sequential approach caused few problems because a new factory or at least new
equipment would be used to produce any new product. With more variety and more
frequent product change, modern design engineers need to know more about the costs
and capabilities of existing equipment and about what designs can be brought rapidly
into production. To achieve this, firms are replacing the traditional sequential pattern
of separated specialists with teams of designers, product engineers, and manufacturing
experts who work together to develop products that can be made well and efficiently.
This pattern is supported by yet another technical development-the introduction of
computer-aided design and manufacturing (CAD-CAM) programs and the computer
hardware on which to run them. Use of these technologies is complementary to
frequent product change because it reduces the costs of product design and facilitates
bringing new designs into production quickly and cheaply.
These technologically induced changes are also affecting jobs and career paths.
Many engineers are now rotated frequently among jobs to create a deeper understanding
of the capabilities and needs of different parts of the company. The same changes
affect production workers, who are expected to master a variety of skills and to be
willing to move from task to task, often working in teams and being paid for the skills
they have mastered rather than for the particular job they are doing. It is the workers'
range of knowledge and skills, more than any other single factor, that determines how
quickly a firm can work out the bugs in making a new product and how easily it can
introduce new technologies and improved methods. Progressive firms, recognizing
these facts, are moving to entrust their workers with more responsibility and discretion,
allowing them to make better use of the special local knowledge they alone possess.
In industries where the pace of product change is rapid, the individual product
market is no longer a suitable focus of strategic decision making. Instead, management
must anticipate its customers' needs and invest in building the firm's capabilities to
fulfill those needs. The adoption of this new focus entails far-reaching changes. These
firms develop systems to maintain effective, ongoing communication with customers
to learn about their changing plans and requirements. They develop a capability to
design and manufacture new products quickly, and they train and reward \Yorkers to
be multiskilled. They work to coordinate with suppliers and distributors to ensure
coherence of plans and actions in the face of changing circumstances. They may also
build alliances with other businesses to gain quick access to capabilities that they lack.
All of these changes are promoted by the modern design and manufacturing
technologies, and all increase the rate of product change and the demand for further

The Service I ndustries

improvements in these technologies.

Although manufacturing provides the most striking examples of complementary

technological and organizational changes, these patterns are present in the service
sector as well. Financial services provide a case in point. Easy, high-speed voice and
data communication has resulted in increasingly close linking of financial markets
around the world. Investors and corporate treasurers in North America can keep
abreast of developments on the Tokyo and London stock markets as easily as of those
on the New York and Toronto exchanges. This has created new opportunities for
 589
securities firms, encouraging them to expand and reorganize to enable them to seek
funding and investment opportunities for their local clients on a world scale. The
The Evolution of
Business and
techn ical cha nges and the new services offered by securities firms encourage both Economic Systems
investors a11d borrowers to adopt a globa l approach to their financial decisions. This
further increases the flow of information, the demand for global communications,
and the changes occurring in the markets a nd the securities firms.
Globalization of Econom ic Activity
In nineteenth century North America the fall ing costs of commun ication (telegraphy)
and transportation (railroads) combined with the constitutional prohibition on the
individual states' regulating interstate commerce in the United States and the "National
Policy" of conscious government sponsorship of national economic integration and
development in Canada to stimulate the growth of national markets and national scale
firms. A similar phenomenon is occurring on a global scale in the late twentieth
century under similar impetuses.
The spread of jet airline service, fax machines, international overn ight parcel
services, cheap and reliable international telephone links, wide-ranging scheduled air­
freight operations, and computer data links via satellites have made moving information ,
goods, and people unprecedentedly easy and quick. Among centers of commerce, any
place on the globe can be reached in a day from any other spot, and information can
pass between any two points at the speed of light. The effects of the huge improvements
in the technologies of transportation and communication have been ampl ified by
reductions in economic and political barriers to international movements of goods
and capital (and, to a lesser extent, labor). The United States and Canada have agreed
to free trade between them, and in 1 99 1 negotiations were under way to incl ude
Mexico and to create a North American free trade zone. Western Europe in 1 992
would take another major step toward becoming a single, united market in goods,
services, capital, and labor. More broadly, trade barriers have been progressively
diminished under successive rounds of international negotiations under the General
Agreement on Tariffs and Trade (CATT). Together, these developments have resulted
in the beginnings of global markets.
Firms al ready ra ise money in whatever part of the world they find most attractive:
Japa nese firms have issued bonds in Europe and arranged for their shares to be traded
on the New York Stock Exchange, and the Walt Disney Company has put together
limited partnerships of Japanese investors to finance production of its films. Investments
and acquisitions occur across national boundaries, as do strategic all iances. A Canadian
firm, Olympia and York, was one of the biggest land owners in New York City in
1 99 1 and had undertaken the largest real estate development in London . Western­
European and U . S . firms had begun setting up joint ventures and subsidiaries in the
former social ist countries of Eastern Europe and the Soviet Un ion . Apple Computer
and Sony were working together, as were GM and Toyota, and Renault and Volvo .
Following their cl ients-and sometimes leading them-consulting and accounting
firms and investment and commercial ba nks were reaching out internationally,
establishing themselves around the world. Goods and services markets were becoming
global as well. World brands were emerging, with such names as Panasonic, Louis
Vuitton , Coca Cola, Gucci, Levi's, Reebok, and Mercedes Benz known and sought
worldwide.
As companies develop internationally, they face new kinds of challenges.
Products need to be developed to meet the special ized needs of d ifferent markets. For
example, automobiles need to be adapted to the driving regulations and conditions in
different countries-the width, condition , and congestion of the roads, which side of
the road traffic moves on, speed limits, safety and fuel-economy regulations, local
 590
The Design and content rules that prescribe what fraction of any good must be produced locally to
Dynamics of receive favorable tax treatment, weather conditions, and so on. At the same time, the
Orga nizations gains in productive efficiency from standardization on a global scale can be hard to
ignore, even though they conflict with the need to meet local tastes. Attempts to serve
a national or regional market by establishing a research center or factory there create
a need for developing enough flexibility in procedures to accommodate local patterns
of business relations, labor-market practices, and government regulation. Additional
demands are imposed on the organization by the need to develop means to coordinate
information flows, make decisions, and implement plans across national borders and
time zones, to surmount linguistic and cultural differences, and to take advantage of
the particular skills and resources that are most prevalent in each region.
Even firms that decide to stick to serving their local or national markets are
affected by the growing internationalization of business because their competitors or
the lowest cost supplier may be from other parts of the globe. This means that the
ability to operate in global markets is of very general value. It also means that there
are strong forces driving firms to adopt the strategic and organizational adaptations
that help them to operate in international markets. Just what these adaptations may
be is not yet clear: These are relatively new management problems, and international
firms are experimenting with many kinds of solutions. Whatever solutions do emerge
will further promote the globalization of economic activity by lowering the cost of

I nnovations in Ownership, Financing , and Control

doing business internationally.

Even after allowing for the influences of technology and markets, business organizations
evolve in response to organizational entrepreneurship-the development and imple­
mentation of new management ideas. As we have seen, the 1 980s were an especially
active period for innovation in financing and control. In the English-speaking
developed countries, for example, highly leveraged firms created by LBOs emphasized
incentives for top management and close scrutiny by large investors in an effort to
improve business performance. The same ideas lay behind a series of other innovations
in financing and incentives that spread in the late 1 980s. Some firms now provide an
equity stake to their divisional managers, encouraging them to behave like the
managers of highly leveraged firms but retaining sufficient resources in the central
office to bail out failing divisions in hard times. In the insurance industry, mutual
companies that were formerly owned by their policy holders are being demutualized,
that is, sold to private investors who will exercise tighter financial control over the
managers than the dispersed policy holders were motivated to provide. These trends
illustrate how ideas developed in one sector of an economy are adapted to promote
change in other sectors.
If the effectiveness of these new organization forms is still unproved, the success
of the keiretsu in Japan is not. These groups of firms have proven to be extremely
agile in responding to new market opportunities and new technologies, transferring
information, technology, and financial resources among members, rescuing partners
in financial trouble, cooperating in exploiting new opportunities for investment, and
maintaining employment and high levels of productivity and growth. In an environment
characterized by rapid change, the ability to transfer technology and knowledge is a
new aspect of the organization problem, whose importance may equal that of the
traditional aspects of coordination and incentives.
Even as firms around the world study Japan's system for hints about its success,
that system continues to change. Reliance on debt financing among Japanese firms
fell sharply during the l 980s, and some of the most successful Japanese industrial
firms have completely escaped their dependence on bank financing. At the same time,
the cost of capital rose m arkedly in the early 1990s, un til by som e estimates it exceeded The Evolution of
that in the Un ited States, while the boom in lan d an d stock-m arket prices of the l 980s Business and
faded and, in the latter case, reversed. Deprived of th e capital gains they had enj oyed Economic Sy�tems
in the past, in stitutional investors' re turn s from holding Japanese corporate stocks
became depen dent on divi dend payouts, which had tradition ally been very l ow. S ome
observers wondered whether this woul d lead the institution s to increase pressure on
managemen t to generate an d distribute current earn ings. The l ikely con seq uen ces of
these chan ges in J apan are as hotly debated as are the conseq uences of the 1980s'
financial restructuring in the United States an d United Kin gdom.

H u m a n Resources
Japan is facing maj or changes in its labor markets as well. The population i s growing
older, · and many forecast a labor shortage in future years. This might well con tribute
to the emergence of a very active labor market for experienced workers, similar to that
in other developed countries. Further, the lifetime employment system and the
employee's devotion to the interests of the employer are being challenged by a younger
generation that has grown up in affluence and has been exposed to other countries'
ways. As we have argued, the absence of an active outside labor market interacts
cruciaJ ly wi th other elements of the Japanese employment and human-resource
management system. These developments thus raise serious challenges to the J apanese
system and to the organizations within it.
In other countries, the increasing role of women in the labor force is forcing
reconsideration of old patterns. Traditional career paths were often based on a model
of a male employee who focused on his j ob while his wife stayed home to keep house
and raise children, supporting him in hi s j ob, and followi ng wherever her husband's
career took them. As these patterns change, firms are havi ng to revamp their human­
resource policies to account for the needs of dual-career couples and the demands of
child care.
In the U nited States and Canada especially, firms are also concerned that the
educational systems are not producing potenti al employees with the basic skills in
language, mathematics, and simple reasoning to meet the demands of the new
technologies and methods. Many are increasing their expenditures on training
pr ograms, even when the human capital being developed i s nonspecific. M eanwhi le,
although the J apanese educational system has been successful in teaching students
facts, it has also been criticized for not encouraging creativity. J apanese criti cs have
called for change on that di mensi on.

THE PRESENT AND FUTURE OF ECONOMIC

RESTRUCTURING IN EASTERN EUROPE AND THE USSR
Although the changes in the structure of capitalist firms and economies in the late
twenti eth century have been quite profound, they are overshadowed by the pervasi ve
and momentous economic reorganization occurring in the formerly communist states
of Eastern Europe and the Soviet U nion. The transformation of these economies is
not solely a matter of transferring the state- owned means of production into private
hands. Rather, one coherent system is being junked, and another coherent system
must be found and implemented to replace it. The process is not reform, but
metamorphosis.
In Chapter 4 we discussed the possibility that there could be multiple coherent
patterns of organization. Within each pattern, the various elemen ts of organizational
design and strategy are mutually supporting, and yet one pattern might be much more
productive than another in a particular en vironment. M oreover, mi xing elemen ts
 592
The Design and from two different coherent patterns would not typically yield another coherent pattern,
Dynamics of but instead might more likely result in a mismatch that performs miserably. We also
Organizations suggested that the process of moving from one coherent pattern to another might be
quite difficult, especially if there were little relevant experience within the organization
about how the pieces in the new solution might fit together and support one another.
The discussion in Chapter 4 was at the level an individual firm, but the same point

The Communist and Capitalist Systems

applies with even more force to the restructuring of an entire economy.

State ownership, centrally planned resource allocation, authoritarian political decision

making, and the monopolization of power by a pervasive communist party together
constituted an internally consistent, coherent solution to the general economic prob­
lem of organization: It was a system. In later years it became clear that it was not a
very efficient system, even from a purely materialistic point of view, and it always
involved tremendous costs in terms of human freedom, human dignity, and human
lives. Nevertheless, it did perform well on some dimensions. Centralized decision,
allocation, and control mechanisms allowed the U SSR to focus its physical and
human resources with remarkable effect on narrow objectives: the rapid development
of heavy industry in the period through World War II and of space rockets and military
hardware in the following decades. The overwhelming difficulties created by the
inability to use local information appropriately and by the dysfunctional incentives,
however, meant that achieving these successes used up inordinate amounts of
resources. Moreover, these same problems meant that the resources that remained
could never be put to very effective use in meeting the varied needs of consumers:
The complexity of the task was too great.
Private ownership, decentralized resource allocation, democratic political deci­
sion making, and a distribution of powers among a variety of different institutions and
organizations together constitute another coherent system, with the various pieces
again mutually supporting one another. These two solutions differ on almost every
relevant dimension, however. Moreover, few of the people involved in reshaping the
former communist economies have experience with any version of the alternative
pattern. This is especially true in the Soviet Union, where there is essentially no one
with personal experience and first-hand knowledge of the operation of a market
economy and a democratic political system. In the language of Chapter 4, these
nations face a design problem with innovation attributes: A host of different pieces
must be introduced and structured to fit together, and no one has much experience
with identifying the relevant pieces and how they fit.
To illustrate, consider only the issue of the ownership of the means of production.
As we have seen, the very definition of ownership is a subtle matter. What rights and
duties will owners have? What rights and duties will be assigned to workers, managers,
and government? The answers given to these questions have very real effects on
incentives, behavior, and performance. If share ownership is dispersed, as many of
the reformers advocate, how will business organizations be controlled? What mecha­
nisms will ensure they function in a way consistent with efficiency? If ownership is
concentrated in mutual funds or other institutions, how will those institutional
managers be controlled? Competition in input, output, and corporate-control markets
are im portant parts of the solution in the English-speaking economies. In Japan and
continental Europe, however, ownership is not as dispersed, and bankers play a larger
role in overseeing corporations. It is important to understand that different control
systems arc adapted to different ownership patterns and legal structures. Once
ownership is established, how will investments be safeguarded? A largely legal approach
based on court enforcement of explicit contracts is one possibility. This requires a
legal system that is very different from what ha s been in place. Primary reliance on The Evolution of
mul tiface ted, l ong- term relation ships is more in the Japanese mo de. This approach, Busi ness and
too, demands a se t of legal and institutional supports that arc missing in Eastern Economic Systems
Europe.

Managing the Transition

We described the move from one system to another as a m etamorphosis. Both
ca terpillars and butterflies arc coheren t, successfully function ing crea tures; the thin g
in transition in side the ch rysalis is n either. Similarly, econ omies halfwa y on th e
tran sition between commun ism and capitalism are apt to be monsters, una ble to
function. It is crucial to go all the way to a new, coherent form and not to da wdl e
or ge t stuck en route . .
. During the commun ist era in Ea stern Europe, in cen tives in the enterprises were
bl unted because any losses suffered would be financed by the state and most profi ts
would be taken by the sta te, with only meager rewards for good performance and
trivial penalties for failure. In the tran sition between communism and private
ownership, incen tives for profi ta ble en terprises ha ve often become even worse. With
a change of ownersh ip imminent, any resources that the firm can squirrel awa y will
reduce current profi ts and taxes and increa se the amount even tua l l y a vaila ble for
workers, managers, and the new own ers. In Poland, partly as a result of these incen tives
and partly from the turmoil in the econ omy, to tal en terprise profi ts in 1 99 1 were
foreca st to be virtually z ero. This elimina ted virtually all of the govern men t's income
from the profi ts tax, one of its main sources of revenue, and contributed to a ma ssive
fiscal crisis. The longer the transition takes, the worse these problems will be. At lea st
in this respect, Poland's radical, "cold turkey" approach seemed wise. Despite serious
unemploymen t and other econ omic woes, shortages had ceased, and there were signs
of econ omic recovery.
The problems were much worse at the same time in the USSR, largely because
decisive action had n ot been taken. Thr ough 5 years of half-hearted economic reform
and maj or political change, President Gorbachev had attempted to sa ve essential
elemen ts of the communist system without any apparen t coheren t plan as to
where perestroika would lea d and what system would result. Pressures moun ted for
aban donmen t of commun ism and adoption of a th orough- goin g market system on
one side, and for return to tigh t state and party control of the economy on the other.
By 1 99 1 the state min istries had essentia l l y ceased to function as a means of
coordinating econ omic activity on even the inefficient level they had previously
attained. They and their allies were still powerful en ough, h owever, to en sure that
well-fun ction in g markets had yet to replace them. Consumer goods of the most basic
sort were becomin g very scarce in man y parts of the coun try, and there were fears of
real hunger during the winter of 1 99 1 - 1 992. The sh orta ges were not j ust a result of
much of the suppl y being drawn off to the nascen t free markets in the richer area s,
although the differen tial between the free-market prices and th ose paid by the state
was more than sufficien t to pa y the fines imposed for failin g to deliver the sta te's
orders. Even more important was that the decrepid tran sportation system could not
get food from the farms to the people: Huge fractions of the crops from the collective
farms were rotting in fields. Real output was shrinkin g rapidly, and infla tion wa s
becomin g a maj or problem. Comp oun ding the problem wa s a deteriora tin g balance
of pa ymen ts, as the previousl y captive markets in Ea stern Europe were lost and oil
production , one of the few major sources of hard currency earnings, fell precipitousl y.
Meanwhile, the th reatened pol itical collapse of the Soviet Union and the afterma th
of the attempted coup d'etat of August 1 99 1 were divertin g attention from the crucial
tasks of establishin g a coherent econ omic system for coordination and mo tiva tion,
 .594
The Design and whether through private ownership and markets or state ownership and central
Dynamics of planning.
Organizations Throughout Eastern Europe, new systems are being implemented by people
with no experience and little understanding of how the system works or what it takes
for a capitalist system to succeed. The social experiment of economic transition is an
event of great historical importance, but many severe disappointments on the road to
capitalism are likely.

THE FUTURE OF EC0NOt\llCS, ORGANIZATION.

AND MANAGEMENT
The economi c analysis that goes under the label of the "theory of the firm" in
traditional micr oeconomics textbooks is one in which firms act like si ngle deci'-ion
makers gu ided by prices and costs, buying labor and other inputs in impersonal spot
markets, converting them into output via given and largely unchanging technologies,
and selling this produ ction in more imperso nal markets. This has pr oven to be a very
useful abstraction for many purposes, but of course the real ity of actual firms is very
different. Large firms bring together a mu ltitude of individu als with differing
information and interests, and it is only throu gh careful design of the incentive and
coordination systems that coherent, fruitfu l action can be achieved. Accommodating
the diversity of hu ndreds or thou sands of uniqu e local markets increases these demands
becau se it means that deci sion making mu st be divided among many parties. Long­
term relations with empl oyees, suppliers, and cu stomers are commonplace and
fu ndamentally important for developing systems in which prices are not the primary
gu ide of individual behavi or. With constantly changing technologies and long-term
relationships, long-term strategic decisions are based on bu ilding the systems,
capabilities, and alliances to respond flexibly and coherently to the challenges of a
changing and uncertain future. Recognizing all this is crucial to understanding
organizati ons as they really are. This understanding, in turn, is necessary if we are to
help structure and manage organizations so that they better serve peoples' interests.
Stu dents and practitioners of management have traditionally been well aware of
the actual features of organizations, but they have lacked any u nified and powerful
framework for organizing and understanding the facts and pattern s they observed.
Without this, anecdotes too often become the basis for general pronou ncements.
Theory-a sorting out in a logically consistent fashion of the forces that are important,
of how they interact, and of what consequences resu lt-is necessary for a real and
useful understanding of the bu siness environment. As this book has demonstrated,
economics offers uniquely powerful tools and methods to bui ld such a theory.
This pinpoints yet another complementarity: Theory is enr iched and made
usefu l by institutional knowledge, and this knowledge is organized and made generally
applicable by theory. Economists have too long ignored the study of how firms and
economic systems actu ally operate in a dynamic, tum ultuou s environment. Those
who have studied these matters and those who have managed organizations in these
envir onments have too long labored without the benefits of useful theories to gu ide
their investigations and their decisions. Bringing the two pieces together will give
much more th an the simple sum of the parts. The serious study of economics,
organization, and management has j ust begu n.
 GLOSSARY

Terms in italics correspond to other en tries in the glossary. agent: One who acts on behalf of another. See agency
relationship.
accounting rate of return (on assets): The rate of return on allocation: In the neoclassical model of a private ownership
an investment for a particular period-such as a year­ economy, an allocation consists of lists for each consumer
computed by dividing the net income before interest charges and each firm of the amount of each commodity to be
for that period from an income statement (prepared ac­ bought and the amounts to be sold. More generally, an
cording to the applicable accounting rules) by the accounting allocation is a complete specification of how resources are
value of the assets. to be used.
accounting rate of return (on equity): The rate of return antitakeover statutes: Public laws designed to make hostile
on an investment for a particular period-such as a year­ takeovers more difficult. Many U. S. states have adopted
computed by dividing the net income after interest charges such legislation.
from an income statement (prepared according to the arbitrage: Originally, buying and selling the same item in
applicable accounting rules) by the accounting value of the different markets simultaneously in order to profit from a
firm's equity. difference in prices between the markets. A pure arbitrage
adverse selection: Originally, an insurance term referring transaction involves no risk and no net investment. The
to the tendency of those who seek to buy insurance to be a term is now applied more broadly to trading that takes
nonrandom selection from the population-more particu­ advantage of discrepancies in pricing among groups of assets
larly, to be those who expect to have the highest expected that are close substitutes.
claims. Adverse selection now refers also to the kind of asset: A potential future Aow of benefits and services. Also,
precontractual opportunism that arises when one party to a the article giving rise to the stream. For example, shares of
bargain has private information about the something that stock or machines are assets.
affects the other's net benefit from the contract and when assignment problem: A situation in which efficiency re­
only those whose private information implies that the quires that one or more tasks be done and that only a single
contract will be especially disadvantageous for the other individual or group do each task. The problem is to ensure
party agree to a contract. first that all the tasks are done without duplication of effort
agency relationship: As used in economics, an agency and second that no other assignment of people to tasks
relationship is one in which one person (the agen t) acts on yields greater output or incurs lower costs.
behalf of another (the principal). For example, an employee authority relation: One in which one party (the superior)
is an agent of his or her employer and a doctor is the agent has the right to direct the behavior of the other (the
of a patient. (The term principal-agent relationship has a subordinate), at least within bounds, and to supervise,
narrower meaning in law. ) monitor, and punish or reward the subordinate.

595
backward integration: B ri nging the supply of an input under on a riskless investment plus an excess return proportional
the ownershi p and manage ment of the input purchaser. A to the beta of the stock.
form of vertical in tegration. cash flow: Accounting net income plus any allowances for
bankruptcy: A set of legal provisions designed to come into depreciation. Essentially, the amount of money generated
effect when a firm or individual is unable to meet its debt by the operation that is currently available for investi ng,
obligations. Under Chapter 7 bankruptcies in U nited States servicing any new debt taken on, or disbursing to owners.
law, the debtor firm's assets are sold , and the proceeds centralization: The vesting ofcontrol for ind ividual activities
distributed to claimants in order of their priority. ln a with a higher authority who communicates or i mposes the
Chapter 1 1 bankruptcy, for which the firm applies to the decisions on the individuals. The higher authority might
courts, the firm is protected from its creditors for a period be a person at a higher rank in a hierarchy, or the collectivity
in which it attempts to reorganize and negotiate a settlement of individuals themselves acting through some group deci­
of claims with the creditors . sion process. Contrast with decentraliza tion, where each of
bargaining costs: The transaction costs involved in negotia­ the individuals themselves would make the decisi ons .
tions between or among different parties. These include CEO: Chief Executive Officer, the highest ranking officer
the time spent on bargaining, resources expended during of a corporation . The CEO is typically ei ther the company
bargaining or in trying to i mprove bargai n ing position , and president, or chair of the board, or both .
any losses incurred as a result of failure or delay in reaching certain equivalent: Gi ven a choice between an uncertain
otherwise efficient agreements. or random income and a certai n , nonrandom one, the
beta: The expected percentage change in value of a financial amount of certain income that would make the chooser
asset when the value of the ma rket portfolio of all assets just ind ifferent between the two alternatives . Also called the
changes by one percent. See capital asset pricing model. certainty equivalent.
bounded rationality: The lim itations on human mental classical firm : A conception of the firm in which the actions
abilities that prevent people from foreseeing all possible of the firm are those that would be taken if a single individual
contingencies and calculating their opti mal behavior. had decision making authority and paid fixed wages to
Bounded rationality may also include those limitations on workers and prices to suppliers and received as profit any
human language that prevent perfect communication of excess of the firm's receipts over its expend itures .
those thi ngs that are known . classified board of directors: A board i n which only a
brittleness: The extent to wh ich the perfor mance of an mi nority of the members are up for re-election by the
econ omic system deteriorates when the information supplied stockholders in any year, so that the votes of even a maj ority
to the system is slightly i naccurate. of the shares cannot immed iately replace a majority of the
business j udgment rule: ln U . S . corporate law, the unwill­ board . Also called a staggered board.
i ngness of the Courts deciding in lawsuits against a corpora­ Coase Theorem: A proposition that if there are no wealth
ti on's board of di rectors to "second-guess" the board's effects and no signi ficant transaction costs, then (apart from
business decisions and rule that those decisions were un wise distributional considerations) the outcome of bargaining or
or mistaken , provided they were made in good faith and in contracting is independent of the initial assignment of
the honest belief that they were in the best interest of the ownership, wealth , and property rights and is determi ned
company. solely by efficiency.
bust-up : A form of takeover in which the purchaser of a coefficient of absolute risk aversion: Twice the amount that
company resells its individual divisions or other assets to an individual would pay to avoid having to bet $ 1 on the
other buyers. toss of a fair coin .
call option: A financial contract in which the buyer of the coercive tender offer: A tender offer structured so that i t is
option receives the right (which he or she may choose not in the best interests of each shareholder to tender , even
to exercise) to purchase some particular asset (often, shares though he or she believes the offer is for less than the firm
of a particular company) from the seller of the option on is worth.
a particular date (a European call) or on or before a collateral: Property of a borrower that is contractually
predetermined date (an American call) at a predeterm ined forfeited to the lender if the loan payment terms are not
price, called the striking price . See also option, put option, met. Also called security for the loan.
wa rrants. common-resource problem: A situation in which several
capital asset pricing model (CAPM) : A model of stock different parties can use a resource for their individual
markd prices accord ing to wh ich the expected or average benefit and property rights are not sufficiently well defined
return to investing in any stock is equal to the rate of retu rn and enforced to ensure that individuals bear the full costs
of the actions and receive the full benefits they create. Also activities well. Typically, this refers to the firm's ability to
called a free-rider problem (especially in situations where design, make, sell, or distribute a certain kind of product.
excluding parties who fail to pay for the resource is difficult) corporate control: Authority over the decisions of a firm,
or a public good problem. The resulting inefficiencies have typically attained by purchasing alone or with allies a large
led to the term the tragedy of the commons. fraction of the firm's shares.
comparative performance evaluation: The practice of evalu­ corporate culture: A set of shared beliefs and values,
ating an individual's performance by comparing it to the precedents, expectations, stories, routines, and procedures
performance of others doing similar work. in a firm that help define that firm's way of doing things
competitive equilibrium: A list of prices, consumption and serve as a guide to behavior for tho�e within the firm.
plans, and production plans such that ( 1 ) every individual corporate raider: An individual who undertakes hostile
consumes the goods he or she most prefers subject only to takeovers.
the limits of his or her budget, (2) every firm makes goods corporate strategy: The determination of which business
and uses inputs in the way that maximizes its profits, and activities the firm will undertake. Contrasts with business­
(3) the total _quantity supplied of each good is equal to the unit strategy, which is concerned with how the firm will
total quantity demanded. compete in a given business.
complements: A set of activities with the property that doing corporation: An organizational form that allows the enter­
more of any subgroup of the activities raises the marginal prise to act as a legal entity separate from its owners, who
return to the other activities. enjoy limited liability for the corporation's debts. See
i
complete contract: A hypothetical contract that describes also not-for-prof t corpora tions, public corpora tions, priva te
what action is to be taken and payments made in every corporations.
possible contingency. cospecialization: A condition of two assets each of which
complete markets: A hypothetical set of markets, one for is more productive when used with the other. Cospecialized
each possible commodity at each future date and for each assets must be unique in some respect and must also be
possible realization of the uncertainty in the world. complemen ts.
conglomerate firm: A firm operating in several unrelated cost of capital: The cost to an organization of obtaining
lines of business. financial resources. Usually stated as an interest rate.
connectedness: The expected loss incurred from failing to covariance: A measure of the extent to which the realizations
coordinate a particular group of decisions. of two random variables are linked. Measured as the
consumption plan: A pair of lists showing the amounts of expected value of the product of the deviations of the
every good that a consumer plans to buy and to sell. variables from their means.
Actual consumption is calculated by adding purchases and debenture: A document indicating that a firm has incurred
subtracting sales from the consumers's endowmen t. a debt and setting forth the required interest payments and
contracts: Formally, contracts are legally enforceable prom­ repayment schedule.
ises. They may be oral or written, and they typically must debt overhang: A situation in which a borrower that
involve obligations on each party-for example, to provide does not have enough resources to meet its current debt
a good or service on the one hand, and to pay for it on the obligations nevertheless has profitable investment opportu­
other. See complete contract, incomplete con tract, and nities but does not undertake these because its debt makes
implicit con tract. it unable to obtain financing or because the profits would
convertible bond: A corporate debt instrument which accrue to the creditors rather than to the borrower.
carries the right to convert the obligation to common (or, debt workouts: Negotiations between a debtor and creditors
perhaps, preferred) stock at a contractually specified ex­ designed to avoid bankruptcy by revising the creditors'
change rate. claims.
convertible preferred stock: Preferred stock with the right decentralization: See centralization.
to convert the claim to common stock at a specified rate. design problem: A decision problem in which there is a
cooperatives: A form of commercial organization in which large amount of a priori information about the relationship
only customers (or sometimes suppliers) are eligible to be among the variables in an optimal solution and in which
among the owners and any earnings are distributed in failing to achieve the desired relationship is costlier than
proportion to sales or membership or through price reduc­ other kinds of errors.
tions, rather than in proportion to the owners' investments. differential voting rights: A system of corporate govern­
core competencies of the firm: A form of economy of scope ance in which not all common stock has equal voting rights.
that arises out of a firm's ability to carry out some types of For example, recently acquired stocks might be accorded

Glossary 597
fewer votes than those that have been held for an extended m trust for employees as the basis for their retirement
period. Used to make hostile takeovers harder to win, mcomes.
because a majority of shares, if newly purchased, will not end-game problem: The difficulty that a person in a
carry a maj ority of the votes and so cannot alone unseat the relationship that is about to end may face different incentives
board. than in an on-going relationship and may be led to act
diversification: For an individual investor, the division of dishonestly or inefficiently.
invested wealth among a variety of different assets. For a endowment: In the competitive equilibrium model, the
firm, the operation in several different lines of business. In amounts of vari ous goods that a consumer owns initially,
either case, diversification is designed to reduce risk. before trade opens.
double marginalization: The tendency of two firms, one of equal compensation principle: The principle in incentive
which supplies the other, to set prices at such a high level contracting that if an agent is to allocate effort among
as each adds a profit margin to its own marginal cost that different activities, then each must bring the same marginal
their combined profits would be increased if the final price return to effort. Otherwise, the agent will focus exclusively
were lowered. on the one that yields the greatest impact on his or her
downstream: An activity that follows the reference activity mcome.
in the sequence of steps from producing raw materials to equity: ( 1 ) The value of real property in excess of any legal
delivering a finished product to the customer. claims against it for debts owed. (2) Resources contributed
economies of scale: The reduction in average cost that is to a firm in exchange for an ownership claim. (3) Securities
achievable when a single product is made in large quantities. issued by a firm that represent ownership rights (such as
economies of scope: The reduction in total cost that is stocks) or that are convertible into such securities (such as
achievable when a group of products are all made by a warrants).
single firm, rather than being made in the same amounts event study: A statistical and economic methodology under
by a set of independent firms. which the market's evaluation of the value created or
efficiency principle: The working hypothesis that organiza­ destroyed by some event or action is estimated by examining
tions and institutional arrangements that persist tend to be abnormal returns to assets whose values might plausibly
efficient ones. The logic is that if an arrangement is reflect the effect of the event or action. These abnormal
inefficient, then there are gains to be realized from changing returns are determined by comparing the actual price of an
it. asset around the time of the event with a statistical estimate
efficiency wage: Higher pay than required to attract and of what this price would otherwise have been.
hold workers in the particular employment, provided the excess return: In the capital asset pricing model, the expected
higher pay is designed to induce higher productivity. For return above the risk-free interest rate.
example, pay in excess of the recipient's market opportunities expected value (or expectation): The weighted average of
could encourage greater effort and productivity through the the possible realizations of a ra ndom va riable, where the
fear of losing quasi rents if the relationship is terminated for weights are the probabilities.
poor performance. externalities: Actions of one party that affect the welfare of
efficient: An allocation, contract, or organization is efficient others and are not mediated through markets.
if there is no feasible alternative one that everyone finds to financial leverage: The use of debt financing to increase
be at least as good and that at least one person strictly the resources available to the firm for a gi\'en level of equity
prefers. provided by owners.
efficient markets hypothesis: The hypothesis that prices in firm-specific capital: Human or physical capital that is less
securities markets, and particularly stock markets, fully and productive when it is used outside a particular firm.
accurately reflect all information relevant to forecasting Fisher separation theorem: The proposition that a decision
future returns. The weak, semi-strong, and strong form of maker \Vho is able to borrow and lend at the same interest
the efficien t markets hypothesis differ in the specification of rate can base investment decisions solely on a forecast of
what information is asserted to be embodied in prices. cash flows and interest rates, without regard to his or her
elimination tournaments: A sequence of contests with only preferences about the timing or composition of con­
the winners at each stage being allmved to compete at the sumption.
next. focus: The extent to which a firm limits itself to a relatively
employee-owned firm: A fir m in which the providers of few lines of business.
labor services hold at least a controlling interest. forward integration: Bringing a downstream activity, such
employee stock ownership plan: A form of pension plan in as distributing or selling a firm's product, under the owner­
which stock in the employing firm is purchased and held ship and management of the firm.
free cash flow: ( I ) Cash flow, plus after-tax interest expenses, horizontal integration: ( I ) In antitrust econom ics, expansion
less investments. (2) In Michael Jensen's free cash flow of a firm by acquisition of or merger with competitors. (2)
hypothesis, the amount of a firm's cash flow in excess of In business usage, an expansion into a related activity that
what can be profitably invested in the firm. does not involve vertical integration.
free-rider problem: See common-resource problem. hostile takeover: A change in corporate ownership that is
functionally organized firm: A firm in which the traditional opposed by the current management and board. Usually
functions, such as accounting, sales, manufacturing, and accomplished by buying a sufficiently large fraction of the
so on, are each controlled by a single department, in shares from the current stockholders to be able to control
contrast to the multidivisional fzrm. the election of board members.
fundamental theorem of welfare economies: The proposi­ human capital: Acquired skills and knowledge that make
tion that the allocation associated with a competitive an individual more productive.
equilibrium is efficient. human-resource management: The formulation and
gain :-sharing plans: Compensation plans designed to moti­ administration of human resource policies and of exceptions
vate employees by giving them some fraction of the returns to these.
to improved performance. human-resource policies: Policies relating to recruitment,
game theory: A general analytical approach to modeling employment, training, compensation, employee benefits,
social situations in which the information, possible actions, job assignments, promotion, and termination.
and motivations of the actors or players and how those Hurwiez criterion: A standard for measuring the amount
actions lead to outcomes are all specified in detail. In of information required by a decision-making or resource
contrast, the competitive equilibrium model does not specify allocation system. The standard measures the number of
what would happen if the demands of consumers exceeded variables which must be communicated at the last stage
the available supply. of the planning process to check whether the plan is effi­
general partners: The members of a partnership who have cient.
decision-making authority and who have unlimited personal idiosyncratic risk: The portion of any financial risk that is
liability for the partnerships' debts. independent of the total financial risk born in the economy.
general-purpose (human) capital : See non-specific asset. Also called unsystema tic risk, as contrasted with systematic
goal congruence: A situation in which the objectives of risk.
different individuals or organizations are sufficiently aligned imperfect commitment: Parties' limited abilities to bind
that they are led to pursue common goals. themselves to future courses of action, especially to bind
golden handcuffs: Compensation designed to reduce turn­ themselves to avoid opportunistic behavior.
over by paying exceptionally large amounts or by giving implementation problem: The mathematical problem of
substantial amounts of deferred pay that would be forfeited minimizing the cost born by a principal while still inducing
if the employee were to leave the firm. a self-interested agent to perform in a particular way. Also
golden parachutes: Contracts that promise large severance called a minimum cost implementation problem.
payments to employees ( usually senior executives) who lose implicit contracts: Shared understandings that are not
or leave their jobs shortly after a change in corporate control. legally enforceable but that the parties consider to be binding
greenmail : Payments by a corporation to a potential corpo­ on one another's conduct.
rate raider to induce him or her to give up an attempted incentive compatibility constraints: Limitations on the set
takeover. Usually involves a targeted share repurchase, in of contracts that can be implemented that arise from the
which the firm pays a premium over the market price to necessity of giving individuals appropriate incentives to
buy back the shares accumulated by the raider. induce them to adopt the desired course of action. These
hierarchy: ( 1 ) An idealized arrangement of authority in constraints are particularly important when there are infor­
which each person has only one boss and the organization ma tional asymmetries or incompleteness, so that individuals
has a single top officer. (2) A system of ranking employees. might misrepresent their private information or take unob­
holding company: A company which owns several other servable actions that are different from those desired by the
companies but exercises little or no management control other parties.
over them. incentive efficient: Efficient when the constraints implied
hold-up problem: The problem that one who makes a by the necessity of providing incentives are recognized.
relationship-speci fic investment is vulnerable to a threat by incentive intensity principle: The principle in incentive
other parties to terminate that relationship. This threat then contracting that the intensity of incentives should increase
permits these parties to obtain better terms than were initially with the marginal productivity of effort and with the agent's
agreed. ability to respond to incentives and should decrease with
Glossary 599
the agent's risk aversion and the vanance with which junior debt has no legal claim until the claims of more
performance is measured. senior debt have been met. Also called subordinated debt.
influence activities: Self-interested activities designed to junk bonds: Corporate debentures that are considered
influence others' decisions. Within organizations, these are particularly risky and so carry a high interest rate. More
often aimed at redistributing rents and quasirents and formally, bonds that are not rated as being of investment
take the form of political activity or misrepresentation or grade by one of the bond rating services (Moody's or
distortion of information. Standard and Poor's).
influence costs: The costs incurred in attempts to influence just-in-time manufacturing: A production system in which
others' decisions in a self-interested fashion, in attempts to inventories of goods in process are minimized because the
counter such influence activities by others, and by the required inputs to each stage of manufacturing are delivered
degradation of the quality of decisions because of influence. to each work station just as they are needed.
informational asymmetries: Differences among individuals LBO association: Michael Jensen's term for the collectivity
in their information, especially when this information is consisting of the management of a firm that has gone
relevant to determining an efficient plan or to evaluating through an LBO, the firm that arranged the transaction and
individual performance. purchased the bulk of the equity, the banks that provided
informational incompleteness: Lack of complete informa­ loans, and the various purchasers of the firm's debt obliga­
tion, especially when the actions of one party may not be tions.
observable by others. leveraged buyout (LBO): The purchase of the stock in a
informationally efficient: A system for communicating firm when the resources to finance the purchase come
information to support efficient resource allocation decisions mostly from borrowing that is debt of the firm. The result
is informationally efficient if no other system also results in of the borrowing means that the firm is highly leveraged.
efficient decisions and requires that less information be limited liability: The condition of a person whose liability
communicated. See Hurwicz criterion. for the debts of a partnership or other organization is limited
informational rent: A return in excess of opportunity costs to the amount of capital that the person has invested.
that accrues by virtue of an individual having access to limited partner: A partner in a limited partnership who
precontractual private information. The private information supplies financing and enjoys a share of the partnership
means that the individual must be given incentives not profits but who exercises no control of partnership decisions
to take advantage of the informational asymmetry, and and who has limited liability for partnership debts.
providing these incentives results in the rents. limited partnerships: A partnership consisting of both
informativeness principle: The principle of incentive con­ general partners and limited partners.
tracting that holds that payments under a contract should liquidate: Sell all the assets of a firm.
depend on the value of a variable if and only if accounting local nonsatiation: A property of preferences that means
for that variable allows a reduction in the error with which that near any consumption bundle there is another one that
performance is measured. is strictly preferred.
innovation attribute: The attribute of a decision problem management buyout: A purchase of a firm by its managers.
that the information required for an optimal decision is not Often these transactions are also leveraged buyouts because
directly available to people within the organization. the purchase was financed by heavy borrowing.
intensity of incentives: The rate at which expected income management by objectives: A technique for setting perform­
changes with improved performance under an incentive ance standards for employees in which the individuals have
contract. a role in identifying the objectives they will try to meet.
internal labor markets: A complex of administrative proce­ market failures: Failures of markets to achieve efficient
dure, and rules governing the allocation of labor, invest­ allocations. Sources include economies ofscale, externalities,
ments in it, and its cc,mpensation within an organization. and missing markets.
The key idea is that these procedures and rules supplant market for corporate control: Refers to the possibility of
the operations and directives of the outside labor market. changing corporate control through buying the stock of the
investment: An expenditure of resources that creates an firm in the securities markets.
asset. market portfolio: A portfolio that contains dollar amounts
job ladder: A sequence of jobs of increasing rank through of different assets in proportion to their aggregate values in
which employees climb by gaining promotions. the overall economy.
junior debentures: Debt that contractually has a lower market socialism: A hypothetical economic system in which
priority than other, senior debt in claims on the firm's assets the means of production would be owned socially (or by
in bankruptcy. When assets are inadequate to pay all claims, the state) but resources would be allocated by prices.
600 Clo!i!-IHf"\
 measurement costs: Costs involved in determining the net present value: The worth today of a cash Aow stream,
quality of a good or service that a party incurs to improve computed as a weighted sum of the cash flows in each
its bargaining position. future period, using weights that depend on interest rates.
menu of contracts: A system for compensation in whieh nexus of contracts: A nexus is a connected group. Armen
individual employees may choose whieh of several different Alchian and Harold Dcmsetz have identified the firm as
formulas will be used to compute their pay. such a connected group of (explicit or implicit) contractual
minimum cost implementation problem: See implementa­ relationships among suppliers, customers, and workers.
tion problem. nominal: Amounts measured in some currency, rather than
missing markets: A situation in which no market exists in in units of actual purc hasing power.
which to transact in a particular good or service. non-specific assets: Assets that arc equally useful when
modem manufacturing strategy: Any of a collection of employed in combination with any of various other assets
manufacturing strategies that seek to exploit the capabilities or in any of several different relationships. See specific asset,
of flexible equipment and rapid data communication and firm-specific capital, and cospecialization.
processing to increase the rate of new product introduction, normal form game: A list of players, their available strategies,
improve quality and reduce inventories and overhead costs. and the payoffs that each player stands to earn from each
Modigliani-Miller theorems: Two propositions in corporate possible combination of individual strategies.
finance theory. The first is the proposition that, but for not-for-profit organization: An organization (often a corpo­
taxes, the total market value of a firm's debt and equity ration) whose net proceeds from operations must remain in
would be independent of how much of the financing takes the organization and be used to advance the organization's
each of these forms. The second proposition is that the total objectives. Often used for organizing the provision of
market value of the firm's debt and equity is unaffected by charitable activities. A not-for-profit has no owners in the
its dividend policy, provided its investment policy is held sense of residual claiman ts.
fixed. no wealth effects: The condition on preferences that means
monitoring: An activity whose aim is determining whether that choices among non monetary alternatives are unaffected
the contractual obligations of another party have been met. by the individual's wealth or income.
monitoring intensity principle: The principle of incentive one-fund portfolio theorem: The conclusion that, for
contracting that indicates that more resources should be investors who care only about the mean and variance of
used to reduce the errors in measuring performance when the returns on their investments, the optimum portfolio
stronger performance incentives are being given. always consists of shares of a riskless asset with the balance
moral hazard: Originally, an insurance term referring to invested in predetermined proportions among the other
the tendency of people with insurance to reduce the care securities. These predetermined proportions are the "one
they take to avoid or reduce insured losses. Now, the term fund" that underlies every efficient investment portfolio.
refers also to the form of postcontractual opportunism that opportunistic behavior: Self-interested behavior uncon­
arises when actions required or desired under the contract strained by morality.
are not freely observable. option: A financial contract giving a right which need not
multidivisional organization: A form of firm organization be exercised unless the holder chooses to do so. See call
in whieh there are multiple business units with autonomy option and put option .
to make day-to-day operating decisions and with control over Pareto dominated: A situation from whieh it is possible to
their own functional departments (accounting, marketing, increase strictly the welfare of some party without diminish­
etc). Planning and coordination among these units and ing that of any other party.
performance evaluation of the unit managers are the Pareto optimal: A situation from whieh it is impossible to
responsibility of a central office. increase the welfare of any party without decreasing that of
mutual insurance company: An insurance company that is some other party.
legally owned by its policyholders. participation constraints: Limitations on contracts or other
Nash equilibrium: A strategic situation in which each organizational arrangements arising from the fact that
decision maker's planned strategy is best from his or her participation is voluntary and so individuals must expect to
point of view in light of the strategies that he or she expects do at least as well as under their next-best alternatives or
others to employ, and these expectations are correct. they will refuse to participate.
neoclassical market model: A model of market exchange in participatory management: A policy of allowing employees
which utility-maximizing consumers and profit maximizing to participate in all important management decisions,
firms transact at prices over which each party perceives itself usually accompanied by arrangements to ensure goal con­
to exercise little control. gruence.
Clossar\' 601
partnership: A form of organization in which some or all strategies results in an outcome that is Pareto dominated by
of the multiple owners, the general partners, accepted an outcome that would be achieved under some other
unlimited liability for the organization's debts and exercise (dominated) strategies.
management control . private corporation: A corporation whose common stock is
perfect capital markets: A theoretical ideal in which all not offered for sale to the general investing public.
individuals can borrow and save on the same terms, with private information: Information that is relevant to deter­
these terms being unaffected by the amounts involved . mining efficient allocations that is known only to some
piece rate: The amount paid for each unit of a product that subset of the parties involved .
is produced . Also the system of compensating people in privately held corporation: See private corporation.
proportion to the amounts they produce. privatization: The transfer of firms from government to
poison pill: A takeover defense that greatly reduces the private ownership.
value of the firm in the event of a hostile takeover by giving production plan: A listing of inputs to be purchased and
stockholders a right to acquire shares in the firm (or some outputs to be sold by a firm.
other financial claim on the firm) at a greatly reduced price profit sharing: An element of a compensation plan in which
in the event of a control change. employees receive bonuses that are in aggregate tied to firm
postcontractual opportunism: Opportunistic behavior by a profits.
party that takes place after a contract is signed. Moral hazard proxy fight: A contest among competing parties to gain the
and the hold up problem are two particular problems of proxies of a majority of a corporation's shareholders, giving
postcontractual opportunism. the successful party the right to vote the shares in a
posting of bonds: Setting aside a sum of money (a bond) to shareholder vote. Usually launched as a challenge to
guarantee performance under a contract. management (which normally gets the proxies of sharehold­
precontractual opportunism: Opportunistic behavior by a ers who do not attend shareholder meetings) and the current
party that takes place before a contract is signed . Adverse board of directors.
selection is a problem of precontractual opportunism. public corporation: A corporation whose shares are bought
preferred stock: Securities issued by a firm which, like and sold through an organized exchange and so may be
common stock, need not be redeemed by the corporation held by any investor.
but which do not carry voting rights and must be paid public-goods problem: See common-resource problem.
dividends before any dividends can be paid on common publicly held corporation: See public corpora tion.
stock. put options: A financial contract that gives the buyer the
present value: See net present value. right (which the buyer may choose not to exercise) to sell
principal: ( 1 ) The party whose interests are meant to be a particular asset at a specified price on a certain date (a
served in an agency relationship. (2) The initial amount of European put) or on or before a certain date (an American
a loan or investment. put).
principle of risk sharing: The principle that when many quality circles: Groups of workers who meet to identify and
people share in a number of independent risks, with each implement ideas for improving the quality of their product.
taking a small part of each risk, the total cost of bearing the quasirent: A return in excess of the minimum needed to
risk is reduced . keep a resource in its current use.
principle of unity of responsibility: The principle that final random variable: An amount that depends on the outcome
responsibility for all the jobs needed to accomplish a of an uncertain event.
particular task should reside with a single person. This ratchet effect: The tendency of performance standards in
policy both clarifies \\'hat might otherwise be ambiguous an incentive system to be adjusted upwards after a particu­
responsibilities and makes it easier to evaluate performance larly good performance, thereby penalizing good current
and provide incentives. performance by making it harder to earn future incentive
principle of value maximization: The principle that in the bonuses.
absence of wealth effects an allocation is efficient if and rf'al: Real, as opposed to nominal amounts, represent actual
only if it maximizes the total value obtained by all of the purchasing power by allowing for price changes.
parties. relational contract: A contract that specifies only the
prisoners' dilemma: A strategic situation modelled as a general terms and objectives of a relationship and specifies
normal form game in which each party or player has a mechanisms for decision making and dispute resolution.
dominant strategy-one that is best no matter what behavior renege: Deliberately choose not to carry out a promise or
is expected from the others-but play of these dom inant contract to the detriment of the other party.
renegotiate: Bargain to determine new terms to replace of sales is expected to respond to changes in certa in variables,
those of an existing contract. If no new agreement is such as the salesperson's efforts.
reached, the previous one remains in force. Scanlon plans: An incentive plan in which workers are paid
rent: A return received in an activity that is in excess of the a bonus if the ratio of labor cost to sales is less than some
minimum needed to attract the resources to that activity. speci fied target, with the bonus being in proportion to the
reorganization: Redesign of the relationships, authority, costs saved.
responsibilities, and lines of communication in an existing scorched earth policies: A kind of takeover defense that
organization. In a reorganization in bankruptcy, the bank­ deliberately reduces the value of a firm to discourage hostile
rupt firm renegotiates the amount of its debts and the timing bidders.
of its payment obligations and may also reorganize its screening: Offering a menu of contracts or options with the
activities to allow continued operation. intention of encouraging self-selection.
reputation: The view formed of an individual or organiza­ secured debt or loan: A debt for which a particular asset
tion by another based on past experience, especially as a has been pledged as security. If the borrower docs not meet
basi� for forecasting future behavior. the terms of the contract, the lender can lcgal.ly seize that
residual claimant: One who is entitled to receive the residual asset.
return of an asset. selective intervention : The management practice of
residual return: Income from an asset or business that allowing divisions or businesses to operate with nearly
remains after all fixed obligations are met. complete independence, intervening only to correct particu­
residual right of control : The right to make any decision lar problems characteristic of independent firms.
concerning an asset's use that is not explicitly assigned by self-selection: The pattern of choices that individuals with
law or contract to another party. different personal characteristics make when facing a menu
resource allocation problem: The problem of using li mited of contracts or options. For example, workers who face a
resources efficiently or in a way that maximizes some fixed choice between a job with a fixed hourly wage and one
objective. with piece rate incentives will tend to prefer the former if
restructu ring (financial): Changing the financial structure they expect to be relatively unproductive and to prefer the
of the firm. During the 1 980s restructuring typically was latter in the opposite case.
designed to reduce the free cash fl-ow and thereby improve semistrong form efficient markets hypothesis: The hypothe­
managerial incentives or to make the firm a less attractive sis that stock prices (or those of other assets) fully reflect all
target for corpora te raiders, or both. publicly available information. This hypothesis implies that
revelation principle: The principle holding that any out­ unless an investor has inside information, he or she cannot
come that can be achieved by some mechanism under the expect to earn better rates of return (on a risk-adjusted basis)
self-interested strategic behavior that is induced by the from any active investment policy than from one of simply
mechanism can also be achieved by a mechanism employing buying and holding a fully diversified, market portfolio.
an honest mediator to whom the parties willingly report senior debt: Debt which has priority over other, more junior
truthfully and who then implements the outcome that debt in its claims on the firm's assets in bankruptcy.
would have resulted from the original mechanism. separation of ownership and control: A situation in which
risk aversion: The preferring of a sure thing to a risky the residual returns and residual rights of control belong to
outcome with a somewhat higher expected return. different parties. More particularly, the common condition
risk neutral: The characteristic of a person who is indifferent of modern corporations in which the shareholders are the
between receiving a fixed sum of money or a risky prospect residual claimants but effective control of decision making
with an expected value equal to the fixed sum. lies with the top managers of the firm.
risk premium: The excess of the expected value of some separations: Quits, firings, and permanent layoffs.
random income over its certain equivalent. shareholder rate of return: The rate of return on holding a
risk tolerance: A measure of willingness to bear risk. stock over a period, calculated as the change in share price
Measured as the inverse of the coefficien t of absolute risk plus any dividends received, all divided by the initial price
aversion. of the shares.
routines: Standardized rules for decision and action that, short selling: The practice of borrowing an asset in order to
although they may vary to a limited degree with the sell it and later repurchasing a similar asset to repay the
particular circumstances, are applied across a period of time loan.
without further fine-tuning. signaling: Acting in a way that demonstrates to others the
sales response function: A relation indicating how the level actor's intentions or abilities or some other characteristic

Glossary 603
about which the actor has private, unverifi able informati on. the remaining activities. Strong complementarities lead to
For example, a worker who accepts a job that begins with design decisions.
several weeks of unpaid, specific training signals an i ntention supermajority rules: A requirement in many antitakeover
to work for the firm for a long period of time. charter amendments that any change of control be approved
specialization: ( 1 ) The division of tasks on the basis of by more than a si mple majority of the voters. For example,
comparative advantage. (2) The process of narrowing (and , a 2h or ·¼ majority may be required .
presumably, deepening) the range of tasks that a particular synchronization problem: The problem of arranging the
individual or mach ine can perform . ti ming of activities when close coord ination is required .
specific investment: An investment that creates a specific systematic risk: The portion of the variance of return on
asset. an investment that varies di rectly with returns on a market
specific assets: Assets whose val ue is much greater in index. In the capital asset pricing model, the systematic
a particular use or rel ationship than i n the next-best risk of an investment is measured by its beta. See also
alternative. idiosyncratic risk.
specificity: The extent to wh ich assets are specific. When takeover premium: The amount by which the price paid
the principle of value maximization holds, specificity is for a firm's shares in a takeover exceeds the total market
measured by the loss in value entailed in shifting the asset val ue of the firm's shares in the absence of a takeover.
to another use. team production: A production process in which the
spin off: The creation of a separate corporation whose shares individual outputs cannot be separatel y identi fied . For such
are distributed to the original firm's stockholders and into a process, any individual incentives must be based on some
which certain assets are placed . measure of the effort or d iligence of the workers.
spot market contract: A contract for the im med iate ex­ technical analysis: The practice of attempting to identify
change of goods or services at cu rrent prices . patterns in asset price movements for use in forecasting
staggered board of directors: See classified board. future price movements.
stakeholder: Any individual or group who has a di rect tenure: The condition of being protected from termi nation
interest in a firm's continuing profitabl e operations (in­ of empl oyment in a job, regardless of general performance,
cluding stockholders, lenders, employees, customers, sup­ subject only to meeti ng certain minimal standards of
pliers, communities where the firm employs workers, and acceptable behavior.
so on). tournament: A contest in which the pri zes received depend
state of incorporation : I n the Un ited States, the state in only on ordi nal ranking (first, second, third , etc. ) and not
which a company is legally established and whose laws on absolute performance. A policy of promoti ng the person
govern the relationship between the company and its who is judged best qual ified creates a tournament, as do
stockholders . Need not be the location of the headquarters sales contests in which a prize is given for achievi ng the
of the firm . largest sales vol ume.
statistically independent: A condition holding between tragedy of the commons: See common-resource problem.
random variables under which knowledge of the realized transaction: ( 1 ) An exchange involving goods, services, or
value of one vari able gives no information about the money. (2) The l argest unit of economic activity that cannot
probability of different real izations of the other. be subdivided and performed by several different people.
stock options: In executive compensation, options given to transaction costs: Costs of carrying out a transaction or the
the employee by the firm to buy its stock for a specified opportunity costs incurred when an efficiency-enhancing
price during some specified period . transaction is not realized .
strategic investments: ( 1 ) Investments whose benefits par­ transfer prices: The prices used for transactions among
ti ally accrue to parts of the organization not i nvolved in departments or divisions with in a firm.
making the i nvestment. (2) Investments made to create or unlimited liability: The cond ition of a person whose l iabil ity
demonstrate comm itment and thus alter the behavior of for the debts of a partnership or other organ ization is not
competitors or allies. lim ited to the amounts he or she has invested .
strong form efficient markets hypothesis: The hypothesis unsecured: An unsecured loan or debt is one which no
that stock prices (or those of other assets) fully reflect all speci fic asset is pl edged to support repayment.
information , including both publ icly available information unsystematic risk: See idiosyncratic risk.
and informa tion held by insiders. up-or-out rule: An empl oyment pol icy under wh ich em­
strongly complementary: A group of activities is strongly pl oyees who are not promoted or made partners must leave
complementary when increasing the levels of some activi­ the organization.
ties in the group grea tly increase the marginal returns to upstream: An activity that precedes the reference activity
(,04 Clo-,-,arv
in the sequence of steps from producing raw materials from weak form efficient markets hypothesis: The hypothesis
natural resources to delivering a finished product to the that current stock (or other asset) prices reflect fully all the
customer. information that is embodied in past prices. An implication
utility function: A numerical representation of an individ­ is that observing past prices cannot help in forecasting future
ual's preferences over different possible choices or situations. prices. In particular, price changes should not be correlated
variance: A mathematical measure of the amount that a over time and should not display any predictable patterns.
random variable is likely to vary around its mean value. This implies that technical analysis cannot be the basis of
The variance is equal to the expected value of the square a profitable investment strategy.
of the difference between the variable and its mean. wealth effects: The variation in the amount a consumer is
venture capitalist: An investor who specializes in providing willing to pay for some object or in the quantity that the
equity funding to new ventures, often also providing advice consumer may wish to buy at a particular price as a result
and practicing active monitoring. of a change in the consumer's wealth.
vertical integration: Bringing two or more successive stages white knight: A merger partner or acquircr sought out
in production and distribution under common ownership by the target's management or board to thwart a hostile
and management. takeover.
warrants: Rights issued by a firm (often in connection with winner's curse: The tendency of the winning bidder in a
the issuance of new bonds) entitling the holder to buy contracting competition to be one who has underestimated
common shares from the firm at a specified price on or the cost of doing the job, because bidders who overestimate
before a specific date. In contrast to call options, warrants, the cost usually bid too high. Similarly, the tendency of
when exercised, lead to an increase in the number of shares the winning bidder in an auction to be one who has
of the firm that are owned by private investors. overestimated the utility of the object being sold.

Glossary 605
 INDEX

Abegglen, James C., 275n, 5 1 9n Apple Computer, 1 33, 1 39, 309, 424, supplier relations in Japanese, 565-69
Abowd, John, 442 576, 589 See also specific companies
Abraham, Katherine C. , 337n, 405 Applicants, attracting, 34 1 Automobile insurance, 22 1
Accounting, 540 Applied Materials, Inc., 399 Autonomy from intervention, 2 1
Accounting rates of return, 438 Aquifers, ownership rights to underground, Avis, 4 1 , 41 5
Accumulation factors, 45 1-52 296-97 Axelrod, Robert, 264
Ace, 562 Arbitrage, 2 1 1 Azariades, Costas, 3 54
Adverse selection, 1 29, 149-54 Architecture of organizations, 20-2 1
closing of markets and, 1 50-5 3 Arrow, Kenneth, 5 1, 62, 85, 1 96, 24 1, B.A. T. Industries, 484
mathematical example of, 1 5 1-53 3 54 Backward integration, 54 1
moral hazard vs., 1 69 Asanuma, Banri, 3 1 0n, 566n Baily, Martin, 3 54
in principal-agent problem, 237-39 AskComputer Systems, 579 Bain & Co., 408, 409-1 0
rationing and, 1 53-54 Asset(s) Baker, George P. , 162, 3 88, 406n, 421,
See also Precontractual opportunism book values bf, 485n 445
Agent(s) cospecialized, 13 5-36, 1 39, 307, 3 1 0, Bakker, Jim, 524
incentive contracting and, 1 98-20 l 3 19 Baltimore Canyon, drilling rights in, 429
managing distant trading, 2 52 defined, 1 34 Bankruptcy, 183-84, 502-5
risk averse, 222 financial, 1 3 5 Chapter 7, 5 10
risk-neutral, 236-39, 249 information and prices of, 467-75 Chapter 1 1 , 278-79, 503 , 504, 5 1 0
See also Principal-agent problem physical, 1 34-3 5 control structure of corporation and, 5 1 0
Age/wage profiles, 1 57, 1 89, 3 3 5-37 specific, 30-3 1, 1 3 5-39, 307-8 end game problem and, 266
Aghion, Philippe, 1 96, 241 in franchise retailing, 564-65 Banks, monitoring incentives of, 501
Agreement, costs of, 30 1-2 stakeholders' rights and, 3 1 6-19 Bank syndicates, 540
Aguilar, Francis, 54 3n strategic asset destruction, 504-5 Bannock, Graham, 5 1 2n
Air traffic controllers, moral hazard and Asset ownership, 249, 307- 1 9 Bargaining, 140-4 5
shirking among, 1 79-8 1 complex assets, 3 1 3-19 incentive efficient, with private informa­
Akerlof, George, 1 52, 162, 3 54, 373 incentive pay and, 23 1-32 tion, 143-45
Aktiengesellschaften, 490 predicting, 307-1 3 over a sale, 140-4 3
Alchian, Armen, 20, 5 1 , 1 37n, 161, 292- Assignment problems, 91-92 Bargaining costs, 1 47, 300-30 1, 304-5
93, 3 2 1, 322, 3 54 Associated Press, 562 Bargaining position, investments in, 1 49
Alexander, Cordon, 460n, 478 Atchison, Thomas, 42 1 Baron, David, 24 1
Alliances, business, 575-80, 586 AT&T (American Telephone and Tele­ Baron, James, 3 54
Allocation graph), 81, 548, 557, 571, 572 Barze), Yoram, 5 1 , 32 1
Pareto dominated, 23 Auctions, 5 1 4 Base pay, determinants of managerial,
Pareto optimal, 23, 66 Authority 427-28
resource . See Resource allocation delegation of, 1 7, 544 Batch size, optimal, 1 1 1, 112
Alza Corporation, 50 l in employment relations, 330-3 1 Bayer AC, 490n
Analog Devices, Inc., 366, 414- 1 6 i n franchise retailing, 565 Beatrice Corporation, 5 1 1
Anderson, Erin, 5 59n Authority relation, 2 59 Bebchuck, Lucian, 5 3 1
Antitakeover statutes, 5 1 8 Automobile industry, 1 37 Becker, Cary, 322, 3 54
Aoki, Masahiko, 3 17, 3 54, 362n, 385 organizational strategies in, 2-6 Beer, M. , 1 1 6n

607
Beha\'ior sources of, 1 30-3 1 Centralization, 544
consumer, 64 defined , 1 28 communication and , 9 3
modeling human, 42-43 Bowles, Samuel , 2 5 6-57, 282 defined , 1 1 4
Belcher, David, 42 1 Boycotts, 268-69 Central planning, 1 3- 1 4, 72-7 3
Bell Laboratories, 548 Bradley, Allen, 587 CEO. See Chief executive officers (CEOs)
transfer pricing in Bellcore, 8 1 Bradley, M ichael, 5 1 1 n Certainty equivalent, 2 1 0, 246-47
Bell South, deferred compensation and re- Brand names, 2 59 Cha-Kyung, Koo, 543n
tention at, 3-+7 Breach of trust, takeover premia due to, Chandler, Alfred , Jr. , 4n, 73, 88, 540n ,
Benetton , 565, 587 5 1 3- 1 5 583
Bergman, Yaacov, 5 3 1 Brealey, Richard, 478 Chapter 7 bankruptcy, 5 1 0
Berkeley Co-op, 563 Breeden , Douglas, 477 Chapter 1 1 bankruptcies, 278-79, 503,
Berkshire Hathaway, 500 Bresnahan , Timothy, 478 504, 5 1 0
Berle, Adolphe, 1 8 1 , 1 96 Brickley, James A. , 4 3 5 n , 5 3 1 , 5 54n, Chari, V. V. , 1 62
Bernheim, B. Douglas, 454n 563n, 566 Charitable organizations and activities,
Bernoulli, Daniel , 24 1 Bridgestone Tire, 457, 483 , 5 1 1 5 2 3-27
Besanko, David, 24 1 British Petroleum pie, 490n Chernobyl nuclear reactor accident, 1 5n
Beta, 46-+ Brittleness of coordination system, as- Cheung, Stephen, 5 1
Better Business Bureau, 268 sessing, 94- 1 00 Chevron Oil , 48 3
Bhagat, Sanjai, 4 3 5 n , 484n, 5 1 2n, 5 1 4- 1 5 Brooks Brothers, 484 Chidambaram, A. C. , 4 1 0
Bhattacharya, Sudipto, 1 62, 477, 5 3 1 Bruck, Connie, 399n, 489 Chief executive officers (CEOs)
Bias i n promotions, 365 Buchanan , James, 283 compensation of, 423, 424-27, 428,
Bidding, competitive, 1 00, 5 56, 5 58, 567 Bud's Ice Cream of San Francisco, 278 43 3-43
Bieber, Owen, 423 Bulow, Jeremy, 1 1 9 adequacy of, 440-4 1
Biotechnology, 429 Bureaucratic expansion , tendencies toward, intensity of performance incentives,
Black, Fisher, 478 427-28 438-40, 442
Bloom ingdales, 484 Burrough, Bryan, 494 international differences in, 42 5-26
Blue Diamond, 562 Burroughs, 574 in large U . S . fi rms, 424-2 5 , 426
Board of directors, 1 83 , 496-97, 508 Business all iances, 5 7 5-80, 586 long-term incentive plans, role of,
classified or staggered, 5 1 7 Business firm. See Firm(s) 426-27
corporate governance reforms and, 5 2 1 Business judgment rule, 279, 3 1 4 performance pay, 43 3-4 3
directors' liability, 279 Business school admissions, 342 setting, 4 3 3-34
Japanese, 3 1 7 Business strategy. See Strategy(ies) in smaller firms, 426
ownership of corporation by, 3 1 4- 1 5 Business Week, 424 holdings of thei r companies' stock, 49 5
setting CEO pay, 43 3-3 5 Bust-ups, 484-85 promotion to, 376
Boards of trustees, university, 5 2 5-26 Buyers, coordination costs of, 29 China, ownership problems in, 297
Boeing Company, 429, 579 Byrne, John A. , 424n Cho, Dong-Sung, 543n
Boesky, Ivan, 488-89 Chrysler Corporation, 337
Bognanno, M ichael L. , 368 Californ ia, water rights and usage i n , 297, Chung, Kwang S . , 1 82 n , 48 5n, 506n,
Bolton , Patrick, 1 1 8, 1 96 298-300 5 1 7n , 5 3 1
Bond, Alan, 484 Callen, Jeffrey, 5 3 1 Cipolla, C . M . , 258n
Bondholders, monitoring i ncentives of, Call options, 457 Claimants
501 naked , 1 74 in bankruptcy, hierarchy of, 504
Bonding, 1 89-90, 522-2 3 CalPERS, 5 2 1 residual, 29 1
Bond markets, 540 Campeau, Robert, 484 Claris Corporation, 309
Bonds Capital Classical financial economics, 448-8 1 , 482
convertible, 4 5 7 , 490 allocation of i nvestment, by markets, capital asset pricing model (CAPM ),
junk, 1 7 1 , 48 5 , 488-89, 5 1 1 , 520 459-60 464-66, 472
Bond trad ing, 1 0 costs of, 452-54 of financial structure decisions, 4 5 6-60
Bonus systems, I I , 3 89, 4 1 4, 425 human . See Human capital allocation of investment capital by
Bonwit Teller, 484 venture, 500- 50 I markets, 459-60
Book values, 48 5 n Capital asset pricing model (CAPM), 464- Modigl iani-Miller analyses, 448, 4 5 6-
Borch, Karl , 24 1 66, 472 59, 49 1 , 5 0 5
Borrowing Capital gains tax, 507 information and prices of financial assets
in employment relations, 3 38 Capitalism, 592-93 in, 466-7 5
incentives for risk taking by, 1 7 3-7 5 Capital markets efficient markets hypothesis, 467-7 1 ,
im ·estments financed by, 450 classical theory of, 459 ·473_74
monitoring of borrowers, 1 74-7 5 perfect, 450-5 1 , 46 1 shortsighted markets and manage­
Boundaries. See Horizontal scope and Capital structure. See Financial structure ment, 47 1-73, 474
structure; Vertical boundaries and Cardinal information , 367 of investment decisions, 448, 449- 5 5
relations Career paths, 258, 5 8 1 Fisher separation theorem, HS, 449-
Bounded rationality, 1 29-40, 1 59 Carmichael , H . Lome, 1 96, 3 8 5 51
barga i 1 1ing costs and, 1 47 Carrier, 4 1 4 net present values in, 4 5 1- 54
contractual incompleteness and, 1 29-40 Cartels, 540 strategic investments as design deci­
achieving commitment despite, 1 39- Cash flow, 450 sions, 454-5 5
40 cost of capital and required, 4 5 3 investment risk and return in, 460-66
effects of, I 3 3-34 free, 492-94, 496, 507 Classical firm, 293
in im-c\hnent\ a11d specific assets, net present value of, 4 5 1 - 54 Classical theory of wages, employment and
1 34-39 Casio, 107 human capital, 327-29
respomes to, 1 3 1 - 32, 1 39 CBS Records, 280, 483, 570 Classified boards, 5 1 7

608 l 111h
Clayton & Dubilier, 500 incentives for individual performance, Consumer(s)
Clive, Lord, 2 52, 2 54 1 57-58, 391-403 in neoclassical model of private owner­
Coase, Ronald, 28-29, 34, 5 1, 88, 3 21, deciding what to motivate, 391-92 ship economy, 62-63
5 51 eliciting employees private informa­ resource endowment of, 63, 64, 6 5
Coase theorem, 38, 39, 293-306, 346 tion, 400-402 Co11sumer behavior, 64
bargaining costs and limits of, 300-30 I explicit incentive pay, 187-88, 206-9, Consumption plan, 63-6 5, 69
ethics of private property, 30 5 392 Contavcspi, Vicki, 44 3n
ill-defined property rights and tragedy of implicit incentive pay, 402-3 Continental Airlines, 5 54, 5 5 5, 570
the commons, 294-97 managerial performance pay, 400 Continental (German tire firm), 491
legal impediments to trade, 301-3 pay for skills, 399 Contract(s)
transaction costs and efficient assignment piece rates, 236, 388, 392-96 complete, 127-29, 289
of ownership claims, 303- 5 sales commissions, 2 1 6-17, 3 96-99 efficiency wage, 363
untradable and insecure property rights, for scientists and engineers, 399-400 financial, moral hazard in, 183-8 5
297-300 job design and incentive pay, 4 1 0-11 implicit, 1 32, 139-40, 332-33
Coefficient of absolute risk aversion, 210, model of incentive pay, 215- 1 9 incentive. See Incentive contracts
213 objectives of policy, 390-91 labor (employment), 131-32, 157, 329-
Coercive offers, 515 optimal policies for, 378-79 33
Cold War, 1 2, 13 pay attached to jobs, 360-62, 369-71 menus of, 401-2
Collateral, 183. See also Bankruptcy; pay equity and fairness, 2 74-75, 418- 1 9 screening and, 158- 5 9
@Debt performance evaluation and, 403-8 performance, 179
Columbia Pictures Entertainment, 280, with explicit performance pay, 403-4 problems of actual contracting, 1 28-29
483, 5 70 in subjective systems, 404-8 relational, 131-32, 139-40, 2 5 9, 330
Comment, Robert, 487 responsibility and, 258 reputation as enforcer of, 259-69
Commercial practicability doctrine, 130-31 sources of randomness in performance, spot market, 131
Commissions, sales, 216-17, 396-99 207-8, 220-21 uncertainty and complexity of transac-
Commitment, 13 5 See also Incentive contracts tion and, 3 1 -32
contractual incompleteness and, 1 3 3 Competencies, core, 107-8, 554, 570-71 Contract feasibility, 218- 1 9
deferred compensation and problems of, Competition Contracting, organizations and, 20
43 outside labor market, promotions and, Contractual incompleteness, bounded ra-
imperfect, 30, 129, 234, 236 364-66 tionality and, 1 29-40
noncontractual means of achieving, for rents, horizontal expansion and, achieving commitment despite, 1 39-40
1 39-40 5 73-74 effects of, I 33-34
promotion policies and problems with, in S&L industry, perverse effect of, 175 in investments and specific assets, 1 34-
379 Competitive bidding, 1 00, 5 56, 5 58, 567 39
Common law, English, 277 Competitive (corporate) strategy, organiza­ responses to, I 31-32, 139
Common-resource problem, 294-98 tional innovation and, 569 sources of, 1 30-31
Commons, John, 5 1 Competitive equilibrium, 66-67, 69-71, Control, corporate. See Corporate control
Communication 74, 1 0 2 Convertible bonds, 4 5 7, 490
in centralized v. decentralized systems, Complementarity(ies), 11 5, 5 87 Convertible preferred stock, 457
93 asset ownership and, 312 Cook, Clive, 1 4n
controlling influence by limiting, 273- coordination failure and, 11 1 -12 Cooper, Russell, 85, 1 1 9
74 design attributes and, 108- 1 0 Cooperation, group incentives and, 4 1 6
coordination and economizing on, I 00- of discretion (control) and incentives, Cooperatives, 26-27, 413, 562-63
1 06 412- 1 3, 549 Coordination, 88-124
Hurwicz criterion as measure of require­ innovation attributes and, 112- 1 3 brittleness of system of, assessing, 94-
ments of, 10 1-2 momentum and, 543 100
See also Coordination between tasks, job enrichment and, 411 business strategy and, 106-13
Communication technologies, 587, 589 between theory and institutional knowl- design, 3
Communism, 12-16, 592-94. See also edge, 594 economizing on information and com­
Eastern Europe, economic restruc­ Complements, 1 7 munication and, 1 00-1 06
turing of; Soviet Union Complete contracts, 1 27-29, 289 in franchise retailing, 565
Comparative performance evaluation, 220- Complex assets, ownership of, 3 1 3-19 horizontal expansion and problems of,
21, 233, 404 Complexity 571-73
Compensation, 388-446 asset ownership and, 308- 1 0 management's role in, 114-16
balance between risk and incentives in, of transaction, 31-32 in market for medical interns, 4 3-48, 77
207, 208-9 Comprehensive damage coverage, 22 1 in multidivisional firm, 545, 547
deferred, 346, 347, 432-33 Computer-aided design and manufacturing organizational methods for achieving,
equal compensation principle, 228-32, (CAD-CAM) programs, 588 26-28
311, 363, 394, 397, 403, 4 1 2, 44 1 , Computer hardware and software, 309 price and, 2 7, 57-62, 89
47 1 , 473, 5 5 8 Concentrated shareholding, 497-500 price- vs. quantity-based coordination
executive and managerial, 389-90, 423- Conglomerate firms, voluntary restructur- system, 94-100
46 ing of, 519-20 resources used in, 90
CEO, 423, 424-27, 428, 433-43 Conglomerate mergers and acquisitions, specialization and, 2 5-26, 56-57
middle-level executives, 4 27-29 486 with outside suppliers, 5-6
motivating risk taking, 429-32 Connectedness task of, 2 5-28
patterns and trends in, 424-29 asset ownership and, 312- 1 3 variety of problems and solutions in, 90-
forms and functions of, 388-91 of transactions, 32-3 3 94
group incentive pay, 4 1 3-18 Consensus decision making, 3 5 I , 431, 44 3 vertical integration and, 5 56-5 8
as incentive, 8, 9, I 0-11 Constant returns to scale, 99-100 Coordination costs, 2 9

l11cb 609
Cooter, Robert, 321 Covenant House, 524 Deposit insurance program for S&Ls, 170-
Copa�'ments , 207, 221 Crawford, Robert, B7n , 161, 322 76
Core competencies, 1 07-8 , 5 54 , 570-72 Credit, trade , 278-79 Derivative securities, 4 59n
Corporate control, 482-5 37 Credit bureaus, 268 Desai, Anand, 5 1 I n
alternati,·es to publicly-held corporation, Creditors Design
521-27 in Chapter 11 bankruptcy, 278 , 503 design decision
charitable and not-for-profit organiza­ classes of , 279n formal model of , 12 1-24
tions, 523-27 conflicting interests of shareholders and, strategic investments as , 454-5 5
partnerships, 522-23 494-96 job , 408-13
changes in, 483-505 corporate control of, 508 organizational, 115 , 190-92, 273-
current lenders' vs . other capital sup- See also Debt financing 77
pliers interest and, 494-96 Credit rationing, 154 Design attributes , 91 , I 06
debate over, 487-89 Cremer, Jacques, 283 complementarities and , 108- 1 0
default and bankruptcy costs of, 502-5 Crop insurance policies , 177 production planning with, 103-6
incentive and rights aspects of finan- Crown Zellerbach, 484 strategic , 454-5 5
cial structure , 491 , 505 Crystal , Graef , 439, 442 Design connectedness , 33
international patterns of financing and Culture, corporate , 1 0, 11 , 265 , 547, 574 Design coordination , 3
ownership , 489-91 Cycle of poverty , 250 Design problem with innovation attributes,
managers' vs. owners' interests and , Czechoslovakia, 306 592
491-94 Deutsche Bank , 490-91
monitoring incentives for lenders, Developing countries
501-2 Dana Corporation , 414 debt overhang and underinvestment in,
monitoring incentives for owners , Dark , Frederick H. , 554n , 563n , 566 496
496-501 Davis, Evan , 512n production planning in , 75
in 1980s, 483-87 Dayton Hudson, 518 Dewatripont , Mathias , 24 1
rise of debt, 485-87 Dean Witter Reynolds, 575 Diamond, Peter, 85
innovation in, 590-91 De Beers diamond monopoly, 148 Differential voting rights , 517
market for , 508-9 , 520 Debentures , types of, 457 Directors . See Board of directors
mechanics of , 509-10 Debreu, Gerard, 62, 85 Disability insurance, 179-81
in multidivisional firm, 545, 549 Debt-equity ratio, 490 Discretion , 21, 412-13 , 549. See also Cor-
separation of ownership and, 181 Debt financing porate control
signaling and financial decisions, 505-8 classical analyses of financial structure Discrimination, employment, 343-44
takeovers and restructurings in U . S. in decisions and , 456-60 Distribution, efficiency and, 248-50
1980s, 483-84, 5 10-21 conflict of interest between equity and, Distributional implications of decisions,
aftermath of, 520-21 183-84 limiting , 272-75
defenses against, 515-20 excessive risk taking and high ratio of, Diversification of I 960s , 486
Corporate culture, 10 , 11, 265, 547, 574 495 Diversity, product , 3
Corporate governance reforms, 520-21 free cash flow and, 496 Dividends, 459, 492, 507-8
Corporate raiders, 1 81, 182 , 475 , 487-89, international patterns of, 489-91 Divisionalized firm. See Multidivisional
5 I 8-19. See also Takeovers rise of, 485-87 firms
Corporate social responsibility , 3 I 6 signaling from, 506-7 Division managers, 545
Corporation(s) types of, 457 Divisions, defining, 546
alternatives to publicly-held, 521-27 Debt overhang, 495-96 Dixit, Avinash, 138n, 283
not-for-profit, 4 I, 3 I 5 , 523-27 Debt workouts, informal, 503-4 Dobrzynski, Judith H. , 4Hn, 499
ownership of, 314-15 Decentralization, 26 Doeringer, Peter, 359 , 385
in Japan, 317-18 communication and , 93 Donaldson, Lufkin and Jenrette, 576
private, 490n coordination failure and, 111-12 Donaldson's, 484
privately held, 483 defined, 114 Donations , voluntary, 524-25
to limit influence , 275 Double marginalization, 5 50-5 I , 560
Cortes, I B, I 39 organizational routines and, 92 Double monopoly, 7
Cospecialized assets , 135-36, 139, 307, Decentralized pay determination , 370 Down-sizing, 427-28
310, 319 Decisions, decision making Downstream integration, 541
Cost(s) by consensus, 3 5 1, 43 1 , 443 Drexel Burnham Lambert , 399, 485, 488-
of agreement and enforcement, 301-2 efficient routines to optimize influence 89 , 520
bargaining, 147, 300-301, 304-5 activities, 273-77 du Pont , Pierre, 2
of capital, 452-54 product and pricing, 277 Du Pont. See E. I . Du Pont de Nemours
influence . See Influence costs under uncertainty, 209-11 and Company
information, 147-48 Deconglomeration, 484-8 5 Duration of transaction, 31
mca�urement, 147-49 Decreasing returns to scale , 99
mobility, H6-47 Deductibles , 207, 221 E. 1. Du Pont de Nemours and Company,
monitoring, 254, 2 56 Default, 502-5 79, 410, 417 , 540, 541, 542, 5H,
production, B-H Deferred compensation, H6, 347, 432-B 546
ri\k, 188 Delegation of authority, 17 , 544 East and West Germany, union of, 15
tramaction, 28-3 5, 38, 39, 301-3 Demand , �upply and, 327-28 Easterbrook, Frank, 5 31
efficient assignment of ownership Demming, W . Edwards , 442-43 Eastern Airlines, 5 I 0
claims and, 303- 5 Demographic changes, job design and, 408 Eastern Europe , economic restructuring
Cmt accounting, 540 Demsctz, Harold, 20, 5 I, 292-93 , 321, of, 12-16, 305-6 , 591-94
Cmt crntcrs, 225-30 3 54, 498, 531 East India Company, 539
Cm a nancc, 219, 463 Demski, Joel, 241 Eastman Kodak, 540

610 l 1 11 h
Economic organization, 19-5 3 Efficient separations, 347-48 Esca lator clau�c, 136
case study of, 43-48 Effort ESOPs, 280, 389, 413-14, 415, 485, 5 18
coordination and motivation, tasks of, incentives for, 218-19 Ethics of private property, 305
2 5-28 responsiveness to, 222 European Economic Community, 565
economy as highest-level, 19-20 profitability of incremental, 221-22 Evaluation, performance. See Performance
efficiency and, 22-2 5 E gg ed (Israeli bus company), 41 evaluation
evolution and persistence of, 48 Ehrenberg, Ronald C., 368 Event studies, 470
formal organizations, 20-21 Eihokai, 568 Excess returns, 462-63, 464
human motivation and behavior, model- Eisner, Michael 0. , 425, 439 Executive compensation, 389-90
ing, 42-43 Electrical power generation industry, 13 5- CEOs, 423, 424-27, 428
level of analysis of, 21-22 37 adequacy of, 440-41
market failures and nonmarket, 77 Electronic Data Services (EDS), 189, 483, intensity of performance incentives,
neoclassical model and theories of , 73- 498, 5 54, 5 5 5, 570, 574, 575, 5 76 43 8-40, 442
77 Elimination tournaments, 3 76 international differences in, 42 5-26
organizational objectives, 39-4 2 Ellickson, Robert, 265-66, 268n in large U . S. firms, 424-2 5 , 426
transaction costs analysis, 28-3 5 , 38, 39, Embargoes, 268-69 long-term incentive plans, role of,
301-5 Emmott, Bill, 489n 426-27
wealth effects and, 3 5-39 Employee cooperatives, 26-27 , 413, 561- performance pay, 43 3:-43
Economies of scale, 74-75, 106-7, 565 63 setting, 433-34
business alllances for, 579 Employee ownership, 315, 413, 415 in smaller firms, 426
cooperatives and, 562 Employee Retirement Income Security Act debate on, 434-3 5
in multidivisional firm, 545, 548 of 1 974 (ERISA), 1 77 deferred, 432-33
organizational and technological change Employee shirking, 1 79-81, 189, 292, evidence on performance and, 437-41
in manufacturing and, 586 371-73 forms of, 4 2 5
of simple market procurement, 5 5 3 Employee Stock Ownership Plans middle-level, 427-29
strategic decisions with, 106 (ESOPs) , 280, 389, 4 1 3-14 , 415, patterns and trends in, 424-29
Economies of scope, 1 07, 5 54 48 5, 5 1 8 tasks and temptations of senior execu­
Economy as highest-level organization , Employee turnover , 157 , 344-47, 3 50 tives, 43 5-36
19-20 Employment value maximization and incentives,
Education classical theory of levels of, 327-28 436-37
adequacy of, 591 efficiency wages for incentives in, 250- Executive recruitment market, 430-3 1
selection on basis of, 342-4 3 61 Expected rates of return, 466
as signal for productivity , 15 5-56 lifetime, 2 5 3, 279-80, 339-40, 3 50-52, Expected utility theory, 246-47
Edward I of England, King, 269 3 58, 393 , 5 9 1 Expected value (mean), 209, 246
Edwards, Richard, 360, 3 8 5 long-term, 3 58-59, 363 Expertise, academic tenure based on, 380-
Efficiency, 3 4 See also Human resource management 81
in agreements with large numbers of Employment and income security, 3 3 3-38 Explicit incentive pay, 187-88, 206-9,
participants, 1 45-46 Employment contract, 1 31-32, 1 57, 329- 392
concept of, 22 33. See also Compensation; Incen- Ex post opportunism, 136, 139
in coordination and motivation, 2 5-28 tive contracts Ex post renegotiation, 134
distribution and, 248-50 Employment discrimination, 343-44 Externalities, 75, 316
economic organization and, 22-25 Employment relations, 329 Exxon Corporation, 426, 565
incentive, 14 3-45 authority in, 3 30-31
informational, 72-73, 100-103 borrowing and lending in, 338
insider trading and market, 468 risk sharing in, 3 3 3-38 Fairness, labor market segmentation and,
labor market segmentation and, 360 End-game problem, 266 360
of long-term employment, 363 Endowment, resource , 63, 64, 65 "Fallen angels, " 485n, 488
in market for medical interns, 43-48 Enforcement , costs of, 301-2 Fama, Eugene, 470n, 5 3 1
menus of contracts and , 158-59 Engineers, compensation for, 399-400 Farrell, Joseph, 34n, 118, 321
moral hazard and, 166 English common law, 277 Fast-food business, 260-61
need for information for, 26 Entry deterrence, vertical integration for, Feasibility, contract, 2 1 8- 1 9
of organizations, 23-24 561 Federal Aviation Authority, 1 80
of ownership patterns, 498 Equal compensation principle, 228-32, Federal Crop Insurance Corporation
of resource allocations, 23 311, 363, 394, 397, 403, 4 1 2, 441, (FCIC), 177
separating distributional aspects of deci­ 471 , 473, 5 58 Federal Deposit Insurance Corporation,
sions from, 275-76 Equilibrium, competitive, 66-67, 69-71, 520
signaling and, 156 74, 102 Federal Employees' Compensation Act,
Efficiency principle , 24-2 5, 28-36 Equilibrium behavior, 3 73-74 179-80
Efficiency wage contracts , 363 Equity, financing with Federalist, The, (Madison) , 271
Efficiency wages, 250-61 classical analyses of financial structure Federal Reserve, 1 7 5
comparative statics for, 2 54-56 decisions and, 456-60 Federal Savings and Loan Insurance Cor­
in fast-food business, 260-61 conflict of interest between debt and, poration (FSLIC), 170-76
incentive payments vs. , 2 52-5 3 183-84 Filene's, 484
Marxian view of, 2 56-57 international patterns of, 489-91 Finance, classical theory of. See Classical
unemployment and, 253-54 signaling from, 506-7 financial economics
Efficient markets hypothesis , 467-71, 473- types of, 457 Financial accounting, 540
74 Equity, pay, 274-75, 4 1 8-19 Financial contracts, moral hazard in, I 8 3-
evidence on, 469-70 Equivalent value index, 36 85
forms of, 467-69 Errors of fit, 123-24 Financial leverage, 485

lmb 61 1
Financial risks, e\ aluation of, 209-1 1 , Ford Motor Company, 2, 3, 1 6, 78, ! 09, Coldsmith, James, 484
246-47 3 37, 542, 545, 5 5 3 Goodyear Tire and Rubber, 484
Financial sef fices, 588-89 Formal organizations, 20-2 1 Gorbachev, Mikhail, 1 5 , 297, 593
Financial stock exchanges, 29 Formula-based incentive pay, 399-400 Cordon , Dona ld, 3 54
Financial structure Forstma nn, Little & Company, 500 Could, J. R. , 8 1 11
classical economics of decisions about, Forward integration , 541 Governance reforms, corporate, 520-2 1
4 5 6-60 Foster, George, 85, 47011 Covemment(s)
allocation of investment capital by Foulkes, Fred, 44 5 creation, alteration, and assignment of
markets, 459-60 Foust, Dean, 499 property rights, 30 3-5
Modigliani-Miller analyses, 448, 4 56- France, 340, 490 funding of not-for-profit organizations,
59, 49 1 , 5 0 5 Franchise retailing, 5 54, 56 1 , 563-6 5 , 566 526-27
incenti\·e and rights aspects of, 49 1 , 505 Frank, Robert H . , 4 1 811 influence costs of, 270-7 1 , 274
international patterns in, 489-9 1 Franks, Julian, 50411 means of coordi nation used by, 89
objectives in selecting, 508 Fraud, 1 7 1 , 1 76, 1 77 production planning by, 89
See also Corporate control Free cash flow, 492-94, 496, 507 public insurance programs, 1 70-78
Financing, i nnm·ation i n , 590-9 1 Free-rider problem , 278, 294-98, 3 9 5 , rents created by, 270-7 1
Fireman, Pau l , 424 500, 5 1 0, 564, 566 See also Public sector
Firestone Tire & Rubber, 4 57, 48 3 , 5 1 I Freixas, Xavier, 2 4 1 Government National Mortgage Associa-
Firing, tenure and, 380-8 1 French, Kenneth, 47011 tion (CNMA), 1 77
Firm(s), 5 38-84 Frequency of transaction, 3 1 Greenmail , 439, 484, 5 1 8- 1 9
as bearer of reputation, 3 3 1 -32 Fringe benefits, 389, 390 Greif, Avner, 2 5211 , 269n
busi ness all iances, 575-80, 586 Frost, Peter J . , 282 Crossman, Sanford, I 62, 24 1 , 3 2 1 , 322,
career paths in, 2 58, 58 1 Fudenberg, Drew, 1 1 9, 24 1 53 1
changing nature of, 5 39-4 3 Functional organization, 78 Cross return, 1 72
classical, 293 Fundamental theorem of welfare econom­ Croup incentive pay, 4 I 3- 1 8
core competencies of, 1 07-8, 5 54, 570- ics, 62-7 1 , 8 3 , 89, J OO, 1 04 Croup ownersh ip approach, 295
72 Funding of not-for-profits, government, GTE, 5 76
horizontal scope and structure, 54 1 -42, 526-27 Cuesnerie, Roger, 2 4 1
569-7 5 Fur trading businesses in Canada, 6-9 Guilds, 268
competitive (corporate) strategy and Gulf Oil, 483
organizational inno\'ation, 569 Cain-sharing plans, 4 1 4- 1 6 Cutfreund, John, 9, I 0
directions of divisional expansion, Galen , M ichele, 489
5 70-7 1 Callini, Nancy, Hn Hall, Bronwyn H . , 5 1 2 n
disadvantages of expansion in, 5 7 1 -7 5 Came theory, 262, 283 Haloid Corporation, 429
internal structure of, 78-79, 544-5 2 Gas tax-shelter programs, 1 84-8 5 , 224-26 Hammer, Armand, 493
multidivisional form, 78-8 3, 544-52 Ceanakoplos, John, 1 1 9 Hannaway, Jane, 23 1 11 , 445
development of, 5 40-42 GenCorp, 485 Hanseatic League (Hansa), 268-69
problems of managing, 546-5 2 Cenentech, 4 29 Hansma nn, Henry, 53 1 , 56 1 n
multi product, 542-4 3 General Agreement on Tariffs and Trade H anson Croup, 49011, 5 00
in neoclassical model of private owner- (CATT), 589 Hardware, computer, 309
ship economy, 6 5 General Dynamics Corporation, 429 Harris, Jeffrey, 3 54
a s nexus o f contracts, 20, 3 3 1 General Electric Company, 1 07, 1 08, 394, Harris, M ilton, 1 62 , 24 1 , 3 54, 3 8 5 , 50611,
present and future of, 586-9 1 48 3 , 540 53 1
scale and structure of, I 06-7 General Motors (CM), 2-4, 5 , 1 6, 1 7, 26, Hart, Oliver, 1 6 1 , 24 1 , 3 2 1 , 322, 3 54,
transaction with other firms, 89-90 27, 79, 1 06, 1 1 1 , 1 37, 1 89, 308, 531
vertical boundaries and relations in, 3 1 0, 337, 483, 498-99, 540-4 1 , Harvard School , 39
5 5 2-68 542, 544, 545, 566, 567, 5 70, 573 , Hayek, Friedrich, 5 1 , 56, 72, 85, 1 00,
cooperati\'es, 26-27, 4 1 3 , 562-63 574, 5 7 5 , 576, 577, 578, 589 1 18
franchise retail ing, 5 54, 56 1 , 5 6 3-6 5, General partners, I 84 Hazard, moral . See Moral hazard
5 66 General-purpose human capital , 328, 344- Health insurance, 1 49, I 50, 1 67-68, 22 1
simple market procurement, advan­ 45 Health Maintenance Organizations
tages of, 5 5 3- 56 General Telephone and Electronics (HMOs), 4 1 8
supplier relations i n Japanese auto in­ (GTE), 5 57 Helyar, John, 494
dustry, 56 5-68 Geographically defined d ivisions, 546 Hewlett, William, 547
\'ertical integration, 1 59, 54 1 , 542, Germany, financing and ownership pat- Hewlett Packard Company, 1 1 6, 276-77,
549, 5 52-5 3 , 5 56-6 1 terns in, 490-9 1 HO, 546, 547, 548, 579, 587
Fmn-specific human capita l , 328, 3 3 3 , Cetschow, George, 1 9411 H ierarchy, 29, 2 59
34 5 , 3 50-5 1 , 363-64 Gibbard, Allan, 1 62 H igh-defin ition television (HDTV), 1 08
Fisher, Irving, 450, 477 Gibbons, Robert, 266, 2 8 3 , 4 3811 H ildenbrand, Werner, 8 5
Fisher Body, I 37, 308 Gilson, Ronald, 27411, 3 8 5 , 4 1 811, 5 2 1 11 H iram Walker, 483
Fisher separation theorem, 448, 449-5 1 Glasnost, 1 5 H iring
Fishing rights, 294-96 Class-Steagall Act ( I 9H ), 9 in Japan, 349-50
Fit, error\ of, I 2 3-24 Globalization of economic activity, 589-90 selection criteria in, 342-43
Fk:xiblc manufacturing tcclmologics, 1 1 0, Coal congruence, 1 88 See also Recru iting
587-88 Goldberg, \'ictor, 1 6 1 , 3 54 Hirschma nn, Albert, 509n
Fou1s, prod uctivity and, 486-87 Colden handcuffs , 346 Hoag, Susan E. , 1 8211 , 48 5 11, 50611, 5 1 711 ,
1-'<,rhes magazine, 424. 4 3 5 11, 5 1 9 Golden pa rachutes, 390, 437, 5 1 8 531
FcJTcl, l kmy, 4, 1 6, 1 1 1 , 1 1 2, 587 Goldman Sachs, 380 Hoffmann-La Roche, 429

611 l 1 1d1·x
Holding companies , 78 366, 401, 497, 546 , 571-72, 574, Incomplete contracts. See Contractual in­
Hold-up problem , 136-39, I 59 , 307-8, 575-76, 576, 577, 586 completeness, bounded rationality
312, 568 Icahn, Carl , 484, 488 , 499 , 513 and
Holm , Kaare, 410 Idiosyncratic risk , 465 lncorpuration , state of, 5 1 8
Holmstrom , Bengt , 5 1, 1 6 1 , 24 1 , 3 54, Imperfect commitment, 30, 1 29 , 234 , 236 Increasing returns to scale , 74-7 5 , 99, 100
373, 42 1 , 441n , 445, 477 Imperfect information, 9 5-96 Independent contractors, incentives for,
Honda Motor Company, 578 Implementation problem, 2 1 8 232
Honesty, reputation for , 263 Implicit contracts, 1 32 , 139-40 , 332-33 Individual incentive compensation. See
Honor, merchant, 2 57 Implicit incentive pay , 402-3 under Compensation
Horizontal merger , 569n Incentive compatibility constraint, 200 Individuals , analysis on level of, 2 1 -22
Horizontal scope and structure , 541-42 , Incentive constraints , 1 29, 1 41 , 144, 145 , Industrial enterprise , emergence of, 539-
569-75 146, 2 1 8, 2 54 40
competitive (corporate) strategy and orga­ Incentive contracts, 1 79, 206-36, 249 Industrial planning, national, 1 08
nizational innovation , 569 to control moral hazard , 1 87-88 , 206-9 Influence activities, 192-94, 249
directions of divisional expansion, 570- explicit , 1 87-88, 206-9 optimizing , 273-77
72 incentive pay, 2 1 4-36 structuring decision processes to limit,
disadvantages of expansion in, 571-7 5 efficiency wages vs. , 2 5 2-53 275-77
Horngren, Charles, 8 5 equal compensation principle, 228- subjective performance evaluation and ,
Hostile takeov�rs, 181-83, 484 , 490 , 5 1 I , 32, 311 , 363 , 394 , 397, 403 , 4 1 2 , 406
515 44 1 , 471, 473, 5 58 varieties of, 272-73
Hostile tender-offer takeover, 509 incentive-intensity principle , 221-26, Influence costs, 1 92, 193-94, 375
Houston Oil and Minerals Corporation , 2 27, 2 28, 234, 395 , 398, 549 conflicting managerial interests and, 4 5 5
194, 5 74 informativeness principle of, 2 1 9-2 1 , decentralized pay determination and ,
Houston-Tenneco, 274, 277 233-34 , 235 , 368 , 432 370
Hudson's Bay Company , 6-9, 16, 26, 78 , intertemporal incentives and, 232-36 of government , 2 70-71, 274
292, 331, 483, 539, 544 job design and, 410- 1 I horizontal expansion and problem of ,
Hughes Aircraft , 48 3 measured performance, basing pay 573-74
Human capital, 135, 328-29 on , 2 1 4-15 Japanese methods of limiting, 279-80
classical theory of , 327-29 for middle-level managers , 428-29 legal system and , 277-80
firm-specific , 328, 333, 345 , 3 50-5 1, model of , 215-19 in private sector, 271-73
363-64 monitoring intensity principle , 226- of promotions in subjective evaluation
general-purpose, 328 , 344-45 28 , 403 system, 407
nonspecific, 328 See also Compensation restructuring and, 520
nontransferability of, 3 1 3 mathematical example of , 198-201 Informal debt workouts, 5 03-4
partnership and, 5 23 reasons for failure of , 2 53 Information
protection of , in Japan , 350-51 Incentive efficiency , 143-45 cardinal , 367
turnover and , 344-45 Incentive-intensity principle , 221-26, 227 , comparing coordination systems with
See also Human resource management 228 , 234 , 395 , 398 , 549 perfect and imperfect , 94-96
Human capital investments, ambiguous Incentive problems competing sources of, 186
separations and, 348-49 adverse selection, 129, 147-54, 237-39 coordination and economizing on , 1OO­
Human capital risk, managerial investment private information, 140-47, 150 I 06
decisions and, 430-31 lncen,tives, 8, 9-12, 16 horizontal expansion and problems of,
Huma1 1 resource management , 326-57 , balancing risks and, 207, 208-9 571-73
548-49 bonding as , 189-90 in multidivisional firm, 544-4 5
changes occurring in , 591 delegated authority and, 17 need for , 26
classical theory of wages, employment discretion and, 412-13 ordinal, 367
and human capital, 327-29 for effort , 218-19 prices of financial assets and, 467-75
contracts, 329-33 employment, efficiency wages for , 250- efficient markets hypothesis , 467-7 I ,
employment, 1 3 1 -32 , 1 5 7 , 329-32 61 473-74
implicit , 132, 1 39-40 , 332-33 financial structure and , 49 1 , 505 shortsighted markets and manage­
employment relations , 329 in franchise retailing , 564-65 ment , 471-73, 474
authority in, 330-3 1 intensity of , 2 1 6 , 221-26 private. See Private information
borrowing and lending in , 338 in markets, 27-28 , 71-72 required in production planning, 100-
risk sharing in, 333-38 in multidivisional firm, 544-4 5, 549 IOI
in Japan, case study of , 349-52 optimal risk sharing ignoring, 213-14 Informational asymmetries, 30, 140, 141,
recruiting , 339-44 , 430-31 ownership and , 8, 9 , 11-12 , 190-91, 142-43, 159
retention, 344-47 , 349-50 291-93 Informational efficiency , 72-73, 100-103
separations , 347-49 partial insurance of income security and , Informational rent , 141n
See also Compensation; Job assignments; 335 Information costs, 147-48
Motivation; Promotion policies promotions as, 366-67 Information systems, 540
Hungary, 306 ratchet effect and , I 4-15 Informativeness principle , 219-21 , 233-
Hurwicz, Leonid, 85 , 10 1 , I 18, 1 22 for risk taking with borrowed funds , 34 , 23 5 , 368 , 432
Hurwicz criterion, 1 0 1-2 1 73-75 Ingersoll , Bruce , 177n
Hutterites , 305 socialist , 14- 1 5 Ingres, 576
vertical integration and need for perfor­ Innovation, 569, 590-9 1
mance, 5 58-59 Innovation attributes , 92-93 , 106 , 1 1 2-13 ,
I. Magnin, 484 See also Compensation 592
IBM Corporation, 159 , 309 , 340, 341, Income security, employment and , 333-38 Insider trading, 468, 489

lndC'x 613
Institutional investors, 474, 497-98, 500, classical economics of risk and return Junk bonds, 1 7 1 , 485, 488-89, 5 1 1, 520
5 20, 5 2 1 on, 460-66 Just-in-time (JIT) system, 4-5, 6, 92-93,
Institutions, reputations aided by, 266-69 financed by borrowing, 450 110, 312- 1 3, 394
Insurance human capital, ambiguous separations
ad\'erse selection and, 1-+9, 1 50, 1 53 and, 348-49 Kahn, Lawrence, 429n
automobile, 22 l in specific assets, 1 35-39 Kaiser-Permanente, 4 1 8
copayments, 207, 22 1 vertical integration and protection of, Kakalik, James S. , 503
deductibles, 207, 22 l 5 56- 58 "Kanban" (JlT) system, 4- 5, 6, 92-93,
disability, 1 79-8 l Investors 1 1 0, 3 1 2-13, 394
of employment and income security, active, 500, 520-2 l Kanter, Rosabeth Moss, 585
333-38 institutional, 474, 497-98, 500, 5 20, Kaplan, Stephen N. , 5 1 3n
government (public), 170-78 521 Katz, Michael, 5 53n
health, 1-+9, 150, 167-68, 221 perspective o n financial structure deci­ Kaufman, Henry, 485n
life, 1 78-79 sion, 458 Kawasaki, S ., 223n
moral hazard and, 1 66, 1 76-79 venture capitalists, 500- 50 1 Kazamaki, Eugenia, 340
partial, 334-37 Invisible hand, 28, 72, 77 Keiretsu, 577, 579-80, 590
risk sharing and, 211 -14 Iraq, 296 Kenney, Roy, 1 48
Insurance contract, 206-7 Israel Khrushchev, Nikita, 12
Insurance industry, 5 90 distorted prices in, 73 Kim, E . Han, 5 1 1 n
Integration public policy toward layoffs in, 340 Kimberly-Clark, 5 53
upstream (backward) or downstream (for­ ITT, 486 Kirman, Alan, 85
ward), 541 Klein, Benjamin, 1 37n, 1 48, 161 , 2 5 9,
vertical, 1 5 9, 54 1 , 542, 549, 5 52- 53, James, Estelle, 53 l 282, 322
5 56-61 Japan Klemperer, Paul, 1 1 9
Intel, 34n cost of capital in, 453- 54 Knight, Frank H . , 1 9, 5 1 , 333, 358
Intensity of incentives, 2 1 6, 22 1-26 financing in, 490, 491, 5 90-91 Koenig, Richard, 417
performance incentives for CEOs, 4 38- labor markets in, changes in, 591 Kohlberg, Kravis, Roberts & Company
40, 442 national industrial planning in, 1 08 (KKR), 437, 484, 48 5, 489, 493,
Interdependence of transactions, 32-33 ownership patterns in, 490, 491 500, 5 1 1
Interest rates, net present value and, 4 5 1- privatization of industry in, 306 Koito Manufacturing, 499
52 production planning in, 75 Korea, production planning in, 7 5
Internal labor markets, 3 5 9-7 5 Japanese firms Korn/Ferry International, 434
rationale for, 362-7 5 CEO compensation in, 425-26, 442-43 Kovac, Edward, 8 1
efficiency o f long-term employment, consensus decision making in, 431 , 443 Kraakman, Reinier, 5 2 1 n
363 goals of, stakeholders' interests and, 41 Kreps, David, 24 l, 282, 3 54
firm-specific human capital, 328, human resource management in, case Krueger, Alan B., 26 1
333, 345 , 3 50- 51, 363-64 study of, 349-52 Kuwait, 296
pay attached to jobs, 360-62, 369-7 l incentive contracts for suppliers of, 223- Kyohokai, 568
promotion policies, 364-69, 373, 24
374, 37 5, 376 keiretsu, 577, 579-80, 590 Labor contracts, 1 3 1 -32, 1 57, 329-33
as systems, 371-7 5 managerial risk taking in, 43 1 Labor market
Internal structure of firm, 78-79, 544- 5 2 monitoring by concentrated sharehold- internal . See Internal labor markets
Internationalization o f business, 589-90 ers, 497 Japanese, changes in, 591
International Telecommunications Satellite no-layoff policy in, 338 promotions and competition in outside,
Project (INTELSAT), 577 ownership and responsibility of, 317- 1 8 364-66
Intertemporal incentives, 232-36 participatory management in, 279-80 Labor market segmentation patterns, 359-60
Intervention(s) pay and dual hierarchy in, 362 Laffont, Jean-Jacques, 241
autonomy from, 2 1 Jarrell, Gregg A . , 487, 5 1 7, 53 l Lal, Rajiv, 397, 398
influencing, 1 93 Jensen, Michael C ., 162, 182n, 1 83n, Lamb, Charles, 248
selective, 1 92-94 388, 406n, 42 1, 435n, 439-42, Lambert, Richard A. , 376, 437
Inventory, JIT system of, 4- 5, 6, 92-93, 445, 495, 500, 5 1 I n, 531 Land O'Lakes, 562
110, 3 1 2-13, 394 Job, pay attached to, 360-62, 369-7 l Langton, Nancy, 274n
Investment Job assignments, 3 5 1 , 364, 376-77 Lapin, Lisa, 297n
defined, I 34 Job design, 408- 1 3 Larcker, David F. , 376, 437
motivating risk taking in, 429-32 Job enrichment programs, 4 1 1 Law, the, and legal system
savings and loan crises and risky, 171-7 5 Job ladders, 374-7 5, 376 as impediment to trade, 301-3
specific, 31 parallel, 366 influence costs and the, 277-80
Investment banking industry, 9- 1 2 Job rotation, 236, 5 8 1 merchant, 267, 277
Investment proposals, paying for, 43 1 -32 John, Andrew, 8 5 , I 1 9 reputations and, 265-66, 267
Investments John, Kose, 531 Law and economics movement, 277-78
in bargaining position, 1 49 Johnson, F. Ross, 437, 493-94 Law firms, dividing profits among partners
classical economics of investment deci­ Johnson, "Magic, " 4 1 9n of, 4 1 8
sions, 448, 449- 5 5 Johnson, Robert W . , 423 Lawler, Edward E ., Ill, 392n, 395-96,
Fisher separation theorem, 448, 449- Joint ventures, 577, 586 42 1, 423, 445
51 Jones, Daniel, 538 Layoffs, 190, 334-35, 338, 340, 348, 5 1 4-
net present values in, 451-54 Jones, Larry, 162 15
strategic investments as design deci­ Joskow, Paul, 137 Lazear, Edward, 166, 1 89, 1 96, 340, 385,
sions, 454- 5 5 Junior debentures, 457 42 1

614 l wl1·x
LBO association, 500, 522 McConnell, John J . , 474n Ma rket efficiency, insider trading and, 468
Learning, organizational, 43 McDonald, Jack, 49 1 n Market failures, 28, 73-77
Learning explanation for tenure, 38 1-82 McDonald's Corporation, 400, 564, 565 Market-oriented strategy, 8
Lease, Ronald C. , 435n McDonnell-Douglas, 429 Market portfolio, 464
Leases, 308, 502 MCI, 570 Market procurement, advantages of, 5 5 3-
Leffler, Keith, 2 59, 282 McKinsey & Company, Inc. , 408, 409- 1 0 56
Legal-entity approach, 20-2 1 MacLeod, Bentley, 385 Ma rket-segmentation strategy, 17
Lehman Brothers Kuhn Loeb, 574 McMillan , John, 223n, 24 l Market socialism, 72-73, 94
Lehn, Kenneth, 498, 5 1 5n, 5 3 1 Macmillan Press, Ltd. , 490n Market-value maximization , 436--37
Leland, Hayne, 5 3 1 Madison, James, 27 1 Markowitz, Harry, 477
Lenders, monitoring incentives for, 50 1-2 Maghribi traders, 252, 258 Marriott, 5 1 2
Lenin, Nikolai, 1 3 Maidique, Modesto, 4 1 4n Marshall, Alfred, 248, 252, 282
Lenscrafters chain, 579 Mailath, George, 1 62 Martin, Joanne, 282
Lerner, Abba, 72, 85 Majluf, Nicholas, 5 3 1 Marx, Karl, 1 3, 50
Leroy, Stephen F. , 470n, 477, 478 Malcomson, James, 385 Marxian approach, 38, 39
Leveraged buyouts (LBOs), 272, 48 5-86, Malkiel, Burton , 470n , 477 Matching problems in market systems, 76--
. 486, 488, 489, 5 1 1 , 5 1 5 , 590 Management 77
Levering, Robert, 5 5 3n large shareholders and discipline of, 497 Matsusaka, John G. , 487
Levin , Daron , · 576 participatory, 27 5 , 279-80 Matsushita, 483, 570
Lewis, Meryl, 282 role in coordination , l l 4- l 6 Maxfield, Robert, 5 5 8
Lewis, Michael, IOn shortsighted, 47 1-73, 474 MBA student admissions, 342
Lex mercatoria, 267 Management buyouts (MBOs), 483, 484, Mean, 2 1 0, 246
Liability(ies) 488, 509 Means, Gardner, 1 8 1 , 1 96
corporate, 456, 457 Management by objectives, 40 I , 402 Mean-variance analysis, 46 1
of corporate officers and directors, 279 Managerial compensation Measurement costs, 147-49
limited, 1 84, 1 85 deferred compensation, 432-33 Meckling, William, 5 3 1
in partnership, 522-23 middle-level , 427-29 Medical interns, market for, 43-48, 77
unlimited, 1 84 motivating risk taking and, 429-32 Medoff, James L. , 337n, 40 5
Lichtenberg, Frank, 486, 487 performance pay, 400 Menus of contracts, 40 1-2
Life insurance, moral hazard in private, See also Executive compensation efficiency and, 1 5 8-59
1 78-79 Managerial hierarchies, 364 Mercer, Robert, 482
Lifetime employment, 2 5 3 , 279-80, 339- Managerial interests, conflicting, 45 5 Merchant gu ilds, medieval, 268
40, 3 50-52, 3 58, 393, 59 1 Managerial misbehavior, 1 8 1-83 Merchant honor, 257
Limited liability, 1 84, 1 85 Managers Merchant law, 267, 277
Limited partners, 1 84 conflicting interests of owners and, 49 1 - Mergers and acquisitions, 1 92, 483-84,
Limited partnerships, 1 84-8 5 94 486--87, 540
LIN Broadcasting, 424 division, 545 conglomerate, 486
Lincoln Electric Company, 236, 338, 340, multifaceted jobs of, 39 1 -92 failed, influence costs and, 1 93-94
393, 403 ownership of corporation by, 3 1 5 golden parachutes following, 390, 437,
Linden , Dana Weschler, 443n Mandatory retirement, 1 89-90 518
Linear compensation formulas, logic of, Mankiw, N. Gregory, 470n horizontal, 569n
2 1 6-- 1 7 Manne, Henry, 468, 478, 53 l special problems of, 574-76
Lintner, John, 477 Manufacturing Mergerstat Review, 5 1 I
Liqu idation, 278, 502, 503, 504, 5 1 0 flexible, I I O , 587-88 Merit raises, 389
List (or vector), 63 modern manufacturing strategy, 1 09- 1 0 , Merrill Lynch, 499
Local nonsatiation, 64-65 1 1 2, 1 1 5- 1 6, 408 Mesa Petroleum, 484
Lockheed Corporation, 429, 5 2 1 technological and organizational change Mesters, Mace, 385
Long, William F. , 5 I I n in, 579-80, 586--88 Metropolitan Life Insurance Company,
Long-term employment, 3 5 8-59, 363. See Marathon Oil, 1 82, 483, 5 1 7 502
also Internal labor markets Marginalization, double, 5 50-5 l , 560 Meyer, Margaret, 365
Long-term goals, assessing contributions Market(s) Michelin, 483
to, 363-64 adverse selection and closing of, 1 50- 5 3 Microsoft Corporation , 309
Long-term incentive compensation for allocation o f investment capital by, 459- Midcon, 5 1 2
CEOs, 426--27 60 Middle-level managers, compensation of,
Long-term lenders, monitoring incentives complete and competitive, 40 427-29
of, 502 coordination through system of, 27 Midwest Airlines, 5 5 3
Lublin, Joann S. , 4 34n for corporate control, 508-9, 5 20 Milbank, Dana, 568n
Lucas, Charlotte-Ann, 1 74 for executive talent, 4 30-3 l Milgrom, Paul, 1 62, 24 1 , 267n, 269n,
Lucas, Robert, 477 incentives in, 27-28, 7 1-72 3 54, 42 1 , 478
Lucky-Goldstar chaebol (industrial group), informational efficiency of, 72-73 Milken , Michael, I O , 399, 485, 488-89,
542-43, 569 for medical interns, 4 3-48, 77 5 1 I , 520
Lundberg, Craig, 282 missing, 75-76 Miller, George, 300n
Lundberg, Shelly, 3 54 mon itoring by, 1 86-87 Miller, Merton, 477, 478
perfect capital, 450-5 1 , 46 1 Mineral rights, 296
MCA, 483 shortsighted, 453-54, 47 1 -73, 474 Minimum cost implementation problem,
McAfee, Preston, 24 l transfer pricing with competitive outside, 2 54, 256
Macaulay, Thomas, 252 82-83 Minow, Nell, 494
McCaw Cellular Communications, 424 Market-clearing price, 60-6 1 Mirrlees, James, I 96, 24 l

l ud,·x 6 15
t-.tisrepresentation, strategic, 46-48, 141 Motivation, 126-27 North West Company, 6, 7-9, 16, 17,
Missing markets, 7 5-76 deciding what to motivate, 391-92 292
t-.1itel Corporation, 5 71 in market for medical interns , 4 3-48 Norway , 340
Mitsubishi Chemical, 275, 519 modeling human , 42-43 Not-for-profit organizations, 41, 315, 523-
Mitsubishi Corporation, 576, 579-80 prices in neoclassical model as, 67 27
Mitsubishi Electric Corporation , 577 , 579 task of, 2 5-28 No wealth effects, 217, 248-49, 513
Mitsubishi Heavy Industries, 580 See also Bounded rationality; Incentives Nucor Corporation, 4 1 4
Mitsubishi keiretsu , 577 , 579-80 Motivation costs, 29-30
Mnookin, Robert, 274n, 385, 418n Multidivisional firms , 78-83, 544-52 Objectives, organizational, 39-4 2
Mobility costs, 346-47 deciding on scope of, 5 5 1-52 Occidental Petroleum, 493, 511-12
Mobil Oil, 182, 483, 517 development of, 540-42 Ocean fisheries, economics of, 294-96
Modem manufacturing strategy, 109-10, at General Motors, 3-4 Offers
112, 115-16, 408 problems of managing, 546-52 outside, 345-46
Modigliani, Franco, 477, 478 transfer pricing in, 79-83 inadequate and coercive , 515
Modigliani-Miller theorems, 448, 456-59, Multiproduct firm , 542-43 tender, 509, 511
491 , 505 Murphy, Kevin J. , 162, 266 , 388, 406n, Officers , liability of corporate, 279
Momentum, complementarities and, 543 407n, 421 , 43 5n, 438n, 439-42, O'Flaherty , Brendan, 379n
Monetary policy, 17 5 44 5, 495 Ohno, Taiichi, 4 , 111
Monitoring Muscarella, Chris J. , 473n Oil and gas tax-shelter programs , 184-8 5,
of borrowers, 174-7 5 Myers, Stewart , 478 , 531 224-26
to control moral hazard, 186-87 Myerson, Roger, 162, 241 Oil industry , motivating risk taking in, 43 1
dividends and, 507-8 Okuno-Fujiwara, Masahiro, 253, 254, 282
lenders' incentives for , 501-2 Nabisco Brands, 483, 493 Olympia & York Development, 483
owners' incentives for, 496-501 Naked call options , 174 One-fund portfolio theorem, 462
Monitoring costs, 254, 2 56 Nalebuff, Barry , 138n, 283 Operational scale, 106, 107, 408
Monitoring intensity principle , 226-28, Nash equilibrium, 263 Opportunism
403 National Broadcasting Company (NBC), ex post, 1 36, 139
Monopoly, 6-7, 71, 148, 5 59, 562 483 postcontractual , 137-38, 167
Monopsony, 71 National industrial planning, 108 precontractual, 140-4 7, 148
Monteverde, Kirk, 31On National Intern Matching Program Opportunistic behavior , 128-29, 136 , 137
Montgomery, Cynthia, 486, 487 (NIMP), 44-48, 77 Optimal batch size, 111, 112
Montgomery Ward, 483 NCR Corporation, 5 7 1 Optimal incentive intensity, mathematical
Moore, John, 322 NEC, 571 derivation of, 222-23
Moore, Larry F. , 282 Nelson, Philip, 162 Optimal price, determining, 5 9-60
Moral hazard, 129, 150, 154n. 158, 159, Nelson, Richard, 43 , 51 Optimal variety choice curve, 111-12
166-203, 549 Nenko (lifetime employment) , 2 53, 279- Options
adverse selection vs . , 169 80, 339-40 , 350-52, 358 , 392, 591 call, 174 , 457
concept of, 166 Neoclassical market model, 57-69 options
controlling, 18 5-92 economic organization problem in, 58 nonmatching, 34 5-46
bonding, 1 89-90 extensions and difficulties with, 61-62 put, 457
explicit incentive contracts, 187-88, market-clearing price in, 60-61 stock, 390, 42 5, 431, 441
206-9 , 392 with one objective and single scarce re- Ordinal information , 367
monitoring, 186-87 source, 58-60 Organizational architecture, 20-21
ownership changes and organizational prices in, assumptions about, 71 Organizational design, 1 15, 273-77
redesign, 190-92 of private ownership economy, 62-69 redesign, 190-92
efficiency effects of, 166 scope of, 68-69 Organizational politics, 271
efficiency wage and, 253 theories of organization and, 73-77 Organizational size, executive pay and,
group ownership approach and, 295 Nestle , 5 I 2n 438
incidence of, 167-70 Net present values, 451-54 Organization (organizational structure), 2-
influence activities and unified owner­ Net return, 172 17
ship, 192-94 Netter, Jeffry , 531 business
m msurance Newman, Peter C . , 6n historic examples of, 2-9
misbehavior and, 1 66 New products, 110, 121-24 patterns of internal organization, 78-
public vs. private, 176-79 New United Motors Manufacturing, Incor­ 79, 544-52
in not-for-profits, 5 2 5-26 porated (NUMMI), 577, 578 strategies of modem firms, 9-12
in organizations, 169-70 , 179-85 Nexus of contracts, organization as, 20, changing economies of Eastern Europe
employee shirking, 179-81, 189, 292 , Bl and, 12-16, 305-6, 591-94
371-73 Nissan, 578 formal, 20-21
financial contracts, 18 3-8 5 Nominal (money) value, 466 manufacturing and changes in, 579-80,
managerial misbeha\·ior, 181-83 Nonmatching option, 34 5-46 586-88
with risk-neutral agents , 236-39 Nonprofit organization, 41, 315, 523-27 patterns of success and failure in, 16-17
savings and loan crisis, case study of, Nonsatiation, local, 64-65 See also Economic organization; Firm(s)
170-76 Nonspecific human capital, 328 Organizations
7\1orck, Randall, 512n Nordstrom, 265 efficiency of, 23-24
t-.1orgcnstern, o�kar, 241 Normal form game, 261-62 internal use of price systems, 78-83, 89
Mortgages, government insurance North, Douglass C . , 267n, 288 moral hazard in, 169-70, 179-85
(GNMA) on, 177-78 North America, production planning in, employee shirking, 179-81, 189, 292,
Mmkowitz, Milton, 5 53n 75 371-73
Mossin, Jan, 477 Northern Telecom, 572 financial contracts, 183-8 5

616 l 1 1d1·.x
Organizations (continued) limited, 1 84-8 5 Plant closing decisions, �takeholders' right�
managerial misbehavior, 1 8 1-83 tenure in form of, 379-80 i n, 3 1 8- 1 9
See also Firm(s) Pascale, Richard T. , 547 Poison pills, 1 82-83, 499, 5 1 7
Oshman, Kenneth, 572 Past performance, standards based on, 2 3 3 , Pola nd, 14, 1 5 , 1 6, 306, 593
Oster, Sharon, 3 54 234 Polaroid Corp. , 4 1 5 , 485
Outland, Donald, 398n Patterson, Gregory A. , 1 30n, 337n Politics, organizational , 27 1
Outside labor market competition, promo­ Paul, Jonathan, 478 Pollack, Andrew, 577
tions and, 364-66 Pauly, Mark, 1 96 Pollution, 304-5, 3 1 6
Outside offers, 345-46 Pay. See Compensation; I ncentive con- Porter, Michael, 1 93-94
Outside opportun ities, wages relative to, tracts Porter, Richard D. , 470n, 477
253 Pay differentials between ranks, 376 Portfolio
Overbidding, winner's curse of, 5 1 4 Pay-for-skills programs, 399 market, 464
Overpaying, takeover premia due to, 5 I 2 Pay ranges, 370-7 1 one-fund portfolio theorem, 462
Owner-operated suppliers, 5 56 Pels, Donald A . , 424 Portfolio choice, 46 1
Ownership, 288-324 Pension Benefit Guaranty Corporation Positive external effects, 454n
alternatives to publicly held corporation , (PBGC), 1 77 Postcontractual opportunism, 1 37-38, 1 67.
, 5 2 1-27 Pensions, 3 38, 389 See also Moral hazard
asset, 249, 307- 1 9 Employee Stock Ownership Plans, 280, Postlewaite, Andrew, 162
complex assets, 3 1 3- 1 9 389, 4 1 3- 1 4, 4 1 5 , 485 , 5 1 8 Poulsen, Annette B . , 5 1 5n
incentive pay and, 23 1-32 Pepsico, 434 Pound, John, 478, 500n, 5 3 1
predicting, 307- 1 3 Perelli, 49 1 Poverty, cycle of, 2 50
Coase theorem and, 38, 39, 293-306, Perelman , Ronald, 484, 488, 489 Pratt, John , 241
346 Perestroika, 1 5, 593 Precontractual opportunism, 1 40-4 7, 1 48
bargaining costs and limits of, 300- Perfect capital market, 450- 5 1 , 46 1 Preferred stock, 457
301 Perfect information, comparing coordina- Prejudice within workforce, discrimination
ethics of private property, 305 tion systems with, 94-9 5 due to, 343-44
ill-defined property rights and tragedy Performance, sources of randomness i n , Premium
of the commons, 294-97 207-8, 220-2 1 risk, 2 1 0- 1 1 , 246-47
legal impediments to trade, 301-3 Performance contracts, 1 79 . See also Com­ total, 2 1 2, 2 1 3
transaction costs and efficient assign­ pensation; Incentive contracts takeover, 1 82, 5 1 1
ment of ownership claims, 303-5 Performance evaluation Present val ue, 4 5 1
untradable and insecure property comparative, 220-2 1 , 2 3 3 , 404 net, 4 5 1 -54
rights, 297-300 compensation and, 403-8 Price(s), price system, 56-87
concept of, 289-9 3 frequency of, 407-8 capital asset pricing model (CAPM),
conflicting interests of managers and Japanese auto supplier relations and 464-66, 472
owners, 49 1 -94 problems of, 567 competitive equil ibrium in, 66-67, 69-
of corporation , 3 1 4- 1 5 subjective, 404-8 7 1 , 74, 1 02
determinants of, 1 9 1 -92 Performance measurement coordination and, 27, 57-62, 89
efficiency of patterns of, 498 basing pay on, 2 1 4-1 5 price- vs. quantity-based coordination
employee, 3 I 5, 41 3, 4 1 5 pay attached to jobs vs. , 369-70 system vs. , 94- 1 00
group approach to, 29 5 difficulty of, 32 design decisions and, 1 04-6
incentive effects of, 8, 9, 1 1- 1 2, 190- asset ownership and, 3 1 1 - 1 2 economizing on information and com­
9 1 , 29 1-93 grouping tasks according t o ease of, 4 1 1 munication through, 1 00- 1 06
innovation in, 590-9 1 horizontal expansion and cost of, 573 of financial assets, information and,
international patterns of, 489-9 1 precision of, 222 467-75
neoclassical model of private ownership Performance pay, 1 0, 1 1 , 399 as fully decentralized, 1 14
economy, 62-69 for CEOs, 43 3-43 fundamental theorem of welfare eco-
overcoming ratchet effect using, 236 for groups, 41 3- 1 8 nomics and, 62-7 1 , 89
restructuring in Eastern Europe and for managers, 400 internal organizational use of, 78-83, 89
USSR and issue of, 592-93 screening and, 1 57-58 market-clearing, 60-6 1
separation of control and, 1 8 1 Performance standards. See Standards market failures and, 73-77
single, 296-98 Permanent employment. See Lifetime em- optimal, determining, 59-60
socialist, 1 3 , 297 ployment resale price maintenance, 39
stock, 1 1 - 1 2 Perot, Ross, 1 89, 498-99, 5 7 5 socialism and, 72-73
unified, 192-94 Perry, Nancy, 393, 4 1 3n, 4 1 7 transfer, 79-83 , 549- 5 1
See also Corporate control Persian Gulf War ( 1 990- 1 99 1 ), 296 Pricing decisions, 277
Owners' incentives for monitoring, 496-50 1 Peter Principle, 276, 374 Pricing formula, 462-64
Petroleum deposits, ownership rights to, Primary labor market, 3 59-60
Packard, David, 1 1 6, 547 296 Principal-agent problem, 2 14
Parallel job ladders, 366 Pettway, Samuel H . , 400n adverse selection in, 237-39
Paramount Communications, 499, 5 1 7 Pfeffer, Jeffrey, 274n incentive contracting and, 198-20 1
Parenting, commitment i n , 1 3 5 Phantom stock plans, 425 moral hazard in, 1 70
Pareto, Vilfredo, 23n Phibro, 9 See also Incentive contracts
Pareto dominated allocation, 23 Phillips Brothers, 9 Prisoners' Dilemma game, 1 3 8-39
Pareto optimal allocation , 23, 66 Physical assets, 1 34-3 5 Pritzger's, 290n
Participation constraint, 200 Pickens, T. Boone, 484, 488, 499-500, Private branch exchange (PBX) systems,
Participation incentive constraints, 141 568n 5 57-58
Participatory management, 275, 279-80 Piece rates, 236, 388, 392-96 Private corporation , 483, 490n
Partnerships, 4 1 3 , 4 1 8, 5 22-23 Piore, Michael , 3 59, 385 Private i nformation, 1 29, 1 59, 1 69, 34 1

liub 617
Pri, ate information (continued) Rasm usen, Eric, 283 Resource allocation problem, 90-9 3
precontractual opportunism and, 1 40- Ratchet effect, 14- 1 5, 23 3-36, 395 design attributes of, 91-92
47, 148 Rates of return innovation attributes of, 92-93
strategies for revealing, 1 54-59, 400-402 accounting, 4 3 8 Resource endowment, 63, 64, 6 5
Privately held corporation, 483 expected, 466 Responsibility
Private ownership economy, neoclassical shareholder, 4 3 8 compensation and, 2 5 8
model of, 62-69 See also Return(s) corporate social, 3 1 6
Private property, ethics of, 3 0 5 Rating agencies, reputation system replaced i n multidivisional firm, 545
Private sector, influence costs in, 27 1-73 by, 267 unity of, 4 1 0
Private-sector insurance programs, publi c Rationality, bounded. See Bounded ratio­ Restructuring, 5 1 7
VS. , 1 76-79 nality economic, in Eastern Eu rope and
Privatization process in Eastern Europe, Rationality-based theories of motivation USSR, 1 2- 1 6, 30 5-6, 591-94
30 5-6 and behavior, 42-43 voluntary, 5 1 9-20
Probabil ity mass function, 246 Rationing, adverse selection and, 1 5 3-54 See also Takeovers
Prodigy, 5 7 5 Rat race behavior, 372-73, 374 Retailing, franchise, 5 54, 5 6 1 , 563-6 5 ,
Product(s) Ravenscraft, David J . , 5 1 1 n 566
new, 1 1 0, 1 2 1-24 Raviv, Arthur, 1 62, 241 , 506n, 5 3 1 Retenti on, 3 44-47, 349- 50
steps in creation and delivery of, 5 52 RCA Corporation . 483 Retirement, mandatory, 1 89-90
Product-based definition of divisions, 546 Real value, 466 Return(s)
Product design decisions, 277 Recession of I 990, 487 excess, 462-63, 464
Product diversity, 3 Recruiting, 3 39-44, 430-3 1 gross, 1 72
Production costs, 3 3-34 Reebok, 424 net, 1 72
Production planning, 6 5 , 70, 7 5 , 89, 1 00- Reimer, Blanca, 499 owning complex return streams, 3 I 3- 1-f
1 0 1 , 1 0 3-6 Reisner, Marc, 299n residual, 290-93 , 3 1 5
Productivity, 1 5 , 1 5 5-56, 486-87 Relational contracts, 1 3 1-32, 1 39-40, risk and, 460-66
Product reputations, 259 2 59, 3 3 0 Returns to scale
Professional partnerships, 522-23 Relative performance evaluation, 438n constant, 99- 1 00
Profitability of incremental effort, 22 1-22 Renault, 589 decreasing, 99
Profit centers, 225-30 Reneging, 1 3 3, 1 3 5 , 1 39, 1 86 increasing, 74-7 5 , 99, 1 00
Profit maximization, 40-4 I Renegotiation, ex post, 1 34 Revelation principle, 14 2n
Profit sharing plans, 4 I 3-1 -f Rent(s), 3 3 3 Revlon, 484
Promotion policies, 364-69, 373, 3 74, competition for, horizontal expansion Revzin, Phi lip, 565n
3 7 5 , 376 and, 573-74 Rewards, 1 86, 366-67. See also Incentives
in Japanese firms, 3 5 1 defined, 269-70 Rey, Patrick, 24 I
optimal, 3 78-79 earned on "rising stars , " 3 7 5 Ricart i Costa, Joan, 445
subjective evaluation systems and, 406-7 quasi-rents vs. , 270 Rice, Valerie, 400n
up-or-out rules in, 379, 380, 382 vertical integration to capture, 560 Rich's, 484
Promotion tournaments, 367-69, 404, 428 Rent seeking Rights, property. See Property rights
Property rights Japanese methods of lim iting, 279-80 Rights offering, 499
for California water, 298-300 optimizing influence activities and, 273- Riordan, Michael, 24 I
creating, altering and assigning, 303-5 77 "Rising stars," 3 7 5
ethics of private property, 3 0 5 in public and private sectors, 270-73 Risk(s)
ill-defined, tragedy o f the commons and. Reorganization in bankruptcy, Chapter 1 1 , balancing incenti ves and, 207, 208-9
294-97 278-79, 503, 504, 5 1 0 commission system and, 396, 397
untradable and insecure, 297-300 Reporting relationships, defining, 547-48 of explicit i ncentive contracts, 1 87-88
Proudhon, P. J . , 3 0 5 Reputation and reputation effects, 1 86, financial , evaluation of, 209-1 1 , 246-4 7
Proxy, 1 87n 1 90, 2 5 7-69 piece rate and, 394-9 5
Proxy contests, 509- 1 0 as contract enforcers, 259-69 return and, 460-66
PTL, 5 24 aided by institutions, 266-69 systematic, 46 5
Public goods, investment projects to create, ambiguity, complexity and lim itations unsystematic, 464-65
1 4 5-46 of, 264-66 Risk aversion , 1 87, 2 1 0 , 222, 430
Public-goods problem, 294-98 in repeated transactions, 2 59-64 coefficient of absolute, 2 1 0, 2 1 3
Public policy toward layoffs, 340 firm as bearer of, 3 3 1-32 Risk costs, 1 88
Public sector for honesty, 263 Risk neutrality, 1 87, 2 1 0, 2 1 3 , 249
government insurance programs, 1 70-78 the law and, 265-66, 267 moral hazard with risk-neutral agents,
influence costs in, 270-7 1 , 274 product, 259 236-39
rent seeking in, 270-7 1 role of, I 39-40 Risk premium, 2 1 0-1 1 , 246-47
Punishment, commitment problem with, Resale price maintenance, 39 total, 2 1 2, 2 1 3
1 86 Research and development, 548 Risk sharing
Put options, 4 57 Reserve army of the unemployed, 2 5 3 efficient, 2 1 2- 1 3
Pyle, David, 5 3 1 Residual claimant, 29 1 i n employment relations, 333-38
Residual returns, 290-93 , 3 1 5 insurance and , 2 1 1 - 1 -f
Quality circles, 279 Residual rights of control , 289, 29 1-93, optimal, ignoring incentives, 2 1 3 ,
Qua�i-rcnts, 269-72, 3 1 6, 3 3 3 , 3 50, 375 3 1 -f 2 1 -f
Resolution Trust Corporation, 520 principle of, 2 1 1
Ra jm·c\h, B hagwan Shrcc, 524 Resource allocation within work group, 4 1 6- 1 7
Ra 11d01 11 11c�, i11 performance, source� in, efficiency of, 23 Risk taking
207-k, 220-2 1 parcto optimal, 66 creditor-shareholder conflicts of intere�t
Ra11do1 1 1 variable, 246 in private ownership economy, 65-66 and excessive, 49 5

61H liuh
Risk taking (continued) Scorched earth policies, 517, 520 Shortsightedness, technological, 454
motivating managerial, 429-32 Screening, 1 57-59 Shaven, John 8., 454n
by saving and loan associations, 171-75 employment discrimination and, 343-44 Showa Denko, 275, 519
deposit insurance and, 171-73 in recruiting, 341-42 Shultz, George, 1 5n
incentives for, 1 73-75 Sculley, John, 424 Siconolfi, Michael, 399n
Risk tolerance, 213, 416-17 Seapointc Savings and Loan, 174 Siegal, Donald, 487
Ritter, Bruce, 524 Search problems in market systems, 76-77 Siemens AC, 490n, 572
RJ R Nabisco, 437, 484, 485, 489, 493, Sears Roebuck, 79, 540, 541, 542, 544, Signaling, 155-56, 342-43
502, 511, 513 570, 575 financial decisions and, 505-8
Rob, Rafael, 162 Secondary labor market, 359 Simmons, I l arold, 521
Roberts, John, 85, 162 Second source, creating, 34 Simon, Herbert, 5 I, 161, 326, 3 54
Roll, Richard, 531 Secured creditors, 279n Simon and Schuster, I nc. , 370-71
Rolm Corporation, 5 56, 5 57- 58, 571, Securities, underwriting, 9, 10 Single ownership, 296-98
574, 575, 576 Securities and Exchange Commission Siow, Aloysius, 379n
Romer, David, 470n (SEC), 186, 434, 468, 489, 5 1 0, Skills, pay for, 399
Roos, Daniel, 538 521 Sloan, Alfred, 2 , 3, 4, 5, 16, 17, 111,
Rose-Ackerman, Susan, 531 Security market line, 464, 465 544, 545
Rosen; Sherwin, 354, 385, 438n , 44 5 Segmented-market strategy, 3 Smith, Abbie, 5 12n
Rosenthal, Robert, 162 Selection criteria in hiring, 342-4 3 Smith, Adam, 2 5, 26, 28, 50, 62, 69, 72,
Rosett, Joshua, 514n Selective intervention, 192-94 8 5, 257, 2 5 9
Ross, Stephen J., 196, 24 1 , 424, 477, 531 Self-employment, 236 Smith, Randall, 399n
Roth, Alvin E. , 43n Self-enforcing agreements, 3 3 2-3 3 Smith, Roger, 499, 576
Rothschild, Michael, 162 Self-selection, 169, 371-73 Smith, Vernon, 85
Routines, organizational, 43, 92, 265, in recruiting, 341-42 Socialism
273-77 via screening, 1 59 building, 13- 1 5
Rowntree , 5 1 2n Self-selection constraint, 15 5-56 market, 72-73, 94
Royal Dutch/Shell, 426 Sellers, coordination costs of, 29 ownership and, 13, 297
Royalties, 2 38 Sematech, 579, 586 price system and, 72-73
Rubinstein, Mark, 478 Semiconductor industry, 34, 579, 586 Social responsibility, corporate, 3 1 6
Semistrong form efficient markets hypothe- Software, computer, 309
Safeway, 484, 511 sis, 467, 469, 474 Sony Corporation, 20-21, 280, 426, 483,
Saga, 5 1 2 Senior debt, 457 490n, 570, 575, 589
Salary, 425 Senior executives, compensation of. See South Korea, 542
Sale, bargaining over a, 140-43 under Compensation Soviet Union
Sales commissions, 2 1 6- 1 7, 396-99 Seniority, 1 89, 190 coordination failure in, 77
Sales response function, 397-98 layoffs and, 190, 348 coup d'etat of August 1991, 5 93-94
Salomon Brothers, 9-12, 1 6, 17, 28, 78, Senior management, role in coordination distorted prices in, 73
292, 39� 432, 48� 48� 499 of, 115-16 economic restructuring in, 12-16, 591-
Salop, Joanne, 157, 162 Separation of ownership and control, 181 94
Salop, Steven, 157, 162 Separations, efficient, 347-49 ownership problems in, 297
Sample space, 246 Service industries, technological and orga- ratchet effect in, 233, 234
Samuelson, Paul, 477 nizational changes in, 588-89 Space Communications Corporation, 579
San Francisco Forty Niners football team, Severance pay, 437 Specialization, 2 5-26
265 Shapiro, Carl, 2 50, 2 53, 282 coordination and, 56-57
Sappington, David, 241 Shapiro, Matthew, 470n in multidivisional firm, 544-45
Sapulkas, Agis, 5 5 5 Shapiro-Stiglitz model of efficiency wages, Specific assets, 30-3 1
Sasser, W. Earl, 400n 250-54 in franchise retailing, 564-65
Satellite Business Systems, 570 Shared expectations, corporate culture and, hold-up problem and, 136-39, 307-8
Sato, Fumikata, 122 265 investments in, 135-39
Satterthwaite, Mark, 162 Shareholder-power movement, 499 stakeholders' rights and, 3 1 6-19
Savings and loan associations (S&Ls), crisis Shareholder rate of return, 438 Specific investment, 31
of, 154n, 170-76, 183, 292, 485, Shareholder resolutions, 510, 520-21 Spector, B. , 116n
488, 495, 520 Shareholder Rights Plans, 182 Spence, Michael, 15 5, 1 62, 1 96, 24 1, 3 54
Scale Shareholders. See Stockholders Sperry, 574
constant returns to, 99-100 Shareholding, concentrated, 497-500 Spin offs, 27 5, 5 1 7, 519. See also Restruc-
decreasing returns to, 99 Sharpe, William, 460n, 477, 478 turing
economies of. See Economies of scale Shavell, Stephen, 241 Spot market contracts, 1 3 1
increasing returns to, 74-75, 99, 1 00 Shearson American Express, 574 Spot market transactions, 2 5 9
operational, 106, 107, 408 Shepard, Andrea, 34n, 554n Sputnik, 12
structure of firm and, 106-7 Sherer, Peter D. , 429n Srinivasan, Venkantaraman, 397, 398
Scandals in not-for-profits, 524, 525 Sherman Antitrust Act (1890), 540 Stability, National I ntern Matching Pro-
Scanlon Plans, 414 Shirking, employee, 179-8 1, 189, 292, gram and, 4 5-46
Scherer, F. M., 531 37 1 -73 Staelin, Richard, 398n
Schevardnadze, Edvard, 15n Shleifer, Andrei, 477, 484n, 497n, 512n, Staggered boards, 517
Schiller, Robert, 477 5 1 3 n, 514-1 5, 5 31 Stakeholders
Scholes, M yron, 11, 415, 433n, 478, Short selling, 461 interests of, organizational objectives
491 n Shortsighted markets and management, and, 41-42
Scientists, compensation for, 399-400 471-73, 474 specific assets and rights of, 3 1 6-19
Scope, economies of, I 07, 5 54 cost of capital and, 4 53-54 Stalin, Joseph, 12, 13

I ndex 619
Stalk, George, Jr . , 275n, 519n Suppliers Toyoda, Ei ji, 4, 1 1 1
Standard Brands , 493 competitive use of independent, 5 5 5-56 Toyota Motors, 4-6, 16, 1 7, 27, 1 06, 1 1 1 ,
Standard Oil of New Jersey, 79, 540, 541- coordination with outside, 5-6 3 1 0, 3 1 2- 1 3, 490, 499, 566 , 567,
42, 5 42, 544 for Japanese auto industry, 565-68 569, 576, 578, 589
Standards owner-operated, 5 56 Trade credit, 278-79
in franchise retailing, 563, 564- See also Vertical boundaries and rela- Tragedy of the commons, 294-98
objective ways to set, 233 tions Training of Japanese permanent employ-
in piece-rate system, 395, 396 Supply and demand, 327-28 ees, 350-5 1 . See a/so Human
ratcheting up of, 233-36 Surplus sharing, 347-48 capital
Stanford University, 5 24 Synchronization problems, 91, 1 04-6 Transaction(s), 2 1
Start-ups, financing, 500-501 Systematic risk, 465 dimensions, of , 30-33
Startz, Richard, 354 System One, 5 5 5 between firms, 89-90
Stata, Ray, 4 1 4n Systems, internal labor markets as, 37 1-75 reputations in repeated, 259-64
Staten, Michael E., 1 79n Transaction costs, 28-35, 38 , 39, 30 1 -3
State of incorporation, 5 1 8 Takeover premium, 182, 51 I efficient assignment of ownership claims
Statistical concepts, 246 Takeovers, 483-84, 488 and, 303-5
Statistical independence, 211 , 246 bust-ups, 484-85 Transfer-line production system, 394
Stein, Jeremy, 478 golden parachutes following, 390, 437, Transfer pricing, 549-5 1
Steinberg, Saul, 484, 488, 5 I 9 518 in multidivisional firms, 79-83
Stigler, George, 85 hostile, 1 81 -83, 484, 490, 5 1 1, 5 1 5 Transfers of value, takeover premia due to,
Stiglitz, Joseph, 1 52, 1 62, 241, 250, 2 53, interpretation and evaluation of, 474-75 5 1 3-1 5
282, 477 in U . S . in 1 980s, 483-84, 5 1 0-2 1 Transportation technologies, 5 87, 589
Stock, preferred , 457 aftermath of, 5 20-2 1 Trans Union, 290n
Stock appreciation rights, 425 defenses against, 5 1 5-20 Trans World Airlines (TWA), 484, 502,
Stock awards (restricted) , 425 Tandem Computer, 309 5 1 3, 5 20
Stockholders Tax(es) Trench warfare in World War I, 264
conflicting interests of creditors and, capital gains, 507 Troy, Henry, 81
494-96 deferred compensation and, 4 33 True Value, 562
corporate control of, 508 financial structure decisions and, 458-59 Trust, 1 39
free cash flow and, 492 Tax shelter programs, gas and oil, 1 84-85, offering, 26 1-62
options open to dissatisfied, 509-1 0 224-26 in trench warfare in World War I, 264
ownership of corporation by, 3 1 4 Taylor, Frederick, 1 79 Tullock, Gordon, 283
poison pill and, 1 82-83 Teachers, incentives for, 230-3 1 Tully, Shawn, 427n
Stockholders' proxies, 1 87n Teagle, Terry, 4 1 9n Turnover, employee, 344-47
Stock market(s), 540 Team production, 292-93 human capital and, 344-45
CEO pay and performance in, 44 1-42 Technical analysis , 467 Japanese hiring policy and, 350
crash of October 1 9, 1 987, 470 Technological shortsightedness, 454 mobility costs in, 346--47
golden parachutes and, 437 Technology screening designed to reduce, 1 57
takeover premia due to mispricing by, divisions defined by, 546 TWA, 484, 502, 5 1 3, 520
5 1 2-13 globalization of, 5 89 Twain, Mark, 206n, 448n
Stock options, 390, 425, 43 1, 441 manufacturing and changes in, 579-80, Two supplier policy, 567, 568
Stock ownership, 1 1 - 1 2 586--88
Stock performance, tying pay to, 441-42 Teece , David, 3 1 0n UAL, 424
Strategic asset destruction, 504-5 Tehranian, Hassan, 43 5n Umbeck, John, 1 79n
Strategic Defense Initiative ("Star Wars"), Tender offers, 509, 5 1 1 Uncertainty
1 49 Tenneco, Inc, 1 94, 486, 575 asset ownership and, 308- 1 0
Strategic effects, 454n Tenure, 276, 379-82 decisions under, 209- 1 1
Strategic investments as design decisions, Terrorism, 135 of transaction, 3 1-32
454- 5 5 Theory , 594 Underinvestment, debt overhang and,
Strategic m isrepresentation, 46-48, 1 4 1 Third World, debt overhang and underin- 49 5-96
Strategy(ies) vestment in, 496 Underwriting securities, 9, 10
business Thomas, Robert Paul, 288 Unemployed, reserve army of the, 2 53
coordination and, 1 06-- 1 3 Thompson, Donald B . , 482 Unemployment, efficiency wages and,
core competencies and, 570-72 Thomson Corporation, 483 2 53-54
competitive (corporate), organizational 3M Company, 366 Unified ownership, 1 92-94
innovation and, 569 Time-and-motion studies, 233, 236 Union contracts, 329-30
modern manufacturing, 1 09-1 0, 1 12, Time Inc. , 5 1 7 Unions, 334
1 15-1 6, 408 Time Warner, 424, 499 Uniroyal Goodrich, 483
in normal form game, 26 1-62 Timing, strategic expansion and, 570--72 Unisys, 574
Strikes, 334 Tirole, Jean, 5 1, 1 1 9, 162, 241 United Automobile Workers (UAW), 337,
Strong form efficient markets hypothesis, Tobin, James, 477 340, 578
436, 467 , 469, 470-7 1 "Tobin's q," 486 United States, government insurance and
Student loans, 1 78 Tolerance, risk, 2 1 3, 4 1 6-- 1 7 guarantee programs in, 1 70-78
Subjective performance evaluation, 404-8 Tollison, Robert, 2 83 U . S. Constitution, 30 1
Subordinated debentures, 457 Torous, Walter, 504n U . S. Food and Drug Administration, 501
Sum itomo Chemical, 275, 519 Total risk premium, 2 1 2, 2 1 3 U . S . Memories, 586
Summers, Lawrence, 477-78, 51311, 531 Tournaments, 367-69, 404, 428 U . S. Supreme Court, 541
Su11kist, 562 elimination, 376 United Technologies Corporation, 429,
Supem1a jority rules, 5 1 7- 1 8 Towers, Perrin, Forster & Crosby, 426 486, 5 5 7

620 I nch
Unity of responsibility, principle of, 4 I 0 vertical integration , 1 59, 54 1 , 542, 549, Welfare economics, fu ndamental theorem
University boards of trustees, 52 5-26 5 52-5 3, 5 56-6 1 of, 62-7 1 , 83, 89, 1 00, 104
Unlim ited liability, 1 84 Vishny, Robert W. , 484n, 497n , 5 1 2n, Welles, Chris, 489
Unobserved characteristics, 1 49, 1 50 5 1 4- 1 5 , 5 3 1 Werncrfclt, Birger, 486, 487
Unocal, 484 Voluntary donations, 524-2 5 Western Europe
Unsecu red creditors, 279n Voluntary restructuring, 5 I 9-20 production planning in, 75
Unsecu red debentures, 457 Volunteers in not-for-profits, 5 2 5 public policy toward layoffs in, 340
Unsystematic (idiosyncratic) risk, 465 Volvo, 589 Westinghouse, 1 3 1 , 1 3 3
Up-or-out rule, 379, 380, 382 Von Neumann, John, 24 1 Weston, J. Fred, 1 82n, 48 5 n, 506n, 5 1 7n,
Upstream integration, 541 Voting rights, differential , 5 1 7 531
USX (former U. S. Steel), 1 82, 483, 499, Whalers, enforcement of norms among,
5 1 7, 540, 542 Wachter, Michael , 3 54 267-68
Utility fu nctions, 22, 36, 42 Waegelein, James, 4 3 5 n White knight, 5 1 2
Utility of consumption plan, 63-64 Wages Will iams, Joseph, 5 3 1
age/wage profiles, 1 57, 1 89, 3 3 5-37 Will iamson, Oliver, 19n, 5 1 , 1 26, 1 6 1 ,
Value, 456 classical theory of, 327-28 1 94n, 1 96, 322, 3 54
net present, 45 1-54 efficiency. See Efficiency wages William the Conqueror, 1 3 3, 1 39
nominal (money) vs. real, 466 Walt Disney Company, 42 5 , 439, 484, Wilson, James Q. , 1 66
Value creatio11, takeover premia due to, 5 1 9, 570, 589 Wilson, Robert, 24 1 , 283
515 Warner Brothers, 5 1 7 Winner's curse, 5 1 4
Value index, 36 Warranties, adverse selection and, 1 49-50 Winter, Sidney, 43, 5 1
Value maximization principle, 3 5-38, 74, Warrants, 457, 490 Wolf, Stephen M . , 424
83, 1 40, 148, 2 1 1 , 248, 249-50, Water rights in California, 298-300 Wolfson, Mark, 1 84n, 224n , 4 1 5 , 433n,
346, 436-37 Weak form efficient markets hypothesis, 49 1 n
Van Horne, James C. , 4 5 3n, 478 467, 469, 470, 471 Womack, James, 5 38
Varian, Hal, 478, 5 3 1 Wealth, distribution of, 249-50 Women in labor force, 59 1
Variance, 2 1 0, 246, 463 Wealth effects, 3 5-39 Wood, Robert, 54 1
Vector, 63 no wealth effects, 2 1 7, 248-49, 5 1 3 World War 1 , trench warfare in, 264
Venture capital, 500-50 1 Weigelt, Keith, 376 World War II, 89
Vertical boundaries and relations, 5 5 2-69 Weingast, Barry, 267n, 269n
cooperatives, 26-27, 4 1 3 , 562-63 Weintraub, Roy, 85 Xerox, 429
franchise retailing, 5 54, 5 6 1 , 563-6 5 , 566 Weirton Steel Company, 4 1 5
simple market procurement, advantages Weisbach, Michael S . , 563n, 566 Youngstown Sheet and Tube, 5 1 3
of, 5 5 3-56 Weiss, Andrew, 1 5 2, 1 62
supplier relations in Japanese auto in­ Weiss, Yoram, 385 Zeckhauser, Richard J. , 1 96, 24 1 , 478,
dustry, 565-69 Weitzman, Martin, 88, 89, 1 1 8, 241 500n , 5 3 1

l ndi·x 621
(continued from front flap)

Professor Milgrom is co-editor of the American

Economic Review and an associate editor of Games
and Economic Behavior and The Journal of Financial
Intermediation and is a Fellow of both the
Econometric Society and the Society of Actuaries. He
has received a Guggenheim Fellowship and
fellowships at the Center for Advanced Study in the
Behavioral Sciences and the Institute for Advanced
Studies in Jerusalem and has also served as an
associate editor of Econometrica, the Journal of
Economic Theory and the Rand Journal of
Economics. Recognized for his outstanding teaching,
he has also consulted for maj or corporations and
served as an expert witness in legal cases.

John Roberts is the Jonathan B. Lovelace Professor of

Economics in the Graduate School of Business at
Stanford University. He formerly taught at the
Kellogg Graduate School of Management at
Northwestern University and has served as associate
dean of Stanford's business school. He has been the
Bass Faculty Fellow at Stanford and a visiting
researcher at ·the Center for Operations Research and
Econometrics of the University of Louvain, the
Center for Advanced Studies in Jerusalem, and the
University of California, Berkeley. He has also
lectured on economics and management at
universities, research institutes, and government
agencies in East Asia, North America , and Western
Europe.
The author of more than fifty scholarly publications,
Professor Roberts has contributed to public
economics, general equilibrium theory, the
foundations of macroeconomics, the economics of
planning, and especially the theory of industrial
competition. With Professor Milgrom, he has recently
made major contributions to the economics of
organizations and management, incentives, and
manufacturing, and to the strategic analysis of
complementarities.

A fellow of the Econometric Society, Professor

Roberts is an associate editor of the American
Economic Review, Games and Economic Behavior,
the Journal of Economic Theory, the Journal of
Economics and Management Strategy and, formerly,
Econometrica. He has received faculty fellowships
from CORE and the Center for Advanced Study in
the Behavioral Sciences. He is a successful teacher in
MBA , PhD and Executive Education programs, has
consulted for major corporations and the United
States government , and has served as an expert
economic witness.

PRENTICE HALL
Englewood Cliffs, NJ 07632
Printed in the U.S.A.