PRICING IN COMPETITIVE

ELECTRICITY MARKETS
Topics in Regulatory Economics and Policy Series

Michael A. Crew, Editor

Graduate School of Management, Rutgers University
Newark, New Jersey, U.S.A.

Previously published books in the series:

Crew, M.:
Incentive Regulation for Public Utilities
Crew,M.:
Commercialization if Postal and Delivery S eroices
Abbott, T. A.:
Health Care Poliry and Regulation
Goff, B.:
Regulation and Mmroeconomir Peiformance
Coate, M.B. and A.N. Kleit:
The Economics if the Antitrust Proress
Franz, R. S.:
X-Efficienry: Theory, Evidenre and Applications (Second Edition)
Crew, M.:
Pricing and Regulatory Innovations Under Inmasing Competition
Crew, M., and P. Kleindorfer:
Managing Change in the Postal Delivery Industries
Awerbuch, S. and A. Preston:
The Virtual Utili!J
Gabel, D. and D. Weiman:
Opening Networks to Competition: The Regulation and Pricing ifArcess
Zaccour, G.:
Deregulation if Electric Utilities
Young, W.:
Atomic Energy Costing
Crew, M.:
Regulation Under Increasing Competition
Crew, M.A. and P.R. K.leindorfer:
Emerging Competition in Postal and Delivery S eroices
Cherry, B.A.:
The Crisis in Telecommunirations Carner Liabili!J:
Historical Regulatory Flaws and Recommended Riform
Loomis, D.G. and L.D. Taylor
The Future if the Telecommunications Industry:
Forecasting and Demand AnalYsis
Alleman,]. and Noam, E,
The New Investment Theory if Real Options and its
Implications for Telecommunications Economics
Crew, M. and P. Kleindorfer
Current Directions in Postal Riform
PRICING IN COMPETITIVE
ELECTRICITV MARKETS

edited by

Ahmad Faruqui
Retail and Power Markets Area
Science and T echnology Development Division
EPRI
Palo Alto, California, U.S.A.

and

Kelly Eakin
Christensen Associates
Madison, Wisconsin, U.S.A .

....
"
Springer Science+Business Media, LLC
Library of Congress Cataloging-in-Publication Data

Pricing in competitive electricity markets / edited by Ahmad Faruqui and Kelly Eakin.
p. cm. -- (Topics in regulatory economics and policy series)
Includes bibliographical references and index.
ISBN 978-1-4613-7043-7 ISBN 978-1-4615-4529-3 (eBook)
DOI 10.1007/978-1-4615-4529-3
1. Electric utilities--United States--Costs. 2. Electric utilities--Rates--United States. 3.
Competition--United States. 1. Faruqui, Ahmad. II. Eakin, Kelly. III. Topics in
regulatory economics and policy.

HD9685.U52 P74 2000

333.793'231 '0973--dc21
00-035709

Copyright © 2000 by Springer Science+Business Media New York

Originally published by Kluwer Academic Publishers, New York in 2000
Softcover reprint of the hardcover 1st edition
AII rights reserved. No part of this publication may be reproduced, stored in a
retrieval system or transmitted in any form or by any means, mechanical, photo-
copying, recording, or otherwise, without the prior written permission of the
publisher, Springer Science+Business Media, LLC

Pr:nted an acid-free paper.

Table Of Contents

List of Figures IX

List of Tables X1l1

List of Contributors xvii

Foreword XXXlll
J. ROBERT MALKO

Preface xxxv

Section I: Overview

1. Pricing Retail Electricity: Making Money Selling a Commodity 5

KELLY EAKIN AND AHMAD FARUQUI

Section II: Industry Restructuring and Its Pricing Implications

2. Pricing and Revenue Management 35

ROBERT G. CROSS

3. The Role of Price in the Restructured Electricity Market 39

LEONARD HYMAN
4. Competitive Infrastructure: As an Enabler of Market-Based Pricing 47
ERIC P. CODY

5. Competitive Rates: A Break from the Past? 65

JOHN NEUFELD

Section III: New Issues

6. Anticipating Competitor Responses in Retail Electricity Price

Design 85
KEN SEIDEN AND AHMAD FARUQUI

7. Understanding Latent Market Power in the Electricity Pool of

England and Wales 103
DEREK W. BUNN, CHRISTOPHER DAY AND KIRIAKOS VLAHOS

8. Market Design and Price Behavior in Restructured Electricity

Markets: An International Comparison 127
FRANK A. WOLAK

9. The Effect of Technology on Energy Pricing in a Competitive

Energy Market 153
CHRIS KING

Section IV: Risk Management in Volatile Markets

10. Managing Total Corporate ElectricitylEnergy Market Risks 165

ALEX HENNEY AND GREG KEERS

11. Managing Weather Risk in Energy Pricing: A Consumer Oriented,

Value Added, Energy Service 183
DUNCAN P. MACARTHUR

12. An Econometric Study of Weather's Effect on Prices 191

SCOTT P. MARTELLO

13. Electric Market Simulation 197

LANCE S. MUCKELROY
14. Energy Derivatives and Price Risk Management 211
1 AMES READ AND ART ALTMAN

15. Statistical Approaches to Electricity Price Forecasting 249

1. STUART MCMENAMIN AND FRANK A. MONFORTE

Section V: Case Studies

16. Using Customer-Level Response to Spot Prices to Design Pricing

Options and Demand-Side Bids 267
ROBERT H. PATRICK AND FRANK A. WOLAK

17. How to Buy Low and Sell High 295

MICHAEL T. O'SHEASY

18. Real Time Pricing - A Unified Rate Design Approach 307

STEVEN V. HusO

19. Dynamic Pricing and Profit Maximization Choices for the Investor
Owned Electric Disco 323
GEORGE R. PLEAT

20. Developing and Pricing Distribution Services 335

LAURENCE D. KIRSCH AND ROBERT 1. CAMFIELD

21. Pricing Throughout the Product Lifecycle: When Mature Markets

Meet Innovation 349
WILLIAM LEBLANC

22. Residential TOU Price Response in the Presence of Interactive

Communication Equipment 359
STEVEN BRAITHWAIT

23. Retail Pricing Tools to Meet Customer Needs 375

CHRISTOPHER 1. HOLMES

24 Pricing Options for the Baltic Electric Market 381

CHARLES ZIMMERMAN AND FLOYD DAVIS
Section VI: Pricing of Energy Services

25 Value-Added Services in a Competitive Electric Industry 397

ANNE SELTING

26 Measuring How Customers Value Electricity Service Offers 415

LISA WOOD, SUZANNE GAMBIN, AND PATRICIA GARBER

27 Electricity Marketing: Is the Product the Price? 427

JAMES LONG, BRYAN SCOTT, AND BERNIE NEENAN

Index 447
List of Figures

Figure 1-1 Time-of-Use and Seasonal Pricing ................................. 14

Figure 1-2 A Continuum of Risk-Differentiated Products .................... 16
Figure 1-3 The Retail Product Space ............................................ 28
Figure 4-1 Corporate Restructuring Roadmap................................. 52
Figure 4-2 Competitive Infrastructure Requirements -
Overview of Transaction Flows ..................................... 56
Figure 4-3 Forces Threatening Core Utility Systems .......................... 57
Figure 4-4 Activities Requiring Information Sharing..................... .... 58
Figure 4-5 Daily Load Estimation and Reconciliation Process .............. 59
Figure 4-6 Comparison of Generation Supply Options ....................... 61
Figure 5-1 Percentage of U.S. Electric Motor Power Used in
Manufacturing Supplied by Self-Generated Electricity .......... 71
Figure 6-1 Competitive Market Dynamic ....................................... 86
Figure 6-2 Oligopoly Market Equilibria ........................................ 88
Figure 7-1 NGC Representative Bidding Function ........................... 109
Figure 7-2 Marginal Cost Bidding Function ................................... 110
Figure 7-3 Profitability of Bidding-Up ......................................... 111
Figure 7-4 SMP Effects of Bidding Up ........................................ 112
Figure 7-5 Output Effects of Bidding-Up ..................................... 112
Figure 7-6 Revenue of Bidding-Up ............................................. 113
Figure 7-7 Profitability ofNP Bidding-Up .................................... 118
Figure 7-8 Profitability ofPG Bidding-Up .................................... 118
Figure 7-9 Profitability of NP Bidding-Up after Divestment. .............. 119
Figure 7-10 Profitability ofPG Bidding-Up after Divestment.. ............ 119
Figure 8-1(a) Average Prices throughout the Day for UK. .................. 142
Figure 8-1(b) Average Prices throughout the Week for UK ................ 142
Figure 8-2(a) Average Prices throughout the Day for NW .................. 143
x Pricing In Competitive Electricity Markets

Figure 8-2(b) Average Prices throughout the Week for NW ............... 143
Figure 8-3(a) Average Prices throughout the Day for VICT ............... 144
Figure 8-3(b) Average Prices throughout the Week for VICT ............. 144
Figure 8-4(a) Average Prices throughout the Day for NZ .................. 145
Figure 8-4(b) Average Prices throughout the Week for NZ ................. 145
Figure 8-5 Price STD throughout the Day for UK. .......................... 146
Figure 8-6 Price STD throughout the Day for NW ........................... 147
Figure 8-7 Price STD throughout the Day for VICT ......................... 147
Figure 8-8 Price STD throughout the Day for NZ ............................ 148
Figure 10-1 Risk and Regulation ................................................ 167
Figure 10-2 Expected Profit, Downside Profit and Profit at Risk.......... 178
Figure 10-3 Profit and Risks from a Set of Portfolios ........................ 178
Figure 10-4 Portfolio Profit Forecast for the Existing Position
of the Portfolio and After Possible Change (112) .............. 179
Figure 10-5 Portfolio Profit Forecast for the Existing Position
of the Portfolio and After Possible Change (2/2) ..... ......... 179
Figure 13-1 Price Duration Curves Example ................................. 199
Figure 13-2 Regulatory-Based Planning Process ............................. 200
Figure 13-3 Market-Based Price Development Process ..................... 201
Figure 13-4 Natural Gas Forward Curve ...................................... 203
Figure 13-5 Natural Gas Forward Curve with an Options Based
Uncertainty Range .................................................. 204
Figure 13-6 WSCC Forward Curve Projection .............................. 205
Figure 13-7 WSCC Forward Curve Projections .............................. 205
Figure 13-8 WSCC Forward Curve Projection with
Confidence Intervals ............................................... 207
Figure 13-9 Expected Versus Book Case WSCC Forward
Curve Projection ................................................... 207
Figure 13-10 WSCC Forward Curve Distribution ........................... 208
Figure 14-1 A Portfolio of Instruments ........................................ 213
Figure 14-2 Value At Risk (VaR) .............................................. 215
Figure 14-3 Cash Flow At Risk (CFaR) ....................................... 216
Figure 14-4 Terminal Payoffs: Forward Contract Long Position .......... 219
Figure 14-5 Terminal Payoffs: Forward Contract Short Position .......... 219
Figure 14-6 Terminal Payoffs: Call Option ................................... 222
Figure 14-7 Terminal Payoffs: Put Option .................................... 222
Figure 14-8 Short Hedgers ...................................................... 230
Figure 14-9 Long Hedgers ...................................................... 230
Figure 14-1O(a) Forward Price Curve (On Peak Power) .................... 237
Figure 14-1O(b) Forward Price Curve (Off-Peak Power) ................... 237
Figure 14-11 Forward Price Curve (Gas Market) ........................... 238
Pricing in Competitive Electricity Markets xi

Figure 14-12(a) Base-Load Genco Cash Flow-

Low Power Price Volatility .................................. 238
Figure 14-12(b) Dispatchable Genco Cash Flow-
Low Power Price Volatility ................................... 239
Figure 14-13(a) Base-Load Genco Cash Flow -
High Power Price Volatility .................................. 240
Figure 14-3(b) Dispatchable Genco Cash Flow -
High Power Price Volatility .................................. 241
Figure 14-14(a) Base-Load Genco Cash Flow -
Power and Gas Price Volatility .............................. 241
Figure 14-14(b) Dispatchable Genco Cash Flow -
Power and Gas Price Volatility .............................. 242
Figure 14-15(a) Base-Load Genco Cash Flow -
Correlated Power and Gas Markets ......................... 243
Figure 14-15(b) Dispatchable Genco Cash Flow-
Correlated Power and Gas Markets ......................... 243
Figure 14-16 Base-Load Genco Cash Flow Distribution (July 1998) ...... 244
Figure 14-17 Dispatchable Genco Cash Flow Distribution (July 1998) ... 245
Figure 14-18 Dispatchable Genco Cash Flow Distribution (July 1998) ... 246
Figure 15-1 PJM Average On-Peak Price ($/MWh), April 98 to May 99 252
Figure 15-2 PJM On-Peak Energy Demand (GWh), April 98 to May 99 252
Figure 15-3 Available Nuclear Capacity (MW), April 98 to May 99 ..... 253
Figure 15-4 Gas Prices at Henry Hub ($/mmBtu), April 98 to May 99 ... 253
Figure 15-5 Scatter plot of On-Peak Price vs. Energy ....................... 254
Figure 15-6 Actual and Predicted Values - Neural Network Model. ...... 260
Figure 15-7 Contribution of Linear Terms to Predicted Value .............. 261
Figure 15-8 Contribution of Node 1 (On Peak Energy) to
Predicted Value .................................................... 261
Figure 15-9 Contribution of Node 2 (Supply Terms) to Predicted Value 262
Figure 16-1 (a) Pool Selling Prices, April 1991 - March 1992......... .... 276
Figure 16-1(b) Pool Selling Prices, April 1992 - March 1993 .......... '" 276
Figure 16-1(c) Pool Selling Prices, April 1993 - March 1994 ............. 277
Figure 16-1(d) Pool Selling Prices, April 1994 - March 1995 ............ 277
Figure 16-2(a) Total System Loads, April 1991 - March 1992 ............ 278
Figure 16-2(b) Total System Loads, April 1992 - March 1993 ........... 278
Figure 16-2(c) Total System Loads, April 1993 - March 1994 ....... , .... 279
Figure 16-2(d) Total System Loads, April 1994 - March 1995 ............ 279
Figure 16-3 Mean Own Price Elasticities for Water Supply Firms ........ 283
Figure 16-4 Mean Own Price Elasticities for Copper, Brass, and
Other Copper Alloys Manufacturing Firms ................... 283
Figure 16-5 Mean Own Price Elasticities for Hand Tools and
Finished Metal Goods Manufacturers ........................... 284
xii Pricing In Competitive Electricity Markets

Figure 16-6 Mean Own Price Elasticities for Steel Tubes

Manufacturing Firms ............................................... 285
Figure 16-7 Mean Own Price Elasticities for Timber and Wooden
Furniture Manufacturing Firms .................................. 285
Figure 16-8 Mean Own Price Elasticities for Food, Drink, and
Tobacco Manufacturing Firms ................................... 286
Figure 16-9 Demand Response to Energy Price Changes ................... 288
Figure 16-10 Demand Response to Demand Charge Changes .............. 288
Figure 16-11 The Effect of Demand-Side Bidding on the Market
Clearing Spot Price ................................................ 289
Figure 16-12 Demand Response to a Price Increase in Load Period 27 ... 290
Figure 17-1 Customer Demand Profile ......................................... 298
Figure 17-2 Unit Cost by Hour Versus Unit Price by Hour ................ 302
Figure 17-3 Rate Complexity and Load Shape Risk ......................... 303
Figure 17-4 RTP Totals .......................................................... 304
Figure 22-1 Average Hourly Usage (kWhlhr)-Treatment and
Control Groups. Non Critical Weekdays, Month 1
(June 29-July 19, 1997) ............................................ 364
Figure 22-2 Average Hourly Usage (kWhlhr)-Treatment and
Control Groups. Critical Weekdays, Month 1
(June 29-July 19, 1997) ............................................ 364
Figure 22-3 Average Hourly Usage (kWh/hr)-By Rate Group.
Non Critical Weekdays, Month 1 (June 29-July 19, 1997) ... 365
Figure 22-4 Average Hourly Usage (kWhlhr)-Treatment and
Control Groups. Weekends, Month 1
(June 29-July 19, 1997) ............................................ 365
Figure 23-1 Impact Of Varying Per Unit Margins on
Overall Profitability ................................................ 378
Figure 24-1 Generation Mix- 1998 ............................................. 382
Figure 25-1 Monthly Costs of Green Power Versus Standard
Electric Service Line Break Pacific Gas and
Electric Territory ................................................... 406
Figure 26-1 WTP (Yo-Y\) is the Amount an Individual Would Pay
to Move to an Improved State of Electricity Services (ES *). 417
Figure 26-2 Risk Free Supply ................................................... 418
Figure 26-3 Customer Allocation of "Votes" Across Energy Offers ...... 420
Figure 27-1 ValueChoice sM Energy Products ................................ 435
Figure 27-2 Peak Price .......................................................... 436
Figure 27-3 Target Segment for SelectChoice sM ............................ 442
List of Tables

Table 1-1 A Menu of Retail Pricing Options ................................... 13

Table 1-2 Customer Choice and ESP Profit. ................................... 25
Table 1-3 Profit and Market Share Impacts of Response Strategies ......... 29
Table 3-1 Calculation of Price in the Regulated Framework ................... 40
Table 7-1 Percentage Change in Generation
(for marginal cost merit order) ........................................ 113
Table 7-2 Percentage Change in Profit Contribution
(for marginal cost merit order) ...................................... 114
Table 7-3 Percentage Change in Revenue (for marginal cost merit order) 114
Table 7-4 Percentage Change in SMP (for marginal cost merit order) .... 115
Table 7-5 The Herfindahl Index (non-nuclear companies) ................... 115
Table 7-6 Percentage Change in Generation (using NGC data) ............ 116
Table 7-7 Percentage Change in Profit Contribution (using NGC data) ... 116
Table 7-8 Profit Contribution ................................................... 120
Table 7-9 Generation .............................................................. 120
Table 8-1 Annual Means and Standard Deviations (SDs) of Spot
Price of Electricity in Home Currency per MWH ................ 130
Table 8-2 Annual Means and Standard Deviations (SDs) of Spot
Price of Electricity Converted to US$IMWH using Daily
Exchange Rate ........................................................ 132
Table 8-3 Ratio of (Highest Price - Lowest Price) + (Average Price)
over Various Time Horizons ......................................... 133
Table 8-4 R-Squared, Standard Error, and Sample Mean of
Dependent Variable for Regression Forecasting Half-Hourly
Pool Selling Price in England & Wales ........................... 136
XLV Pricing In Competitive Electricity Markets

Table 8-5 R-Squared, Standard Error, and Sample Mean of Dependent

Variable for Regression Forecasting Hourly Spot Price from
Nord Pool. ............................................................. 137
Table 8-6 R-Squared, Standard Error, and Sample Mean of Dependent
Variable for Regression Forecasting Half-Hourly Spot Price
from VicPool. ......................................................... 138
Table 8-7 R-Squared, Standard Error, and Sample Mean of Dependent
Variable for Regression Forecasting Half-Hourly Spot Prices
from NZEM-North Island Reference Node ........................ 139
Table 8-8 R-Squared, Standard Error and Sample Mean of Dependent
Variable for Regression Forecasting Half-Hourly Spot Prices
from NZEM-South Island Reference Node ........................ 140
Table /0-1 Value at Risk (confidence limit 99%, for a 2-week period) ... 168
Table 13-1 Key Electric Price Dri vers .......................................... 201
Table 13-2 WSCC Hydrological Generation by Year. ....................... 202
Table 13-3 MACC Nuclear Generation by Year. ............................. 202
Table 14-1 Valuation of Firm Fuel Contract. ................................. 225
Table 14-2 Valuation of Flexible Fuel Contract. ............................. 226
Table 15-1 Exponential Smoothing Summary ................................. 257
Table 15-2 ARIMA (0,1,4) (1, 0, 0) Summary .............................. 257
Table 15-3 Regression Model Results .......................................... 258
Table 15-4 Neural Network Model Results ................................... 259
Table 16-1 Sample Means and Standard Deviations of Components of
PSP .................................................................... 273
Table 17-1 RTP Customers ..................................................... 296
Table 17-2 Bill Reduction Impacts Due to Price Response ................. 297
Table 17-3 Commodity Purchasing Using a Combination of
Derivatives & Spot Purchases ...................................... 299
Table 18-1 One-Part RTP ....................................................... 316
Table 18-2 Two-Part RTP ....................................................... 316
Table 22-1 Specific Tier Prices for TOU Rates ............................... 361
Table 22-2 Daily Average Usage (kWh) ....................................... 366
Table 22-3 Elasticities of Substitution ......................................... 372
Table 23-1 Summary Impact from New Product Offering
(based on $0. 18IMMbtu incremental margin) ................... 378
Table 25-1 Types of Value-Added Services ................................... 398
Table 25-2 VAS Retailing Versus Commodity Retailing ................... 399
Table 25-3 VAS for Large Customers in England & Wales ................ 401
Table 25-4 Rating of VAS by Industrial Customers in England & Wales 402
Table 26-1 Hypothetical Energy Offers for Power Shopping Exercise .... 419
Table 26-2 List of Attributes by Cluster. ...................................... 422
Pricing in Competitive Electricity Markets xv

Table 26-3 Hypothetical WTP Calculations for Alternative

Electric Offers ........................................................ 424
Table 27-1 SelectChoice sM Formulation ...................................... 438
Table 27-2 SelectChoice SM Base Product Features & Design Challenges439
Table 27-3 SelectChoice SM Prices for the Initial Pilot Offering ........... .444
List of Contributors

Art Altman is the Manager of the "Asset and Risk Management Target"
for EPRI in Palo Alto, California. Art's background is in mathematical
finance and information technology and his focus is on R&D in valuation
and financial risk management. In this role, Art helps power producers
understand the risks that they face in the newly competitive bulk power
markets and helps them develop business and hedging strategies along with
associated software to manage value, risks and profit-making opportunities.
Art has been with EPRI since 1990. He is an Associate of the Society of
Actuaries and his financial experience includes four years with Metropolitan
Life Insurance where he developed strategies and models for pricing
financial products sensitive to a variety of contingencies such as interest rate
fluctuation. Art has also been employed by Wells Fargo Bank and Rockwell
International's Science Center. Art has an M.B.A. from the University of
California at Berkeley where he specialized in derivative valuation and risk
management. He also has a Masters degree in Computer Science from the
University of Rochester and a Bachelors degree in Mathematics from the
State University of New York at Buffalo.

Steven D. Braithwait is a Vice President and has managed projects at

Laurits R. Christensen Associates Inc. for the past nine years in the areas of
competitive pricing, measuring customer response to innovative pricing
programs, (e.g.: real-time pricing; time-of-use pricing with interactive
communications; wholesale market price forecasting; sales and load
forecasting; and demand-side management (DSM) impact evaluation). He
has also provided testimony before public service commissions in the areas
of forecasting and least-cost planning guidelines, and has recently reviewed
load forecasting techniques for clients in the U.S. and internationally. Prior
xv III Pricing In Competitive Electricity Markets

to joining Christensen Associates, Dr. Braithwait managed numerous

projects in the load forecasting, demand-side management and planning
areas at EPR!. He has delivered papers at numerous industry conferences on
the topics of innovative pricing, load forecasting, and DSM. Dr. Braithwait
holds a Ph.D. in Economics from the University of California, Santa
Barbara.

Derek W. Bunn is a Professor of Decision Sciences at London Business

School, where he directs the activities of the Energy Markets Group. He has
been involved in developing new approaches for forecasting electricity
demand for over twenty years, and for the past ten years he has also led a
modeling program seeking to understand strategic behavior in the
competitive market of England and Wales. He is Editor of Energy
Economics and holds degrees from Cambridge, Oxford, and London.

Robert J. Camfield is a Senior Economist with extensive experience in

the electric services industry. He manages projects involving market
organization, pricing strategy, cost assessment, and price forecasting, and is
responsible for coordinating the Network Economics and Power Engineering
Practice. He has recently managed a large restructuring project in Eastern
Europe. He has developed an innovative approach to pricing unbundled
distribution services utilizing marginal cost methods. Before joining Laurits
R. Christensen Associates, Inc., he worked with numerous organizations
within Southern Company including Pricing and Economic Analysis, Cost
Analysis, and Strategic Planning where he initiated, organized, and managed
system-wide projects. His tenure with Southern Company also includes
overseas assignments through Southern Energy International. Prior to
Southern, Mr. Camfield was chief economist of the New Hampshire Public
Utilities Commission, and has testified on numerous occasions, most
recently on transmission interface congestion. He is a graduate of
Interlochen Arts Academy and holds an M.A. in Economics from Western
Michigan University.

Eric P. Cody is President of the Retail Access Advisory Group, a

division of Way finder Group, Inc., which provides products and services to
utilities and energy retailers transitioning to competitive energy markets,
emphasizing detailed business processes, information flows, and new
operating models. Clients of the Retail Access Advisory Group serve more
than 30 million customers in twenty-nine states, Canada, Europe, and Asia.
Mr. Cody holds a B.A. (cum laude) from Amherst College and a Masters
degree in City and Regional Planning from Harvard University, where he
specialized in energy planning and policy analysis. He previously served for
Pricing in Competitive Electricity Markets XIX

five years as top information officer for the Wayfinder Group, Inc. (formerly
NEES Global, Inc.) companies.

Robert G. Cross is the Chairman of Talus Solutions, Inc. He is an

internationally recognized expert in the essential business practice of Pricing
and Revenue Management. He is the author of the New York Times
Business Best Seller, "Revenue Management: Hardcore Tactics for Market
Domination" (which has been translated into Chinese, French, German,
Japanese, Korean, Portuguese, and Russian) and is a respected and sought
after lecturer. He has been a consultant to many of the world's largest
airlines, hotels, and rental car companies, and his concepts have been
successfully extended to many other industries worldwide. Mr. Cross
attended Texas Tech University, where he received his B.A. in Chemistry,
with an emphasis on quantitative methods. After a tour in the Air Force,
where he was trained as a pilot, Mr. Cross earned his Juris Doctor degree
(cum laude) from Texas Tech. He is a member of both the Texas and
Georgia Bar Associations.

Floyd Davis currently manages electric power planning and policy

consulting activities for Bechtel Consulting. For the past twenty years, he
has conducted assessments of power markets in the U.S., Asia, Africa,
Europe, the Middle East and Central America and provided consulting
advice on strategic planning, project feasibility, tariff design and power
sector restructuring. He has most recently been an advisor to the energy
regulatory commissions in Latvia and Lithuania. Prior to working for
Bechtel, Mr. Davis has held positions at the MITRE Corporation and at the
Tennessee Valley Authority. He obtained a B.S. and an M.S. in Industrial
Engineering from the University of Oklahoma.

Christopher Day is a postdoctoral researcher at the University of

California, Berkeley, within the POWER group. His research has focussed
upon developing agent-based simulation techniques to understand dynamic,
strategic behavior in electricity markets, where market power issues are
significant. He was previously a research student at London Business
School.

Kelly Eakin is a Vice President at Laurits R. Christensen Associates Inc.,

and is a specialist in price theory, organization, and regulation of industry,
and environmental economics. He has worked on several projects involving
innovative service design, customer price responsiveness, and market
assessment. Dr. Eakin has managed real-time pricing projects with major
U.S. utilities. He also directed an EPRI project on the measurement, costing,
xx Pricing In Competitive Electricity Markets

and pncmg of ancillary services. Currently, Dr. Eakin is managing

development of EPR!' s Product Mix Model, a pricing tool for the energy
merchant. Dr. Eakin joined Christensen Associates in 1994 from the U.S.
Department of Agriculture where he developed expertise in environmental
and resource economics. Prior to his service with the U.S.D.A., Dr. Eakin
was on the faculty at the University of Oregon for seven years. At the
University of Oregon, Dr. Eakin was active in the graduate programs of the
Economics Department and the College of Business. His scholarly writings
have been published in a number of prestigious journals including The
Review of Economics and Statistics, The Journal of Human Resources and
The Southern Economic Journal. Dr. Eakin holds a B.A. in History from the
University of Texas at Austin and a Ph.D. in Economics from the University
of North Carolina at Chapel Hill.

Ahmad Faruqui is the Manager of the Retail and Power Markets Area at
EPR!. Over the past two decades, he has worked with more than fifty
energy companies and government agencies on restructuring energy markets,
retail business strategy, marketing tactics, pricing design, and demand
forecasting. In his career, he has held senior management positions at
several consulting firms, including A. T. Kearney, Barakat & Chamberlin,
Battelle-Columbus Division, and Hagler Bailly. He has also worked at the
California Energy Commission and the Applied Economics Research Center
at the University of Karachi. He has authored or co-authored more than one
hundred articles on energy issues and is co-editor of the book, Customer
Choice: Finding Value in Retail Electricity Markets. He graduated from the
University of Karachi, Pakistan with a B.A. in Economics and has a Ph.D.
from the University of California, Davis.

Suzanne Gambin is a consultant at PHB Hagler Bailly, a leading

management, and economic consulting firm serving energy companies
worldwide. Her expertise is in designing and implementing customer choice
studies, market research projects and database marketing programs to
support the formulation and execution of retail marketing strategies. Ms.
Gambin holds a B.A. in Economics and an M.B.A from the University of
North Carolina at Chapel Hill.

Patricia B. Garber is the Manager of Retail Products Marketing at

EPR!. She has over fifteen years experience in designing, managing, and
utilizing market research of manufacturers, distributors, advertising
agencies, utilities and other public agencies. In addition to the energy field,
she has worked for computer hardware and software manufacturers,
automobile manufacturers and distributors, financiaUeducational services,
Pricing in Competitive Electricity Markets xxi

and consumer packaged goods manufacturers and their advertising agencies.

Dr. Garber studies retail marketing strategy, customer loyalty, market
segmentation, and new products. She has a Ph.D. in Sociology from the
University of California, Los Angeles.

Alex Henney, formerly a director of London Electricity, published

Privatise Power in 1987 and was the first to advocate a competitive
restructuring of the electricity industry in England and Wales, including a
Pool as a competitive spot market. Subsequently, he has advised on
competitive restructuring in a number of countries, including the U.S., where
he has a joint venture, Competitive Electric Strategies Inc., with the
Resource Dynamics Corp. of Vienna, V A. He was founding secretary of the
International Association of Power Exchanges.

Christopher J. Holmes is Director, Pricing and Economic Analysis for

UtiliCorp United based in Kansas City, Missouri. In this capacity Mr.
Holmes' responsibilities include designing and implementing market based
pricing products for retail markets with particular emphasis on energy
delivery systems. Prior to this effort Mr. Holmes was Principal Economist
in the Retail Market Management Department at Cinergy Corp designing
retail price product and developing asset management metrics. In addition,
Mr. Holmes assisted in market planning, resource planning, and demand-
side management implementation activities. Prior to his marketing
experience, Mr. Holmes served five years as a rates analyst, designing
residential and commercial time differentiated rates as well as performing
marginal and embedded cost of service studies. Mr. Holmes has an M.S. in
Economics from Arizona State University, and a B.A. from the University of
Colorado.

Steven V. Huso is the Administrator, Rate Research, for Northern States

Power Company. In this position, he is responsible for developing and
managing new and refined pricing designs. Mr. Huso has been with NSP for
twenty years, in various pricing positions. He has represented NSP as an
expert witness for pricing in the States of Minnesota, Wisconsin, North
Dakota, and South Dakota. He has served on EPRI and Edison Electric
Institute pricing committees and presented papers on real-time pricing and
interruptible service pricing at conferences sponsored by EPRI and the
Association of Energy Engineers. He has also advised state-owned Taiwan
Power Company on real-time pricing and other electric pricing issues. Mr.
Huso received a B.A. degree in Mathematics and Economics from St. Olaf
College and an M.B.A. degree with a Finance concentration from the
University of St. Thomas.
XXll Pricing In Competitive Electricity Markets

Leonard S. Hyman is a Senior Industry Advisor to Salomon Smith

Barney's Global Power Group. From 1978 to 1994, as head of the Utility
Research Group and First Vice President at Merrill Lynch, he supervised and
maintained research on foreign and domestic energy and
telecommunications utilities. He was, also, a member of privatization teams
for offerings of British, Spanish, Mexican, Argentine, and Brazilian utilities.
Mr. Hyman has testified before Congress and has served on four advisory
panels for the U.S. Congress Office of Technology Assessment, and for a
study undertaken by the National Science Foundation. In addition, he was a
member of a Pennsylvania State task force on electric utility efficiency, a
NASA task force on fusion and other energy sources, and a blue ribbon task
force advising on the reorganization of the North American Electric
Reliability Council. He is the author of America's Electric Utilities: Past,
Present and Future (in its sixth edition), co-author of The New
Telecommunications Industry (in its second edition), The Water Business
and Unlocking the Benefits of Restructuring: A Blueprint for Transmission
and editor of The Privatization of Public Utilities. Far more than ten years,
Mr. Hyman was selected by Institutional Investor magazine as one of the
leading research analysts in his field. He is a Chartered Financial Analyst
(CFA) and is listed in Who's Who in Finance and Industry, Who's Who in
Science and Engineering, Who's Who in the World, and Who's Who in
America. He holds a B.A. from New Yark University (where he was elected
to Phi Beta Kappa) and an M.A. in Industrial Organization with a minor in
Latin American Studies from Cornell University.

Greg Keers is a founding director of KW International (KW). KW has

supplied software for trading and risk managing energy portfolios since
1993. This software, KW2000, is a front to back-office solution that is used
by more than thirty-five major energy trading companies in ten countries.
Since 1989, Greg has worked as a business consultant specializing in
deregulated power markets and has published several papers on the subject.
In the u.K., he has worked on the design and implementation of the England
and Wales electricity pool and several business planning assignments for
investor owned power companies. Since 1992, he has carried out trading
and risk management consulting assignments for many of the major players
in the Scandinavian (Nord Pool) market. Since 1996, he has also done
similar assignments for companies in other deregulating markets around the
world, including Australia, Austria, France, Germany, Netherlands, New
Zealand, North America, and Switzerland. He graduated from
Loughborough University of England with a B.Sc. honors degree in
Mathematical Engineering.
Pricing in Competitive Electricity Markets XXlll

Chris King is Founder, Director, and Chief Executive Officer of

Utility.com, Inc. a leading Energy Service Provider in deregulated energy
markets. Mr. King is former Vice President-Regulatory Affairs and
Strategic Planning for CellNet Data Systems, the nation's leading
independent provider of advanced metering and data services to electric
utilities and competitive electricity sellers. Mr. King, as VP-Sales &
Marketing, developed CellNet's business plans and helped lead the company
in obtaining contracts now totaling over five million advanced meters from
utilities nationwide. Prior to CellNet, Mr. King directed several rate
programs at Pacific Gas & Electric, including marketing time-of-use,
interruptible, and other pricing options to residential and commercial
customers. Mr. King is a nationally recognized expert in utility
deregulation, having published widely and testified regularly before state
and federal regulatory commissions, as well as the Commerce Committee of
the U.S. House of Representatives. Mr. King holds a bachelor's and
master's degree in environmental sciences from Stanford University, as well
as a master's degree from its Graduate School of Business, where he was a
Sloan Fellow.

Laurence D. Kirsch, who since 1982 has specialized in economic

analysis of the electric utility industry, leads Christensen Associates'
research on the changing structure of electric power markets. This research
has encompassed studies of bulk power markets, power pool operations,
electric power system cost structures, and reliability costs. Dr. Kirsch has
developed and applied methods for estimating the real-time marginal energy
and reliability costs of both generation and transmission; has developed
methods for costing and pricing unbundled ancillary services; has evaluated
the relative merits of various schemes for auctioning wholesale power; has
participated in the development and implementation of pricing policies for
independent power producers; and has assessed a wide variety of utility
pricing practices at both the wholesale and retail levels. Dr. Kirsch holds a
Ph.D. in Economics from the University of Wisconsin at Madison.

William J. LeBlanc, Vice President of E source, heads E source's

Research Department which produces the company's products and services.
He also develops reports and workshops on strategic topics for the energy
services industry, as well as guiding much of E source's product
development. Mr. LeBlanc has over a dozen years of experience in strategic
marketing, new product development, pricing, market research, and demand-
side management. Prior to joining E source in January 1998, he worked for
six years as a director at Barakat & Chamberlin, a national consulting firm.
xxiv Pricing In Competitive Electricity Markets

He founded and served for several years as president of the Association of

Energy Services Professionals, one of the industry's major professional
organizations. Some of Mr. LeBlanc's recent publications include Pricing
Strategies for Competitive Energy Markets, Perrier or Plain Water?
Branding in the New Energy Marketplace, and Guide to New Product
Developmentfor the Emerging Market Environment. Mr. LeBlanc served as
a project manager at the EPRI from 1988-91. He managed projects in
demand management, rates, marketing, and customer behavior; developed
promotional programs for EPRI products and services; and conducted
numerous conferences and workshops. Mr. LeBlanc holds both an M.S. and
a B.S. in Mechanical Engineering from Stanford University and a B.A. in
Management Engineering from Claremont McKenna College.

James B. Long (Ben) has over thirty-four years experience in the electric
utility business. Over twenty years of this experience has been in the pricing
and costing areas, with major emphasis on the development of special
pricing programs for targeted market segments. Past experience has
included managing the Rate Department at Public Service Company of
Oklahoma where he was responsible for directing the pricing, costing, and
load research activities for regulatory activities and filings both at the state
and federal level. Currently he directs the pricing product activities for
Central and South West Corporation. Other experience includes electric
distribution design and operation, industrial marketing, and pricing for
Georgia Power Company. He has served as chairman for the EPRI
Innovative Pricing group and is currently chairman of the Producing
Successful Retail Products and Services group. He is a member of the
Professional Pricing Society. He has testified before several jurisdictions on
costing, pricing, and policy issues. Ben holds a B.S.E.E. from Auburn
University and an M.B.A. from Georgia State University.

Duncan P. MacArthur is developing international partners in the Pacific

Rim, Europe, and Canada for sales and service of WeatherWise USA LLC's
weather risk management services. WeatherWise helps innovative energy
marketers differentiate their retail products, hedge corporate weather risk,
and quantify weather risk in trading and purchase. Mr. MacArthur has
managed business development for a broad range of new ventures and
turnarounds in the energy industry. His experience includes consumer
energy products, financial services, generation equipment, and fiber-optic
grid control/communications systems. He also rebuilt sales and service
operations to turnaround sales of power equipment in the u.s. and Europe.
Mr. MacArthur has a B.S. in Mechanical Engineering from the University of
Pricing in Competitive Electricity Markets xxv

Virginia, an M.B.A from Harvard University and is a Registered

Professional Engineer.

J. Robert Maiko is a Professor of Finance in the College of Business at

Utah State University. He serves as an Advisory Council Member of the
Society of Utility and Regulatory Financial Analysts, and was president of
the society from 1988-1990. He is on the Board of Directors of the National
Regulatory Research Institute at The Ohio State University and serves on the
Advisory Council of the Center for Public Utilities at New Mexico State
University. Dr. MaIko has served as Chief Economist at the Public Science
Commission of Wisconsin (1975-1977 and 1981-1986) and has served as
Chairman and Vice Chairman of the Staff Subcommittee on Economics and
Finance of the National Association of Regulatory Utility Commissioners.
He also served as Program Manager of the Electric Utility Rate Design
Study at the EPR!. Dr. MaIko received a B.S. in Mathematics and
Economics from Loyola College, and an M.S. and Ph.D. in Economics from
the Krannert Graduate School of Management at Purdue University.

Scott Martello is a risk management specialist at the Tennessee Valley

Authority. In the past five years in the profession he has developed several
econometrics based models for application to risk management, load
forecasting, and asset optimization. In addition to analysis work Mr.
Martello has also worked as a power trader at TV A. He also teaches
economics at Northwestern Technical Institute in Rock Spring, GA. In 1992
he earned a B.A. in Economics from the University of Pittsburgh at
Johnstown and in 1994, earned an M.S. in Economics from Auburn
University.

J. Stuart McMenamin is Executive Vice President at Regional

Economic Research, Inc., where he specializes in the fields of energy
economics, statistical modeling, and software development. Over the last
twenty years, he has managed numerous projects in the areas of system load
forecasting, price forecasting, retail load forecasting, end-use modeling,
regional modeling, load shape development, and utility data analysis. He
has also directed the development of several software packages, including
the EPRI end-use models, time-series forecasting models, load shape
development tools, and market analysis tools. In prior jobs, Dr. McMenamin
worked for Criterion, Inc., where he specialized in statistical modeling of
telecommunications and energy markets. He worked for the President's
Council on Wage and Price Stability under the Carter administration. He
also was a lecturer in economics at University of California, San Diego,
where he taught courses in Microeconomics, Finance, and International
xxvi Pricing In Competitive Electricity Markets

Trade. Dr. McMenamin received a B.A. in Mathematics from Occidental

College and a Ph.D. in Economics from U.C.S.D.

Frank A. Monforte is Vice President of Forecasting at Regional

Economic Research, Inc., where he specializes in the areas of energy and
price forecasting, end-use forecasting, and statistical and mathematical
modeling. Dr. Monforte directs the development and support of RER's
short-term forecasting tools and services. His forecasting expertise includes
the application of neural networks in the areas of short-term load and price
forecasting, retail scheduling, and dynamic load profiling. In addition to his
forecasting responsibilities, he is a nationally recognized authority in the
area of industrial end-use analysis. Reflecting this expertise, he manages the
software and data development efforts for EPR!' s industrial end-use
forecasting model, INFORM. Prior to joining RER, Dr. Monforte worked
for Southern California Gas, where he developed a series of models dealing
with competition in the natural gas industry, including a pipeline capacity
bidding model and a least cost supply-planning tool. Dr. Monforte received
a B.A. in Economics from the University of California, Berkeley, and a
Ph.D. in Economics from the University of California, San Diego.

Lance Muckelroy received a B.S. in Petroleum Engineering from the

University of Texas at Austin in 1987 and an M.B.A. from the University of
Texas at Austin in 1990. He joined Houston Lighting & Power, now a
division of Reliant Energy, Incorporated, in 1990. His current position is
Director of Market Analysis in the Regulatory Planning and Analysis Group.
His primary responsibilities involve the development of market clearing
price projections for electricity and assisting the organization in its transition
to competition in Texas.

Bernie Neenan is the General Manager ofAXS Marketing, L.L.c. He

has twenty years experience working on electricity industry restructuring and
pricing issues ranging from refining and adapting traditional rates to meet
new challenges to designing innovative pricing platforms that open up new
market opportunities and respond to competitive pressures. His clients
include utilities, customers, and industry groups throughout the U.S. and
Canada and other countries including South Africa, England, New Zealand,
and Ecuador. Bernie was recognized by an EPRI Innovator's Award
program for his contributions to developing and testing dynamic pricing
products such as RTP and priority services. Prior to forming AXS, Bernie
managed Electrotek Concept's utility pricing practice following five years
service as Manager of Rates at Niagara Mohawk Power Corporation. He
Pricing in Competitive Electricity Markets xxvii

holds advanced degrees in Agricultural Economics from Cornell University

and the University of Florida.

John L. Neufeld is a Professor of Economics at the University of North

Carolina at Greensboro. He is a member of the North Carolina Energy
Policy Council. Dr. Neufeld has served as a senior economist at the
Research Triangle Institute in North Carolina. He has authored a number of
articles on energy issues including electricity rate design. Much of his
current research has concerned the historical development of the U.S.
electric power industry. He graduated from Yale University with an AB. in
Economics and from the University of Michigan with a Ph.D. in Economics.

Michael (Mike) T. O'Sheasy is the Manager of Product Design for

Georgia Power Company, an operating company in the System Company
system. His responsibilities include pricing strategy development and future
rate planning; rate research, design, and evaluation; the preparation and
filing of retail rates with the Georgia Public Service Commission and the
forecast of base rate revenues for the corporate budget. He joined Southern
Company Services in 1980 as an engineering cost analyst and progressed
through various positions in the Marketing and Regulatory Support
Department, specializing in allocated cost of service studies. While at SCS,
he was selected for the Southern's Superlative Award. Since joining GPC,
Mike has been selected for the Team Excellence Award. In addition, he has
won numerous EPRI awards related to pricing, including EPRI's Innovator
and Technology Transfer awards. Mike is a graduate of the Georgia Institute
of Technology with a B.S. in Industrial Engineering and an M.B.A from
Georgia State University.

Robert H. Patrick is an Associate Professor in the School of

Management at Rutgers University. His current research on electricity and
natural gas markets includes pricing options under competition, predicting
consumer demands under alternative pricing options, designing market
structure and rules in restructured markets, and incentive regulation. Prior to
joining Rutgers' faculty, he was a Manager at EPRI and held academic
positions at Purdue and Stanford Universities. He has published numerous
articles on pricing, regulation, energy, and environmental economics in
professional journals and books; served on editorial, governmental, and
private advisory boards, nationally and internationally; and is a charter
member of the New Jersey Council of Academic Policy Advisors. He
earned a Ph.D. in Economics from the University of New Mexico and a B.A
(magna cum laude) from Blackburn College.
XXVlll Pricing In Competitive Electricity Markets

George R. Pleat is a Principal Pricing Analyst with Baltimore Gas and

Electric Company (BGE) where he directs the Company's efforts in retail
price unbundling. In 1984, Mr. Pleat joined BGE to provide the lead for the
Company on strategic cost studies and subsequent price design applications.
From 1981 to 1984 Mr. Pleat worked with the Minnesota Public Service
Department as a utility rates analyst testifying on electric and gas rate design
and cost of service issues representing the broad public interest before the
Minnesota Public Utilities Commission. Prior to 1981, Mr. Pleat was a
market specialist for the U.S. Postal Rate Commission. Mr. Pleat has an
M.A. in Economics from George Washington University (1981) and a B.S.
in Business Administration from Duquesne University (1975). George is
currently Secretary of the Southeastern Electric Exchange Rate Section and
the Chairperson of NERA's Marginal Cost Working Group. He has
published three articles in Public Utilities Fortnightly magazine: "Pricing
and Profit Strategies of a Stand-Alone Electric Distribution Company"
January 15, 1997; "Should Metering Stay at the Stand-Alone DISCO?"
February 1, 1998; and "Unbundling Retail Prices: An Electric Utility
Prepares for Life as a Disco" May 15, 1999.

James A. Read is a Principal with The Brattle Group, an economic,

environmental and management consulting firm in Cambridge,
Massachusetts. He is an expert in capital budgeting, corporate finance, and
regulatory economics. His consulting practice has focused on the public
utility, natural resource, and transportation industries. Mr. Read is currently
working with the EPRI to develop methods and software for commodity
price risk management. He is also author of related EPRI reports, including
Valuation and Management of Nuclear Assets and Option Pricing for
Project Evaluation. Mr. Read has also provided legal counsel with advice
on asset valuation, cost of capital, and damages. He has worked extensively
with academic experts in finance and economics. Mr. Read was formerly a
Principal with the consulting firm Incentive Research Incorporated.
Incentive Research and The Brattle Group merged in January 1995. Mr.
Read came to Incentives Research from the firm of Charles River
Associates, where he was the Director of Financial Consulting. He holds a
B.A. in Economics from Princeton University and an M.S. in Finance and
Statistics from the Sloan School of Management at the Massachusetts
Institute of Technology.

Bryan J. Scott has twenty years experience in the electric utility

business. Nineteen years of this experience has been in the pricing and
marginal costing areas, with major emphasis on the development of special
pricing programs for retail customers. Past experience has included
Pricing in Competitive Electricity Markets XXlX

managing the Pricing section at Public Service Company of Oklahoma

where he was responsible for directing the rate design and pricing research
activities for regulatory activities and filings and responsibility for managing
the Pricing and Costing areas for Central and South West Corporation.
Currently he leads the pricing development activities for Central and South
West Corporation. He has testified before several jurisdictions on pricing,
marginal costing, and product development issues. He is a member of the
Professional Pricing Society. Bryan has a B.S. in Economics from the
University of Tulsa.

Ken Seiden is a Vice President at Quantec; a consulting firm that

provides economic, engineering, statistical, and strategic planning services.
He has over fifteen years experience appraising the demand for new products
and services, performing competitive and market assessments, conducting
program evaluation, pricing, forecasting, and market segmentation analyses,
developing economic software, and formulating strategic business plans.
Prior to joining Quantec, he was President of Essential Economics and a
project director with Barakat & Chamberlin. In these positions, he directed
dozens of consulting projects over a ten year period. Earlier in his career he
worked as a market economist for Pacific Bell Telephone and the Bonneville
Power Administration. Dr. Seiden has published articles in Energy Policy,
Energy Economics, The Electricity Journal, Applied Economics, and Public
Choice. He has presented nearly twenty papers at various conferences, and
has co-authored several methodology reports for EPRI dealing with utility
resource planning, forecasting, and pricing issues. He has a Ph.D. in
Economics from the University of Oregon.

Anne Selting is a Consultant at National Economic Research Associates

(NERA) in San Francisco. Prior to joining NERA, Ms. Selting worked in
the independent power sector and for the California Energy Commission. At
NERA, Ms. Selting has also been involved in issues related to retail
commercial strategy, customer switching behavior in newly competitive
markets, and market share forecasting methods. She has also worked in the
areas of information requirements for direct access, customer load profiling,
and retail metering issues. She is a co-author of numerous publications
related to retail energy markets. Ms. Selting received a B.S. (with honors) in
Journalism and Political Science from the University of Colorado and an
M.S. in Applied Economics at the University of Minnesota. As a U.S.
Department of Agriculture Fellow, she also completed a year of doctoral
study in the Agricultural and Resource Economics department at the
University of California, Davis.
xxx Pricing In Competitive Electricity Markets

Kiriakos Vlahos is assistant professor of Decision Sciences at London

Business School. He has spent over ten years as a researcher and consultant
on electricity planning and privatization issues. His particular expertise is in
the joint application of optimization and simulation techniques, and for some
of this work, he was awarded the Goodeve medal of the u.K. Operational
Research Society in 1993. He received a Ph.D. from London Business
School in 1991.

Frank Wolak is a Professor of Economics at Stanford University. His

fields of research are industrial organization and empirical economic
analysis. He specializes in the study of privatization, competition, and
regulation in network industries such as electricity, telecommunications,
water supply, natural gas, and postal delivery services. He is the author of
numerous academic articles on these topics. He has worked extensively on
the empirical analysis of market power in the England, Wales, California,
Norway, Sweden, Spain, Zealand, and Australian electricity markets. He is
a Research Associate of the National Bureau of Economic Research and an
associate of the University of California Energy Institute in Berkeley.
Professor Wolak has served as a consultant to the California and U.S.
Departments of Justice on market power issues in the telecommunications,
electricity, and natural gas markets. He has also served as a consultant to the
Federal Communications Commission and Postal Rate Commission on
issues relating to competition in network industries. He is the Chairman of
the three member's Market Surveillance Committee for the Independent
System Operator of newly restructured California Electricity Supply
Industry. In this capacity, he has testified several times at the Federal
Energy Regulatory Commission on issues relating to market monitoring and
market power. He has also served as an advisor to the governments of
various countries around the world, including Mexico, Indonesia, and Spain
on issues relating to electricity industry restructuring.

Lisa Wood is a Principal at PHB Hagler Bailly, where she specializes in

the design, implementation, and interpretation of market research. She has
extensive experience in the areas of retail customer choice, willingness to
pay for new products and services, customer prioritization, and customer
valuation. Lisa has directed a wide range of studies in the electric utility
industry over the past ten years and has published in leading journals and
spoken at numerous marketing and electric utility industry conferences.
Prior to joining PHB Hagler Bailly, Dr. Wood directed the Decision and
Market Analysis group at Research Triangle Institute. Dr. Wood holds a
Ph.D. from the Wharton School of the University of Pennsylvania.
Pricing in Competitive Electricity Markets XXXI

Charles Zimmermann currently works for Bechtel Consulting as resident

advisor to the Energy Regulatory Commission in Lithuania. In that capacity,
he has worked closely with other regulatory entities and electric power
utilities in the Baltics in exploring the potential structure of the Baltic energy
market. He has directed numerous projects related to the transition from a
socialist economy to a market economy in the energy sectors of central
European countries and newly independent states and has also conducted
assignments in Egypt, Pakistan, New Zealand, u.K., and Canada. Prior to
working for Bechtel, he held a senior position with Hagler Bailly and has
worked for Foster Associates and the Council on Environmental Quality. He
graduated from Brown University with an A.B. in Architecture and City
Planing and has a Ph.D. in Resource Economics from Cornell University.
Foreword

During the past twenty-five years, the pricing of electricity has received
significant attention and has addressed various challenges. i Some of these
pricing challenges include: time-of-use considerations, conservation, lifetime
issues, co-generation, retention of large use customers, reliability
considerations, incentive regulation, de-regulated markets, and customer
choice. Pricing by objective has been an important theme in addressing
changing challenges. ii '
During the decade of the 1990' s, some electricity markets are being de-
regulated and other electricity markets are facing innovative regulatory
frameworks. Financial and economic restructuring are occurring in
electricity markets. Pricing issues and strategies increasingly focus on
customer choice, business risks, and economic value-added services.
Technological, marketing, and financial considerations are clearly playing
important roles in emerging pricing strategies.
In changing electricity markets, new pricing strategies will need to
consider cost factors (including marginal costs and accounting costs), value
of service factors (including price elasticity of demand), specific market
factors (including customer usage characteristics), and pricing by objective
(including efficiency and equity). Insights and lessons from the twentieth
century will be useful in the development of effective pricing strategies for
electricity markets in the twenty-first century.iii
xxxiv Pricing In Competitive Electricity Markets

A variety of chapters in this book discuss important theoretical pricing

issues and practical pricing applications in the changing electricity markets.
This book presents useful information that should help the reader address
challenging and complex pricing issues during the twenty-first century.

J. Robert Maiko
Professor of Finance
College of Business
Utah State University

NOTES

i For a discussion of some pricing challenges and related issues. see:

I. Richard D. Cudahy and J. Robert Maiko, "Electric Peak-Load Pricing: Madison Gas
and Beyond," Wisconsin Law Review, Volume 1976, Number 1, Spring 1976, pp.
47-78.
2 J. Robert Maiko, Darrell Smith, and Robert G. Uhler, Costing For Ratemaking:
Topic Paper #2, a report to the National Association of Regulatory Utility
Commissioners, Electric Utility Rate Design Study, Report No. 85, EPRI, Palo Alto,
California, August 1981.
3. Ahmad Faruqui and J. Robert Maiko, editors, Customer Choice: Finding Value In
Retail Electricity Markets, published by Public Utilities Reports, Inc., Vienna,
Virginia, 1999.

ii For an insightful discussion of pricing by objective within the traditional regulatory

framework, see James C. Bonbright, Principles of Public Utilities Rates, Columbia
University Press, New York City, New York, 1961, pp. 291-292.

iii For an overview of restructuring issues facing the electricity power industry, see Gregory B.
Enholm and J. Robert Maiko, editors, Electric Utilities Moving Into the 21 SI Century,
published by Public Utilities Reports, Inc., Arlington, Virginia, 1994.
PREFACE

This book introduces a new family of pricing concepts, methodologies,

models, tools and databases that fall under the rubric of market-based pricing.
Vertically integrated utilities for whom tariffs were established by regulatory
commissions using cost-of-service concepts are rapidly giving way to power
marketers, energy service providers and utility distribution companies. These
new players have to compete in order to earn a profit.

In traditional rate making, utility profits were almost guaranteed, being

the product of a rate of return and a rate base, both established through a
regulatory process. Prices were a residual, set to recover a revenue
requirement that would ensure the realization of this rate of return. If
operating costs went up, prices would go up, some times automatically
through devices such as the fuel adjustment clause. Quantity sold was held
fixed, based on a utility demand forecast for a "test year."

By contrast, under market-based pricing, profits are a residual concept,

and represent what is left over when the costs of doing business are deducted
from revenues. Prices are determined through the free interaction of demand
and supply. Initially, prices tend to go into a free fall, as price wars break out.
Losses are pervasive. Over time, the emergence of new products and
services, coupled with market segmentation, restores profitability and create
customer value.

Many but not all of the papers in this book were presented at an EPRI
conference. We express our sincere thanks to the authors who have
contributed to this book. Our understanding of prices and competition in
energy markets has benefited greatly from discussions and interactions with
our colleagues at EPRI and Christensen Associates.
XXXVI Pricing in Competitive Electricity Markets

Ahmad would like to acknowledge his debt to Bob Maiko and Rene Males
who introduced him to the topic of electricity pricing as he joined the staff of
EPRI's Electric Utility Rate Design Study in 1979. We have also benefited
enormously from the editorial assistance of Pamela Arauz, Peggy Prater, and
most notably Joan Stephens.

We would like to acknowledge the enormous sacrifices that our spouses,

Nuzhat and Marla, have made as we have devoted evenings and weekends to
reading, editing, and writing the papers that appear in this book. Without
their emotional support and forbearance, this book would not have been
possible. Finally, Ahmad would like to thank his college-going daughters,
Furah and Saba, for sharing with him some of their unbounded youthfulness.

A.F.
Danville, California

K.E.
Madison, Wisconsin
PRICING IN COMPETITIVE
ELECTRICITY MARKETS
SECTION I

OVERVIEW
Chapter 1

Pricing Retail Electricity: Making Money Selling a

Commodity

Kelly Eakin and Ahmad Faruqui

ulUrits R. Christensen Associates Inc. and EPRI

Key words: Break-even Prices; Building Blocks; Bundling; Deregulation; Energy Service
Provider; Forward Contracts; Guaranteed Price Contracts; Price Caps; Price
Floors; Price Competition; Product Differentiation; Risk-based Pricing; Spot
Price; Value-added Services.

Abstract: This paper draws from observations of other deregulated industries to describe
the likely transition to competitive retail electricity markets. The paper
describes risk-differentiated products and lays out the principles for risk-based
pncmg. The paper also addresses the strategy of bundling value-added
services with retail electricity. Finally, the paper presents conclusions of how
successful energy service providers might be able to differentiate their
products and avoid the "commodity trap" of price wars and low profit margins.

1. INTRODUCTION

As competition unfolds in retail electricity markets, designing an optimal

line of energy products presents a unique set of profit-making challenges and
opportunities for energy service providers (ESPs). ESPs need to design lines
of attractive products that keep and draw customers while simultaneously
generating profits for their shareholders. It therefore becomes essential to
know how customers will accept and respond not only to these individual
product offerings but also to different combinations of product offerings.
In U.S. electricity markets, the "standard" retail product has heretofore
been a guaranteed-price product under which the customer buys units of
consumption (energy or demand) at a fixed price that is announced well in

A. Faruqui et al. (eds.), Pricing in Competitive Electricity Markets

© Kluwer Academic Publishers 2000
6 Pricing in Competitive Electricity Markets

advance and which applies to all units of consumption. Almost all retail
consumers have obtained service under such guaranteed pricing, which
meant they could use electricity without regard to the variation in underlying
marginal costs and economics. As discussed by Neufeld in Chapter 5, since
the inception of the electric business, utilities have indeed experimented with
several "innovative" rate forms that better reflect underlying marginal costs.
But these rate forms, such as time-of-use, real-time pricing, and
curtailable/interruptible service, have been treated as adjuncts to the basic
service. In addition, customers and regulators have viewed them as
somewhat risky in terms of acceptance.
Under competition, retail consumers of electricity have choices among
ESPs and among products. For ESPs, success hinges on the extent to which
customers are able to discern some desirable difference in one product over
another. When product distinctions are blurred or inconsequential to the
consumer, the "eye-catching" feature of the product becomes price. And
when ESPs compete on price alone, the result is price wars and concomitant
paper-thin profit margins.
Consequently, ESPs find that innovation in pricing and product design
becomes increasingly important in maintaining and increasing market share
and profitability. ESPs now offer a wide array of electricity products,
tailored to their customers' specific needs. New electricity products are
being developed using new pricing and product design practices that reflect
the realities of the new competitive environment. All retail electricity
products will be constructed, either explicitly or implicitly, as combinations
of a few basic building blocks. Today's so-called "innovative rates", based
on estimates of system marginal costs in the case of vertically integrated
utilities and on spot market prices in the case of de-integrated companies,
will evolve to become the fundamental products that ESPs offer. These
innovative products mayor may not necessarily pose significant risks for the
ESP, as the "riskiness" of the ESP's portfolio will depend on the
composition of the entire portfolio, including purchase obligations, sales
obligations, and hedging positions.
This paper is organized as follows. Section 2 draws from observations of
other deregulated industries to describe the likely transition to competitive
retail electricity markets. Section 3 describes risk-differentiated products
and lays out the principles for risk-based pricing. Section 4 addresses the
strategy of bundling value-added services with retail electricity. Finally,
Section 5 presents conclusions of how successful ESPs might be able to
differentiate their products and avoid the commodity trap.
Pricing in Competitive Electricity Markets 7

2. THE TRANSITION TO COMPETITION: THREE

NOT-SO-PERFECT DAYS

The last twenty-five years have seen several major industries go through
a process of deregulation. These industries include airlines, trucking,
railroads, banks, and telecommunication. i Each of these industries displays a
unique deregulatory process determined in large part by its unique regulatory
history. Now, as the electricity markets are on the verge of a competitive
transformation, what can ESPs learn from these other industries'
experiences?
The transformation of a heavily regulated industry to one based on
competition and market forces can be organized into three stages or "days"
of transition.

2.1 Day 1: Opening of Markets

Deregulation typically proceeds in an evolutionary manner. The

inefficiency and inflexibility embedded in the complex web of an industry's
regulatory history provide the driving forces for change. Eventually the
weight of regulation becomes too great for customers and the regulated
entity. Even though the process has been slow, the protective walls of
regulation appear to collapse rapidly and those in the industry perceive
revolutionary change. Day 1 occurs as the walls of regulation come down
and markets open up to competitive entry.
As markets open up, competitive entry occurs, intensive price
competition results and profit margins evaporate or even tum into losses.
Entry has at least two forms. Completely new players enter the industry.
Also, existing players expand, along geographic and product scope, into
markets previously foreclosed by regulation.
The initial objective of the new entrants is to get a toehold in the new
markets. Consequently, their focus tends to be on gaining market share
rather than maximizing profit. Entrants introduce themselves to the
customers by offering lower prices. Long-time incumbents respond (and
perhaps panic) by matching or beating the entrants' low prices. Price
competition becomes intense and escalates into a full-blown price war. The
product essentially becomes a commodity with price being the only
distinguishing characteristic.
Day 1 of competition also involves the resolution of regulatory holdover
issues, such as "stranded costs." These costs, most of which may have been
prudently incurred with regulatory approval, may not be recoverable with
market-based pricing. Consequently, Day 1 often involves transition charges
to recover stranded costs. Inclusion of these charges obfuscates competition
8 Pricing in Competitive Electricity Markets

and leads to non-transparency of prices. Such charges may represent a

substantial portion of customer bills. For example, in California, they
represent about 25 percent of a typical residential bill. Inclusion of transition
charges may in fact slow the beginning stages of competition and keep the
unstable walls of regulation upright a bit longer. However, the walls do
eventually crumble and the end result of Day 1 is intense price competition.

2.2 Day 2: Consolidation

The second day of competition begins to see the easing of the bloody
price wars of Day 1. Many of the new entrants do not survive Day 1. Also,
a few of the long-time providers are unable to adapt to market-driven
choices and become fatal victims of the price wars. Day 2 involves a
Darwinian struggle in which only the fittest companies survive, by
exploiting their unique core competencies and sometimes by engaging in
draconian cost-cutting measures. Consolidation (i.e., horizontal mergers)
becomes one of the main mechanisms to extend competencies, cut costs and
clear the Day 1 battlefield of its victims.
Day 2 profit margins are low compared to those achieved in the days of
regulation, but positive profits for the industry do emerge from the sea of
losses from Day 1. Occasional price wars still occur in Day 2, but these are
more in the nature of skirmishes, rather than campaigns designed to
annihilate the enemy. Competing firms begin to recognize that a pure
market share strategy is not sustainable. Survivors shift some emphasis back
to profitability. Also, as consolidation occurs, suppliers look for methods of
competition other than price. That is, with product differentiation, attempts
begin to escape the commodity trap.

2.3 Day 3: Stabilization

Day 3 sees further industry consolidation. Often just a handful of major

industry players remain, but entry barriers are low. The concentration of the
industry returns part way to the pre-deregulation levels. However, both
geographic and product scopes have expanded. The major players typically
expand to serve national markets. Niche players may survive serving
smaller markets and specialized product lines.
Price wars become increasingly infrequent. Prices do remain in check
because of potential entry by "discount firms," but the major players
compete mostly on non-price characteristics such as service, product quality,
bundling and brand-name recognition. With product and merchant
differentiation, direct price comparison becomes more difficult.
Consequently, price ceases to be the only focal point for customers.
Pricing in Competitive Electricity Markets 9

Day 3 returns profits to a competitive level. That is, the rate of return to
shareholders is, on average, commensurate with other financial investments
with similar risk. For deregulated industries, the expected rate of return is
significantly higher than the regulatory guaranteed rate of return of earlier
days. However, financial risks are now borne by shareholders.

3. RISK-BASED PRICING

In this section we describe the financial risks that are inherent in

competitive electricity markets. The risks, which are unmasked by
deregulation, actually provide the ESP with opportunities to create value by
developing a risk-differentiated line of products to serve customers with
diverse risk tolerances.
ESPs are exposed to many types of uncertainties or risks. These risks
include event risk, operational risk, credit risk, wholesale price risk, and
customer load risk. Event risk is associated with particular types of events
such as natural disasters and theft. Operational risk involves the uncertainty
in performance of equipment and operational systems. Credit risk concerns
the ability of customers to pay and vendors to deliver. Wholesale price risk
reflects the uncertainty in the market price of wholesale electricity, which is
the primary input required to satisfy retail electricity service contracts.
Customer load risk arises from uncertainties in customer demands.

3.1 What is Risk-Based Pricing?

Risk-based pricing is a market-based approach to pricing retail electricity

products that simultaneously embodies the risk costs borne by the ESP and
the risk aversions of the retail customers. Specifically, risk-based pricing
incorporates wholesale price risk and customer load risk into costing and
pricing. Event, operational and credit risk are not directly incorporated into
risk-based pricing.
Risk-based pricing involves four steps. The first step develops a line of
retail products that differs in how the financial risks are shared between the
ESP and the retail customer. The second step determines the market-based
cost to the ESP of providing each product. Third, the ESP assesses the
customer's or customer segment's willingness to pay for a particular
product. This willingness to pay will be determined in part by the degree of
risk aversion. Fourth, the ESP makes some judgements on competitors'
actions and reactions in response to the ESP's price and product decisions.
10 Pricing in Competitive Electricity Markets

One additional note is that risk-based pricing is largely distinct from

retail risk management. Risk-based pricing focuses on the expected cost of
providing a particular product and on the expected profitability of a complete
line of retail products. A particular product creates a cash-flow risk
exposure because of the uncertainty in wholesale price and customer load.
While each contract generates risk exposure, risk management generally
does not occur at the contract level or even the retail portfolio level. Instead,
a retail price manager undertaking risk-based pricing has an additional
responsibility to measure risk exposures and pass that information along to
the corporate risk manager. The risk manager can then manage all risks at
the corporate level. Chapter 10 by Henney and Keers covers the issue of
corporate risk management. Anticipating competitor response is the topic of
Chapter 6 by Seiden and Faruqui.

3.2 Building Risk-Differentiated Products

Competitive retail electricity services will be available under a

continuum of terms that offer the customer different levels of hedging
against price and quantity uncertainties. Although there is a large range of
specific product offerings that ESPs might offer, this continuum can be
viewed as combinations of four basic building blocks: spot pricing, forward
contracting, guaranteed price service, and financial instruments. Building
block combinations can be used to develop a portfolio of customized
electricity products.

3.2.1 Building Blocks

The spot price is the contemporary price established for immediate

purchase and delivery. The expected spot prices of electricity provide the
foundation for the market-based price of all retail electricity products. This
is true even though only a small fraction of retail customers are (or will
ultimately be) on spot price rates. Experience of other countries such as
Australia, New Zealand, and the United Kingdom indicates that only a
fraction of consumption (around 10 percent) will actually be purchased at
spot. Under spot pricing, retail customers either buy power at the prevailing
wholesale spot price plus a sales mark-up, or sell power at the wholesale
spot price less a mark-down. The merchandising margins (mark-ups and
mark -downs) cover the ESP's costs of arranging services, incl uding the
financial costs of the ESP's working capital.
Forward contracts provide for trades of specified quantities of power at
specified prices for a specified future period, generally under terms that are
tailored to the circumstances of the wholesale or retail parties to the
Pricing in Competitive Electricity Markets 11

contracts. Forward contracts differ from the electric power industry's

traditional retail tariffs in that the contracts are for specific quantities of
electricity. They more closely resemble a typical contract for purchasing a
commodity like wheat, com, or aluminum. However, a retail electricity
forward contract for a particular month load shape is itself a combination of
about 720 hourly forward contracts.
Guaranteed price service, also called flip-the switch (FTS) service,
allows consumers to buy flexible quantities of power at specified prices
under specified conditions. It is similar to a forward contract except that the
quantity of power available to the customer is not rigidly constrained as
under a standard forward contract. Traditional retail electricity tariffs have
been of this type, allowing consumers unlimited flexibility in choosing their
consumption levels. Guaranteed price contracts may, however, have either
unlimited or limited flexibility, depending upon the needs and circumstances
of the parties to the contract. Retail consumers will find guaranteed price
contracts attractive because they manage both price and quantity risks: with
guaranteed prices, consumers do not have to worry about exactly how much
power they will want in the future. ii
Financial instruments include options to buy or sell specified quantities
of electricity at a specified time. iii These options may be in the forms of
price caps, price collars, and certain forms of interruptible service. Financial
instruments may also include swaps, under which buyers and sellers "swap"
two different streams of payments for a specified quantity of electricity.
Examples of swaps are arrangements that tie the price paid by the customer
for a specified quantity of electricity to a (possibly non-electricity) price
index. Swaps may also cover the risks that arise from locational price
differences by swapping payment for power at one location for another
location.

3.2.2 Building Block Dimensions

The organization of wholesale electricity commodity markets will

determine the basic building blocks and the terms under which they are
traded. The terms of retail contracts, by contrast, will be set by ESPs'
standard tariffs, by individual-customer negotiation, and by competition.
And these terms will vary over time.
Retail contract terms will reflect market participants' forecasts of future
wholesale spot prices. The key dimensions of these contracts will be
contract length, price notification period and exclusivity.
Contract Length. Contracts will be offered for different terms, ranging
from days to years. In essence, a contract for a certain period of time - like a
year - is equivalent to 8,760 contracts for each of the hours of the year.
12 Pricing in Competitive Electricity Markets

Variations in contract length can therefore be regarded as equivalent to

variations in the building blocks that are combined into a single contractual
product.
Price Notification Period. Prices may be announced years in advance of
the periods to which they apply; or they may not be announced until after the
fact. Virtually no advance price notification exists for spot pricing, while
each of the other building blocks have contractually-specified notification
periods. Customers will, in the aggregate, be willing to pay more for products
that offer longer notification than those that offer shorter notification. Longer
notification allows customers time to plan changes in electricity production and
consumption in response to price. There are exceptions, however, including
consumers who can rapidly respond to price and thereby benefit from price
variations, and other consumers for whom electricity costs are relatively small
and who can therefore afford to gamble on spot prices being lower than prices
guaranteed in advance. For the ESP, longer notice exposes the ESP to more
wholesale price risk. This may be particularly serious if the ESP has
substantial guaranteed price contract obligations.
Exclusivity and No-Resale. When a contract is exclusive, the generator
or consumer may trade only with the ESP. A companion to exclusivity is a
no-resale restriction. Exclusivity and no-resale restrictions are essential for
guaranteed price contracts to be sustainable in the competitive market place.
Without the exclusivity restriction a customer could buy on the spot market
when the spot price was less than the guaranteed price and buy on the
guaranteed price contract when the spot price was above the guaranteed
price. This would leave the guaranteed price ESP selling only when it was
losing money on each kWh. Similarly, without a no-resale restriction, a
customer could, in principle, "go into business" during high spot price hours
and buy at the guaranteed price and sell at or below the spot price and make
a profit. Thus, without these two restrictions, an ESP could not offer a
profitable (or break-even) guaranteed price product. Only guaranteed price
type contracts need exclusivity and no-resale provisions.

3.2.3 Retail Electricity Products

Retail electricity products are constructed as combinations of the four

basic building blocks. Table 1 classifies the market-based products we
expect to see in competitive retail electricity markets. One set of products
falls into the category of guaranteed price products. Another category is
spot-priced products. Finally, there are customer risk management products
that help customize the retail offering. These three types of products are
analogous to interest rate products that are available to the homeowner:
Pricing in Competitive Electricity Markets 13

fixed-rate mortgages, adjustable-rate mortgages with no rate protection, and

adjustable-rate mortgages with an interest rate cap.

Guaranteed Price Products Spot Price Products

Flat Spot
TOU Occasional RTP

Seasonal Buy-back
Fixed Bill Dispatchable Interruptible

Customer Risk Management

Price Caps & Collars Weather Hedges
Table 1. A Menu of Retail Pricing Options

Guaranteed Price Products. Guaranteed pncmg products have two

distinguishing characteristics: 1) prices are specified in advance and hold for
a specified contract period; and 2) no quantity restrictions are imposed on
the customer. Because of these characteristics, these products are sometimes
called "flip-the-switch" (FrS) products. A flat rate product is both a
building block and a complete retail product. The product consists of a per
kWh energy charge that is constant for all hours across the entire contract
period (e.g., five cents/kWh).iv Flat rates appeal to customers for their
simplicity and appear to be popular in the long distance telecommunication
market.
A time-oj-use product (TOU) also has the distinguishing characteristics
of a guaranteed price product. However, in contrast to a flat rate, TOU
products have different prices for different time periods within a day or a
week. TOU products are simple building block combinations of guaranteed
price contracts for each time period. Such products typically require special
metering that tracks consumption by the hour. A simple TOU rate is
illustrated in Figure 1.
14 Pricing in Competitive Electricity Markets

Guaranteed Price Guaranteed Price Time-of-Use

+ =
7 a.m. - 7 p.m. 7 p.m. - 7 a.m. Rate

Guaranteed Price
Guaranteed Price Seasonal
for Summer
+ for Non-Summer = Rate
Months

Figure 1. Time-of-Use and Seasonal Pricing

A seasonal rate product is similar in structure to a TOU product. The

key difference is that the seasonal price periods correspond to different
calendar months rather than to different hours within a day or week. Figure
1 also illustrates a simple seasonal rate as a combination of two guaranteed
price contracts. Obviously the building block structures of TOU and
seasonal rates could be combined to yield a seasonal TOU rate. Seasonal
rates do not require special metering for implementation.
A fixed bill product is a product where the customer's bill is invariant of
usage. This is sometimes referred to as "all-you-can-eat pricing." The
product can be technically decomposed into a forward contract for expected
usage and a balancing contract with a zero marginal price. In its pure form,
the fixed bill completely eliminates the customer's bill uncertainty.
However, the customer still faces considerable quantity risk. The fixed bill
is based on the cost to serve the expected load with a significant premium
attached to protect the ESP. If actual usage turns out to be unusually low,
then the fixed bill looks quite costly to the customer. Additionally, the fixed
bill products for electricity and cellular phones typically have some usage
limits. Usage beyond the limit is billed at a pre-specified per unit price.
Thus, the typical "fixed bill" product is more aptly described as a take-or-
pay forward contract with a guaranteed price contract for additional usage.
Spot Price Products. Spot price products are products where the retail
customer sees a price that is directly related to the wholesale spot price.
Leading examples are real-time pricing products and many interruptible and
curtailable products.
Real time pricing (RTP), retail customers pay a marginal price equal to
the expected wholesale price plus some mark-up.v The mark-up typically
takes the form of a simple adder. A day-ahead forecast of the expected
wholesale price is used to set the RTP. Some RTP programs use hour-ahead
Pricing in Competitive Electricity Markets 15

forecasts and notice. Also, there are some programs that use day-ahead
forecasts, but charge ex post, based on actual system marginal costs or spot
prices.
One-part RTP rates charge hourly prices based on wholesale prices for
all electricity consumed. The spot price building block is the retail product.
In contrast, a two-part RTP product is a combination of a forward contract
and a balancing contract for incremental or decremental load priced at
wholesale spot with a mark-up or mark-down (i.e., a bid-ask spread).
Historically, the forward contract has been the customer's historic
consumption pattern priced at the traditional tariff. This resulted in
customer-specific revenue neutrality (on historic usage) and re-assured
regulators that the customer continued to make at least the same contribution
to overhead costs.
As markets move toward competition, two-part RTP products will
continue to be offered and will likely grow in popularity. However, the
forward contract will be negotiated between the ESP and the customer, or
ultimately be determined strictly by market forces.
Interruptible and curtailable (1IC) products provide the ESP with the
right to stop or restrict service if certain wholesale market or power system
conditions prevail. In essence, the lie customers are selling call options to
the ESP. When the options are exercised, the retail lie customers become
wholesale suppliers of electricity. The payments received by the lie
customers are in the form of up-front credits, credits for actual interruption
or curtailment, and reduced prices for power actually taken. Under
competition, the bill savings and credits for accepting interruptions and
curtailments will reflect the market value of the energy procured by
exercising the lie options.
Risk Management Products. The third classification of retail energy
products falls into the category of customer risk management products.
These "add-on" features can customize retail products to bring the retail
customer into his or her comfort zone with respect to risk. This category
includes price caps, floors and collars, swaps, and weather hedges.
A price cap is a maximum price that can be charged. A cap is obtained
by the ESP selling a call option to the retail customer. The call option gives
the customer the right, but not the obligation, to buy electricity at a pre-
specified price (the "exercise price"). A price cap limits the retail
customer's exposure to high retail spot prices.
A price floor is a minimum price that can be paid. The customer selling
a put option to the ESP creates a price floor. The put option gives the ESP
the right, but not the obligation, to sell electricity at a pre-specified price. A
price floor limits the electricity ESP's exposure to low retail prices.
16 Pricing in Competitive Electricity Markets

A price collar is a combination of a price cap and price floor. If the spot
price is within the collar, then the customer buys power at spot. If the spot
price is outside the collar, then the customer buys power at the relevant
- • vi
exercIse pnce.
As described above, swaps are arrangements under which buyers and
sellers "swap" two different streams of payments. An example would be an
aluminum manufacturer who has his purchase price for electricity directly
indexed to the world price for aluminum, or inversely indexed to the world
price of bauxite (the key input), or some combination of both.
Weather hedges are financial contracts that supplement guaranteed price
contracts. A weather hedge reduces the risk in the customer's bill
attributable to weather uncertainties. The weather-related bill uncertainty
results from the correlation of weather with customer demand. In essence,
the ESP is selling the retail customer a guaranteed price contract with a
weather insurance policy. MacArthur, in Chapter 11, describes one such
weather hedge product.
Figure 2 summarizes the discussion of constructing risk-differentiated
products.

o
Risk to Customer

Figure 2. A Continuum of Risk-Differentiated Products

At one extreme of the continuum is the pure spot product (one-part RTP).
With this product, all of the wholesale price risk is passed through to the
retail customer. At the other extreme is the flat rate product. With this
product the ESP completely bears both the wholesale price risk and the
customer load risk. In between are building block combination products.
Pricing in Competitive Electricity Markets 17

By changing the dimensions of the building blocks and by adding financial

options, the amount of risk sharing can be continuously shifted.

3.3 Costing Risk-Differentiated Products: Calculating

Break-even Prices

Having determined how to construct risk-differentiated products from

building blocks, it is now crucial to figure out the costs. This is
accomplished by calculating the break-even price for each building block
and then adding these together to obtain a market-based version of the retail
product cost.
The building block products, other than the spot pricing contract, are
derivative products of spot electricity. Thus, the building block break-even
prices are functions of the expected spot prices, that is, derivative prices.
The complete list of break-even price determinants is the hourly wholesale
forward price curve, the coincidence of hourly customer load and hourly
wholesale price shapes, wholesale price volatility, customer load volatility,
and the correlation between the uncertainty in wholesale price and the
uncertainty in customer load. Thus, break-even prices are typically
customer-specific even though the underlying asset price, the expected spot
price of electricity, is a market variable.

3.3.1 Hourly Forward Prices

Wholesale electricity price forecasting is the topic of Chapter 15 by

McMenamin and Monforte. The hourly forward price curve is one
embodiment of expectations about future spot prices. The forward price
curve depicts how today's prices for forward contracts depend upon the
delivery date. The formal relationship between expected spot price for time
t, E{SI}' and today's forward price for time t, Fo,I' is:

(1) F =E{S }e(rr-r,)1

0,1 I

where e is the natural number (2.71828), rr is the risk-free discount rate and
ra is a risk-adjusted discount rate. vii
The expected spot price and the forward price are equivalent expressions
of value. Because the risk-free discount rate is both objective and
observable, using the forward curve to embody price expectations simplifies
the present-value discounting problem.
18 Pricing in Competitive Electricity Markets

3.3.2 Retail Forward Price

A retail forward contract is a contract with a single per unit price agreed
upon today for specific hourly quantities delivered in future time periods.
The retail forward break-even price depends upon the degree of coincidence
between the forward load shape and the wholesale price shape. Specifically,
the retail forward break-even price is a load-weighted average of the hourly
forward prices. That is:

(2)

where RF is the break-even price, Fo,t is the hourly forward price, and L t is
hourly load. If the customer's load shape is coincident with the expected
hourly price shape, then the per unit retail forward break-even price is
greater than the average hourly forward price. If the customer's load is non-
coincident, then the break-even price is less than the average hourly price.

3.3.3 Guaranteed Price Product

The guaranteed price product has a pre-specified price (or prices) that
holds for the contract period. However, no quantity restrictions or
obligations are imposed on the customer. As a result, the break-even price
for this product consists of an expected component and an uncertainty
component. Specifically,
(
Pcrp(ht I
(3) Pfts =RF(lIT) of T e Jdt

where T is the length of the contract, t is a time indicator, O"p is the volatility
in wholesale price, O"L is the volatility in customer load, and p is the
correlation between the uncertainty in wholesale price and customer load. viii
The first factor of equation (3), RF, is from equation (2) and captures the
cost of serving the expected load shape. The second factor (called the FTS
PcrpcrLt
effect), e , captures the fact that the ESP faces uncertainty in
wholesale prices and uncertainty in customer load and these uncertainties
may be correlated. The integration and the scaling by liT is merely an
averaging procedure to reflect the condition that a single per kWh price is
going to be charged for the entire contract period [O,T].
Pricing in Competitive Electricity Markets 19

Examination of equation (3) confirms that if price and load uncertainty

are positively correlated, then the FrS effect inflates the retail forward
break-even price. If the correlation is negative, then the break-even price for
the guaranteed price product is less than the retail forward price for the
expected load shape. If there is no correlation, then the break-even price is
the same as the retail forward break-even price for the expected load shape.
The FrS effect is an actuarially fair risk premium and does not have any
connection to the risk aversion of either the customer or the ESP.
Even if there were no correlation between the customer's load and the
wholesale price, equation (2) and equation (3) differ. Charging a price given
by equation (2) on a defined load guarantees a contract with a value (today)
of $0. Equation (3), on the other hand, is the break-even price for an
uncertain load. Charging a price given by equation (3) yields a zero-mean
probability distribution (today) of expected profit. If tomorrow's forward
price curve differs from today's, then the contracts based on today's break-
even price change in value, regardless of whether the contract load is certain
or uncertain.

3.3.4 Price Caps

A price cap is a call option sold by the ESP to the customer. Under a
certain set of assumptions, the value (i.e., cost or break-even price) of a price
cap (or floor) has an explicit analytical expression. ix
The per kWh break-even price for a price cap in hour t on a defined load
is:
(4) Defined Load Price Cap Cost I

where X is the price cap level (also called the exercise price), N is the
cumulative normal distribution operator, and

In( Fo,1 / X) + V2 (J'p 2 t

Again, O"p is the volatility in wholesale price.

20 Pricing in Competitive Electricity Markets

If the price cap protection applies to all hours in the contract period or
to any subset of hours, then the cost of that price cap is the sum of the break-
even prices for the individual hourly price caps. For a price cap that applies
to a defined load shape in all hours over the period [O,T], the per kWh break-
. • x
even pnce IS:

(5) Defined Load Price Cap Cost

In(RFIX) + ~ 0"/ t

The break-even price for a price cap that applies to an undefined load
has a similar equation, but it is necessary to account for the FTS effect. The
per expected kWh cost of price cap protection on an undefined load xi is:

(6) Undefined Load Price Cap Cost

In( PjiJ X) + ~ 0"/ t

(6a) dl =

Note that equation (6) is the same as equation (5) except for the replacement
of RF with P fts . It is through Pfts that load uncertainty and the correlation of
price and load uncertainty are introduced into the break-even price formula
for a price cap. It is also worth noting that price uncertainty, O"p, now has a
more complex role because it enters into the pricing formula via P fts in (6),
(6a) and (6b) as well as directly in d 1 and d 2 .

3.3.5 Price Floors

A price floor is a put option sold by the customer to the ESP. The per
kWh break-even price for a price floor that applies to a defined load shape in
all hours in the period [O,TJ is:
Pricing in Competitive Electricity Markets 21

(7) Defined Price Floor Cost

In( RFIX) + V2 a/ t
(7a) d] =

The per expected kWh break-even price formula for an undefined load
price floor xii is:

(8) Defined Price Floor Cost

(8a) d] =

Again, the flip-the-switch effect enters the break-even pricing formula

through Pfts .

3.3.6 Swaps and Other Financial Instruments

Determining the break-even prices for other financial instruments

involves modeling the other price process and correlating that process with
the wholesale electricity price process and possibly the customer load
process. Pricing a weather hedge requires a model of the weather process
and correlating that with wholesale price and customer load. Statistical
approaches to estimate the relationship of weather and wholesale price is the
topic of Chapter 12 by Martello.
22 Pricing in Competitive Electricity Markets

3.3.7 Conclusions on Break-even Prices

The break-even prices given by the equations in this section are prices that
would yield the ESP zero expected profit. Break-even prices do not (and
should not) contain any element of overhead, fixed, or sunk cost. Break-
even prices generally are not the prices that should be charged, but instead
provide the supply-side foundation from which prices can be marked up
according to demand considerations. In short, break-even prices are the
market-based version of marginal costs, from which retail prices can be
marked up to generate profit or margin to help cover overhead costs.

3.4 Pricing Risk-Differentiated Products

The break-even prices just discussed provide the foundation for
profitable pricing. The ability of the ESP to charge a retail price in excess of
the break-even price depends upon three interacting concepts. First, the
ability to price above break-even depends upon the customers' price
sensitivity or the price elasticity of demand for electricity. Second, pricing
strategy depends upon the choices the customer segment has. Third, the
degree of customer risk aversion also contributes to the margin that may be
attainable.
The price elasticity of demand is a measure of how customer purchases
change in response to a change in retail price. xiii If the ESP were the only
provider, then the profit-maximizing mark-up over the break-even price
would be inversely related to the price elasticity of demand E. Specifically,
the profit maximizing price would be set such that:

(9) (Retail Price - Break-even Price) / Retail Price = liE.

However, the price elasticity of demand may suggest greater pricing latitude
than a profit seeking ESP is likely to have in practice. This is because the
customer may have choices either among ESPs or a choice among several
products offered by a single ESP, or both. With choice, a customer with
relatively inelastic demand for the generic electricity may nevertheless have
very elastic demand for the electricity product offered by a particular ESP.
Two influences in the choice process are worth noting. First, customers
may have some difficulty in making accurate distinctions in comparing
alternatives. Second, customers may have some biases for particular
providers, or for particular products, or for just staying with the current ESP
and product. These two influences, evaluation uncertainty and provider and
Pricing in Competitive Electricity Markets 23

product biases, may result in an ESP having some pricing power over its
electricity product, despite the presence of competition.
Risk aversion may also influence the customers' willingness to pay above
the break-even price. Strongly risk-averse customers may be willing to pay
a premium well in excess of the FrS effect on break-even price. Other
customers may be risk neutral and therefore unwilling to pay even the
actuarially-fair FrS premium if a market-based spot price alternative is
available. Differences among customers in their risk aversion provide a
basis for an ESP to profit from offering a line of risk-differentiated energy
products.
Some version of equation (9) (above) appears in most microeconomic
textbooks. The equation looks straightforward, but implementation of a
pricing strategy based on is actually quite involved. Real-world
complications to implementation include the fact that customers may have a
choice of ESPs and products and that customers may have biases and risk
aversions that affect how they make their choices. Also complicating
implementation is the calculation of the break-even prices.

3.4.1 EPRI's Product Mix Model

The ESP pricing specialist needs to process and integrate a considerable

amount of information. This includes information about wholesale prices,
customer load characteristics, price responsiveness, choice behavior, the
product choices facing each customer segments, and what competitors are
doing and are likely to do in response to the ESP's price and product
decisions. This presents a formidable, but necessary task for the pricing
specialist. To assist in this task, EPRI has developed the Product Mix
Model, an analytical tool for designing and pricing retail energy products.
The Product Mix Model is a state-of-the-art tool that allows the user to
build, cost and price the building block products described above. The
Product Mix Model brings together modem finance theory in its
representation of wholesale price and customer load uncertainties, neo-
classical microeconomics in its price response module, and market research
methods in its customer choice module.

3.4.2 Customer Segmentation

A successful segmentation strategy needs to be based on meaningful

differences among customer groups. The meaningful differences involve the
24 Pricing in Competitive Electricity Markets

variables contained in equation (9) (above). Specifically, customers may

differ in their marginal cost to serve (i.e., break-even price), in their price
responsiveness, in their set of competitive alternatives, in their risk aversion
and in their choice behavior. Segmenting customers and developing and
targeting products based on these criteria can simultaneously enhance profits
and increase customer satisfaction. Segmentation on non-meaningful
differences, such as right-handedness and left-handedness, does not increase
profits and may actually lead to simultaneously lower profit and increased
customer dissatisfaction.

3.5 Creating Value by Sharing Risk

Evidence suggests significant diversity among customers in the implied

demand for risk management products. The diversity arises from at least
three sources. First, customers may have differences in their capability to
alter electricity usage in response to changing retail prices. Those with less
capability to respond will be more inclined toward a less risky product.
Second, customers may have differences in their risk tolerance levels. More
risk-averse customers have a greater demand for a less risky product. Third,
risk-averse customers may have different access to hedging instruments. A
multinational multi-product corporation that is active in world financial
markets may want only commodity energy because it is undertaking all its
risk management activities in the financial markets. On the other hand, a
small, yet energy intensive, manufacturer may want electricity price risk
management bundled with the commodity electricity.
The diversity in customers' demands for risk management products
creates opportunities for the ESP. We demonstrate this in the following
stylized example developed using EPRI's Product Mix Model. The base
case for the example is an ESP earning $10 million profit by serving three
equal-sized customer segments with a single FTS product.
The three seemingly identical customer segments are actually quite
different. One segment is "inflexible and risk averse." This segment lacks
the ability to alter usage in response to price changes. Furthermore, this
segment does not like financial uncertainty. A second segment is "flexible
yet risk averse." This segment does have significant capability to change
usage in response to price changes. This flexibility includes the ability to
shift usage across hours in response to inter-hour price differences. This
segment has the same degree of risk aversion as the first segment. The third
segment has the same flexibility as the second segment has to alter and shift
Pricing in Competitive Electricity Markets 25

usage in response to price changes. This segment however is not bothered

by financial uncertainty. Thus, this segment is "flexible and risk neutral."
The ESP can create value by introducing new risk-sharing products.
Introducing new products requires careful attention to their pricing. Pricing
too high results in no-takers and a wasted marketing effort. Even worse,
pricing too low results in lost margin from existing product lines. A tool like
Product Mix Model can help in designing and pricing new products to
enhance profits and in avoiding cannibalizing margin.
In this example, the ESP can enhance profit by 8 percent if it introduces
and judiciously prices a spot product and a spot product with a price cap.
This gain in profit is realized while offering customers a choice of products.
Table 2 summarizes the results of this analysis.

Segment Inflexible Flexible Flexible ESP Profit

Risk Averse Risk Averse Risk Neutral (millions)
Product
FTS 99% 36% 19% $ 5.1
Spot 0% 32% 62% $ 3.3
Spot + Cap 1% 32% 19% $ 2.4
TOTAL 100% 100% 100% $10.8

Table 2. Customer Choice and ESP Profit

When offered the choice of three products, the customers self-select the
product that best meets their needs. The inflexible and risk-averse customers
stick with the FTS product. The flexible and risk-neutral customers switch
overwhelmingly to the spot products. The flexible but risk-averse customers
see gains from expanding in low-priced hours and contracting in high-priced
hours. However, these same customers are made uneasy by the price roller
coaster. This segment is tom apart by the choice, with the spot + cap being
the most popular choice.
In this example, having the flexible customers and the risk neutral
customers take some of the price risk has created value for both the
customers and the ESP. However, all customers present the ESP with
market opportunities. The break even price formulas contain a measure of
the risk costs faced by the ESP. If a customer is willing to pay more to avoid
risk than the risk cost to the ESP, then profits are increased by the ESP
taking on the risk and charging the customer appropriately for the bundle of
electricity and risk management.
If, on the other hand, the customer is not willing to pay the actuarially
fair risk cost, then the ESP can profit by passing some price risk on to this
26 Pricing in Competitive Electricity Markets

customer. The required discount in retail price is less than the avoided (per
unit) risk cost.
This section has laid out the process for constructing, costing and pricing
risk-differentiated products. Offering a risk-differentiated menu or retail
products, priced according to risk-based pricing principles, results in the
win-win outcome of increased profit and increased customer satisfaction. As
competitive markets for retail electricity mature, ESPs will offer a more
diversified set of retail products. This in tum will contribute to Day 3
stabilization for the industry.xiv

4. BUNDLING VALUE-ADDED SERVICES WITH

COMMODITY ELECTRICITY

ESPs continue to lose money in the United States, two years after the
introduction of competition in California, the New England states, and
Pennsylvania. According to publicly available documents, these losses
collectively exceed $500 million annually. Two of the largest firms in this
market, ENRON and PG&E Energy Services, are respectively losing about a
$100 million a year and about $50 million a year.
In the English market, ten years after the introduction of competition,
margins in the retail energy supply business are razor thin. Profits average
less than l.5 percent of sales. In Australia and New Zealand, about five
years after competition, similar percentages exist.
How then does one get beyond the commodity trap? One approach
discussed in the previous section is to create a portfolio of risk-differentiated
products to satisfy diverse customer needs. Another approach that holds
considerable promise involves the introduction of new products and services
that would expand the business horizons of ESPs. This approach is the topic
of this section.
Restructuring of regulated industries has often led to an explosion of
novel products and services for consumers. Telephone deregulation resulted
in enhanced services such as call waiting and call forwarding that created
entirely new markets and positioned Telecom suppliers as more than just the
company that delivers dial tone. In banking and financial services,
deregulation led to product differentiation that gave us on-line banking,
discount brokerage, and automated teller machines. The entry of new
competitors, and the reduction of constraints on incumbents, acts to spur
innovation in product and service design.
In each case, new or novel use of technology together with a deep
understanding of unarticulated customer needs led to breakthrough products
that created entirely new lines of business bringing significant competitive
Pricing in Competitive Electricity Markets 27

advantages for those who created them and substantially enriching their
relationships with their customers.
In the electricity industry, suppliers are conceiving new product designs
that "bundle" the core commodity energy product with a wide range of
"value-added" services. The core commodity product is either electricity by
itself, or electricity and natural gas. Value-added services range from items
that are close to the functions one typically associates with energy supply
companies to those that are progressively far afield. Examples of the former
include enhanced power quality services, energy efficiency services,
engineering services to ensure compliance with environmental regulations.
Examples of the latter include facilities management, productivity audits,
Internet access, distance learning, and cable TV. There are several examples
that lie in the middle of these extremes. One of the most popular ones is risk
management services. Through these services, the ESP helps to mitigate
price volatility for some customers and to provide lower energy costs for
others. These services have been discussed earlier in this chapter.

Typical examples of bundled products include the following:

• Commercial segment: Bundling real time pricing of electricity with a
device for controlling HV AC systems such as the EPRI RTP controller
and time-flexible end use technologies such as Thermal Energy Storage.
• Industrial segment: Bundling off-peak pricing of electricity with high-
efficiency motors for gas pipeline compression to make it competitive
with gas engines.
• Residential segment: Bundling information technologies and energy
efficiency services with sale of electricity on a fixed-bill basis.

The creation of these bundles relies on imaginative integration of the

energy commodity with value-added services, with a view to creating a
unique position in the energy market place. Three dimensions are
simultaneously brought to bear in the field, with one dimension being
various ways to price energy, another being various value-added services,
and the third being the firm's unique image. For example, some firms
emphasize their knowledge of local conditions because they are the
incumbent providers; others project a "green" image to appeal to the
environmentally sensitive customers, while still others project an image of
being a nationally well-known and financially stable company. The new
"market space" created by these three dimensions is portrayed graphically in
Figure 3.
28 Pricing in Competitive Electricity Markets

VALUE·ADDED SERVICES

Power Quality Service

Environmental Compliance

Facility Management

Energy Efficient Services

SUPPLIER OPTIONS

Figure 3. The Retail Product Space

To see the importance of bundling core energy with value-added

services, we ran a simulation with EPRl's Product Mix model. We assumed
that initially the incumbent company is making $30 million in profit selling
the core energy product. By definition, this company is a regulated
monopoly, and owns 100 percent of the market. Then we introduce
competition into the market, and assume that the challenger firm proceeds to
lure away the incumbent's customers by undercutting its price by 10 percent.
The model then tells us that such behavior will permit the challenger to take
away 40 percent of the business from the incumbent, and reduce the
incumbents' profits to $18 million.
In response, the incumbent can execute four strategies. xv These are:

• Match the lower price offered by the challenger

• Introduce a new core energy commodity product, (i.e., engage in risk-
management services)
• Introduce another value-added service
• Offer both a new risk-management service and a non-risk value added
service
Pricing in Competitive Electricity Markets 29

Table 3 shows the results of these various moves.

Response Strategy Profit Market

(millions) Share
Do Nothing $18 60%
Match Price Cuts $20 66%
Introduce a Risk-Management Product (RMP) $21 69%
Introduce Value-Added Services (VAS) $22 72%
Introduce both RMP and VAS $23 75%
Table 3. Profit and Market Share Impacts of Response Strategies

In this simulation exercise, the best strategy is the fourth one.

5. CONCLUSIONS

The introduction of competition in the heretofore-regulated monopoly

business of electric utilities will initially result in significant profit-erosion as
challenger firms engage in price wars with incumbent suppliers and their
unregulated affiliates. As in other industries that have made a transition
toward deregulation, such price wars will destroy shareholder value and
result in large-scale mergers and acquisitions between the existing and new
players. Eventually, the surviving firms will invent new ways of selling and
pricing electricity that allow them to get outside the commodity trap.
Some of these innovations will involve the provision of risk-management
services designed to take the risk out of buying an inherently volatile
product: spot priced electricity. The design of these services involves the
incorporation of a number of concepts alien to historical cost-of-service
pncmg. Some examples include the customer's willingness to pay for
various product attributes, price elasticities of demand, risk tolerance,
correlations between loads and prices, and load and price volatilities.
Other ways in which the new players will seek to boost earnings will
involve the bundling of non-electric attributes with the core energy product.
Such bundling activities will result in the creation of a new "market space"
that gives electric customers access to a wide range of "value added"
services such as enhanced power quality, increased energy efficiency,
management of facilities, environmental compliance, and Internet access.
30 Pricing in Competitive Electricity Markets

REFERENCES

i For reviews of the effects of deregulation in other industries, see C. Winston, Economic
Deregulation: Days of Reckoning for Microeconomists," Journal of Economic Literature,
Volume 31, Number 3 (September 1993), and R. Crandall and J. Ellig, "Electric
Restructuring and Consumer Interests: Lessons from Other Industries," in Customer
Choice: Finding Value in Retail Electricity Markets, (A. Faruqui and R. Maiko, editors),
Public Utility Reports, Inc., Vienna, Virginia, 1999.
ii For further discussion of this traditional standard product, see S.D. Braithwait, D.W. Caves,

K. King, and L.D. Kirsch, Pricing the Riskiest Retail Electricity Product - Flip the Switch,
TB-I06922, EPRI, Palo Alto, California, 1996.
iii More complete descriptions of financial options and techniques for valuation of these

derivatives can be found in J. Hull, Options, Futures and Other Derivative Securities,
Prentice-Hall, Englewood Cliffs, New Jersey, 1997.
IV There may also be customer and demand charges. Customer charges are non-volumetric
and merely increase the customer's average cost per kWh but not the marginal price of a
kWh. Demand charges, on the other hand, can distort the perceived marginal price in
hours with demands near the peak. Thus a "flat rate" with a demand charge may not be
perceived as a flat rate at all.
v An in-depth discussion of real-time pricing, both one-part and two-part, can be found in Real

Time Pricing QuickStart Guide, TR-105045, EPRI, Palo Alto, California, 1995. Also, M.
0' Sheasy describes the Georgia Power two-part RTP rates in Chapter 17 and S. Huso
provides further discussion of one-part and two-part RTP rates in Chapter 18.
vi A combination of a price cap and a price floor applied to the same quantity and with the
same exercise price creates a forward contract. Thus, technically speaking the forward
contract is a combination product and the only fundamental building blocks are the spot
price contract, the guaranteed price contract and the financial instruments.
vii Equation I is obtained by using the present value equivalence of the expected spot price
discounted with a risk-adjusted discount rate and the forward price discounted at the risk-
free discount rate.
viii The guaranteed priee break-even equation given by (3) assumes that processes displaying
lognormal distributions characterize both wholesale price uncertainty and customer load
uncertainty.
ix The path-breaking work on options pricing can be found in F. Black and M. Scholes, "The
Pricing of Options and Corporate Liabilities," Journal of Political Economy, 81 (May-June
1973), 637-59 and in F. Black's, "The Pricing of Commodity Contracts," Journal of
Financial Economics, 3 (March 1976), 167-79. The key underlying assumptions are that
prices (and loads) are log-normally distributed and that financial markets are efficient so
that no arbitrage opportunities persist. These are the same assumptions underlying
equations (I), (3), (4), (5), (6), (7) and (8) in our paper. A more complete discussion of
options pricing and a critique of the underlying Black-Scholes assumptions can be found
in J. Hull, Options, Futures and Other Derivative Securities, Prentice-Hall, Englewood
Cliffs, New Jersey, 1997.
x The formulas given by equations (5) and (7) further assume that the price volatility is the

same for all hours in the period [O,T].

xi An undefined load price cap must be accompanied by a no-resale condition.
Pricing in Competitive Electricity Markets 31

xii A break-even price of an undefined load price floor is related to the expected load of the
customer. It does not give the ESP the right to sell an unlimited quantity at the price floor
level. Instead, an undefined load price floor applies to the customer's entire load,
whatever that turns out to be. An undefined load price floor must have an exclusivity
requirement.
xiii Fonnally, the price elasticity of demand E =-(percentage change in quantity
purchased)/(percentage change in price). We choose the convention of reporting the price
elasticity as a non-negative number. Estimation of price response measures for retail
electricity customers is the topic of Chapter 16 by R. Patrick and F. Wolak.
xiv Market simulation indicates that a small percentage of being sold at spot-based prices
would significantly dampen wholesale price volatility. For more details, see D. Caves, K.
Eakin and A. Faruqui, "Block That Spike," EPRI Inside Pricing Series, Volume 2, Issue I,
Palo Alto, Cali fornia, 1999.
xv In the real world, each of its strategies would invite a corresponding competitive response.
In this simulation, we have "turned off' such a response. If one wanted to pursue this
analysis further, one could deploy the full complement of tools afforded by game theory.
For a general introduction to this subject, see John McDonald, Strategy in Poker, Business,
and War, W.W. Norton, 1996.
SECTIONU

INDUSTRY RESTRUCTURING AND ITS

PRICING IMPLICATIONS
Chapter 2

Pricing and Revenue Management

Robert G. Cross
Talus Solutions, Inc.

Over the next decade, there will be a complete transformation of the

electric utility industry. Monopolies, which had been regulated for over a
century, will be open to competition. While there are valid arguments for
and against this transformation, its occurrence is part of an immutable trend
of replacing centrally controlled economic systems with still imperfect, but
more efficient decentralized free-market systems.
These transformations have already taken place in the airline, trucking,
gas, and telecommunications industries. They resulted in radical alterations
in the economic models, which drove the industries prior to deregulation.
For example, in the twenty years prior to airline deregulation, there were no
new airline start-ups and no airline failures. In the twenty years after
deregulation, over 200 airlines began service, and over 160 airlines failed or
were merged out of existence. Capacity was up 167 percent while traffic
was up almost 240 percent. This resulted in a much higher utilization of
capacity as airline load factors rose from 55.4 percent to over 70 percent.
The average price of an airline ticket fell in real dollars by over 42 percent.
Intense competition and changing consumer preferences resulted in
increased fare volatility as airlines constantly chased higher fares when they
could and lower fares when they had to. Today, an average of 50,000 fares
change daily, and the average airline fare lasts only ten days. Still, despite
this apparent chaos, airlines are more profitable than ever.
Similar outcomes have been observed in the other industries that have
gone through deregulation. This is a consequence of the shift from the
relatively stable regulatory environment to the disorder of the marketplace.
As such, we can expect the same from the deregulation of electricity. And
we should learn from the previous experiences.

A. Faruqui et al. (eds.), Pricing in Competitive Electricity Markets

© Kluwer Academic Publishers 2000
36 Pricing in Competitive Electricity Markets

Because of the inherent limitations on cost-cutting and productivity

increases, managing the shift in pricing power from regulators to consumers
has proven to be the most important factor in a company's ability to
transition from a regulated economy to a competitive one. The most
successful companies have employed the new science of Pricing and
Revenue Management to profitably manage this transition.
This new science has been driven by market needs resulting from more
intense competition, and it has been facilitated by the tremendous growth in
information technology. Pricing and Revenue Management techniques
segment customers by buying characteristics, predict consumer behavior at
the micro-market level, optimize product availability, and set prices to
maximize revenue growth. Ultimately, it involves selling the right product
to the right customer at the right time for the right price.
The paradigm of Pricing and Revenue Management reverses the familiar
business equation, where costs are viewed as the primary driver of
profitability. This new paradigm contends that revenue is the key driver of
profitability as well as growth. It recognizes that there is a limit to how
much you can drive down costs, but no limit to how much you can grow
revenues.
This new concept originated in the deregulated airline industry as "Yield
Management." Robert Crandall, Chairman of American Airlines, estimates
that this concept generates over $500 million in incremental revenue for
American. This equates to creating an additional 4 to 5 percent in revenue
growth from its existing capacity. Fundamentally, advanced Pricing and
Revenue Management techniques utilize vast volumes of customer behavior
data to determine where the company can charge more based on customers'
perceived value and where they can discount with discretion to increase
market share. This concept has proven its application to many industries that
have a high proportion of fixed costs in their business operations. Besides
airlines, other industries that have successfully made use of this concept
include shipping, hotel, and car rentals. The application of this concept is
currently being explored for companies in the electricity business.
There are several reasons for the emergence of Pricing and Revenue
Management as a core competence for those in electricity markets. These
include the spread of deregulation across all boundaries (geographical and
political), the emergence of substitutable products and services, advances in
information technology, and the need for timely information about future
developments.
There are several core concepts of Pricing and Revenue Management that
apply directly to the sale of electricity. The first is using price, not costs, to
balance supply and demand. Currently, seasonal and peak prices are
infrequently used to lessen demand at peak times or to incentivize movable
Pricing in Competitive Electricity Markets 37

demand to shift off-peak. These core concepts have been used very
successfully in other industries to level loads while increasing utilization and
revenues. Could a consumer agree to heat water only at night for a reduced
rate?
Another core concept of Pricing and Revenue Management is the
targeting of sales to micro-markets, not mass markets. This requires
understanding customer values by market segment, and pricing your product
according to the differential values placed by the different segments. This is
part of a trend to one-to-one marketing or "mass customization", as
consumers require solutions tailored to their personal perceptions of value,
convenience, quality, and risk. This allows a firm to significantly raise its
revenues and profits from a wide variety of market segments. Can a
manufacturer with a flexible operation agree to move its utilization on short
notice for a rate cut?
Some of the greatest power derived from the new science of Pricing and
Revenue Management involves a meticulous analysis of customer data to
predict future customer behavior. Armed with this analysis, a company can
far better respond to a diverse marketplace with optimal products and prices.
The process of Pricing and Revenue Management is a continuous one,
where market response is immediately monitored and measured. This
response is used to dynamically update forecast market demands under a
variety of scenarios of customer behavior. Capacity planning can then be
optimized and recommendations for price and availability of resources can
be created in a real-time environment.
In his business bestseller, The Post-Capitalist Society, Dr. Peter Drucker
reveals that the world's economy is transitioning away from a capital-driven
model to a knowledge driven one. Knowledge, not capital, will be the basis
for the creation of wealth. The electric power industry will be following this
trend. More important than tremendous amounts of capacity will be market
knowledge of which segments are willing to pay how much at what time.
This is the realm of Pricing and Revenue Management. The firms that
master this new paradigm will be able to get all the capacity they want, when
they want it. They will be the creators of wealth and the successors to the
old guard.
Chapter 3

The Role Of Price In The Restructured Electricity

Market

Leonard Hyman
Salomon Smith Barney Inc.

Key words: Competition; Price Signals; Restructured Markets.

Abstract: The price of the products in a regulated market serves a subordinate function,
as long as regulators find that total revenue will cover all costs and return
earned will attract needed capital. Pricing in a competitive market, however,
serves the function of bringing about the efficient allocation of economic
resources. The mixed competitive-regulated pricing scheme contemplated for
the restructured electricity market may neither efficiently allocate resources
nor attract needed capital to certain segments of the market. Furthermore,
regulators who do not appreciate the value of market signals and system
operators who downplay the commercial consequences of reliability decisions
may demand a command-and-control overlay that could inhibit development
of an economically efficient market.

1. DEFINING THE PROBLEM

What role did price play in the regulated electricity industry? That is an
easy question. Price was a residual, as shown in Table l.
Bonbright, in his classic text about public utilities, allocated five out of
406 pages to the role of price, mostly to discuss the role of price in the
competitive market, which is "that of controlling the distribution of scarce
resources among multiple and competing uses." j

A. Faruqui et al. (eds.), Pricing in Competitive Electricity Markets

© Kluwer Academic Publishers 2000
40 Pricing in Competitive Electricity Markets

Step I Determine cost of doing business excluding cost of capital

Step 2 Determine rate base

Step 3 Set appropriate return on rate base

Step 4 Determine cost of capital by multiplying rate of return by rate

base

Step 5 Add cost of doing business excluding cost of capital plus cost
of capital to determine overall cost of service

Step 6 Divide overall cost of service by expected sales units to

determine price per unit

Table 1. Calculation of Price in the Regulated Framework

Why not accept that definition as the role of pricing in the new electricity
market, and move on? That, too, is an easy question. In the new industry
structure, roughly one-third to one-half of the business retains a regulated
pncmg regime. Many of the regulated and unregulated functions can
substitute for each other. That, in itself, should lead to inefficient allocation
of resources. As an example, distributed resources (unregulated) can
substitute for distribution plant (regulated), or nearby generation
(unregulated) can substitute for a combination of distant generation
(unregulated) plus transmission (regulated). Bonbright said that in a
competitive market, "the market price will tend to come into accord with
production costS."ii That means that in a competitive market, society can
choose the lowest cost products needed to produce goods and services,
thereby increasing its economic efficiency. Competition, furthermore, "will
impel rival producers to strive to reduce their own production costs in order
to maximize profits. "iii In a mixed competitive-regulated marketplace,
consumers will choose between products with competitive prices (which,
overtime, should approximate costs) and those with regulated prices, (which
reflect regulators' determination of cost or price, and rarely reflect the cost
of that product to society). Under those circumstances, consumers will pick
the lowest priced alternative to them, which mayor may not represent the
lowest cost alternative for society as a whole.
Now that we understand the different roles of price in the two systems, let
us examine why the supposedly competitive, restructured system now under
construction will not produce the benefits that I, at least, had expected from
the introduction of competition, at least not for a while.
Pricing in Competitive Electricity Markets 41

2. OLD STYLE UTILITY PRICING

Supplying electricity involves a number of functions (such as generation,
transmission, distribution, or power quality), some of which consumers
could have produced themselves or bought from a supplier other than the
local utility. The utility might have furnished consumers with individual
prices for each of those services, but in reality, it could not have done so
easily, because both regulator and utility dealt with total price. Thus,
consumers had no way of determining whether they or their designated
agents could perform some of the supply functions for less cost than the
utility charged. All consumers could decide was whether they could self-
produce or buy the entire product more economically from the utility, with
"economically" defined not in any economics textbook sense that implies
producing the lowest cost to society, but rather meaning cheaper for
customers. Utility pricing, then, was a blunt instrument that sent few
rational signals to society as a whole, and certainly not enough signals to
allow consumers to make sensible decisions.
Returning to societal issues, the rate making process' relentless emphasis
on historical average costs led to one disaster after another in the past three
decades. Long run incremental costs began to rise in the 1960' s, but prices
did not move up in the same manner, thereby inducing uneconomic demand
for the product. When the utilities finally raised prices to cover the new
costs, demand fell below expected levels, thanks to price elasticity, and the
utilities had to explain excess capacity to regulators. The long run
incremental cost of power production began to fall in the 1980' s but utility
prices did not reflect that drop. The utility pricing structure started to drive
away customers, which led to price discrimination in favor of certain
customers who had options to self-generate or had political influence, and
finally led, to the collapse ofthe old system. iv

3. THE ROLE OF PRICING IN A COMPETITIVE

MARKET

Remember all those sterile discussions of supply and demand in

Economics 10 1, all accompanied by graphs without numbers on the axes,
which spared us the realization that our economics professor could not
measure any aspect of the process in real terms? Price, however, does serve
a purpose in that textbook world, in that economists assume that consumers
generally consume more of a product when price falls, and less when price
rises, and producers supply more of the product when price rises and less
when it falls. In times of shortages, when many consumers want the
42 Pricing in Competitive Electricity Markets

product, suppliers raise prices, and as they do so, consumers back off when
the price gets too high for them. Eventually, new suppliers enter the picture,
due to the high prices, and the increased supply causes prices to fall.
That seemingly simpleminded view of how markets work describes what
a competitive electricity market should look like, at least in the way the late,
legendary Professor Fred Schweppe of M.LT. described it to me, or at least,
the way I understood what he said. v Basically, in his concept of the
homeostatic Vi system, the network ties together all customers and producers.
They know the price of electricity all the time. They react to prices, by
producing more or consuming less. Now we get to the interesting part. The
new system need not require reserve margins because, as demand rises
towards maximum supply, price rises steeply and customers drop off the
system, either cutting down demand or turning on their alternative
generating units, thereby preventing an emergency. In other words, pricing
sends signals. As an economist of sorts, I like that solution. It is
unbelievably efficient. So I asked this question of a world famous systems
expert: "Can we be sure that enough customers will get off fast enough to
prevent the system from collapsing?" She answered, "No." So much for
pure pricing signals. We seem to have come across a glitch, also known as a
"market imperfection" in economic jargon. (Maybe, though, pricing can
keep the market from getting to that dangerous point and system controllers
step in only to prevent breakdowns.)

4. NOW FOR REALITY

Now, let us examine real issues. To begin with, the restructured

electricity market is nowhere near a textbook case of the competitive
marketplace. The regulators and industry participants have created
balkanized markets, in which generators could exercise market power in
their local areas. Regulators neither provide the economic incentives to
assure reliability at the lowest cost nor encourage expansion of the
transmission system, which would enhance reliability and reduce the market
power of local generators. Organizations with no incentive to operate
efficiently, but with a mandate to maintain reliability, will run key
components of the system, and may set reliability standards without regard
to economic consequences. Hundreds of marketers will compete to convince
consumers of the superiority of their electrons over the electrons of the other
suppliers. Certain producers will obtain privileges over the other producers,
depending on the political correctness of their output. And prices will reflect
previous errors, social policy, market power, and regulation. This will not
resemble a simple textbook market. In that jungle out there, the players will
Pricing in Competitive Electricity Markets 43

have to remember those immortal words of Leo Durocher, "Nice guys finish
last. "

5. WHAT WILL PRICE DO TO THE MARKET?

The correct price signals could and should make our electricity system
more efficient. They could cause entrepreneurs to insert electric and gas
transmission, storage, traditional generation, reliability services, distributed
reliability and distributed generation into the system at the right places and
times. They could replace cumbersome rules of the road designed by
regulators, security coordinators, and incumbents. With the right
measurement and communications facilities in place, the right price signals
would induce consumers to take rational steps to control demand or put on-
site generation into service, to react as normal consumers would do, and
those reactions would dampen the incredible volatility in the electricity
markets.
Price considerations have played a major and often perverse role in the
competitive generation market to date. The competitive entrants into the
electric business (most of whom are associated with existing utilities or
utility-like firms) and the incumbents seem to have placed big bets on the
direction of prices based on these assumptions:

• Transmission constraints will remain

• Anti-trust and regulatory authorities will not express concern about
market power as long as consumers pay less than they do now
• Low generating prices are a temporary phenomenon
• Global warming is not a serious business consideration
• Administrative reliability rules, which market players can manipulate,
will impede the development of market-oriented reliability
• Interactivity between customer and supplier will stay at a minimal level
• Distributed generation and distributed reliability will play insignificant
roles in the market

What if they are wrong?

6. CONSEQUENCES OF MISPRICING

A combination of regulators with a profound misunderstanding of market

phenomena, ineffective trust-busters and well-positioned incumbents could
lead to mispricing of the product, inefficient markets, and high prices for
44 Pricing in Competitive Electricity Markets

many consumers. But that same combination will create opportunities for
those who decide to beat the system rather than join it. Virtual utilities, for
instance, could aggregate customers, meter them, control their loads at peak,
put in distributed generation where necessary, and cut out of the system at
peak periods, saving their customers some money and pocketing the balance.
Doing so would not only reduce the peak, eventually, but also devalue power
facilities purchased for peak market conditions. Underground, underwater,
DC and superconductor transmission and distribution lines, some placed
without need for use of eminent domain, could bypass or enhance the
existing transmission and distribution grid, thereby undercutting the value of
facilities that benefit from congestion. Addition of flexible AC transmission
system (FACTS) devices and similar enhancements could open up jammed
transmission routes. Unfortunately, though, additions to the transmission
network might not follow without changes in regulatory attitudes toward
attracting capital.
At this stage, the regulators and the independent system operators in
charge of transmission have put their bets on a system of congestion charges,
plus command-and-control plus regulated return, which, so far, has failed to
induce investments in significant enhancements to the transmission system.
Are we setting up a system that will grow with the market? The Electric
Reliability Panel commissioned by the North American Electric Reliability
Council commented that:

"We cannot depend on market forces to provide incentives to

enhancement while transmission is as regulated as it is ... The
future of the transmission grid requires far more attention than it
has gotten to date."v;;

Some consumers will react to price by setting up distributed generation.

Mini-turbines now, and fuel cells soon, would enable them to do so, without
need for elaborate infrastructure. The introduction of an automobile-based
fuel cell would threaten the incumbent power suppliers because the device
could generate many times the electricity needed in a normal home. If the
electricity system cannot react to price signals, someone else will.
The Clinton Administration, in 1998, added still another wrinkle to the
picture by coming out in favor of net generation. The customer would have
a two way meter, and sell electricity into the grid, at will, presumably at
retail prices, although I do not know what would qualify as the purchaser in
the new system. Net generation, son of PURP A, might mix an administered
price supply that gets first shot at the market with competitive generators
that fight for the rest. Just think of the additional distortions to the market.
Think of the entrepreneurial opportunities, though, for sellers of distributed
generation.
 Pricing in Competitive Electricity Markets 45

The prices set in an imperfect, semi-regulated, semi-competItIve,

congested, balkanized, semi-oligopolistic marketplace can - and will - set
off competitive moves that will undermine the prices in those markets. The
first round of electricity pricing will encourage the commercialization of
technologies that will take business from those complacent and
conventionally minded people who populate this business, providing that the
emerging technologies come in at low enough costs.

7. CONCLUSION

A while ago, Marija Ilic and I co-authored an article that urged

policymakers to get it right the first time. We pointed out the many
instances of governments that deregulated in a hurry, with predictable
consequences, producing in the process "notable examples of misguided
public policies that neglected a systems overview and did not get it right the
first time."viii A year later, I attended an industry meeting at which an
unconcerned regulatory expert - unconcerned except for procedural issues -
assured us that nobody had any intention of getting it right the first time. I
am sorry about that, because we will incur unnecessary expenses, deprive
consumers of benefits, and drive many to make decisions they would not
have otherwise made, to the detriment of an efficient electricity supply
system.
In the end, I, as a consumer, have less faith in the ability of people who
are supposed to assure me reliable, efficient electric service and improve my
lot through industry restructuring, than in the efficacy of price signals that
will bring in new competitors, real entrepreneurs and technologies that are
close to commercialization. The second round of industry restructuring may
not put us in an optimal position, but it will produce an energy system more
competitive than the one we are about to get.

REFERENCES
i James C. Bonbright, Principles of Public Utility Rates, New York, Columbia University
Press, 1961, p. 44.

ri Bonbright, op. cit., p. 53.

iii Ibid., p. 53.

iv Leonard S. Hyman, America's Electric Utilities: Past, Present and Future (Arlington,
Virginia: Public Utilities Reports, 1997).
46 Pricing in Competitive Electricity Markets

v The concept may have been first described in Fred C. Schweppe, "Power systems '2000':
hierarchical control strategies", IEEE spectrum, July 1978, although that article does not
use the term "homeostatic."

vi The Oxford Dictionary and Thesaurus defines "homeostasis" as "the tendency toward a
relatively stable equilibrium between interdependent elements."

Vl1 North American Electric Reliability Council Electric Reliability Panel, Reliable Power:
Renewing the North American Electric Reliability Oversight System, Final Pre-publication
Copy, December 22, 1997, pp. 34-35.

V111 Marija Ilic and Leonard Hyman, "Getting It Right the First Time: The Value of
Transmission and High Technologies", The Electricity Journal, Vol. 9, No.9, November
1996, p. 9.
Chapter 4

Competitive Infrastructure: As An Enabler of

Market-Based Pricing

Eric P. Cody*
Wayfinder Group, Inc.

Key words: Billing Systems; Customer Choice; Information Technology, Load Profiling;
Settlement Protocols.

Abstract: A competitive energy market featuring lower cost energy and superior
environmental performance will only be realized through market-based
pricing, where prices are determined by the interplay between supply and
demand forces rather than by allocating costs of service into rates. New retail
pricing structures signal the fundamental economic characteristics of resources
available to retailers and reflect actions taken to hedge financial risk, while the
acceptance of pricing by consumers indicates their willingness to adopt
behavioral patterns consistent with their individual or corporate values.
Market pricing needs to be dynamic enough to capture the interplay of these
two forces and produce a more optimal outcome than that achieved by the
guiding hand of regulation. However, an innovative pricing structure will only
be meaningful if customers can be switched to the correct retailer and issued
an accurate bill.

The author's first hand exposure since mid-1997 to early retail access
implementation experiences and planning approaches being followed by
utilities in thirty states and several foreign countries, strongly reinforces the
notion that competitive infrastructure (the operational processes and
information systems) must be in place to enable fulfillment of a customer's
choice of energy retailer, as well as daily market clearing, and are critical
prerequisites for market-based pricing. If this infrastructure is not delivered
successfully, then customers' I requests cannot be acted upon effectively,
competitive pricing cannot be supported, and customer confidence will quickly
erode. Yet, in few locations worldwide is adequate attention being paid to
competitive infrastructure as an enabler of market pricing.

* "The opinions expressed in this chapter are those of the author and are not necessarily those
of Wayfinder Group, Inc. Global, Inc., or its affiliated companies."

A. Faruqui et al. (eds.), Pricing in Competitive Electricity Markets

© Kluwer Academic Publishers 2000
48 Pricing In Competitive Electricity Markets

This chapter uncovers some of the key tradeoffs that must be carefully
considered during the design of rules and procedures for the transition to the
competitive energy market to be completed successfully. Tradeoffs include
market pricing vs. market settlement, accuracy vs. predictability, and
centralized management of information vs. diffusion of control. The context
for competitive pricing is described, along with a summary of fundamental
challenges that have not been definitively answered by early retail access
experiences worldwide. Utilities and retailers who acquire a superior
understanding of these issues will be in a better position to mitigate the costs
of operating in an inherently risky business.

1. INTRODUCTION

The purpose of this chapter is to bring into sharp focus the interplay
between competitive electricity pricing and the business process and
systems, which enable it. Specifically, it looks at the industry's readiness to
support a new competitive market, not just to redefine how utilities look or
what they do. It considers three questions:

• Are restructuring activities taking place across the industry likely to

lead to a consistent and efficient structure for the competitive market?
• Will market participants be able to compete successfully when the go-
aheads are finally issued by state commissions, the PERC, and the
regional ISOs or transmission operators?
• How can the approach to market opening improve the odds that the
desired economic benefits of the new market will be realized within
the first few years?

Put bluntly, I explore how an innovative pncmg structure will be

meaningful only if customers can be switched to the right retailer on a timely
schedule, and customer bills are calculated accurately. The key message is
in two parts. First, competitive infrastructure is complex and needs to be
done right the first time, or the costs and system impacts to utilities and
retailers will be enormous. Second, the new market rules must be forward
leaning and recognize that a competitive market is being created in which
the nature of consumer protection needs to be rethought. Many regulatory
commissions, while very experienced in setting and enforcing rules for
monopoly businesses, are less experienced in designing a workable
framework for a competitive market, although staffers with longstanding
experience may find some of the fundamental lessons of telecommunications
or natural gas deregulation to be transferable. Moreover, some commissions
are being pushed strongly by consumer advocates and environmental
Pricing in Competitive Electricity Markets 49

interests to prescribe requirements such as disclosure labeling into the

unregulated domain. To the extent that regulatory prescriptions for the
competitive market are applied, in an environment characterized by
competitively thin margins, small retailers may view them as barriers to
entry. Larger retailers, utility affiliated or not, will experience significant
margin pressure.

2. COMPETITIVE ELECTRICITY PRICING - THE

ENABLER OF ECONOMIC BENEFITS

Pricing of electricity as a commodity for sale in a competitive market is

the key instrument by which the desired economic benefits of utility re-
regulation - namely, consumer savings and economic development - will be
achieved. It is pricing that enables the balancing of supply resources,
generating plants and other assets, with customer demand. Pricing reflects
how the value of an asset is projected into the marketplace and how signals
are sent to consumers to influence their behavior toward desirable patterns of
use in ways that were impossible under a regulatory process that depended
upon test periods, or cross-sectional snapshots, for rate making purposes.
The shift to more dynamic, market-based pricing is as fundamental in its
nature as the transformation of a centrally planned economy into a capitalist
economy. One of the key premises underlying the creation of a competitive
electricity market, then, is that innovative, market pricing will be supported
and enabled by the infrastructure (the "competitive infrastructure") of
business processes and systems being specified through the legislative and
regulatory de-monopolization process. If the magic of price-induced
economic efficiency is to be realized, there can be little margin for error in
creating this infrastructure, especially in regions that have below-average
electricity costs. In those locations, inefficient infrastructure could lead
prices to rise under competition.
Early experience worldwide, however, shows that it is not safe to assume
that competitive pricing will be enabled by the competitive infrastructure
designs typically being specified. Sophisticated, hourly pricing of
commodity electricity, to align economic signals to the consumer with the
hourly cost characteristics of the assets, is often quite difficult and evokes
conflicts with market settlement mechanisms which rely heavily on
representative usage patterns, called "load profiles." Measurement of energy
consumption at the level of granularity necessary to calculate hourly bills on
the one hand, and market settlements (defined in Section 7) on the other,
requires either massive technology investments in real-time metering
networks or stop-gap estimation techniques such as load profiling. Even
more importantly, the majority of customers appear to expe(;t simpler pricing
50 Pricing In Competitive Electricity Markets

structures rather than pricing which is even more time differentiated than
their pre-existing, bundled tariffs (a preference clearly recognized by
telephone long distance providers with their set price per minute offers).
One last point is indisputable - the high overhead costs that will be
experienced by retailers serving customers in multiple jurisdictions with
varying rules and procedures will ultimately be passed on to customers.

3. THE UNFOLDING MARKET LANDSCAPE

The newly competitive electricity markets differ in fundamental ways. In
Pennsylvania and New Jersey (where the market opened in the fall of 1999),
the initial emphasis is on rapid creation of a competitive retail electricity
market, driven by a substantial energy "shopping credit." In New York,
Rhode Island, Massachusetts, and California, the emphasis was on gradual
introduction of a competitive market in a manner that protects consumers
and provides savings for all during a transition period lasting several years.
Both offer some degree of savings, delivered by vastly different designs. In
the first year of Massachusetts' new electricity regime, mandated 10 percent
bill reductions yielded savings to consumers of around $500 million - yet
few customers actually switched generation suppliers.
Under the first model, which I call the "kindle the market" model, many
have tended to judge success by the number of retail suppliers offering
electricity and the volume of customers switching. Pennsylvania, in setting a
relatively high benchmark price for power, the so-called "shopping credit"
shown on the bill, has elicited an enthusiastic response from competing
suppliers able to proffer savings in their sales pitches to consumers.
Those with the "protect the consumer" model often judge success by the
amount of money all consumers are saving. These are guaranteed savings
that come with a lower benchmark power supply price (sometimes called the
"standard offer"). In Massachusetts, where the standard offer price trends
steadily up over its seven year life, the market is developing at a moderate
rate. Several thousand business customers have switched suppliers for
incremental savings, and buying cooperatives of residential customers,
chambers of commerce, and non-profit organizations are exploring
discounted, bulk buying opportunities.
These two general approaches and early results in five states have fed
what I believe is the most widely held misperception of re-regulation across
North America - that success is defined by the number of customers
switching in the near term. However, customer switching alone cannot tell
us whether competition is superior to monopoly regulation in its ability to
leverage lower prices and increase value to consumers. Ironically, it is the
benchmark price predetermined by law or regulation that creates the price
Pricing in Competitive Electricity Markets 51

differential the consumer sees and therefore determines how many switch
while the market is getting up to speed, plain and simple. But have they
saved? To conclude that New York and Massachusetts are less successful
based on infant market experience is naive since the economic development
goals of regulators and legislators in those states are clearly being achieved.
Competitive retailers view higher benchmark prices favorably since they
inject an economic stimulus and aid in the launch of the new retailing
industry. However, it would be an egregious mistake to allow such self-
interest to sway other state regulators to focus only on the rapid development
of the market, creating plans that may be more supplier-friendly than
consumer-friendly. And half the country already enjoys electric power at
rates below the national average, by definition. In these emerging markets,
overzealous stimulation of a competitive market for its own sake might well
lead prices upward.
Each state has to grapple with its own starting point conditions. Having
worked directly on competitive market issues with utilities in thirty states, I
can testify that no two face exactly the same situation or, for that matter, the
same prospect of economic opportunities through competition. Utilities in
different regions have different fuel sources, plant types and vintages,
weather conditions, regulations, social programs, and customer
demographics - it is no wonder that prices vary widely. And let's not forget
the unique role of rural electric cooperatives nationwide, whose customers
already make plenty of choices - they own the utility!
So while comparisons may be useful, it is too early to make judgements
about a long-term transition from electric monopoly to competitive market
environment. Market creation is not a sprint. It is a marathon. Those who
attempt to predict a winner after the first mile may be in for a rude
awakening at the finish line.

4. NAVIGATING FROM BUNDLED TO

UNBUNDLED PRICES
Figure 1 provides a schematic view of the journey which is traveled by
most utilities as they transition from a vertically integrated, regulated
monopoly structure with bundled tariffs to a structural model in which the
delivery business - transmission and distribution - remains regulated while
the generation supply business is fully competitive. From the point in time
when a state legislature or regulatory commission indicates its intention to
adopt some form of retail access and customer choice, this complex
navigational exercise typically takes a minimum of two years. In some
states where a more aggressive schedule was originally adopted for
52 Pricing In Competitive Electricity Markets

implementation, delays have often extended the total time for

implementation to a period in excess of two years.

Overview
OV1 Corporate Restructuring Roadmap

we

PROPRU;':TARY & CONFIDIii:NTIAL. Capy..-iuht

Figure 1. Corporate Restructuring Roadmap

Where does competitive pricing fit into these transition efforts? The
simple answer is pervasively. Competitive prices are inevitably viewed by
customers against the backdrop of the "benchmark" price, the shopping
credit or standard offer price which is generally set through an adjudicatory
proceeding or settlement negotiation. Beyond this single reality, a great deal
of detail work must also take place to simply enable the utility to break out
the other cost components which must be itemized on the customer's new,
unbundled bill. This process begins with restructuring of the corporation
into distribution, transmission, and generation entities; reassignment of costs
and financial assets to the appropriate business entities; development and
submission of an unbundled cost of service allocation study reflecting how
these reassigned costs are to be allocated to customer classes; design of
unbundled rates; and issuance of unbundled bills, preceded by the necessary
Pricing in Competitive Electricity Markets 53

customer education. Few utilities correctly anticipated that introduction of

unbundled rates would require significantly more customer communication
than the introduction of Choice of electricity supplier itself, and learned
through experience that greater call volumes and call durations were
typically associated with the delivery of the first month's unbundled bills.
Also, if customers' bills are not unbundled several months prior to the
advent of Choice, it should come as no surprise that some customers will
believe their previous total rate of, say, 94 per kilowatt-hour may be reduced
to the 44 per kilowatt-hour advertised for energy only by a competitive
retailer, simply because they have not assimilated an understanding of how
pricing has changed. There are many opportunities to create customer
confusion during the transition to a competitive market, any of which can
reduce customers' confidence at a time when they are being presented with
exciting new choices.
Unbundling of rates and competitive choice create two other types of
work: (1) Revisiting of rate policies, and (2) creation of billing plans.

1. Rate policies and customer terms and conditions need to be revised for
the following purposes:

• Line extensions
• Customer contracts
• Construction advances
• Demand management incentives
• Customer deposits
• Credit and collections

2. Billing plans need to be created with the following features:

• Budget monthly payment
• Credit card and direct debit forms of payment
• Consolidated billing of multiple accounts
• Internet billing

Many other work streams flow through this transitioning project, which
must be executed successfully to establish the starting point framework for
competitive pricing. Utilities and retailers who fail to pay attention to the
critical path steps in development of the market infrastructure and procedural
rules, embodied as navigational way points in the figure, often subsequently
learn that their competitive strategies cannot be easily realized or that
implementation will cost far more than planned. The devil is truly in the
details and this argues persuasively for direct involvement by stakeholders in
the development of detailed market rules and procedures.
54 Pricing In Competitive Electricity Markets

5. THE UNFORGIVING NATURE OF THE RETAIL

ELECTRICITY MARKET

Understanding the environment for competitive electricity pncmg also

requires an appreciation for the challenges faced by energy retailers, those
firms that acquire energy from generation owners and resell it to consumers
individually or through aggregators. Early experiences with competitive
electricity across the U.S. describe an environment for retailers which is
harsh and unforgiving of even tactical oversights. Moreover, rules that have
been crafted by legislative committees, regulatory proceedings and
settlements, and even collaboratives of utilities, retailers, consumer
advocates and environmentalists, have proven to be uninformed about the
true complexities of operating the new market. Rules that are inconsistent
across regulatory jurisdictions, even distribution franchise areas, add
overhead cost burdens to an already difficult situation. Most retailers face an
anxiety-inducing combination of these factors, including:

• Razor thin margins on the commodity

• High acquisition costs for residential and small business customers
• Potential for high customer turnover, or "chum"
• Predetermined benchmark prices that may be difficult to compete
against
• Cash flow volatility due to load profiling practices
• Referendum questions that threaten to overturn enabling laws
• Market 'rules' for the retailing of electricity that vary considerably
from those for natural gas and other commodities that will almost
undoubtedly be jointly marketed by some retailers
• Unique system interfaces for enrollment, billing, and other logistical
requirements in each utility service territory
• Customer confusion and uncertainty

The piecemeal pattern of market implementation, with its non-standard

procedures and rules, makes life even more unpleasant for energy retailers as
they struggle to find ways to survive the bruising fight for initial market
share. And ironically, should pricing evolve toward more sophisticated, time
differentiated options for the mass market, the level of investment in the
systems to support them would likely dwindle the number of retailers.
Fortunately, there are means for addressing many of these threats to the
viability of the competitive market. Failure to address them soon, however,
will most likely lead to a future that is dominated by a small number of large
national energy retailers, with significantly more market power than the
utilities they are supplanting. Implementation decisions are currently being
made routinely by regulatory commissions, utilities, and ISOs in the twenty
Pricing in Competitive Electricity Markets 55

or so states pursuing retail access. These will profoundly influence the ways
in which the new market will operate, and what competitive participants will
be able to do. To the extent that competitive energy retailers do not
participate alongside utilities at every step of these rulemakings, future
options may be precluded and the successful operation of the competitive
supply market may be impacted. Competitive pricing depends on metering,
timely data collection, electronic information transfers, and sensible rules,
and these are being set now. The popular debate over whether to also make
metering and billing competitively provided services bypasses many, more
fundamental questions that will ultimately determine the fates of most
retailers.

6. COMPETITIVE INFRASTRUCTURE A
CRITICAL PREREQUISITE FOR COMPETITIVE
PRICING

"Competitive infrastructure," as I use the terminology here, refers to the

business processes and computer systems that are necessary for the new
energy market to function correctly before, during, and after each trading
day, as shown schematically in Figure 2. These processes must tightly
integrate the various operations of utilities, retailers, customers,
ISOs/transcos, and power exchanges and must work well for all these
stakeholders, not just the incumbent utility. Accordingly, the center of
gravity must shift away from the utility, which, in practical terms, no longer
has the unilateral ability to specify all of the basic processing requirements.
Delivering a competitive infrastructure that works well, and is cost-effective
to implement, is challenging. Many have underestimated the complexity
associated with enrollment of customers with their chosen retailers, passing
of metering information for billing purposes, unbundled billing, information
sharing, allocating load and losses among retailers, transrrusslon
arrangements and market settlement. Virtually all of the arrangements that
need to be made for these market functions to work well, impact pricing to
some extent.
Moreover, all of these physical and financial functions are enabled by
electronic transactions that flow across the Internet or some other form of
open, electronic information trading network. So, simply preparing
computer processing systems is insufficient. The project encompasses core
business application systems, databases for information mining and
reporting, and electronic trading of information (E-commerce). For this
reason, the degree of risk rivals the now infamous Y2K millennium project
in some IT executives' minds. Figure 3 illustrates some of the distinct forces
56 Pricing In Competitive Electricity Markets

that will impact most utilities' core business systems during the next five
years.

Competitive Infrastructure Requirements:

Overview of Transaction Flows

NEESGlobal

Figure 2. Competitive Infrastructure Requirements: Overview of Transaction Flows

Pricing in Competitive Electricity Markets 57

Forces Threatening Core Utility Systems

f('!' NEESGlobal

Figure 3. Forces Threatening Core Utility Systems

What are the key attributes of this new competitive infrastructure? Some
activities are new and unfamiliar, such as enrollment or load profiling and
reconciliation. Some of the transaction volumes, by their nature, may be
high even where only a small minority of customers initially chooses to
switch. For many utilities, 25 percent or more of the total customer
population relocates each year. This process must now involve third parties.
Electronic information transfer must occur between the utility and its new
trading partners, some of whom have never delved into E-commerce
previously. ISOs and power exchanges are being built simultaneously, and
within compressed time frames. The adopted schedule for rollout of the
market may be unrealistic given the sheer amount of work which must be
performed. This particular combination of attributes describes a pattern any
veteran IT manager will instantly recognize - a high risk project. In fact,
there is no way to make the implementation of competitive infrastructure
into a low risk project. However, careful design can avoid implementation
becoming unnecessarily complex and overly risky.
As if the pressures acting on utilities' core application processing systems
were not enough, the new market also creates an insatiable demand for
customer information that is stored on the utility'S computers. These data
describing past customer use of electricity literally enable the retail market
58 Pricing In Competitive Electricity Markets

since without access to them, retailers assume inordinate risk, certainly out
of proportion to the narrow margins which are possible today. The data are
vital to decisions made by market participants, in that they allow profitability
screening of customer segments, determine pricing approaches, support
billing, and help marketers and aggregators to understand settlement
calculations and resulting cash flows. Utilities are traditionally very good at
processing enormous volumes of transactions, like issuing 60,000 bills daily,
but this kind of slicing and dicing of the raw information is a different story.
The information warehousing technology needed to deal with such market
demands is not yet fully in place at many utilities. And ironically, computer
systems that are optimized for high volume transaction processing are
generally very inefficient at handling ad hoc information queries, of the types
suggested by Figure 4.

Activities Requiring Information Sharing

{; NEESGlobal

Figure 4. Activities Requiring Information Sharing

Pricing in Competitive Electricity Markets 59

7. THE DILEMMA OF MARKET PRICING AND

MARKET SETTLEMENT
Traditional rate making and cost recovery practices produced tariffs based
most often on monthly energy usage and a cents per kilowatt-hour charge,
with time-of-use rates and demand based structures for large customers and
those with load patterns of special significance, like electric heating
customers. Metering, therefore, was heavily skewed toward the mechanical
dial, single register watt-hour meter. Noteworthy exceptions exist, such as
those few utilities that have installed automated meter reading networks for
improved customer service and closer-to-real-time rate designs.

Figure 5. Daily Load Estimation and Reconciliation Process

The new electricity market depends upon hourly values for customer
usage, either measured or estimated, as total system loads and delivery
system losses are allocated among the various retailers of electricity
operating within a control area. Pricing too may require hourly
measurement of loads for some customers, perhaps all customers eventually.
Load profiling, an estimation methodology, has been invented to meet this
60 Pricing In Competitive Electricity Markets

need for hourly or half hourly usage infonnation. Figure 5 illustrates

graphically how the various inputs are combined in load profiling to produce
trading day supplier load curves for imbalance reporting and financial
settlements. However, a central problem that has become evident is that
load profiling disconnects pricing from costs. This problem becomes critical
as the market evolves and hourly costs become more volatile.
I will leave the issue of which customers to directly meter and which to
profile-estimate for another day, but confront the pressing pricing issue
head-on here. For those customers being profiled, and there will be many
for years to come, how can hourly loads be estimated in a manner that
assumes they are price-inelastic when the most central motivation of the
market is to change market-based of consumption behavior? A secondary
problem also arises in that there is always a discrepancy between the ex ante
estimates and the ex post reality as reported by the meter when it is read. We
know a priori that there is error and bias in load estimation; in fact, we know
that part of the error will never be corrected even when the trading day is
ultimately closed.
The problems that load profiling presents for retailers are important to
understand, since profiling is a necessary evil for Customer Choice (unless
new metering technology can be financed and brought to bear across the
entire population of customers by the time the market opens). These
problems include but are not limited to the following:

• Profiled market-based of use will not match actual patterns of use,

hour by hour. Since customer bills are still in most instances based on
metered use and wholesale power bills on profiled use, volatility will
be introduced into cash flows.
• Balancing of retailers' wholesale power bills against customer bills
will be difficult, if not impossible, due to calendar month basis vs.
meter reading cycle.
• Load shapes used to represent individual customers and customer
classes are often drawn from some historic period (a so-called
"proxy" day), which we know cannot be perfectly representative of
the trading day being settled. i
• Even if customers' hourly loads were all measured directly for use in
daily settlements, line losses must still be allocated across retailers
and these can be a significant percentage of load requirements.
• Profiling errors in individual hours can be significant and when these
errors combine with spikes in hourly imbalance charges, very large
distortions to cash flow will occur. These may never be fully
corrected, even at the time of final trading day settlement.
Pricing in Competitive Electricity Markets 61

The one thing we do know is that the load profiling methodology is a

stop-gap arrangement that is necessary for the opening of markets. Its
primary weakness is its inability to capture dynamic responses to price
change, and this is precisely the goal of industry restructuring. As the
market becomes liquid and volatile over time, the evolution itself promises
to render the technique less acceptable from the perspective of retailers.
Regulators and the rules which are being put in place should anticipate the
need to reduce the market's dependence on load profiling within three to five
years after market opening, if not sooner.

Comparison of Generation Supply Options

Typical Structures
Standard Rate Option Default Generation Senice Competitive Pricin2
• Non-competitive supply option • Non-competitive supply option +C-Ompetitive supply option
+ Provided by successful bidder or + Provided by successful bidder or • Provided by competitive
incumbent utility incumbent utility retailers. including Affiliated
Retailer
+ Eligibility: Residential and + Eligibility - All customers • Eligibility: All customers
small business customers
.3-5 year tenn of availability + Permanent availability • A vailability subject to contract
tenns. market price
• Known price. market proxy • Routing market price +Competitive priee
• Regulated tariff • Unregulated price + Unregulated priee
• No risk to customers • Risk borne by customers • Balanced priee and risk
• Reflects average margins for • No margin (pass-through) +C-Ompetitive margins
residential/small business class

~) NEESGlobal

Figure 6. Comparison of Generation Supply Options

8. DEFAULT GENERATION OPTIONS AND

COMPETITIVE SUPPLY

Figure 6 (above) underscores the reality that this is a market in transition

by comparing generation supply options commonly emerging across the
industry. Virtually all models of Customer Choice include provisions for
62 Pricing In Competitive Electricity Markets

some form of default generation service in the event that some customers
prefer not to choose a retailer or that they are not offered any competitive
supply option. Some, but not all, also provide a discount, or stable, rate
option for generation supply, an element introduced initially in the New
England settlement agreements reached during 1997, beginning with
Massachusetts Electric Company.
The market effects, and logistics, of these supply options must be clearly
understood by retailers since this is often the "path of least resistance" for
consumers who are uncompelled by sales promotions or difficult for retailers
to reach. It is equally important for utilities setting up their systems to
recognize that these supply options do not always fit easily into the
transaction structure that has been designed for competitive retailers.
Default generation service does not "market" customers, may not have
billing or customer inquiry center capabilities, and may require the
distribution utility to assume some or all of the risk of market price
fluctuation, even where the utility has divested its generation assets and
developed a "wires-only" corporate vision.

9. CUSTOMER EXPECTATIONS ABOUT PRICING

The competitive electricity market, it is hoped by many, will be the

magical instrument which will accomplish what regulation has been unable
to do over a period lasting two decades - price a product whose costs are
highly time sensitive in a way that will encourage consumers to modify their
patterns of use. This balancing of production cost against product utility, if
done successfully, offers greater economic efficiency. During the 1980's,
the concept of real-time pricing (RIP) intrigued regulators with its promise
in achieving a similar outcome; however, two stumbling blocks became
apparent. First, metering would require massive upgrading investment to
support the monitoring and communications attributes central to RTP.
Second, and in spite of some R&D experimentation, it never became clear
that the majority of customers would prefer such an hour to hour pricing
regimen, although the sophistication of certain very large customers and
their ability to save under RTP were obvious. The same two factors
continue to apply today as the industry restructures and unbundles the
services previously provided under monopoly regulation. Experiences in the
u.K. and Australia indicate that perhaps 5 percent of electricity is being sold
under RTP, although this share can be expected to change with the
introduction of new measurement technology and better access to detailed
information on patterns of use.
Pricing in Competitive Electricity Markets 63

Do customers want hourly pricing signals? Evidence from pilot programs

and early Customer Choice experience suggests the answer, "perhaps not."
Flat pricing of kilowatt-hours, with or without a monthly customer charge,
appears to be winning. The potential savings associated with changing
consumption behavior, (e.g., running a dryer in the middle of the night to
take advantage of low cost, off-peak energy, simply are not large enough to
warrant significant adjustments in consumer behavior). Even if appliances
were time clock synchronized with energy rates, someone still has to unload
the dryer.
The most central theme may tum out to be that predictable cost is worth
more to customers, particularly small and medium sized businesses, than
lowest possible expected cost, subject to some uncertainty. Hence, some
retailers operating in the Massachusetts and Rhode Island markets have
committed to a price that is only marginally lower than the published
Standard Offer price over its seven year expected life, yet significant
numbers of municipal customers have signed up. As more detailed customer
switching data become generally available, across the variety of market
models being witnessed, it will be become clearer how consumers weigh
price reductions against hassle avoidance and predictability.

10. ENABLING THE VISION OF COMPETITIVE

PRICING TO BECOME REALITY

Several fundamental issues that must be overcome for market

restructuring to be successful have been presented in this chapter. To recap:

• Failure to standardize and implement in a timely manner the complex

infrastructure of new business processes and systems that are needed to
support competitive pricing will seriously undermine the ability of
retailers to participate in and the market to deliver the desired economic
benefits;

• Consumer confidence is already challenged by the scale of changes

inherent in industry restructuring, so it is important to manage
implementation risk to avoid creating an uncertain situation in which
consumers lose confidence in the new market; the result of such
confidence erosion will be customers who 'stay put;'

• Retailers should carefully assess whether specific consumer segments

really want more sophisticated pricing structures before insisting that the
enhanced infrastructure necessary to support them be put in place;
64 Pricing In Competitive Electricity Markets

• The effect of default generation service arrangements on pricing must

not be overlooked, as these represent the benchmark price which
consumers see during the transition away from monopoly service;

• Consistency in rules and procedures across the restructured industry is

sadly lacking and only being pursued as an afterthought today;
importantly, this weakness must be addressed to keep many retailers
alive in the competitive marketplace long enough for real pricing
innovation to be witnessed;

• The discrepancies that occur between load profiling and customer

billing, and pricing and costs, are serious enough to derail many new
entrants before they become established in the marketplace, if they are
not understood and planned for; and

• The unfolding electricity market is a harsh, unforgiving environment that

will punish those who do not master the technical challenges inherent in
customer acquisition, enrollment, billing, energy procurement and risk
management, and customer renewal. Only when these technical
prerequisites are passed can a retailer move toward truly competitive
pricing and margin levels which provide for a sustainable, ongoing
business.

NOTES

i "Dynamic profiling," in which the hourly loads of customer class samples are downloaded
nightly and class mean hourly loads calculated for the actual trading day, is an
improvement over the proxy day approach but involves substantial metering and
operational costs.
Chapter 5

Competitive Rates - A Break from the Past?

John Neufeld
University of North Carolina at Greensboro

Key words: Competition and Rate Design; Demand-Charge Rates; Electric Rate Design;
Electric Utility History; Hopkinson Rates; Louis Brandeis; Marginal Cost;
Regulation and Rate Design; Samuel Insull; State Rate Regulation; Time-of-
Day Rates; Wright Rates.

Abstract: Although competition seems to require a major shift in the ways electric
utilities set prices, the industry is really returning to the situation which
prevailed in its infancy at the beginning to the twentieth century. Most
traditional rate structures were designed during that period according to basic
pricing principles still in operation today, and were designed to deal with the
competition the industry faced during that period. The adoption of state rate
regulation and changes in technology eliminated or reduced much of that
competition, and eliminated the incentive for rates to change as market
conditions changed. As a result traditional rate structures became an
anachronistic relic of earlier conditions.

To many within the electric utility industry the issue of pricing has
seemed long settled. Although that process varies somewhat from firm to
firm, the steps by which rates are determined have become very standard.
The process starts by determining the utility'S revenue requirements or total
costs (including return to capital). The next step allocates these costs in
various ways: function (transmission, generation, distribution), type (demand
cost, energy cost, customer cost), and customer class. The final step is to
design rates that will recover these costs from each customer class. The long
experience utilities have had with this methodology means the process is
well understood, if not entirely automatic. My experience with rate
"engineers" indicates that most will agree that some aspects of cost
allocation are arbitrary, that competing theories can lead to different
allocations, and that it is often not clear that anyone theory is more "correct"
than the others. There is widespread awareness that economists and other
A. Faruqui et al. (eds.), Pricing in Competitive Electricity Markets
© Kluwer Academic Publishers 2000
66 Pricing In Competitive Electricity Markets

have criticized this traditional "embedded cost" approach, but most utility
practitioners still favor it. It seems to them to have a clear objective basis
that ties the price an electricity user must pay to the costs of serving that
user.
The move to a situation in which electric utilities must compete with one
another is likely to have a profound effect on pricing practices and strategies,
as the other chapters in this book make clear. The best pricing system for
any particular electric utility will depend on the precise institutional
framework within which it competes-especially the nature of any remaining
regulation and the degree of competition. Two principles should govern
market pricing when not prohibited by regulation:

1. No electricity customer should be charged a price less than marginal

cost (except, perhaps as part of a short-term promotion). Any
customer willing to pay a price that exceeds marginal cost should be
eagerly accepted.

2. The extent to which any customer should be charged a price

exceeding the marginal cost of supplying him or her depends entirely
on that customer's price elasticity, that is, sensitivity to price. This,
in tum, is primarily determined by the alternatives available to the
customer - other fuels and other electricity suppliers.

These principles are consistent with a variety of pricing schemes, but not
traditional embedded cost rates. The less a utility is regulated and the more
it faces competition, the more important these principles are and the less able
the utility will be to adhere to a traditional embedded cost pricing
methodology. Furthermore, pricing is likely to never again be as settled as it
apparently has always been in the electric power industry.
How could there be such a difference between the pricing system the
industry grew up on and that which appears to lie in its future? The answer
to this question lies in the industry's past, when the traditional pricing
system was developed and implemented. Most details of traditional rate
structures were determined prior to regulation, when firms in the industry
faced the greatest level of competition they would face prior to now. The
early industry was engaged in an active search for rate structures before that
time and used the two principles of competitive rate design to evaluate new
designs. Many, if not most, of the rate structures now regarded as "new"
(such as time-of-day rates) were well-known and used prior to the end of the
19 th century. State regulation of electric rates changed the nature of the
incentives facing electric utilities; stability became more important than
adjusting rates to meet changing circumstances.
Pricing in Competitive Electricity Markets 67

In its earliest days, the chief competition to electric utilities came not
from other electric utilities but from other energy sources. The electric
power industry actually precedes Thomas Edison's well-known
achievements. The power produced by that industry went to arc lights,
whose harsh, brilliant light was unsuitable for use in most indoor spaces. In
urban areas, most indoor lighting was provided by centrally distributed
manufactured gas-itself a relatively new technology (gas lighting first
started in New York in 1824). Incandescent lighting required a system
different from arc lighting, and Edison established the first successful
incandescent system on Pearl Street in New York in 1882. Edison based his
price on the cost of gas lighting, which was then $2.25 per 1,000 cubic feet.
His original price was 1.2¢ per lamp hour, based on a sixteen candle-power
bulb. In other words, the first modem American electric utility did not base
prices on cost of production; it based them on the cost of a competitive
alternative.
As we shall see, this approach remained important and is reflected in
modem rate structures. Interestingly, in 1879 a price war existed among gas
producers in New York, and the price had fallen to 75¢ for 1,000 cubic feet.
That price war wasn't ended until the following year. Could Edison have
competed with gas if he had developed incandescent lighting two years
earlied By 1899 the price of gas had fallen to 50¢, and the invention of the
gas mantle substantially improved the quality of gas light. In at least some
New York districts, electric lights lost ground to gas lights. By 1900,
reduced selling prices for both bulbs and electricity made the cost of electric
light only 17 percent of what it had been in 1882.
Edison wanted to sell light rather than electricity because he anticipated
substantial improvements in efficiency of the light bulb, and he wanted this
benefit to accrue at least partially to the utility.;; Edison also failed to fully
anticipate the non-lighting demand for electricity. His original rate was
equivalent to 24¢ per kWh, (uncorrected for inflation). The selling of light
rather than electricity, however, did not become standard within the industry,
although the practice of supplying light bulbs without additional charge
remained a relic of Edison's original plan for many decades with some
utilities.
The issue of exactly how electricity should be priced surfaced quite early
in the industry's history, about ten years after Edison's Pearl Street Station,
and was a major topic of discussion around the tum of the century. A record
of those discussions exists in the engineering journals of the time and in the
proceedings of professional and trade associations. The discussion was
international in scope, with English engineers often taking a leading role,
and was often very sophisticated. There were virtually no participants from
outside the electric power industry, and no evidence that those engaged in
the discussion had tried to familiarize themselves with pricing issues in other
68 Pricing In Competitive Electricity Markets

industries. In particular, there were practically no references to the pricing

experiences of the railroads, which faced some strikingly similar problems.
Those in the U.S. electric power industry were kept aware of the parallel
discussions also occurring within the power industries of Britain and
continental Europe. Although the discussions were quite wide-ranging, of
particular interest are the genesis of the modem "demand charge" rate
structure and early discussions of time-of-day rates. iii
The brilliant British engineer John Hopkinson made one of the earliest
contributions to the theory of electric rates, perhaps the first derived from a
theoretical model of the economics of electric power, in his Presidential
address to the Junior Engineering Society on November 4, 1892.
Hopkinson's understanding of the relationship between pricing and
competition is, perhaps, even more relevant to the industry today than it was
in his time:
The charge for a service rendered should bear some relation to the
cost of rendering it. If it is a matter of open competition the matter
will settle itself, for no one will for long be able to supply some
customers at a loss, and recoup himself by exorbitant profits from
others. If the matter be a case more or less of monopoly, the
adjustment is less certain; thus the Post office charges Y2d. postage
for a printed circular and Id. for a written letter, the two costing
the Post Office exactly the same. iv
He furthermore recognized that although electricity faced competition
from other energy sources, competition from other providers of electricity
would have the biggest impact:
There is no object in reducing the cost of electricity for lighting in
the case of any customers much below the cost of equivalent
lighting by gas, unless there are competitors in the field willing to
do it. v
The heart of Hopkinson's analysis involved the well-known peak-load
problem, which he compared to the situation involving rapid transit:
... the Metropolitan District Railway must be prepared to bring in
its thousands of passengers to the City at the beginning of the day
and to take them back in the evening, and for the rest of the day it
must be content to be comparatively idle. The line must be of a
carrying capacity equal to the greatest demand, and if this be great
for a very short time the total return for the day must be small in
comparison with the expense of rendering the service. In such a
case it would not be inappropriate to charge more for carrying a
person in the busy time than in the slack time, for it really costs
more to carry him. vi
Pricing in Competitive Electricity Markets 69

Despite this excellent analogy, Hopkinson's proposed rate structure

contained an important error. Hopkinson advocated "a fixed charge per
quarter proportioned to the greatest rate of supply the consumer will ever
take, and a charge by meter for the actual consumption."Vii It was common
practice in those days to base electricity rates either on the basis of the
number of installed lights (which would determine the maximum possible
power needed by that user) or on the basis of energy consumption as
measured by a meter. Hopkinson's proposal can be regarded as simply
noting that both factors should be used. But this ignores the crucial fact that
the "greatest rate of supply the consumer will ever take," may occur at a
different time from the peak demands from all other electricity users. Of
course, during Hopkinson's time, when utility-generated electricity was used
overwhelmingly for artificial light, the peak demand for nearly all users
came at about the same time after dusk. As will be seen below, other
engineers quickly rectified this error. Despite this, Hopkinson's rate
structure is still in wide use today and is typically referred to as the
"Hopkinson" rate. It is certainly an irony of history that Hopkinson's name
would primarily live on today attached to a mistake he made in rate design
given his major contributions to electrical engineering. viii I have been unable
to find any record of Hopkinson responding to his critics or ever saying
anything further on the topic of electricity rates. He was unfortunately killed
at the age of forty-nine in 1898 in a mountain climbing accident.
Discussion of Hopkinson's paper appeared in several letters to the British
journal, Electrician. Two noteworthy letters were those from Arthur Wright
and Gisbert Kapp, both of whom essentially advocated new metering
technologies to support new rate structures. ix Gisbert Kapp, who deserves to
be known as the father of time-of-day pricing, advocated two prices for
electricity: a higher price during the time of day when the central station
experienced maximum total power demand and a lower price at other times.
This was to be supported through the use of a meter that contained a clock. x
Surprisingly, Kapp's interest in electricity rates apparently did not extend
beyond this one letter, although other engineers continued to champion the
time-of-day rate, and frequently referred to it as the "Kapp rate."xi Arthur
Wright, on the other hand, became a prominent and vocal supporter of the
demand charge rate. In his letter Wright supported Hopkinson's proposal
but claimed to have improved upon it by developing a meter capable of
. . . xii
measunng maxImum power consumptton.
Wright's meter offered an important benefit to the demand charge rate
structure because it eliminated all of the practical problems associated with
using the number of light bulbs (connected load) as a measure of maximum
power use (such as on-site inspections). Wright also introduced an
influential variation on Hopkinson's demand charge. Rather than simply
applying a charge per kilowatt of maximum power consumption, Wright
70 Pricing In Competitive Electricity Markets

employed a declining block rate in which the size of the initial (higher-
priced) block was determined by maximum power consumption. As long as
an electricity user's consumption exceeded the amount of this initial block,
there was no difference between bills rendered under a Wright rate rather
than a corresponding Hopkinson rate. Why, then, did Wright introduce this
complication? It is clear that Wright realized that the two methods were
equivalent. xiii The British Parliament, however, had begun setting a
maximum price which electric utilities could charge per kilowatt-hour.
Wright was able to implement his demand charge in accord with these laws
by using the legal maximum as the rate for the initial block. xiv It is another
historical irony of electricity rate design that the Wright rate structure
became widely used in the US. (and is still widely used) even though US.
utilities did not face the problem which led to its creation.
Although the work of British engineers on electricity rate structures was
reported in the US., some time passed before Americans picked up the
debate in earnest. In the earliest days of electric power, the primary
competitive concern for electric utilities was gas lighting, not other electric
utilities. As indicated above, important technological advances were made
in gas lighting as well as electric lighting.
The two different pricing schemes most used at that time were either to
price based on energy use alone, measured by meter, or to price based on the
number of installed lights. As mentioned above, Hopkinson's contribution
was to note that both factors should be used. Charging by meter priced
electricity in a manner like that used for pricing gas. xv It did not, however,
track the marginal costs of providing electricity and had some unfortunate
effects when the costs of gas and electric lighting were so close. Lighting
has a natural peak, and failing to differentiate the extra cost of providing
energy during the system peak aggravated the peaking problems faced by the
early utilities. A particularly frustrating development for the industry was
the development of lighting fixtures that could use both gas and electricity.
Electricity was used during the early evening hours, when guests were likely
to be visiting. At other times, gas was used. This merely aggravated the
peaking problem. Pricing solely on the basis of the number of installed light
bulbs avoided this problem but introduced others. Such pricing discouraged
the installation of light bulbs and required the utility to prevent customers
from surreptitiously installing lights. Technological progress eventually
brought down the cost of electricity, but the most important factor was the
development of daytime demand for electric power.
The major users of electricity during daytime were industries.
Unfortunately for electric utilities, they faced a very serious competitor for
the provision of industrial electricity: the self-generation of electricity by the
plant itself in a so-called isolated plant. xvi The optimum size for generating
equipment was very small by modem standards, and a factory could easily
Pricing in Competitive Electricity Markets 71

provide its needs by installing the same type of equipment used by an

electric utility operating in an urban area. xvii The advantages to the factory
were many: it avoided the distribution and billing costs that a utility would
have to incur, and the waste heat could be used as process steam. In the
early days of electricity there was probably no wayan electric utility could
get this business; the marginal cost to a utility of supplying an industrial user
was likely greater than the total cost to the factory of generating its own
requirements. As shown in the chart below, isolated plants in manufacturing
remained a very important competitor to electric utilities for quite some
time. It wasn't until after 1914 that power provided by electric utilities
exceeded that produced in isolated plants. Even as late as 1929, isolated
plants provided 35 percent of the power used to power motors in
manufacturing.

80%

70%

60%

50%
..,OJJ
..,'"
4()%
C
..,u'- 30%
i:l..
20%

10%

0%
1899 1904 1909 1914 1919 1923 1925 1927 1929

Year
Source: U.S. Census of Manufactures. 1929, Washington: GPO, p. 112.

Figure 1. Percentage of U.S. Electric Motor Power Used in Manufacturing

Supplied by Self-Generated Electricity

The importance of the isolated plants problem to electric utilities is

reflected in the ample contemporary discussions which have been preserved
in trade journals and in the records of industry trade groups. Consulting
engineers made a business of advising electricity users about whether their
particular interests were best served by an isolated plant or by purchasing
power from a utility. Engineering and trade journals frequently considered
the question of which source was cheaper and reported on interesting
72 Pricing In Competitive Electricity Markets

isolated plant installations. xviii In 1909, at the meeting of the National

Electric Light Association (NELA - later to become the Edison Electric
Institute) several papers were presented on the issue of isolated plants. xix
The NELA was eventually to form a committee on purchased power.
Technological improvements increased the optimum generator size until
they became too large for factories. At this point, the utilities had an at least
theoretical advantage over isolated plants. The industrial business was
absolutely critical to the long-run viability of electric utilities; if, however,
they got that business by simply charging industrial users marginal cost, all
of the benefits would flow to those industrial users. The lower marginal cost
of serving industrial users justified a lower price, and the alternative isolated
plants offered such users made them very sensitive to a utility's prices. If
electric utilities were to avoid ceding all the benefits of industrial daytime
use of electricity to those users, the prices charged those users would have to
reflect that sensitivity. A sophisticated pricing scheme would be needed, and
the demand charge rate structure fit that need perfectly.
The individual perhaps most responsible for publicizing the demand
charge rate structure in the U.S. was Samuel Insull, one of the most
interesting utility executives in the history of the U.S. electric power
industry.xx Insull became an advocate for Arthur Wright's pricing system
and acquired an interest in the American rights to Wright's meter. Wright
attended the NELA meeting in 1898 and strongly advocated the use of his
meter and pricing system. At this meeting and at other times, Wright
engaged in debates with advocates of time-of-day pricing, including,
interestingly, Insull's chief engineer, Louis Ferguson. xxi Despite this, Insull
adopted Wright's system in Chicago, promoted it elsewhere in the U.S., and
took credit for its widespread use in this country in 1922. xxii
Wright invariably defended his pricing scheme as the best way of
reflecting the costs of serving a particular customer even when he was
clearly being out-debated on this point by advocates of time-of-day pricing.
As long as lighting was the primary application for electricity provided by
utilities, there was likely only a slight benefit to time-of-day pricing over
Wright's pricing scheme. Ironically, it was the opportunity to sell electricity
for daytime industrial motor use that gave the demand charge rate structure
its real advantage. Wright had a meter to sell, and he may not have realized
that the benefit of the demand charge rate structure was not that it tracked
the cost to the utility of serving that electricity user but rather that it tracked
the user's sensitivity to the price the utility charged. It was the perfect
device for automatically discounting the price to exactly those for whom the
discount was necessary. The cost to an electricity user of electricity from an
isolated plant was largely determined by the maximum power needed (which
determined the plant size) and the total amount of energy which would be
generated (which determined fuel costs). The cost to a utility of providing
Pricing in Competitive Electricity Markets 73

electricity to a user is largely determined by the amount of energy required

during the system peak (which requires a larger plant as well as adding to
fuel costs) and the amount of energy required off-peak (which adds to fuel
costs). xxiii The demand charge rate structure bases the price of electricity to a
user not on the cost to the utility of supplying that electricity but on what the
costs to the user would be of obtaining electricity from an isolated plant. It
thus automatically provided a lower price to those for whom isolated plants
were most economical.
In order for the demand charge rate structure to work effectively for a
utility, the utility had to have certain characteristics. First, it had to enjoy
economies of scale sufficient to ensure that its marginal costs were below the
prices it would charge using the demand charge rate structure even if the
customer's peak coincided with the system peak. No utility was in a better
position to do that than Insull' s Commonwealth Edison. Second, the utility
had to be large enough that in these pre-regulation days there was an
advantage to having a published rate schedule rather than simply negotiating
with each potential customer. Commonwealth Edison was probably the
largest electric utility at the time and was located in a major industrial area.
Isolated plants still offered the advantage of providing steam which could be
applied to other uses, but Commonwealth Edison met that challenge by
creating a subsidiary to operate boilers to provide steam under contract to the
utility's electricity customers. xxiv This service was clearly used as a means of
granting additional discounts to those for whom the steam by-product was
important. On its own, the subsidiary did not cover its expenses but was
justified by the profits it generated for the sale of electricity. Insull
recognized the importance of pricing structures and even claimed that proper
rates "may possibly have had as much to do with reducing operating costs
and reducing interest and depreciation costs as have the wonderful work of
the inventors and the marvelous skill of the engineers."xxv
The interest in rate structures in the early industry led to other
suggestions as well. In 1901, L.R. Wallis proposed an interesting variant on
the demand charge rate structure. xxvi In this version, an electricity user
would contract with the utility for the maximum power which, instead of
being measured with a meter, would be controlled by a current limiter. Such
a system has been widely employed in Europe. Although good quantitative
data is not available, clearly time-of-day pricing was used in the early
industry. xxvii Metering technology was an active area of investigation, and
many ingenious meters were developed. xxviii
Shortly after the tum of the century a hiatus occurred in discussions on
electricity rates, perhaps because the industry became relatively less
concerned with the energy markets and more concerned with capital
markets. Between 1881 and 1912, gross capital expenditures by electric
utilities grew at an annual rate of almost 18 percent as the industry created
74 Pricing In Competitive Electricity Markets

the nation's first electricity infrastructure. Furthermore, until 1915, each

year's gross capital expenditures exceeded that year's total revenues. xxix At
this time, utilities were subject to regulation by the municipalities in which
they were located. Two characteristics of this municipal regulation were
worrisome to utilities. First, utilities were generally not granted a protected
monopoly. Although consolidation between 1900 and 1906 led to de facto
monopolies in many cities, competition was always a possibility. The other
problem was corruption, of which utilities were both instigators and victims.
The large immovable capital expenditures that electric utilities had to make
in a city made them vulnerable to extortion, both from corrupt politicians
and from competing utilities (who would battle using their powers of
eminent domain). xxx Both of these problems contributed to financing
difficulties since it was investors whose funds were tied up in the utility's
capital. There was also a perception that municipal regulation was not
effective. Setting maximum prices could lead to reduced quality of service
and tended to be ineffective in an industry where rapid technological
progress naturally led to lower prices. A municipality's bargaining position
was strongest at the time a fixed-term franchise was being awarded or re-
awarded. This led to the unfortunate disincentive for the franchise holder to
undertake investment (or even maintenance) when the term was close to
ending since there was no way of ensuring that capital costs could be
recovered if the franchise was lost. Consolidation sometimes led to a single
utility holding multiple franchises. This might make an award to an alternate
franchisee difficult if the multiple franchises had different expirations since
each territory might be too small to permit efficient operation.
The rise of Progressivism led reformers to seek an end to corruption and
inefficiency in city government. Many Progressives favored municipal
ownership and operation of utilities, and considerable discussion occurred
over the relative advantages of municipal versus private ownership.
Municipally owned electric utilities had a long history, and many of them
were established because of the reluctance of private interests to undertake
the risky investment. Municipalities had an advantage in capital markets in
that they could pledge other assets (including especially their powers of
taxation) as backing for their bonds. By 1902, municipalities owned about
23 percent of the nation's electric utilities; they were smaller than average,
however, and provided less than 8 percent of total industry output. xxxi State
regulation of railroad rates had already been in existence for some time, and
was seen as a model for utilities both by some utility executives and by
Progressive politicians. xxxii In 1905, the National Civic Federation
established a blue-ribbon group of prominent leaders to look at the issue of
whether the nation would be better served by utilities which were privately
or municipally owned. Included were future Supreme Court Justice Louis
Brandeis, United Mine Workers president John Mitchell, and Samuel Insull.
Pricing in Competitive Electricity Markets 75

A "committee of investigation" consisting of twenty-one individuals equally

divided among those who had expressed an opinion in favor of municipal
ownership, in favor of private ownership, and those who were undecided or
had no opinion set out to investigate utilities both in the u.s. and in England.
The final report comprised three volumes. Although the committee could
not agree on the central issue of ownership, they did agree that utilities
should operate as monopolies and that private utilities should be subjected to
regulation which applied uniform accounting rules and which made their
records public. One of the people who worked on the report was the noted
economist John R. Commons. Commons used the recommendations of the
yet unpublished study to formulate a Wisconsin law, adopted in 1906,
establishing state commission regulation of electric utilities.xxxiii This law,
and the NCF report, served as models for subsequent state commissions.
Because, perhaps, of the lack of progress of the municipal ownership
movement, a number of prominent Progressive politicians, including Robert
M. LaFollette and Charles Evans Hughes, became advocates of the
regulation of electric utilities by state commissions. xxxiv In 1907, the NELA
Subcommittee on Public Regulation and Control issued a report favoring
regulation which would protect the capital investment of utility corporations.
Although there was some opposition within the industry, utilities and their
executives were frequently in the forefront of advocacy for the establishment
of state regulatory commissions. xxxv
State rate regulation seemed designed to reassure investors. Utilities were
protected from competition (from other electric utilities), and the rate-
making process explicitly recognized the necessity of a utility repaying its
investors and should have reduced the risk of investing in utility stock or
bonds. On the other hand, regulation did not protect the utility from other
forms of competition and opened its rate making process to a new level of
scrutiny. This led the industry to a new phase in the evolution of rate design.
The issue was no longer one of determining the ideal structure to use in
pricing electricity; the issue instead became one of defending the industry's
ability to continue using rates appropriate for the Cnon-utility) competitive
conditions in which it operated. Once the issue of rates moved from the
strictly business sphere to the political and legal sphere, utilities faced a very
real challenge over their ability to control rate design.
The challenge to utilities came from an association of manufacturers of
machinery for isolated plants that formed the "Uniform Electric Rate
Association." They championed the legal and political argument that
utilities should be required to base rates solely on the cost of service and not
on a user's demand characteristics. The Association obtained and published
as a pamphlet a legal opinion prepared by Louis D. Brandeis (not yet
Supreme Court Justice) who argued that charging different rates to different
users was only legally justifiable if it could be shown to be cost-based.
76 Pricing In Competitive Electricity Markets

Differences based solely on differences in the characteristics of demand

(including the feasibility of using isolated plants) were not legal. The
publication of this pamphlet led to over twenty letters to the editor of
Electrical World on the issue of uniform rates. xxxvi Several state legislatures
debated bills that would have stripped from power companies the ability to
engage in any differential pricing. xxxvii
The NELA met the pricing challenge by forming a special committee on
"Rate Research." Despite its name, the committee's purpose was to forge an
industry consensus on the proper form of rate design. The committee
opened an office in Chicago and, for several decades, published a weekly
periodical, Rate Research, that reported on all news affecting electricity
rates, especially the actions of regulatory commissions. Rate Research also
reprinted many of the now classic papers on electricity rates, including
Hopkinson's original paper that argued (incorrectly) that the demand charge
rate structure reflected the cost of serving a customer. In 1914, the Rate
Research Committee's report strongly advocated "value-of-service" as the
primary basis for structuring rates. The committees specifically defined
value-of-service as the cost to an electricity user of obtaining an equivalent
or substitute means of service and regarded "demand" (in the sense of the
demand charge) as a measure of that value:
The value of the service to the customer depends on what it would
cost him to make it himself, and this cost clearly depends in part
on the size of plant that he would need. The size of plant that he
would need is determined by his maximum demand and necessary
reserve ... The demand is at least a rough measure of this cost, and
is therefore a test of the value to the buyer.
The committee judged time-of-day rates as undesirable even though they
did reflect differences in the cost of service because the value of electricity
to the buyer did not change by time of day.xxxviii
Fortunately for the industry, early decisions by utility commissions were
quite receptive to the notion that utility rates had to deviate from the cost of
providing service to certain customers. A number of these decisions
explicitly accepted the notion that rates for wholesale service should be
based on the cost to the purchaser of providing the service for himself. xxxix
An early regulatory decision (1909, Massachusetts) made the role of the
demand charge clear:
The demand system, whatever its faults in determining the
individual's cost to the company, has at least the merit of
recognizing the most essential elements determining the probable
cost to the individual of supplying himself, and therefore operates
to fit the price which the company must make to get his business,
to his actual condition. xl
Pricing in Competitive Electricity Markets 77

Regulation inverted the normal relationship between the prices a business

charges and the revenue it receives. For competitive firms operating in a
market, the total revenue the firm receives is a function of the prices it
charges. For such firms, pricing is very often an extremely important
determinant of commercial success. xli For regulated electric utilities, the first
step in a rate hearing is typically the determination of the company's
allowable earnings. It is that, rather than the prices charged individual
customers, which determines the firm's total revenue, and that is where
electric utilities have devoted their energies. Once the total amount has been
determined, dividing that among the utility's customers has become almost
an afterthought. In order to determine the utility's total return, and insure
revenue sufficient to compensate investors, regulatory commissions
undertake to determine total costs (including cost of capital). While this is
appropriate to prevent excessive profits and ensure adequate return to
investors, it does not provide the information needed to determine individual
customer rates.
Once the utility's total return is determined, it has little interest in
continuing the hearing process. Although individual customers or groups of
customers may participate in a rate hearing, arguments about the rates
charged anyone group are a zero-sum game: any decreases achieved by one
group require offsetting increases on another group of customers. Utilities
and utility commissions have managed to minimize this controversy by
increasingly embracing the fiction that rates charged individual electricity
users are in some meaningful sense equal to the cost of providing service to
those individual customers. Controversy is also avoided at each hearing by
continuing the rate structures that the regulatory commission approved
earlier. Thus the demand charge rate structure remains to this day the most
popular way of pricing electricity to industrial users. As a history of the
development of electricity rate structures makes clear, these rate structures
were not developed to be cost-based; they were developed to reflect the
sensitivity of industrial users to prices (price elasticity). The demand charge
rate structure is an anachronism of economic conditions facing the electric
utility industry seven decades ago. Its continued importance is evidence of
the extent to which the regulatory system has inhibited work on pricing
within the industry.
Success in competitive pricing will go a long way to determine the
winners and the losers in competitive electricity markets, and knowledge of
how to price electricity has been lost since the industry became regulated.
The exact form of pricing schedules will depend on the way the structure of
the market evolves, but the bases of electricity pricing must be a knowledge
of the marginal costs of serving different electricity users and a knowledge
of each user's sensitivity to price. These principles were understood by the
early industry, which faced serious competition for the provision of
78 Pricing In Competitive Electricity Markets

electricity, but regulation made price setting a relative backwater. Those

within each utility responsible for setting rates have been lulled into
believing the preservation of anachronistic practices is a scientific exercise
in determining the cost of providing service to each individual user. For
many of them, competition will be a rude and upsetting experience.

REFERENCES
1 Charles E. Neil, "Entering the Seventh Decade of Electric Power." Edison Electric Bulletin.
vol. 10. Sept. 1942, pp. 321-332.

ii Ibid .. p. 330

111 A very large proportion of electricity rate structures for commercial and industrial users
contain a demand charge (or equivalent) which basis a user's bill on the maximum amount
of power used (the user's "peak"). In its most common form, no account is made of the
time in which that maximum power demand occurred. The rate thus fails to account for
the major factor determining the marginal cost of providing electricity to that user: the
user's contribution to the system peak.

IV John Hopkinson, "The Cost of Electric Supply," Transactions of the Junior Engineering
Society, vol. 3, 1892-3, P 39.

v Ibid., p. 40.

vi Ibid., p. 35.

Vl1 Ibid .. p. 39.

viii Hopkinson is credited with inventing the three-wire system for electricity distribution and
(with his brother) working out the general theory of alternating current and the operation
of ac generators in parallel. A biography of him can be found in the Encyclopedia
Britannica. including its web site at "Hopkinson, John" Encyclopa:dia Britannica Online.
<http://www.eb.com:180fbol!topic?eu=41936&sctn=I>[AccessedJuly81999].This
biography does not mention his work on electricity rates.

IX Electrician, (December 23,1892), vol. 30 pp. 201 and 221.

, Kapp was born in Vienna and pursued his electrical engineering career in both Germany
and Britain. He became the first professor of electrical engineering at the University of
Birmingham, which currently has a building named after him. The University's then Head
of the Department of Electronic and Electrical Engineering wrote an unpublished
biography of Kapp in 1972. That biography credits Kapp's primary achievements as being
associated with the multi-polar slow-speed dynamo and the understanding of "magnetic
circuits." No mention is made of his work on rate design.
Pricing in Competitive Electricity Markets 79

XI
Of particular note is a series of articles written by Alfred H. Gibbings in a British journal.
Gibbing's analysis, including his criticism of Hopkinson's rate, is very sophisticated.
Alfred H. Gibbings, 'The Various Methods of Charging the Public for Electricity from a
Central Station," The Electrical Review, vol. 35, no. 870, July 27, 1894, pp. 96-97; no.
871, pp. 125-127; no. 872, pp. 157-159.

xu Wright's ingenious meter essentially wound a resistance wire in series with the customer's
load around the bulb of a maximum registering thermometer. Since the thermometer
could not register a temperature increase instantaneously, Wright's meter tended to forgive
power spikes which were very short in duration, much as modem demand meters tend to
integrate power use over some period such as fifteen minutes. "Methods of Charging for
Electricity: The Wright Rebate Indicator," The Electrical Review, (November 6, 1896),
Vol. 30, pp. 595-598.

xiii He made this clear in Arthur Wright, '"Some Principles Underlying the Profitable Sale of
Electricity," Journal of the Institution of Electrical Engineers, London: E. and F.N. Spon,
Ltd., 1902, vol. 31, p. 481.

XIV Ibid., pp. 489-490.

In 1900, one of the leaders of the electric utility industry made this point at a meeting of
the leading industry trade group:" ... when meters were procurable, they (electric utilities)
adopted the system used by the gas companies, and have since gradually awakened to the
fact that while this system may be suitable to gas business (which is questionable), it is not
suitable to the electric business." Henry L. Doherty, "Equitable, Uniform, and Competitive
Rates," Proceedings of the National Electric Light Association, Twenty-Third Convention,
1900, New York: The James Kempster Printing Company, p. 292.

xvi Isolated plants were used not only in factories. Institutions, hotels, and, occasionally, large
residences also employed them. Industrial users were most important for the utility
industry because of their daytime non-lighting use of electricity.

xvii The Census Bureau made this precise point in 1902. U.S. Bureau of the Census, Special
Reports, Central Electric Light and Power Stations: 1902, Washington: GPO, 1905, p. 3.

xviii For example, R.S. Hale, "Isolated Plant vs. Central Stations Supply of Electricity: A
Suggestion for Obtaining Estimates of Costs on a Competitive Basis," Electrical World
and Engineer, vol. 42, Sept. 5, 1903, pp. 383-4; H.S. Knowlton, "The Central Station and
the Isolated Plant," Cassier's Magazine, vol. 32, 1907, pp. 359-363; "Electrical Plant in
the Newark Free Public Library," Electrical World and Engineer, vol. 42, Aug. 15, 1903,
pp. 271-272.

xix Various authors, HIsolated Plants," National Electric Light Association, Thirty-Second
Convention: Volume II, Technical and Commercial Sections, Papers, Reports, and
Discussions,

" Born in Britain, as a teenager Insull became the indispensable personal secretary to
Thomas Edison. As a young adult he was in charge of production and distribution for
80 Pricing In Competitive Electricity Markets

Edison General Electric, Edison's primary manufacturing company. When Edison

eventually lost the battle between AC and DC, Edison General Electric merged with
Thompson-Houston in what might now be called a hostile takeover to form the modem
General Electric. Insull was the only Edison man offered a position in the management of
the new company. Loathe to work for Edison's business enemies, Insull accepted a 75
percent cut in pay to become president of Chicago Edison, one of several electric utilities
in that major city. An incredible risk-taker, Insull outmaneuvered corrupt politicians and
pushed technology to create in Commonwealth Edison the first modem large-scale utility
serving an entire urban area, making Chicago the most electrified city in the world. His
reputation and skill made him a spokesman for the industry and eventually put him at the
center of a fast-growing electric utility holding company, Middle West Utilities, which
was a leader in bringing electricity to small towns and some rural areas. Middle West's
financial structure was shaky, and missteps following the 1929 Stock Market Crash led to
its bankruptcy, at the time the biggest business failure in American history. Now at the
center of the public utility holding company imbroglio of the '20s and '30s, Insull's fame
changed to infamy, and he was indicted for securities fraud. Fleeing the country to escape
prosecution, Insull became the object of a massive high-publicity international manhunt.
After his capture and return to the U.S., he was acquitted of all charges. A sympathetic
and well-written biography is Forrest McDonald, Insult, Chicago: University of Chicago
Press, 1962.

'Xl Arthur Wright, "Profitable Extensions of Electricity Supply Stations" with discussion,
Proceedings of the National Electric Light Association, Twentieth Convention, 1897, New
York: The James Kempster Printing Company, 1897, pp. 159-189, 190-209,213-221.

xxii Samuel Insull, "Thirty Years of Chicago Central-Station History," speech given on
September 20, 1922 reprinted in Samuel Insull, Public Utilities in Modem Life, Chicago:
Privately Printed, 1924, pp. 347-348.

xxiil This simplifies the situation for an electric utility by ignoring the issue of different
generation types and equipment used for transmission and distribution. A more
sophisticated account would not substantially change the story, however. Furthermore,
these issues were less important to utilities in the early days.

'XlV S.M. Bushnell, "Central Station Operation of Steam Plants in Connection with Lighting
Company's Service," NELA Proceedings: Thirty-Second Convention, Volume II,
Technical and Commercial Sections, 1909, pp. 778-815.

xxv Samuel Insull, "Twenty-Five Years of Central-Station Commercial Development," NELA

Proceedings, 33 rd Convention, 1910, vol. I, p. 216.

XXVI L.R. Wallis, "The Foresee (4-C) System of Charging," NELA Proceedings, Twenty-
Fourth Convention, 1901, New York: The James Kempster Printing Company, 1902, pp.
34-51. Discussion, pp. 255-280.

xxvil W.J. Hausman and J.L. Neufeld, "Time-of-day pricing in the U.S. electric power industry
at the tum of the century, " Rand Journal of Economics, vol. 15, Spring 1984, pp. 116-126
Pricing in Competitive Electricity Markets 81

XXVIII One interesting meter was designed to allow the central station to control (and thus
change) the exact time of the peak when a premium was charged. See E. Oxley, "Multiple
Metering of Electric Currents," U.S. Patent 593852, issued November 16, 1897.

XXIX Calculated from data provided in M.1. Ulmer, Capital in Transportation,

Communications, and Public Utilities: Its Formation and Financing, Princeton: Princeton
University Press, 1960. pp. 320-321 and 476-477.

xxx M. Glaeser. Outlines of Public Utility Economics, New York: Macmillan. 1927. p. 204,
232; D.F. Wilcox. Municipal Franchises. New York: McGraw Hill. 1910, vol. 1, pp. 101-
132.

xxxi U.S. Department of Commerce and Labor, Bureau of the Census, Central Electric Light
and Power Stations, 1902, Washington: UPGPO, 1905, pp. 7, 24.

xxxn Samuel Insull had called for state regulation of electric utilities in his capacity as
president of the NELA in 1898. His primary argument was that it would increase the
industry's access to capital.

xxxin John R. Commons, Myself, New York: Macmillan, 1934, p. 120.

xxxiv M. Glaeser, op.cit., p. 234.

xxxv Douglas D. Anderson, Regulatory Politics and Electric Utilities (Boston: Auburn House,
1981), pp. 39-47.

xxxvi See letters to the editor section of Electrical World from October 25, 1913 to July 31,
1915.

xxxvii "Central-Station Rates Discussed at Boston," Electric World, vol. 57, Mar. 9, 1911, p.
604; William H. Winslow, "Rate Making for Central Stations," Electrical World, vol. 63,
Jan. 3,1914, pp. 12-13.

xxxvin "Report of the Rate Research Committee and Discussion," Proceedings of the National
Electric Light Association, Thirty-Seventh Convention, New York: James Kempster
Printing Company, 1914, pp. 59-116. Quoted material appears on p. 88.

xxxix A list can be found in L.R. Nash, Public Utility Rates, New York: McGraw Hill, 1933, p.
321. A particularly interesting view of the demand charge rate structure by the
Massachusetts regulatory commission in 1909 can be found in Rate Research, vol 2, (Oct
23,1912

xl "Electric Rates-Massachusetts," Rate Research, vol. 2, Oct. 23,1912, pp. 52-53.

xli An important exception is when numerous sellers are selling a standardized commodity. In
such a case, a single market price will prevail.
SECTION III

NEW ISSUES
Chapter 6

Anticipating Competitor Responses in Retail

Electricity Price Design

Ken Seiden, and Ahmad Faruqui

Quantec, and EPRI

Key words: Anticipating Competitor Actions; Game Theory; Market Experiments; Market
Simulations; Retail Pricing.

Abstract: The process of anticipating competitor response in retail pricing strategies can
be improved by incorporating both theoretical and experimental findings
within market simulation tools. Working in a fashion similar to flight
simulators, which accelerate pilots' learning processes, price- and product-
choice simulations explicitly capture the essential mechanics of competitive
dynamics, providing a valuable training ground for marketing and pricing
strategists.

1. INTRODUCTION

Analysts in competitive industries have incorporated the responsiveness

of customers to prices, advertising, and product features in strategic business
plans for decades. However, inclusion of the tactics and strategies of
competitors in retail price design is a recent development. Modem
competitive strategy explicitly recognizes the actions and responses of
competitors in the battle for customers and profits. New analysis tools
combine recent theoretical contributions from economists, game theorists,
and marketers with field data to simulate potential market interactions and
outcomes.
This chapter summarizes research sponsored by EPRI, 1999 to
incorporate competitive dynamics and anticipate competitor responses in
pricing strategies. It includes an overview of oligopoly and game-theoretic
A. Faruqui et al. (eds.), Pricing in Competitive Electricity Markets
© Kluwer Academic Publishers 2000
86 Pricing In Competitive Electricity Markets

modeling approaches that can be used to anticipate competitor reactions to

pricing decisions, and insights from market experiments, including results
from competitive retail pricing experiments using EPRI's Product Mix
Model.

2. MARKET MODELS IN THE ELECTRIC POWER

INDUSTRY

Market simulation modeling is not new in the electric power industry.

Responding to the energy price shocks of the 1970' s, the industry has been
at the forefront of the estimation of customer demand functions and the
responsiveness of demand to changing electric prices and rate designs. Path-
breaking studies include the estimation of energy and peak demand models
across market sectors and time-of-day, and more recent analyses of industrial
customer responsiveness to real-time prices.

Custom ers

Incumbent Challenger
Energy Energy
Com pany Companies

Figure 1. Competitive Market Dynamic

Most of the research to date has assumed a bilateral relationship between

customers and their incumbent energy company (i.e., utility). Recent
industry restructuring and the development of competitive retail and
wholesale markets essentially render the bilateral models useless. Those
pricing studies that have incorporated an element of competition typically
assume that competitors' prices passively remain unchanged as the
Pricing in Competitive Electricity Markets 87

incumbent energy company seeks to meet profit and other corporate

objectives through innovative pricing designs.
In the multilateral market model illustrated in Figure 1, competitors are
viewed as active market participants. From a profitability standpoint they
are as important as customers are, and no decision is made without
considering all aspects of competitive dynamics.

3. POTENTIAL METHODOLOGICAL
APPROACHES

There are several potential approaches that can achieve the desired
elements of a market simulation tool for the retail electricity industry. We
classify prospective methods within two main groups: (1) Market outcomes
from classic economic models of oligopolistic rivalry, and (2) Analytical
solutions based on applications of non-cooperative game theory to industrial
organization and marketing.

3.1 Classic Oligopoly Models

Oligopolistic markets occur when there are few enough sellers that each
firm believes its fortunes are noticeably affected by the actions of other
firms, and that its actions influence the fortunes of other firms. Classic
oligopoly theory offers particularly useful insights into firm behavior as an
industry moves from a regulated monopoly market to a more competitive
environment. The models can generally be viewed as describing market
outcomes given the nature of firm interdependencies. Each model has a
distinct set of assumptions that describe these interdependencies, and as the
assumptions change, so do market outcomes.
As illustrated in Figure 2, there are an infinite number of theoretical
possibilities ranging from joint monopoly to perfectly competitive outcomes.
Ultimately, industry structure depends on whether or not barriers to entry
exist, the types of customers in the market, what differences exist in products
(costs), and on the nature of competitive rivalry.
Cournot developed the first formal model of oligopolistic behavior over
150 years ago (Cournot 1838). His duopoly (two-firm) model explicitly
recognized firm interdependencies through the specification of competitive
dynamics. Cournot assumed that each firm would select its output given that
it has no control over the rival's production decision; the rival's output is
assumed to equal its previous production level. The optimal choice or
88 Pricing In Competitive Electricity Markets

reaction for each firm is summarized through its "guess" at the rival's level
of output.

Price

M . Joint Monopoly
c· Coumot
B - Bertrand
I S . Stackelherg
I
_ _ _ _ _ 1_ PC - Perfect Competition
I
I
. • • . . . • . • . . J . . . . 1. ...
: I Supply or MC

Output

Figure 2. Oligopoly Market Equilibria

The Coumot duopoly equilibrium exhibits certain features that are quite
reasonable for oligopolistic markets. As anticipated, the market price-
quantity pair lies between the competitive and monopoly outcomes.
Additionally, the assumptions of linear demand, homogeneous products, and
constant marginal cost allow the Coumot duopoly equilibrium to be
generalized into a more general oligopoly equilibrium where the market
approaches the competitive equilibrium as the number of firms increases.
However, three aspects of the Coumot model have been subjected to
criticism:

1. The model assumes that the firms' products are identicaL

2. Each firm selects its level of output rather than the price it will charge.
3. The model assumes both firms' myopia will continue indefinitely.

Bertrand changed the rivalry notions of the Coumot duopoly such that
firms react to one another's prices rather than quantities, and each firm
assumes that its rival will not change price (Bertrand, 1883). Consumers
still perceive that the products are identical, so Firm l' s demand function is
assumed to exhibit the following properties:
Pricing in Competitive Electricity Markets 89

• Demand for Firm l' s output is equal to the entire market's demand for
the product if its price (PI) is less than the price set by Firm 2 (P2),
• Demand for Firm l' s output is equal to zero if PI > pz, and
• Demand for Firm l' s output is one-half of market demand if PI = pz.

Firm 2 faces an identical demand function. Suppose Firm 1 tries to price

as a monopolist (Pm in Figure 2). Firm 2 could choose to price at the same
level and split monopoly profits with Firm 1, but since it thinks its rival will
not change prices (Bertrand assumed the same myopia as Cournot), it can
improve profits by slightly reducing price and serving the entire market. If
this occurs, however, Firm 1 would retaliate by charging an even lower
price. This process continues until neither firm has an incentive to change
price-at the competitive equilibrium where price equals marginal cost, as
shown in Figure 2.
Slight extensions of the basic Bertrand model yield outcomes that are
more plausible. First, if the product homogeneity assumption is relaxed and
the firms have differentiated products (i.e., different demand functions), the
model will yield a price equilibrium above marginal cost. Second,
Edgeworth showed that if firms are capacity-constrained, a Bertrand price-
rivalry can result in prices above marginal cost even if products are identical
(Edgeworth 1925, pp. 111-142).
If firms are capacity-constrained, they cannot serve all customers. Some
customers will have to buy from the higher priced firm. Returning to a
Bertrand-style price war, at some point one of the firms will find it more
profitable to raise its price and serve a small market instead of lowering its to
serve a large market share. When that occurs, the other firm will raise its
price to a level just below that of the first firm, triggering another price war.
These "Edgeworth cycles" continue indefinitely.
Each of these classic models assumes that rivals are essentially equal.
Another branch of classical oligopoly theory considers market outcomes
when one firm is dominant and is considered the market leader. The
Stackelberg leader-follower model builds upon the basic Cournot framework
by allowing one firm to know how the other will react (Stackelberg 1934).
Returning to the Cournot duopoly model, Firm 2 still assumes that Firm l's
output will remain fixed, and it maximizes profits given this belief. Firm 1,
however, knows this to be true and uses this information to find its optimal
level of production. The well-known Stackelberg solution asserts that if the
market leader knows the follower's reaction function, it can capitalize on
that knowledge. Notice that this knowledge is not sufficient to produce the
desired outcome. The leader must also provide information-in this instance
its level of production-to facilitate the intended reaction from the follower.
90 Pricing In Competitive Electricity Markets

The models presented here illustrate the range of possible market

outcomes contained in classical oligopoly theory. There are other models,
such as dominant firm price leadership and contestable markets. In the
dominant finn model, a market leader sets price, and all other competitors
set theirs at or just below the dominant firm's price. A contestable market is
characterized by freedom of entry, so it is impossible to maintain industry
prices and profits above perfectly competitive levels because of the threat of
rivals entering the market.
One of the main criticisms of classic oligopoly theory has been that it
consists of a hodgepodge of unrelated models where prices and quantities
exist across the broad range between monopoly and perfectly competitive
market outcomes. This lack of specificity led industrial organization
theorists to seek a more comprehensive theory of competitive rivalry. Game
theory provides the methodological framework for the analysis of
competitive rivalry. Seemingly disparate classical oligopoly models are now
seen as special cases within this general framework.

3.2 Game Theory

Each of the models described above recognizes the central feature of an

oligopolistic market structure: interdependence and rivalry among finns.
They can be viewed more generally as describing market outcomes given the
nature of firm interdependencies. Each model has a distinct set of
assumptions that describe these interdependencies, and as the assumptions
change, so do market outcomes. Decision making can be described as
interdependent if the choices of individual decision makers, each of whom
exercises self-interested behavior (which might include concern for the
welfare of others), affects the interest of other decision makers.
Game theory is the study of this interactive decision making. In addition
to industrial organization, game theory has been broadly applied to economic
issues such as competitive bidding, collective bargaining, and auctions.
Additionally, it has been applied to non-economic fields including political
science, biology, psychology, and military strategy. Almost all modem
models of oligopoly have a game-theoretic foundation.
Game theory remained the purview of academic mathematicians and
social scientists until the 1980' s. The mathematical foundations of the
theory date to the publication of The Theory of Games and Economic
Behavior (von Neumann and Morgenstern, 1944). In the half-century since
this publication academicians have shown that there is virtually no limit to
the behavioral complexity that can be analyzed with game theoretic models.
In practice, however, very simple models are often sufficient to shed
considerable light on real world behavior. Marketing theorists, who were
primarily interested in modeling consumer behavior, began to explicitly
Pricing in Competitive Electricity Markets 91

incorporate game-theoretic competitive concepts into their models about

fifteen years ago. The theory has been applied to traditional marketing
problems such as the nature of price wars, product positioning and pricing,
advertising, entry, and product distribution. While it cannot always provide
a definitive, prescriptive strategy, game theory can help executives and
managers think strategically and make well-informed decisions by providing
insights into complex market issues. For an extensive exposition of game
theory, see Fudenberg and Tirole (1991). Bierman and Fernandez (1998),
Tirole (1988), and Moorthy (1985) provide very readable applications of
game theory to economics, industrial organization, and marketing.
In recent years, game-theoretic concepts have moved beyond providing
material for academic journals into furnishing substance for government
policy, bestsellers, and textbooks. Game theory has also moved into the
forefront of economic theory. John Nash, John Harsanyi, and Reinhard
Selten were awarded the 1994 Nobel Prize in Economics for their
contributions to game theory. While the Nobel committee was making their
decision, the Federal Communications Commission (FCC) was using game
theory to design the auction of the electromagnetic spectrum for personal
communications services (PCS); this led bidders to employ game theorists as
strategy consultants.
We group game-theoretic modeling approaches into two areas based on
their applicability to retail electricity market simulation modeling:

1. Static games. A game is static if no player can observe what other

players do or change his strategy in response to others' strategy, and if
the players play the game only once and are uninterested in future
interactions. If each player knows who the other players are, the
strategies they have available, and the payoffs for each strategy
combination or profile, the game is one of complete information.
2. Dynamic games. A game is dynamic if a player can observe the moves
of other players and react to those moves. Any static or one-shot game
can be used to generate a dynamic game by repeating it over and over
again. Dynamic games can have a finite or infinite time horizon. If a
dynamic game of complete information is also characterized by
uncertainty about game history-the move or moves opposing players
have made-the game is said to have imperfect information. It has
perfect information if all players know the history of the game at each
stage of the game.

The classic oligopoly models developed by Cournot and Bertrand are

static games. These models yield solutions that are "stable." No firm would
choose to change strategies given the strategies of other players. The game
is played once; there is no opportunity to improve one's profit the next
92 Pricing In Competitive Electricity Markets

period because the next period does not exist in a static-modeling

framework.
Dynamic games improve upon this unrealistic assumption. Multi-period
games have been built that expand upon the basic Coumot, Bertrand, and
Stackelberg assumptions. Important dynamic considerations such as firms'
discount rates and the number of periods affect market outcomes. For
example, a market characterized by Bertrand behavior can yield prices above
marginal cost if the firms believe that the competitive situation they now
face will be repeated indefinitely. The key insight of dynamic games of
complete information is that credible, future actions influence the present. A
strategy is thus distinguished from individual actions. As with the game of
chess, a strategy represents a plan for every possible contingency.

4. GAME THEORETIC EXPERIMENTS

Analysts typically apply econometric modeling techniques to test

behavioral hypotheses with actual market data. However, these data are
often unsuitable for testing the predictions of game-theoretic models because
it can be difficult to isolate reactions to rivals' behavior or obtain enough
events from which statistically valid results will emerge. Consequently,
human experiments are now the primary tool by which the behavioral
predictions of the theory are tested. Kagel and Roth (1995) provide a
comprehensive review of experimental methods and results.
Two strands of experimental research provide guidance toward the
development of retail energy market simulation models: (1) the reactions of
competitors to one another in posted-price situations, and (2) the reactions of
competitors in auction markets. The first area illustrates the key parameters,
assumptions, and data necessary to model markets where customers
purchase energy and energy services from a set of posted or known
alternatives. Most residential and small non-residential customer segments
would be included in this group. The second area reveals information about
market outcomes where large customers or aggregators use bidding
procedures to select providers.

4.1 Dynamic Bertrand and Cournot Experiments

Bertrand games typically use a posted-offer auction format where all

sellers must simultaneously announce prices. Buyers are chosen randomly
and matched to sellers or quantities are simulated from a market demand
function or a matrix of possible outcomes. In Coumot games, sellers choose
Pricing in Competitive Electricity Markets 93

quantities simultaneously and prices are determined from an aggregate

demand function.
Dynamic Bertrand price-choice and Cournot quantity-choice games have
been studied through experiments. Some of these games have prisoners'
dilemma-like discrete alternatives, but most have a more expansive set of
'continuous' choices and payoffs. Fouraker and Siegel (1963) performed
some of the first experiments. In their Cournot games, subjects chose
quantities of a homogeneous product simultaneously and payoffs were
determined from an aggregate demand-price-profit table. Each game had
several stages and players did not know when the game would end until the
ending was announced. Both duopoly and triopoly games of complete
information were played, as were triopoly games of incomplete information
(about others' payoffs and quantity decisions). The results, which were
reported for the twenty-first stage of each game, were as follows:

• In games of complete information, duopoly market outcomes ranged

evenly from the competitive industry solution to the joint-monopolist
solution.
• In games of complete information, a majority of triopoly market
outcomes were produced market quantities that exceeded the Cournot
outcome, and most of these were at the competitive level. The addition
of a single player restricted tacitly collusive behavior.
• Incomplete information about rival's individual quantities and payoffs in
the triopoly games outcomes tended toward the Cournot solution.

More recent quantity-choice studies confirm the trend found in the

Fouraker and Siegel experiments: collusive behavior often occurs in duopoly
markets; however, in markets with three or more firms, outcomes are closer
to the competitive equilibrium than the Cournot equilibrium. Binger, et af
(1990) study markets with forty repetitions and find that duopoly markets
outcomes are, on average, at the Cournot level, but that markets with five
sells have outcomes that are between the Cournot and competitive levels.
Holt (1985) and Mason Phillips and Redington (1991) analyze repeated
duopoly markets and find that outcomes are between the Cournot and joint-
monopoly levels. Beil (1988) studies markets with four firms and finds that
outcomes are between the Cournot and competitive levels.
Price-choice experiments are more prevalent in the industrial organization
experimental literature. As with the Cournot experiments, consumers are
typically represented by mathematical or tabular demand relationships that
define firms' products as homogeneous or heterogeneous. In some studies,
sellers have complete information regarding available price choices, the past
price levels, and profits of all competitors. Other studies have incomplete
94 Pricing In Competitive Electricity Markets

information about previous price levels and profits; each seller only knows
whether his price was higher or lower than competitors' prices.
Fouraker and Siegel (1963) also present price-choice results in their study.
In these games subjects chose prices of a homogeneous product
simultaneously, with the lowest price seller winning all demand. In duopoly
markets with incomplete information, prices converged at or near the
competitive level; however, with complete information, prices were evenly
distributed between the monopoly and competitive outcomes. Prices tended
toward the competitive equilibrium in triopoly markets with and without
complete information.
Murphy (1966) used a similar price-choice design for duopolists with
incomplete information that had additional stages. He found, consistent with
Fouraker and Siegel, that prices tended toward the competitive level initially,
but continued repetitions caused prices to move toward the joint-monopoly
outcome. Alger (1987) found that cooperative pricing behavior often
occurred in duopoly price-choice experiments. Stoecker (1980) also looked
at duopolies but under conditions of complete information where subjects
played several ten-period games and switched partners each game. Subjects
learned from repeated play the benefits of tacit collusion; after a few games
most players priced at the joint-monopolist level until near the end of game.
However, continued research by Selten and Stoecker (1986) indicates that
this learned 'ability' to effectively collude unravels in later rounds of ten-
period games. This finding is consistent with the theoretical proposition that
in finitely repeated games the process of backward induction precludes
rational players from colluding.
Most price-choice experiments that compare duopoly markets with
markets with more sellers have outcomes consistent with Cournot-
experiments: with three or more firms collusive behavior is less likely.
Davis, Holt, and Villamil (1990) find that the presence of one additional firm
leads to more competitive outcomes; additionally, Stoecker (1980) also finds
that duopolists are unable to generate joint-monopoly outcomes when a third
firm is present. However, Davis and Holt (1994) find that prices are
consistently above competitive levels in markets with five firms.
In summary, multi-period price-choice duopoly experiments tend to have
more collusive outcomes than markets with more sellers, but more firms
does not necessarily eliminate joint-monopoly behavior. End game (in finite
games) and learning effects also influence market outcomes. Furthermore,
games with a high probability of continuing at each stage do not necessarily
result in the play of collusive trigger or tit-for-tat strategies. Despite trends
that are consistent with theory and intuition, experimental outcomes are
always subject to the peculiarities of individual players.
The experimental evidence also indicates that decision-makers may not
initially recognize the consequences of market actions. Behavior changes
Pricing in Competitive Electricity Markets 95

over time in repeated simulations. For example, many pricing experiments

reveal that new players often engage in fairly destructive and unprofitable
price wars in response to a price cut by one party. However, repeated
interactions show that individuals become more 'forward thinking' as they
learn of the potential dynamic consequences of certain actions.

4.2 Auctions

Large customers might not simply purchase electricity and other energy
services from the provider who offers the best posted-price value. Suppliers
may signal their value of these buyers by posing custom-tailored offers that
may be entertained by the customers before selecting the best overall bid.
Several types of auction formats or trading mechanisms might be used such
as a single round of sealed bids with the lowest bid winning the customer, a
single round with the second lowest bid winning the customer, or a
descending sequential auction.
The first method is perhaps the most popular in commercial settings such
as consulting and construction. If all other matters are equal, the low bidder
wins most of these kinds of contracts. Similarly, a large energy customer
might request sealed bids from various suppliers and choose the one offering
the best overall price. If bidders have independent, private assessments of
the value of obtaining the customer and also have perfect information about
all bidders' private valuations, the winning bid is equal to a fraction below
what the second-lowest bidder is willing to offer, and the winning bid is
profitable. Additionally, if the same valuation and information assumptions
are present with the exception that the winner gets the price offered by the
second lowest bidder, the price paid by the customer is the same as in the
case of the first-price auction.
The situation changes dramatically when uncertainty over the value of the
item exists. In a common value auction, all or most of the value of the item
being auctioned is common across all bidders, but each bidder has different
beliefs about the item's value. For example, several general contractors may
use the same labor pool to construct a building but each differs in the
assessment as to the length of the construction period. Retail electricity
providers will purchase power from a single wholesale market, but have
different assessments of future wholesale price levels and volatility. In some
instances the winning bidder is beset by the "winner's curse" and loses
money because his valuation overestimates the item's value.
Researchers have postulated that the critical element determining the
existence of a winner's curse is game experience. Kagel and Levin (1986)
conducted an experiment where the value of the auctioned item was drawn
from a known uniform distribution. Bidders were given an initial sum of
cash and allowed to earn more if they won and bid an amount less than the
96 Pricing In Competitive Electricity Markets

random value of the item, but would go bankrupt and be out of the game if
they lost their endowment via the winner's curse. After each round, each
bidder was allowed to observe all bids and learn from one another's
behavior. The results indicated that although the winner's curse was present
in early rounds, loss levels diminished with additional play. In addition,
overall profits were positive when there were two or three bidders, but were
negative when six or seven individuals were allowed to bid. Kagel (Kagel
and Roth 1995, Chapter 7) notes that the provision of bidding information-
both winning and losing bids-is a factor that allows players to reevaluate
bidding strategies in repeated auctions. Another important market element is
bankruptcy: as overly aggressive bidders are eventually forced out of the
game, the winner's curse diminishes.
Another factor that can reduce the winner's curse is to reduce uncertainty
about common values. For example, further development of wholesale
markets will provide better estimates of future costs and reduce uncertainty
as whole price volatility is revealed. Industry organizations such as EPRI
might play an important information clearinghouse role by performing
volatility analyses and other research to reduce valuation uncertainty.
Auction formats also influence the presence and size of a winner's curse.
Kagel and Levin (1991) find that an English auction format, where bids are
publicly announced, yields similar results as a series of sealed bid auctions
where bidder's valuations are announced after each round. Information in
the form of the losing bids in the English auction is analogous to the bid data
provided between rounds in the sequential sealed bid auctions.
A descending sequential auction would provide similar benefits in a
deregulated electricity markets. In this format, bidders provide sealed bids
in each round. If the bids are sealed, the amounts (but not necessarily who
bid what) are revealed before the next round begins. Bidding continues as
long as two or more bidders' offers are less than the low bid from the
previous round. This process essentially speeds up the learning process
(relative to a series of one-shot sealed bids with several customers) because
players obtain information in early and intermediate rounds without actually
winning the bid. The winner's curse is therefore reduced relative to standard
sealed bid auctions.
In summary, auction experiments offer another way to view rivals'
behavior in artificial markets. This strand of the experimental literature is
particularly relevant to the development of bidding strategies for large
customers. Experiments reveal that the winner's curse is reduced as
information increases and that the amount of information available to
bidders is dependent on the auction format.
Similarly, the evidence suggests that some winning auction bids suffer
from the "winners curse," because the winner's valuation overestimates the
item's true value or cost. The critical element determining the existence of a
Pricing in Competitive Electricity Markets 97

winner's curse, however, is game experience. Repeated auction experience

mitigates the potential downside of the winner's curse.

5. PRODUCT MIX MODEL EXPERIMENTS

We conducted three pricing experiments using EPR!' s Product Mix

Model (EPRI, 1998). The players in these experiments were participants in
Product Mix User Group meetings and workshops. All players are involved
in retail pricing decisions at their utilities.
The first two experiments had the same initial set up, rules and initial
number of teams. The first had eight two-person teams while the second had
eight one-person teams. In each round of the game, each team had to make a
product and price declaration. There were three products that could be
offered - a guaranteed price product, a spot price product, and a combination
product consisting of a forward contract plus a spot product. A team could
change products between rounds, but could offer only one product per round.
Elimination of teams occurred if they earned less than half the average
profit level for that round. The game concluded when either no elimination
occurred for two consecutive rounds or when only one team remained. At
the beginning of the game, the teams were informed of the monopoly (or
cartel) profit-maximizing price. Explicit collusion among teams was
forbidden and none was observed or detected.
Low margin products offered in the first round of each game set the tone
for all future rounds. In both experiments, the surviving teams offered low
margin products. As a result, total industry profits steadily declined in the
first few rounds. The first experiment lasted nine rounds with industry
profits stabilizing at 10 percent of the maximum possible after five rounds.
The second experiment lasted only six rounds with industry profits declining
until the fourth round. In the final two rounds, only two teams remained and
industry profit steadily increased to about 30 percent of the maximum
possible.
The first experiment concluded with three teams surviving, one offering a
spot-price product and two offering forward-plus-spot combinations. Th~
second experiment concluded with only two surviving teams, one offering a
spot product and the other offering a forward-plus-spot combination. The
non-survival of the guaranteed price product reflected the customers'
flexibility to respond to price changes coupled with the absence of customer
risk aversion.
The third experiment began with four teams with identical customer
bases. In Round 0, each team earned $1.7 million with a three cent
guaranteed price product in regulated and geographically isolated markets.
In Round 1, regulation and geographic barriers were removed and teams
98 Pricing In Competitive Electricity Markets

were free to offer anyone of three products in the consolidated open-access

market.
This third experiment introduced several important differences from the
previous two experiments. First, the customer was moderately risk averse.
This made the guaranteed price product more attractive, other things the
same. Second, the experiment began with only four two-person teams.
Third, the elimination rule was weakened so only teams whose loss in a
round exceeded their cumulative profits were eliminated. The loosening of
the elimination rule made pricing gambles more acceptable. Fourth, an
incumbency bias was introduced, indicating that a customer would tend to
stick with his or her current provider, even if a slightly lower price were
available elsewhere. If the customer switched, the incumbency bias applied
to the new provider in the next round. Finally, the game had no stated
conclusion and players did not know with which round the game would end.
As with the first two experiments, lower margin products set the tone of
the game. Two teams pursued an extreme form of buying market share by
pricing below cost. These two teams captured almost the entire market and
lost a lot of money. One of the pair survived because of prior profits, but the
other went bankrupt and was eliminated from the game. The "winners" of
the third round were the two teams that priced above cost. They had very
small market shares and small positive profits.
With the elimination of one of the four original teams, the industry profit
picture began to stabilize. Furthermore, the remaining three teams each
chose a different pricing product. When the game was halted after the fifth
round, the spot product had the lowest margin, the greatest market share and
the greatest profit. The guaranteed price product had the highest margin, the
lowest market share, but significantly greater profit than the much lower
margin combination product.
Generalizing from these simple experiments to real-world pricing
behavior is a tenuous endeavor. In fact, the absence of consistent patterns
may be the most important finding. However, four conclusions do seem to
be suggested by these experiments. First, it only takes one player pursuing a
low margin strategy to begin the process of competing away most of the
industry profits. Second, total industry profit eventually stabilizes when
only a few players remain. Third, after stabilizing, industry profits are only
a small fraction of the maximum possible. Fourth, the rules of elimination
appear to affect the pricing strategies. Looser elimination rules allow
players to experiment with both higher margin products and negative margin
products.
Pricing in Competitive Electricity Markets 99

6. CONCLUSIONS
Anticipating competitor response to pncmg strategies is receiving
increasing attention in the academic and business community. Simple cost-
plus pricing rules previously used in regulated and some non-regulated
industries are obsolete, as are those pricing strategies that focus too intently
on winning customers and ignore other aspects of profitability.
Firm interdependence has been long recognized by economists, and
insights from game theory are now being incorporated into the strategic
plans of corporations. Together, the analytical and experimental branches of
the theory provide a sound, logical framework to anticipate and understand
price dynamics in competitive electricity markets. The inherent flexibility of
output of each firm in retail electricity markets suggests that the industry be
modeled as competing on prices (Bertrand) as opposed to quantities
(Coumot). Price-choice experiments also explicitly capture the essential
mechanics of price determination from rivals' behavior, providing a valuable
training ground for pricing analysts.
An infinite number of outcomes might occur in retail electricity markets,
ranging from cutthroat price wars to a lack of price competition. In addition
to product, customer, and market characteristics, outcomes are dependent on
the rules of the competitive game: the number of periods, the sequence of
moves, market information, payoffs, and the players themselves. As
revealed by numerous experiments, competitive dynamics in each situation
are unique. It is important to recognize that each game is different and that
the predictive power of simulation results is limited. The great value of
simulating market behavior will not be derived from results per se, but will
instead be derived from the learning process or the play of the game.

REFERENCES

Alger, D., "Laboratory Tests of Equilibrium Predictions with Disequilibrium Data,"

Review of Economic Studies, Vol. 54 (1987). pp. 105-145.

Beil, R.O .• "Collusive Behavior in Experimental Oligopoly Markets" (working paper,

Texas A&M University, 1988).

Bierman, H.S., and L. Fernandez, Game Theory with Economic Applications (Addison-
Wesley, 1998).

Bertrand, J., "Theorie Mathematique de la Richesse Sociale," Journal des Savants, (1883)
pp.499-508.

Binger, B.R., E. Hoffman, G.D. Libecap, and K.M. Shachat, "An Experimental Study of
the Cournot Theory and Firm Behavior" (working paper, University of Arizona, 1990).
100 Pricing In Competitive Electricity Markets

Cournot, A., Recherches sur les Principes Mathematiques de La Theorie des Richesses
[Researches into the Mathematical Principles of the Theory of Wealth] (Paris: Hachette,
1838).

Davis, D.O., and C.A. Holt, "Market Power and Mergers in Laboratory Markets with
Posted Prices," Journal of Economics, Vol. 20 (1994), pp. 467-487.

Davis, D.O., c.A. Holt, and A.P. Villamil, "Supracompetitive Prices and Market Power in
Posted-Offer Experiments" (working paper, University of Illinois, 1990).

EPRI, Product Mix ModeL: Version O.2{J, Palo Alto, California (February 1998).

EPRI, Anticipating Competitor Response in Retail Electricity Price Design, TR-112087,

Palo Alto, California (April 1999).

Fouraker, L.E., and S. Siegel, Bargaining Behavior (New York: McGraw Hill, 1963).

Fudenberg, D., and J. Tirole, Game Theory (Cambridge: The MIT Press, 1991).

Holt, C.A., "Industrial Organization: A Survey of Laboratory Research," in Kagel, J.H.

and A. V. Roth, eds., The Handbook of Experimental Economics (1995).

Holt, C.A., "An Experimental Test of the Consistent Conjectures Hypothesis," American
Economic Review, Vol. 75, pp. 314-25.

Kagel, J.H., "Auctions: A Survey of Experimental Research," in Kagel, J.H. and A.V.
Roth, eds., The Handbook of Experimental Economics (1995).

Kagel, J.H., and D. Levin, "Revenue Raising and Information Processing in English
Common-Value Auctions" Working Paper, University of Pittsburgh (1991).

Kagel, J.H., and D. Levin, "The Winner's Curse and Public Information in Common
Value Auctions," American Economic Review, Vol. 76 (1986), pp. 894-920.

Kagel, J.H. and A.V. Roth, eds., The Handbook of Experimental Economics (New Jersey:
Princeton University Press, 1995).

Moorthy, K. S., "Using Game Theory to Model Competition," Journal of Marketing

Research, Vol. 22 (August 1985), pp. 262-282.

Murphy, lL., "Effects of the Threat of Losses on Duopoly Bargaining," Quarterly

Journal of Economics, Vol. 80 (1966) pp. 296-313.

Neumann, John von and Oscar Morgenstern, The Theory of Games and Economic
Behavior (Princeton: Princeton University Press, 1944).

Shaked, A., and J. Sutton, "Relaxing Price Competition Through Product

Differentiation," Review of Economic Studies, Vol. 49 (1982), pp. 3-13.

Selten, R, and R. Stoecker, "End Behavior in Sequences of Finite Prisoners' Dilemma

Supergames: A Learning Theory Approach," Journal of Economic Behavior and
Organization, Vol. 7 (1986), pp. 47-70.
Pricing in Competitive Electricity Markets 101

Stackelberg, H. Yon, Marktform und Gleichewicht (Vienna: Julius Springer, 1934).

Stoecker, R. Experimentelle Untersuchung des Entscheidungsverhaltens im Bertrund-

O/igopol, Yol. 4 of Wirtshaftstheoretische Entscheidungsforschung (Bielefeld: Pfeffersche
Buchhandlung, 1980).

Tirole, 1., The Theory of Industrial Organization (Cambridge: MIT Press, 1998).
Chapter 7

Understanding Latent Market Power in the Electricity

Pool of England and Wales

Derek W. Bunn, Christopher Day and Kiriakos Vlahos

London Business School. u.K.; University of California. Berkeley; and London Business School.
u.K..

Key words: Competition; Electricity; Herfindahl Index; Market Power Concentration;

Restructuring.

Abstract: Conventional approaches to assessing market power often refer to market

concentration. for which the Herfindahl index is a generally applied measure.
However. in looking at competitive markets for electricity generation. based
upon marginal time-of-day bid prices. as in the electricity pool of England and
Wales. we find that market concentration is an insufficient measure and can
underestimate actual market power. Using a plant-by-plant industry
simulation model. extra insights can be gained with respect to the potential for
supranormal profits and the circumstances upon which they depend. In
particular. the conventional wisdom that an industry structure with a 20
percent upper bound on individual market share seems generous given the
special price-setting properties of the market rules and a target closer to 12
percent of the potential price-setting plant might be needed to ensure efficient
competition.

1. INTRODUCTION

The creation of daily markets for electricity is becoming a common

ingredient in the extensive restructuring of energy utilities, which we have
seen gather pace worldwide in the 1990's. Motivated simultaneously by the
desire to create competitive marketplaces for wholesale electricity, and to
facilitate efficient time-of-use pricing responses by customers, the pricing
mechanisms of the new electricity pools are invariably based upon the
marginal bid price offered by the generators. In the electricity pool of
England and Wales, for example, from 1990-1999, price has been set half-
A. Faruqui et al. (eds.), Pricing in Competitive Electricity Markets
© Kluwer Academic Publishers 2000
104 Pricing In Competitive Electricity Markets

hourly, from a uniform day-ahead auction, as that of the marginal plant

which was scheduled to run at each period. All plants scheduled to run at
that time were then paid at this price. Other pools have been designed upon
hourly prices (e.g., California, Colombia) and may have different lead times
in publishing the prices (e.g., Norway) or may even be ex post (e.g.,
Victoria). However, the combination of competition between generators
submitting bids and a marginal price-setting mechanism has been a recurring
and core feature in most market designs.
One of the inevitable questions that this design has been raising is its
vulnerability to market power by large generators owning portfolios of plant.
Since the supply function for electricity costs is steeply increasing, a large
generator that holds a spectrum of plant may have considerable influence in
price-setting at the margin, and reap the benefits through the profits so-
created on its baseload plant. The susceptibility of competitive electricity
pools to this activity clearly depends upon market concentration; if all plants
are separately owned, then open competition will lead to an efficient market
and prices close to marginal cost. However, economic analyses of oligopoly
in such markets (e.g., Green and Newbery, 1992; von der Fehr and Harbord,
1993; Borenstein et aI, 1996) have suggested that prices above marginal cost
could result, depending upon the number of generators actively competing at
the margin.
In the British context, the pool suffered repeated criticism for being a
duopoly, with only two generators, National Power and Powergen, initially
owning most of the marginal plant. Ever since the first official pool price
review (OFFER, 1991), the regulatory office recognized the ability of these
generators to influence pool prices at will, so much that by 1994, an average
price cap and a commitment to the divestment of 6GW of their plant (about
17 percent of their capacities) had to be imposed (OFFER, 1994). The fact
that such a price target could be met over the subsequent two year period
was testimony by itself to their market power. By 1996, however, with the
rising market share of new independent power plants, and the divestment of
the 6GW to Eastern Electric, a distribution company which already had
some generating plant, an official inquiry into the acquisition by National
Power and Powergen of two distribution companies, looked at the issue of
their market power and concluded that their " ... ability to influence prices
over a sustained period will be small" (MMC, 1996). Nevertheless, the
Government rejected the recommendations of that inquiry, implicitly
concerned about market power, by alluding to the need for competition to
become more fully established (see Financial Times, 1996). Furthermore, in
evidence to a government inquiry shortly afterwards, the Director General of
Electricity Supply stated that he still was " ... not satisfied with the present
extent of competition either in the Pool or in generation" (Littlechild, 1997a)
Pricing in Competitive Electricity Markets 105

and indeed further price controls were then being considered to coincide
with the full liberalization of the retail market (Littlechild, 1997b). By 1998,
the change of Government brought a new initiative to reduce prices to
consumers, by reforming some aspects of the market mechanism and making
a further attempt to erode the market power of the generators. Thus, in 1999,
National Power and Powergen were each making plans to divest a further
4GW (about 45 percent of their capacities).
All of this raises the question of how market power should be evaluated,
in contexts such as this, as it is evidently such a crucial factor limiting the
pace of regulatory liberalization. Market shares have usually been the
starting point in looking at this issue, and when the industry was initially
restructured in 1990, with generating capacity being split approximately 50
percent, 30 percent, 20 percent between National Power, Powergen and
Nuclear Electric, respectively, most analysts immediately criticized this
concentration. Suggestions of an equal split between 5 companies had been
widely held (e.g., Henney, 1987; Robinson, 1988). This conventional
wisdom has persisted within the industry and seems to pervade the rationale
for worldwide unbundling, (e.g., Benavides, 1996). But subjectivity in
defining an acceptable market share is unsatisfactory. Regulatory confusion
was created in 1996 when the u.K. government allowed Eastern Electric to
become vertically integrated with a generating market share of 11 percent,
but blocked Powergen's similar aspirations, having a generating market
share of 17 percent (Financial Times, 1996).
The Herfindahl index, defined as the sum of the squared market shares in
an industry (cf. Stigler, 1968; Borenstein, 1996), is a common measure of
concentration and has been used in the u.K. electricity context (MMC,
1996). However, again the issue remains as to what is an acceptable level.
If percentage market shares are used, then the index has a maximum of
10,000, and a guideline of 1,800 is used by the U.S. Department of Justice
for merger analysis. In the u.K., it was about 2,000 in 1995, and its fall
from 3,000 five years earlier is one of the reasons that the MMC took a
benign view of market power. It has also been used by the u.K. Office of
Electricity Regulation to justify the level of divestment imposed upon
National Power and PowerGen in 1996, and in the U.S. concerning the
restructuring of the California utilities (Joskow et ai, 1996). In the u.K.
context, referring to the 6GW transfer of power stations to Eastern,
Littlechild (1996) asserts that these, "take the index to 1,600, equivalent to
about six equal size firms. So, on this basis competition has roughly
doubled" [since 1990]. However, the Herfindahl index can be misleading if
the competition does not conform to the simple assumptions of Cournot
competition, especially for the low demand elasticity levels that characterize
short-term electricity markets (Borenstein, 1996). Indeed, even the U.S.
106 Pricing In Competitive Electricity Markets

Department of Justice suggests that it should be interpreted as an incremental

measure to assess specific mergers/de mergers rather than an absolute
statement of the competitive status of the industry.
In our specific electricity pool context, the problem is compounded by the
market rules of price setting, and for this reason, the percentage of time that
a company sets the marginal plant has been looked at carefully in the u.K.
(OFFER, 1994, MMC, 1996). Thus, in 1995/96, National Power with 34
percent of the capacity, set the pool price about 50 percent of the time.
However, by itself, frequency of price setting does not indicate excessive
market power, just the cost and role of the plant involved. For example, in
1996, 15 percent of the time the price was set by the 1.8 GW pumped storage
facility, which is the only plant owned by First Hydro, but one that is
generally expected to run mostly at the margin.
Empirical attempts to monitor market power often involve comparing
market prices with estimates of marginal costs. For the England and Wales
market, Brealey and Lapuerta (1997) undertook one of the most thorough
analyses of this sort and suggested that the dominant generators were
bidding at about 50 percent above short-run marginal costs. Similarly,
Borenstein, Bushnell and Wolak (1999) have suggested that prices in the
California power exchange are 15 percent above competitive levels.
Analyses like this are not easy to undertake. Estimating a reasonable
competitive baseline involves not only assessments of fuel and thermal
efficiency parameters (apart from the problems of evaluating hydro
resources in some systems), but also the issues of how to incorporate no load
and start-up costs (which can add about 20 percent; Kahn, 1998) and make
reasonable provisions for availability. Furthermore, whilst it can be useful
for retrospective monitoring, it does not give an indication of the potential
which might exist within the market for the abuse of market power. An
understanding of this potential is sometimes important in suggesting how
much regulatory oversight needs to be maintained.
Analytical evaluations of potential market power at a stylized industry-
wide level, based upon theories of supply-function equilibria (Green and
Newbery, 1992: Newbery, 1995) have looked at the "deadweight loss" to
welfare which may result from the market failing to clear at marginal cost.
Again, the suggestion of five companies, actively competing at the margin,
has been suggested as being required to reduce market power to an
acceptable level. However, the analytical assumptions of differentiability,
demand elasticity and linearity have been criticized as being quite restrictive
(von der Fehr and Harbord, 1993). Moreover, while analysis of deadweight
loss can provide a useful regulatory perspective, it is a very indirect measure
of market power.
Pricing in Competitive Electricity Markets 107

In this paper, we take a more direct, market simulation approach. We

interpret market power quite explicitly in terms of the potential supranormal
profit that a generator can obtain through increasing bid prices across some
or all of its plant. In this respect, we are able to address the more pragmatic
of the U.S. Department of Justice's guidelines for competition, namely that a
player should not have the power to raise market prices more than 5 percent
above marginal cost. Using a plant-by plant model of the full system in
England and Wales in 1995, and averaging the results across a full year
consisting of several seasons, a number of simulations are undertaken. The
first series is a stylized analysis of hypothetical companies to investigate
how profitability, market share and the Herfindahl index relate to each other
as the number of equivalent competing companies increases from one to
eight. Just how much market power (in terms of the ability to raise prices,
increase revenue and profits), is associated with different level of market
concentration (Herfindahl indices) is the open question in the generalized
context of this type of market.
The second series of simulations then applies this approach to an analysis
of the market power debate that surrounded National Power and Powergen in
1996, and which lead the 6GW divestment. There is no doubt that these
generators have exercised market power, as various commentaries on the
history of the u.K. Electricity pool noted (Newbery, 1995; EPRI, 1996), but
it is also accepted that they have not exercised it to the full. Our analysis
therefore reflects market power as a latent property insofar as it analyses the
potential for profits in the short term. It is not an empirical commentary on
whether and to what extent such power has been exploited, and indeed the
sustainability of exercising to the full any latent market power is a strategic
and regulatory issue open for debate. Some comments on that are offered in
the final section.

2. SIMULATING MARKET POWER WITH

STYLISED MARKET CONCENTRATIONS

The first series of experiments is grounded in a detailed model of the

electricity pool of England and Wales, but under a number of stylized
scenarios of market concentration and bidding behavior. The electricity
industry of England and Wales was unbundled in 1990 in a way that is also
beginning to occur in many other parts of the world. Generation is split from
transmission, and distribution has had to separate its local monopoly in the
physical distribution of electricity from the commercial activity of retailing
power to end-users. All generating plant, except for a small amount of old
nuclear (7 percent of the market in 1996), is now privately owned.
108 Pricing In Competitive Electricity Markets

Transmission and distribution are regulated by price, while generation and

retail supply are increasingly competitive. At the beginning of 1996, the
market shares of generation were National Power (32 percent), PowerGen
(24 percent), nuclear (23 percent in total, split between British Energy and
Magnox Electric) with the remaining generation being spread across a
number of independents and imports from France and Scotland. The major
structural change in 1996 was the 6GW plant transfer (from National Power
and Powergen) to Eastern (initially a distribution and retail supply
company), raising its generating market share to 11 percent.
The National Grid Company (NGC) runs the transmission business and is
responsible for operating the daily "power pool," a day-ahead forward
market place for setting the half-hourly prices. Thus, each morning, NGC
invites bids from generators for the price of each generating unit and its
availability to supply power over the next twenty-four hours. These are
matched against the demand forecast to produce a least cost, half-hourly
generation schedule. The price per half-hour is fixed at that of the marginal
unit scheduled for that period, (i.e., the System Marginal Price [SMP]).
These prices, for the next day, are available at 4:00 P.M. and published in
the newspapers the following morning. To the extent that a generating unit's
actual marginal costs are below the bid price of the marginal unit, then that
unit will profit from the difference on the energy supplied.
In addition, as an incentive to the generators to provide sufficient
capacity to the system, an extra "availability payment" is added to SMP, so
that all the generators who are scheduled day-ahead receive the Pool
Purchase Price (PPP) determined (half hourly) as:

PPP = SMP + LOLP(VLL - SMP) with

VLL: Value of Loss of Load (initially set at £2IKWh)2
LOLP: Loss of Load Probability per half-hour

The Loss of Load Probability is computed half-hourly, a day in advance,

taking account of demand uncertainty and the probabilistic reliability of
individual plant in meeting the load as planned. While the SMP mechanism
is becoming quite widespread in other countries, the LOLP formula has been
quite idiosyncratic to the u.K. For this reason, we will not incorporate it
into this analysis, although it has presented large generators with another
opportunity for the exercise of market power (Newbery, 1995). On the day,
there are several other payments to generators for changes to the schedule
and "ancillary services", such as spinning reserve, congestion, frequency and
voltage control, comprising what is referred to as "Uplift". This is passed on
Pricing in Competitive Electricity Markets 109

to the demand-side of the pool, so that the Pool Selling Price consists of PPP
plus Uplift.
In this study, however, we are concentrating upon market power in
setting SMP. One approach to modeling SMP, which seems to be quite
often used in the industry, is to stack all the plants in the system according to
increasing marginal cost (or by actual bids) to provide a generators' supply
function. Figure 1, which is a representative bidding function made
available by NGC (1995), can be interpreted as indicating what SMP would
be for any given demand level.

Merit Order - NGC data

lor-----,------,--~~~~~~~==~,------,,----.

6
~
Bid Price
(p/KWh)
4

-
2

o~--~--~-----~--~----~--~--~j
o 10000 20000 30000 40000 50000 60000 70000
Plant Capacity in merit order (MW)

Figure 1. NGC Representative Bidding Function

Notice the llGW of nuclear plant bidding in at zero, to be certain of

running at baseload; in contrast to the very high bid prices of the open cycle
gas turbine peaking plant corning in above 60GW. Normal winter peaks are
about 48GW. This is a typical bidding profile that clearly varies throughout
the year according to the availability of plant, and strategically according to
the bidding policies of the generators. We will use this function in the next
section when we look explicitly at the latent market power in the 1996
division of plant between the main generators. In this section, however, it is
useful to see how market power and market concentration are related, given
radically different dispersions of plant into companies. We have also
formulated as a baseline, a marginal cost reflective bidding function, Figure
110 Pricing In Competitive Electricity Markets

2, on the basis of previous work (Bunn, Larsen and Vlahos, 1994) updated
with recent data (Electricity Association, 1995). This estimate of marginal
costs produces a function that is clearly flatter in mid merit, and assigns
estimated costs to the "must run" and other baseload plant that, as Figure 1,
tend to bid into the market at zero. We used this marginal cost function as a
baseline in the first series of simulations.

-----,-----
Merit Order
I
- Marginal
I
Cost dataI

I
~

J
Bid Price I
(p/KWh) ~
4 I
I

2~
i~
oI I I
o 10000 20000 30000 40000 50000 60000 70000
Plant Capacity in merit order (MW)

Figure 2. Marginal Cost Bidding Function

We then simulated a year's operation of pool price behavior, with demand

profiles for seven separate seasons under different conditions of market
concentration. The starting point for understanding the economic
implications of these strongly convex supply functions in a marginal price
setting mechanism, is to consider the situation of increasing concentration in
which progressively more of the plant is owned by one company, and this
company is free to bid up its plant. We therefore, first of all envisage the
stock of generating plant split with one company who operating several
plants (the "portfolio generator") and the rest of the plant operating
individually. Furthermore we allow this "portfolio generator" to
progressively own more of the plant, starting from the baseload and working
up the merit order. Figure 3 shows the percentage change in an operating
Pricing in Competitive Electricity Markets III

profit index from pool trading as this artificial company, which progressively
owns more of the plant on the capacity axis, from the lowest cost upwards,
increases the bid price of all its plant by various percentages. Clearly when
the portfolio generator just owns the baseload, this increase in bid price
makes no difference, as long as the base load plant does not set SMP. Figure
4 shows the associated SMPs. Finally, when all the plant is owned and bid
up, does the percentage increase in SMP match that of the bid mark-ups.
Evidently there is no pool-based market power in just owning baseload\
which in this case amounts to about 25 percent of the market.

300

250

200 Company Bid Up 10 %

I Company Bid Up 20 %
I Company Bid Up 30 %
Company Bid Up 40 %
150 ~
% Change I
in Profit l_ /'---

100 i

I
50 I -- -
--...;::----- -----
o t-'-----'------_:-~~--'---·::
:-~~;~./~,f
~.-'"- ~
o 10000 20000 30000 40000 50000 60000
Plant Capacity in merit order

Figure 3. Profitability of Bidding-Up

The slight dip in profitability observable around 20GW reflects the drop
in market share that would follow bidding up some mid-merit plant. This
displacement of marginal plant is more evident in Figure 5. Above baseload,
it is only in the monopolistic situation of owning all the plant that a drop in
generation through bidding up is restored. The revenue plot in Figure 6
reinforces this insight. Overall, this first series of experiments was just to
reinforce our intuition on the trade-off between increasing profitability and
losing market share, if a portfolio generator is tempted to increase its bid
mark-up across its stock of plant.
112 Pricing In Competitive Electricity Markets

:: I,I---'-~--.-----'-----.-----r --,----1
!

2-41
Company
Company
Bid Up 10 %
Bid Up 20 % 1
l
Company Bid Up 30 %
Company Bid Up 40 %
Avg. Annual2 .3
Demand
l
Weighted SMf.2 /- ----
(p/KWh) /

2.1
L ~/
/

2~
l.91

l.8
I
o 10000 20000 30000 40000 50000 60000
Plant Capacity in merit order

Figure 4. SMP Effects of Bidding Up

" \

-40 I
% Change in
Generation -60 Company
Company
Bid
Bid
Up 10 %
Up 20 %
~
I
Company Bid Up 30 %

l'
Company Bid Up 40 %
-80

-1 00 ~,--------------------,-------------------,--,--_----"-- ______________________________ ~-J

o 10000 20000 30000 40000 50000 60000
Plant Capacity in merit order

Figure 5. Output Effects of Bidding-Up

Pricing in Competitive Electricity Markets 113

I
40

20

o
-20
% Change
in Revenu~40

-60 Company Bid Up 10 %

Company Bid Up 20 %
Company Bid Up 30 %
Company Bid Up 40 %

10000 20000 30000 40000 50000 60000

Plant Capacity in merit order

Figure 6. Revenue of Bidding-Up

Table i: Percentage Change in Generation (for marginal cost merit order)

Percentage of Bid Increase

No Companies 10% 20% 30% 40%
I 0.00 0.00 0.00 0.00
2 -21.08 -31.69 -43.71 -49.45
3 -30.10 -45.55 -62.32 -69.07
4 -34.55 -52.15 -70.38 -76.10
5 -36.71 -55.37 -75.37 -81.78
6 -36.98 -56.27 -75.60 -80.73
7 -36.51 -58.74 -75.02 -80.95
8 -36.62 -61.55 -78.11 -84.85

The next set of simulations distributed the non-base-Ioad plant evenly

across an increasing number of companies. Thus, for example, with four
non-base-Ioad companies (in addition to the 20 percent owned by nuclear),
each would have 20 percent of the plant and a similar cross section of
marginal costs. Tables 1 through 4 summarize the effects of one company
114 Pricing In Competitive Electricity Markets

out of the set bidding up all its plant. For example, in Table 1, if there are
four similar non-base-Ioad companies and one bids up all its plant by 10
percent, it would lose 34 percent of its generation, but it would still be more
profitable by 3 percent (Table 2) and increase SMP for the benefit of all the
others by 2.2 percent (Table 4). Moreover, this will not be at the expense of
the other companies as they collectively gain the extra market share and
benefit from the higher SMP. The market mechanism induces benign
collusion with any generator who is able and willing to increase profitability
by bidding up. We will return to the strategic issue of willingness later.

Table 2. Percentage Change in Profit Contribution (for marginal cost merit order)

Percentage of Bid Increase

No Companies 10% 200/< 30% 40%
1 72.76 144.61 216.20 289.43
2 24.16 43.86 54.31 70.45
3 10.02 12.45 -0.06 -3.06
4 3.01 -3.69 -25.05 -30.93
5 -0.61 -12.56 -38.96 -49.91
6 -2.58 -15.57 -42.36 -51.96
7 -2.64 -21.41 -44.33 -55.98
8 -6.22 -26.83 -48.73 -61.88

Table 3. Percentage Change in Revenue (for marginal cost merit order)

Percentage of Bid Increase
No Companies 10% 20% 30% 40%
1 10.07 20.02 29.93 40.07
2 -16.47 -23.53 -32.94 -35.56
3 -26.64 -40.05 -56.41 -62.29
4 -31.58 -48.22 -66.82 -72.39
5 -34.12 -52.16 -72.71 -79.47
6 -34.56 -53.34 -73.39 -78.86
7 -34.26 -55.92 -72.96 -79.27
8 -34.61 -58.98 -75.95 -83.18
 Pricing in Competitive Electricity Markets 115

Table 4. Percentage Change in SMP (for marginal cost merit order)

Percentage of Bid Increase

No Companies 10% 20% 30% 40%
1 10.10 20.00 29.91 40.04
2 4.70 10.49 16.68 22.95
3 3.00 7.01 10.81 14.39
4 2.21 4.74 7.10 9.32
5 1.64 3.70 5.28 6.56
6 1.38 3.00 4.07 4.81
7 1.21 2.50 3.41 4.06
8 1.01 2.23 3.03 3.52

Table 5. The Herfindahl Index (non-nuclear companies)

No Companies Herfindahl Index (Generation)

1 0.64
2 0.35
3 0.25
4 0.20
5 0.18
6 0.16
7 0.14
8 0.13

One might also expect that such latent market power depends upon the
convexity of the aggregate supply function bid into the market by all the
generators. Indeed this seems to be the case. Tables 6 and 7 replicate Tables
1 and 2, but on the basis of the more convex representative bidding function
of Figure 1. Even with eight non-baseload companies, it is more profitable
for one to bid up, and there is a much smaller drop in market share than in
the less convex supply function case. Thus, whether a particular value of the
Herfindahl index, 1,000 or 2,000, suggests market power depends very much
upon the actual shape of the supply function.
 116 Pricing In Competitive Electricity Markets

Table 6. Percentage Change in Generation (using NGC)

Percentage of Bid Increase

No Companies 10% 20% 30% 40%
1 0.00 0.00 0.00 0.00
2 -9.47 -14.82 -21.95 -25.69
3 -1 \.12 -18.53 -26.35 -30.41
4 -15.11 -21.92 -30.44 -35.92
5 -14.99 -22.12 -29.33 -36.16
6 -14.99 -23.81 -30.66 -35.70
7 -15.39 -23.97 -30.86 -36.90
8 -17.48 -26.04 -33.51 -40.37

Clearly, one way to reduce latent market power is to encourage the supply
function to become "flatter," (i.e., less convex). In this respect, it is perhaps
curious that when the price-cap was introduced in 1994 (OFFER, 1994), it
stipulated both time-weighted and demand-weighted average annual price
targets, thereby signaling permission to increase the convexity of the supply
function. This was meant to encourage the building of a new peaking plant,
but, rather than reducing market power by increasing the tendency to cost-
reflective bidding, OFFER actually increased the profitability that could
result from potential bid-price manipulation (compare Table 7).

Table 7. Percentage Change in Profit Contribution (using NGC data)

Percentage of Bid Increase

No Companies 10% 20% 30% 40%
I 24.44 49.00 73.41 97.52
2 10.43 19.59 23.90 27.67
3 7.07 12.18 11.64 11.79
4 4.32 7.61 6.02 4.22
5 3.23 5.26 4.25 1.20
6 2.48 2.64 \.30 -1.02
7 1.71 \.31 -1.38 -5.23
8 1.38 0.21 -2.28 -8.35
Pricing in Competitive Electricity Markets 117

3. LATENT MARKET POWER IN THE UNITED

KINGDOM (U.K.) DUOPOLY

Figure 7 shows a similar analysis based upon the actual ownership of

plant by the companies operating in 1996. If the largest generator, National
Power, were to bid up by lO percent successively more of its plant, starting
from the top (peaking plant) and working down, the increase in notional
operating profits from pool trading would be as shown. Clearly it has
market power to increase its profitability, but it has to increase bids on all its
plant to do relatively better than PowerGen, who benefits quite considerably
from National Power's action. This suggests that only in peak periods when
almost all their plant will be scheduled, would NP profit relatively more than
PG. The situation for PowerGen, Figure 8, is not symmetrical. While it has
the market power to increase profitability, because it is smaller than National
Power, it gives NP even greater profits, especially in peak periods when all
their plant is bid up.
To the extent, therefore, that relative market share and comparative
financial performance are concerns, they will tend to curb the attraction of
exercising market power for short-term profitability. Figures 9 and lO show
the situation after the divestment of 6GW of plant during the summer of
1996. It has made very little difference to the market power of PowerGen,
but has eroded the attraction to National Power of exercising its market
power. Although both still have the power to increase their own profitability
at will, both now loose out in relative terms. The duopoly still has
considerable latent market power, but it is in the interest of each player for
the other to exercise it. Again we see the situation encouraging benign
collusion in the exercise of market power. Tables 8 and 9 reveal the
situation in more detail. Four scenarios for NP and PG each bidding up all
of their plant by 10 percent are summarized; Case 1 using the marginal cost
bidding function before divestment, Case 2 after divestment; Case 3 using
the NGC representative bidding function before divestment, Case 4 after
divestment. In a sense, Cases 3 & 4 represent a baseline where the
generators have already exercised market power in moving from the
marginal supply functions of Cases 1 & 2 (note the 20 percent difference in
the SMP base). Comparing Cases 1 and 3, there are clearly decreasing
returns to the exercise of market power, in the relative attractiveness to NP's
exercise is less in Case 3. Alternatively, even in Case 4, there is latent
market power that could be an attractive option to both of them. Overall,
these tables give a useful annual average view as they take into account the
seasonal pattern of demand.
118 Pricing In Competitive Electricity Markets

National
I
Power
I
Bids ___Up
"____ 1
10% Before Divestment
I
45
I
40

35

30

25 f--
r- NatIOnal Power
PowerGen
British Energy
1
% Change in
Profit 20

15

10
\
5- \,
I
Ok------L-----~------~----~-----~-~.~.'~.~--~~-L----~
~,

o 10000 20000 30000 40000 50000 60000 70000

Plant Capacity in merit order

Figure 7. Profitability of NP Bidding-Up

National Power
25 PowerGen
% Change British Energy
in Profi~W

15

10

5r-
\.
"-
-----,
\
,
I
I \

O------~------~------~-------L------~--~'~·~~--~----~

o 10000 20000 30000 40000 50000 60000 70000

Plant Capacity in merit order

Figure 8. Profitability of PO Bidding-Up

Pricing in Competitive Electricity Markets 119

National Power Bids Up 10% After Divestment

I
--j

l
\
25 \

20
~~~
N'tio",1 Pow" - \ - - - :\
15 PowerGen \
% Change British Energy \
in Profit
10

\
\
5 \

o~----~------~------L-----~------~~~~~----~

o 10000 20000 30000 40000 50000 60000 70000

Figure 9. Profitability of NP Bidding-Up after Divestment

30

25
\ I
20 National Power
% Change PowerGen
British Energy
in Profit
15

10

5 - - --- ---

oL-----~----~------L-----~--~~~~--L---~

o 10000 20000 30000 40000 50000 60000 70000

Plant Capacity in merit order

Figure 10. Profitability of PG Bidding-Up after Divestment

120 Pricing In Competitive Electricity Markets

Table 8. Profit Contribution

National PowerGen Nuclear SMP SMP

Power Electric base
% chg % chg %chg
NP Bids up 10% (1) 42.42 42.36 8.31 1.91 1.81
PG Bids up 10% (I) 41.45 10.92 4.90 1.87 1.81
NP Bids up 10% (2) 20.90 26.88 5.30 1.88 1.81
PG Bids up 10% (2) 32.67 8.21 4.36 1.86 1.81
NP Bids up 10% (3) 14.73 24.88 4.89 2.28 2.17
PG Bids up 10% (3 ) 15.18 14.62 4.88 2.27 2.17
NP Bids up 10% (4) 7.81 14.65 2.71 2.23 2.17
PG Bids up 10% (4) 11.85 12.66 4.60 2.26 2.17

( I ) Using the marginal cost bidding function

(2) Using the marginal cost bidding function, after both NP and PG have divested plant.
(3 ) Using the NGC representative bidding function
(4) Using the NGC representative bidding function, after both NP and PG have divested
plant

Table 9. Generation

National Power PowerGen Nuclear Electric

%chg Abs % chg Abs % chg Abs

NP Bids up 10% (I) -16.77 -12.38 18.11 12.10 0.00 0.00
PG Bids up 10% (I) 29.88 22.06 -33.77 -22.56 0.00 0.00
NP Bids up 10% (2) -26.82 -15.59 13.80 8.99 0.00 0.00
PG Bids up 10% (2) 24.87 14.46 -32.30 -21.04 0.00 0.00
NP Bids up 10% (3 ) -9.04 -7.12 11.66 7.02 0.00 0.00
PG Bids up 10% (3) 10.99 8.66 -14.50 -8.73 0.00 0.00
NP Bids up 10% (4) -4.13 -2.60 2.37 1.41 0.00 0.00
PG Bids up 10% (4) 6.93 4.36 -15.74 -9.41 0.00 0.00

(I) Using the marginal cost bidding function

(2) Using the marginal cost bidding function, after both NP and PG have divested plant.
(3 ) Using the NGC bidding function
(4) Using the NGC bidding function, after both NP and PG have divested plant.
Pricing in Competitive Electricity Markets 121

4. CONCLUDING COMMENTS

There are a number of virtues in taking a industry simulation approach to

analyzing market power. It allows the more direct measure of profitability
per price mark-up, rather than indices of market share or market
concentration, to be used as a basis of evaluation. The Herfindahl index of
market concentration clearly has to be used with care for electricity markets
of the u.K.type. Even for what are usually considered low values, around
1600, we can see significant market power. The diversity of plant owned is
as important as the amount owned. We see that 25 percent of baseload
nuclear does not give market power, whereas 25 percent of mid-merit plant
gives considerable potential. In particular, conventional wisdom that five
companies are sufficient for market efficiency may be an underestimate in as
diverse a system as the u.K. It may be that only when no company has more
than about 12 percent (i.e., eight or more companies) of a diversity of base
and price-setting plant, could the market be operating efficiently. The
generalization of this to other contexts depends upon the cost and technology
mix of the plant. As we have observed, the more convex and steeply
increasing the marginal cost function, the more market power there is for a
given amount of concentration, and this convexity is created by the stock of
plant having different marginal costs. However, once market power exists,
there is the potential for this function to become more convex through
strategic bidding.
The analytical approach taken here is close to how one might expect the
day-to-day bidding tactics to be undertaken by the major players. They
know the history of bids and have models of how NGC will schedule plant.
They will inevitably simulate the payoff from variations in bidding tactics.
It is unlikely that short-term demand elasticity will have been considered,
especially as it was not even part of the demand forecasting model that NGC
actually used to set SMP. Its reliable estimation has been elusive in the u.K.
and observations from NGC suggest that it is only for SMP changes several
orders of magnitude greater than the SMP scenarios investigated here that a
demand-side response had been factored into the day-ahead forecast. ii
While investigations of latent market power might therefore appear more
pragmatic than Coumot-based formulations of oligopolistic equilibria, they
do leave unanswered the implications of any subsequent multistage
balancing reactions to the exercise of latent market power, repeated gaming,
and indeed the existence of stable solutions. On the latter point, it might
reasonably be observed that the evolution of the this market in the u.K. can
be better characterized as a discontinuous process of structural changes, and
as such the evaluation of the potentials inherent in each state of disequilibria
is a more realistic form of analysis. The approach evidently provides
122 Pricing In Competitive Electricity Markets

complementary insights to the more aggregate, longer-term theoretical

studies of imperfect competition.
Being model-based, therefore, the approach facilitates analysis of what
could happen, rather than what has happened or what should happen. Latent
market power is an important issue in understanding the strategic forces that
could influence the market. In terms of the factors that could mitigate or
preclude the exercise of latent market power, we can identify: (1) short-term
competitive reactions, (2) longer term market-balancing tendencies (i.e.,
market entry) and (3) regulatory threat. We have seen that in terms of
profitability, the exercise of market power can be advantageous both to the
instigating player and to the rest of the generators. This can induce benign
collusion. However, the loss of relative market share will be the major
inhibitor here even if it is profitable. Longer term considerations of demand
elasticity, substitution away from electricity, and attracting new entrants are
less likely to moderate behavior than shorter term regulatory pressure. In the
Pool of England and Wales, it does indeed seem to have been the threat of
regulatory response that has moderated the exercise of market power and
resulted in prices that are quite modest compared to the potential that exists
within the generators. Regulatory pressure appeared quickly in 1994 when
average annual prices moved up from £24/MWh to £28IMWh, and again in
1998 when a similar increase emerged over the winter. These price levels,
however, are far below the threshold for any significant demand-side
elasticity to appear, although if sustained they could increase the attraction
of new entrants. It is fair to say that in the u.K., the generators have not
demonstrated excessive market power to date, but an understanding of their
potential to do so is crucial to maintaining the appropriate balance between
regulation and the progress to full liberalization.
The analysis presented here has sought to generalize some insights that
follow from the SMP price-setting mechanism, per se, in electricity pools.
Market power in practice is influenced by several other factors, (e.g., the
contract positions of the generators with respect both to sales and fuel
purchase), and is exercisable on other aspects of the market, (i.e.,
geographical location and the restriction of available capacity). However,
the SMP market-power issue is core and has been the main focus of
regulatory restraint on market power in generation in the u.K. Furthermore,
insights on this are fundamental and generalizable to many other similar
pools which are being created around the world.
Pricing in Competitive Electricity Markets 123

NOTES
Although outside the pool in the market for contracts, substantial baseload capacity is a
source of market power.

U Demand-side participation in the Pool of England and Wales has been quite limited,
although there was some day-ahead bidding by large industrial users that in principle
should have reduced peak SMPs (Bunn, 1997). However, these demand-side bids tended
to be above SMP and generally greater than £50/MWh. Furthermore, according to private
communication from NGC, it is only for expected SMPs above £100/MWh that some
elasticity heuristics are utilized for operational planning. In this study, SMPs were much
lower, around £20/MWh.

REFERENCES

Benavides, J., (1996), Lessons from the Colombian Electricity Industry and Regulatory
Reform, Paper presented at the 17th USAEElIAEE North American Conference, available
from the author, Universidad de Los Andes, Bogota, Colombia

Borenstein, S., Bushnell, 1., Kahn, E., and Stoft, S, (1996) Market Power in California
Electricity Markets, Utilities Policy Vol. 5, No. 3/4, pp. 219 -236.

Borenstein, S., Bushnell, J. and Wolak, F., (1999) Diagnosing Market Power in
California's Deregulated Wholesale Market. Mimeo. Email: [email protected]

Brealey, R.A., and Lapuerta, C. (1997), A Report on Generator Market Power in the
Electricity Market of England and Wales. The Brattle Group, 28 Grosvenor St., London WIX
9FE U.K.

Bunn, D.W. (1997) Rewarding Demand-side Participation in the Electricity Pool of

England and Wales, Decision Technology Centre, London Business School, NWI 4SA, U.K.

Bunn, D.W., Larsen, E., and Vlahos, K (1994) Complementary Modeling for Analysing
Several Effects of Privatisation on Electricity Investment, 1. Opl. Res. Soc, Vol. 44 No. 10,
pp. 957 - 971.

Electricity Association (1995) Annual Report, 30 Millbank, London SW 1P 4RD

EPRI (1996) The British Privatisation Experiment, 5 Years Later, EPR! TR-106528,
EPRI, 3412 Hillview Avenue, Palo Alto, CA 94304 USA

Financial Times (1996) DTI Blocks Power Bids, April 25, p. 24

124 Pricing In Competitive Electricity Markets

Green, R., and Newbery D. M. (1992) Competition in the British Electricity Spot Market.
Journal of Political Economy Vol. 100, No.5, pp 929-953.

Henney, A., (1987) Privatise Power, Policy Study Number 83, Centre for Policy Studies,
London

10skow, P., Frame, R., lurewitz, 1., Walther, R., and Hieronymous, W. (1996) Report on
Horizontal Market Power Issues, Federal Energy Regulatory Commission Docket No. ER96-
1663-000

Kahn, E.P. (1998) Numerical Techniques for Analyzing Market Power in Electricity. The
Electricity Journal Vol. II (6) pp. 34-43

Littlechild, S. (1996) Regulation of the Electricity Market, Speech to the lEA Third
Annual Electricity Conference, available from the library at OFFER, Hagley House,
Birmingham B16 8QG

Littlechild, S. (l997a) Evidence to the House of Commons Trade and Industry Select
Committee on Energy Regulation, HMSO, London SW8 50T

Littlechild, S. (l997b) Fourth Consultation on Price Restraint, Office of Electricity

Regulation, OFFER, Hagley House, Birmingham B 16 8QG

Monopolies and Mergers Commission (1996) National Power and Southern Electric:
Report on the Proposed Merger HMSO London SW8 50T

National Grid Company (1995) Seven Year Statement, National Grid Company, Coventry
CV48lY

Newbery, D.M., (1995) Power Markets and Market Power Energy Journal Vol. 16 No.3
Office of Electricity Regulation (1991) Report of Pool Price Enquiry, OFFER Hagley
House, Birmingham B 16 8QG

Office of Electricity Regulation (\ 994) Decision on Monopolies and Mergers Reference,

OFFER Hagley House, Birmingham B 16 8QG

Robinson, c., (1988) Competition in Electricity, lEA Inquiry 2, Institute of Economic

Affairs, London.

Stigler, G.1., (1968) The Organisation of Industry, Ch. 4, Irwin, Homewood, Illinois
Pricing in Competitive Electricity Markets 125

Von der Fehr, N-H. M., and Harord, D., (1993) Spot Market Competition in the U.K.
Electricity Market, The Economic Journal, Vol. 103 pp. 531-546
Chapter 8
Market Design and Price Behavior in Restructured
Electricity Markets: An International Comparisoni

Frank A. Wolak
Stanford University, Department of Economics

Key words: International Markets; Market Design; Price Volatility; Strategic Behavior;
Time Series Analysis.

Abstract: This paper argues that the market rules governing the operation of a re-
structured electricity market in combination with its market structure can have
a substantial impact on the behavior of market-clearing prices. It provides an
assessment of the relationship between market rules and market structure and
the behavior of prices in several markets, including England and Wales,
Norway, the State of Victoria in Australia, and New Zealand. The paper
closes with a discussion of the available evidence that the behavior of prices in
each country is the result of the exercise of market power. Several empirical
regularities are suggested by this cross-country analysis. One result is greater
price volatility in systems dominated by fossil-fueled power plants relative to
those dominated by hydro-electric power plants. I also find that electricity
supply industries with a larger component of private participation in the
generation market tend to have more volatile prices, although the evidence
presented also seems to suggest that markets with less participation by
government-owned firms also have lower mean electricity prices after
controlling for differences in generation technologies. Electricity spot markets
with mandatory participation also tend to have more volatile prices than
systems with voluntary participation.

1. INTRODUCTION

Regulators in the U.S. and in several other countries have recently

implemented new regulatory schemes and organizational forms to improve
the incentives for efficient operation of electric utilities. Until very recently,
A. Faruqui et al. (eds.), Pricing in Competitive Electricity Markets
© Kluwer Academic Publishers 2000
128 Pricing In Competitive Electricity Markets

in the U.S., this restructuring took the form of performance-based or

incentive-regulation plans, where the revenue a utility is allowed to earn is
tied less to the cost of providing electricity and more to the attainment of
performance goals as quantified by total factor productivity or some other
measure of productive efficiency.
Other countries have undertaken more radical approaches to restructuring
their electricity supply industries. Following the privatization of the
majority of the generating assets of the formerly state-owned Central
Electricity Generating Board (CEGB), the privatization of all of the Area
Boards (the local electricity distributors), and the introduction of a market
for generation in England and Wales (E& W) on April 1, 1990, many
Organization for Economic Cooperation and Development (OECD) member
countries have formed wholesale markets for electricity and introduced
varying degrees of competition into the retail side of the electricity supply
industry. Most other OECD countries are currently in the process of
implementing similar reforms.
The U.S. has been slow to undertake this radical restructuring process.
As of May 1, 1999 there were only three regions of the U.S. operating
competitive wholesale markets for electricity: California, PJM (all or part of
Pennsylvania, New Jersey, Maryland, Delaware, Virginia, and the District of
Columbia), and New England (Massachusetts, New Hampshire, Vermont
and Maine). The California market was the first to begin operation on
March 31, 1998. As of this same date, the incumbent investor-owned
utilities in California faced competition for their retail customers. Electricity
industry restructuring activity exists throughout most all of the u.S. ii
Retail competition of some form exists in all of the states with a
competitive wholesale electricity market, as well as in many other states.
Several other states in the U.S. have enacted legislation to re-structure their
electricity industry. The vast majority of states have ongoing activity in the
regulatory body that oversees the electricity industry (usually the public
utilities commission) or in the state legislature dealing with electricity
industry restructuring. There are only two states with no significant ongoing
activity on electricity industry restructuring.
All of the industry restructuring that have taken place in the U.S. and
abroad are consistent with the view that competition should be introduced
into the electricity supply industry wherever it is technologically feasible.
Only those portions of the production process most efficiently supplied by a
single firm should remain regulated. The prevailing view is that the
technologies for electricity generation and retailing are both such that
competition is feasible. As discussed above, economies to scale in
generation are exhausted at levels of production significantly below current
levels of industry output. Assuming that all retailers have equal access to the
Pricing in Competitive Electricity Markets 129

transmission and distribution network and electricity from the wholesale

generation market, significant increasing returns to scale in electricity
retailing are unlikely to exist. On the other hand, because competition in the
transmission and distribution of electricity would require duplication of the
current network, these two portions of the electricity supply industry are
thought to possess the features of a natural monopoly. Therefore, the
transmission and distribution sectors of the electricity supply industries in all
of these countries are regulated to varying degrees.

2. AN INTERNATIONAL COMPARISON OF THE

BEHAVIOR OF SPOT ELECTRICITY PRICES

This section characterizes the time series properties of the spot electricity
prices from England and Wales, Norway and Sweden, Victoria, and New
Zealand electricity markets since their inception. My ultimate goal is to
relate differences in the time series behavior of electricity prices across the
four markets to differences in market structure and market rules across the
four markets. Although this is an extremely difficult task, the analysis to be
presented does appear consistent with the view that market structure and
market rules cause significant differences in the behavior of spot prices for
electricity across the four markets.
One of the most striking features of prices from these electricity markets
is their tremendous volatility within days and across days within the week. I
would like to understand the extent to which this variability in prices is
forecastable and how this forecastability varies across the four markets.
Table 1 gives the annual average half-hourly (hourly in the case of Nord
Pool) price and standard deviation of price for each year in our sample in
terms of the home currency of that country. For the Nord Pool, prices are
quoted in Norwegian kroner per MWH. The "na" entries in the table are due
to the fact that the electricity market did not operate during that year. For all
markets, I only have data for a portion of the year in which the market
began, and data for only the first few months of 1997. The E&W market
data runs from April 1, 1990 to March 31, 1997. The Norwegian Spot
Market data runs from May 4, 1992 to May 16, 1997. The Victoriil data
begins July 1, 1994 and ends May 3, 1997. The New Zealand data begins
October 1, 1996 and ends May 31, 1997.
Several conclusions flow from Table 1. First, the mix of generation
technology has an impact on both the mean and standard deviation of market
prices. Prices in the two markets dominated by fossil fuel technology -
E&Wand Victoria - tend to be much more volatile than the prices in the two
markets dominated by hydro-electric capacity, Nord Pool and New Zealand.
 130 Pricing In Competitive Electricity Markets

The coefficient of variation, the standard deviation divided by the mean, for
almost all years in E&W and Victoria is larger than those in Nord Pool and
New Zealand.

Table 1. Annual Means and Standard Deviations (SDs) of Spot Price of Electricity in Home
Currency per MWH
Year Mean SO Mean SO Mean SO Mean SO Mean SD
(EW) (EW) (NW) (NW) (VIC) (VIC) (NZN) (NZN) (NZS) (I'ZS)

£ £ NKr NKr $AU $Al! $I'Z $NZ $NZ SI'Z

1990 17.38 5.38 na na na na na na na na

1991 22.50 12.65 na na na na na na na na
1992 23.42 6.28 58.10 44.38 na na na na na na
1993 27.14 7.86 80.28 41.02 na na na na na na
1994 24.73 18.73 182.67 49.29 36.72 18.24 na na na na

1995 26.15 50.89 117.69 38.92 41.94 30.02 na na na na

1996 25.18 27.85 253.52 44.62 21.11 19.30 39.36 17.00 28.53 6.53

1997 29.27 27.97 150.63 42.90 22.96 59.05 46.97 8.71 41.97 8.87

Notes: EW = England and Wales Pool. units =£/MWH; NW = Nord Pool. units = NKr/MWH;
VIC =Victoria Power Exchange. units =$AUlMWH; NZN =New Zealand North Island, units =
$NZlMWH; NZS = New Zealand South Island. units =$NZlMWH.

With the exception of Victoria in 1994 and 1995 versus 1996 and 1997,
mean prices in the fossil fuel dominated markets tend to be more stable
across years than prices in the hydro-electric dominated systems. iii The
mean prices in the E& W market are much more stable across the years than
those in the Nord Pool. A major determinant of the mean of prices in hydro-
electric capacity-dominated markets is the amount of water available. If
there is little water, then the reservoirs tend to be low and flow volumes in
the rivers are reduced, so that hydro-electric generators tend to be very
reluctant to sell into the spot market during the winter season and spot prices
remain high until the late spring and summer when electricity demand is
much lower. The supply of energy inputs to fossil fuel-based systems is not
nearly as sensitive to local weather conditions. Because there are relatively
integrated international coal, natural gas and oil markets, prices for these
fuels tend to be stable across years, so that the mean price of electricity from
fossil fuel-based markets should be stable across years. The more variable
annual mean prices across years and smaller variance in prices within years
in hydro systems versus fossil-fuel dominated systems is consistent with this
VIew.
There are three alternative explanations for the lower level of volatility in
the Nord Pool and NZEM relative to the E&W market and VicPool. First,
Pricing in Competitive Electricity Markets l31

both fossil fuel-based systems, the E&W market and VicPool, are mandatory
pools, whereas the two hydro-electric-based systems, the Nord Pool and
NZEM, have optional day-ahead markets. Consequently, the lower relative
volatility in the Nord Pool and NZEM could be explained by holders of
bilateral contracts for electricity standing ready to sell into the spot
electricity market if prices become sufficiently high. This willingness to sell
into the spot market at high prices increases the elasticity of the supply
response that any generating company might face if it attempts to raise its
bid prices, so that much of the adjustment to high bids in the spot market
will come in the form of reduced amounts transacted rather than increased
prices, as is the case in mandatory pools with little, if any, demand-side
bidding such as the E&W market, and to a lesser extent VicPool.
Second, the vast majority of generating capacity in the E&W market is
privately-owned and an increasing (over time) fraction of the capacity in the
VicPool is privately-owned, whereas both the Nord Pool and NZEM are
dominated by large state-owned generation companies. One would expect
the large state-owned companies to pursue other objectives besides
maximizing profits, whereas the major goal of the privately-owned firms
would be to maximize profits. Therefore, some of the volatility in the E&W
market and VicPool could be explained as episodes of the successful and
unsuccessful attempts to exercise market power. State-owned enterprises
may also be unwilling to engage in the risky bidding behavior necessary to
set these occasional high prices, and may instead settle for lower, but more
certain revenue streams than privately-owned firms.
Third, there are differences in the bidding process across the four
markets. In E& Wand VicPool generators can alter the quantity supplied
from each bid increment on half-hourly basis. Whereas in the Nord Pool and
the New Zealand markets both the prices and quantities associated with the
hourly supply curves submitted by generators are fixed for the duration that
a generating unit's supply curve is valid. The greater flexibility afforded to
bidders in the E&W and VicPool to vary their supply curves on a half-hourly
basis may allow generators to tailor their bids to set market prices that more
closely follow the within-day pattern of total system load than in the Nord
Pool and New Zealand market where supply functions are generally fixed for
longer periods of time during the day.
A final aspect of Table 1 deserves comment. Consistent with the
description of the differences in market structure between the North and
South Islands in New Zealand (cheap, abundant hydro-electric power in the
South Island and most of the population in the North Island along with more
expensive fossil fuel-based plants), the mean price in the North is
significantly higher than the mean price in the South for both years. In
addition, prices in the North are also more volatile than those in the South,
132 Pricing In Competitive Electricity Markets

particularly for 1996. This reflects the use of fossil-units to meet system
peaks in the North Island. Consequently, even for an integrated system such
as the New Zealand market, the region with the greater share of total
electricity production from fossil-fuel unit's experiences greater spot price
volatility.
To determine which market sells electricity at the lowest price, I convert
each hourly or half-hourly price to U.S. dollars (US$) using the relevant US$
to home currency exchange rate at noon that day obtained from the PACIFIC
web-site. iv Table 2 lists the mean and standard deviations of the US$IMWH
half-hourly or hourly prices (in the case of the Nord Pool). E&W
consistently has the highest US$ price for electricity for the years in which I
have comparable data. For both 1994 and 1995, the US$ prices in E&W are
significantly higher than those in the Nord Pool or Victoria. In both of these
years, Nord Pool set lower prices on average, although in 1996 and 1997,
this order reverses, with VicPool US$ prices significantly lower than the
US$ prices in either the Nord Pool or the NZEM. These low prices in
Victoria can be explained, in part, by the extremely inexpensive Australian
brown coal and natural gas purchased to generate electricity. The coal used
to produce electricity in E&W is considerably more expensive. u.K. coal is
more costly to mine and purchasing coal from abroad entails significant
transportation costs which increases its price in E&W relative to Victoria.

Table 2. Annual Means and Standard Deviations (SDs) of Spot Price of Electricity
Converted to US$/MWH using Daily Exchange Rate
Year Mean Std Mean Std Mean Std Mean Std Mean Std
(EW) (EW) (NW) (NW) (VIC) (VIC) (:'-iZN) (NZN) (NZS) (]\;ZS)

1990 31.84 10.25 na na na na na na na na

1991 39.80 22.71 na na na na na na na na
1992 41.32 11.31 9.20 6.73 na na na na na na
1993 40.80 11.91 11.28 5.62 na na na na na na
1994 38.00 29.37 25.97 7.04 27.42 13.65 na na na na
1995 41.10 79.15 18.50 5.89 30.95 21.94 na na na na
1996 39.37 44.06 39.26 6.86 16.53 15.03 27.77 11.93 20.13 4.59
1997 47.96 46.95 22.50 7.14 17.83 46.00 32.75 6.04 29.25 6.10
Notes: EW = England and Wales Pool, NW = Nord Pool, VIC = Victoria Power Exchange,
NZN = New Zealand North Island, NZS =New Zealand South Island.

In order to better understand the pattern of volatility in the electricity

prices (in home currency per MWH) in the four markets, I compute the ratio
of the difference between the highest and lowest price over a given time
horizon divided by the average value of prices over that same time horizon.
Pricing in Competitive Electricity Markets 133

Table 3. Ratio of (Highest Price - Lowest Price) + (Average Price) over Various Time
Horizons

(Highest Price in Day - Lowest Price in Day) + (Average Price for Day)
Day: Mean SD Min Max
NW 0.18 0.19 0.00 2.04
NZN 0.58 0.65 0.03 3.15
NZS 0.37 0.41 0.01 2.86
EW 1.51 1.34 0.23 12.12
VIC 1.78 1.45 0.03 26.58

(Highest Price in Week - Lowest Price in Week) + (Average Price for Week)
Week: Mean SD Min Max
NW 0.44 0.38 0.04 2.21
NZN 1.49 1.04 0.23 3.31
NZS 1.06 1.00 0.18 3.90
EW 2.83 3.29 0.54 37.84
VIC 3.97 8.51 0.80 102.22

(Highest Price in Month - Lowest Price in Month) + (Average Price for Month)
Month: Mean SD Min Max
NW 0.86 0.54 0.12 2.22
NZN 2.66 1.06 0.66 4.09
NZS 2.28 1.43 0.52 3.94
EW 5.23 6.53 0.89 45.08
VIC 7.81 19.20 1.96 117.29

(Highest Price in Year - Lowest Price in Year) + (Average Price for Year)
Fiscal Year: Mean SD Min Max
NW 2.48 0.99 1.14 4.00
NZN na na na na
NZS na na na na
EW 18.80 15.26 4.07 46.37
VIC 43.16 72.23 4.37 151.46
Notes: Mean = Sample Mean, SD = Standard Deviation, Min = Sample Minimum, Max =
Sample Maximum. All prices in home current per MWH.
134 Pricing In Competitive Electricity Markets

For example, for each day in the sample, I compute the difference
between the highest and lowest day and divide that by the average price for
that day. Repeating this calculation for each day in the sample for each
market and computing means, standard deviations, the sample minimum,
and the sample maximum, yields the values given in Table 3. This table
shows that over all time horizons the prices in E& Wand VicPool are
considerably more variable than those in the Nord Pool and NZEM. By this
measure of variability, the VicPool prices are more volatile than the E&W
prices. The Nord Pool prices exhibit the least amount of average variability
over the four time horizons.
Because I do not have a complete year's worth of data for the NZEM, I
cannot compute the ratio of the difference of the highest and lowest prices
within the year divided by the average price for the year for the NZEM
prices. However, the greater variability in the North Island versus the South
Island NZEM prices shows up in this measure of price variability for all
available time horizons. Although the average variability of these prices is
less than that magnitude in either the E&W market or the VicPool, these
prices are substantially more variable that the Nord Pool prices. These
results illustrate the differences in the time series behavior of prices in
systems where fossil-fuels are used to meet peak demands as in the North
Island of New Zealand relative to systems where hydro-electric capacity is
used to meet system peaks as in the South Island of New Zealand and Nord
Pool.
The next step in the across-country analysis of the behavior of prices
focuses on the relative forecastability of the daily vector of prices in each
country. This requires a model for the time series behavior of the (48xl)
vector of half-hourly prices or (24xl) hourly prices for Nord Pool, which I
denote Yt • After some preliminary analysis of each vector of prices, I settled
on a time-varying mean for Yt which depends on the day of the week and
month of the sample period. I hypothesize that, once M t , the (48xl) [(24xl)
for the case of the Nord Pool] vector of means of Yt. is subtracted from Yt ,
the resulting stochastic process is a vector autoregressive model of order 8.
The statistical model I hypothesize for Yt is:

where E t is a (48xl) [(24xl) for the case of the Nord Pool] vector-valued
white noise process with mean zero and covariance matrix L, <P(L) = I - <PJL
- ... - <ppI}, where each <Pi is a (48x48) [(24xl) for the case of the Nord Pool]
matrix of coefficients and L is the lag operator function which is defined by
e
Yt- k = Yt. The remaining discussion of the model is for the case of forty-
eight half-hourly prices, although the modifications necessary for twenty-
Pricing in Competitive Electricity Markets 135

four hourly prices are straightforward. Let Mti denote the ith element of Mt.
In terms of our above notation, Mti = XtNBi where X t is a vector of day of the
week and month indicator variables for load period i is and Bi is the vector of
coefficients associated with these indicator variables. Excluding the Bi
coefficients associated with M ti , for each element of M t, there are 16,120 =
8x(48)2 elements of <1>], <1>2, ... ,<1>8 to estimate. More than 18,000 coefficient
estimates (including the Bi for each of the 48 load periods) are developed for
this model, by least squares applied to each of the 48 load period price
equations. I provide several summary measures of the adequacy of this
model and summarize what insights it provides about the forecastability of
Yt for each market.
To investigate the adequacy of (1) for each country, I compute the
multivariate analogue of the Box-Pierce (1970) portmanteau statistic derived
by Hosking (1980) for the (48x 1) vector of residuals from equation (1). This
statistic is computed as

Ltrace( C L
M T
P =T r' cr/ C r C~l ), where C r =r 1 Et Et "
r=l t=r+1

where Et is the residual vector from equation (1) for period t and Co is the
sample covariance matrix of Et. Hosking has shown that the asymptotic
distribution of P is X2 with N2x(M - p) degrees of freedom where p is the
order of the autoregressive process and N is the dimension of Yt. For all of
the models estimated, I find little evidence against the null hypothesis that Et
is multivariate white noise.
Table 4 presents the R2 , the standard error of the regression and mean of
the dependent variable for each of 48 ordinary least squares regressions of
the half-hourly price on 8 lags of this price and all other half-hourly prices. I
find the largest R 2,s -all in excess of 0.84- are associated with load periods
33 to 38 which run from 4:00 P.M. to 7:00 P.M., the load periods in the day
with highest prices on average as indicated by sample mean of the PSP in
each load period given in the third column. Load periods 33 to 38 are also
the periods with the six largest estimated regression standard errors. The
combination of these two results suggests that the explanatory power of the
model is highest for those load periods i=33, .. ,38 with the highest
unconditional variance in Yti. However, despite the superior explanatory
power of the model for these load periods, the level of the estimated forecast
variance is higher for these load periods than for any others. Past values'of
Yt therefore improve the predictive power of the load period regressions for
periods 33-38 significantly more than they do for the other load period
regressions, but despite this fact, these load periods are still the most
136 Pricing In Competitive Electricity Markets

unpredictable in terms of the estimated level of their day-ahead forecast

variance. This result is consistent with the view that there are short periods
within the day when PSP is above or below its unconditional mean, and the
occurrence of these extreme prices in certain load periods within a day make
them more likely to occur in the same load periods in neighboring days.

Table 4. R-Squared, Standard Error and Sample Mean of Dependent Variable for
Regression Forecasting Half-Hourly Pool Selling Price in England and Wales
Load Period R-Squared Standard Sample Load Period R-Squared Standard Sample

of Error of Mean of of Error of Mean of

Regression Regression Price in Regression Regression Price in

£/MWH £IMWH

I 0.81 2.43 15.06 25 0.70 8.64 31.31

2 0.78 3.91 17.32 26 0.70 7.73 29.71

3 0.80 4.44 18.91 27 0.70 5.47 25.62

4 0.80 4.77 19.92 28 0.71 4.65 23.65

5 0.80 4.03 18.54 29 0.70 4.83 22.41

6 0.80 3.91 17.96 30 0.73 4.56 21.83

7 0.79 3.80 17.16 31 0.75 4.47 21.22

8 0.79 3.10 15.78 32 0.79 7.69 22.97

9 0.79 2.68 14.90 33 0.84 19.07 32.20

10 0.83 2.21 14.42 34 0.86 31.34 44.61

11 0.80 2.61 14.55 35 0.87 35.45 50.19

12 0.82 2.45 14.56 36 0.87 28.72 46.24

13 0.80 2.68 15.02 37 0.84 17.81 35.51

14 0.80 3.49 17.66 38 0.85 9.10 30.24

15 0.82 3.30 20.31 39 0.82 6.81 27.67

16 0.76 4.55 22.15 40 0.81 4.97 25.52

17 0.74 5.81 24.25 41 0.76 4.55 23.89

18 0.72 6.52 25.96 42 0.75 4.29 23.91

19 0.66 8.05 27.88 43 0.76 4.23 24.35

20 0.66 8.57 29.45 44 0.79 4.05 24.10

21 0.66 8.22 29.36 45 0.77 4.12 22.36

22 0.66 8.10 28.60 46 0.77 3.75 19.85

23 0.69 8.09 29.37 47 0.74 3.05 17.20

24 0.69 8.93 30.99 48 0.78 2.52 15.43

Table 5 presents this same information for the (24xl) vector of daily
Nord Pool spot prices. The most striking feature of this table is the
uniformly high explanatory power of these twenty-four regressions. In all
Pricing in Competitive Electricity Markets 137

cases, the R2 is at least 0.99, which implies that almost all of the movements
in hourly prices across days in the Nord Pool can be forecasted. In addition,
none of the hours appear to be significantly more predictable using past
prices than other hours. For all hours during the day, the standard errors of
the regressions are very similar in magnitude, although the hours during the
day with higher average prices do have slightly larger estimated residual
variances.

Table 5. R-Squared, Standard Error, and Sample Mean of Dependent Variable for
Regression Forecasting Hourly Spot Price from Nord Pool
Load Period R·Squared Standard Sample Mean Load Period R-Squared Standard Sample Mean

Error of Price in Error of Price in

Regression NKrIMWH Regression NKrIMWH

1 0.99 7.31 137.99 13 0.99 9.41 149.57

2 0.99 9.72 136.28 14 0.99 9.46 149.07

3 0.99 8.87 135.38 15 0.99 9.39 148.76

4 0.99 7.46 134.86 16 0.99 9.33 148.53

5 0.99 7.40 135.05 17 0.99 9.03 148.00

6 0.99 7.65 136.82 18 0.99 9.06 148.70

7 0.99 8.84 142.69 19 0.99 8.92 148.98

8 0.99 10.33 148.90 20 0.99 8.69 148.41

9 0.99 10.72 151.05 21 0.99 8.44 147.81

10 0.99 10.65 151.93 22 0.99 8.48 147.69

11 0.99 10.17 151.87 23 0.99 7.99 145.43

12 0.99 10.01 151.49 24 0.99 10.01 141.79

Table 6 presents the information in Table 4 for the Vic Pool prices. The
R2 from the 48 regressions used to estimate the 8th order vector
autoregressive process indicate that VicPool prices are less forecastable than
the Nord Pool prices. The magnitude of the R2 is similar to those for the
E&W system in Table 4. However, different from the results in Table 4, I
find that the higher average price periods do not have higher R Z from the
regression forecasting that price. In fact, the highest average price period,
load period 26, has by far the lowest R2 = 0.44. Different from the case of
the E&W market, the highest R2 >'s occur for load periods with both low
and high average prices.
Because I only have a very short time series of prices for the NZEM, it is
not possible to estimate an 8th order vector autoregressive process for these
prices. I estimate the vector autoregression with the largest number of lags
possible given the time series of prices available to me. In this case, I am
able to estimate a 4th order vector autoregressive process. Table 7 presents
l38 Pricing In Competitive Electricity Markets

the R2, the standard error of the regression and mean of the dependent
variable for each of 48 ordinary least squares regressions of the half-hourly
spot price on 4 lags of this price and all other half-hourly prices for the North
Island spot prices. Table 8 produces this same information for the South
Island spot prices. The South Island price results resemble the results for
Nord Pool, consistent with the fact that the South Island is dominated by
hydro-electric capacity. The North Island results also resemble the Nord
Pool results, but they show more variability across hours in the R2 and the
standard error of the regression than do the South Island results. This is
consistent with the use fossil-fuel plants in the North Island.

Table 6. R-Squared, Standard Error. and Sample Mean of Dependent Variable for
Regression Forecasting Half-Hourly Spot Price from VicPool
Load Period R-Squared Standard Sample Mean Load R-Squared Standard Sample Mean

Error of Price in Period Error of Price in

Regression 5AUfMWH Regression SAUIMWH

I 0.89 9.23 32.22 25 0.89 16.56 37.50

2 0.89 9.12 29.06 26 0.44 113.75 40.80

3 0.87 10.49 33.47 27 0.81 18.47 38.20

4 0.87 9.72 29.36 28 0.81 18.39 37.86

5 0.87 8.59 24.82 29 0.80 18.73 37.21

6 0.86 7.77 20.92 30 0.78 18.80 36.31

7 0.86 7.15 17.40 31 0.79 18.03 35.90

8 0.84 6.94 14.73 32 0.79 18.77 35.92

9 0.83 6.34 12.85 33 0.80 17.62 35.76

10 0.80 6.64 11.95 34 0.79 17.22 36.15

II 0.77 9.10 13.69 35 0.80 17.49 37.59

12 0.78 9.51 16.28 36 0.82 17.55 39.92

13 0.79 11.16 22.35 37 0.77 21.82 39.84

14 0.81 13.29 28.56 38 0.81 18.28 3808

15 0.84 11.35 29.47 39 0.84 14.66 35.50

16 0.85 \3.39 34.02 40 0.81 14.64 34.39

17 0.83 14.97 36.46 41 0.86 14.24 33.80

18 0.82 16.37 37.00 42 0.86 13.85 32.33

19 0.84 15.55 37.81 43 0.83 13.26 29.87

20 0.84 15.71 3803 44 0.83 12.28 26.52

21 0.85 15.43 37.99 45 0.82 12.09 25.21

22 0.85 16.08 38.19 46 0.80 12.08 24.74

23 0.84 15.93 37.38 47 0.79 14.60 34.18

24 0.87 15.06 37.26 48 0.80 13.76 33.31

Pricing in Competitive Electricity Markets 139

Table 7. R-Squared, Standard Error, and Sample Mean of Dependent Variable for
Regression Forecasting Half-Hourly Spot Prices from NZEM-North Island Reference Node

Load Period R- Squared Standard Sample Mean Load Period R- Squared Standard Sample Mean

Error of Price in Error of Price in

Regression $NZlMWH Regression $NZlMWH

1 0.98 4.06 42.55 25 0.96 6.30 46.14

2 0.98 4.22 41.71 26 0.97 5.22 46.10

3 0.97 4.19 41.09 27 0.96 4.45 45.30

4 0.97 4.26 40.54 28 0.97 4.38 45.36

5 0.97 4.14 39.61 29 0.97 5.06 45.14

6 0.97 3.85 38.71 30 0.98 5.00 45.32

7 0.96 4.48 38.42 31 0.97 5.32 45.24

8 0.97 3.90 38.08 32 0.96 7.54 46.42

9 0.97 4.20 37.87 33 0.93 11.18 46.17

10 0.97 4.19 38.03 34 0.97 7.46 47.67

11 0.96 4.19 38.28 35 0.94 9.36 47.08

12 0.96 4.12 38.80 36 0.92 10.06 47.55

13 0.96 3.91 40.11 37 0.95 7.49 46.60

14 0.96 4.23 42.31 38 0.95 6.19 45.90

15 0.96 7.20 45.34 39 0.96 5.35 45.33

16 0.97 10.50 50.84 40 0.97 4.61 45.03

17 0.96 11.02 50.33 41 0.98 6.38 47.00

18 0.95 11.57 49.27 42 0.98 6.80 47.53

19 0.97 6.43 47.35 43 0.98 5.19 46.35

20 0.96 7.33 47.04 44 0.98 5.15 46.03

21 0.98 4.84 46.46 45 0.96 4.33 43.63

22 0.97 5.19 46.30 46 0.97 3.76 42.32

23 0.97 5.30 46.22 47 0.97 5.07 43.44

24 0.96 6.56 46.76 48 0.96 5.92 41.59

140 Pricing In Competitive Electricity Markets

Table 8. R-Squared, Standard Error, and Sample Mean of Dependent Variable for
Regression Forecasting Half-Hourly Spot Prices from NZEM-South Island Reference Node
Load R-Squared Standard Sample Mean Load Period R- Squared Standard Sample Mean

Period Error of Price in Error of Price in

Regression $NZlMWH Regression $NZlMWH

I 0.99 3.13 37.40 25 0.98 3.35 39.25

2 0.99 2.86 36.95 26 0.99 3.00 39.03

3 0.99 2.88 36.48 27 0.99 2.93 39.21

4 0.99 3.16 36.09 28 0.99 3.03 38.97

5 0.99 3.04 35.23 29 0.99 3.12 38.56

6 0.99 3.07 34.48 30 0.98 3.28 38.26

7 0.98 3.39 33.87 31 0.98 3.33 38.29

8 0.98 3.39 33.54 32 0.98 3.26 38.71

9 0.98 3.48 33.32 33 0.98 3.36 38.90

10 0.98 3.78 33.30 34 0.99 3.05 39.05

II 0.98 3.30 33.34 35 0.97 5.53 39.56

12 0.99 2.75 33.66 36 0.96 6.38 40.19

J3 0.99 2.78 34.74 37 0.99 3.56 39.60

14 0.98 3.56 36.33 38 0.99 3.38 39.19

15 0.98 4.89 37.98 39 0.99 3.22 38.85

16 0.96 7.59 39.73 40 0.99 2.83 38.28

17 0.97 5.35 38.81 41 0.99 3.16 38.07

18 0.96 5.48 38.85 42 0.98 3.74 37.94

19 0.99 3.21 38.98 43 0.98 4.13 37.90

20 0.98 3.22 39.01 44 0.98 4.02 38.03

21 0.98 3.43 39.16 45 0.98 3.87 37.52

22 0.98 3.46 39.17 46 0.98 3.43 36.62

23 0.98 3.30 39.13 47 0.98 3.64 37.31

24 0.98 3.48 39.34 48 0.98 3.60 36.21

Pricing in Competitive Electricity Markets 141

I now characterize differences in the behavior of the spot prices within the
day and week across the peak and off-peak months of the year. Figure l(a)
plots the average behavior of normalized prices throughout the day for the
E&W market in Winter (December, January and February) and Summer
(June, July and August). To compute the normalized price for any load
period, I divide the actual price by the sample mean price of electricity in the
E& W market. Figure 1(b) plots the behavior of normalized prices
throughout the week in Summer and Winter. These plots illustrate an
important feature of behavior of prices in the E&W market. During the
winter months, all weekday prices become very high during load periods 35-
37. The average high price during weekdays (excluding Fridays) is more
than 4.0 times the sample mean of the spot price in load periods 35-37. In
Wolak and Patrick (1996b), we argue that this pattern of prices represents
the exercise of market power by National Power and PowerGen, the two
major generators in the E&W market. (See figures 1 below).
Figures 2(a) & (b) present the day 2(a) and week 2(b) normalized price
plots for the Nord Pool. There appears to be little predictable variation in the
spot prices within the day and across days of the week in the Nord Pool. The
major movements in prices appear to be across the peak and off-peak
seasons, with average summer prices significantly below average winter
prices within the day and within the week. This reflects the view that water
scarcity is a major determinant of prices in the Nord Pool. [See Figures 2(a)
and (b) below].
Figures 3(a) & (b) present the day 3(a) and week 3(b) normalized price
plots for the VicPool. The average pattern of prices within the day 3(a) and
week 3(b) in VicPool shares features with both the Nord Pool and the E&W
market. For consistency with the other figures, I have defined Summer to be
the months of June, July and August and Winter to be December, January
and February. For the most half-hours, average prices in June, July and
August (Summer) are higher than those are in December, January and
February (Winter). The differences in predictable price fluctuations within
the day and week across the peak and off-peak seasons is not nearly as
pronounced in the Vic Pool as it is in E&W market. Both seasons exhibit
more predictable variation within the day and week than do E&W prices in
the Summer, but less than E&W prices in the Winter. [See Figures 3(a) and
(b) below].
Because no data exists for New Zealand for the months of June, July and
August, Figure 4(a) & (b) plot the average prices in the North and South
Islands throughout the day 4(a) and week 4(b). The pattern of average prices
within the day for both prices in New Zealand is very similar to the pattern
of prices within the day for the Nord Pool. A similar statement can be said
142 Pricing In Competitive Electricity Markets

about the behavior of both New Zealand prices within the week. (See
Figures 4(a) and (b) below).

Average prices throughout the day for UK

5.0

4.5

4.0

·
3.5
0
3.0
~
"·
N 2.5

1
z
2.0

1.5

1.0

0.5

0.0

Load period

Figure I(a). Average Prices throughout the Day for U.K.

Average prices throughout the week for UK

5.0

4.5

4.0

3.5

t: 3.0

~ 2.5

j 2.0

o 4 II , , 2 2 3
8 I 4
4
II
2 "o 8
8
3
6
W_1y load period

I-Wlntw p
•. - --- Summw,Pr1~
,- I
Figure J(b). Average Prices throughout the Week for U.K.
Pricing in Competitive Electricity Markets 143

Average prices throughout the day for NW

5.0

4.5

4.0

.3
't:
3.5

3.0
II.

Ja
2.5

~ 2.0
•
z
1.5

1.0
--------------------------------------------------------------
0.5

0.0
0 6 1 1 2
2 8 4
Load portod

- WInter price.
- Summer pric.

Figure 2(a). Average Prices throughout the Day for NW

Average prices throughout the week for NW

5.0

4.5

4.0

.=
3.5

't:
II.
3.0
11
.!!a 2.5
~ 2.0
•
z
1.5

1.0
____ ------, .. - ... --......--r---- ... -'-----....--r---- ... -,-------,.,-----
0.5

0.0
o 2 4 7 9 1 1 1
4 8 2 6 2 4 6
o 4 8
WHkty load period

- Wlnt.r prlc••
- Summer prfce.

Figure 2(b). Average Prices throughout the Week for NW

144 Pricing In Competitive Electricity Markets

Average prices throughout the day for VICT

5.0

4.5

4.0

3.5
e 3.0
't:

...a.
..E
.
~ 2.5

z
2.0

1.5 .... / . . ,
../
/ . . - - - - - - .7'~-----.::--"'----..:~
-"-
",,-
... --.......... /--
1.0
"
",- ' -//- ' ---'
.... ----../
0.5

1 2 3 3 4
8 4 o 6 8
Load period

- - Winter prien
- - Summer prieD

Figure 3(a). Average Prices throughout the Day for VICT

Average prices throughout the week for VICT

5.0

4.5

4.0

3.5
~
'I: 3.0
...a.
..E
.
~ 2.5

2.0
z
1.5

1.0

0.5

0.0
0 4 1 1 2 3
8 4 9 4 3
4 2 0 8

Weekly load period

- - Winter prle..
- - Summer pr1c••

Figure 3(b). Average Prices throughout the Week for VICT

Pricing in Competitive Electricity Markets 145

Average prices throughout the day for NZ

5.0

4.5

4.0

3.5
I
-.: 3.0
...
Q.

S 2.5

"~ 2.0
:2
1.5

1.0

0.5

0.0
0 1
2
1
8
2
4
3
o
3
e
4
2 "
8
Load perlod

- final prlca: N_ 1
1...... final prl_: So~

Figure 4(a). Average Prices throughout the Day for NZ

Average prices throughout the week for NZ

5.0

4.5

4.0

3.5
~

"t. 3.0

11.5
~
41 2.5

2.0

1.0j
0.5
1
O.O~,~~~~~~~~~~~~~~~~~~~~~~~~~~~
o 4 1 2 2
8 4 4 8
4 o 8

Weekly load period

- - final prices: North

- - fino I prices: South

Figure 4(b). Average Prices throughout the Week for NZ

Figures 5 through 7 (below) plot the period level standard deviations of

the normalized prices within the day for the E&W market, Nord Pool and
146 Pricing In Competitive Electricity Markets

VicPool. Each point on this plot is the standard deviation of the normalized
(by the overall sample mean price) price for that load period within the day
for all days within that season. Figure 5 illustrates that although normalized
prices in load periods 35-37 are known to be very high, there is considerable
uncertainty precisely how high they will be. Figure 6 tells a similar story to
case of the mean prices within the day for Nord Pool. The uncertainty in
normalized prices is uniform within the day in both Summer and Winter, but
the uncertainly in normalized prices is uniformly higher in the Summer than
the Winter.
Figure 7 illustrates that for the most part the degree of uncertainty in
normalized prices is very similar across load periods in the VicPool. The
only exception is that during the high-priced periods in December, January,
and February (the months of peak demand in Victoria) the uncertainty in
normalized prices is highest during the highest priced periods of the day.
Figure 8 computes the load-period level standard deviations in normalized
prices for the North Island and South Island prices in New Zealand. The
pattern of uncertainty in these prices is very similar to the within-day
uncertainty in prices in the Nord Pool. The North Island standard deviations
tend to be higher than the South Island standard deviations particularly for
the peak periods of the day, reflecting the use of fossil-units during these
time periods in the North Island.

Price STD throughout the day for UK

7.0
6.5
6.0
5.5
5.0
4.5
4.0

load period

- Winter price srD

- Summer price STD

Figure 5. Price STD throughout the Day for U.K.

Pricing in Competitive Electricity Markets 147

Price STD throughout the day for NW

7.0
&.5
&.0
5.5
5.0
4.5
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
0 2
4
Load period

- Wlnlor p- STD 1
1- Summer price STD

Figure 6. Price STD throughout the Day for NW

Price STD throughout the day for VieT

7.0
&.5
&.0
5.5
5.0
4.5
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
0 1 2 3 3 4 4
8 4 o & 2 8
Loadpertod

- - Wlnler prl .. STD

- - Summer price STD

Figure 7. Price STD throughout the Day for VIeT

148 Pricing In Competitive Electricity Markets

Price STD throughout the day for NZ

7.0 1
6.5 ~
6.0
5.5
5.0
4.5
4.0
3.5
3.0
2.5
2.0,
1.51

1. 0 1
0.5 '
0.0 ~~.~-~-.~-~~--,~~~~~~~~~~~~r~--~/~---,--:-:-=--=--=--=--=--=--::--,~=~~~---,-~~~~~---,-~~~-~-r-c

Load period

-~ Price STO: Norlh

- Price STO: South

Figure 8. Price STD throughout the Day for NZ

3. MARKET STRUCTURE AND MARKET RULES

AND THE EXERCISE OF MARKET POWER
A significantly more detailed analysis of each of these markets is
required to draw conclusions about the exercise of market power in any of
these markets. However, the strong influence that both market structure and
the market rules appear to exert on the behavior of prices in thes~ markets
suggests that such cross-country analysis should yield insights about how
market rules and market structure combine to allow the exercise of market
power.
The dramatically different pattern of average electricity prices within the
day and within the week in the E&W market relative to the other three
markets lends further support to the conclusion reached in Wolak and Patrick
( 1996b). For example, the two largest generators in the England and Wales
market - National Power and PowerGen - possess significant market power
that they are able to exercise when certain conditions in the E&W market
make the residual demand they jointly face extremely large relative to the
capacity of these two large generating companies.
The relatively flat pattern of average prices throughout the day in the
Vic Pool and the very low US$ prices for electricity from these market in
1996 and 1997 is indicative of a competitive electricity market. The
 Pricing in Competitive Electricity Markets 149

relatively high degree of volatility in prices throughout the day in the

VicPool indicate that generators are sometimes successful at obtaining high
prices, but just as often their efforts yield very low prices. Thus, on average,
prices are very low. Consequently, the VicPool is an example of a market
where the efforts of generators to exercise market power are on average
unsuccessful. It is more competitive market than the E&W market.
Nord Pool and the NZEM present a more difficult puzzle because both
markets are dominated by large state-owned enterprises. We would not
expect these firms to exercise market power with the same vigor as
privately-owned firms. Nevertheless, both of these electricity supply
industries produce the vast majority of their power from very inexpensive
hydro-electric resources, so the higher US$ dollar prices in these two market
relative to Victoria does raise suspicions about the exercise of market power
by the large state-owned firms. As discussed earlier, in the fall of 1992,
Statkraft publicly announced a policy to keep the spot price above 100
NKrlMWH, although subsequently prices have fallen below this level for
long periods of time. The evidence from the behavior of prices in both Nord
Pool and the NZEM relative to prices in Victoria and the E&W market
indicates that the large state-owned generators in Nord Pool and the NZEM
are price-leaders with the remaining firms serving as a competitive fringe.
Further analysis of both of these markets is necessary to reach a more
definitive conclusion about whether these outcomes represent market power.

NOTES

Frank A. Wolak is a Professor of Economics at Stanford University and a Research

Associate of the National Bureau of Economic Research. The author would like to thank
Marshall Yan for the outstanding research assistance. Severin Borenstein, Jim Bushnell
and Richard Green provided very helpful comments on a previous draft. Partial financial
support for this research was provided by the National Science Foundation. This paper
reflects the state of electricity industry restructuring in England and Wales, Norway and
Sweden, Australia and New Zealand as of January 1, 1998, unless otherwise noted. An
up-to-date description of the status of electricity industry restructuring in these countries
can be obtained from http://www.stanford.eduJ-wolak.

ii The Energy Information Agency (EIA) of the U.S. Department of Energy (DOE) maintains
an up-to-date web-site on the status of electricity industry restructuring in each state. See
h ltp:/Iwww.eia.doe.gov/cneaflelectricitv/page/restructll re .hlInl.

iii As discussed in London Economics (1996), there are several reasons to believe that there
was a regime shift in the VicPool before and after January 1, 1996. Before this date, very
few of the generators had been sold off, so that the SECV was effectively bidding all
plants. In addition, before this date, there were high levels of vesting contracts at prices
between $AU 35/MWH and $AU 40 MWH.
150 Pricing In Competitive Electricity Markets

IV Policy Analysis and Computing and Information Facility in Commerce (PACIFIC) at the
University of British Columbia, Faculty of Commerce and Business Administration
(http://pacillc.commerce. lInc.calxr/ ).

REFERENCES

Armstrong, Mark, Cowan, Simon, and Vickers, John (1994) Regulatory Reform:
Economic Analysis and British Experience, MIT Press: Cambridge, MA.

Baker, A.B. (1992) "The Methodology and Process of Forecasting Nominal Demand,"
The National Grid Company, Grid System Management, National ControL GSMNC/ABB/DF
5.1.2.9, Coventry, England.

Box, G.E.P. and Pierce, David A. (1970) "Distribution of Residual Autocorrelations in

Autoregressive-Integrated Moving Average Time Series Models," Journal of the American
Statistical Association, 65, 1509-1526.

Brough, Martin and Lobban, Seumas, (1995) Guide to the Economic Regulation of the
Electricity Industry, OXERA Press, Oxford, U.K.

Bunn, Derek W. and Larsen, Erik R., (1992) "Sensitivity of Reserve Margin to Factors
Influencing Investment Behavior in the Electricity Market of England and Wales." Energy
Policy, May, 420-429.

Edison Electric Institute (1996) Retailing Wheeling & Restructuring Report, December,
1996.

Electricity Association (1994) U.K. Electricity, Electricity Association Services Limited,

London.

Electricity Association (1995 U.K. Electricity, Electricity Association Services Limited,

London.

von der Fehr, Nils-Henrik Morch and Harbord, David (1993) "Spot Market Competition
in the U.K. Electricity Industry," The Economic Journal, 103,531-546.

von der Fehr, Nils-Henrik Morch and Harbord, David (1997) "Competition and
Regulation in Electricity Markets: Economic Theory and International Experience", Mimeo,
June.
Pricing in Competitive Electricity Markets 151

Green, Richard 1. and Newbery, David, M. (1992) "Competition in the British Electricity
Spot Market," Journal of Political Economy, 100(5), pp. 929-953.

Green, Richard J. (1996) "Increasing Competition in the British Electricity Spot Market,"
The Journal of Industrial Economics, Vol. XLIV, No.2, pp. 205-216.

Green, Richard, J. (1999), "The Electricity Contract Market in England and Wales,"
Journal of Industrial Economics, Vol. XLVII, No. I, pp. 107-124.

Helm, Dieter and Powell, Andrew (1992) "Pool Prices, Contracts and Regulation in the
British Electricity Supply Industry," Fiscal Studies, 13(1), 89-105.

Hosking, J.R.M. (1980) "The Multivariate Portmanteau Statistic," Journal of the

American Statistical Association, 75, pp. 602-608.

Joskow, Paul (1987) "Productivity Growth and Technical Change in the Generation of
Electricity," The Energy Journal, 8(1), pp. 17-38.

Lee, Byung-Joo, (1995) "Separability Test for the Electricity Supply Industry," Journal of
Applied Econometrics, 10,49-60.

London Economics (1997) The Australian Electricity Markets.

London Economics (1997) Overview of Nord Pool.

National Grid Company (1995) 1995 Seven Year Statement, The National Grid Company
pic, Coventry.

Nord Pool (1997) The Spot Market.

Nord Pool (1997) The Futures Market.

Newbery, David M. (1995) "Power Markets and Market Power," The Energy Journal,
16(3),39-66.

OX ERA Press (1996) "The Stock Market Goes Nuclear," Energy Utilities, July 15, No.
31196.

OX ERA Press (1997) "The Pool Feels the Cold," Energy Utilities, January, 13.
152 Pricing In Competitive Electricity Markets

Patrick, Robert H. and Wolak, Frank A. (1997) "Estimating Customer-Level Demand for
Electricity Under Real Time Pricing," Mimeo, March 1997. (Available from
http://www.stanford.edul-wolak).

Reed, E. Grant, Drayton-Bright, Glenn R., and Ring, Brendan J. (1998) "An Integrated
Energy and Reserve Market for New Zealand," Energy Modeling Research Group Working
Paper. EMRG-WP-98-0 l. Department of Management, University of Canterbury,
Christchurch, New Zealand.

Powell, Andrew (1993) 'Trading Forward in an Imperfect Market: The Case of

Electricity in Britain," The Economic Joumal, \03,444-453.

The Electricity Pool (1997) Schedule 9: The Pool Rules, Pooling and Settlement
Agreement for the Electricity Industry in England and Wales, Issue 45, Version 6.20,
February 1.

Wolak, Frank A. and Patrick, Robert H. (1996a) "Industry Structure and Regulation in
the England and Wales Electricity Market," in M.A. Crew, editor, Pricing and Regulatory
Innovations Under increasing Competition, Kluwer Academic Publishers, 65-90.

Wolak, Frank A. and Patrick, Robert H. (l996b) "The Impact of Market Rules and
Market Structure on the Price Determination Process in the England and Wales Electricity
Market," Mimeo, June 1996. (Available from http://www.stanfonl.edu/-wolak).
Chapter 9

The Effect Of Technology on Energy Pricing In A

Competitive Energy Market

Chris King
Utility. Com

Key words: Advanced Metering, Demand Reduction; Demand Response; Electric

Deregulation; Internet; Market Power; Price Spikes; Time-of-Use Pricing.

Abstract: Three types of market power threaten effective electric competition: vertical,
horizontal and the cartel-like market power of a group of market participants,
generators, imposed on another group, consumers, during times of system
peak loads. This chapter addresses means to mitigate these issues; they
include the Internet, which provides consumers with free access to a wealth of
information about their own energy use and how better to manage that use;
advanced metering, which provides consumers the ability to reduce usage, and
their bills, during system peak hours; functional separation of the elements of
electricity, including generation, transmission, distribution, and revenue cycle
services (billing, metering, and customer service); and, finally, ensuring the
economic separation of regulated and competitive services.

1. INTRODUCTION

Three types of market power threaten effective competition in electricity,

The first is the vertical market power of companies that own virtually the
entire value chain of electricity, including generation, transmission,
distribution, and revenue cycle services. The second is horizontal market
power in which companies leverage assets and resources acquired to provide
regulated services to compete for services open to competition, including the
provision of retail electricity via default or standard offer service. The third

A. Faruqui et al. (eds.), Pricing in Competitive Electricity Markets

© Kluwer Academic Publishers 2000
154 Pricing In Competitive Electricity Markets

is cartel-like market power of a group of market participants, generators,

imposed on another group, consumers, during times of system peak loads.
Fortunately, there are powerful means to mitigate these market power
issues. First and foremost is the ability of competitive electric suppliers to
provide consumers with technology tools. Via use of the Internet,
consumers have free access to a wealth of information about their own
energy use and how better to manage that use. Via advanced metering,
suppliers provide consumers with the ability to reduce usage, and greatly
reduce bills, during system peak hours. Use of these tools enables the
customers of one Internet-based provider, for example, to save an average of
close to $100 per year on their electric bills.
There is no doubt that the Internet puts the power of choice in the hand of
the consumer. That in itself is going to affect prices that for many years
have been the domain of a utility. Coupled with deregulation and low
marginal costs of disseminating information will allow the consumer to
"shop around" for the best deal possible, increase competition for the
utilities and build the path to a better informed and educated decision
making consumer.
Second, vertical market power can be mitigated through functional
separation of the elements of electricity, including generation, transmission,
distribution, and revenue cycle services (billing, metering, and customer
service). Competitive services should be fully exposed to the forces of the
marketplace. Another mitigation is to still regulate natural monopoly
services, specifically the transmission and distribution wires.
Third, horizontal market power can be mitigated by ensuring the
economic separation of regulated and competitive services. There should be
no cross subsidies between such services, and the entire cost of each type of
service should be allocated to that service. Most importantly, customers
who choose to take a service from a competitive supplier should not be
required to continue to pay the regulated company even a portion of the
amount previously paid for that service.
To make it possible to use technology in bringing down prices of energy,
standards adoption is reducing the high transaction costs that are caused by
the use of differing data formats and data transport mechanisms in each
distribution utility service area. A number of states in the U.S. are already
beginning to converge on the use of the Electronic Data Interchange formats
of the Utility Industry Group. Such standards also result in greater access to
information by consumers.
To understand the role of technology in enabling price competitiveness, it
is important to understand the structure of the energy industry for
consumers. This begins with looking at market power issues.
Pricing in Competitive Electricity Markets 155

2. DEFINING POWER MARKET

Market power, generally, is that situation in which market participants are

able to earn "excess profits" as a result of market inefficiencies. Two
generic types of market power occur in the electric industry: vertical and
horizontal. A third type, similar to that enjoyed by a commodity cartel, is
caused by the limitations of today's installed information technology; this
type allows power generators, as a class of market participants, to earn
excess profits at times of system peaks by taking advantage of the lack of
demand response by consumers - which, in turn, is a result of the lack of
information. The information consumers need to exercise the demand side
of the supply and demand equation is greater detail on usage - such as how
much is during peak times - and on pricing - such as how much more
expensive is power at those times.
Vertical market power results when a single participant, generally the
incumbent utility, controls all or most elements of the electricity value chain
in a way that prevents economically efficient consumer decisions. This
value chain starts with power production, extends through transmission and
distribution, and concludes with revenue cycle services, including billing,
metering, and customer service. By owning all elements of the value chain,
a single market participant can raise prices above competitive levels. That
participant can also exert market power by controlling a single, scarce
element, such as transmission or distribution wires, or even detailed energy
usage information.
Horizontal market power results from the geographical or breadth of
services scope of a market participant that provides that participant with
certain competitive advantages. A common example is leveraging resources
deployed for one service to reduce the costs of entry for another. In
electricity, for example, a utility could use its service trucks and personnel to
support services similar to but unrelated to the distribution of electricity,
such as appliance maintenance. Since other companies do not have the same
opportunity, (e.g., an appliance repair company cannot use its trucks and
personnel to perform electricity system maintenance), the utility'S horizontal
market power gives it a competitive advantage.
Cartel-like market power differs from vertical and horizontal market
power in the sense that, instead of being a situation in which a single
company has market power, it is one in which a group of companies have
market power as compared to consumers. Electricity is unique in two
respects that result in this cartel-like market power. First, power cannot be
stored; with few meaningful exceptions, power production and use must be
balanced every four seconds. Accordingly, during the peak hours of the
year, almost all the power plants in an area are running, and very few plants
156 Pricing In Competitive Electricity Markets

are available to serve the last few kilowatts of demand. Second, even though
this lack of producers results in very high power costs, consumers have no
reason to reduce their usage, since they pay a price that is averaged over the
year. Thus, generators can charge as much as seventy-five times the normal
rate for energy. Moreover, all producers are paid these high, marginal
clearing prices in those markets, such as California or the u.K., where most
(California) or all (U.K.) power must flow through the officially approved
exchange.
Economics professors Frank Wolak of Stanford University and Robert
Patrick of Rutgers University studied such market power in the U.K. and
found that the lack of price signaling to power users enables generators to
manipulate market prices for energy and capacity, resulting in excess profits.
They found that the lack of price signals provided via time-of-use or hourly
(half-hourly in the U.K.) metering has resulted in serious market
inefficiencies in the u.K., including forcing consumers to pay high market
prices - sometimes exceeding $1,500 per MWh - during peak periods:

... One of the problems in the United Kingdom is that most

electricity consumers, including all residential customers, pay a
price for electricity to their retailer that does not change in
response to half-hourly variations in the market-clearing price of
electricity. Consequently, under the current system a very high
market price brings about little, if any, demand reduction, because
the final consumer of electricity does not pay this price for its
electricity. i

3. CONSUMER RISKS RESULTING FROM

MARKET POWER

Throughout human history, open competitive markets have consistently

delivered lower prices and greater innovation than regulated monopolies.
The success of such markets motivates the current trend in states in the U.S.
toward adopting retail electricity competition. Market power, if not
mitigated, presents two dangers. First, in the absence of effective
competition, the desired price and innovation benefits of competition will
not materialize. Consumers will not exercise choice, or their choices will
not be economically efficient. Second, with the restraints of regulation
removed, companies with market power could charge even higher prices and
earn excess profits if consumers have no effective tools to combat that
market power.
Pricing in Competitive Electricity Markets 157

Three examples of market power in electric competition are of particular

import. The first is the vertical market power of companies who own all the
major elements of the electricity value chain in a limited geographic region,
including generation, transmission, distribution, and revenue cycle services.
Such vertical market power has been addressed extensively in electric
restructuring proceedings at both the U.S. State and Federal levels, with
consensus that generation, transmission, and distribution must be unbundled
from one another, with or without divestiture requirements. Without equal
and non-discriminatory access to transmission and distribution systems, the
jurisdictions have agreed that there can be no effective competition between
power generators.
The second important example is horizontal market power in which
regulated and competitive utility functions are cross-subsidized, intentionally
or not, and which results in anti-competitive effects. One such situation is
the provision of standard offer or default service where some or all of the
costs, such as revenue cycle services, are embedded in regulated distribution
rates. In this situation, competitive suppliers are at a major disadvantage;
they must recover all of their revenue cycle service costs from competitively
provided services, while the standard offer service does not include those
costs.
Another such situation of horizontal market power is the sale of
competitive services such as advanced metering or any other competitive
service, where the sale uses the regulated utility's name. In this case, the
brand equity inherent in the name, and the association of that name with
electricity services, reduces the company's cost of acquiring a new customer
or selling a new service to an existing customer. Because the customer
places a value on this brand equity, the customer is willing to pay more for
service. For example, in Pennsylvania, all small businesses and residential
customers would save 10 percent on their electricity by switching to a wide
range of competitive suppliers, yet over 80 percent of these customers have
not switched suppliers. On average, these non-switching customers are
paying approximately $100 per year for name brand and other incumbency
equity. Naturally, customers should be allowed to choose freely to pay extra
for brand equity and do so in almost all competitive markets. The difference
is that, in those other markets, customers are not required, by government-
regulated monopoly, to take a portion of their service (electricity distribution
and, so far, at least some revenue cycle services) from the named entity.
One actual customer story illustrates the strength of this brand equity.
An industry expert living in the East suggested to his California-dwelling
relative that the latter sign up for service from a competitive Electric Service
Provider (ESP) to obtain savings on his electric bill. By way of background,
the relative is very bright; in fact, he was a Rhodes scholar and was well
158 Pricing In Competitive Electricity Markets

aware of the California Public Utility Commission's educational efforts

regarding deregulation. The expert explained that, under the rules of electric
competition, the regulated utility is still responsible for repairing service
after outages and ensuring reliability. Nevertheless, the Rhodes scholar
mistakenly believed that his service might somehow be less reliable if he
switched to an ESP.
States in the U.S. have developed varying approaches to mitigating such
horizontal market power of incumbency and brand equity. One approach is
to allow utilities to use their names for unregulated competitive affiliates,
provided they disclose clearly that those affiliates are not the same as the
regulated utility, operate completely independently, keep entirely separate
accounts, and obtain no financial benefits from the regulated entity,
including credit - a key requirement in wholesale electricity markets. In the
spirit of compromise, Utility.com does not oppose the ability of utilities to
continue to use their names under these conditions.
The third important example of market power is that of the cartel-like
market power of power generators. In this case, during times of system
peak, consumers are forced to pay excess prices because there is no demand
response in spite of excessive wholesale power prices. This occurs because,
except for the less than one percent of customers that have time-of-use or
hourly meters, consumers have no awareness that wholesale prices are so
high. These small consumers simply pay the same, averaged price
throughout the year - including the very high costs incurred during the
system peak hours. A similar effect occurs with respect to the cost of
reliability, which is the price that grid operators must pay for backup reserve
energy and other ancillary services (functions regulated by the PERC). As
with electric competition as a whole, where more offerings are made to large
electricity users, small consumers are the ones who suffer from not having
the advanced metering that allows them to respond to price signals and - if
they so choose - to avoid paying the high costs of on-peak power.
The result of this lack of price signals is that consumers pay very high
prices for very inefficient use of capital invested in power plants. Electric
generating plants are among the least efficiently used capital in the country,
operating on average only 46 percent of the time. This low figure compares
to average industrial capacity utilization in the U.S. of about 83 percent.
Improving this efficiency represents one of the most important sources of
savings in the deregulated electric industry. History shows that price signals
will accomplish this result. For example, following deregulation the U.S.
airline industry increased its capacity use from 48 percent to 73 percent, over
a 50 percent improvement.
Pricing in Competitive Electricity Markets 159

4. MITIGATING MARKET POWER THROUGH

NEW TECHNOLOGY

Fortunately, new technology enables innovative competItive electricity

suppliers to deliver, and consumers to take advantage of; capabilities that
can help combat market power. The first of these, the Internet, enables very
low cost information sharing and data exchange. The second, low cost
advanced metering, enables consumers to respond to high peak power costs.
Should they choose, they can just say no to paying for those costs by
reducing or even eliminating energy consumption at those times.
Internet: ESPs are beginning to pioneer the use of the Internet as an
operating model for retail electricity sales and customer service. In contrast
to the simple addition of a Web site to an electricity retailing operation based
around paper and telephones, via the Internet, companies can recruit, sign
up, serve, bill, and support customers at costs that are as much as 90 percent
lower than traditional utility customer service costs. The Internet operating
model can even involve collecting information that enables its users to
forecast peak power consumption and offer savings commensurate with
those estimates. Via its Web site, the Internet-based company educates its
customers regarding the use of energy and peak energy and how those
customers can reduce such usage.
However, the real power of the Internet will be felt when the customer has
access to real time energy data, prices and availability and easy to use tools
to manage their own demand and supply. Coupled with deregulation (read
open competition), easy to use Internet technologies can enable the
consumer to search, choose and manage their energy suppliers. Some firms
are pursuing the Priceline-motivated reverse bidding procedure, where
consumers decide that they only want to pay x amount for their energy and
utilities and ESPs from around the country bid on that consumer. With the
Internet, we could probably see this model being serviced for even
residential consumers. Power cannot be stored, so it lends itself to a model
where the surplus can literally be auctioned off. Technology, educational
tools and the Internet can create a paradigm shift in the Industry, as we know
it.
Innovative metering: The Internet can be combined with offering
innovative metering technology to customers. At a cost as low as one to two
dollars per month in many situations, advanced metering services are
affordable to even the smallest energy users.
Wireless technology also enables many other data services, including
smart, communicating thermostats. These devices are the homeowner's
equivalent of a building energy management system, but at a cost and level
of simplicity suited for the small consumer.
160 Pricing In Competitive Electricity Markets

This technology exists and is being deployed in scale today. Millions of

residential, commercial, and industrial energy users now have their meters
read remotely via radio technology as often as every five minutes. With
their meters on line, these customers have the technology in place to receive
several new services, some of which are already being offered to them by
innovative ESPs.
These energy consumers receive detailed energy usage information to
help them better manage their bills. They could receive an energy budget,
updated daily. In some cases, they no longer have to call the utility to report
an outage - and, after an outage, the utility knows for sure that the
customer's power is back on. Some ESP customers receive discounts for
charging electric vehicles during off-peak hours. Utility.com believes that
two of the most important ways for customers to realize the full benefits of
competition are to mitigate the market power of incumbent suppliers and to
make new technologies. It enables customers to reduce costs and increases
the number of choices ESPs can offer them.

5. MITIGATING MARKET POWER THROUGH

DEMAND RESPONSE

Consumer demand response has great potential as a tool to mitigate

wholesale price spikes. Such spikes typically occur during critical peak
times when systems reserve margins are reduced. If during the Midwest
wholesale price spikes of June, 1998, only 8 percent of the customers had
faced real-time prices, and on average exhibited an elasticity of demand
equal to .2, wholesale prices would have been reduced by almost 67 percent
from the peak value of $7,500 per MWh. California's competitive
wholesale market, the Power Exchange (PX), has exhibited similar price
responsiveness to customer demand; on July 28, 1998, for example,
wholesale prices increased by 83 percent from noon to 1:00 P.M., even
though demand increased by less than 2 percent. In an internal study, the PX
found that as little as a 3 percent reduction in peak demand could save
almost $8 million per day during the summer critical peak period.
Significantly, wholesale price spikes - in the absence of demand
response - are not an isolated problem confined to events in the Midwest;
deregulating markets around the world, including the u.K. and Australia,
have experienced such wholesale price spikes. Importantly, such price
~pikes are not any different from the regulated past; they simply allocate the
cost of the peaking power plants - many are used less than 100 hours per
year - to the hours in which they are used (under regulation, those costs are
averaged over the year and paid by all customers, regardless of whether they
Pricing in Competitive Electricity Markets 161

are using energy at times of system peak). Moreover, every customer

benefits from reductions in hourly wholesale prices, even though the peak
demand reductions are provided by only a subset of customers.
Demand response has great potential to mitigate price spikes in the
ancillary services markets as well. In California, such prices have reached
$9,999 per MWh. Utilities have always called on customer load reductions
during critical peak times through curtailable and interruptible rates,
resulting in thousands of megawatts of additional peaking power in the U.S.
Until recently regulators have placed little emphasis on demand-side bidding
for ancillary services. However, the Market Surveillance Committee of
California's Independent System Operator has called for increased ability for
market participants to bid into the ancillary services market; demand-side
bidding would be a simple and cost-effective source of ancillary service
bidders.
Federal agencies have already called for further emphasis on demand-
side activities as an important tool to mitigate market power. For example,
the Department of Justice and Federal Trade Commission advocate time-of-
use rates as one of the two most important ways of combating anti-trust
issues and market power - the other being open transmission access.
Customer Response to Price Signals: In contrast to some common
beliefs, customers do change their demand for electricity depending on its
price, just as they do for other products - making it an effective tool to
mitigate the cartel-like market power of generators. Such price responses
have been documented in a wide range of studies going back to the early
1980' s. For example, Pacific Gas & Electric (PG&E) conducted a series of
studies of customer load shifting under voluntary time-of-use rates for all
customer classes over several years beginning in 1983. All of these studies
demonstrated significant load reductions during peak periods. Of particular
note is the study of such rates for residential customers, where PG&E found
an average 21 percent reduction in peak use among program participants.
This reduction is much larger than the amounts needed to influence
significantly wholesale price spikes, which usage must be in the 2 to 5
percent price range to yield significant savings. EPRI surveyed scores of
time-of-use pricing studies conducted during the 1980' s; these studies found
consistently that customers shift load to off-peak time periods in response to
higher peak prices, with residential customers having the greatest inclination
to shift load. Now, with retail competition, competitive suppliers such as
Utility.com have the opportunity to promote such pricing to consumers.
Studies of real-time pricing have revealed similar and equally compelling
results. Studies of large commercial and industrial customers found price
elasticities as high as 0.35 (that is, a 3.5 percent decrease in consumption for
every 10 percent increase in price). Virginia Power found in its study that
162 Pricing In Competitive Electricity Markets

large commercial and industrial customers "reduced their on-peak load

during the 'critical' days by approximately 40 percent"!
As noted above, residential customers are especially price sensitive.
Fewer, but some, real-time pricing studies have been conducted on these
customers. The results are consistent with studies of time-of-use pricing for
residential customers and real-time pricing for commercial customers. An
example is American Electric Power's (AEP) study. AEP used technology
that automatically responded to price signals, making it as simple as possible
for customers to benefit from real-time prices. An example is automatic
adjustment of the thermostat in summer: 72 degrees for low electricity
prices, 74 for medium, 76 for high, and 80 for critical peak prices. Peak
demand reductions were dramatic: between 50 and 60 percent during peak
times - and savings even more so: customers in the program saw bill savings
of approximately $175 per year.

6. CONCLUSION

Electric deregulation coupled with technological advancement has great

promise, as it has in other industries, for reducing prices and unleashing
markets to develop innovative products and services. Market power -
vertical, horizontal, and the cartel-like market power of power generators in
the absence of demand response - threatens to reduce or eliminate the great
potential for the benefits of competition. The Internet and development of
user-friendly technological tools put the power in the hands of the consumer,
shifting the monopoly/cartel like power base of incumbent utilities.
Economists often use the word versioning and we have seen it being used on
many product offerings, most obvious being software. Energy may soon be
"versioned" to provide for group or individual "personalized" pricing. It is
now up to the Federal and state governments to work with Utilities as
consumers are getting informed and are demanding personalization rather
than accept traditional and limited offerings.

REFERENCES

1 Press Statement. Stanford Center for Economic Policy Research. Professor Frank Wolak.
January 17. 1997.
SECTION IV

RISK MANAGEMENT IN VOLATILE

MARKETS
Chapter 10

Managing Total Corporate ElectricitylEnergy Market

Risks*

Alex Henney, and Greg Keers

Competitive Electric Strategies Inc., and KW International

Key words: Correlation; Co-Variance; Downside; Generation; Greeks; Monte-Carlo;

Profit at Risk; Simulation; Retail; Risk; Utilities; VAR; Variance; Volatility.

Abstract: This paper starts with a short history of the use of value-at-risk techniques in
financial risk management. The specific and often unique risk management
challenges faced by electricity companies are then described along with a
description of how they can be tackled using a VAR approach. There are
many and varied methods for making value-at-risk calculations. The paper
gives an overview of each, along with a description on their particular
suitability for electricity portfolios. In the final section some conclusions are
drawn and recommendations made as to how best to implement a VAR based
risk management approach in a power company.

The banking industry has developed a tool kit of very useful "value at
risk" techniques for hedging risk, but these techniques must be adapted to
the special complexities of the electricity market.

Alex Henney and Greg Keers

1. INTRODUCTION

This paper starts with a short history of the use of value-at-risk techniques
in banking risk management. In this second section, it examines the specific
and, in many instances, complex risk management challenges faced by
electric companies from the behavior of prices in electricity markets, and the
* This article previously appeared in the October 1998, Volume 11, Number 8 issue (pp. 36)
of the Electricity Journal, and is reprinted with the pennission of the publisher, Elsevier
Science, Inc.

A. Faruqui et al. (eds.), Pricing in Competitive Electricity Markets

© Kluwer Academic Publishers 2000
166 Pricing In Competitive Electricity Markets

character of generation and electric retailing risks. The third section

describes the main methods for making V AR calculations along with an
analysis of their suitability for analyzing the risks of electricity portfolios
and the case for using "profit-at-risk" and "downside-risk" as measures of
risk. The final section draws the threads together and explains how to look
at managing total corporate electricity market risk, which is a big step
towards managing total corporate energy market risk.

2. VALUE AT RISK i

The concept of "value at risk" was developed at the instigation of the

central monetary authorities, who were concerned about the risk exposure of
the portfolios of currencies and financial instruments of banks engaged in
derivative trading. It was formally adopted by the Basel Committee for
Banking Supervision of the Bank for Industrial Settlements in 1995, and is
used to determine the level of risk capital required to support a bank's
trading. Subsequently it has been adopted by companies involved in
commodity trading, life insurance, and pension funds.
V AR is the amount of money a trading company might lose due to the
market risk of asset prices moving against it before it can close its positions.
In practical terms V AR measures for a portfolio the worst expected loss to a
specified confidence level during a given time period under normal market
conditions. The concept incorporates two central elements of risk:

• The sensitivity of a portfolio to changes in underlying prices, which

reflects how well the portfolio is hedged (the more fully it is hedged the
less sensitive it is to price changes)
• The volatility of the underlying prices which reflects the likelihood of
large price changes

The Basel Committee for Banking Supervision recommends banks to

calculate ten working days 99 percent confidence V AR numbers - thus a
bank which has a $10 million 99 percent confidence VAR could lose more
than $10 million in 1 out of 100 periods.
Measuring risk is one thing, managing it is another. The concept is
simply illustrated in Figure 1 by showing a normally distributed profit curve
that has an expected profit of $30 million but a 5 percent chance of losing
$10 million or more. In this example, risk is managed by hedging with
contracts so that there is now an expected profit of $20 million, but the
probability of losing $10 million or more has been reduced to 2Y2 percent.
V AR methods are now applied in many risk management related activities
with varying objectives including:
Pricing in Competitive Electricity Markets 167

• Avoiding large losses through activItIes for which the risks are not
accurately reported or have not been properly controlled (e.g., the rogue
trader)
• A voiding trading strategies and procedures that could potentially result
in large losses through inaccurate risk quantification (e.g., a hedging
strategy with large and unmeasured basis risk)
• Identifying optimal hedging strategies or assessing particular hedging
opportunities in order to control financial risks as cost effectively as
possible

Profit Distribution

·20 ·10 0 10 20 30 40 50 60 70 80

Profit ($m)

Figure 1. Risk and Regulation

The regulatory drive has naturally focused on the first two since they have
caused sudden and large losses (e.g., Barings, Orange County, and
Metallgesellschaft) and many financial institutions have implemented risk
management practices based on V AR principles. The choice of V AR
method and the way it is implemented depends in part on the relative priority
placed on the three objectives together with other factors such as accuracy,
time to calculate, and the costs of implementation. Many companies have
gone further and now use the approach as the key component of accurate risk
quantification and hedging optimization in their search for more cost
effective risk management in a competitive market.
168 Pricing In Competitive Electricity Markets

2.1 VAR Methodology for Banking

With considerable fanfare, in 1994, J.P. Morgan published its approach to

analyzing risk, which it called "RiskMetrics". Subsequently it has
developed the approach, which it now publishes and updates with Reuters,
providing data sets of volatility and correlation estimates for over 400
instruments ii . In its basic form RiskMetrics involves marking trades to
market, and then analyzing the exposed position against short term price
forecasts to measure change in value of a portfolio in terms of price changes.
The V AR calculations are based on a 1.65 standard deviation one-tailed test
"since risk only measures a negative outcome, one side of the distribution is
excluded." The approach is often termed "Delta-Normal," discussed further
in Section 4.1. Following its introduction, RiskMetrics has become the
umbrella name for a series of VAR methodologies, from the Delta-Normal
method to "full valuation" methods for portfolios whose price and return
distributions are neither symmetrical nor normally distributed. The closer
the models match economic reality, the more accurate are the estimated
VAR numbers but they are never completely accurate iii , as shown in Table 1.

The following table shows the significant differences in the V ARs calculated in two ways for
the equity and foreign exchange derivati yes portfolios of a bank.

Table I Value at Risk (confidence limit 99 percent, for a 2 week period)

Methodology Equity Foreign Exchange
($m) ($m)
Correlation 16 48
Historic Simulation 27 29

3. ELECTRICITY MARKET RISKS iv

Many North American electric companies first came into contact with the
concept of V AR through setting up affiliates whose purpose is to trade
physical or financial wholesale contracts speculatively for a profif. Many of
the traders have been recruited from financial companies where people are
familiar with the analytical VAR methodology, and from oil and gas trading
where the financial concepts have become part of the culture. In
consequence the concepts of RiskMetrics and of valuing option contracts
(and portfolios) using "delta equivalents" and perhaps calculating one or two
other "greeks"vi have been adopted by many power marketers.
That said, analytical V AR methodologies should not be treated as a
cookbook solution just because they are incorporated into a number of
trading packages. Rather they should be handled with caution. J.P. Morgan
Pricing in Competitive Electricity Markets 169

advises it is not appropriate "if the user's portfolio is subject to non-linear

risk to the extent that the assumption of conditional normality is no longer
valid. Then the user can choose between two methodologies - Delta Gamma
and structured Monte-Carlo ... [these] approaches, typically referred to as
"full valuation" models, rely on revaluing a portfolio of instruments under
different scenarios. How these scenarios are generated varies across models,
from basic historical simulation to distributions of returns generated from a
set of volatility and correlation estimates. Full valuation models typically
provide a richer set of risk measures since users are able to focus on the
entire distribution of returns instead of a single V AR number."
In reality, wholesale trading is but a small part of the risk for most
sizeable electric companies which own generation assets and have retail
books, because the inherent risks in generation and in retailing are likely to
be very much greater than a prudent company should allow its speculative
traders to incur. V AR methods have been developed in Scandinavia from an
entirely different perspective to North America, namely for managing total
corporate risks of generation assets and portfolios of customers along with
wholesale trading in an electricity market which has volume risk of +20
percent in production and enormous price risk (within the last six years the
annualized average of the spot price has varied by a factor of four) with an
aim of maximizing returns while controlling risks. This has lead to
modifications and extensions of V AR methodology that are now firmly
established, and that have been used for several years. These are described
below.

3.1 Electricity Market Price Risks

What then are the characteristics of electricity markets and prices?

• Most markets are immature, are often subject to structural change as part
of an ongoing process of liberalization, and in some places are affected
by political intervention
• There are a number of price drivers which function in a complex
manner. Electricity spot and forward prices are not generally normally
or log-normally distributed. They are skewed, fat-tailed and exhibit
mean reversion. There can be large and sudden step changes in prices
including price spikes, and these - rather than random volatility - are the
major risk concern
• There is frequently a lack of relevant historic data on forward price
movements, and consequently the volatilities of forwards products and
170 Pricing In Competitive Electricity Markets

accurate estimates of the inter-product correlations required for

analytical V AR methodologies are not available
• Electricity markets are often decentralised across different regions with
transmission constraints between them, and consequently there is basis
risk and possibly also currency risk

Furthermore there are fundamental differences between electricity and

most other traded products, some of which are the consequence of the
inability to store electricity:

• The relationship between spot and forward prices is driven by the

physical dynamics of generation and by demand, and is complex. This
situation is unlike that for products that can be physically stored
relatively easily, where the spot and forward prices are linked by the cost
of storage. Nor is it like money markets, where the spot and forward
prices are linked by relevant interest and exchange rates

• Electricity spot prices exhibit high degrees of daily, weekly, and

seasonal variability (i.e., there is typically a consistent shape to daily,
weekly and annual spot price profiles), and depending on the particular
regional spot market, volatility may exhibit similar variability. Because
the spot price variability is expected, electricity forward price curves
also exhibit daily, weekly, and annual seasonal profiles. Furthermore
the extent and nature of this seasonality in market prices and volatilities
can change dramatically under different market conditions. The extent
of the forward price variability is such that whereas in currency or crude
oil forward price curves the price twelve hours into the future might be
0.1 percent different from that twenty-four hours into the future,
differences of 1000 times greater than this (i.e., 100 percent) are not
untypical in electricity forward price curves due to the daily pattern of
the spot price (viz. the daily peak price might be twice or more as high
as that for off-peak), and the size of the difference between day ahead
forward peak and off-peak prices is not constant and predictable but
changes hour to hour and day to day'ii. Although there are three layers
of variability in electricity prices, even single layer variability is often
avoided by published works on forward price theory, and the variability
component is frequently left as an input into forward price models. Yet
the variability is by far the most significant component of the electricity
forward price curve, swamping all the relatively subtle effects of risk
free interest rates and volatility based risk adjustment. Since the
variability is driven by physical market dynamics, it can only be
effectively analyzed by modeling the physical dynamics of a regional
market.
Pricing in Competitive Electricity Markets l71

3.2 Generation Risks

V AR methods were originally designed to measure how much money

might be lost before the risk in a trading position can be closed. But for a
fundamental player in the market with assets that give it a natural position,
the concept of closing a position and eliminating risk is not usually practical.
For example, a gold mining company has a natural long position in gold and
an electricity generator has an underlying long position in power. Short of
selling out of the business or selling a life of plant contract, it is not
generally possible to close these natural positions in the medium term
because there are volume risks due to:

• Reliability uncertainties such as forced outages

• Fuel cost and market demand uncertainties for mid-merit and peaking
plant
• Fuel volume uncertainties for hydro plant

Although analytical models based on forwards prices can effectively

approximate the operation of some type of plant that have little or no
dynamic or fuel constraints (e.g., highly reliable open cycle gas turbines with
reliable gas supply at constant cost might be analytically modeled as options
using Black-Scholes formulae), they are the exception. The production from
many plants will not only depend upon the spot price, but may be correlated
both positively and negatively with it (e.g., if a large base load plant falls
over the spot price may increase, or conversely if the spot price increases a
peaking plant may start up). Thus unlike (say) a gold mine where
production may be for stock and sales may be from stock, frequently the
output and the sales income of a plant is instantaneously dependent upon the
spot price in complex waysviii, and analytical methods are rarely available for
describing the relationship. Furthermore there may be highly complicated
physical contracts incorporating considerable flexibility in terms of price,
volume and timing of deliveries that may dictate how some mid-merit and
peaking plants operate and which cannot be priced or risk modeled by
established analytical methods. Consequently it is necessary with many
plants to simulate the dynamics of their behavior and response to spot price
(and vice versa) in order to provide accurate profit forecasts and V AR
calculationsix . Thus trying to model plants in forward markets as if they
were tradeable derivative instruments introduces inaccuracies and modeling
risks. Normally portfolios that include generation will require simulation of
output under different spot price scenarios to calculate V AR.
Since the volume risk in a generation portfolio cannot be eliminated, a
seemingly closed position may not be a minimum risk position, and for many
generators (especially hydro units) the minimum risk position is likely to be
172 Pricing In Competitive Electricity Markets

long compared with average expected output. This strategy reduces the risk
of being caught short when probably others are also short and consequently
prices are high, which can result in lower profits, if not losses. Thus the
fundamental player should devise a suitable hedging strategy to aim for an
acceptable expected risk/expected profit position - a very risk averse
company may aim for a minimum risk position, while a risk acceptor would
accept more risk for the prospect of more reward.

3.3 Power Retailing Risks

In power retailing the main risks are the scope for mispncmg,
administrative confusion when many quotations (including mUltiple
quotations to the same customer) are made in a short period of time, credit
risks, and uncertainty in the volume and profile of customer demand. Retail
contracts are not straightforward and defined products; usually they are for
fixed prices but variable volume. Generally, they are priced based on
assumptions about volume and the profile of the volume over time, but for
many customers the actual volume and profile will vary depending on
factors such as the economy, the level of electricity prices (for some
customers' contracts there may be complex correlations between load -
hence potential exposure - and spot prices); and the weather. Generally
when demand is high (e.g., when there is cold weather in a winter peaking
system or hot weather in a summer peaking system), spot prices will be high.
But there also can be negative correlations when very large plants are shut
down because of strikes or major maintenance. There are also correlations
between the errors of load and the errors of spot price forecasts, which
results in non-normal profit distributions even if market prices exhibit
normal distributions. The system should thus incorporate profiled load
forecasts along with the facility to allow them to be modified in the light of
the weather and of economic circumstances. Combining this volume effect
with the non-normal price distributions causes even more distortion of the
distribution or profits away from normal.
The power retailing risks may be covered by a mix of relatively
straightforward purchase contracts, some complex contracts, and limited -
but controlled - exposure to spot prices. As with generation, the concept of
eliminating price risk is generally not relevant X short of liquidating the
business.

4. VAR METHODOLOGIES

The most usual ways of calculating V AR are:

Pricing in Competitive Electricity Markets 173

• Delta-Normal method, also referred to as the "variance/covariance"

"volatility/correlation" method. It is a "parametric" or "analytical"
method and is based on simplifying assumptions of market price and
portfolio characteristics discussed below
• Historical simulation is a "full valuation" method that involves revaluing
the full portfolio of instruments under historical conditions. It requires
no simplifying or approximating assumptions
• Monte Carlo simulation is also a "full valuation" method
• Stress testing is based on simulating a set of subjective scenarios of
circumstances that are usually thought to be extreme. Again it is a "full
valuation" method

As stress testing is conceptually simple it does not require explanation,

but the first three do and are described in the following three sub-sections
followed by two sections considering what performance measure should be
used to measure risk - the case for "profit at risk" - and how risk should be
measured in simple terms when distributions are likely to be irregular - the
case for "downside risk".

4.1 Delta-Normal Method

The standard implementation of this method involves the following

assumptions:

• Market prices are normally or log-normally distributed (i.e., over time

they follow a continuous random walk with no jumps)
• Quantities are independent of market price (as with swaps, but not with
options, caps, floors, retail or generation)
• The profitability or returns of the portfolio are normally or log-normally
distributed
• All elements of the portfolio can be accurately represented by some
linear combination of a set of standard products
• There are liquid markets for the standard products from which volatility
and covariance parameters can be estimated

These are convenient assumptions because they allow the portfolio

behavior to be described and manipulated relatively easily, and a VAR
figure for an individual contract or a portfolio can be calculated from the
total position in each of the standard products, from the estimated volatilities
or variances of each of the standard products based on historical data, or
174 Pricing In Competitive Electricity Markets

from a matrix of correlations or covariances between all of the standard

products based on historical data.
The main attraction of the method is its relative simplicity, and this often
results in its use in situations where the assumptions are known to be
somewhat flawed xi . Several extensions to the method are sometimes used, of
which the most common is often referred to as the "Delta-Gamma" method
which allows for portfolio quantities that are dependent on market price.
Thus, in principle this extension largely overcomes the serious errors in the
Delta methods for quantifying portfolios of options, but it does so at the
expense of requiring data that is generally difficult to obtain.
The Delta-Normal and Delta-Gamma methods rely on the availability of
parametric models to map all portfolio items into positions in a set of
standard products. For simple electricity portfolios that only include
baseload swaps and options thereon, this is not too serious a problem. For
example, exchange traded futures products can be used as the underlying
standard products. Most portfolios will not, however, map perfectly to these
products. For example, a portfolio might include:

• Non-baseload contracts (e.g., contacts for peak hours or profiled hours)

• Contracts for non-standard time periods (e.g., February 3 to March 12)
• Contracts, and/or retail sales and/or generation in regions or at nodes that
have basis risk with respect to the standard products
• Complex contracts that include flexibility or risk sharing terms
• Physical contracts that are subject to transmission and dynamic
constraints
• Generation and/or retail portfolios

Even if the portfolio is very simple, the assumption of normally

distributed market prices may be stretched. But assuming that is accepted,
then option contracts can be modeled using models such as the Black-
Scholes option pricing model (along with assumptions of a perfect capital
market and predictable interest rates with constant volatility). Because of
the invalidity of underlying assumptions, the Delta-Normal V AR
methodology is not ideal for managing the risk in electricity portfolios and in
general there are significant modeling risks associated with an over-reliance
on the method. That said, when the absolute accuracy of a risk measure is
not of paramount importance and a trend over time in the relative level of
risk is what is required to provide a wanning signal, then analytic V AR may
be acceptable because it is easy to implement and quick to calculate.
There are, however, no such equivalent models for more complex
electricity contracts, and certainly none for retail contracts based on fixed
prices and metered demand or for generation plants. In consequence
approximating retail and generation to combinations of swaps and options
Pricing in Competitive Electricity Markets 175

can result in significant and unquantified modeling errors, and so an

analytical approach to measuring risk is impractical. Generation and retail
contracts should be modeled against possible spot price scenarios and risk
assessed using a simulation approach which can handle the correlations
between factors, skewed and fat tailed price distributions, and the mean
reversion of electricity market prices that are conserved in competitive
electricity markets. This approach eliminates the errors resulting from the
invalidity of the assumptions made by parametric V AR methods.

4.2 Historical Simulation

With this method, historical prices are applied to the current portfolio to
assess what the outcome would be if history repeated itself. The length of
the historical period chosen is very important - if the period is too archaic, it
may not capture the full variety of events and relationships between the
various assets and within each asset class, while if it is too long the data may
be too stale to relate to the future. The advantage of this method is that it
does not require any explicit assumptions about correlations and the
dynamics of the risk factors because the simulation follows every historical
move. The disadvantages for electricity are two-fold. First, with newly
created and changing electricity markets, little historical information may be
available, and when it is available it may be invalid due to changing market
conditions. Second, there may be obvious reasons (e.g., changes in market
structure, new plant coming on line) why the future will differ from the past.

4.3 Monte Carlo Simulation

The Monte Carlo simulation method calculates the change in the value of
a portfolio using a sample of randomly generated price scenarios that are
assumed to be equally probable. The approach requires making assumptions
about market structures, the stochastic processes the prices follow and the
correlations between risk factors and the volatility of these factors. The
relationships are derived from econometric estimation, using historical data,
and/or inferred from current market variables (for example option prices).
Monte Carlo analysis allows combining historical information with market
expectations, but it is computationally demanding.
A common criticism of Monte Carlo simulation is that it requires
subjective input in the choice of simulation model structure. Where the
V AR outputs are used as part of an ongoing risk monitoring operation it is
important that the model structure is changed as little and infrequently as
possible to avoid spurious step changes in the calculated VAR. Normally,
after an initial period of developing and testing different model structures, a
176 Pricing In Competitive Electricity Markets

structure can be used for some time. If market conditions change

significantly or new model structures become available then the approach is
flexible enough to allow their introduction. It is worth noting that the
sUbjective choice of model structure is invariably made because it is known
to be more accurate than the simplistic structures applied by parametric
methods. The setting of the model parameters could be done subjectively as
could the volatility and correlation parameters to a Delta-Normal approach.
However, in operational circumstances they should clearly be set objectively
based on market analysis or model fitting to relevant historical data where
possible.
Jorion (op cit.) comments "Monte Carlo analysis is by far the most
powerful method to compute value-at-risk. It can account for a wide range
of risks, including non-linear price risk, volatility risk, and even model risk.
It can incorporate time variation in volatility, fat tails and extreme
scenarios ... [but it] is the most expensive to implement in terms of systems
infrastructure and intellectual development".

4.4 Profit at Risk

As explained earlier the ongm of V AR was a concern by banking
authorities about the capital adequacy of banks measured by closing out a
loss in a given period, and financial institutions now revalue and mark their
portfolios to market daily. But:

• As we have discussed, the elements of an electricity company's portfolio

cannot be "closed" in any liquid forwards markefii , and there is no
obvious holding horizon for a V AR calculation. Since market price and
volume uncertainty ends once physical delivery occurs the time to this
event is usually taken for the holding horizon. Consequently market
price simulations that represent how much the price might vary between
now and delivery are more appropriate than between now and some
arbitrary period (say two weeks)
• Most electricity companies track profitability (e.g., retail sales margin or
generation spark spread) as their primary performance measure rather
than the value of their portfolio (no electric company marks to market its
generation assets and retail portfolio on a regular basis to give a real
time asset valuation)

Thus it is more appropriate to use a V AR measure that indicates directly

the risks in achieving expected/forecast results. Using market price
simulations that represent how much the price might vary between now and
delivery can provide a "profit at risk" figure.
Pricing in Competitive Electricity Markets 177

4.5 Downside Risk

Risk is related to dispersion of possible outcomes - the greater the

downside dispersion from the expected average, the greater the risk; the less
the downside dispersion from the expected mean, the less the risk and with
normal distributions it is measured by the standard deviation. But this metric
does not apply to distributions that are very non-normal. Thus it is
appropriate to describe the "downside profit" in a simulation approach as a
parameter such as:

• The Xth worst case or Xth percentage worst case

• The expected mean of the Xth (or Xth percentage) worst cases, which
provides a measure of smoothing

Then the "downside risk" can be defined as the difference between the
expected outcome and the "downside profit" to give a measure analogous to
the standard deviation of a (log) normal distribution, see Figure 2 (next
page).

5. MANAGING CORPORATE ELECTRICITY AND

ENERGY MARKET RISKS

We can now draw the threads together for a company that generates,
retails, has a speculative trading book, and a hedging book, see Figure 3
(below), and may also have some complex bilateral physical contracts that
cannot be readily assigned. The exhibit illustrates that while the
expectations of profit are additive, the expectations of risk are not. There is
some diversity among plants and consequently risk is reduced by owning
more than one unit, and there is diversity of risk among customers and which
attenuates the risks of the retail portfolio. Since generation is inherently
long, and retailing is inherently short they should to a degree provide a
mutual hedge. The speculative trading book is an activity in its own right
which should generally have limited risk exposure and can be controlled
against limits, but the hedging portfolio is designed (as its name indicates) to
hedge the total corporate risk. Consequently it may show a large risk, which
is offsetting the "net risk" of generation and retailing, and an expected loss
(which is the cost of "insuring" to reduce risk of seriously poor
performance). It is of fundamental importance to appreciate that in a fully
deregulated electricity market, the risks of generation and of the retail
portfolio will be an order or two greater than that of a properly controlled
wholesale book.
178 Pricing In Competitive Electricity Markets

9 "Profit at Risk" = Expected - Downside

.
.l!!
7

.
Q.

~ 5

'0 4

E
~
II)

3
z" 2

0
0 a 0 a a a a a a a a a
a a a 0 0 0 0 0 0 0 a
0 0 0 0 0 0 0 a 0 0 0
a 'Q
'" ~ ~
'" '"'"
0 a
'" '"'" ....
0 0
'"
't t Profit $000
Downside Expected
-7000 10000

Figure 2. Expected Profit, Downside Profit and Profit at Risk

Generation Trading Retail

..L..L 66 Hedge Speculate

Profit $900 $2000 $5 $30 $700 $300 $250

Risk $300 $450 $700 $30 $320 $190 $80

Profit $2900 $35 $1250

Risk $700 $725 $590

Profit $4185
Risk $630 Figures in ooos
• Mark the portfolio forward positions to market or to forecast prices
• Consistent forward price assumptions must be used for profit calculations
• The profit figures are additive, risk is not
• Integrated portfolios may Include natural hedges

Figure 3. Profit and Risks from a Set of Portfolios

Pricing in Competitive Electricity Markets 179

.
.!!
7

...
Q.
E
5
'0 4
:,;
.c 3
E
z" 2

0
0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0

.,.
0 0 0 0 0 0 0 0 0 0 0
0
';"
"? '" ~ '" 0
N '"
N
0

'" '"'"
0 0

'"
Profit $000

Figure 4. Portfolio Profit Forecast for the Existing Position of the Portfolio and After
Possible Change (112)

Cummulative profit distribution

35

en
.!!
30

25
- -~-=.....:;:---

."'
Q.
E
20
'0
:,; 15
.c
E 10
z"
5
y
'/
0
0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0
0 U;
'"
0
'" 0
"' '" "''"
0
..,.
0 0
';" N N
"'
Profit $000

Figure 5. Portfolio Profit Forecast for the Existing Position of the Portfolio and After
Possible Change (212)
 180 Pricing In Competitive Electricity Markets

Given the close link between the spot price, generation output and some
retailing loads, the total corporate book of generation, of power retailing, of
speculative trading, and of hedging portfolios should be simulated together
to show the financial outcome for each element of the portfolio and the total
for user defined portfolios (e.g., hydro generation, thermal generation,
industrial customers, residential customers, hedging portfolio etc.).xiii The
simulations of financial outcome are then used to calculate an expected
profit and a downside outcome. Given the appropriate computational tools,
the total corporate electricity market risk can now be managed by analyzing
the consequences on the profit at risk, as in Figures 4 and 5 (above), of:

• Changing generation strategy (e.g., changing outage planning, investing

in new plant, retrofitting, dual fuelling, tolling, leasing plant)
• Altering retailing strategy by volume and type of customer, (e.g., target
more metal meiters, fewer steelworks, more supermarket chains)
• Buying or selling different volumes and types of contracts (swaps,
options, caps and collars, swing contracts) on the wholesale market to
hedge the position. In valuing these contracts, they should be judged not
so much by their general market value as by how well they hedge the
risk of the particular portfolio of plant and retail contracts

If the aim is to achieve a minimum risk position, then this can be found,
but many wish to go for more profit if it can be achieved without incurring
disproportionate risk.
The total corporate market energy risk can be assessed by creating - and
relating - the various electricity books with the books for fuel purchasing,
trading and retailing both for a country and where appropriate across
countries taking account of exchange rate risks.

REFERENCES
I This section draws on:
Managing Derivative Risks: The Use and Abuse of Leverage, Lillian Chew, John
Wiley & Sons, 1996, which is an overview of the commercial occurrence and
management of risk
Value at Risk, Philippe lorion, Irwin, 1997, which is a text book on V AR
VAR - Understanding and Applying Value at Risk, KPMG Risk Publications, 1997,
which comprises a series of papers on financial applications of V AR as well as the
latest source document on RiskMetrics™, which is referenced on p. 3.

ii RiskMetrics™ - Technical Document, Fourth Edition, \996, New York, December 17,1996
available free of charge on l.P. Morgan's web page at http:/www.jpmorgan.comlRisk
ManagementiRiskMetrics.
Pricing in Competitive Electricity Markets 181

iii The Basel Committee for Banking Supervision asked fifteen banks from major G-IO
countries to produce VAR numbers for a sample portfolio of approximately 350 positions,
and to test four different variants of the portfolio; balanced and unbalanced; and each with
and without option positions. The returned results were widely dispersed because each
firm used different correlation assumptions, volatility data and treatment of options.

iv Energy Risks, Dragana Pilipovic, McGraw-Hill, 1998, covers the quantitative ground.

v Note two other different meanings of the word trading are: 1) physical sale of surplus or
purchase of deficit power, and 2) hedging to offset the risks of a portfolio of generation
assets and/or of retailing contracts. Also, selling retail contracts is sometimes called retail
"trading".

vi The "greeks" are as follows:

the "delta" measures the sensitivity of the change of value of an instrument (e.g.,
option) or portfolio to change in either the spot price or the forward price as
appropriate. The delta is thus the partial derivative of portfolio value with
respect to the market price of the underlying product (e.g., the market price of an
electricity futures contract). Delta is related to position in that the product of an
option contract delta (e.g., 0.4) and the contract size (e.g., IOMW) gives an
approximately equivalent position in a swap contract (e.g., 4MW), the "delta
equivalent" position
the "vega" risk measures the sensitivity of change in value of an instrument or
portfolio to change in volatilities, and is thus the partial derivative of portfolio
value with respect to volatility
the "theta" measures the change in value of a portfolio with respect to time as
options tend to decay in value the closer they get to expiration
the "rho" measures the change in value of a portfolio with respect to a change in
the discount rate
the "gamma" is the derivative of delta, and is thus a "second order" analytic.
Portfolios that have non-zero gamma (e.g., options, caps, floors, electricity retail
and generation) exhibit non-linear behavior that causes inaccuracies in analytical
or parametric V AR methods

vo Taken to an extreme, electricity forward prices and their respective volatilities can be
analyzed down to an hourly or half-hourly resolution. But collecting, processing and
forecasting market information to any useful degree of accuracy at this detailed resolution
is impractical beyond a horizon of a few days. It is therefore common practice to group
hours into time slots which prima facie results in a degree of approximation when using
such forward prices and volatilites for V AR calculations. In fact, the accuracy and
precision of VAR calculations is generally improved by working with time slots because
the reduced amount of data allows for higher quality analysis through more accurate
modeling of price uncertainties and distributions. The optimum time slots for risk analysis
depends on regional market characteristics and the makeup of the particular portfolio
being analyzed. In genera!, grouping hours that have similar price levels is largely
consistent with accurate profit forecasting and risk quantification (e.g., weekday peak,
weekday off-peak, weekend).
 182 Pricing In Competitive Electricity Markets

viii Normally irrespective of the degree that output is financially contracted, but possibly
effected by:
physical contractual commitments
market share motivations overriding those for operating profits

IX Because of the dynamic nature, this modeling has to be done on spot prices and cannot be
applied to forwards prices, which have different stochastic behavior. For example,
forwards prices at a year hence will not be as volatile as spot prices and will not spike up
to extreme levels.

x The way of eliminating electricity price and volume risk is to act as a marketing front for a
generator who assumes those risks.

xi According to Jorion (op cit) a part of Baring's portfolio losses were caused by Nick Leeson
selling 35000 put and calls in equal volumes with similar strike prices. A Delta-Normal
V AR calculation of this portfolio would have reported zero risk (the puts have -ve deltas
and the calls +ve delta), but a full valuation simulation approach showed the real V AR to
be $6.3m.

xu Note that trading standard wholesale contracts to match the latest forecast position does not
result in a "closed" position since considerable risks remain due to volume uncertainties.

xiii In no other commodity market are the reasons for an integrated analysis so compelling.
For example, oil companies do not need to undertake mark-to-market and V AR analysis
that includes petroleum stored at highway retail outlets, and a gold mining company is
unlikely to include planned mine output in its market-to-market and V AR analysis. It is
the lack of the buffering or insulating effects of storage that underpins the need for this
approach with power portfolios.
Chapter 11

Managing Weather Risk in Energy Pricing ... A

Consumer Oriented, Value Added, Energy Service

Duncan P. MacArthur
WeatherWise USA, Inc.

Key words: Energy Pricing; Retail Energy Products and Services; Weather Risk.

Abstract: Weather is the critical factor in wholesale and retail energy pricing. Weather
affects the quantity of energy used and the wholesale price. With
deregulation, new ways to manage price risks are emerging for consumers and
energy marketers. These include eliminating the risk of price and weather
changes in consumer energy bills, quantifying the risks in wholesale energy
buying and marketer user of degree day transactions.

Weatherproofing consumer bills shifts risk from consumers to energy

marketers, who can better support sophisticated planning. Weatherproofing
technology also provides a methodology to quantify the risk of various
strategies for supplying energy. In addition, the energy marketer can estimate
the size and value of a degree day transaction based on the risk the marketer is
willing to accept.

" ... everybody talks about the weather, but nobody does anything about
it. "
Mark Twain

1. INTRODUCTION

Weather is the critical factor in the energy industry. We know the energy
industry is weather sensitive, every hot summer brings a new story of
"brown outs" and in cold winters the spot price of gas goes through the roof.
Weather affects the price and quantity of energy used by wholesale and retail
consumers, it affects the economy when industry or user budgets are cut and
A. Faruqui et al. (eds.), Pricing in Competitive Electricity Markets
© Kluwer Academic Publishers 2000
184 Pricing In Competitive Electricity Markets

it even affects stockholders in energy companies when earnings fall and

stock prices drop due to weather. With deregulation and privatization, new
ways to manage price and other risks are emerging for consumers and
energy marketers. These include eliminating the risk of price and weather
changes in consumer energy bills, quantifying the risks in wholesale energy
buying and marketer use of Degree-Day transactions. These weather risk
management instruments give the energy marketer a way to do something
about the weather. They have a profound impact on energy pricing and can
provide a competitive advantage for the innovative energy marketer.
This highlights the market opportunities in the energy industry for
weather risk management as a key element of pricing and marketing
strategy. It also summarizes market conditions that may restrict the growth
of innovative pricing instruments. The conclusion is that the energy industry
needs new products and needs to be more customer driven. In addition, to
capitalize on deregulation, the energy industry needs assistance in marketing
and managing those products - in all functional areas - from personnel to
customer service.

2. THE ENERGY MARKETPLACE

Right now, the trillion dollar energy industry is undergoing tremendous

upheaval, worldwide. Deregulation and privatization are fostering
monumental changes in the way energy is produced, distributed, sold and
used. Utilities are selling off generation operations, becoming pipes and
wires companies, and leaving retail sales to energy marketers, including
affiliates and new entrants. All this massive change is happening in an
industry with a long reputation for stability.
Now, deregulated and privatized retail energy marketers are faced with
competition, new service offerings, customer switching and leftover
regulation. In addition, as utilities are deregulated, they lose the ability to
pass along risk. Deregulation and customer demand for stable pricing is
forcing risk management in offerings, including fixed forward unit pricing,
swing volumes and spot market pricing. Energy marketers must respond to
these increased market demands with "thin" margins on gas and electricity.
They do not have new, high margin products.
At the customer level, deregulation is causing confusion and uncertainty.
Consumers are confused. For example, press coverage of Con Ed's retail
energy choice program has concluded that governmental and company
presentations have caused more confusion than understanding.
In the deregulated marketplace consumers are faced with new names,
new price claims, new services (telephony, Internet connections, football
Pricing in Competitive Electricity Markets 185

tickets) fluctuating prices and, of course, weather risk. Add to this a whole
new round of dinnertime phone calls - and you have a confused, uncertain
and irritated customer. This is the opposite of what customers want!
In some respects energy companies that currently serve these uncertain
customers prefer the marketplace confusion. When customers are in doubt,
they typically stay with what they know. As a result, the new energy
marketer entering the market has a tougher sales job.
The energy marketplace needs weather risk management. For example,
heating energy consumers are worried about a cold winter and increased
prices for energy. In contrast, heating energy marketers are worried about a
warm winter and decreased prices for energy. Similarly, cooling energy
consumers want cool summers and cooling energy providers want hot
summer temperatures.
Real problems are real opportunities. Consumers clearly want to avoid
confusion. They prefer certainty. Even sophisticated derivatives traders pay
a premium to go home to fixed rate mortgages. At the same time, energy
marketers need new tools to compete. We think weather financial
instruments address customer and energy marketer opportunities. Energy
marketing is entering a new age. Providing reliable supply and a
competitive price are now basic market requirements. The task is to lead the
industry into innovative, value added, energy offerings.

3. WEATHER RISK MANAGEMENT

The typical energy bill is made up of two factors, the unit price of energy
(including transport in pipes or wires) and the quantity used. For the
majority of customers the quantity used is dependent on weather. Typically,
however, price and weather related uses are multiplicative. For example,
during a warm summer, the demand for electric power increases and so does
the price. Similarly in a cold summer demand and price drop. Typically, the
end result can be a total energy cost swing of plus or minus 30 percent. We
have seen ±50 percent.
Upstream industry segments already manage weather risk. Energy
producers and transporters are active in trading markets and vary production,
storage, etc. Similarly, industrial customers have limited weather risk, since
their energy use is dependent on production rather than weather. Finally,
energy retailers were (and many still are) insulated from the financial effects
of weather. Margins are at risk, but commodity risk is passed to customers.
For example, if the summer is warm and the price of power rises, those
increases are passed along to customers. Similarly, some companies have
weather normalization, which assures earnings, regardless of weather.
186 Pricing In Competitive Electricity Markets

4. RETAIL COMPETITIVE ADVANTAGE WITH

WEATHER FINANCIAL INSTRUMENTS

It is the downstream end -residential, commercial and institutional

customers- which offer the best potential for innovative pricing strategies.
These energy consumers are not expert energy buyers. They know what
they spend, but energy is a necessary expense that they have little control
over. Most of these consumers are billed for energy but they buy comfort.
For these customers, hedging can fix the commodity price and weather
financial instruments can fix the volume of the commodity used. This can
have a strong advantage for the marketer in providing a value added energy
service (risk management) that is now only enjoyed by the largest industrial
customers. This is a service that can also generate revenue for the energy
marketer. A stable energy bill is important to residential and small
commercial customers that do not want to worry about energy costs, and it is
critical to institutions that do not want to face program changes because of
energy costs driven by weather.
In general, weather financial instruments fall into three major groups.
First is the WeatherProof Bill sM , which is a product for retail energy
consumers. The WeatherProof Bill sM also assists in buying a hedge, so the
energy marketer can limit weather risk. Weather insurance is another type of
financial instrument and is in general use for catastrophic weather
occurrences. The insurance industry is now extending the scope to include
extreme, but not catastrophic, weather conditions. Finally, options are a tool
for managing weather risk. These can be caps, which limit upside risk,
floors, which limit downside risk or collars, which combine a cap and floor.
One other factor is important. In the past, retail energy companies
essentially had one customer - the local public utility commission. As a
result, critical energy marketing was to the regulators. If one can add
millions to revenue with a rate case instead of selling to hundreds of
residential customers, who are going to "hook up" anyway, rate cases are the
most important function. Consumer marketing was (and is) minimal and
largely information dissemination on safe and efficient energy use.
Although most utilities report high customer satisfaction, satisfaction is not
marketing. Satisfaction does not retain or acquire new customers or target
new products and services to sell. In 1998, The Wall Street Journal wrote a
story that noted the last innovation in power marketing was Reddy Kilowatt
in 1926.
Pricing in Competitive Electricity Markets 187

5. SATISFYING MARKET REQUIREMENTS FOR

WEATHER FINANCIAL INSTRUMENTS

Let's summarize this picture of the market. It is the energy marketer's

customer that offers the potential for innovative pricing using weather risk
management and the only way to reach that customer is through an
organization that has limited marketing expertise.
Specific market requirements are that the consumer has weather
sensitivity. We define this sensitivity for heating as the ratio of energy use
in the three winter months to that used in the three summer months. We
look for a 2: 1 ratio. For cooling, we look for a summer to winter ratio of
2: 1. In addition, the buyer of a weather financial instrument must have the
ability to evaluate various HDD/CDD (Cooling Degree Day) strike levels
and relate these to the users own requirements. Finally, the minimum
weather instrument offering from traders we have seen is $100 per HDD
(Heating Degree Day.) A typical residential consumer has a value at risk in
the range of $0.10 per HDD. It is doubtful that most weather traders will
market direct to customers. Most will look to the energy industry to
aggregate risk, much like the mortgage banking industry aggregates
mortgages for securitization.
This is the service WeatherWise provides. WeatherWise offers a fixed,
total energy cost payment option product that is sold to retail consumers
through energy marketers. Our product, the WeatherProof Bill sM . enables
the energy marketer to eliminate weather risk for customers. Technically,
the WeatherProof Bill sM is a full requirements fixed unit energy price
forward contract multiplied by a Heating Degree-Day Swap. Eliminating
weather risk for energy consumers at the retail level gives a powerful
competitive marketing advantage when the cost effect of weather swings is
as much as ± 30 percent. Add to that the simplicity of quoting a fixed total
price, and the energy marketer has a powerful marketing tool. Simplicity is
valuable in the marketplace. Energy marketers now are having difficulty in
explaining new, dis aggregated billing terms, such as, transport price,
commodity price, and stranded cost. This confusion at the retail consumer
level is increased with weather price fluctuations and weather related energy
use. The confusion is topped off with claims that deregulation really is a
benefit. Finally, the whole confusing package is telemarketed to the
customer - at dinnertime!
WeatherWise can take all the risk including the risk of non-weather
related energy use. The degree of risk taken is reflected in WeatherWise's
fee structure. In any case, if WeatherProof Bill sM customers use more
energy because of a hot summer, WeatherWise pays the energy marketer the
difference. If they use less, the marketer pays WeatherWise.
188 Pricing In Competitive Electricity Markets

In addition to providing a powerful retail marketing tool, WeatherWise

helps the energy marketer aggregate weather risk, so that the marketer can
use the sophisticated energy trading options and degree day hedges now
coming into the market. WeatherWise's modeling software also helps power
traders create a firm cost per unit forward price. It analyzes purchasing
strategy based on the quantified level of risk the buyer is willing to accept.
The quality of these usage projections and buying strategies is, of course,
dependent on the quality of the mathematical models used.

6. REQUIREMENTS FOR RETAIL ENERGY

MARKETERS

It is important to recognize key issues in marketing weather financial

instruments. First, the quality and cost of weather risk management depend
on the quality and accuracy of forecasting models. Accurate models in tum
require accurate input data. Improving the quality of input data was a major
requirement for WeatherWise. About one-third of our 1.5 million lines of
code is for cleaning up utility information. Utility billing data is frequently
bad. It has irregular billing periods (multi-months, short corrections, partial
months, etc.,) corrections and rebills, duplicate records, misread meters and
bad estimates. In addition, utility billing and information systems are
typically the product of years of add-ons, debugging and modifications.
Access to data without disrupting the system. This requires IT expertise to
modify systems to accept, process and bill new technology.
The result of WeatherWise's technology is very accurate usage estimates
with regard to weather - within ±0.5 percent on groups of 100 or more
customers. We have found most utility forecasting systems are lucky to get
±6 percent. We run up to 500 regressions on each individual's energy usage
to determine the best model.
The final key issue is to understand consumer reactions. The energy
marketer needs to know if the customer will "relax their efficient energy use
patterns?" Knowledge of how customers will react without the forced
economies of pay-as-you-go is critical. In the insurance industry, this is
called moral hazard. WeatherWise has developed methodologies to manage
energy use moral hazard.
In addition, WeatherWise has found that in developing the market for
weather financial instruments we have to assist the energy industry and its
customers in realizing the value of weather financial instruments. This
requires bringing an in-depth knowledge of the energy business, retail
marketing expertise and political sensitivity to the marketplace. In short, our
investment in technology and marketing cannot be easily duplicated. It
Pricing in Competitive Electricity Markets 189

requires funding, time and extensive testing to Insure that model types
actually work.
It is important to note that retail marketing is still a new concept to many
energy marketers. In the past marketing was confined to consumer
satisfaction surveys and image improvements. There is little need for
marketing in a regulated, single supplier market, since customers will hook
up in any case. In the future energy marketers will find an increasingly
competitive marketplace. This will demand retail marketing skills. These
skills must include targeting new opportunities to increase margins, since the
margins on energy commodities are "thin." In addition, energy marketers
will need to learn new ways of presenting, promoting, pricing and collecting
for their products and services. As a supplier to the energy industry, we
have found we must bring this expertise with our product.
Most utility energy buyers and financial operations understand the
mechanics of trading, hedging and managing risk. However, this
understanding is not a typical skill of energy marketing operations. It is
these marketing operations that will "sell" weather financial instruments to
consumers. Hence, market developers for weather financial instruments
must have training in weather financial transactions. At the same time, the
energy marketer's traders need training in weather risk management tools for
energy purchase and trading.
Political sensitivity is another key requirement. Many public utility
commissions regard the cost of hedging and weather financial instruments as
an unnecessary increase in the price of energy. Many customers are still
served by regulated operations. Overcoming this obstacle requires
"marketing" to these public organizations. It is a delicate task. For example,
weather financial instruments are derivatives. Derivatives have a very bad
reputation. with governmental organizations. Not only is the prospect
frightening after Orange County, Barings Bank and Union Bank of
Switzerland, but also most regulators know they do not have the staff or
ability to monitor sophisticated financial transactions for public use.
In summary, developing innovative pricing and weather risk management
for the energy industry requires a broad scale internal and external marketing
effort. It is - or will be - a necessary effort for effective competition in a
deregulated environment. It is an opportunity because even the most
regulated companies will be asked by stockholders, "How do you manage
weather risk?" Which is, of course a better question than, "Why didn't you
manage the risk?" It is also important to note the success to date of non-
energy offerings, such as telephony, Internet, brand advertising, etc. has been
very weak. The energy industry needs new, energy-related products and
weather financial instruments can meet that need.
Chapter 12

An Econometric Study of Weather's Effect on Prices

Scott P. Martello
Tennessee Valley Authority

Key words: ARCH; Box-Cox; Correlation; Econometric; GARCH; Technique;

Transformation.

Abstract: This paper researches how weather affects the spot price of electricity.
Determining a detailed correlation between the two would certainly be a
valuable tool in forecasting prices. We can draw conclusions about the
relationship by applying different econometric techniques and specifications.

This paper serves as a survey of methods that an analyst may use to address
the weather-price relationship. It lists several simple model specifications that
prove quite informative. It discusses the Box-Cox Transformation method
which is a very versatile technique and can be widely applied. This technique
is considered most valuable to a broad constituency like these conference
attendees that may have a wide range of data available. Also, the paper
discusses ARCH and GARCH estimation of volatility - very insightful
technique especially valuable in financial markets.

The material is presented briefly to suggest directions of analysis to the

readers. They can obtain more rigorous information from the references or by
contacting the author.

1. INTRODUCTION

A crucial element to innovative electricity pncmg is, of course,

information. This paper seeks to provide better information concerning
electricity pricing to decision makers. Possibly the most important factor in
determining daily electricity price movements is weather. Especially in the

A. Faruqui et al. (eds.), Pricing in Competitive Electricity Markets

© Kluwer Academic Publishers 2000
192 Pricing In Competitive Electricity Markets

peak seasons of summer and winter. By exploring the econometric

relationship between weather and prices this paper attempts to advance the
search for better information by detailing the correlation and recommending
policy based on its conclusion.
This paper will not focus extensively on the academics of the matter.
Instead, it will suggest directions for research.

2. THE DIFFERENT MODELS

The models are relatively simple and straightforward. They all follow the
basic form:

price =f (c d d , h d d) (1)

or
(2)
price = f(t d d)

Where cdd is cooling degree days, hdd is heating degree days, and tdd is
total degree days. The simplest of these are basic linear regressions
including, but not limited to, the following:

price = a + {31 cdd + {32 hdd (3)

price = a + {31 tdd (4)

price = a + {31 cdd + fh cdd2 + {33 hdd + {34 hdd2 (5)

price = a + {31 tdd + {32 tdd2 (6)

2.1 Box-Cox

The analysis also includes some Box-Cox transformations. Box-Cox is a

generalized non-linear specification (Pindyck and Rubinfeld 1991) and,
therefore, is a versatile specification.
Pricing in Competitive Electricity Markets 193

Box-Cox follows the general form:

y A 1
(7)
A

It can be used for maximum likelihood estimation of non-linear models

and /... can be allowed to vary by variable. So its versatility is appealing
when exploring a new analysis such as the relationship between weather and
prices. The first step in applying Box-Cox is to try the most commonly used
values of /..., 1 and O. Of course, if /...=1 then the estimated equation is

(y-l)=a+fj(x-l) (8)

When A=O it appears indeterminate. However, through Taylor Series

Expansion you arrive at is:

lny=a+fjlnx (9)

So, when /"'=0 the Box-Cox transformation yields the log-linear model.

2.2 ARCH and GARCH

Most models in econometrics estimate the mean of a random variable.

ARCH (Autoregressive Conditional Heteroscedasticity) and GARCH
(Generalized Autoregressive Conditional Heteroscedasticity) models are
markedly different. They estimate the variance of the dependent variable.
The variance is specified to depend on past values of the dependent variable
and other independent variables. This estimation is advantageous because it
provides a direct estimate of volatility. Obviously, volatility is of interest in
the developing electricity market. It helps the analyst evaluate risk and
affects confidence intervals of estimates.
A simple and effective GARCH model is the GARCH (1,1) (Greene
1993). The (1J) refers to one ARCH term and one GARCH term and takes
the following form:
194 Pricing In Competitive Electricity Markets

2
= OJ + a
2
E , _ 1 + (10)

What this model is saying is that the variance. today depends on three
things; a constant «(0), yesterday's variance (crt}, the GARCH term), and
yesterday's "innovation" or error (E t} , the ARCH term). The ARCH term
comes from a conventional mean equation such as equation (3).
Of course, you see no error term when you look at that equation;
however, it is implied. Error terms would exist after estimation. In fact a
more thorough specification of equation (3) would be the following:

price, = a + /31' cdd + ~, hdd + £, (11)

The GARCH (1J) specification makes particular sense in financial

applications. The nature of the specification incorporates the phenomenon
of volatility clustering that is common in financial arenas. That is, large
changes are most often followed by large changes and small changes are
most often followed by small changes.
The author apologizes if the preceding discussion of the ARCH and
GARCH process is too cryptic. A step by step instruction of the process is
not appropriate for this paper. However, the Greene reference in the
bibliography should be helpful or feel free to contact the author.

3. RESULTS

It is most important to mention that results may vary. Any given data set
may dictate any of a number of adjustments that the analyst must make.
Therefore results will be discussed in a manner that considers that caveat. In
this broad discussion the paper reveals the best results achieved with a
particular data set. Of course, prudence dictates that specific results cannot
be revealed. But specific results would not be best for this paper anyway.
More importantly it discusses a recipe for achieving your own best results.

3.1 Price Prediction

The Box-Cox transformation where A=1 (equation 8) produced the best

results. That particular transformation has no effect besides changing the
mtercept of the equation. The diagnostic statistics are identical.
Pricing in Competitive Electricity Markets 195

Using A=O proved indeterminate with the data set used in this paper.
Technically speaking it gave a near singular matrix which cannot be
inverted. The point is that model did not work.

3.2 Volatility

The GARCH(l,l) specification was, in the end, unsatisfactory. The

estimated equations were very sound. All diagnostics were favorable and it
appeared to be a good model. However, its forecasts were illogically
extreme. We must explore alternate specifications to make the model useful.

4. PROBLEMS

Most of the models in the study had to be corrected for autocorrelation.

This will likely be a persistent problem in time series analysis such as this.

5. CONCLUSIONS

The models mentioned here are simple and limited. The primary purpose
is give a direction to begin analysis. The Box-Cox method mentioned above
is used in this paper for its versatility. It would be very worthwhile to
explore other Box-Cox specifications.
Also, as a general recommendation, the analyst can focus on maximizing
the adjusted R2 of the equations. Maximizing R2 is often the first thing
people think of and comes at the expense of overlooking other valuable
diagnostic statistics. However, in this case the diagnostics are very similar
across different specifications. So, that being the case, maximizing adjusted
R2 is appropriate.
Further, the GARCH specification must be explored. The volatility
prediction is a very valuable one and should be addressed. The starting point
should still be the GARCH (1,1) model. This problem may be a data
specific one and not a product of the model specification.

6. THE NEXT STEP

After going through all the research and experimentation listed above,
what is next? Well, that depends on each analyst's specific intentions.
However, the author is going to pursue a few avenues of development.
196 Pricing In Competitive Electricity Markets

First, a model should include some measure of available capacity. This

variable should most certainly influence the daily price of electricity and add
to the explanatory power of the model. If one can find data on this subject,
they may change the model from predicting the mean price to the change in
prices.
Also, a system of models may prove most useful. One may find the case
where a model is effective for small changes in weather, but not large.
Likewise, another may do well with large changes, but not small. If the
analyst applies variations of the models mentioned above, they should be
able to nlitigate that situation.

REFERENCES

Greene, William H. Econometric Analysis, New York, Macmillan Publishing Company,

1993. pp. 568-575.

Pindyck, Robert S. and Rubinfeld, Daniel L. Econometric Models & Econometric

Forecasts, New York, McGraw-Hill Inc., 1991, pp. 240-243.
Chapter 13

Electric Market Simulation

Lance S. Muckelroy
Reliant Energy

Key words: Forward Curve Development; Market Clearing Price; Structural Modeling;
Uncertainty Analysis.

Abstract: Competitive-based decision making must take price uncertainty into account.
For mature commodity markets, participants are able to use the futures,
forwards, and options markets to price assets and product offerings. These
markets provide a market clearing price and implied volatility. The electric
industry is in the beginning stages of its transition to a fully competitive
market. Thus, market clearing price projections and implied volatility are not
available in many markets, and are not available on an hourly basis in any
market. Therefore, structural modeling will be essential in the development of
market clearing price curves and price uncertainty. This chapter details some
of the methodologies and results of using structural models for electric market
simulation.

1. INTRODUCTION

A dramatically new paradigm has emerged in the electricity business.

With the exception of the electric delivery system (i.e., transmission and
distribution), public utility commission (regulatory) defined prudency-based
decision making has become, or is quickly becoming, obsolete. With this
change, decision making must take into account an uncertain, market-based
business environment that no longer offers a "guaranteed" rate or return.
Therefore, a new set of analytical tools and methodologies are required. It is
imperative that these tools address the multifaceted business needs necessary
for competitive-based decision making.

A. Faruqui et al. (eds.), Pricing in Competitive Electricity Markets

© Kluwer Academic Publishers 2000
198 Pricing In Competitive Electricity Markets

2. SPOT VERSUS FORWARD PRICES

Market clearing price forecasts for electricity are now essential for asset
valuation and retail offering purposes. Depending on the objective of the
study, price projections will be based on either spot or forward prices. Spot
electric prices are defined as the price of electricity at delivery. Spot price
projections allow for the forecasting of expected cash flows. While these
can be used for pro forma's and other financial statements, they are not
suited for the types of decisions required in an uncertain business
environment. This is because whether you are evaluating generation assets
or retail offerings, embedded options are inherent in the valuation
determination.
Forward price facilitates the proper pricing of embedded options. Option
valuation is based on, among other things, prices set today, price uncertainty
(volatility) and some time period. Forward prices are defined as the price of
electricity set today for future delivery. There are other advantages to
forward price projections. Forward prices are often observable.
Furthermore, as will be discussed later, some of the key drivers of electric
prices have forward-based markets. Finally, unlike spot prices, forward
prices have a known discount rate. The discount rate is a risk-free rate.

3. MARKETS

Currently, most of the electric trading hubs are very illiquid, or only liquid
a few months out in time. This will change with time. Standard 5(6) x 16 [5
weekday (plus Saturday) for sixteen hours a day] or 7 x 24 (seven days per
week, twenty-four hours per day) contracts will eventually trade with greater
liquidity. However, many embedded options will need to be evaluated on an
hourly basis. These hourly price profiles will change over time as supply
options change and as demand responses shift and new sources of usage
(e.g., electric vehicles) are brought to market. An example of the importance
of the price profile is demonstrated in Figure 1 (next page).
As Figure 1 shows, for the same annual (8760 hours) market price of
$28IMWh, the hourly price profile can be dramatically different. This will
result in vastly different valuations for certain assets like dispatchable
peaking resources or retail contracts with hourly volume usage. For
example, if one has a dispatchable generating resource with variable costs of
$301MWh the profit of dispatching the resource given these two different
price profiles is $42,505IMW and $ O.OO/MW. This example clearly shows
the necessity of capturing the hourly price behavior. To accurately estimate
the hourly price profile, structural modeling will be of great importance. It
Pricing in Competitive Electricity Markets 199

will allow for a mechanism to capture the hourly price profile, the
uncertainty around that hourly price profile, and various ways the expected
profile changes over the years as new technology and unit retirements take
place.

Annual Price Duration Curve

100

90

80

70

\
\
60
:;;
;t
::&
~ 50

------
~

---- ------
0-
0
::&
40

30

20

1
10

1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000 9,000

Tim. (Houra)

Figure 1. Price Duration Curves Example

4. STRUCTURAL MODELING

Structural modeling of the electric system is nothing new to the electric

industry. These models attempt to simulate a least-cost production solution
for a given hourly demand forecast and set of available generating resources.
The resource are modeled to capture their operating characteristics (heat rate
curves, maximum/minimum capacity levels, minimum up/down times,
forced and maintenance outage rates, fuel costs, start-up costs, variable
operations and maintenance costs, among other variables). The resulting
solution was used so that regulated electric utilities could take to the
regulating entity a solution that would justify the supply or demand-side
option(s) that they felt they needed to meet their obligation to serve their
captive retail customers in a reliable, prudent manner.
200 Pricing In Competitive Electricity Markets

With certain changes, these models, or models like them, will still be used
in the post-deregulation electric industry. While the previous solution was
least cost, the new paradigm requires a profit maximization solution. Also,
most of the current models represent demand without any price elasticity.
As the industry provides hourly price signals to its customers, the ability to
shift electric usage to avoid the higher price hours will need to be captured in
the models.
Another improvement will be shifting from a cost-based dispatch to a bid-
based dispatch. While the economists can debate whether or not the solution
will be driven to the same long-run market equilibrium price under either
dispatch methodology, in the short run, gaming theory and other pricing
dynamics will be required to be captured in these models.

5. MARKET CLEARING PRICE PROJECTION

METHODOLOGY

Most decision making by the electric utility industry in the past has been
to develop a cost to serve estimate, run a few sensitivities, and use those
results to justify to the regulating entity the need to put the resource
alternatives being proposed into rate base. A schematic of the process would
look like Figure 2.

Base Case
Integrated
and Market
Resource
Sensitivity Model ~

Plan
Development

Figure 2. Regulatory-Based Planning Process

The base case inputs came from various departments within the
organization. The load forecast group would have their models generate the
load forecast. The fuels group would generate the various fuel forecasts.
The energy production group would give generating unit operating
characteristics. Then, the corporate planning group would incorporate these
inputs, run the structural model, and provide the documentation in support of
the resource plan that was generated.
Having a base case market clearing price projection with various
sensitivity runs is not adequate to develop price volatility. Internally
Pricing in Competitive Electricity Markets 201

consistent scenario development will be needed. Thus, we need to modify

the previous schematic of the price process as in Figure 3.

Scenario InDut Market Scenario. Statistical

• Development Scenario Model Prices Analysis

i I
Figure 3. Market-Based Price Development Process

In order to integrate this price process, the first task will be to identify the
key drivers of the electric price. Some are listed in the table 1.

Table 1. Key Electric Price Drivers

Supply Drivers Demand Drivers

Generating Units (Existing, Additions and Load

Requirements)

Fuel Prices Price Elasticity

Hydro Conditions Weather

Unit Availability

Market Area

New Technology

If time permitted, uncertainty analysis on all possible variables could be

performed. However, given the time intensive nature of the market models,
identifying a few key drivers and concentrating on them will adequately
capture the major sources of price uncertainty. Depending on the region of
interest, different variables will be chosen as uncertain variables.
For instance, for the Western U.S. (WSCC - Western System
Coordinating Council), hydrological based generation is a key driver of
electricity and it can vary greatly in how much water will flow (primarily
depending on winter snowpack) and when it will flow (depending on
202 Pricing In Competitive Electricity Markets

spring/summer snowmelt). Table 2 shows just how much hydrological

based generation can vary by year.
In the Mid-Atlantic region, hydrological conditions may not play a key
role and thus can be excluded as an uncertainty variable. It would simply be
modeled as an expected value forecast across all scenarios.
Conversely, nuclear outage uncertainty may be a key electric price driver
for the Mid-Atlantic (MAAC) region but not so in the WSCC. Table 3
shows just how much nuclear generation can vary by year.

Table 2. WSCC Hydrological Generation by Year

Year Generation MWh

1994 155,262,006

1995 214,501,214

1996 239,573,494

1997 249,033,455

1998 217,630,893

Source: Resource Data International, Inc.

Table 3. MAAC Nuclear Generation by Year

Year Generation MWh

1994 88,042,735

1995 84,839,019

1996 81,398,260

1997 84,619,924

1998 97,127,345

Source: Resource Data International, Inc.

Pricing in Competitive Electricity Markets 203

In most areas, however, natural gas will be a key price driver. Unlike the
previously mentioned variables, natural gas has the advantage of being an
actively traded commodity in both the futures and over-the-counter (OTC)
markets. The New York Mercantile Exchange (NYMEX) trades gas at
Henry Hub for contracts on a monthly basis for 36 months. The OTe
markets trade natural gas at numerous trading hubs for terms as long as ten
years or more. A typical natural gas price curve for delivery at Henry Hub
might look like Figure 4.

Gas Forward Curve

2_60 , - - - - - - - - - - - - - - - - - - - - - - - - - - - - ,

2.50 f-----------------------------1

~,.O r--------~~-~------~"'------------l
------
::;;:230 f--------- -
~~-------I

'" 220 1_ _ _ _ _ _ _ _ _

2_10 f-----------------------------1

200 f----------------------------1

1.90 ',---~_-_--~--_--_--~--~--~--------I

1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

Figure 4. Natural Gas Forward Curve

For the price uncertainty around the curve, the options markets can be
used to calculate the implied volatility. The resulting natural gas price
uncertainty can them be displayed as in Figure 5 (next page).
Thus the key price drivers are represented as a distribution of possible
values. For variables like natural gas the futures and options markets will
provide a market based distribution of the possible values. For variables like
hydro, load growth and nuclear outages, historical analysis can be used to
arrive at a distribution of the values. The key is to identify the key price
drivers, obtain an estimation of the distribution that these values can
represent, and develop internally consistent scenarios that have an equal
204 Pricing In Competitive Electricity Markets

likelihood of occurring. Performing a structured Monte Carlo approach to

the scenario selection accomplishes this goal. This approach also allows for
a wide range of scenarios yet few enough to allow for the use of structural
modeling. Furthermore, internally consistent scenarios are critical for the
development of meaningful investment rules that can be derived from this
combination of possible scenarios.

Gas Uncertainty

1998 1999 2000 200t 2002 2oo) 2004 2005 2006 2007

Figure 5. Natural Gas Forward Curve with an Options Based Uncertainty Range

6. MARKET CLEARING PRICE PROJECTION

Using scenarios developed in a manner that anyone scenario is as likely

to occur and another, the expected market-clearing price is simply the
average of the scenario market clearing prices. An example of a market
price projection is given in the Figure 6 (next page).
The noticeable dip in prices occurring around 2003 for this region occurs
as a result of market entry that is projected to occur as market participants
realize that this region lends itself to a very volatile market. Because of that,
a higher capacity stock is justified. Monthly prices summarized as some of
the more popular contracts are in Figure 7 (next page).
Pricing in Competitive Electricity Markets 205

Market Clearing Price Forecast

~.oo ~--------------------------------------------------------------

3100 '------------------------------------------------------------,~-

i ~oo
f
D..
Q ~.oo ----~~------------~~~--------------------------------------
~
.;
i
()

2800

2700 ~--------------------------------------------------------------

26.00
2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020

Figure 6. WSCC Forward Curve Projection

Market Clearing Price Forecast

00.00 , - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

moo +---------------------~------------------------~._-----------

J 1\

I
00.00 t-----~------------~~----------t_i-----------+__r-----------

ro.oo t-----~~--------~~~~--------~~_f~------~~~---------

\ .....
10.00 t--------------------------------------------------------------

Feb-99 "ug-99 Mar-OO 001.(J() >.pr·01 May..Q2 Dec.<J2 Jun.()3

1- -7>24 --(6xI6) - - - . (ron 6X16) - - -Needle Peak (6 Hrs.) 1

Figure 7. WSCC Forward Curve Projections

206 Pricing In Competitive Electricity Markets

With the multiple scenario aspect of the simulation, the uncertainty

around the expected forecast can also be represented. Figure 8 (below) is a
graph with confidence intervals around the projected (6x16) market price
forecast.
Implementing an internally consistent, scenario-based market clearing
price projection approach to forecasting requires significant support and
effort. Many may find that it requires more effort than is needed and will
continue with a base case forecast approach with sensitivity analysis to
explain how prices will react with certain changes in the input variables.
They will do so at their own peril. A comparison of an expected value
market -clearing price forecast and a base case forecast is as in Figure 9
(below).
The short-term difference in the expected price and the base case price
forecasts is due to the asymmetric effect on prices of many of the key inputs.
For example, an equal MW increment/decrement in load will have a
different effect on prices. Because of the log normal distribution in capacity
prices, the incremental increase of a certain amount of load will produce a
higher proportional increase in prices (significantly so when capacity is
extremely tight) than will a similar decremental amount of load. Other key
inputs have a similar effect on prices. Natural gas prices have a floor in how
low the prices will fall (presumable not below zero). However, at least in
the short-term, there is no ceiling to which prices may increase (in the long-
term one would not expect prices to be significantly above the cost of new
production).
Furthermore, those that choose the base case route to market price
forecasting will have no basis for calculating true market volatility nor will
they have a basis for developing correlations between market clearing prices
for electricity and the key drivers of the electric price. The price uncertainty
that will be missing with the base case approach is illustrated with the Figure
10 (below).
An example of the importance of taking into account price volatility is
given with the following example. Let's say we have a dispatchable natural
gas fueled peaking unit with a 15,000 BtU/kWh heat rate. A base case
analysis will show the value of this resource to be $ 37/kW. However, the
uncertainty-based forecast will show the value of the resource to be $
53/kW. As this example shows, price uncertainly for peaking capacity
provides for significantly more upside than downside. This unit is
undervalued by 43 percent in the base case approach.
Pricing in Competitive Electricity Markets 207

Uncertainty in the Market Clearing Price (6x16) Forecast

12000

100,00

8000

J:
~
~

40.00

10% Probability -20% Probability -Expected Price -80% Probability ··90% Probability 1

Figure 8. WSCC Forward Curve Projection with Confidence Intervals

Alternative Market Clearing Price Forecasts

32.00

:2
~
:: ----------------------------------------~~~~~~----------------------
2900 ~
~
.g
11.
28,00

C
'"
:; 27.00
o
~ woo -------~~---------------------------------------------------------
:l!
25.00
,
24.00
\
\
,
'.'
.
23.00
2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020

1- -Expected Price - - ·Base Case --Lon;J-Tenn Eqlilibrium 1

Figure 9. Expected Versus Book Case WSCC Forward Curve Projection

208 Pricing In Competitive Electricity Markets

Price Distribution Curve

012 t----------:.F--kjf----~------------------_c

0.1

~ 0.08 +--------f----Hf-----\:----------------~
:c ~
<>
o
Ii. 0.06 t---------.f-----Hf------\.------------------------<

0.04 +-------f-----Hf-----"'----------------'

002 t------jr------Hf--------'~-------------<

oL---~----__--Ll----__----__--~~~~--~~~
o 10 20 30 40 50 60 70 80 90
Market Clearing Price (SlMWh)

I-Price Distribution -Expected Price -Base Case Price I

Figure 10. WSCC Forward Curve Distribution

7. CONCLUSION

The need to take into account price uncertainty in a market-based

economy is essential to good decision making. To be able to use price
uncertainty for asset valuation, forward prices must be used. This is because
options are evaluated using risk-free prices. Forward prices accomplish this
criterion since they are defined as the price of electricity set today for future
delivery. Thus, the forward price is certain, allowing a risk-free discount
rate.
Since forward prices are observable, and will increasingly trade on the
open market, some might suggest a lack of need to develop methodologies
for projecting market clearing prices. However, many generating assets and
retail offerings are not dispatched or used on a fixed basis. Generating assets
will dispatch differently depending on the market price in a given hour or set
of continuous hours. Retail consumers will use electricity at varying rates
throughout the day depend on many factors, only one of which is weather.
Thus, an hourly price projection is paramount to the valuation process.
Since markets are not envisioned to trade on an hourly basis other than a few
days ahead at most, market simulation using structural modeling approaches
Pricing in Competitive Electricity Markets 209

are an excellent way to tackle the hourly market clearing price projection
need.
Chapter 14

Energy Derivatives and Price Risk Management

James Read and Art Altman*

The Brattle Group and EPRI

Key words: Asset; Call; Correlation; Derivative; Forward; Fuel; Future; Generation;
Hedge; Market; Option; Put; Speculate; Value; Volatility.

Abstract: The business and market settings in which fuel management decisions are
made has changed radically in just a few short years. It used to be that most
electric power companies were vertically integrated public utilities. They held
exclusive franchises in their service areas in return for which they accepted the
obligation to serve and agreed to be subject to cost of service rate making.
Now, in contrast, wholesale power markets are competitive, and some states
have introduced competition at the retail level as well. New kinds of players
have emerged, including power marketers and independent power producers.

This chapter explores the implications of volatility and correlation for fuel
planning and asset management by introducing and then building on the
concept of derivative assets. It will show that all major electric power
resources-fuel supply contracts, power purchase agreements, and generating
units, for example--can be viewed as derivatives. It will also show that
volatility and correlation affect both the value and the risk of these resources.

In particular, volatility creates a premium on flexible resources-resources

that can be adapted as market conditions evolve. Flexibility in resources in
tum influences the kinds of instruments that can be used to manage risk.

* The authors are grateful to EPRI and Project Manager Jeremy Platt for permission to adapt
this material from their contribution to EPRI report number TR-l 11564. Jeremy Platt is the
Manager of EPRI's 1999 R&D Target #77, Fuel and Power Supply. Portions of this material
were originally developed under EPRI funding for Target #49, Power Markets and Risk
Management, with Art Altman as R&D Manager.
A. Faruqui et al. (eds.), Pricing in Competitive Electricity Markets
© Kluwer Academic Publishers 2000
212 Pricing In Competitive Electricity Markets

1. INTRODUCTION

The business and market settings in which fuel management decisions are
made has changed radically in just a few short years. It used to be that most
electric power companies were vertically integrated public utilities. They
held exclusive franchises in their service areas in return for which they
accepted the obligation to serve and agreed to be subject to cost of service
rate making. Power pools were common, of course, but they were
established chiefly to enhance reliability rather than to improve profitability.
Now, in contrast, wholesale power markets are competitive, and some states
have introduced competition at the retail level as well. New kinds of players
have emerged, including power marketers and independent power producers.
While rate regulation still plays a role in the electric power industry, the
advent of competition creates the potential for greater profitability. It also
creates greater risks.
In particular, the volatility of power prices exceeds that even of natural
gas, which was widely regarded as the most volatile commodity before
electricity trading took off. Regulation, in addition to controlling entry and
pricing, had an important role in risk management from the perspective of
shareholders, due to fuel adjustment clauses and periodic rate cases. In the
absence of regulation, power companies will have to devise new risk
management methods.
This section explores the implications of volatility and correlation for
fuel planning and asset management by introducing and then building on the
concept of derivative assets. It will show that all major electric power
resources-fuel supply contracts, power purchase agreements, and
generating units, for example--can be viewed as derivatives. It will also
show that volatility and correlation affect both the value and the risk of these
resources.
In particular, volatility creates a premium on flexible resources-
resources that can be adapted as market conditions evolve. Flexibility in
resources in tum influences the kinds of instruments that can be used to
manage risk.

2. A PORTFOLIO PERSPECTIVE

One of the themes of this section is that fuel planning and management
should be viewed from the perspective of the corporation as a whole. Fuel
management is not an end in itself but rather a means to an end.
Specifically, fuel resources are acquired and managed to facilitate the
production of electric power. Accordingly, the value and risk of fuel
Pricing in Competitive Electricity Markets 213

resources need to be judged in the larger context of a generation company

and perhaps in the context of an integrated electric power company.
An electric power company-or any other business for that matter--can
be viewed as a portfolio of instruments. (See Figure 1: "A Portfolio of
Instruments") An integrated power company typically holds generation
assets, fuel supply contracts, wholesale power contracts, and retail service
agreements, among other resources. Each of these can be a source or a sink
of value. For example, a generating unit has value in a competitive market
to the extent that it can produce power at a cost below the market price. That
in tum depends on the type of fuel it bums, the efficiency with which it
bums it, and other factors, such as start-up costs. Each instrument can entail
exposures to one or more sources of risk. A steam generating unit has
exposures to power and fuel markets, for example.

A Portfolio of Instruments

'" Remll ...1M

.J~D~ Y =~,...,....-"---,
V .~ I·~
c~. D ..,:-'t~ D ,;.
) 6~= Ma~rkets
Merchant Portfolio
p::r P::ase Ma>rkets
Contracts Contracts

Generation Portfolio

Figure 1. A Portfolio ofInstruments

As Figure 1 suggests, an integrated power company can be viewed as a

combination of a generation company (GenCo) and a retail company
(RetaiICo). The GenCo produces power and sells it to the RetailCo while
the RetailCo buys power from the GenCo and sells it to retail customers.
Many incumbents have recognized this and adopted functional unbundling in
response to competition-that is, they have divided their companies into
business units to isolate sources and sinks of value and thereby focus
management attention-though they remain integrated companies. Even in
214 Pricing In Competitive Electricity Markets

these cases it is important to evaluate risk from the perspective of the overall
portfolio. The reason is two-fold. First, some risk exposures "cancel out"
when business units are combined. Second, some of the underlying risks
may be correlated, in which case there may be "natural hedges" in a
portfolio that are overlooked when the components are evaluated as separate
business units.

3. MEASURING PORTFOLIO VALUE & RISK

In principal it is straightforward to calculate the value of a portfolio:

calculate the value of each asset in the portfolio and then add the values up.
Calculating asset values can be difficult in practice, however. Later in this
section we describe principles for calculating the value of fuel and power
resources based on the fact that they are derivative assets.
The overall risk of a portfolio can be described in a number of ways. The
most popular portfolio risk metric is called "value at risk" or "VaR".
Another portfolio risk metric is called "earnings at risk" or "cash flow at
risk" ("CFaR"). We will describe both in tum.

3.1 Value at Risk

Value at risk is a measure of the potential for loss on a portfolio. It can

be defined as the maximum loss that will be realized over a specified holding
period (e.g., five days) with a specified level of confidence (e.g., 99 percent).
Thus a value at risk metric is specified by two parameters: the portfolio
holding period and the confidence level. A statement that the value at risk
for a given portfolio is $10 million based on a five-day holding period and a
99 percent confidence level could be translated as follows: "The probability
that the value of the portfolio will decline by more than $10 million over the
next week is less than one percent." (See Figure 2).
VaR evolved in the financial industry. It is widely used by commercial
banks and other financial companies to measure the risk of their trading
portfolios. The motivation for using a metric like VaR is that these
companies are active in many markets and thus have exposures to a large
number of risks. Simply measuring and reporting exposures to a large
number of risks does not provide a satisfactory summary of the risk that the
owners of a corporation are bearing. VaR evolved as a response to these
concerns. Since it is a portfolio risk metric is nets out any offsetting
positions (i.e., longs versus shorts) and takes account of correlations amongst
underlying exposures.
Pricing in Competitive Electricity Markets 215

VALUE AT RISK (VaR)

Value at Risk
Probability

, 1
Frequency I,.-'--f--I~,

m,\ :
Value

Figure 2. Value At Risk (VaR)

3.2 Cash Flow at Risk

Many companies acquire assets with a view to holding them indefinitely.

In contrast to financial companies, they are not engaged in trading
businesses. It is natural in these settings for managers to think in terms of
the risk associated with earnings or cash flow as distinct from the market
value of their portfolio. For example, they may be concerned with the risk
of failing to meet payroll or debt service. Earnings and cash flow at risk are
portfolio risk metrics that focus on financial flows rather than stocks.
Cash flow at risk can be defined as the probability that the cash flow
generated by a portfolio during a specified planning period (e.g., a month, a
quarter, or a year) will be less than a specified target level (e.g., $2 million).
A cash flow at risk metric has two parameters: the planning period (e.g.,
January 1999) and the target cash flow level (e.g., $2 million). (See Figure
3.) It is intended to measure the risk of a portfolio of relatively illiquid
assets and liabilities-assets and liabilities that are difficult to trade in a
short time period (e.g., one week).
216 Pricing In Competitive Electricity Markets

CASH FLOW AT RISK (CFaR)

Currulative
Probability

Threshold Cash
Cash Row Row

Figure 3. Cash Flow At Risk (CFaR)

3.3 Calculating Portfolio Risk Metrics

Like any other random variable, the uncertainty associated with portfolio
values or cash flows can be described in the form of a probability
distribution. There are several methods for estimating the probability
distribution of portfolio value (as the basis for VaR) and cash flow (the basis
for CFaR). These include relatively simple methods, such as constructing
frequency distributions based on historical observations of portfolio results,
to more sophisticated but computationally demanding methods, such as
modeling the underlying sources of risk explicitly, using a computer to
simulate a large number of joint outcomes, and computing the portfolio cash
flows. Calculating the portfolio payoffs requires calculating the payoffs to
each instrument in the portfolio and then adding them up.

4. BASIC CONCEPTS REVISITED: SPOT,

FORWARD, AND OPTION MARKETS

From a financial perspective, fuel and power supply contracts are energy
derivatives. A derivative is an asset the payoffs to which are completely
Pricing in Competitive Electricity Markets 217

determined by the price of some underlying asset. A coal supply contract,

for example, entails the exchange of coal for cash, so its payoffs depend on
the relationship between the contract prices and the market prices of coal
over the life of the contract. Thus, if one knew with certainty what the
market prices of coal would be over the life of the contract, one could
calculate the contract payoffs with certainty. In this case coal is the
underlying asset. To take another example, a power purchase agreement
entails an exchange of electric power for cash. In this case power is the
underlying asset. If the contract prices are indexed to the price of natural gas
rather than fixed, then both power and gas are underlying assets.
Recognizing that fuel and power supply contracts are derivative assets is
crucial because there are well-established methods for calculating the value
and risk of many derivatives.
From a financial perspective, all derivative assets are equivalent to some
combination of elementary derivatives-forward contracts and options.
Therefore, if you know how to price and hedge elementary forwards and
options, you can price and hedge more complex derivatives, too.
Elementary derivatives are sometimes called "financial building blocks" to
emphasize this idea.

4.1 Elementary Forward Contracts

An elementary forward contract is an agreement to exchange a fixed

quantity of a commodity for a predetermined price on a specified future date.
The fixed quantity is called the "contract quantity", the predetermined price
is the "forward contract price", and the specified date is the "delivery date".
No money changes hands until delivery.
Consider an example: In January, Party A agrees to buy from Party B
10,000 MMBtu of natural gas in July at Long Beach, California for a price
$3.00IMMBtu. The underlying asset in this case is natural gas, the contract
quantity is 10,000 MMBtu, the delivery date is July, the delivery place is
Long Beach, and the contract price is $3.00IMMBtu. Party A will take
delivery of 10,000 MMBtu of gas at Long Beach in July and pay $30,000
whereas Party B will deliver 10,000 MMBtu of gas and accept $30,000 in
payment.
There are two sides to every contract. The buyer is said to have a "long"
position in a forward contract whereas the seller has a "short" position. The
long position is better off when the price of the commodity goes up (since
she takes delivery of the commodity) whereas the short position is worse off.
In the preceding example, Party A has the long position and Party B has the
short position.
218 Pricing In Competitive Electricity Markets

The forward contract in the preceding example entails physical

settlement. That is, the short position will make delivery of natural gas and
the long position will take delivery. Some commodity contracts specify
financial settlement, in which case the long and short positions exchange a
cash payment equal to the difference between the cash value of the
underlying on the delivery date and the contract price. If the cash value of
the commodity were $2.90/MMBtu on the delivery date, for example, the
long position would collect $1,000 from the short position. If the cash value
of the commodity were $3.1O/MMBtu on the delivery date, the long position
would pay $1,000 to the short position.
A forward contract is an obligation, so the buyer and seller (long and
short positions) will exchange the commodity regardless of its price on the
delivery date. The long position takes delivery of the commodity and pays
cash. The short position makes delivery of the commodity and receives
cash. If we use the symbol ST to denote the spot price of the commodity on
the delivery date and X to denote the contract price, then the payoffs on the
delivery date (C T) to the long position in a forward contract can be written as
follows:

Forward Contract (long):

The gains on the long position in a contract are the losses on the short
position, and vice versa. Therefore, the payoffs to the short position can be
written as follows:

Forward Contract (short):

The payoff profiles of long and short positions in a forward contract are
depicted in Figures 4 and 5.
Viewed in isolation a forward contract is like a bet. If the spot price of
the underlying is greater than the contract price on the delivery date, then the
long (short) position "wins" ("loses") in the sense that she receives (delivers)
a commodity with cash value greater than its price. If the spot price is less
than the contract price, on the other hand, then the long (short) position loses
(wins). Thus the risk profile of a forward contract is symmetric or "two-
sided."
Pricing in Competitive Electricity Markets 219

TERMINAL PAYOFFS: FORWARD CONTRACT

LONG POSITION
Payoff ($)

30

20

10

0 Spot
Price ($)
-10

-20

Figure 4. Terminal Payoffs: Forward Contract Long Position

TERMNAL PAYOFFS: FORWARD CONTRACT

SHORT POSITION
Payoff ($)

20

10

0 Spot
Price ($)
-10

-20

-30

Figure 5. Terminal Payoffs: Forward Contract Short Position

220 Pricing In Competitive Electricity Markets

4.2 Forward Prices

The cash or spot price of a commodity is the price for immediate

delivery. The forward price of a commodity is the price for delivery of the
commodity on a specified future date. Forward prices are established in
forward markets, that is, through buying and selling for future delivery. In
fact, the forward price on a given date is the contract price set in forward
contracts struck on that date.
Whereas there is only one spot price for a commodity at any time, there
is a schedule of forward prices, one for each possible delivery date. The
schedule of forward prices for a commodity is often depicted by a graph with
delivery dates along the horizontal axis and prices along the vertical axis.
The delivery date at the origin is the current date and the corresponding price
is the spot price. Dates to the right on the horizontal axis are future dates
and the corresponding prices are therefore forward prices. This graph is
called the "forward price curve" or "forward curve" for short.
It is costly to move commodities from one place to another. As a
consequence, there can be substantial differences in the prices for delivery of
commodities to different locations. These price differentials may be
temporary or persistent. Power and natural gas are good examples.
Therefore, it is essential that the delivery point to which a forward price
curve pertains be identified.

4.3 Elementary Option Contracts

An elementary option contract confers the right to buy (if it is a "call"

option) or sell (if it is a "put") a commodity for a predetermined price on (or
before) a specified future date. i The predetermined price is called the
"strike" or "exercise" price. The specified date is the "expiration" date. The
holder of an option contract-the party that decides whether or not to
exercise-is said to have a long position and the writer of the option is said
to have a short position.
Here is an example of a call option. In January, Party A pays Party B
$0.30IMMBtu to obtain an option to buy 10,000 MMBtu of natural gas in
July at Long Beach for a price of $3.00IMMBtu. The contract quantity in
this case is 10,000 MMBtu, the expiration date is July, and the strike price is
$3.00IMMBtu. Party A has the long position and Party B has the short
position. The option premium is $O.30IMMBtu. If the spot price of gas is
greater than $3.00IMMBtu on the expiration date, then Party A will exercise
the option. If the spot price is less than $3.00IMMBtu, Party A will let the
. . ii
optIon expIre.
Pricing in Competitive Electricity Markets 221

The salient feature of an option is that it is a right rather than an

obligation. Accordingly, the holder will exercise an option only if it is
profitable to do so. If the spot price of the underlying at expiration is greater
than the strike price, the holder of a call will choose to exercise whereas the
holder of a put will allow it to expire. If the spot price is less than the strike
price at expiration, on the other hand, the holder of a call will allow the
option to expire whereas the holder of a put will choose to exercise. Payoffs
to long positions in elementary calls and puts, respectively, can be written as
follows:

Call Option: CT = max{O,(ST - X)}

Put Option:

Whereas forward contracts have two-sided risk profiles option contracts

have one-sided risk profiles. These payoff profiles are depicted in Figures 6
and 7.

4.4 Option Prices

The payoffs to the holder of an elementary option contract will positive

or zero (since the holder gets to decide whether or not to strike), so the
payoffs to the writer will negative or zero. Therefore, the holder will have to
pay the writer a price to induce her to enter into a contract. By convention
the price of an option contract is called the "option premium."
In a well-functioning commodity market there is only one forward price
for delivery of the commodity to a specified place on a specified date. In
contrast, there may be an array of option premiums because there may be a
number of different option contracts. To begin with, both call options and
put options may be traded. In addition, there may be calls and puts with
different strike prices. Thus option markets may offer a much greater variety
of tools for managing risk and betting on prices than do forward contracts.
222 Pricing In Competitive Electricity Markets

TERMINAL PAYOFFS: CALL OPTION

Payoff ($)

Figure 6. Terminal Payoffs: Call Option

TERMINAL PAYOFFS: PUT OPTION

Payoff ($)

:~
o +---+----"'t----+---+----l~ Spot
Price ($)
10 20 30 40
-10

-20

Figure 7. Terminal Payoffs: Put Option

Pricing in Competitive Electricity Markets 223

5. OTHER CONTRACTS & ASSETS CAN BE

VIEWED AS DERIVATIVES

To illustrate the power of the financial building block framework, let's

look at how some standard derivative products can be replicated by
constructing portfolios of forwards and options.
Consider first a swap contract. A swap is an agreement to exchange cash
flows according to a specified formula. In fact, there are several different
kinds of swaps, but the most common is a fixed-for-floating-price swap, in
which one party receives a floating (spot or index) commodity price each
month over a specified period and pays a fixed price while the counterparty
receives the fixed price and pays the floating price. Actually, a swap of this
sort is equivalent to a portfolio of serial one-month forward contracts, each
of which has a different delivery date. For example, a one-year swap
agreement struck at the end of 1998 is equivalent to twelve forward
contracts: one contract for January 1999 delivery, a second contract of
February 1999 delivery, and so on, up to December 1999. Therefore, if we
can observe forward prices for delivery over the next twelve months, then
we can figure out what this swap agreement is worth.
Caps and floors are two other examples. A cap periodically pays an
amount equal to the difference between the floating (spot or index) price of a
commodity and a fixed "cap" price, if the difference is positive. A floor, on
the other hand, pays an amount equal to the difference between a fixed
"floor" price and the floating price, if the difference is positive. Upon
reflection it should be clear that these agreements are analogous to the fixed-
for-floating-price swap we just described, except that they are equivalent to
bundles of options rather than forward contracts. Specifically, a cap is
equivalent to a portfolio of serial call options whereas a floor is equivalent to
a bundle of serial puts.
Since these contracts are equivalent to combinations of forward or option
contracts, we can figure out what they are worth by finding the relevant
forward prices and options premiums and then adding up the values of the
parts. In fact, this basic approach can be used to evaluate a wide variety of
assets, as we will now see.

6. VALUING FUEL & POWER RESOURCES

Many fuel and power contracts are equivalent to combinations of

elementary forward contracts. For example, consider a contract struck in
December 1998 to supply 10,000 MMBtu of natural gas a month for the
224 Pricing In Competitive Electricity Markets

following year at a price of $1.80IMMBtu. In each of the next twelve

months the supplier will deliver 10,000 MMBtu of gas and receive payment
of $1.80IMMBtu. This contract is equivalent to a series of twelve one-
month forward contracts for natural gas: one contract for January 1999
delivery, another contract for February 1999 delivery, and so on. Therefore,
we can calculate the current market value of this contract by calculating the
market value of the twelve component forward contracts and then adding
them up.

6.1 Fixed-Price Fixed-Quantity Contracts

If there is an active forward market for a commodity, it is straightforward

to calculate the market value of a contract that specifies delivery of fixed
quantities. Since the parties enter into forward contracts willingly, the net
market value of a forward contract is zero at its inception. Thus the current
forward price (FO,T) for delivery of a commodity on a specified date (T) is the
contract price on newly written forward contracts:

PV{ST - FO,T} =0
The forward price is fixed in advance, so the present value of the forward
price can be computed by discounting at the rate of return prevailing on
bonds (r) over the same time horizon. This implies that the present value of
the commodity on the delivery date is equal to the current forward price
discounted to the present at the "risk-free" rate:

Since fixed-price fixed-quantity contracts entail a series of payoffs of the

form ST-X, where ST is the cash value of the commodity and X is a contract
price, the current market value of such a contract is equal to sum of a series
of terms like the following:

PV{ST - X} = (Fo,T - X)/(l + r)T

Table 1 shows how this valuation rule can be applied to calculate the
current market value of a hypothetical supply contract that delivers 10,000
MMBtu of gas a month for a period of one year at a contract price of
$1.80IMMbtu. In this example, the forward curve for gas is "flat" with
forward prices for delivery in each month over the next year equal to
Pricing in Competitive Electricity Markets 225

$2.00IMMBtu. Thus, the difference between the current forward price and
the contract price each month is $0.20IMMBtu, which is $2,000 a month for
the contract, before discounting. When the payoffs are discounted at a rate
of 5 percent a year and then added up, the total is $23,376. Notice that all
we needed to know to calculate the value of the contract was (1) the contract
prices, (2) the current forward prices for delivery of the underlying
commodity over the time horizon of the contract, and (3) the current interest
rates in the bond markets.

Table 1: Valuation of Firm Fuel Contract

Forward Supply
TiIre Discount Forward Contract Contract Contract Contract
Delivery Remaining Factor Pree Pree Value Quantity Value
Period (years) ($IMMBtu) ($IMMBtu) ($/MMBtu) (gallons) ($)

January 0.083 0.996 2.00 1.80 0.199 10,000 1,992

February 0.167 0.992 2.00 1.80 0.198 10,000 1,984
March 0.250 0.988 2.00 1.80 0.198 10,000 1,976
April 0.333 0.984 2.00 1.80 0.197 10,000 1,968
May 0.417 0.980 2.00 1.80 0.196 10,000 1,960
June 0.500 0.976 2.00 1.80 0.195 10,000 1,952
July 0.583 0.972 2.00 1.80 0.194 10,000 1,944
August 0.667 0.968 2.00 1.80 0.194 10,000 1,936
September 0.750 0.964 2.00 1.80 0.193 10,000 1,928
October 0.833 0.960 2.00 1.80 0.192 10,000 1,920
November 0.917 0.956 2.00 1.80 0.191 10,000 1,913
December 1.000 0.952 2.00 1.80 0.190 10,000 1,905

Total 23,376

The value computed in Table 1 is an estimate of what the contract is

worth today. It indicates, for example, that if the supplier wished to get out
of the contract, she would have to pay $23,376 to do so. The process of
periodically re-evaluating contracts based on current market conditions is
referred to as "marking to market."

6.2 Flexible Contracts

Many contracts have embedded options. That is, they are equivalent to
combinations of forward contracts and option contracts. For example, a
contract to supply natural gas that requires the customer to take a minimum
226 Pricing In Competitive Electricity Markets

of 5,000 MMBtulmonth at $1.80/MMBtu but allows the customer to take up

to 10,000 MMBtulmonth at $1.801MMBtu is equivalent to a combination of
(1) twelve one-month forward contracts for 5,000 MMBtu of gas at a
contract price of $1.801MMBtu and (2) twelve one-month call options on
5,000 MMBtu of gas with strike prices of $1.80/MMBtu.

Table 2: Valuation of Flexible Fuel Contract

Forward Minimum Can Optional Supply
Time Forward Contract Contract Contract Option Contract Contract
Delivery Remaining DiscOlUlt Pri::e Pri:e Value Quantity Vahle Quantity Value
Pernd (years) Factor (SIMMBtu) ($IMMBtu) (SIMMBtu) (gallons) ($) (gallons) ,$1

January 0.083 0.996 2.00 1.80 0.199 5,000 0.208 5,000 2,036
February 0.167 0.992 2.00 1.80 0.198 5,000 0.200 5,000 1,992
March 0.250 0.988 2.00 1.80 0.198 5,000 0.232 5,000 2,148
April 0.333 0.984 2.00 1.80 0.197 5,000 0.236 5,000 2,164
May 0.417 0.980 2.00 1.80 0.196 5,000 0.252 5,000 2,240
June 0.500 0.976 2.00 1.80 0.195 5.000 0.252 5,000 2.236
July 0.583 0.972 2.00 1.80 0.194 5,000 0.268 5,000 2,312
August 0.667 0.968 2.00 1.80 0.194 5,000 0.272 5,000 2.328
September 0.750 0.964 2.00 1.80 0.193 5,000 0.268 5,000 2,304
October 0.833 0.960 2.00 1.80 0.192 5,000 0.280 5,000 2,360
November 0.917 0.956 2.00 1.80 0.191 5,000 0.276 5,000 2,336
December 1.000 0.952 2.00 1.80 0.190 5,000 0.288 5,000 2,392

Total 26,848

We have already seen how to calculate the value of the forward

components of the contract. If there is an active options market for gas-in
particular, if there are traded call options with $1.80 strike prices and
expiration dates in each of the next twelve months-then we can calculate
the value of the option components of the contract too. This is illustrated in
Table 2 (above), in which the premiums for call options with $1.80 strike
prices range from about $0.20 to about $0.30IMMBtu, depending on the
expiration date. The analysis indicates that the contract has a current market
value of $26,848, substantially more than the case in which the customer
was required to take all 10,000 MMBtu. This example illustrates an
important point: a contract with flexibility (i.e., embedded options) is always
worth at least as much as an otherwise identical contract without flexibility,
and in general it will be worth more.
Pricing in Competitive Electricity Markets 227

6.3 Other Fuel & Power Resources

Many other resources owned and managed by electric power companies

can be viewed as energy derivatives. For example:

• Generating units bum fuel in order to produce electric power, so they

are derivatives for which fuel and power are the underlying assets.
Since the manager will dispatch a generating unit if the price of power
is greater than the marginal cost of generation, a dispatchable unit is
equivalent to a portfolio of call options on energy, where the "strike
price" is equal to the generation cost. If a unit is not dispatchable,
then it is equivalent to a portfolio of forward contracts for power and
fuel.
• Storage facilities allow fuel to be carried from the present into the
future, so they are derivatives for which fuel is the underlying asset.
(Forward contracts are sometimes referred to as "synthetic storage"
for this reason.)
• Tolling agreements entail the exchange of fuel for power at a
contractually specified heat rate. Like power plants, they are
derivatives for which the underlying assets are power and fuel.

The recognition that these resources are derivative assets is crucial

because it implies that (1) the value of these resources depends on the
current forward prices of the underlying assets and (2) the risk of these
resources depends on the volatility and correlation of the underlying assets.
Some of the most important resources (e.g., generating units, tolling
agreements) are driven by the spread in power and fuel prices, hence the
emphasis in this report on managing price spreads to achieve profitability
and manage risk.
In the last example we were able to calculate the value of the contract
because the prices of the embedded options were directly observable. Often
we will not be so lucky, and it will be necessary to estimate the values ofthe
options.

6.4 A Formula for Estimating Option Values

There are formulas for calculating the theoretical value of a European

option when the underlying asset price can be described as a log-normal
random variable. iii To start with the simplest case, the value of a European
call option with a fixed strike price can be computed using a variant of the
well-known Black-Scholes option pricing model. Like a forward contract,
the value of a call depends on the current forward price for delivery of the
228 Pricing In Competitive Electricity Markets

underlying asset on the option expiration date (Fo.T) , the contract (strike)
price of the option (X), the expiration date of the contract (7), and the risk-
free rate of return (r). Unlike a forward contract, the value of a call also
depends on the volatility of the underlying commodity price (a):

CT = max{O,(ST - X)}
PV {C T} =(Fo'T N{a}- X N{b }YO + r)T
a ~ I{ F;T }"rr + ~"rr
b=a -aJf

The symbol N {'} denotes the cumulative probability for a standard normal
variable-a normal random variable with a mean of zero and a variance of
one. It lies between zero and one. This formula could be used, for example,
to value options to buy power or fuel at fixed prices.
Here is an example. What is the current value of a call option on natural
gas with a strike price (K) of $2.00IMMBtu when there are three months
remaining to contract expiration (T = 3/12 = 0.25 years)? Suppose the
forward price for delivery of gas (FO,T) three months from now is
$2. 151MMBtu, the interest rate is 5 percent/year (r = 0.05), and the volatility
of gas prices is 40 percent/year (<1> = 0.40). Using the formula above, we
calculate that parameter a is 0.4616 and the parameter b is 0.2616. We can
find a table of cumulative probabilities for the standard normal distribution
in the back of any statistics book, which we can use to determine that N(a) is
approximately equal to 0.6778 and N(b) is equal to 0.6032. When we put it
all together we calculate that the option is worth $0.25/MMBtu.
With minor modifications, essentially the same formula can be used to
value European options for which the strike price is risky rather than fixed.
The modifications required to value these exchange options are (1) the
current forward price of the second underlying asset (F2,o.T), is substituted for
the fixed strike price and (2) the volatility parameter depends on the
volatilities of both of the underlying assets (<1>1 and <1>2) and on their
correlation (Pu):
Pricing in Competitive Electricity Markets 229

CT = max{O,(SI,T-SZ,T )}
PV{ CT } = (F;,O,T N{a} - F2,O,T N{b })/(l + rl

a =In( F;,O,T }aJf + !aJf

F2 ,o,T 2
b=a-aJf
a = ~a; + a~ - 2 a /a2 P/,2
This formula might be used, for example, to approximate the value of the
cash flows to a dispatchable generating unit In that case the first underlying
asset is the price of electric power and the second underlying asset is the
price of fuel (e,g., natural gas) multiplied by the relevant heat rate (e.g.,
10,000 Btu/kWh). It conveys the critical insight that the value of a
generating unit depends not only on the volatility of the power market but
also on the volatility of the underlying fuel market and the correlation of the
power and fuel markets. This same observation applies to many other
resources, including tolling agreements with embedded option terms.

7. HEDGING VERSUS SPECULATION

Derivatives can be used to take exposure to risk or reduce exposure to

risk. They can, in other words, be "risky" or "safe", depending on how they
are used. This observation raises the important distinction between hedging
and speculative motives for participating in commodity markets.
A hedger is a market participant who has an exposure to risk and wishes
to reduce that exposure. The owner of a coal-fired generating unit is a
possible example. If the coal unit is the only asset in the owner's portfolio,
then the portfolio is "long" electricity and "short" coal. This means that,
other things being equal, the value of the portfolio will increase if electricity
prices increase and the value of the portfolio will decrease if coal prices
increase. The owner is a natural "short hedger" in electricity and a natural
"long hedger" in coal, since she can reduce her portfolio risk exposures by
selling electricity forward and buying coal forward. In fact, by taking "equal
but opposite" positions, the owner can eliminate ("neutralize") the
exposures. The idea is to combine two (or more) positions that have
opposite exposures so that the net portfolio exposure is reduced. This is
depicted graphically in Figures 8 and 9.
230 Pricing In Competitive Electricity Markets

SHORT HEDGERS
Payoff ($)
Production (S)
,,
30

Hedge (X)
20 ,
,
,, '
, ''
10 ,>~
,,
,, ,
o -l'---+-----'-t-,--t----I--..... Spot
10 20 " " 30 40 Price ($)
,,
-10 ,,
,,
,,
-20 Forward
Position (X-S)

Figure 8. Short Hedgers

LONG HEDGERS

Payoff ($)

Forward
20 ,, Position (S-X)
,
, ,,
10 ,,
,,
0 ,, , Spot
,, 10 /,' 20 30 40 Price ($)
, ,
"
,," ,,
-10
, ,,
,,
-20 Hedge (-X)
,,
,,
,,
-30 ,,
,,
,, Consumption (-S)

Figure 9. Long Hedgers

Pricing in Competitive Electricity Markets 231

A speculator is a market participant who has a view on the market and

therefore wishes to increase market exposure. The owner of a coal-fired
generating unit is again a possible example. If the owner has no view on the
electricity market, for example, then she might choose to sell all of the
production of the coal unit forward. On the other hand, she might believe
that current forward prices for electricity are "low" and hence wish to retain
some exposure to the electricity market. Alternatively, she might believe
that electricity prices are "high" and therefore wish to increase the exposure
of her portfolio by selling forward more than the generation capability of the
coal unit. Thus the terms "hedging" and "speculative" describe motives for
participating in a market. An individual player can be influenced
simultaneously be both hedging and speculative motives.

8. MANAGING PORTFOLIO RISK

At the highest level of generality, we can identify four basic steps

involved in measuring the risk of a portfolio of assets:

1. Identify the underlying sources of risk

2. Describe the possible outcomes for each source of risk
3. Calculate portfolio exposures to the sources of risk
4. Calculate overall portfolio risk

There are many possible sources of risk. Broad categories include event
risk, operational risk, credit risk, and market risk, to name a few. Event risk
is risk associated with specific types of events, such as fire, flood, and theft.
The traditional approach to managing event risk is to purchase insurance
policies from an insurance company. Operational risk is risk associated with
the failure of physical or managerial systems. The traditional approach to
managing operational risk in the electric power industry included
maintaining reserves of generation capacity and holding substantial
inventories of fuel, for example. Credit risk is risk associated with the
ability of customers to pay and vendors to deliver. Market risk is risk
associated with uncertainty about future commodity prices.
Measuring underlying sources of risk boils down to describing them in
statistical terms. For example, event risks can be measured in terms of the
possible amounts of loss and the associated probabilities. Market risks are
usually measured by the standard deviation of percentage price changes, a
topic addressed elsewhere in this report under the rubric of "volatility."
Measuring exposures to risk requires an understanding of how those risks
affect the value of the assets that make up the portfolio. In the event of a
232 Pricing In Competitive Electricity Markets

forced outage, for example, how much energy would be lost and what would
be the value of the lost energy? With respect to market risks, one might ask
how much the value of an asset will change if the price of an underlying
asset changes by one dollar. Measures of risk exposure can be derived from
a comprehensive asset valuation model.
Measuring portfolio risk (e.g., VaR) involves combining measures of risk
exposure with measures of the underlying risks. For all but the simplest
portfolios, this task requires a portfolio evaluation tool that can integrate
information about the underlying risk factors (e.g., market prices, forced
outages), evaluate the assets in the portfolio, calculate the risk exposures of
the assets, and calculate portfolio risk recognizing the correlations as well as
own risks of the underlying factors.

8.1 Price Risk Exposures

There are several standard measures of price risk exposure. The most
frequently used of these measures-which are referred to collectively as the
"Greeks"-are the delta and gamma parameters. We will discuss both in
tum.
Delta (~) is defined as the rate of change in the value of a derivative
asset (e.g., a generating unit) with respect to the price of an underlying asset
(e.g., power or fuel):

It is a measure of the slope of the function that relates the value of a

derivative asset to the price of an underlying asset. A delta equal to one, for
example, indicates that the value of the derivative will increase by one dollar
when the underlying price increases by one dollar. A delta equal to minus
one indicates that the value of the derivative will decrease by one dollar
when the underlying price increases by one dollar. A delta equal to zero
indicates that the value of the derivative will be unaffected by changes in the
underlying price.
The delta of a derivative need not lie between plus and minus one. To
see this, consider the delta of a forward contract for delivery of Q units of a
commodity at some time t in the future. Since the value of a contract to
deliver Q units of the commodity is equal to Q times the value of a contract
Pricing in Competitive Electricity Markets 233

to deliver one unit, the delta of a contract to deliver Q units is equal to Q

times the delta of a contract to deliver one unit.
One can compute as many deltas for a derivative asset as there are
underlying assets. For example, a forward agreement to exchange one
commodity for another commodity will have two deltas, one with respect to
each of the underlying commodities. A commodity contract that calls for
delivery each month over a one-year time horizon will have a delta with
respect to the forward price for delivery in each month.

~ = JV
I JFI
~ = Jv
2 JF2

Delta indicates how large an exposure a derivative or a portfolio of

derivatives has to a particular market price. Suppose, for example, that a
generating unit has a delta equal to - 100,000 with respect to the price of
natural gas in July. This indicates that the value of the unit will go down by
$100,000 when the price of July gas goes up by $1. The gas price exposure
could be "delta hedged" by buying or selling an appropriate quantity of
another gas derivative. An obvious example of another gas derivative is a
forward contract for gas. Buying 100,000 units of gas forward would offset
the delta exposure of the generating unit.
Gamma ( r ) is defined as the rate of change in delta with respect to a one
dollar change in the price of an underlying asset:

It is a measure of the rate of change in the delta of a derivative with

respect to the price of an underlying asset. The gamma of a forward contract
is zero whereas the gamma of a call or put option that is at or near the money
is greater than zero. The gamma of an option that is deep in or out of the
money is close to zero. iv Gamma is an indicator of stability. It measures the
extent to which the delta of a derivative asset changes when the price of an
underlying asset changes.
234 Pricing In Competitive Electricity Markets

A derivative asset has a gamma with respect to each underlying price and
with respect to each pair of underlying prices. For example, a derivative
with two underlying prices will have two "diagonal" gammas-one with
respect to each underlying-and one "cross" gamma with respect to the pair
of underlying prices:

a~J
111
aFI
a~2
122
aF2
a~J a~2
112 121
aF2 aFI
Gamma has important implications for managing a portfolio that includes
options or option-like instruments. A delta-hedged portfolio with substantial
gamma exposure will need to be rebalanced frequently to maintain a hedged
position. To reduce the frequency with which rebalancing is required, a
portfolio can be "gamma hedged" by taking offsetting positions in
instruments that have gamma exposure as well as delta exposure.

8.2 Portfolio Risk Exposures

Ultimately the risk that matters to the owners of a business is the risk of
the portfolio of assets and liabilities that they hold, not the risks of the
individual assets and liabilities themselves. The risk exposures of the assets
and liabilities in a portfolio are an important determinant of portfolio risk but
they are not the whole story. For one thing, portfolio risk is a function of net
risk exposures, which can be found by adding up the exposures of the
components of the portfolio:

Since portfolios may contain both "long" and "short" pOSltIons with
respect to a source of risk, some of the exposures "cancel out". For another
thing, the underlying sources of risk may be correlated. For example,
portfolios of generation assets have long exposures to power and short
Pricing in Competitive Electricity Markets 235

exposures to fuel. Power and fuel prices are positively correlated-when

fuel prices go up or down, power prices tend to go up or down too--so fuel
price risk partly offsets power price risk in such a portfolio. In order to
calculate the net risk exposures of a portfolio, it is necessary to assess the
exposures of each of the components of the portfolio. To calculate the
overall risk of a portfolio, it is necessary to assess the volatilities and the
correlations of the underlying risks. That assessment, in turn, will be
facilitated by an understanding of those markets and their relationships to
other energy markets.

8.3 Portfolio Risk: A Recap

To see how it all fits together let's consider a common method of

computing value at risk called the "delta-normal" approximation. This
method exploits a well-known result from portfolio theory that if the returns
to the securities in a portfolio are normally distributed, then the portfolio
returns are also normally distributed with a variance equal to a weighted
average of the covariances of the security returns. If portfolio returns are
normally distributed, it is straightforward to construct a confidence interval
for the value of the portfolio:

The symbol (Jp is the standard deviation of the portfolio, ~t is the length
of the holding period, and k is a constant that corresponds to the specified
one-tailed confidence interval for the normal distribution. In the delta-
normal methodology, the portfolio standard deviation is computed using the
volatilities and correlations of the underlying risk factors in conjunction with
the deltas of the portfolio with respect to those factors. For example, if there
are two factors underlying a portfolio--the prices of two commodities, like
gas and electricity, for example-then the portfolio standard deviation can
be computed as follows:

The deltas (~) in this expression are the rates of change in the value of the
portfolio with respect to the underlying commodity prices. The symbols FI
236 Pricing In Competitive Electricity Markets

and F2 denote the forward prices of two commodities. Again, notice that the
portfolio risk depends on (1) the net risk exposures, (2) the volatilities of the
underlying risks, and (3) the correlations of the underlying risks.

9. GENCO PORTFOLIO EXAMPLES

To illustrate these ideas we will now examine two stylized generating

companies, each of which owns a single 100 MW gas-fired generating unit.
One company owns a base-load generating unit, the other a dispatchable
unit. The base-load unit runs around the clock throughout the year whereas
the dispatchable unit runs only when power prices exceed the sum of fuel
and variable operating costs. The heat rate for the base-load unit is 8,000
BtU/kWh while the heat rate for the dispatchable unit is 10,000 Btu/kWh.
Fixed operating and maintenance costs are $50,000/month for both units.
Figures 10 and 11 (below) show the forward price curves for the relevant
power and gas markets over a twelve-month time horizon (assuming that
today is December 1997). Notice that the forward curve for power exhibits a
summer peak whereas the forward curve for gas exhibits a winter peak.
Having set the stage, we will now do two things. First, we will explore the
implications of alternative levels of volatility and correlation for the value
and risk of the two GenCos. Second, we will demonstrate that the power
and fuel contracts that hedge the price risks associated with the two GenCos
differ in a qualitative as well as a quantitative way. In particular:

1. The value of the dispatchable generating unit is very sensitive to the

volatilities and correlation of the power and gas markets. In contrast,
the value of the base-load unit is not sensitive to these parameters.
2. The risk of both the dispatchable unit and the base-load unit are
sensitive to both the volatilities and correlation of power and gas.
3. The market risk associated with the base-load unit can be largely
eliminate through a combination of forward sales and purchases. To
obtain a comparable degree of risk reduction for the dispatchable
unit, an option contract is required.

These results could be anticipated if one thinks in terms of the building

block framework discussed earlier. In particular, the base-load generating
unit is analogous to a bundle of forward contracts to buy power and sell fuel,
whereas the dispatchable unit is analogous to a portfolio of options to
exchange fuel for power.
Pricing in Competitive Electricity Markets 237

Forward Price Curve (On Peak Power')

$'1v1Wh
35

30

25

20

15

10

0
3l 3l
...,~ ~

Figure JO(a). Forward Price Curve (On Peak Power)

Forward Price Curve (Off-Peak Power')

$lMWh
18
16
14
12
10
8
6
4
2
0
<Xl <Xl <Xl <Xl <Xl
83
q>
...,.,
c
0>
.6 .,
0>
~
0>
~ t., C:
...,
If :; ~ :; ::J

Figure IO(b). Forward Price Curve (Off-Peak Power)

238 Pricing In Competitive Electricity Markets

Forward Price Curve (Gas Market)

$IMWh
$2.05l
$2.00
$1.95
$1.90
$1.85
$1.80
$1.75
$1.70
$1.65
$1.60
$1.55
~ ~ ~ ~ ~ ~
I
C: .6 C:
'"
-, Cll
Ll. ~ ~ :J
-,

Figure 11. Forward Price Curve (Gas Market)

Base-LoacI Genco Cash Row

$1,600
Low Power Price Volatility

$1,200

I
$800

$400

-$400

-$800
8l
i i 8l
CD CD
~ ~ ~
~ '" ~ ~
q>
~ !
C: c:: ....,
~
"5
....,
'" ~ ~ ....,
:J

1- Expected - High - Low 1

Figure 12(a). Base-Load Genco Cash Flow - Low Power Price Volatility
Pricing in Competitive Electricity Markets 239

Dispatchable Genco cash Row

$1,000 Low Power Price Volatility

$1,200

-$400

-$000 +--+----t-+--+----t-+--+----t-+--+----t---l

1- Expected - High - Low 1

Figure 12(b). Dispatchable Genco Cash Flow - Low Power Price Volatility

9.1 Implications of Volatility and Correlation for Value

and Risk

To begin, suppose the volatility of prices in the on-peak power market is

25 percent/year and the volatility of off-peak prices is 12 percent/year, for all
delivery dates. Figure 12 (above) shows the present value of the expected
cash flows from the two GenCos over a twelve-month time horizon.
Specifically, the solid lines represent the present value of the cash flows each
month and the "bands" around the solid lines represent the range of possible
outcomes, with the upper and lower "bars" representing the 95 and 5 percent
confidence levels, respectively. There is pronounced seasonality in the cash
flows to both GenCos, due primarily to the seasonality in the underlying
power market. Cash flows are higher for the base-load unit than for the
dispatchable unit, due to the heat rate differential. However, the range of
possible outcomes is greater for the base-load unit in the shoulder and winter
months than it is for the dispatchable unit. In particular, the cash flows to
the dispatchable unit are never less than -$50,OOO/month whereas the 5
percent bar is as low as -$200,OOO/month for the base-load unit (in
December). This is due to the dispatch flexibility inherent in the
240 Pricing In Competitive Electricity Markets

dispatchable unit, which allows the owner to limit cash losses in the event
power prices are low in relation to gas prices.
Now suppose instead that the volatilities of peak and off-peak prices are
twice what they were in the preceding example (50 percent/year and 25
percent/year, respectively). Figure 13 (below) shows that the cash flows to
the base-load unit in this case are the same as they were in the preceding
example, but the range of possible cash flow outcomes is much wider. In
contrast, the cash flows to the dispatchable unit in the winter and shoulder
months are substantially larger in the "high" volatility case than in the "low"
volatility case. This is especially noticeable in the month of December,
where the present value of cash flows is about $150,000/month in the latter
case versus $100,000/month in the former. It is due to the fact that the
owner of the dispatchable unit can capture the benefits of favorable price
movements and avoid the costs of unfavorable movements. (The value of
cash flows in the summer months is not affected by the change in volatility,
since the probability that power and gas prices in the summer will not justify
running is very low.) Notice, too, that although the range of cash flow
outcomes for the dispatchable unit increased with higher volatility, the
down-side is still limited to the $50,000/month fixed operating cost. The
down-side for the base-load unit, in contrast, is much greater.

Bas&LoacI Genco Cash Row

$1,600 Hig-. Pov.Ier Price Volatility

$1,200

r. . ·. . . V. . . . . . . . . . . .""I'--.-
..rT....Ctr......1 . . . . =.... -='=
.....=""
...........

-$400

-$800 +--+-t---+---+-+--+-+--+-t---+---+--l
~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~
~ ~ ~ ~ I ~ ~ ~ ~ 8 ~ ~
I-~ed - fig, - Low I

Figure J3(a). Base-Load Genco Cash Flow - High Power Price Volatility
Pricing in Competitive Electricity Markets 241

Dispatchable Genco Cash Row

$1.600
High Po\wr Price Volatility

$1.200

-$400

-$800
l!i gs IX)
l!i l!i gs l!i
....,~
iJ
~
~
~
~
Jt § c':
....,::l -5
...., ~ 6.
8l ~ ~ ~
[- Expected - High - LON [

Figure J3(b). Dispatchable Genco Cash Flow - High Power Price Volatility

Base-! oacI Genco Cash Row

$1.600
Power and Gas Price Volatility

$1,200

$600

I
'Ui'

V "",,-
$400

$0
TTl I
--F----·T-------t-----'l1·--·---·----·------------·----· ------. -_..... ------- ------

-$400

-$600
gs IX)
0> l!i
IX)
0>
IX)
0> l!i ~ l!i l!i IX)
0> l!i i§l
C:
~ ~ ~ ~ § C:
....,::l ....,
'5
~ ! B ~ ~
I-Expected - High - Low I

Figure 14(a). Base-Load Genco Cash Flow - Power and Gas Price Volatility
242 Pricing In Competitive Electricity Markets

Dispatchable Genco Cash Aow

Power and Gas Price Volatility
$1,600

$1,200

..
U>
"E
::I
0
.c
t::..
$800

$400

$0 ..~.....l .......:.......I.......
!1 I
u . . . . .~l. . . 1. . d. . .
.$400

.$800
co co co co co co co co co co co co
0>
'6 'JOJ: "l' 0> 0>
c "l' "l' "l' 0> 0>

~
C
OJ
-, <Il
u. ~ ~
>-
OJ
~
~
-,
s-, Ol
«
~
c.
c?5 8 ~
z 0

I-Expected - High - Lowi

Figure 14(b). Dispatchable Genco Cash Flow - Power and Gas Price Volatility

Up to this point we have ignored volatility in the gas market-that is, we

have assumed it was zero. If we suppose instead that the volatility of gas
prices is 40 percent/year for all delivery periods, we obtain the results
depicted in Figure 14 (above). The results are analogous to those we
observed when we increased volatility in the power market. The expected
cash flows to the base-load unit are unaffected by the change in volatility,
whereas the expected cash flows to the dispatchable unit have increased.
The range of possible cash flow outcomes has increased for both units, but
as before the losses on the dispatchable unit are limited to the fixed operating
cost.
Figure 15 (below) shows the results of a fourth case in which we
recognize not only volatility in power and gas prices but also correlation.
These results are based on an assumption that the correlation between gas
and power prices is 0.75 in each delivery period. In contrast, the previous
results were based on an assumption that power and gas prices have zero
correlation. Notice that adding positive correlation is similar to reducing
power or gas price volatility. Since power and gas prices tend to move
together, price uncertainty in the gas market tends to offset price uncertainty
in the power market. The gas and power exposures in this case are partial
hedges, since one is being produced while the other is consumed.
Pricing in Competitive Electricity Markets 243

Base-Load Genco Cash Flow

$1,600
Correlated Power and Gas Markets

$1,200

...
co
c
$800

::
1r............. >=V
$400
~r-..........+==1
::J
0
.c
t:. $0 ·-r-....
T
.................................:::t::= .............

·$400

·$800
'"~ '" '".!. '"'" '" '" '"-'- '"'"6> '" '" '"'" '"'"6
..,'c:" IfiJ'" :::;'"'" ~ :::;'"'>." ..,'"c ..,'" <t: C/l'"6-
::J ::J
~
::J
Z Cl
Q)
8 Q)

I-Expected . High - Low I

Figure 15(a), Base-Load Genco Cash Flow - Correlated Power and Gas Markets

Dispatchable Genco Cash Flow

$1,600
Correlated Power and Gas Markets

$1,200

I
$tm

$400
T 1 L V
.............
t--
T..
3D ··.L····Cf······-l.--·····t······t······ ...... ..................... ._----- ------- ------

.$400

.$tm
3l 3l 3l 3l 3l 3l 3l 3l 3l 3l 3l
~ ~ ~ ~ :f ....,5 ....,
~
~
c.
c?l ~ ~ ~
1- E>q:Jected - H\t1 - low1

Figure 15(b). Dispatchable Genco Cash Flow - Correlated Power and Gas Markets
244 Pricing In Competitive Electricity Markets

9.2 Hedging Price Risks

We have already observed that since the base-load generating unit runs
throughout the year in all hours it is analogous to a portfolio of forward
contracts. Specifically, it is equivalent to a portfolio of contracts to buy 100
MW of power and sell the equivalent quantity of gas (at a 8,000 BtU/kWh
heat rate). Offsetting the price risk of this GenCo is straightforward: sell
forward 100 MW of power and buy forward the corresponding quantity
(about 600,000 MMBtu) of gas. Figure 16 shows the results. It is a set of
three cumulative probability distributions for the portfolio cash flow for the
month of July 1998. One ("unhedged") shows the cash flow distribution
when the only asset in the portfolio is the base-load unit. Another ("with
power contract") shows the cash flow distribution when a 100 MW fixed-
price, fixed-quantity power sale contract is added. The third ("with power
and gas contracts") shows the portfolio when there is a gas purchase contract
in addition to the generating unit and power sale contract. The results are
clear. Selling power forward reduces the portfolio risk somewhat, but a
great deal of uncertainty remains. However, if gas is purchased forward too,
then the risk associated with the GenCo portfolio is essentially nil.

Base-Load Genco Cash Flow Distribution

July 1998
100% ,------~-----"-.-----:;;;:;--_::::::==_=t

75%
With Power
and Gas
Contracts
- - - -....
\
Unhedged

50%

25% With Power

Contracts

O%+---==4=~~~---+~-~---+--~
·$800 ·$400 $0 $400 $800 $1.200 $1.600
(Thousands)

Figure 16. Base-Load Genco Cash Flow Distribution (July 1998)

Pricing in Competitive Electricity Markets 245

The preceding is an example of "delta hedging." We bought and sold

sufficient quantities of gas and power to offset the delta exposures of the
initial (unhedged) portfolio. The net risk exposures of the portfolio are
therefore zero. Since the portfolio has no gamma (or higher order) risk
exposures, changes in the market prices of gas and power while the hedge
contracts are in place will not affect the market value of the portfolio. The
portfolio is said to be "delta-neutral."
The sizes of the delta hedge positions were easy to figure out for the
base-load GenCo because the power production and fuel consumption are
deterministic. Delta hedging is more complicated for the dispatchable
GenCo because the dispatch decisions depend on future power and fuel
prices, which are uncertain. Delta hedge positions can be derived from a
valuation model for the generating unit. Our valuation model indicates that
we need to sell forward about 95 MW of power on-peak and about 25 MW
of power off-peak, and we need to buy forward about 375,000
MMBtulmonth of gas to delta hedge the dispatchable GenCo. Figure 17
shows the results.

Dispatchable Genco Cash Flow Distribution

July 1998
100% ~----"---------:r::--"7-----::==~---'

With Power ~
75% and Gas .
Contracts Unhedged

50%

With Power
25% Contracts

0% +----+--===:::Y------.l'-+----+-----If-----j
-$800 -$400 $0 $400 $800 $1,200 $1,600
(Thousands)

Figure 17. Dispatchable Genco Cash Flow Distribution (July 1998)

The delta hedge contracts reduce the risk of the dispatchable GenCo
considerably, but substantial risk remains. Upon reflection this is not
surprising since this portfolio contains embedded options. A delta-neutral
portfolio of options will be hedged only for "small" changes in the
246 Pricing In Competitive Electricity Markets

underlying commodity prices. A more robust hedge could be established by

adding contracts that offset the gamma exposures as well as the delta
exposures-that is, by constructing a "gamma-neutral" portfolio. In fact, we
could hedge the price risk associated with the dispatchable GenCo
completely by selling power under a contract that mimics the financial
characteristics of the unit. This contract would be an optional sale of power
with the contract (strike) price indexed to the price of gas. v Figure 18
compares and contrasts the risk characteristics of the initial (unhedged)
portfolio, the delta hedged portfolio ("with power and fuel contracts"), and
the gamma hedged portfolio ("with indexed option").

Dispatchable Genco Cash Flow Distribution

100% ,--_ _ _---;c--_~Ju=IL...:_'19:o:9~8----_===___,

75%
Unhedged
Option
50%

With Power
25% and Gas
Contracts

O%+---~-~~-~~--~---+--~

.$800 .$400 $0 $400 $800 $1,200 $1.600

(Thousands)

Figure 18. Dispatchable Genco Cash Flow Distribution (July 1998)

The preceding examples illustrate some basic principles of managing

price risk. To obtain the results we assumed that there was no uncertainty
about the performance of the two units. In reality, of course, there is
operational risk (the possibility of forced outages) as well as market risk.
The delta and gamma hedge positions could be modified to reflect the
possibility of forced outages, but the resulting portfolios would not be
completely hedged. Nevertheless the examples make clear that a very large
amount of risk reduction can be achieved through judicious design and use
of commodity contracts.
Pricing in Competitive Electricity Markets 247

REFERENCES

i An option that can be exercised any time prior to the expiration date is called an "American"
option. An option that can be exercised only at expiration is called a "European" option.
The discussion in this section is limited to European-style options.

ri If Party A needs oil, she can purchase it in the spot market rather than exercising her call
option.

iii The log-normal probability distribution has the property that prices cannot be less than zero
but are unbounded from above. In other words, the distribution of prices is "skewed to the
right".

iy To say that an option is "in the money" ("out of the money") means that it would yield a
positive (negative) payoff if exercised immediately. To say that an option is "deep" in or
out of the money implies that the exercise decision is essentially a foregone conclusion.

Y To mimic the unit perfectly, the contract price would be equal to the product of the gas price
and a constant equal to 10,000 BtulkWh.
Chapter 15

Statistical Approaches to Electricity Price Forecasting

J. Stuart McMenamin, and Frank A. Monforte

Regional Economic Research, Inc.

Key words: Electricity Price Forecasting; Neural Networks; PJM Prices; Price Forecasting
Methods.

Abstract: In this paper, a variety of modeling approaches are applied and evaluated for
forecasting electricity prices. Methods include time-series models, regression
models, and artificial neural network models. The paper discusses the nature
of the price forecasting problem and identifies reasons why flexible
approaches, such as neural network models, are well suited to this application.

1. INTRODUCTION

With the advent of competition, hourly electricity prices are being

determined by a variety of market mechanisms, rather than cost-based
engineering calculations. As a result, electric utilities, generators, and
traders face a new set of short -term forecasting problems. These problems
are unlike those in other industries, since electricity must be produced at the
same time that it is consumed. As a result prices are determined on an
hourly basis, twenty-four hours a day, seven days a week.
Historically, price forecasting has been performed with least-cost
optimization models. These models compute marginal cost based on
assumptions about system loads, power plant availability, and fuel prices.
These models do not explain price variations related to market strategy and
to buyer and seller behavior in a market system. Statistical models, which
by their nature reflect actual market outcomes, are better suited to short-term
forecasting in this dynamic environment.

A. Faruqui et al. (eds.), Pricing in Competitive Electricity Markets

© Kluwer Academic Publishers 2000
250 Pricing In Competitive Electricity Markets

In this paper, a variety of modeling approaches are applied and evaluated

for forecasting electricity prices. Methods include time-series models,
regression models, and artificial neural network models. The paper
discusses the nature of the price-forecasting problem and identifies reasons
why flexible approaches, such as neural network models are well suited to
this application.
For each approach, model estimates and forecasts are developed using
hourly price data for the PJM (Pennsylvania, New Jersey and Maryland)
power pool area. The modeling approaches are compared based on accuracy
for day-ahead forecasting.

2. PRICE MODELING APPROACHES

In competitive electricity markets, the market-clearing price is the hourly

bid of the last generation unit to meet system demand. In most systems, all
suppliers are paid the hourly market-clearing price. In a perfectly
competitive market, the market-clearing price will be equal to the marginal
cost of the last supplier. But existing electricity markets are far from
meeting the requirements for perfect competition. Reflecting the fact that
there are a limited number of suppliers and that customer demand is highly
inelastic with respect to the market price, there is significant room for
exploitation of market power, especially in periods of high demand.
There are two approaches being used to forecast market prices. The first
is a simulation method based on models of production cost. The second
involves application of statistical methods to historical market data.
The production cost method involves a simulation of plant dispatch and
inter-pool exchanges to meet hourly demands. These methods typically
assume that plants are dispatched based on lowest running cost of the next
available generation unit, subject to operating and transmission constraints.
This approach requires detailed data and assumptions about inventory of
generation plant, including operating capabilities, operating costs, and
geographic location with respect to transmission facilities. While these
models have proven to be very useful for assessing long-term market
options, they are not well suited to the modeling of bidding strategies in a
market setting.
In contrast, statistical methods relate market prices to observed factors
that are believed to impact prices. These factors can include both demand
side and supply side variables, and a variety of model specifications and
techniques are available. Since bidding strategies are embedded in the
observed market outcomes, these methods will work well as long as
strategies, constraints, and market rules remain stable or evolve slowly. In
Pricing in Competitive Electricity Markets 251

what follows, we look at several approaches to short-term statistical

modeling using data for the PJM market.

3. DATA

The dependent variable data is the average on-peak price in the PJM
market. The on-peak period is defined to be the hours between 8:00 A.M.
and 11:00 P.M., which is a sixteen hour block. Data values are available
from April 1998 through the present. The PJM market is currently in
transition from a power pool using least-cost dispatch to a competitive
market based on generator bidding. At this point PJM prices still reflect
dispatch costs more than competitor bidding. Still these prices pose a
significant modeling challenge, and it is reasonable to believe that methods
that work well with these data will also work well in a full bidding context.
Explanatory variables fall into three categories. First, from a time-series
perspective, there is the history of the market price itself. In a day-ahead
market, lagged price data often have high explanatory power, although the
pattern of weekdays, weekends, and seasons introduces some interesting
modeling problems for time-series models.
The second set of explanatory factors is demand-side factors. Hourly
electricity use reflects the life patterns of people, mechanical systems
interacting with weather, cloud cover, timing of sunrise and sunset, water
temperatures, and other similar factors. Because system load can typically
be modeled and forecasted with high accuracy, we proceed here using the
actual demand levels as an explanatory variable, rather than the indirect
variables for weather and calendar effects.
The third set of explanatory factors is supply-side factors. These factors
include fuel prices, generation unit availability, transmission constraints, and
in some markets, hydro flows. Also, in periods of high demand the load
levels in surrounding areas can have a significant impact on local prices,
reflecting the high price of imported energy and the high opportunity cost of
bidding into the local market. The supply-side variables included here are
nuclear capacity on-line and natural gas prices. The data are presented
below.
Figure 1 (below) shows the PJM average on-peak price. Over the
historical period, the mean value is about $25 per MWh, and most
observations are between $15 and $30. On several days, however, the price
shows a significant and short-lived spike, with hourly values nearing $1000,
bringing the average on-peak price for the day above $100 on occasion. (In
the numbers shown here, the hourly price has been capped at $200 before
computing the daily average).
252 Pricing In Competitive Electricity Markets

$120,----------------------------,

$100

$80

$60

$40

$20

Figure 1. PJM Average On-Peak Price ($/MWh). April 98 to May 99

Figure 2 shows the corresponding values for on-peak energy use. These
data have an average value of about 500 billion watthours (GWh), implying
an average load of about 31 GW during on-peak hours. The data show a
strong weakly cycle as well as a weather-driven seasonal cycle.

800,-------------------------,
700

600

500

400

300

200

100

Figure 2. PJM On-Peak Energy Demand (GWh). April 98 to May 99

Figure 3 (below) shows available nuclear capacity measured in million

watts (MW). The average value is about 11.5 GW, with a maximum value
Pricing in Competitive Electricity Markets 253

of about 14 GW. Unit availability reflects both planned and unplanned

outages.

14,000 r-------------------------,
12,000

10,000

8,000

6,000

4,000

2,000

Figure 3. Available Nuclear Capacity (MW), April 98 to May 99

~.OOr_-----------------------_,

$2.50

$2.00

$1.50

$1.00

$0.50

$0.00 L----''--........_ - - I . _........_ - - ' - _........._ ......._~_''___''____'____'_~

~ _ ~ ~ ~ _ ~ b ~ ~ ~ ~ ~ _

Figure 4. Gas Prices at Henry Hub ($/mmBtu), April 98 to May 99

Finally, Figure 4 (above) shows natural gas prices at the Henry Hub. The
average price over this period was about $2.00 per mmBtu, which would
254 Pricing In Competitive Electricity Markets

translate to a marginal fuel cost of about $20 per MWh at a heat rate of
10,000 BtU/KWh.
Comparing Figures 1 and 2, it is clear that high prices occur in periods of
high demand. The relationship is not perfect, however, as shown in Figure 5
(below). This figure provides a scatter plot of on-peak prices versus on-peak
loads, coded by type of day. The figure shows that all of the high load and
high price days are weekdays. However, not all high-load days have high
prices. For example, on days with on-peak energy near 700 GWh, the
average on-peak price ranges from $40 to more than $100. Despite this wide
dispersion, the chart does suggest that the relationship between loads and
price is nonlinear.

140

120
.r:
:;:
100
~
.~
Ii: 80
'"co
CI)
11. 60
C
0

'""
40
I!
CI)
>
c( 20

o~-----+------~----~------~----~------~----~
400 450 500 550 600 650 700 750
On Peak Energy (GWh)

Figure 5. Scatter plot of On-Peak Price vs. Energy

4. MODEL SPECIFICATIONS
A series of models is estimated using these data. We begin with an
exponential smoothing model and an ARIMA model. These data-driven
models do not take advantage of the demand and supply-side variables.
Next a linear regression model is estimated. Finally, a neural network
extension of the linear model is estimated to capture key nonlinearities and
interactions.
The forecasting equation for a one-day-ahead forecast with an exponential
smoothing model is as follows:
Pricing in Competitive Electricity Markets 255

pI = (Level l-1+ Trend 1-1 )x DayMule-7 (1)

where P is the value of on-peak prices, t is the current period, and Level,
Trend, and DayMult are variables generated by the smoothing process. The
smoothing equations use the multiplicative seasonal form, sometimes
referred to as the Holt-Winters method. In this application, the seasonal lag
is set to seven, so that the DayMult variables point to the same day in the
previous week. With this modification, the smoothing equations are as
follows:

Levell =ax[ pI )+(1-a)X(Level l- 1 +Trend l- 1 ) (2)

DayMult l_7

pI
Trend I = bx [ (3)
DayMult l_7

DayMule =cx[ pI 1+ (1-c)xDayMule-7 (4)

Levell)

The forecasting equation for a one-day-ahead forecast with a seasonal

ARIMA model is as follows:

pI = SARIMA(p, d, q)(sp, sd, sq) (5)

where SARIMA represents a seasonal autoregressive (AR) moving average

(MA) process with autoregressive terms of order p and sp, moving average
terms of order q and sq, differencing of order d, and seasonal differencing of
order sd. In this application, the seasonal periodicity is seven days, implying
that seasonal lags point to the same day in the preceding week.

The linear regression model is as follows:

(6)

where X represents a set of demand and supply-side explanatory variables.

256 Pricing In Competitive Electricity Markets

The extension of the regression model to include nonlinear nodes from a

neural network model can then be represented as:

(7)

The first summation in (7) repeats the linear regression from equation (6).
The second summation includes a set of N nonlinear nodes from a simple
feedforward artificial neural network. The specific form uses logistic
transformation functions, which can capture a variety of nonlinear responses.
By construction, the variables included in a node are multiplicatively
interactive if they have nonzero parameters in the sum that appears in the
logistic exponent.
The neural network approach and this specific functional form are widely
used in day-ahead forecasting of system loads. The approach is well suited
to this problem because the response of system load to weather is nonlinear
and because there are significant interactions among explanatory variables.
(For a complete discussion of the neural network functional form, see
McMenamin and Monforte, 1998.) As seen in Figure 5, the relationship
between system load and prices also appears to be nonlinear. Further, there
may be important interactions between demand and supply variables that
help explain daily variations in price. The neural network equation provides
a simple way to allow nonlinearities and interactions in the model without
imposing restrictive assumptions about the structure of the relationship.

5. ESTIMATION RESULTS

The first model estimated is an exponential smoothing model using a

Holt-Winters multiplicative form with trend and seasonal elements (See
Table 1, below). As discussed above, the periodicity of the seasonal term is
set to seven days for this application with daily data. The results are
summarized below. This na'ive model will serve as a reference point. The
mean absolute percent error (MAPE) is 22.6 percent and the mean absolute
deviation is about $6 per MWh. These statistics reflect the accuracy of the
model for purposes of day-ahead forecasting.
Pricing in Competitive Electricity Markets 257

Table 1. Exponential Smoothing Summary

Exponential Smoothing Model Summary

Simple Smoothing Parameter .619
Trend Parameter -.006
Seasonal Parameter .321
Adjusted Observations 397
Deg. of Freedom for Error 394
Adjusted R-Squared 0.379
Std. Error of Regression 9.76
Mean Abs. Dev. (MAD) 6.02
Mean Abs. % Err. (MAPE) 22.57%
Durbin-Watson Statistic l.816

Table 2. ARlMA (0,1,4) (1,0,0) Summary

Variable Coefficient StdErr T-Stat
CONST -0.007 0.025 -0.265
SAR(1) 0.219 0.052 4.225
MA(1) -0.380 0.051 -7.460
MA(2) -0.247 0.052 -4.7l7
MA(3) -0.296 0.053 -5.627
MA(4) -0.031 0.051 -0.595
Summary Statistics
Adjusted Observations 391
Adjusted R-Squared 0.415
AlC 4.528
BlC 4.589
Std. Error of Regression 9.55
Mean Abs. Dev. (MAD) 5.78
Mean Abs. % Err (MAPE) 22.03%
Durbin-Watson Statistic 1.992

The second model estimated is an ARIMA model. After examination of

time-series diagnostics and experimentation with various forms, the final
model is a (0,1,4) (1,0,0), implying that the data are differenced, and a model
is fit with an MA4 and a seasonal ARl. As with the smoothing model, the
seasonal periodicity is set to seven days. The coefficients and summary
statistics for this model are presented in Table 2 (above). As is evident in
258 Pricing In Competitive Electricity Markets

Table 2, the ARIMA model provides only a modest improvement with

respect to day-ahead accuracy, with a MAPE of 22 percent and a MAD of
$5.8 per MWh.
The regression model uses a combination of lagged dependent variables
and explanatory variables. The three explanatory variables are on-peak
energy use (OnPeakEnergy), nuclear availability (NukeAvail), and the price
of natural gas (HHPrice). As shown in Table 3 (below), this specification
improves the MAPE value to 19.5 percent and reduces the MAD to about $5
per MWh.
The final specification introduces two nonlinear nodes to the linear model
presented above. The first node includes only on-peak energy as an input
variable. This node will allow representation of nonlinear effects, to the
extent these effects are present. The second node includes the two supply-
side variables, gas prices and nuclear availability, allowing the modeling of
nonlinearities and interactions with respect to these variables. The extended
model is estimated using nonlinear least squares applied to normalized data.

Table 3. Regression Model Results

Variable Coefficient StdErr T-Stat

CONST -28.576 5.896 -4.847
Saturday -1.060 1.574 -0.673
WkDay -2.017 1.287 -1.567
LaglOnPk 0.306 0.050 6.159
Lag20nPk -0.021 0.050 -0.419
Lag30nPk -0.099 0.050 -1.975
Lag40nPk 0.095 0.050 1.877
Lag50nPk -0.012 0.054 -0.232
Lag60nPk -0.032 0.056 -0.582
Lag70nPk 0.046 0.051 0.890
HHPrice 6.595 1.822 3.620
NukeAvail -0.002 0.000 -4.974
OnPeakEnergy 0.116 0.011 10.920
Summary Statistics
Adjusted Observations 392
Adjusted R-Squared 0.569
Durbin-Watson Statistic 1.695
AIC 4.267
BIC 4.531
Std. Error of Regression 8.18
Mean Abs. Dev. (MAD) 5.06
Mean Abs. % Err. (MAPE) 19.52%
Pricing in Competitive Electricity Markets 259

Table 4. Neural Network Model Results

Coefficient Value StdErr T -Stat

Linear: Intercept 2.417 1.169 2.068
Linear: Saturday -0.008 0.039 -0.205
Linear: WkDay -0.029 0.045 -0.651
Linear: Lag 10nPk 0.162 0.047 3.469
Linear: Lag20nPk -0.094 0.045 -2.065
Linear: Lag30nPk -0.065 0.045 -1.442
Linear: Lag40nPk 0.090 0.045 2.021
Linear: Lag50nPk 0.007 0.048 0.137
Linear: Lag60nPk -0.003 0.050 -0.053
Linear: Lag70nPk 0.013 0.047 0.283
Linear: HHPrice -0.263 0.230 -1.141
Linear: NukeAvail 0.003 0.165 0.018
Linear: OnPeakEnergy 0.479 0.068 7.086
Node1: Slope -3.243 0.953 -3.404
Node1: Bias 8.425 2.472 3.408
Node1: OnPeakEnergy -3.470 1.226 -2.830
Node2: Slope 1.659 1.427 1.163
Node2: Bias -0.348 0.326 -1.066
Node2: NukeAvail -0.695 0.431 -1.615
Node2: HHPrice 1.612 1.093 1.475
Summary Statistics
Adjusted Observations 392
Adjusted R-Squared 0.663
AIC 4.037
BIC 4.372
Std. Error of Regression 7.23
Mean Abs. Dev. (MAD) 4.49
Mean Abs. % Err. (MAPE) 17.15%
Durbin-Watson Statistic 1.619
260 Pricing In Competitive Electricity Markets

As shown in Table 4, this specification provides a further improvement in

model accuracy, with day-ahead MAPE values dropping to 17 percent and
MAD values dropping to $4.5 per MWh. The actual and predicted values
are presented in Figure 6 (below). As is evident, the model does not fully
predict the price spike values that occurred in the summer of 1998. This is
not surprising given the price dispersion that is evident for high load levels
in the scatter plot in Figure 5. Otherwise, however, the day-ahead model
tracks actual outcomes and changes in price fairly well.
Comparing the auto-regressive terms, the coefficient on the one-day lag
of price (Lagl0nPk) is about half the level in the neural network model as it
is in the regression model. This indicates that the neural network model
places higher reliance on the explanatory variables and lessor reliance on the
time-series properties of the residuals.

lWr-------------------------------------------~

100

.- -
o~~--~--~--~~~~--~--~--~~--~--~~

~ ~ ~ ~ ~ ~ ~ ~ ~ .-
Figure 6. Actual and Predicted Values - Neural Network Model

To understand the role of the parts of the neural network model, the
contribution of each component (the linear component and the two nonlinear
nodes) to the total predicted value is presented in Figure 7, Figure 8, and
Figure 9 (below). The linear node appears to account for most of the
seasonal variation and also captures weekly cycles. The first nonlinear node
fires only under price-spike conditions. The second nonlinear node is driven
mostly by natural gas prices. This node has the greatest contribution when
gas prices are high and nuclear capacity is low, as was the case in both the
spring of 98 and the spring of 99.
Pricing in Competitive Electricity Markets 261

100.00 . - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - ,

80.00

Total Predicted Value

60.00 - - Linear Terms

40.00

20.00

Jan Feb Mar

-20.00 ~------------------------------------------------------~

Figure 7. Contribution of Linear Terms to Predicted Value

90.00

80.00

70.00
_... Total Predicted Value
60.00 - - Nodel Contribution

50.00

40.00

30.00

20.00

10.00

0.00
Apr May Jun

Figure 8. Contribution of Node 1 (On Peak. Energy) to Predicted Value

262 Pricing In Competitive Electricity Markets

90.00

80.00

70.00
Total Predicted Value,
60.00 - Node2 Contribution

50.00

40.00

30.00

20.00

10.00

0.00
Apr May Jun

Figure 9. Contribution of Node 2 (Supply Terms) to Predicted Value

6. CONCLUSION

As competitive electricity markets evolve, interest in price forecasting

will increase significantly. Whereas under the regulatory compact, utilities
are responsible for prudent planning and are guaranteed a reasonable rate of
return on investment decisions, in a competitive generation market, profits
will be determined by the relationship between price and cost. Owners of
generation assets will want the best possible price forecasts to make the best
decisions about contracting and bidding strategies. Retailers will want the
best possible price forecasts to develop strategies for covering the loads of
their customers. And utilities, which may be both owners of generation and
retailers, will be interested in price forecasting for purposes of trading and
risk management.
As the analysis above suggests, electricity price forecasting is a
significant challenge. Price variation is significant on a day-to-day basis,
and prices are even more volatile on an hourly basis. The analysis suggests
that more advanced modeling methods will produce better forecasts.
Moving from naIve methods to advanced methods, such as neural networks
reduces the day-ahead forecasting error from about $6 per MWh to $4.5 per
MWh. By developing better data about supply side factors, it is reasonable
to expect that further improvements can be made. Because of the nonlinear
Pricing in Competitive Electricity Markets 263

and interactive nature of the price responses, this is a good application for
neural network approaches, which provide a flexible nonlinear method.

REFERENCES

Azoff, E. M. Neural Network Time Series Forecasting of Financial Markets. John Wiley
& Sons, 1994.

Kuan, C and H White, "Artificial Neural Networks, An Econometric Perspective."

UCSD Discussion Paper, June, 1992.

McMenamin, J. S. and F. A. Monforte, "Short-term Energy Forecasting With Neural

Networks," Energy Journal, Volume 19, Number 4,1998.

White, H, "Neural-Network Learning and Statistics." AI Expert, December, 1989.

SECTION V

CASE STUDIES
Chapter 16

Using Customer-Level Response to Spot Prices to

Design Pricing Options and Demand-Side Bids

Robert H. Patrick and Frank A. Wolak

Rutgers University and Stanford University

Key words: Competition; Demand-Side Bids; Demand Elasticities; Forecasting; Pricing

Options; Retail Pricing; Spot Pricing; Restructured Electricity Markets.

Abstract: This chapter presents estimates of customer-level demands for electricity by

large and medium-sized industrial and commercial customers purchasing
electricity according to half-hourly spot prices and demand charges from the
England and Wales electricity market. The resulting models can be used to
measure demand-altering effects arising from the imposition of arbitrary time
varying and consumption-dependent energy and demand pricing structures,
including alternative levels of supply reliabilitylinterruptibility and stability or
certainty of prices. Knowledge of customer-level electricity demands is
critical to success in the evolving competitive business environment, as well as
for the success of restructured electricity markets as a whole. Such knowledge
is required to effectively forecast customer price responsiveness and demands,
design pricing options to attract and maintain a profitable portfolio of
customers, and demand-side bid to build a price response into the market
demand used to determine market-clearing electricity and ancillary services
prices.

1. INTRODUCTION

Electric utilities in the U.S. and worldwide are faced with dramatic
changes in market structure and regulations governing their form and
operations. Such restructuring is intended to provide incentives for efficient
pricing, investment, and operations by introducing competition into these
markets. The predominant view is that competition is feasible for
A. Faruqui et al. (eds.), Pricing in Competitive Electricity Markets
© Kluwer Academic Publishers 2000
268 Pricing In Competitive Electricity Markets

structurally or functionally separated firms generating (wholesale) and/or

supplying (retail) electricity, while network services (transmission and
distribution) are most efficiently provided by open access firms facing
alternative (to cost-based) forms of regulation (e.g., price caps). Examples
of this view include the England and Wales (E&W) and California electricity
markets. i These changes in market structure and rules lead to some rather
dramatic changes in the operations of electric utilities relative to the
regulated integrated monopoly status of the past. ii Prospering in this new
environment will require utilities to develop comprehensive strategies aimed
at matching profitable service options to diverse customer needs.
Knowledge of customer-level electricity demands is critical to success in the
evolving competitive business environment, as well as for the success of
restructured electricity markets as a whole. Such knowledge is required to
effectively:

• Forecast customer price responsiveness and demands.

• Design pricing options to attract and maintain a profitable portfolio of
customers.
• Demand-side bid to build a price response into the market demand used
to determine market-clearing electricity and ancillary services prices.

A difficulty in restructured electricity markets has been the lack of a

demand response in the price determination process in wholesale markets.
Demand-side bidding is important in order to introduce a price response into
the price-determination process in markets such as the E&W (which
currently uses a perfectly inelastic day-ahead demand forecast in
determining each half-hour's market price). A difficulty in incorporating a
demand-side into the price determination process has also been that few
consumers face prices that vary with the half-hourly wholesale price of
electricity. Customers are not provided with incentives for efficient
conservation and substitution of electricity away from peak periods if they
do not face half-hourly prices that reflect the real-time cost of purchasing
wholesale electricity. This can have adverse consequences for system
reliability, particularly during peak demand periods, when the transmission
and distribution network becomes extremely stressed.
Accurate measurement of the within-day price response of customers is
an important necessary ingredient for any electricity retailer to aggressively
demand-side bid into a wholesale electricity market. Active demand-side
participation builds a price-response into the market price-setting process,
which leads to less capacity being called upon to generate in response to
higher bids by generators. Benefits to retailers from significant demand-side
bidding include the reduced magnitude and variability of market prices and
 Pricing in Competitive Electricity Markets 269

decreased costs of contracts protecting retailers (selling to customers at fixed

prices) from relatively large market prices.
Setting retail prices for electricity in restructured markets is complicated
by the fact that customers now will have their choice among suppliers.
Moreover, suppliers have limited information on the load characteristics of
customers, particularly new ones, including those that may be attracted from
other retailers. iii This lack of information has always existed in regard to
new utility customers. However, in the new competitive retail environment,
cross-subsidizing certain customers at the expense of others is no longer
viable because the customers providing subsidies can now purchase from
another supplier. If the supplier sets too high a price, existing and potential
customers will choose to purchase their electricity from a competitor.
Conversely, setting the price too low will at least imply foregoing available
profit, and possibly lead to the extreme of subsidizing the customer's
electricity consumption. Understanding the structure of customer-level
demand is necessary to effectively design pricing options in markets with
competItIon. Even if all competitors have exactly the same costs, the
supplier with superior information concerning customer demands over time
will be able to design pricing options which attract and maintain a profitable
collection of customers, at the expense of less well-informed suppliers.
The supplier with superior demand information will also be able to more
accurately forecast future customer demands, which remains necessary for
planning and investment. Forecasting customer demands is necessary in
restructured markets for all of the reasons forecasts were made in past.
Precise forecasts are even more critical because utilities are unlikely to be
, able to pass on the cost of all load shape forecast errors on to customers who
now have a choice of suppliers. In a competitive retail market, forecast
errors will be absorbed largely by utility shareholders.
This chapter presents some of the results and implications of our
estimates of customer-level demands, which are fully developed in Patrick
and Wolak (1997, 1999). In these reports we estimate customer-level
demands for electricity by large and medium-sized industrial and
commercial customers purchasing electricity according to half-hourly spot
prices from the E&W electricity market. iv The E&W market was established
in 1990 and has served as a model for restructuring worldwide.
The demand models developed and estimated in Patrick and Wolak
(1997, 1999) quantify the extent of intertemporal substitution in electricity
consumption between pricing periods within the day due to changes in the
E&W pool prices, and distinguishes between the demand-altering effects of
changes in pool prices and changes in demand charges. These customer-
level demand relationships are essential in the design of pricing options to
attract and maintain customers in a market with competing suppliers. They
270 Pricing In Competitive Electricity Markets

can be used to measure the demand altering effects arising from the
imposition of arbitrary time varying and consumption-dependent pricing
structures. In particular, the models can be used to forecast customer-level
electricity loads, revenues from electricity supply, and customer benefits
under alternative rate structures; estimate the own-price and cross-price
elasticities, which can be used in rate simulation and optimization programs;
forecasting; and develop demand-side bids.
The remainder of this chapter proceeds as follows. In the next section we
describe the spot price determination, contracts based on these prices, and
the data used to estimate the model. Section 3 discusses the impossibility of
predicting price responsiveness of customers in restructured markets using
aggregate data. This is followed by a discussion of the price elasticity
estimates for a sample of these customers in Section 4. Section 5 contains
brief examples to illustrate how the econometric model could be used to
compare alternative pricing structures and to implement demand-side
bidding. This chapter closes with a discussion of future research on
designing pricing options and demand side bids under competition.

2. SPOT PRICES AND POOL PRICE CONTRACTS

Generators offer prices at which they will provide various quantities of

electricity to the E&W pool during each half-hour of the following day.
These prices and quantities submitted by generators are by the National Grid
Company (NGC) to determine the merit order of dispatching generation and
reserve capacity. NGC computes a forecast of half-hourly system demands
for the next day. The system marginal price (SMP) for each half-hour of the
next day depends on the price bid on the marginal generation unit required to
satisfy each forecast half-hourly system demand for the next day. For each
day-ahead price-setting process, the 48 load periods within the day are
divided into two distinct pricing-rule regimes, referred to as Table A and
Table B periods. During Table B periods, the SMP is equal to the bid price
of the last generating facility necessary to meet NGC's forecast of demand
for that half-hour. In Table A periods, average start-up and no-load costs
over an expected dispatch horizon are added to bid prices to produce
adjusted offer prices and the SMP is the least adjusted bid price necessary to
meet NGC's forecast of total system load. v The SMP is one component of
the price paid to generators for each MWh of electricity provided to the pool
during each half-hour.
The price paid to generators per MWh in the relevant half-hour is the
Pool Purchase Price, defined as PPP = SMP + Cc. CC is the capacity
charge, where CC = LOLPx(VOLL - SMP), LOLP is the loss of load
Pricing in Competitive Electricity Markets 271

probability, and VOLL is the value of lost load. SMP is intended to reflect
the operating costs of producing electricity (this is the largest component of
PPP for most of the half hour periods). VOLL is set for the entire fiscal year
to approximate the per MWh willingness of customers to pay to avoid
supply interruptions during that year. VOLL was set by the regulator at
2,000 £lMWh for 1990/91 and has increased annually by the growth in the
Retail Prices Index (RPI) since that time. The LOLP is determined for each
half-hour as the probability of a supply interruption due to the generation
capacity being insufficient to meet expected demand. The PPP is known
with certainty from the day-ahead perspective.
The pool selling price (PSP) is the price paid by suppliers purchasing
electricity from the pool to sell to their final commercial, industrial and
residential customers. During Table A, half-hours the PSP is:

PSP = PPP + UPLIFT = SMP + CC + UPLIFT.

UPLIFT is a per MWh charge which covers services related to maintaining

the stability and control of the National Electricity System and costs of
supplying the difference between NGC's forecast of the day's demands and
the actual demands for each load period during that day, and therefore can
only be known at the end of the day in which the electricity is produced.
These costs are charged to electricity consumption only during Table A
periods in the form of this per MWh charge. The ex ante and ex post prices
paid by suppliers for each megawatt-hour (MWh) are identical for Table B
half-hours, (i.e., PSP = PPP for Table B periods). Thus, the only energy
price uncertainty from the day-ahead perspective is the UPLIFT component
of the PSP, which is only known ex post and only applies to the Table A
half-hours. vi
By 4 PM each day, the Settlement System Administrator (SSA) provides
Pool Members, which includes all of suppliers, with the SMP, CC, LOLP,
and identity of the Table A and B pricing periods.
Data from a supplier in this market, Midlands Electricity pIc (MEB), over
the fiscal years vii 1991-1995 was used to estimate half-hourly customer-level
demand functions under MEB's real-time pricing program-what MEB calls
a Pool Price Contract (ppC).viii Under this pricing option, customers are
charged according to half-hourly spot prices that clear the E& W electricity
market-the PSP. As discussed above, the PSP for each half-hour of a day
is to a large part known from the day-ahead perspective, but only known
with certainty thirty days ex post of actual consumption. PPC customers also
face a demand charge (termed a "triad charge" in the E&W market) on the
average kWs consumed during the three half-hours coincident with the
largest E&W transmission system demands. ix The PPC was first offered at
272 Pricing In Competitive Electricity Markets

the beginning of the second fiscal year of the E& W market to allow
consumers with peak demands greater than 1 MW to assume the risks of
pool price volatility and therefore allows the electric retailer to avoid the
costs associated with hedging against wholesale price uncertainty. Under the
PPC, wholesale electricity costs for both energy and transmission services
are directly passed through to the customer. For traditional pricing contracts,
the Supplier absorbs all the wholesale price risk associated with the PSP, and
retails it to final consumers according to fixed and deterministic prices (i.e.,
they do not vary with the PSP).
The peak demand size limit on customers given the option to choose their
supplier as well as purchase according to a PPC was reduced to 100 KW in
1994, coinciding with the same limit change in the definition of franchise
customers. This size limit has been phased out over 1998 and 1999 so that
the entire supply segment of the electricity market is open to competition.
Any customer willing to pay the cost of installing meters capable of
recording half-hourly consumption now has the option to pay for electricity
according to pool prices. Patrick and Wolak (1997 and 1999) provide the
number of customer/year pairs in each BIC class (by two-digit BIC code), as
well as a general description of the industries contained in each class, for our
sample of 263 business lines comprising our sample of commercial and
industrial customers. x
The expected PSPs for all forty-eight half-hourly intervals beginning
with the load period ending at 5:30 A.M. the next day until the load period
ending at 5:00 A.M. the following day are faxed to all pool price customers
immediately following the supplier's receipt of the SMP, CC and the identity
of the Table A and Table B periods from NGC. xi The supplier develops
forecasts of the UPLIFT component of the PSP for Table A half-hours and
provides these with the forty-eight half-hourly SMPs and CCs. The PSP
reported in this fax is equal to the PPP in Table B periods and the sum of the
REC's estimate of the UPLIFT and the PPP in Table A periods. The actual
(ex post) PSP paid by electricity consumers on the PPC for Table A periods
is known twenty-eight days following the day the electricity is consumed.
The actual or ex post PSP is equal to the ex ante PSP for Table B periods
because the UPLIFT is known to be zero in these load periods. We also
collected the information contained on the faxes sent to each PPC customer
the day before their actual consumption occurs. In addition, we collected
information on the actual value of UPLIFT for our sample period. Table 1
gives the sample means and standard deviations for the various components
of the PSP for each fiscal year during our sample.
Pricing in Competitive Electricity Markets 273

Table 1. Sample Means and Standard Deviations of Components of PSP

Fiscal Year Mean Std Dev

SMP 1 19.52 4.10

CC 1 1.29 8.76

UPLIFT I 1.61 2.31

PSP 1 22.42 12.72

SMP 2 22.64 4.24

CC 2 0.17 1.70

UPLIFT 2 1.39 1.12

PSP 2 24.19 5.75

SMP 3 24.16 6.71

CC 3 0.28 2.97

UPLIFT 3 2.18 1.62

PSP 3 26.62 8.76

SMP 4 20.78 12.28

CC 4 3.22 24.49

UPLIFT 4 2.38 4.53

PSP 4 26.38 35.08

Customers on PPCs also pay a demand charge. This £IMW triad charge
is levied on the average capacity used by each PPC customer during the
three half-hour load periods ("triads") in which the load on the England and
Wales system is highest, subject to the constraint that each of these three
periods is separated from the others by at least ten days. The precise triad
charge is set each year by NGC (subject to their RPI-X price cap regulation).
274 Pricing In Competitive Electricity Markets

The triad charge faced by these PPC customers was 6,150 £/MW for fiscal
year 1991/92, 5,420 £/MW for 1992/93, 10,350 £IMW for 1993/4, and
10,730 £/MW for 1994/95.
There are various mechanisms that suppliers use to warn their PPC
customers of potential triad periods. Triad advance warnings are generally
faxed to consumers on Thursday nights and give the load periods during the
following week that the supplier feels are more likely to be triad periods.
Triad priority alerts are issued the night before the day that the supplier
considers the probability of a triad period to be particularly high. These
alerts also list the half-hours most likely to be triad periods. To mitigate the
incentive for suppliers to issue triad priority alerts, the regulatory contract
allows a maximum of twenty-five hours of priority alerts each fiscal year.
Historically, all triad charges have occurred in the four-month period from
November to February. xii The actual price for service paid by PPC
customers also contains various other factors, not related to the pool prices,
which are analogous to charges fixed rate customers face. xiii

3. THE IMPOSSIBILITY OF USING AGGREGATE

DATA TO ESTIMATE PRICE RESPONSIVENESS

The price determination process in the E&W market does not use the
actual market demand to set the two largest components of the PSP, the SMP
and Cc. As discussed in Section 2, these prices are set using a perfectly
inelastic (with respect to price) forecast of the market demand for each half-
hour of the next day. In addition, few customers pay for electricity during
any half-hour at the PSP or even at prices that vary with the PSP throughout
the day, month, or year. Consequently, any attempt to estimate a
relationship between the PSP for a given half-hour and the total system load
for that half-hour should not recover the true relationship between final
demand and the half-hourly market price for any customer or group of
customers, because the actual PPP is set by NGC's forecast of total system
load and not actual total system load.
A notable feature of the behavior of the PSP is its tremendous variability,
even over very short time horizons. For example, the maximum ratio of the
highest to lowest PSP within a day is 76.6, whereas the average of this ratio
over all days in our sample period is about 4.1. The maximum ratio of the
highest to lowest PSP within a month is 107.5 and the average of this ratio
over all months in our sample is 11.0. Finally, the maximum ratio of the
highest to lowest PSP within a fiscal year is approximately 117.8.
The E&W total system load (TSL) exhibits dramatically less volatility
according to this metric. For example, the maximum ratio of the highest to
Pricing in Competitive Electricity Markets 275

lowest TSL within a day is 1.89 and the average over all days in the sample
is l.49. Within a month, the maximum of the highest to lowest TSL is 2.38
and the average over all months in the sample is 2.04. Over a fiscal year, the
maximum ratio of the highest to lowest TSL is 3.08. Defining forecasting
accuracy as the standard deviation of the forecast error as a percent of the
sample mean of the time series under consideration, consistent with this
difference in volatility, the TSL can be forecasted much more accurately
over all time horizons than the PSP.
Comparing the time path of PSP to the time path of total system load
bears out our logic for the impossibility estimating price responsiveness
using aggregate data. Figure 1 (below) plots the half-hourly PSP in
(£IMWh) and Figure 2 (below) plots the half-hourly TSL in gigawatts (GW)
of capacity for the more than 17,000 periods for each fiscal year during our
sample period. All of the price graphs are plotted using the same scale on
the vertical axis to illustrate the tremendous increase in magnitude and
volatility of the PSP over the fiscal years. The highest values of PSP within
a fiscal year tend to occur November through February. These are also the
months when there is an enormous amount of price volatility within and
across days. The pattern and the magnitude of this volatility differ markedly
across the four fiscal years in our sample.
Compared to the four graphs in Figure 1, the four graphs in Figure 2
indicate the very predictable pattern of TSL across days, weeks, and years.
In particular, the total demand in a single day in one year is very similar to
the demand in that same day in the previous year. The cycle of demand
within a given week is similar to the cycle of demand within that same week
in another year. Similar statements can be made for the cycles in TSL within
months across different years.
The difference between the four price graphs and the four TSL graphs
illustrates a very important implication of the operation of the E&W market
which prevents a meaningful price-response from being recovered from co-
movements in TSL and the PSP. Despite the large differences in the patterns
of PSP movements, there is no discernable change in the pattern of TSL.
This results from the fact that the vast majority of commercial and industrial
customers, and all residential customers, purchase power on fixed-price
contracts set for the entire fiscal year. These customers do not face any
within-year price changes or even within-day price changes that depend on
within-year changes in the PSP that might trigger a within-day demand
response. In addition, NGC's forecast of TSL, not actual TSL, determines
the half-hourly PPP.
276 Pricing In Competitive Electricity Markets

Price 91-92
PRICE
800

700

600

500

400

300

200

100

0
•1 it tt 'd .L. to
,l••Lj,.• ill.1il L,'it *
i

9 9 9 9 9 9 9 9 9 9
1 1 1 1 1 1 1 2 2 2
0 0 0 0 0 0 1 1 0 0 0 0
4 5 6 7 8 9 0 2 1 2 3 4
0 0 0 0 0 0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 1 1 1 1 1 1
DATE

---PRICE I
Figure 1(a). Pool Selling Prices, April 1991 - March 1992.

Price 92-93
PRICE
800
700
600

500
400
300
200
100
d • ' , i ,I
0
9 9 9 9 9 9
2 2 3 3 3 3

..
0 0
5
0
6
0
7
0
a
a
9
1
0
0
1
0
2
0
3 ..
0

0 0 0 0 0 o 0 0 0
1 1 1 1 1 1 1 1 1
DATE

---PRICE I
Figure 1 (b). Pool Selling Prices, April 1992 - March 1993.
Pricing in Competitive Electricity Markets 277

Price 93-94
PRICE
800

700

600

500

400

.
300

200

100

0
, II I ~ ~.I J II~ I iii II. .....1.... . UIJ.I
9 9 9 9 9 9 9 9 9 9 9 9 9
3 3 3 3 3 3 3 3 3 4 4 4 4

0 0 0 0 0 0 1 1 0 0 0 0
4 5 6 7 8 9 0 2 1 2- 3 4
0 0 0 0 0 0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 1 1 1 1 1 1

DATE

---PRICE I
Figure J(c). Pool Selling Prices, April 1993 - March 1994.

Price 94-95

9 9 9 9 9 9 9 9 9 9 9 9 9
4 4 4 4 4 4 4 4 4 5 5 5 5
0 0 0 0 0 0 1 0 0 0 0
4 5 6 7 8 9 0 1 2 3 4
0 0 0 0 0 0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 1 1 1 1 1 1
OATE

---PRICE I

Figure J(d). Pool Selling Prices, April 1994 - March 1995.

278 Pricing In Competitive Electricity Markets

System Loads 91 -92

50000

9 9 9 9 9 9 9 9 9 9 9 9
1 1 1 1 1 1 1 1 2 2 2 2
0 0 0 0 0 0 1 1 0 0 0 0
4 5 6 7 8 9 0 2 1 2 3 4
0 0 0 0 0 0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 1 1 1 1 1 1
Date

Syatem Load I
Figure 2(a). Total System Loads, April 1991 - March 1992.

System Loads 92-93

9 9 9 9 9 9 9 9 9 9 9 9 9
2 2 2 2 2 2 2 2 2 3 3 3 3
0 0 0 0 0 0 1 1 0 0 0 0
4 5 6 7 8 9 0 2 1 2 3 4
0 0 0 0 0 0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 1 1 1 1 1 1
Oat.

Syatem Load I
Figure 2(b). Total System Loads, April 1992 - March 1993.
Pricing in Competitive Electricity Markets 279

System Loads 93-94

50000

40000

System Load 30000

20000

10000
9 9 9 9 9 9 9 9 9 9 9 9 9
3
0
3
0
3
0
3
0
3
0
3
0
3
1
3 3
1
" " " "
0 0 0 0

"
0
5
0
6
0
7
0
B

0
9
0
0
0 0
2
0
1
0
" 2
0
3
0 0
1 1 1 1 1 1 1 1 1 1 1 1 1
Date

I -- System Load I
Figure 2(c). Total System Loads, April 1993 - March 1994.

System Loads 94-95

System Load

10000
9 9 9 9 9 9 9 9 9 9 9 9 9

" " " " " " " " "

0 0 0 0 10 0 1 1
5
0
5
0
5
0
5
0

"
0
5
0
6
0
7
0
8
0
9
0
0
0 0
2
0
1
0
2
0
3
0
"
0
1 1 1 1 1 1 1 1 1 1 1 1 1
Dot.

- - - System Load I
Figure 2(d). Total System Loads, April 1994 - March 1995.
280 Pricing In Competitive Electricity Markets

The pricing structures under which consumers pay for electricity have
important economic implications for how the supplier chooses to purchase
and/or produce the electricity demanded. Although electricity at any
particular time is a homogeneous good, the reliability of prices and supply,
as well as the relative magnitudes of prices they face, are important
characteristics of the pricing structures from which consumers choose. Since
suppliers provide electricity to most of their customers at rates fixed well in
advance of the realization of pool prices, they normally hedge against this
price volatility by purchasing contracts for differences (CFDs). CFDs are
financial instruments hedging electricity price volatility. CPDs come in
various forms-one-sided, which hedge only against price risk in one
direction or two-sided, which hedge against both upward and downward
price risk. See Wolak and Patrick (1996, 1997) for further detail. CFDs
have been sold by generators as well as financial institutions and traders that
deal in commodity markets and derivatives. xiv They are not contracts to
deliver electricity and do not cover all price uncertainty. For retail
customers under the PPC, the supplier does not face any risk from pool price
fluctuations since the customer has assumed this risk, hence the supplier
would avoid this risk (and the costs of CFDs to cover any of this risk).
Conversely, under a fixed-price contract, a customer faces no price
uncertainty and the supplier faces all of the PPP risk, unless CFDs are
purchased. These fixed prices must include a premium, relative to spot
prices, to cover the costs of the risk the supplier faces and that which is
covered with CFD purchases.
Each supplier offers several fixed-price options to electricity consumers.
For residential customers, suppliers offer a small number of different
standard price contracts, (e.g., the single-price for all load periods contract,
or a two-price contract) (separate prices for day and night load periods). For
business customers, each supplier offers several standard price contracts, but
particularly for very large customers, price contracts are often negotiated on
a customer-by-customer basis. Consequently, for the same half-hour period,
there are hundreds and potentially even thousands of different retail prices
that different customers throughout the E&W system are paying for
electricity. In addition, movements in the PSP, or in any of its components,
generally have no effect on the movements in these contract prices for the
duration of the contract period, usually a fiscal year. The lack of
responsiveness of TSL to changes in PSP does not imply that individual
customers do not respond to price changes. This lack of responsiveness is
indicative of the fact that only a very small fraction of final customers
purchase electricity at the half-hourly PSP, with the remaining vast majority
purchasing electricity on the fixed-price contracts described above and the
fact that forecast TSL, not actual TSL, sets the market-clearing value of PPP.
Pricing in Competitive Electricity Markets 281

An important consequence ofthese two facts is that it makes little, if any,

economic sense to estimate an aggregate demand curve for electricity
involving PSP or PPP as the price variable and actual TSL as the quantity
demanded variable to recover a price-response. Movements in the half-
hourly or the daily average PSP or PPP, which identify the aggregate price
response, are irrelevant to the vast majority of electricity consumers who
instead face prices that are unrelated to any within-year movements in the
PSP or PPP for the entire fiscal year. Consequently, a price response
recovered from regressing the current value of the TSL on the PSP for that
load period is likely to be extremely misleading about the true potential
aggregate price response because only between 5 and 10 percent of TSL is
purchased at PSP and the remaining is purchased according to prices that are
invariant to changes in the PSP for an entire fiscal year.
To estimate the within-day electricity demand response to within-day
changes in the PSP requires a sample of customers actually purchasing
electricity at prices that move with changes in the half-hourly PSP. PPC
customers are ideally suited to this task because the within-day relative
prices that they pay for electricity are based on the half-hourly PSP.
As discussed in Wolak and Patrick (1997), the major source of the large
values of the PSP over the years shown in Figure 1 is the CC, which is
known with certainty on a day-ahead basis. In addition, large values of the
UPLIFT tend to occur in the same load periods within the day that large
values of CC occur, which makes forecasting UPLIFT relatively easier.
Nevertheless, the two largest components of the PSP are known to the
customer before consumption choices are made for the following day, and
the remaining component is forecastable with considerable accuracy. For
example, the sample mean over our four years of data of the half-hourly
difference between the supplier's ex ante forecast of UPLIFT and the actual
ex post value of UPLIFT is 0.07 £/MWh with a standard deviation of 1.16.
The mean absolute deviation of the half-hourly difference between the
supplier's ex ante forecast of UPLIFT and the actual ex post value of
UPLIFT is 0.56 £/MWh with a standard deviation of 1.02. Comparing these
magnitudes to the annual means of the PSP given in Table 1, which are on
the order of 25 £/MWh, shows that the uncertainty between the ex ante and
ex post values of the PSP is relatively small from the day-ahead perspective.
Of course there is still the uncertainty associated with the demand charge.

4. ESTIMATED PRICE ELASTICITIES

In this section we present mean price elasticities for firms across several
industries in the E&W market. Patrick and Wolak (1999) provide the
282 Pricing In Competitive Electricity Markets

complete modeling analyses and results for these and the other industrial and
commercial industries in our sample of ppe customers. Because prices and
demands are extremely variable over the course of the year and within the
day, there is considerable amount variability both within the day and across
days in these own- and cross-price elasticities. In addition, these elasticities
also vary across days due to weather differences across days. The own-price
and cross-price elasticities of demand for any day and load period can be
computed from the models in Patrick and Wolak (1999).
Given the amount of price volatility in the PSP and the expected demand
charge, even the smallest half-hourly within-load-period own-price
elasticities of demand can imply significant load reductions in response to
price increases. Recall the enormous volatility in the PSP shown in Figure 1.
In addition to this variability, the volatility in the expected demand charge
should be taken into account in determining consumer loads.xv In particular,
it would not be unusual to have values of expected prices across days for the
same load period that differ by a factor 20 or 30, which would imply a
substantial reduction in the within period demand. Table 1 gives the sample
mean and standard deviation of expected half-hourly prices for our four
years of data. For some load periods, the standard deviation of the expected
price is more than three times the value of mean, which indicates the
potential for an enormous amount of variability in prices for the same load
peno. d across days. xvi
Firms in the water supply industry pump substantial amounts of water
into storage and sewage-treatment facilities once or twice a day. These firms
generally have the ability to shift this activity to the lowest-priced load
periods within the day at very short notice. As expected, there is a
considerable amount of heterogeneity across the pattern of within-day price
responses, as well as across the firms in this industry. Figure 3 plots the
sample mean own-price elasticities as a function of the load period for two
firms in the water supply industry. Although during the usual peak total
system load periods, 2:30 PM to 6:00 PM (load periods 20 to 26), we find
relatively small mean own-price elasticities for these periods, ranging from
.06 to .26 in absolute value. For the load periods immediately preceding and
following this time period, the mean own-price elasticities increase rapidly
to as large as 0.86, in absolute value. This implies, for example, a 1 percent
increase (decrease) in price during a pricing period may lead to as much as a
0.86 percent decrease (increase) in electricity consumed in that period. In
the next section we provide an example of load shifting in the water supply
industry in response to price changes.
Pricing in Competitive Electricity Markets 283

Mean Own Price Elasticities of Demand for UK Industries

Bic = 17000 : Water Supply Industry

Load Periods
~ Customer Id=2990121724
Customer Id=B4984021 as

Figure 3. Mean Own Price Elasticities for Water Supply Finns.

Mean Own Price Elasticities of Demand for UK Industries

Bic = 22460 : Copper, Brass, and Other Copper Alloys
Elastlc:ltJ••
0.00
-0.02
-0.04
-0.06
-0.08
-0.10
-0.12
-0.'4
-0.16
-0.18
-0.20
-0.22
-0.24
-0.26
-0.28

.
-0.30

12 ,.
-0.32 -.,..~~~.,.~~........,,...~~.,.~~~...,~........,~. . .~,....,,~~~,...,..~~~,.
1. 30 3& •2

Load Periods

.-.-+
e---e----e Custome,ld=2048374275
Customer td=8108634307

Figure 4. Mean Own Price Elasticities for Copper, Brass, and Other Copper Alloys
Manufacturing Firms.

Figure 4 plots the sample mean own-price elasticities as a function of the

load period for ftrms 10 the copper, brass, and other copper alloys
284 Pricing In Competitive Electricity Markets

manufacturing industry. Again, during the usual peak total system load
periods, beginning at 2:30 PM and ending at 6:00 PM, we find a uniform and
relatively small mean own-price elasticity. For the load periods immediately
preceding and immediately following this time period, the mean own-price
elasticity is over 0.2 in absolute value and gets as large in absolute value as
0.3 in load period 4, the period from 6:30 to 7:00 AM. Although these firms
are characterized by significantly smaller mean own-price elasticities
relative to the water industry, these elasticities still indicate substantial load
response to prices, particularly when considering variability in the PSP and
expected demand charge, as discussed above.
Figure 5 presents the analogous information for the seven firms in hand
tools and finished metals goods manufacturing industry. Figure 6 presents
this information for five customers in the steel tubes manufacturing industry.
Figure 7 represents the timber and wooden furniture manufacturing industry.
Finally, Figure 8 presents the own-price elasticities for the food, drink and
tobacco manufacturing. Figures 3 through 8 are indicative of the range and
variability of mean own-price elasticities of firms within and across various
industries in the E&W market. Patrick and Wolak (1999) present further
details and analyses for these industries, as well as for the remaining
industries in our data set. We next provide two examples of how this type of
demand information can be used in restructured electricity markets.

Mean Own Price Elasticities of Demond for UK Industries

Bic = 31600 : Hand Tools and Finished Metal Goods

Load Periods
Custom.r Id-20.48373890 - Cudom.r Id_2068S2.4085 e--e --e Customer Id=81 0863.4458
~ Custom.r Id=S1.40S0.4099 ~ ~ Custom., Id=81 -4-0504551 .............. Customer Id=81S8636297
-L----t-:Z. Customer Id: 1 0750737353

Figure 5. Mean Own Price Elasticities for Hand Tools and Finished Metal Goods
Manufacturers.
Pricing in Competitive Electricity Markets 285

Mean Own Price Elasticities of Demand for UK Industries

Bic =
22200 : Steel Tubes
Elasticities
0.000

-0.001

-0.002

-0.003

I
-0.00.

-0.005

-0006

I~
-0.007

-o.ooa .
-0.009

-0.010

-0.01 I

-0.012

-0.013

,.
-0.0''''
12 •• 30 42 4.
Load Periods
- - Cu.tomer Id=3538537.aO Cuslomer Id=8140S0",302 e---e-s CIJ.lomerld::!180226157
~ Customer Id=8220133821 Customer Id=8660090126

Figure 6. Mean Own Price Elasticities for Steel Tubes Manufacturing Firms.

Mean Own Price Elasticities of Demand for UK Industries

Bic = 46000 : Timber and Wooden Furniture Industries
Elostic!t1e.
0.000
-0.002
-0.00'"
-0.006
-0.008
-0.Q10
-0.012
-0.0''''
-0.016
-0.018
-0.020
-0.022
-0.024
-0.026
-0.028
-0.030
-0.032
-0.03.
-0.03&

-0.038 'r~~~T"'~~~,...,..~~~~~~--r~~~,..,~~~,......~~~~~~,.,
12
•• ,. 30 42 4•

Load Periods
I e----e---a- Customer Id=!U0510S4g

Figure 7. Mean Own Price Elasticities for Timber and Wooden Furniture Manufacturing
Firms.
286 Pricing In Competitive Electricity Markets

Mean Own Price Elasticities of Demand for UK Industries

Bic = 41000 : Food, Drink and Tobacco Manufacturing Industries
Elasticities
0.00000
-0.00002
-0.0000.
-0.00006
-0.00008
-0.00010
-0.00012
-0.0001.4
-0.00016
-0.00018
-0,00020
-0.00022
-O.OOOU
-0.00026
-0.00028
-0.000.30
-0.00032
-0.00034

12 1.
-0.00036 'r-r~~-"-~~'-'-~r-TO~~--r-r~~--r'-'-~-r--'~~-'--'-~~
30 36 48

Load Periods
Customer Id=5590285425 - - Customer Id=8090107725
e--e --e Customer Id=11.68308706 -tr--tr--tr Customer Id=11990537867

Figure 8. Mean Own Price Elasticities for Food, Drink, and Tobacco Manufacturing Firms.

5. USE OF MODEL RESULTS: TWO EXAMPLES

These demand system estimates enable the measurement of the effects of

alternative time-varying and consumption-dependent pricing structures on
customer-level electricity loads and the resulting effects on the electricity
supplier's revenue and customer's benefits. There are many examples we
could present illustrating various uses of the model but, for conciseness, we
restrict ourselves to two in this section, and discuss additional uses in the
conclusions of this chapter. See Patrick and Wolak (1997, 1999) for
additional examples, as well as detail and the procedures used in
constructing the following examples. We first present examples comparing
alternative vectors of energy price and demand (triad) charge changes. We
then present an example where we use this procedure to develop demand-
side bids derived from changes in the real-time pricing (PPC) customers'
demands as a result of price changes.
We first use the model to predict the demand response to changes in
various components of the expected prices-the sum of expected PSP and
the expected demand charge. Figure 9 considers two changes in the
expected PSP. The baseline scenario is the pattern of consumption for a
representative weekday evaluated at the sample mean of the observed
Pricing in Competitive Electricity Markets 287

expected prices. The first scenario is a 50 percent increase in all forty-eight

half-hourly expected PSPs holding the expected demand charges constant.
Consistent with the own-price elasticities, we find significant reductions in
demand relative to the baseline scenario in load periods early in the day and
later in day with only a small reduction in demand during the high priced
periods of the day. The second scenario decreases the expected PSPs in load
periods 30-34 by 50 percent. Significant increases in the electricity
consumption are predicted in these load periods, with very small reductions
in consumption predicted in the immediately adjacent periods.
The second two scenarios, which are given in Figure 10, consider the
impact of changes in components of the expected demand charge the pattern
of within-day electricity consumption. The first considers a 20 percent
decrease in the demand charge. The representative day selected for this
scenario did not have a triad priority alert in any of the load periods, so the
probability of a demand charge was uniformly small for all load periods in
the day. As a consequence, this reduction in the demand charge had no
discernable predicted impact on the pattern of electricity consumption. The
second scenario assumed that a triad priority alert was in fact issued for load
period 24, so using the estimated probability function given in Patrick and
Wolak (1997), the probability of a demand charge in period 24, went from
close to zero to approximately 0.12. As a result of issuing this triad priority
alert, there is a significant demand reduction predicted for load period 24.
There is also predicted to be a slight reduction in electricity demand in load
periods 31 to 36.
These examples illustrate some of the sorts of predicted price responses
that can be computed using these parameter estimates. Given this
information on price responses and the standard errors around these
responses, the electricity supplier can then estimate the effect of alternative
prices on customer load, customer benefits, and supplier revenues as well as
compute the associated uncertainty in these estimates.
We next provide an example of how our demand system estimates can be
used to formulate demand-side bids by electricity suppliers serving
customers on real-time prices (or the PPC). If suppliers purchasing from the
E&W market are able to accurately predict the response of demand to
within-day price changes for their customers on the PPC, this information
can be used to formulate a demand-side bid function for each supplier. xvii If
a supplier is able to entice more customers to face prices for electricity
which reflect the current PSP from the E&W market for that half-hour, given
accurate estimates of the price-responses of these customers, the supplier can
then formulate an aggregate demand-side bid function which has a relatively
larger price response.
288 Pricing In Competitive Electricity Markets

Price responses: BIC 17000

Water Supply
800

o 1
8 .-2 5
6
.-2 .-8
Load poriod

- - Demand at mean observ.d prices

---- Demand at price +50"
- - - _. Demond at price -50" for load period. 30-3.

Figure 9. Demand Response to Energy Price Changes.

Price responses: BIC 17000

Water Supply
800

700

-g
j 600

500

o 1
8 .-2 5
o
4
2
.-8
Load period

- - Demand at mean ob..rved prfce.

---- Demond at demand charve -20"
- - - _. Demand at triad prlortty alert for load period 24

Figure 10. Demand Response to Demand Charge Changes.

Pricing in Competitive Electricity Markets 289

Figure 11 illustrates the effects of demand-side bidding in the E&W SMP

determination process. TSL is the perfectly price inelastic forecasted
demand used along with generator bids to supply electricity to determine the
half-hour's SMP (labeled PSMP). Demand-side bidding effectively introduces
elasticity into the demand-side in the price determination process, leading to
the half-hour's SMP of PDB , which is lower than the SMP set by an inelastic
demand. Recall that this analysis assumes no change in bidding behavior by
generators in response to demand-side bidding, which seems unlikely given
that under demand-side bidding, generators face the likelihood not being
called on to generate as a result of bidding too high of a price into the pool
and should therefore bid more aggressively, resulting in an even lower SMP.
As emphasized by Wolak and Patrick (1997), the current operation of the
E&W market illustrates the sort of price volatility that can occur if the
demand setting the market price is very price inelastic and only a small
fraction of the total electricity consumed in any half-hour is sold to final
customers at prices that vary with the half-hourly PSP. Consequently,
accurate measurement of the within-day price response of customers is an
important necessary ingredient for any electricity retailer to aggressively
demand-side bid, and thereby build a significant price-responses into the
market price-setting process. Our demand estimates provide the necessary
information to effectively demand-side bid.

Price TSL
Demand Bid
Supply
Function

PDB -------------:)(

Quantity

Figure 11. The Effect of Demand-Side Bidding on the Market Clearing Spot Price.
290 Pricing In Competitive Electricity Markets

We now consider the impact of changing a single half-hourly price within

the day on the demand in that load period and all other load periods during
the day. We assume that the base period pattern of prices is the sample
mean of the vector of load period-level expected prices given in Table 1.
We assume that the supplier has ten customers from BIC 17000 (which is the
approximate number of water supply customers on the supplier's PPC in
1994-95). We assume that the price in load period 27, P 27d , increases from
its sample mean of 35 £IMWh to 100 £/MWh, a large but not unheard of
change in prices. The own-price response is a reduction in demand in load
period 27 of 41.13 kWh per customer, or a total of 411.3 KWh for ten
customers. Computing the demand change for each load period in the day
except load period 27 and multiplying by 10 yields the demand schedule
plotted in Figure 12, which indicates where portions of the 411.3 kWh that
are predicted to no longer be consumed in load period 27 are predicted to be
consumed during the other load periods during the day.

Price Responses: BIC 1 7000

Water Supply
120
110
100

.
'i" 90
~
.:::- 80
~ 70
~ 60
c
0
~ 50
E 40
~
0 30
u
20
10
0
0 1 1 2 3 3 4 4
2 8 4 o 6 2 8
Load pertod

- Consumption chang. (KWH)

Figure 12. Demand Response to a Price Increase in Load Period 27.

Proceeding in this manner for a variety of prospective prices, given the

mix of customers on the PPC, the supplier can determine the magnitudes of
price responses it can expect from various changes in the expected prices
aggregated over all of its customers. Coupled with information on the
standard errors of the these predicted price responses, the supplier can then
formulate demand-side bid functions which account for the aggregate
 Pricing in Competitive Electricity Markets 291

estimated price response of all of the supplier's PPC customers and the
uncertainty associated with these responses.

6. DIRECTIONS FOR FUTURE RESEARCH

The demand models discussed in this chapter are essential inputs into the
successful design of market-based strategies for enhancing customer and
utility benefits in increasingly competitive electricity markets. These models
develop customer-level demand relationships which are necessary in the
design of pricing options to attract and maintain customers in a market with
competing suppliers, while insuring that the costs of each customer's
consumption are covered. The financial viability of a wide variety of pricing
policies in a competitive environment can be assessed, which will ideally
account for the potential competitive responses of other suppliers, the
specific characteristics of the population of customers the supplier can
potentially serve, and the stock of generation, transmission, and distribution
capacities available to the supplier and its competitors. In addition, the
demand models' uses include forecasting customer-level electricity loads,
revenues from electricity supply, and customer benefits under alternative
rate structures, including various energy and demand charge combinations;
estimate the own-price and cross-price elasticities, which can be used in rate
simulation and optimization programs; and to develop demand-side bids.

NOTES

i As of August I, 1999, twenty-four states have restructured electricity industries to encourage

competition and provide consumer choice (Energy Information Administration).

ii See, for example, Thomas A. Stewart ("When Change is Total, Exciting and Scary"
Fortune, March 3: 169-170, 1997), which presents an overview of some of these issues and
the views of Duke Power's CEO, William H. Grigg.

iii The costs of serving a customer are heavily dependent on the customer's consumption path
(as is indicated by the pool price paths presented below). Since the customers' prior
consumption path may be unobservable (depending on metering capabilities and access to
information on the consumer), it is easy to envisage situations where the customer's
consumption could be subsidized by a pricing policy offered.

iv These are customers with at least 100 kW demands.

v Wolak and Patrick (1 996a, 1997) provide further on the mechanics of the price
determination process.
292 Pricing In Competitive Electricity Markets

vi To insure that "fixed" costs are not congregated in a few periods, thereby driving up the
relative prices in these periods, there is an upper bound on the number of Table B periods
each day. See Wolak and Patrick (1997) for these details as well as the determination of
Table A versus Table B half-hours.

vii Fiscal years in the E& W market run from April 1 through March 31 of the next year, e.g.,
the 1995 fiscal year runs from April 1, 1995 through March 31, 1996.

viii MEB was both a supply and distribution company over the time period represented by our
data. Supply and distribution companies have since been separated, see Patrick and Wolak
(1999) for details.

ill These three half-hours are subject to the constraint that they are each separated by at least
ten days. Triad charges can only be known in March of each fiscal year, after actual
electricity consumption has occurred.

x Patrick and Wolak (1997) analyzed five of these industries: water supply (BIC 17(00),
copper, brass, and other copper alloys manufacturing (22460), ceramics goods
manufacturing (24890), hand tools and finished metal goods (31600), and steel tubes
manufacturing (22200).

xi Patrick and Wolak (1997) provide more detail on the fax, including a sample fax.

xii Patrick and Wolak (1997) report all triad advance warnings, priority alerts, and actual triad
periods for our sample.

xiii See Patrick and Wolak (1997, 1999) for a description of the specific charges.

xiv The E& W CFD market has been dominated by generating firms. CFDs generally cover the
PPP but not UPLIFT, triad, or other charges.

xv Patrick and Wolak (1997) estimate the expected demand charge.

xvi 3-dimensional plots of the sample mean of the cross-price elasticities, indicating the type of
within-day demand substitution patterns that exist for some PPC customers, can be found
in Wolak and Patrick (1997,1999).

xvii Inaccurate demand-side bidding would lead to a higher half-hourly UPLIFT component of
the PPC. UPLIFT, the component of the PSP that is determined after demand is realized, is
generally not covered by the CFDs contracts used to hedge wholesale electricity price
volatility.

REFERENCES

Government Statistical Service (1991-96). Business Monitor: MM22 Producer Price

Indices, Central Statistical Office, HMSO Publications Centre, London, U.K.
Pricing in Competitive Electricity Markets 293

Patrick, Robert H., and Frank A. Wolak (1997a). "Estimating the Customer-Level
Demand for Electricity Under Real-Time Market Pricing." Mimeo. (Available from
http://www.rci.rutgers.edul-rpatrick/hp.html)

Patrick, Robert H., and Frank A. Wolak (1997b). Customer Load Response to Spot Prices
in England: Implications for Retail Service Design. TR-109143, EPRI, Palo Alto, CA.

Patrick, Robert H., and Frank A. Wolak. (1999a) "Consumer Response to Real-Time
Prices in the England and Wales Electricity Market: Implications for Demand-Side Bidding
and Pricing Options Design Under Competition," in M.A. Crew edited, Regulation under
Increasing Competition, Kluwer Academic Publishers.

Patrick, Robert H., and Frank A. Wolak (1999b). Using Customer Demands Under Spot
Market Prices for Service Design and Analysis. Report prepared for the EPRl, Palo Alto, CA.

Phelps, A.K. (1994). "A Study of Real Time Pricing in the U.K: The Midlands
Electricity Experience." Midlands Electricity pic, Halesowen, U.K. Mimeo.

Wolak, Frank A. and Robert H. Patrick (1996) "Industry Structure and Regulation in the
England and Wales Electricity Market," in M.A. Crew, edited, Pricing and Regulatory
Innovations Under Increasing Competition, Kluwer Academic Publishers: Boston, MA.

Wolak, Frank A., and Robert H. Patrick (1997). "The Impact of Market Rules and
Market Structure on the Price Determination Process in the England and Wales Electricity
Market," Mimeo. (Available from http://www.stanford.edul-wolak.)
Chapter 17

How To Buy Low And Sell High*

Spot Priced Electricity Offers Financial Rewards

Michael T. O'Sheasy
Georgia Power Company

Key words: CBL; CfD; Decrements; Derivatives; Increments; Marginal Cost; Risk; Two-
Part; Volatility; Win-Win.

Abstract: Under Georgia's Power's innovative, two-part marginal priced tariff,

customers have demonstrated an uncanny ability to do just this, resulting in a
classic win-win for the customer and the utility.

The dream of every participant in a capitalistic economy is to buy low and sell
high. Those that do so reap significant rewards. But for the majority of us,
this simple goal remains illusive.

Historically the electric utility industry like most other businesses has not
offered customers this availability of buying low and selling high - until
recently. In June of 1990, Georgia Power Company quietly requested and was
granted approval of an innovative, two-part marginal priced tariff called Real
Time Pricing (RTP). Since its inception as a pilot with twenty-five test
customers and twenty-five control customers (designed to measure price
response characteristics), it has exploded into two permanent tariffs with over
1000 customers and over 4,000 mW's of load. As the benefits of RTP became
known to its customers, GPC hustled to add an hour-ahead RTP to the original
day-ahead RTP, and lower the mW tariff requirements in order to share the
tariff virtues with as many customers as possible. No other utility in the world
promotes RTP and shares it benefits with its customers as does GPc. In fact if
you were to add up RTP usage at all other utilities in the U.S., the sum would
hardly equal that of GPC' salone.

* This article previously appeared in the January/February 1998, Volume II, Number I issue
(pp. 24) of the Electricity Journal, and is reprinted with the permission of the publisher,
Elsevier Science, Inc.

A. Faruqui et al. (eds.), Pricing in Competitive Electricity Markets

© Kluwer Academic Publishers 2000
296 Pricing In Competitive Electricity Markets

1. WHY DO GPC'S CUSTOMERS READILY

ESPOUSE RTP?

Simply put, it allows customers the availability to "buy low and sell high."
And, in doing so, customers have been able to lower their energy cost
significantly. Some of the more successful customers can be seen in
Table 1.

Table 1. RTP Customers

Customer Actual kWh RTP Bill % Change ¢IkWh Standard

Increases RTP

Customer A 82,918,193 $42,549,917 52.73% 3.36 3.08

Customer B l3,646,933 $549,876 143.68% 6.81 4.03

CustomerC 187,432,829 $5,110,064 52.54% 3.20 2.73

CustomerD 219,610,839 $6,599,075 27.21% 3.29 3.00

CustomerE 80,030,792 $2,315,912 30.00% 3.04 2.89

CustomerF 25,399,756 $889,971 176.10% 4.94 3.50

CustomerG 15,255,949 $510,079 44.43% 3.67 3.34

Customer H 4,941,939 $263,655 49.64% 6.66 5.34

2. HOW DOES RTP WORK?

The two part design begins with a customer baseline load (CBL) which
reflects historical consumption. Since historical consumption (i.e.,
electrical load requirements) has already caused cost to be incurred by the
utility, standard embedded cost rates are charged for this unchanging CBL
load shape. Any actual load changes from this historical load shape
however are priced at the utility'S marginal cost which is normally
considerably lower than its embedded cost (although certain hours can
contain marginal cost twenty-five times that of embedded). This becomes
the second part of the tariff.
Notice that "changes" are billed at hourly marginal prices. This means
that both increments and decrements are priced at RTP. Table 1 reveals
Pricing in Competitive Electricity Markets 297

how customers who are increasing consumption above their CBL are able
to lower their overall cents per kWh. However, customers who can only
decrease (decrement) consumption during high prices can also benefit.
(Decrements are reductions in hourly load below that of the historical
CBL.) If decrements are priced at RTP, this translates into the utility
crediting the customer for reductions in hourly usages below that of
historical usage levels (CBL). Figure 1 (next page) presents a shrewd
energy purchaser making incremental purchases whenever energy is priced
below his marginal value of electricity and refusing energy consumption
thereby receiving decremental credits whenever the price of electricity is
above his marginal value of electricity. This decremental crediting is an
extremely powerful feature of the two-part design and should not be
overlooked; this encourages the significant price response attributes of
RTP, and helps customers make significant dents in lowering their cents
per kWh. It also insures that the utility is selling RTP at its marginal cost;
otherwise, the utility is risking selling additional load below cost or
crediting decrements at prices above marginal cost thereby eating into the
utility's fixed cost contribution requirement. Table 2 reveals customers
who have demonstrated a canny ability to lower their overall cents per kWh
primarily by reducing consumption during high prices.

Table 2. Bill Reduction Impacts Due to Price Response

Customer Effective centslkWh credit for Reduction in Total Bill and in

decrements CentslkWh
AA (4.3) (5%)

BB (8.5) (9%)

CC (10.3) (8%)

DD (9.4) (14%)

EE (4.3) (5%)

FF (8.8) 11 %)

GG (4.5) (6%)
 298 Pricing In Competitive Electricity Markets

Figure 1.

Customer Demand Profile

07/22, Thursday

\
16000

-.-
/
14000
c
z 12000
'.
'"
::0
w
10000 ............_........... '. ~
(/)

c 8000
I-
Z
~ W

'" 6000 .................-t- .... U

,
4000

2000

1 2 3 4 5 6 7 6 9 10 11 12131415 16 17 18 192021 222324

ActualkW CBLkW Cents/kWh

The two-part nature of RTP is analogous to a public exchange whereby a

spot priced product is complimented with financial derivatives. Imagine
that you normally purchase around 15,000 bushels of com over the spot
market on a daily basis. Imagine also that you have purchased a financial
derivative i which guarantees a fixed price (say a Contract for Difference
(CtD) or SWAP) for a fixed quantity of com. Now, suppose your daily
purchases rise and fall below your financial hedge depending upon daily
spot prices for com. Your transactions might look like Table 3.
 Table 3. Commodity Purchasing Using A Combination of Derivatives and Spot Purchases ...,
Fi'
'""
Time Spot Financial Guarantee Daily Spot Market Financial Total Bill Average Bill for Bill for Total
S·
~

Period Price Quantity Bill Derivative Price/Bsh. for quantity of quantity of S·

Purchased on Bill quantity of corn above corn above g
Spot Market corn purchased that under the that under the
~
(1l
along with the financial financial
financial derivative derivative
...::::-.
~.
(1l

Quantity Price derivative

~
(1l
'"l
(I) (1) (K) (L)
(A) (B) (e) (D) (E) (F) (G) (H) ......,
Fi'
~.
(B) x(E) I(D)-(B)I(e) (F)+(G) I(B)(e)+(G)1 (I) x (e) (E)-(e)J(B) (J)+(K)=(H)
I!C) ~
$27,500 $13,400 (1l
I $2.68 10.000 $2.75 15.000 $40.200 $700 $40,900 $2.75 $40.900
~
*
2 $2.74 10,000 $2.75 5,000 $13,700 $100 $13,800 $2.75 $27,500 ($13,700) $13,800

3 $2.78 10,000 $2.75 10,000 $27,800 ($300) $27,500 $2.75 $27,500 ----- $27,500

4 $2.65 10,000 $2.75 20,000 $53,000 $1,000 $54,000 $2.75 $27,500 $26,500 $54,000

5 $2.70 10,000 $2.75 25,000 $67,500 $500 $68,000 $2.75 $27,500 $40,500 $68,000

Total ---- 50,000 ---- 75.000 $202,200 $2,000 $204.200 ---- $137,500 $66,700 $204,200

Average $2.71 10,000 $2.75 15,000 $40,400 $400 $40,840 $2.75 $27,500 $13,340 $40,840

tv
\0
\0
300 Pricing In Competitive Electricity Markets

In essence, Table 3 demonstrates how a purchaser of com could use a

combination of spot purchases and financial derivatives to manage his
overall cost of com. The quantity purchased under the financial derivative
is not subject to price volatility since its price is guaranteed at $2.75 per
bushel, and is therefore less risky (even though the average spot price of
$2.71 per bushel was less in this example than the guaranteed price of
$2.75, the spot price could have been higher than the guaranteed price).
The quantity purchased above the financial derivative amount is still
subject to spot price volatility.
The total bill for the financial derivative and spot purchases can be
observed in column (H). One can also examine the total bill from another
perspective as revealed in column (L). This column (L) perspective views
the total bill as a combination of the net bill for the quantity purchased under
the financial derivatives [column (B»); and the bill for the quantity purchased
outside that of the financial derivative [column (K»). Basically column (H)
views the total bill from the perspective of a price difference in the
derivative quantity while column (L) views the total bill from the perspective
of the quantity difference of total purchases versus derivatives purchases.
Either perspective arrives at the same answer.

Now lets examine two-part RTP rate design. Assume the following changes
to Table 3:

l. Column (B) is the RTP price per 100 kWh's.

2. Column (C) is the customers CBL in hundreds of kWh's.
3. Column (D) is the standard bill price under embedded rates per 100
kWh's.
4. Column (E) is actual load shape usage in 100' s of kWh's.

Then, the following conclusions can be observed:

l. Column (J) is the first part of the two-part RTP design, and is referred
to as the Standard Bill.
2. Column (K) is the second part of the two-part RTP design, and is
referred to as the incremental RTP component.
3. Column (L) is the total RTP bill sent to the customer.

Therefore, the structure of the two-part RTP design is very similar to the
common means of purchasing commodities off of public exchanges with the
following two deviations:
Pricing in Competitive Electricity Markets 301

a) The CBL represents historical consumption before going on RTP as

opposed to a customer - stylized financial derivative (in which the
customer selects the quantity of product that he does not want
exposed to spot prices for that specific time period).
b) The RTP incremental price is the utilities marginal cost plus a risk
adder as opposed to a market clearing price off of an exchange.

As can be easily deduced, RTP is a surrogate for and the precursor to

what electricity pricing will evolve to once it becomes traded on a public
exchange.
One of the major reasons that this linkage is so strong is the relationship
of embedded cost pricing to marginal cost pricing. In general, a utility's
marginal costs are much more volatile than its embedded cost. Embedded
cost include all fixed cost and can be thought of as the fixed cost
contribution which allows the customer to purchase an unlimited quantity at
the stated price for a given load shape. Embedded cost must be of a
magnitude sufficient to cover marginal cost over the period in question;
otherwise the utility will be selling at an incremental loss.
Marginal cost however, beside being more volatile, reflect the change in
cost for an incremental change in usage for a specific time period. By
definition they exclude fixed cost. Here too the customer can purchase a
relatively unlimited quantity of product but at prices that vary per time
period (i.e., spot prices) and reflect the incremental cost to produce the
product.
Embedded cost pricing is relatively stable over time periods, and may not
reflect changing marginal cost over the time period (See Figure 2 below).
As a result the seller using embedded cost pricing has considerably more risk
than seller using marginal cost (spot) pricing. One of these major additional
risk is load shape risk: the possibility that the customer will change his usage
from that anticipated by the seller in developing his embedded rates. It
could indeed be such that the overall marginal cost of producing the product
is greater than the embedded price agreed to (or greater than the marginal
cost assumed for the flexible customer's load shape such that the utility does
not receive the necessary fixed cost contribution and therefore cannot payoff
its fixed cost obligations).
This load shape risk of embedded pricing does not occur with RTP (spot)
prIcmg. If the RTP customer changes his usage habits, he pays the
corresponding marginal cost of producing the product. The RTP customer
likewise bears the risk that the utility's marginal cost of producing electricity
changes unexpectedly, and, if it rises, must be able to shift consumption or
may have to pay a high price. This is not true for the embedded priced
302 Pricing In Competitive Electricity Markets

customer whose price disregards unexpected swings in the actual cost of

producing electricity.

Figure 2.

Unit Cost by Hour Versus

Unit Price by Hour

cen ts
kWh

Tim e

So, in essence, RTP (spot) pricing shifts the burden of load shape risk and
marginal cost risk to the customer while embedded pricing (fixed pricing)
places the risk on the utility/seller. [Since the RTP price is the forecast of
the utility's marginal cost of producing the product (plus a small mark-up),
the utility is practically indifferent as to whether the customer shifts or
reduces consumption due to the compensating effect on the utility's cost for
changed consumption.] Obviously between these two extremes are many
alternatives with their own unique risk sharing such as time of use (TOU)
pricing and seasonal pricing. This concept is demonstrated in Figure 3
(below).
Pricing in Competitive Electricity Markets 303

Figure 3. Rate Complexity and Load Shape Risk

Price
Required
by seller

Guaranteed CEO "UD TOU RTP

Bill Guaranteed
Price

RATE COMPLEXITY
Note:CED stands for customer, energy, and demand rates. HUn stands for hours use of
demand rates.

¢/kWh
Sale
Price

Load Shape Risk To Customer (Rate Complexity)

Economics dictates that risk has a cost with it, and to the degree that risk
can be shifted or lowered, prices (costs) can be lowered. Therefore RTP
should in general offer lower prices than embedded pricing, and, in fact, it
normally is lower. (There are however some hours as previously mentioned
when RTP can skyrocket passed embedded pricing.)
This RTP price volatility with a commensurate price crediting when
usage falls below a CBL allows customers to choose to buy when prices are
304 Pricing In Competitive Electricity Markets

low, and sell back or refuse usage below the CBL when prices are high (or
higher than the value that electricity would provide in that time period).
Figure 4 (below) pictures this extraordinary price response capability of RTP
customers. In fact the volatility of RTP (and electricity costs) is remarkable.
On some days, RTP prices can explode from relatively mild to prices twenty
five times that of early hours of that very day (I don't know of many, if any,
products which experience this volatility).

Figure 4.

RTP TOTALS
07/20/93, Tuesday
1000000 ,---------------------------------------,

c 800000 :I:
z ;:
c(

~ 600000 ~
(f)

c I-
Z
~ 400000 W
()

200000
........
o LL-L~~~-L_L~L_~_L_L~l_~_L_L~l_~~

1 2 3 4 5 6 7 B 9 10'112131415161718192021222324

Actual kW CBl kW Cents/kWh

An analogy to RTP price volatility might be the local gasoline station.

Imagine if it experienced volatile spot pricing such that today at 5:00 P.M.
when you stopped in for a fill-up, the attendant said the price for the next
three hours was $25 per gallon instead of the $1 per gallon that you were
used to paying. The attendant also indicated that prices should return to
normal later that night. Well, more than likely, you would bid the attendant
a fond adieu until later that evening. But, what if the attendant yelled out to
you as you were leaving that he would buy any gasoline from you over the
Pricing in Competitive Electricity Markets 305

next three hours, and pay you $25 per gallon? You might think to yourself,
"I am only down to 1,4 tank and I only need a gallon to get home and back
this evening; so I could sell three gallons back to this idiot at $25 per gallon
and replace it later tonight at $1 per gallon." "It's a deal!" This is similar to
what two-part RTP pricing does. You can choose to buy electricity when the
price is low and sell back (i.e., give up your right to buy your CBL quantity
at stable embedded rates) when the RTP price is high.

3. WHAT IS THE NEXT INNOVATION IN RTP?

As stated two-part RTP is a shadow image of a public exchange with

market clearing prices coupled with financial derivatives. The CBL
functions like financial derivatives in that it reduces price risk. However,
over time, a customer's mix of price stability (CBL) and price risk
(incremental RTP) can get out of sync with his prevailing degree of risk
preference. For example, may be his load exposed to incremental RTP has
grown or reduced dramatically, or may be a one-time order is being sought
but input cost must be guaranteed since the order requires a fixed price
requirement and is very large. As the utility industry and customers
experience with RTP grows, the next evolution will indeed be price
derivatives (GPC calls them Price Protection Products (PPP». With PPP,
the customer can reformat his mix of price stability to price risk, and can do
so for specific time intervals.
Some of the commons PPP's are CfD's (SWAP's) which guarantee a
specific price, Price Caps which place a ceiling on price, and Price Collars
which place a ceiling and a floor on price. A cousin to these products is
index pricing in which the RTP price is swapped for another commodity's
publicly traded price. All of these products normally have stated quantities
and time intervals. For example, an RTP customer with a CBL of 15 mW
and incremental RTP purchases of 10 mW might purchase a CfD for 2 mW
from 12 noon to 9:00 P.M. Monday through Friday during June through
September at a guaranteed price of 6¢/kWh. This form of price risk
management can be referred to as price hedging (as opposed to speculation,
whereby there is not a commensurate physical product purchased to the
degree of the financial instrument).
In summary, two-part RTP affords the customer the luxury of buying
above the CBL when the price is low and selling back effectively below the
CBL when the price is too high. RTP customers have demonstrated an
uncanny ability to do just this resulting in remarkable reductions in their
cents/kWh. Since these changes are performed at a price reflecting the
utilities marginal cost, the utility benefits likewise. Imagine a classic win-
306 Pricing In Competitive Electricity Markets

win whereby the seller and the buyer are in perfect accord to buy low and
sell high.

NOTES

i A financial derivative is a financial contract with no physical product delivery associated

with it. The mechanics and philosophy of financial derivatives are left to another
discussion. For this exercise, assume the derivative takes the difference in the spot a price
and the derivative contract price and debits or credits the customer times the contract
quantity.
Chapter 18

Real Time Pricing - A Unified Rate Design Approach

Steven V. Huso
Northern States Power Company

Key words: Access Demand; Allocation; Caseload; Capacity; CBL; Class Responsibility;
Competition; Demand Charge; Economic Efficiency; Energy Management;
Financial Instruments; Forward Contract; Hourly Capacity; Interruptible; Load
Characteristics; Load Management; Load-Following; LOLH; Marginal Cost;
Market-Based; Non-Linear; Off-Peak; One-Part; On-Peak; Peaking; Power
Supply; Price Averaging; Price Incentives; Price Response; Prices; Purchased
Power; Rate Design; Rate Options; Rationing; Real-Time Pricing; Resource
Use; Revenue Erosion; Risk; RTP; Seasonal; Time-of-Day; TOD; Two-Part;
Unbundling; Value.

Abstract: Pricing electricity by hour using market-based capacity values provides fair
and efficient rationing of available power supplies. This approach works by
directly connecting prices with a wide range of market values. Precise load
management incentives. which result from conveying extreme market
valuations when capacity reserves are critically low, also reduces the need for
peaking generation. Real-time pricing can deliver these benefits by resolving
limitations of more traditional rate design approaches, which include price
averaging and monthly demand charges based on customer load patterns.

Developing a real-time pricing design requires many considerations, which

include: 1) the benefit of using market-based capacity values for all hours, 2)
the benefit of a balanced approach that recognizes the advantages of different
RTP design objectives and approaches, 3) the potential for using RTP detail to
develop a consistent set of rate options, 4) options to provide for different
customer risk preferences, and 5) the need for further RTP refinements.
Further RTP design refinements will accelerate recognition of the important
role such pricing can provide for minimizing power supply costs and
improving customer satisfaction.

A. Faruqui et al. (eds.), Pricing in Competitive Electricity Markets

© Kluwer Academic Publishers 2000
308 Pricing In Competitive Electricity Markets

1. INTRODUCTION

Real-Time Pricing (RTP) has tremendous potential for improving

economic efficiency and customer equity. Although RTP has become
commonplace, its potential for dramatically reducing power supply costs by
more efficient resource use is still significantly under-appreciated. There is
also further potential to provide the detailed building blocks to develop a
consistent set of less complex rate options that are more economically
efficient. RTP's potential to induce efficient use of resources comes from
the ability to closely connect hourly prices with market based values. In
conventional rates, this connection is loosely made by using maximum loads
of individual customers to approximate system generation costs and values.
By contrast, RTP matches prices with system or market conditions, to allow
efficient rationing of power supply resources. RTP serves this purpose
during the transition period to a competitive market. After that transition,
RTP concepts and price structures will continue to be a valuable method for
representing competitive market values and prices.
Although now a common rate option, RTP is typically divided into two
very different rate design structures. These designs are usually referred to as
"One-Part RTP" and "Two-Part RT." Each approach has unique advantages
and disadvantages. A promising new approach is an RTP rate that combines
the best features of both one-part and two-part approaches with new rate
design components. This new unified approach is designed to improve the
net benefits of RTP, and to stimulate further RTP design refinements by
encouraging a less polarized view of RTP rate design issues.
This chapter begins with an overview of the inherent limitations of
conventional rates that restrict economic efficiency. This background helps
define objectives for effective RTP rate design, and the challenges associated
with achieving these objectives. A review of one-part and two-part RTP
designs follows, with associated advantages and disadvantages. This review
is intended to provide helpful perspectives for evaluating a unified RTP rate
design approach that is described. The chapter concludes with a review of
promising rate design refinements and concepts.

2. CONVENTIONAL RATE LIMITATIONS

Many advantages of RTP rates are obvious, such as precise tracking of

prices with changing capacity supply and demand conditions. Some RTP
advantages, however, are best appreciated by closely reviewing and
considering how these advantages address certain limitations inherent in
Pricing in Competitive Electricity Markets 309

conventional rates. A design process that considers these limitations will

help produce a balanced and effective RTP rate.
Conventional rates, by their nature, are subject to a number of limitations.
For example, customer risk preferences for price or quantity variations are
generally recognized only through interruptible rates. This and most other
limitations are a product of the following two situations:
1. Price averaging does not convey the full range of costs or market values,
and
2. Prices are not closely connected to daily and hourly capacity supply
variations.

2.1 Price Averaging

Price averaging is a pervasive and obvious rate design limitation.

Generally, when prices are averaged, half of all energy sales represent an
above average profit margin, and half represent a below average profit
margin. Aside from customer equity and pricing efficiency concerns,
customers who are over-charged provide an attractive target for competing
energy suppliers.
Time-of-day (TOD) pricing reduces the level of price averaging. The
level of improvement is modest, and TOD rates are relatively crude in
relation to RTP rates. Even TOD prices are often too high or too low for any
specific hour. Further, on-peak periods may be long enough to be used for
all seasons. More refined TOD rates address a significant part of the price
averaging issue by using a "critical-peak" price that can be imposed for a
limited annual number of hours. The benefit of the critical peak approach is
very similar to the targeted price signals and incentives provided by
interruptible service rates.
Additional price averaging is included in seasonal rates. For example,
four-month summer seasons are common, even though most of the critical
high-cost days occur in the middle two months. The other two months are
used to include the few critical days that may occur during those periods,
and to cover the range of days often used for each billing month.
Demand charges also suffer from price averaging. Generation,
transmission, and distribution costs are often bundled into one averaged
price. As discussed in the following section, these averaged demand prices
are then applied to billing units that do not necessarily measure the relative
capacity requirements that an individual customer imposes on a power
supplier.
310 Pricing In Competitive Electricity Markets

2.2 Price and Variable Capacity Values

Conventional firm service rates, by virtually any definition, have an

exceptionally limited ability to accurately signal variable capacity values.
This shortcoming substantially restricts the economic efficiency of such
rates, which by nature include few price variations. Capacity market values,
by contrast, cover a wide price range (i.e., from perhaps as little as zero to
over $1.00 per kWh), and are highly variable by hour and by day. By
contrast, marginal energy costs are typically in the proximity of one to ten
cents per kWh.
The ability to convey variable market values is an essential requirement
of an economically efficient pricing structure, such as RTP. The broad range
and high variability of both total customer demand and available capacity
supplies largely explains wide variations in capacity market values. The
greatest potential for economic gains is associated with critical supply
conditions. In such conditions, high capacity valuations provide a targeted
incentive for customer load reductions, which in tum reduces the demand for
relatively expensive generation capacity from peaking plants.
Determining an appropriate distribution of capacity values is challenging,
especially before the establishment of a mature competitive market. Another
challenge is wholesale trading approaches that are still moving from the
marginal energy cost focus of traditional power pools to an approach that
includes hourly capacity market valuations. Still, determinations of hourly
capacity values are becoming increasing possible. Actual market prices,
financial instrument prices, and experiences with customer price responses
can provide guides for estimating capacity market values.
The limited potential for including capacity market values in
conventional firm service rates creates additional pricing inefficiencies. For
example, it requires using customer class load characteristics to develop
prices, which further dulls the connection between prices and resource costs
and values. Hourly capacity costs are the same for all customers, regardless
of class. Pricing that is based on hourly capacity values eliminates the need
to use class load characteristics for capacity pricing. The use of such pricing
during a transition to competitive markets would minimize potential
disagreements over class revenue responsibilities and simplify rate design.
Although including these hourly values is a more direct process with RTP,
the use of RTP detail and concepts has the potential to improve the process
for developing conventional rates.
A final consideration on the issue of connecting prices with capacity
values is the need for separate pricing of distribution capacity. The
unbundling of generation and distribution capacity, while often required for
Pricing in Competitive Electricity Markets 311

regulatory purposes, also provides a closer match between prices and

resource costs.

3. RTP OBJECTIVES

Real Time Pricing has long been considered the ultimate method to
accomplish desirable objectives for electricity pricing. These objectives
address the limitations of conventional rates and focus on the principal goal
of equitable and efficient price signals that provide optimal use of resources.
Specific objectives and the role of RTP rates in meeting those objectives are
listed below:

1. Reduce Price Averaging - The hourly price component of RTP rates

greatly reduces price averaging. RTP prices are determined on a day-
ahead or same-day basis, which allows hourly prices to follow marginal
energy cost differences. The improved accuracy from closely matching
variable costs with prices significantly increases customer equity.
2. More Efficient Use of Fixed Power Supply Costs - Hourly RTP prices
that reflect wide variations in market-based capacity values will directly
signal capacity supply variations. High hourly prices serve to ration
capacity resources by allowing customer price responses tailored to the
economic thresholds of individual customers. As hourly prices increase,
a greater number of customer thresholds are passed. This effectively
reduces total demand for capacity, and re-directs some of that reduction
to customers that attach the highest value to the capacity. Conversely, a
greater than average supply of capacity is signaled to customers through
hourly prices that reflect lower capacity market values. This encourages
higher load levels than would result from conventional rates that include
demand charges applied to maximum monthly demands, which occur
during times when system loads are relatively low. The benefits of this
objective include a close alignment of prices with variable market-based
capacity values.
3. Customer Choice of Risk Levels - RTP rates can be developed to
provide prices that vary according to customer specified price risk
preferences. The general premise here is that a lower average price is
the benefit that is balanced by assuming the risk of being subject to
hourly prices. Another example of using RTP for risk management is a
rate design that prices a customer determined load level (similar to base
load) at a fixed take-or-pay amount, while the remaining load (similar to
load-following) is applied to an RTP rate. Although RTP rates provide
more flexibility and options for risk management, it is important to
312 Pricing In Competitive Electricity Markets

recognize that other rate designs can provide risk-based price

differences. For example, interruptible rates provide price discounts for
certain quantity restrictions.
4. Detail for Consistent Rate Options - The detailed information used to
determine RTP hourly prices can be accumulated to improve consistency
between all rate options. Although RTP rates are not yet practical for
many customers, the detail from RTP rates can be used to develop other
rate options that have less complex load management incentives. In
particular, this approach can be useful for providing detailed information
that can be aggregated to develop more accurate TOD and interruptible
rates. A consistent set of rate options allows customer rate selection on
appropriate economic merits and incentives.
5. Experience for Competitive Markets - RTP rates require the type of
complex pricing and administrative structures that are characteristic of
competitive markets. This provides insights into rate options that will
provide a smoother transition to open competition.
6. Institutional Price Discounts - The appropriateness of this objective
can be argued. However, to some extent, utilities have used RTP as an
administrative and politically acceptable method to establish price
discounts. In a regulated environment, these discounts may be targeted
to price sensitive customers or used to lower average costs by reducing
excess capacity. This is not a new rate design concept - interruptible
rates often include above market discounts in concession to the influence
of large and price sensitive customers. As the transition to competition
moves forward, the relevance of this objective will continuously
diminished.

4. RTP CHALLENGES

Achieving these objectives is challenging, due to the complex nature of

RTP rate design and implementation. The challenges include:

1. Revenue Erosion - This is generally associated with one-part RTP in a

regulated environment. Voluntary RTP rates produce revenue
reductions in the same way as voluntary TOD rates. Customers with
lower cost load profiles can reduce their bills with no price response.
This provides more equity, and still provides substantial price incentives
to respond to hourly prices. In contrast, two-part RTP can minimize
revenue erosion and customer price risk, but at the expense of diluted
price incentives. The challenge is to reconcile these different RTP
approaches, or provide an RTP design that allows customers to select
Pricing in Competitive Electricity Markets 313

any combination of one-part and two-part RTP features, which are

priced to recognize inherent differences in risk and value.
2. Updating "Customer Base Load" (CBL) Profiles - A CBL defines the
hourly loads of an individual customer for the year before commencing
RTP service. It is an essential component of two-part RTP rates. A less
detailed CBL, such as monthly peaks is also used for RTP rates that
combine one-part and two-part RTP designs. This feature minimizes
revenue erosion and customer price risk, and helps retain consistency
with conventional rates by preserving differences in average prices
produced by different historical load factors. The challenge is
eliminating or reducing the burdensome historical hourly detail required
for a CBL, and resolving the problem of a CBL becoming increasingly
stale over time. Fortunately, a unified RTP design approach has
significant potential for replacing the CBL requirement. This approach
divides customer loads into base-load and load-following components
during any current RTP billing year, and does not use an historical base
year.
3. Customer Selection of Risk - Risk levels are affected by a number of
RTP design components. The challenge is determining which standard
or optional components to offer for customer choice. Design elements
that affect risk include the relative level of fixed and variable
components, the range of possible price levels, the presence of true-up
provisions, and variability of components used to determine hourly
prices. Risk management options that apply to various rate components
can include all types of financial instruments; with price caps a simple
example. The price and packaging of these financial instruments to
provide various types of insurance will increase the appeal of RTP rates.
4. Conventional Demand Charges Revisions - Converting conventional
monthly demand charges into hourly energy charges allows a greater
representation of variations in capacity market values. Further, a
separate distribution demand charge applied to individual customer
peaks loads will provide more accurate pricing of distribution costs.
This also allows for converting a higher percentage of remaining
demand charges into hourly energy charges.
5. Marginal Energy Cost Definition - A number of marginal energy cost
definitions are possible. Determining the appropriate definition will
provide better price signals and will allow for more effective scheduling
of generating and power purchase resources.
6. Interruptible Service Option - RTP rates, in some respects, serve as a
voluntary interruptible service rate. An RTP rate that includes an
interruptible rate option recognizes the greater value of mandatory
interruptible service. Furthermore, traditional interruptible service
314 Pricing In Competitive Electricity Markets

customers are usually more capable of responding to hourly price

signals.
7. Short Notice Price Changes - The ability to occasionally update hourly
prices, whether normally provided on a day-ahead or same day basis, is
useful for signaling significant and unanticipated changes in power
supply or customer loads.
8. Automation - RTP service requires sophisticated administrative
processes for price communication, bill compansons, and final bill
calculations.
9. Related Services - Considerable customer value enhancement is
possible by being able to provide services that complement RTP rates,
such as energy management systems that can optimize the benefit of
hourly prices. These services can be profitable on their own and can
serve to increase customer demand for RTP rates.

5. ONE-PART AND TWO-PART RTP RATE DESIGN

The following summary of one-part and two-part RTP rate design

approaches is intended to provide both an historical perspective as well as
reference points for discussing RTP rate refinements.
This summary is not a comprehensive review of one-part and two-part
RTP rate design approaches, which are products of extensive research and
development. Rather, the basics are presented as background for discussing
their relative advantages and disadvantages in meeting RTP objectives. As
might be anticipated, the advantages of each approach are often the
disadvantages of the other approach.

5.1 One-Part RTP

One-part RTP rates are characterized by hourly prices that include all or a
substantial level of fixed costs, in addition to marginal energy costs.
Typically, these fixed costs are included in an adder or multiplier, which is
applied to marginal energy costs for each hour. Some one-part rates include
conventional demand charges to recover a portion of fixed costs and reduce
customer bill variations. In some rates, additional adders or multipliers are
applied to a limited number of hours as a load management incentive to
signal the exceptionally high market value of capacity during low capacity
reserve periods.
Pricing in Competitive Electricity Markets 315

5.2 Two-Part RTP

Two-part RTP rates generally have a fixed "take-or-pay" component. This

fixed component is commonly referred to as an access charge. The access
charge is customer specific and is essentially an annual bill determined by
applying the customer's historical hourly load characteristics (the CBL) to
the standard (non-RTP) rate, and has been characterized as a forward
contract. The second rate component is an hourly price that closely reflects
forecasted marginal costs. The variable price is applied to hourly differences
from the CBL, to determine hourly credits or charges.

5.3 Unified RTP Rate Design

In 1997, Northern States Power Company (NSP) established RTP rates in

Minnesota and Wisconsin. These rates include an Access Demand Charge, a
Distribution Demand Charge, and an hourly RTP price. The rates are similar
in structure; the principal difference is the determination of billing kW for
the Access Demand Charge. The Minnesota RTP rate uses historical
monthly on-peak demands. The Wisconsin RTP rate uses historical annual
average demand for the monthly access demand, which leads to a higher
average level of hourly prices. Both rates include one-part and two-part RTP
design components in addition to their other design components.
In the Minnesota rate, the Access Demand kW for individual customers
is determined from historical on-peak period demands. In Wisconsin, a
historical annual average demand is subtracted from the same historical on-
peak period demands as defined in the Minnesota rate. The Wisconsin
Access Demand provision leads to a higher average level of hourly RTP
prices, although that effect is partially offset by other provisions.
Hourly RTP prices are applied to all energy use, and are determined on a
day-ahead basis. The hourly prices are based on two components. The first
component is hourly marginal energy cost component. The second
component designed to represent relative capacity values. The capacity
value component determines a large share of the variation in hourly RTP
prices and is the most critical and unique component of NSP's RTP rate
design.
316 Pricing In Competitive Electricity Markets

Table 1. One-Part RTP

Advantages Disadvantages

• Hourly RTP prices apply to all • Potentially large variations in

energy use by customers, for more customer bills and power supplier
effective price signals. revenues.

• The need to use an historical • Not revenue neutral for individual

reference year is avoided or customers, limiting its appeal to
minimized. customers whose existing load
characteristics produce rate savings.
• All customers with the same load
and service characteristics are • Hourly prices can be inappropriately
charged the same amount. distorted by adders or multipliers.

Table 2. Two-Part RTP

Advantages Disadvantages

• RTP rates are revenue neutral for • Hourly RTP prices apply only to
individual customers. deviations from a base year, reducing
effective of price signals.
• Variations in customer bills and
power supplier revenues are • Historical reference year loads are
limited. required and need updating.

• Hourly prices are closer to • Rate is disproportionately appealing to

marginal energy costs. customers that increase load levels over
the base year.

• Customers with identical load and

service characteristics may have
different bills if their historical CBL's
are different.

• Limited ability to signal hourly

variations in capacity market values.
Pricing in Competitive Electricity Markets 317

6. HOURLY CAPACITY PRICES

The capacity component is determined with a non-linear function that

provides a different capacity price for each hourly system load level. This
method was designed to approximate the capacity-related component of
competitive market prices. System loads are used for practical reasons;
these values are more available, consistent, and predictable. Ideally, an a
more precise measure would be used, such as hourly reserve margins for the
NSP system or regional conditions would more closely approximate market
pnces.
The important goal is establishing hourly RTP prices that more closely
represent competitive market dynamics. Although one-part and two-part
RTP rates may include a capacity component, it is often used for relatively
few hours such as critical supply periods, and is based on marginal capacity
costs, rather than a closer estimate of market capacity values.
Capacity values are increasingly reflected in market prices, but still
appear to be under-represented. This is encouraging, since the importance of
such values has been frequently unrecognized and under-appreciated. In
large part, this results from a mistaken assumption that marginal energy
costs are the only component that should be considered for optimal prices.
Application of economic theory is not that straightforward for electric
pricing.
Marginal energy costs have been inappropriately viewed as the only
variable cost, and therefore the only element to consider for determining the
marginal cost that should equal price for economic efficiency. Electricity
has product characteristics, however, which make it difficult to apply to
economic theory. In my view, efficient electricity pricing requires marginal
costs that address the dual components of energy and capacity. Therefore,
closely approximating and recognizing market-based capacity values
(estimated as the total market price less marginal energy costs), is a critical
step in determining hourly RTP prices that are effective and economically
efficient.
Many determinations of marginal capacity costs attempt to reflect market
values. Unfortunately, methods for determining marginal capacity costs are
not as refined those for marginal energy costs, and are sometimes misguided.
Problems with marginal capacity costs used in some RTP rates include the
use of only a very limited number of hours, or use of probabilistic methods
such as Loss-of-Load-Hours, which are unlikely to approximate market
values over a wide range of load or reserve margin conditions.
NSP originally used a non-linear allocation method (Lauriol) to associate
a separate capacity value to every system load level. This followed the
theory that some capacity value is present in all hours, although such values
318 Pricing In Competitive Electricity Markets

are minimal for relatively lower system load levels. At the same time, the
significantly non-linear aspect of the method provides strong price signals at
the highest load levels, with maximum hourly prices in the range of 30 cents
per kWh. This price is less than recent peak market price levels, to mitigate
customer risk exposure until risk management options are developed.
More recently, NSP represented this capacity allocation with a curve-
fitting equation. Minor adjustments to the equation are applied during the
annual process of updating the hourly pricing algorithm. These adjustments
are based on comparisons that include the on-peak to off-peak price ratio, the
seasonal price ratio, minimum and maximum prices, and the effect on
customer bills. This process maintains continuity with past RTP hourly
prices and accommodates new marginal energy cost forecasts. The trend of
marginal energy costs indicates that increasing levels of capacity values are
being included in hourly market prices.
The inclusion of market-based capacity values in marginal energy costs,
which is limited to purchased power, blurs the distinction between energy
costs and capacity values. Fortunately, the effect of this blurring on year-to-
year price consistency can be well managed. It also indicates an
encouraging trend of market prices being a more useful source of capacity
market valuations. The reasonableness and need for such valuations will
become more clear as competitive markets evolve and mature. As that
happens, significant improvements in the accuracy of capacity value
estimates can also be expected.

7. ACCESS DEMAND CHARGE

The Access Demand Charge combines components of both one-part and

two-part RTP rates. The following is a further discussion of the Wisconsin
version of NSP's Access Demand Charge. That charge divides historical
monthly kW into base load and peaking (load following) components. The
base load component is determined by average demand, which represents a
100 percent load factor. Since that component is consistent for all
customers, it can be recovered through common hourly RTP prices. The
peaking component is applied to a demand charge that is higher than the
demand charge in the standard rate, in part to mitigate the effect of
transferring the base load component to hourly energy prices.
This Access Demand Charge provides continuity with standard rates by
retaining the load factor effect on the average rate for individual customers.
This allows the one-part RTP feature of applying hourly prices to all energy
use, which maximizes the load management incentives provided by hourly
prices, but with much of the rate stability of two-part rates. The Access
Pricing in Competitive Electricity Markets 319

Demand Charge reduces the significance of historical reference years

relative to two-part RTP designs, and provides a compromise on the revenue
neutrality issue.

8. RTP DEVELOPMENT ISSUES

RTP rate design is continually evolving, and NSP's rate was designed to
contribute to that process. Continued development of the competitive power
supply market will provide additional perspectives that will both advance the
refinement of RTP rates and eventually clarify the future role of RTP rates.
Although RTP rates can be cast in the role of providing a transition to a fully
competitive market, that does not diminish the role of RTP concepts in
packaging market prices. A fully developed competitive market will provide
better determinations of pricing and associated financial instruments, which
are now only approximated by RTP rates.
At least until that time, there are opportunities for significant RTP
refinements. An important goal is replacing dependence on historical load
data. Refinements that may address that goal are further forward contracts,
separate risk management options, and restructuring of demand charges to
reflect time-differentiated capacity reserves and customer-specific ancillary
service costs.
Another goal is providing RTP prices to all energy use by a customer, but
in combination with other rate provisions that balance needs for revenue
stability, continuity with standard rates, and an equitable representation of
the cost or value of service. The future distinction between cost and value
will be an interesting evolution, especially in the development of hourly
pricing for capacity or reserve services.

9. NEW RTP RATE DESIGN CONCEPTS

A central difference between one-part and two-part RTP rates is whether
all energy is subject to hourly prices. Directly related is the pricing method
for energy not subject to hourly prices. Two-part RTP is generally
considered to not apply hourly pricing to all energy; though, in some
respects all energy use can be affected by hourly prices. For example,
during a high priced hour, two-part RTP customers have an incentive to use
as much as possible below their historical load level. However, since these
customers do not have a disincentive for reducing their load at that time,
one-part RTP customers have a greater stake in responding to high hourly
prIces.
320 Pricing In Competitive Electricity Markets

An RTP rate that allows customers to determine how much of their total
energy use is billed at RTP hourly prices directly addresses the central
difference between one-part and two-part RTP rates. It also provides the
tremendous benefit of eliminating the need for billing determinations that
depend on historical loads. This concept is similar to interruptible service
designs that allow customers to determine a firm service level that can be
used during interruption periods. This type of RTP design incorporates a
number of specific refinements, although the development process requires
resolving some difficult issues.
A promising approach for developing this new RTP design involves
billing all energy associated with load levels at or below the customer
specified load level at a fixed price per kWh. This "base load" energy could
also have a take-or-pay structure. The fixed price could be based on the
average price per kWh that results from a comparable conventional rate at an
assumed 100 percent load factor, and may be adjusted for its relative risk
differential and other factors. The "load-following" energy, determined
from loads above the customer specified level, would then be applied to
hourly prices defined by the RTP rate.
This RTP rate design would require including most or all capacity costs
in hourly prices, with the exception of distribution costs billed through a
separate distribution demand charge. Hourly prices would span a wide range
of values to provide total billings on an average price per kWh basis that are
consistent, when appropriate, with conventional rates. This can be
accomplished by developing a distribution of hourly capacity prices, using
automated solving techniques with appropriate constraints. Also required
are rate provisions that ensure appropriate cost recovery from customers that
have extremely seasonal or very low load factors. The benefits of these RTP
rate design concepts include:

1. Greater Customer Flexibility - The ability to better control energy

use subject to hourly pricing, which increases customers ability to
manage risk and makes RTP a practical option for more customers.
2. No Historical Base Year Requirement - Separating energy use into
a pre-determined base-load components and load-following
component determined by current period use, serves to replace many
of the key functions that are provided by historical load references.
3. Hourly Prices Representing Capacity Values - The distribution of
capacity related costs to all hours, under the constraint that the
average price from the RTP rate - including base-load and load-
following energy - is consistent with conventional rates across all
customer load factors. This will allow a close representation of
capacity values.
Pricing in Competitive Electricity Markets 321

4. Equitable Billing of Low Load Factor Customers - Prices closely

correspond to the precise relationship between individual customer
peak demands and system or market supply. Some current RTP
design approaches do not function well with low load factor
customers. Yet, many of these customers have above average load
flexibility, because they use less energy relative to their peak loads.

10. SUMMARY

This review was intended to provide background information and

perspectives to evaluate alternative RTP rate design approaches. Specific
issues and potential refinements were described to stimulate development of
more effective rate designs needed to fully realize the great potential of RTP
rates.
Chapter 19

Dynamic Pricing and Profit Maximization Choices for

the Investor-Owned Electric Disco

George R. Pleat*
Baltimore Gas and Electric Company

Key words: Disco Pricing Strategies; Earnings Stability; Lowering Operating Costs;
Revenue Retention; Value of Service.

Abstract: The stand-alone Investor-Owned Electric Distribution Company (Disco) can

pursue a number of pricing strategies that will insure long-run profitability and
ward off potential competitors. Marginal and avoided costing techniques are
still effective tools for retaining and growing current regulated services.
Measuring the marginal cost of substation and feeder facilities enables the
planner to flag to the management how efficient these resources are utilized
and to price signal customers accordingly. Also by applying a two-part pricing
structure to recover marginal infrastructure costs, the planner can help to fund
dire operating and maintenance activities, and infuse the Disco with steady
revenues. The Disco can compete effectively in price elastic service markets
by applying short-run avoided cost pricing techniques. To stabilize and
protect profitable services, the planner can design permanent and moving fixed
monthly prices. By increasing fixed monthly prices for small use, secondary
service customers, the Disco earnings become more weather normalized and
long-run billing cost transactions with generation suppliers get reduced. The
planner can apply a twelve-month moving fixed pricing strategy to insure an
annual revenue stream from the "pay as you go" large use wires shopper. To
minimize long run wires operating expenses, the planner can encourage
increases in customer power factor by applying a generic KV A wires demand
charge. The KY A charge will help lower long run wires capacity costs and
line losses and enables the Disco to get out of the one-on-one customer power
factor enforcement activity. By offering price discounts to its customers in
exchange for interruptible wire's service, the planner can delay cash draining
reinforcement projects in periods during extended price freezes - assuming

* The views and opinions expressed by the author are not necessarily the views and opinions
of the Baltimore Gas and Electric Company.

A. Faruqui et al. (eds.), Pricing in Competitive Electricity Markets

© Kluwer Academic Publishers 2000
324 Pricing In Competitive Electricity Markets

customer tolerance levels are not exceeded. Finally, the planner should begin
to focus on what value-added service does the Disco provide to the customer.
By identifying these value-added services, the Disco may be able to sustain
price over cost.

1. INTRODUCTION

With the emergence of competition, investor-owned vertically integrated

utilities are positioning themselves by restructuring vertically into
generation, transmission, and distribution operating entities. The Investor-
Owned Electric Distribution Company (Disco) is further segmenting its
business lines between its wires, billing, and metering services. With the
wire's business likely to remain a regulated monopoly service in the near
future, the Disco is at a crossroads in determining the correct strategic,
regulatory, and economic path for its billing and metering servicesi .
The executive (or planner) of a Disco will be reporting annual operating
incomes to his or her stockholders. In addition, the executive will be most
likely selling distribution wires service under some kind of performance-
based pricing regulation and may be considering moving the billing and
metering services to a de-regulated arena. Because of these new paradigms,
the Disco executives will have every incentive to design prices for their
product line that will enhance the operation's long-run profitability.
The Disco can choose various pricing strategies to enhance long-run
profits. To accomplish these goals, the Disco planner needs to: (1) design
prices for the Disco's regulated services on marginal costs to promote
economic efficiency and to deflect long run competition, (2) charge a
permanent monthly fixed wires price for the small use secondary service
customer to stabilize revenues and lower billing costs, (3) initiate a twelve-
month moving fixed wires price for large use high voltage service customer
to maximize revenues, (4) price large customers on a unit KVA basis to
improve power factor on the wires system, (5) offer price discounts to
customers in exchange for interruptible wires service, and (6) apply value-
of-service pricing techniques to sustain prices above marginal cost.
Pricing in Competitive Electricity Markets 325

2. MARGINAL COSTING FOR REGULATED

DISCO SERVICES

With the establishment of a stand-alone Disco from a traditional

vertically integrated electric system, the Disco planner has an opportunity to
design regulated prices that will promote economic efficiency, thereby
positioning the Disco to successfully deal with long-run competition. The
planner must begin to look at the Disco's business in terms of long-run
competitors and to price accordingly. Surviving as a regulated monopoly
means evaluating the cost structure of its services on a marginal cost basis.
There are essentially four regulated businesses the Disco will inherit from
a vertically integrated electric system: Substation and Feeder Wires,
Infrastructure Wires, Billing and Metering. The planner should begin to
assume that all of these businesses could potentially go competitive in the
long run. He or she should price these currently regulated services on long-
run and/or short-run marginal costs (depending on the threat of competition)
instead of book accounting costs.

2.1 Substations and Feeder Wires

The Disco will increase its chances of retaining substations and feeder
assets as a regulated monopoly service for sometime to come if it applies
marginal cost techniques today in developing prices. This means estimating
marginal costs of the substation and feeder delivery facilities that bring the
bulk electric power (generally 69KV, 34KV, and 13 KV service) from the
transmission network to the various neighborhoods and business centers.
Typically these facilities are sized on area peak MV A loads. In practice, the
planner should evaluate the long-term additions of these facilities to the
corresponding area peak loads. Moreover these supply and demand
comparisons should be performed within voltage level categories to
recognize service cost differentials. In some geographical areas, where there
is an over abundance of capacity, the marginal cost will tend to be low which
presumes a lower price to customers. In an area where shortages exist the
marginal cost will be high, presuming a higher price for customers.
The application of marginal costing method to substation and feeder
assets has several advantages. First a cost measurement is in place to
monitor how well the Disco planner is matching supply with demand. A
very low marginal cost is a flag to management that there maybe over built
facilities while in contrast a very high marginal cost would indicate that
capacity expansion, either through reinforcements or new additions, is not
adequate to keep up with ever increasing electricity demand.
326 Pricing In Competitive Electricity Markets

Moreover, if the Disco has usage charges to recover substation and feeder
facility costs, the planner has the flexibility to signal to the customer to use
more when capacity is under utilized and to use less when capacity is over
utilized. ii The positive impact for the Disco is an improved load factor and
adhering to a least cost supply curve for the future. Marginal costing by its
very nature requires comparing supply with demand. Pricing based on
marginal cost makes it difficult for third-party wires suppliers to enter the
substation and feeder market - since these competitors will be forced to enter
the market at below marginal cost to entice customers who are otherwise
stringently loyal to the Disco.

2.2 Infrastructure Wires

Just like any successful business in a competitive market, the Disco

planner should aim to recover the full marginal cost of infrastructure
expansion - even in an assumed regulated monopoly market. With a
growing customer connection market (given a continued strong economy),
the planner will be faced with spending too few annual capital dollars to
connect customers to the electrical system. This connection infrastructure
consists of primary (13 KV) and secondary (including transformers)
overhead and underground conductors located in neighborhoods and
business centers. The Disco should price at the full cost (including the
incremental cost of capital) of connecting an additional customer - which is
equivalent to the marginal cost. This allows the Disco to be fully
compensated for debt and equity costs along with providing a sufficient cash
flow into the operation. Pricing on some kind of average embedded cost will
draw dangerously down the cash reserves needed for on-going operations.
Since these infrastructure costs are generally fixed and not variant with
electric use, fixed prices are the more appropriate mechanism to recover
these marginal costs. However, who pays for the marginal extension costs
and how should the fixed pricing be applied is a key strategic issue. Should
the planner charge the builder for the full extension cost up front through a
one-time hook-up charge or should the planner recover the extension cost
from the wires customer through a monthly fixed tariff (or rental charge)?
The benefit of the hook up charge to the Disco is the influx of more
immediate cash to help fund operating and maintenance activities. The
downside is that the revenue collection through this process is entirely
dependent on a growing customer connection market. Also, significant bill
impacts to the builders might create political consequences in the regulatory
process where the Disco is seeking support for other pricing proposals from
the builder associations.
Pricing in Competitive Electricity Markets 327

The alternative is to recover the marginal extension costs through

amortized monthly fixed tariffs (a rental fee) over the life of the asset. This
enables the Disco to derive a steady stream of annual revenue. The
downside to this alternative approach is that the Disco will be visiting
frequently the Public Service Commission to get these assets into the rate
base so an income return can be earned. With the era of electric retail choice
upon us, that option is probably not a desirable path for the Disco.
A good strategic compromise in recovering extension costs is to get the
best of both payment mechanisms. The Disco planner would recover up
front from the builder a one time hook-up charge equal to the net present
value of marginal extension and marginal substation and feeder costs that
exceed the expected tariff wires' revenues from that residential or
commercial development. The remaining costs would be included in the rate
base process. The Disco has cash coming into the door faster through a hook
up charge to the builder but also has the security of consistent annual tariff
revenues to protect against years when the customer connection market is
stagnant.

2.3 Billing and Metering

If the Disco goals are to retain billing and metering services under a
regulated monopoly environment, it must demonstrate to regulators and
potential competitors that a better deal cannot be made elsewhere by the
customer when acquiring these products. This demonstration can be
supported, among other factors, with the Disco designing billing and
metering prices on avoided cost. By applying avoided cost pricing to these
services, the Disco will discourage third party entrants while illustrating to
the regulators that customers benefit with the implicit economies of scale
associated with a 100 percent regulated monopoly market.
The Disco planner should evaluate the multitude of mini businesses that
are included in the billing and metering functions iii and begin to unbundle its
billing and metering prices by applying the inverse elasticity rule.
Consequently, for those service markets characterized with small barriers to
entry, prices should be set at short-run avoided costs; in contrast those
markets characterized with large barriers to entry, prices should be set above
avoided cost to compensate the more elastic markets that do not contribute to
corporate fixed costs.
For example, the bill rendering market is a highly price elastic service
because of the relatively ease entry into the market place. Many credit card
type companies could easily compete with the Disco for this market. The
service is labor intensive and does not require a large capital investment.
The bill rendering service involves printing the billing information on an
328 Pricing In Competitive Electricity Markets

invoice, sending the bill to the customer, collecting the revenue, and making
payment to the appropriate vendors. This is a prized market because of the
small start up costs and intangible asset associated with direct customer
contact.
To compete effectively with the credit card companies, the Disco planner
should price at its short run avoided cost to render a bill. The cost savings
realized to the Disco for rendering one less bill is small (maybe $.50)
because it has 100 percent of the current market (not withstanding California
and some other States) and enjoys the economies of scale factor of serving
virtually hundreds of thousands if not millions of customers. The planner
may want to pursue a strategy that excludes any corporate overhead
expenses. By pricing at short-run avoided cost, the Disco reduces the appeal
for third party entry into this highly price elastic service while sending the
customer a signal that they are getting a good deal. iv
In contrast, if the Disco has more inelastic billing markets, prices can rise
to recover overhead costs not included in more elastic services. These
inelastic services would include more of the capital-intensive services like
the state of the art billing computer infrastructure and the metering
installation business where safety and reliability are important features. The
Disco planner should price at long-run avoided cost to prepare the data for
bill invoicing (computer bill equipment) and price at long-run avoided cost
to install the next meter with appropriate operation and maintenance.
However, the planner should include in the price some contribution to fixed
costs that support the more elastic markets. A price above avoided cost for
these markets can be maintained by the Disco for the immediate future
through high technical computer billing barriers to entry, safety meter
installation expertise requirements, and "cash register" meter reliability.

3. DESIGNING PERMANENT AND MOVING

FIXED PRICES

3.1 Permanent Fixed Price for Small Secondary Service

Customer

The Disco planner can pursue various pricing strategies to protect future
revenues and to minimize billing expenses. One strategy for the planner is
to move toward higher fixed prices for the small secondary service customer
who does not have demand metering.
Pricing in Competitive Electricity Markets 329

For the typical Disco fonned out of a vertically integrated bundled price
system, the secondary service wires' costs have been historically recovered
together with transmission and generation costs via usage charges. In some
instances, even some of the metering and billing costs have been recovered
in usage prices to minimize rate shock to the small use customer. By
moving to recover more wires, billing, and metering service costs through a
fixed monthly fee, the Disco nonnalizes the annual revenue stream by
insulating the business from abnonnal weather conditionsv • With nonnalized
revenues, the task for estimating accurately the future Disco earnings
become less difficult.
Moreover, if the Disco moves to full recovery of the service costs
through some gradual process, the future billing process with third party
suppliers can be simplified. For example, suppose the wires part of the
Disco separates from the metering and billing functions altogether, (i.e., the
billing and metering functions are put "below the line.") Now the wires
business must deal with third party generation suppliers (assuming retail
choice has arrived) for collecting its wires revenue transactions. Verification
becomes an issue if all or some of the wires' costs are collected through
usage charges. Who has the correct read of the supply registered at the
customer's meter, the generation supplier or the Disco? A fixed charge
bypasses that issue altogether. The Disco wires planner bills the generation
supplier for delivery service revenue based on the number of customers at a
particular voltage level assuming one generic fixed monthly price per
customer.
There are some downsides to applying fixed prices to recover the wires,
billing, and metering costs. First, by moving away from usage prices to
fixed charges, small use customers will be impacted significantly with price
shock, since this niche has enjoyed relatively low fixed charges through the
traditional regulatory process. One procedure to minimize the price shock is
to move gradually toward higher fixed charges over several years. Also to
minimize bill increases, the planner might consider designing two or three
unique fixed charges for the same voltage service depending on the
infrastructure requirements of the end-user. Customers living in an
apartment complex will cause less secondary service line cost per dwelling
than a single-unit five-acre lot development. Consequently, lower fixed
charges can apply to apartment dwellers while in contrast higher fixed
charges can be applied to sprawling single-family residences.
Another downside to the fixed monthly price application is the revenue
erosion that might take place if the local economy sours and new house
construction comes to a stand still. Without new connected customers, the
Disco will feel the revenue pinch until the new house construction cycle
• vi
returns to expanSIOn .
330 Pricing In Competitive Electricity Markets

3.2 Moving Fixed Price for Larger Voltage Service

Customers

In the near future the wires part of the Disco will be threatened by
customer on-site generation installation because of the increased efficiency
rates and lower up front costs of these units. Typically, demand metered,
larger use, higher voltage customers will be experimenting with this power
source application, especially when co-generation is a feasible option.
Subsequently, much of the traditional wires business could be by-passed,
with the customer purchasing sporadic backup service from the Disco on a
"pay as you go" basisvii. Most assuredly, the customer will only want to pay
that month's distribution KVA price for that month's load required. The
Disco could get short changed by constructing a standby substation and
feeder sized to the customer's anytime peak demand while only getting
partial rental payments during the year.
One way for the Disco planner to combat the revenue erosion caused by
the "pay as you go shopper" is to apply a twelve-month moving demand
ratchet. The benefit of this price design is that it resembles a moving fixed
charge tailored toward the actual capacity requirement of the customer. For
each month, the planner would charge the customer for wires service based
on their highest peak demand during the last twelve actual months. For each
new month, the oldest month's demand would be dropped for measurement
and the new month's peak demand added.
For the occasional backup service customer, if they only require one-
month wire's service, they would pay the same rate for each of the next
eleven months, withstanding any bypass. Also the Disco enjoys the same
benefit of earnings protection from a fixed price concept, although not to the
degree of the permanent fixed charged discussed above. The only downside
to the Disco is the extra metering and billing cost associated with evaluating
the highest twelve-month peak demand and charging accordingly. This cost
expense may be considered minuscule compared to the potential revenue
loss from on-site wires bypass.

4. GENERIC KVA PRICING TO IMPROVE POWER

FACTORviii
To encourage customers to improve their power factors, the Disco can
design a generic monthly KV A wires charge. A KV A wires charge is
consistent with how area distribution planners typically size substation and
feeder capacity requirements. These capacity requirements are based on
Pricing in Competitive Electricity Markets 331

local MV A peak demand levels. By applying a KV A wires charge to all

connected customers, there are incentives for customers with low power
factors (or high KV A) to reduce demands through the purchase of more
efficient electric motors. Those customers with equal loads but with high
power factors will already have reduced bills because of the associated low
KVA.
By applying a generic power factor charge to all customers regardless of
power factor and KV A use, the Disco will be relieved of the responsibilities
of informing individual customers that they have to be switched from kW
billing to KV A billing.
In addition to the enforcement benefit, the Disco will realize long term
capacitor, line loss, and wires capacity savings if customers with poor power
factors react to the price signal and reduce their KV A consumption.
Distribution planners contend that as customer power factor declines below
.90, the line loss and capacity costs increase exponentiallyix.

5. INTERRUPTIBLE SERVICE PRICING

To encourage improved load factor, the Disco planner might offer area
interruptible wires service to its customers in exchange for discounted
distribution prices. The benefit to the Disco planner is two fold. First, there
is a benefit in delaying substation and feeder reinforcements in local areas
where demand spike shortages and/or transformer overloads exist. This
becomes more critical in time periods of regulated price caps. Unless the
builder (see discussion above) heavily funds new infrastructure construction,
cash will be tight for the Disco. There will be increased pressure to divert
monies toward operation and maintenance programs that support important
service reliability standards. Interruptible distribution wires service offers
time for the Disco to raise needed capital. Second, the planner has the radio
control flexibility to smooth out spikes by interrupting localized peak MV A
demands, thus reducing transformer feeder overloads and increasing the
quality of consistent reliable wires service.
There are several downsides, however, for offering interruptible wires
service to customers. First, the Disco will have to invest in the specific area
load control equipment devices, which could prove to be more expensive
than the traditional system wide generation supply control devices installed
on customer's air conditioners and hot water heaters. Also, interruptions
will have to be constrained to a tolerance level of the customer, which may
inhibit the delay of reinforcement capability. How many times will a
customer tolerate a specific outage during the temperature sensitive season?
The magnitude of the price discount given the customer for the right to
332 Pricing In Competitive Electricity Markets

interrupt might be the deciding factor in that tolerance range. In addition,

how will the Disco interruption be coordinated with the generation supply
arrangement between the retail aggregator and the customer? Will an
uncoordinated effort cause verification problem at the meter in terms on how
much electricity was supplied and what was actually consumed?

6. VALUE-OF-SERVICE PRICING

Because the Disco will be reporting stand-alone profits to its stockholders

in the future, the entity will be searching for pricing schemes to increase
profitability beyond the normal regulated return on investment. Under either
a regulated or de-regulated environment, the Disco should begin to think of
pricing services on customer value instead of purely cost plus. Cost plus
pricing is only effective in the long-run if the Disco is guaranteed a 100
percent monopoly regulated franchise for many years to come. This is a
poor assumption to make in the fast changing electric industry where new
central power facilities are going the way of the dinosaur and mini-
generators are the flavor of the day. Value-of-service pricing enables the
Disco to differentiate its product from other competitors, thereby giving it
some control in setting price above marginal cost.
For example, what are the important values to a "mom and pop" ice
cream parlor shop that purchases electrical service from a centralized
generation supplier moving through the Disco wires. Certainly, the shape,
speed, or origin of the kilowatt of electricity has very little importance to this
type of customer or any other customer. Instead, that ice cream parlor
customer will value electric service in how it adds value to its own retail
product. The ice cream parlor shop will value infrequent distribution
interruptions during very hot days in the summer time where the ice cream
market demand is at its greatest. Also, this type of customer will value a
quick response for service restoration when a wires outage is experienced -
time is melted ice cream. This involves the Disco having in place an
efficient call center response team capable of evaluating the problem and
dispatching the appropriate repair team to get the ice cream parlor back on
line - with priority over other outage problems in other areas.
With the increased perception of quality of service combined with the
dire need for reliable electrical service, the ice cream parlor business will be
more amenable to paying a premium price well beyond the marginal cost of
service. Also, by quality service bundling (fewer outages and better call
restoration call service), the Disco creates a disincentive for the ice cream
parlor customer to by pass the wires system altogether by installing on-site
mini -generation.
Pricing in Competitive Electricity Markets 333

REFERENCES
i For a further discussion on the decision tree facing the Disco concerning retaining the billing
and metering services under a regulated and deregulated, see George R. Pleat, "Should
Metering Stay at the Stand-Alone Disco?," Public Utilities Fortnightly (copyright),
February 1, 1998, p.44.

ii The variability of costs with use is more characteristic in this particular business. Sudden
shocks in end-use can alter the planners' reinforcement activities around substations and
feeders. Not withstanding some significant financial downsides of a variable wires charge
(discussed below), usage charges seem to be a more appropriate pricing mechanism for
recovering substation and feeder costs.

iii Billing functions include: operations, credit and collections, revenue processing, call center,
and customer relations. The metering functions include installation, operation and
maintenance, meter reading, and field services.

iv Meter reading, meter field services and credit collection activities are other elastic services
that the Disco must apply short run marginal cost pricing.

v A large segment of the electricity market is temperature sensItIve (air conditioning and
resistance heat) making usage revenue collection highly volatile from season to season.

vi For discussion on the advantages and disadvantages of fixed pricing to the Disco, see
George R. Pleat, "Pricing and Profit Strategies for a Stand-Alone Electric Distribution
Company," Public Utilities Fortnightly (copyright), p. 23, January 15, 1997.

vii The customer is likely to request backup service from the Disco when the on-site generators
are down for periodic maintenance.

viii George R. Pleat, "Unbundling Retail Prices For a Future Investor Owned Electric Disco,"
Public Utilities Fortnightly (copyright), May 15, 1999.

ix The economic benefits of generic KV A billing will be tempered, however, because the
Disco must installed meters that have KV A reading capabilities.
Chapter 20

Developing and Pricing Distribution Services

Laurence D. Kirsch and Robert J. Camfield

Laurits R. Christensen Associates, Inc.

Key words: Connection Services; Customer Services; Distribution; Incremental Costs;

Pricing; Wires Services.

Abstract: Distribution services as we know them will undergo significant change in

response to electric power industry restructuring. This change will arise from
a variety of competitive pressures: from merchant firms who want to offer
customer services as a means of gaining proprietary rights to customer
information; from engineering contractors who can profitably offer customers
new options for connecting to the power system and controlling power quality;
and from distributed resources, which can allow customers to avoid use of
distribution wires. In response to these pressures, new distribution services
will be offered, and both new and old services will be re-priced. Most
importantly, competition will force firms to tailor and differentiate their
services according to the needs of various customer segments.

1. INTRODUCTION
The unbundling of distribution tariffs in the U.S. is being driven primarily
by two related developments. The first is that the restructuring of the U.S.
electric power industry will ultimately require open access distribution tariffs
that provide both consumers and power suppliers with non-discriminatory
access to power transportation systems. The second development is that
technological changes, such as the development of "distributed resources"
(generators and curtailable loads) that can be located within distribution
systems, are creating direct competition for the services traditionally
provided by distribution utilities. Efficient distribution tariffs must

A. Faruqui et al. (eds.), Pricing in Competitive Electricity Markets

© Kluwer Academic Publishers 2000
336 Pricing In Competitive Electricity Markets

anticipate both the requirements of open access and the imminent

competition.
The purpose of this chapter is to describe how unbundled distribution
services may be packaged and priced. The first section defines the three
categories of distribution services: wires services; connection services; and
customer services. The next three sections describe the cost, demand, and
competitive considerations that will determine the packaging and pricing of
each of these three service categories. The fifth section examines some of
the planning, operations, and service quality issues that are raised by the
unbundling of distribution services. The final section presents conclusions.

2. DISTRIBUTION SERVICE CATEGORIES

Distribution services can be divided into three general categories: wires

services, connection services, and customer services.
Distribution wires services involve investment in and maintenance of
equipment (such as poles, lines, and substations) that are needed for low-
voltage or radial transportation of power through the distribution network
and radial systems. Wires services may be distinguished according to the
locations and voltages at which customers receive service, and according to
the quality of service.
Distribution connection services involve investment in and maintenance
of equipment (such as substations, drop lines, transformers, capacitors,
reactors, inductors, filters, and meters) that serve individual customers or
small groups of customers. "Customers" can be either consumers or
generators that are connected to the distribution system. Connection services
may be distinguished by the particular facilities that serve each customer,
which will depend upon the customer's location, voltage delivery level, size,
and power consumption or production characteristics. While wire services
have a standard level of quality that applies to all customers or to large
groups of customers, connection services can be tailored to meet the
particular service quality needs of individual customers or small groups of
customers.
Distribution customer services generally include services like billing and
customer communications that have costs that depend upon the number of
customers. Customer services generally involve communication of
information between customers and service providers.
Unbundling will require that utility activities and costs be divided among
these services. In a physical sense, this is difficult because it can be
problematic to identify the particular services provided by particular
facilities. For example, it can be difficult to determine whether a particular
Pricing in Competitive Electricity Markets 337

capacitor bank provides a "distribution wires service" by helping control

voltages for the whole distribution system or instead provides a "connection
service" because it primarily controls voltages for nearby consumers.
The difficulties of physical unbundling are accompanied by
commensurate difficulties of financial unbundling. Financial costs, as
maintained within the format of the U.S. Uniform System of Accounts, often
serve as the basis for establishing distribution utilities' overall revenues. It
is these costs that need to be unbundled among service categories and,
perhaps, among cost types such as operations and maintenance expense and
capital expense. Because some facilities arguably provide multiple services
simultaneously, and because certain costs (like corporate overheads)
arguably support multiple services simultaneously, the unbundling of costs
among services can be somewhat arbitrary. A variety of methods, such as
activity-based costing, are available for making the necessary cost
allocations.
The unbundling of financial costs is simplified somewhat, however, by
the emergence of competition. The market prices of competitive services
will tend to approximate the market's incremental costs of these services.
Because the revenue recovered from competitive services must reflect
market prices, the financial costs that are allocated to these services should
also reflect expected market prices. The remaining financial costs can then
be allocated among non-competitive services according to procedures that
satisfy regulatory policy.

3. DEVELOPING AND PRICING WIRES SERVICES

Under regulation, a distribution wires firm's total allowable revenues will

be determined by its financial costs. The ways that these revenues are
allocated among customers or sub-services can be determined by
incremental costs and by demand factors. Incremental costs are important
because economic efficiency, including encouragement of efficient
conservation and distributed resource investments, requires that distribution
wires service prices must reflect their incremental costs. Demand factors are
important because they indicate the relative efficiency and fairness of
different schemes by which the fixed costs of distribution wires services may
be recovered.
This section begins by considering the financial and incremental cost
bases of wires service prices. It then discusses demand factors and their
pricing implications.
338 Pricing In Competitive Electricity Markets

3.1 The Financial Costs of Wires Services

Distribution costs have traditionally been bundled, with generation and

transmission costs, into a delivered electricity product. In restructured
markets, distribution costs will be separated from generation and
transmission costs and will then be subdivided among the costs of wires,
connections, and customer services.
The financial costs that are allocated to distribution wires service serve as
the starting point for distribution wires pricing. In principle, these costs
should equal wires service operating costs plus the costs of financing the
wires service rate base. Financing costs depend upon the appropriate rates
of return on capital (including both debt and equity), while the rate base
reflects the depreciated value of wires service capital.
The revenues that the distribution firm is allowed to collect in any year
will be determined by either of two general methods. First, the allowable
revenues in each year can be set equal to the distribution wires financial
costs estimated for that year. This is the traditional cost-based method,
which would be applied to unbundled distribution wires service rather than
to bundled electricity service.
Second, the allowable revenues in each year can be set according to price
caps. Under this approach, allowable revenues are initialized according to
the traditional cost-based method just described. In subsequent years,
however, allowable revenues would change according to a price cap index
(PCI) that neither the utility nor the regulator can control. The distribution
wires firm would then have to keep its wires service prices, as measured by
an actual price index (API), below the PCI. That is:

API :::; PCI (1)

Although specific applications of price cap regulation vary in their design,

the basic formula for the price cap index is:

MCI M-X±Z (2)

where MCI is the rate of change in the price cap index, L11 is the growth rate
in an inflation measure that is not controlled by the regulated firm, the X
factor represents the annual real price decrease that is promised to
consumers, and the Z factor adjusts the allowed rate of price escalation for
reasons other than inflation and productivity trends. Judging by past trends
Pricing in Competitive Electricity Markets 339

in price cap design, a price cap that is applied to distribution wires service
firms would likely use a general measure of inflation for the whole u.s.
economy, such as the gross domestic product price index (GDP-PI). The X
factor would be set equal to the amount by which the estimated total factor
productivity (TFP) growth of the wires service industry differs from that of
the whole U.S. economy.
The actual price index would combine all of the distribution wires
charges into a single number. These charges can include charges for
different wires services; per-kWh, per-kW, and fixed monthly charges; and
time-varying charges. For example, in the simple case of a wires firm that
sells two services, the API would be calculated as:

API (3)

where Pi and Qi are, respectively, the price and quantity of wires service i.

To protect certain customer groups from being singled out for price
increases under a price cap plan, the services for different customers may be
put into different "baskets." For example, if the distribution services for
residential and industrial customers are in the same basket, the distribution
wires utility might satisfy the terms of the price cap plan by reducing
industrial prices even as it raised residential prices. To protect residential
customers from price increases, residential customers might be given their
own separate basket, so that they have their own API that cannot exceed
their own PCI.
Although the evidence is imperfect, price caps seem likely to provide
incentives that are superior to traditional cost-of-service regulation.
Specifically, price caps can provide stronger incentives to cut costs and can
mitigate problems with cross-subsidization, cost-shifting, and affiliate
transaction abuses. Because cost-cutting incentives might lead some
distribution wires firms to cut service quality, the cost-cutting incentives
need to be accompanied by service quality incentives or rules that assure that
service quality will be maintained.
340 Pricing In Competitive Electricity Markets

3.2 The Incremental Costs of Wires Services

The costs of distribution services partly depend upon the number of

customers served and partly upon the sizes of customers' loads. Incremental
costs are thus either customer-related or load-related.

3.2.1 Customer-Related Incremental Costs

Customer-related incremental costs are defined as the cost changes that

accompany changes in the number of customers using the service:

!1Cost
(4)
!J..Numberof Customers i

Customer-related incremental costs thus depend upon the characteristics

of the customers in segment i. Such characteristics can include the sizes of
the facilities that serve that customer segment, the load patterns of customers
in that segment, and the locations of customers in that segment. Customer-
related incremental costs logically serve as the basis for charges levied on
each customer.

3.2.2 Load-Related Incremental Costs

Load-related incremental costs are defined as the cost changes that

accompany changes in the use of a service:

IC LOAD ~ Cost
Lt
(5)
~ Load Lt

Load-related incremental costs depend upon the location L and time t of

customers' use of the system. Load-related incremental costs logically serve
as the basis for charges levied on each unit of load.
The load-related incremental costs of wires services have two
components: incremental line losses; and incremental capacity costs.
Line losses are not directly relevant to unbundled distribution pricing.
This is because line losses are a form of energy consumption, the prices of
which should be determined in competitive generation service markets.
Unbundled distribution wires services companies should be involved in
energy trading only to the extent required to secure the safe and smooth
operation of distribution systems or to meet continuing obligations to serve.
Consequently, each market participant should be responsible for covering
Pricing in Competitive Electricity Markets 341

the costs of their own line losses, primarily by providing compensatory

energy to the power system. Only as a last resort, or as a vestige of utilities'
traditional obligations to serve, should market participants depend upon the
distribution company to procure energy on the participants' behalf.
With respect to line losses, the only necessary responsibility of the
distribution wires company is that it must quantify the losses for which each
market participant is responsible. This can be accomplished by charging
participants for the incremental losses attributable to their individual
transactions or for the whole distribution system's average losses.
Incremental losses can be estimated through engineering simulations in
which load-flow equations quantify losses as a function of load, distance,
and voltage. System total losses can be quantified, through analysis of actual
metered loads and losses for a sample of radial and loop systems, by
dividing system total losses by system total load, thus deriving losses as a
percentage of load. Pricing according to incremental losses encourages
market participants to avoid transactions that involve large energy losses and
is therefore more efficient than pricing according to average losses.
Incremental capacity costs measure how the fixed capital and operating
costs of serving a market participant (or group of participants) changes with
a sustained change in load or, for distributed resources, for a sustained
change in output. Such a long-run costing approach is justified by the facts
that: a) short-run variable distribution costs are tiny relative to total
distribution costs; and b) the capacities of individual distribution facilities
are often large relative to the loads that they are designed to serve. For load,
incremental capacity costs are defined as:

~ Costs Due to L;
IcCAP,;
LOAD (6)
MWofL;

where L; denotes the load of customer group i. For distributed resources,

incremental capacity costs are generally negative, which means that they are
generally cost savings. The definition of incremental capacity costs for
distributed resources is consistent with that for loads, and is as follows:

~ Costs Due to DR j
(7)
MWof DR j

where DRj denotes the output of distributed resource j. For both loads and
distributed resources, the relevant cost changes include the changes in fixed
342 Pricing In Competitive Electricity Markets

costs that arise from changes in the timing of distribution investments, from
changes in the mix of distribution investments, and so on. Incremental costs
are defined according to the characteristics of loads and distributed
resources, including the locations, load or output patterns, and sizes of those
loads or resources.
There are several approaches for estimating the incremental costs of
equations (6) and (7). One widely used approach estimates the statistical
relationship between aggregate distribution investments, load growth, and
new customer connections over a period of several past and/or future years.
This approach is problematic for two reasons: it generally fails to
disaggregate investments and loads by region; and it does not sufficiently
account for the non-load-related purposes of distribution investments, such
as serving new customers sites.
Another approach uses planning studies to determine how regional load
changes and new customer sites affect the future stream of distribution
investments. This approach has the virtue of estimating how an increase in
customers' loads, if sustained over time, will affect future costs; and it can
also provide incremental cost estimates that vary by customer location.

3.3 The Demand for Wires Services

Customers' demands for distribution wires service depend upon customer

willingness-to-pay for power at the distribution system's standard level of
reliability, quality, and convenience in each region. These demands also
depend upon the prices of generation and transmission services: the higher
the prices of these other services, the less that customers are willing to pay
for distribution. Estimating the demand for distribution wires service
basically involves estimating customers' demand for delivered power, and
then subtracting the sum of the prices of generation and transmission
services to obtain a residual demand function. The residual demand for
distribution wires service thus shifts over time as the demand for delivered
power shifts and as the prices of generation and transmission services vary.
The demand for the distribution wires services provided by any particular
wires firm also depends upon competition from other wires firms and from
distributed resources. This competition will usually be most intense for
customers who are large, who have multiple plants that are located in the
service territories of different distribution utilities, who are located near
service territory boundaries, or who are located near abundant and cheap fuel
supplies.
Pricing in Competitive Electricity Markets 343

3.4 Pricing Wires Services

Pricing wires services involves three basic steps.

The first step is to determine the extent to which the wires firm's different
services should be "segmented". For example, there can be separate pricing
for services that are provided:

• to different customer locations,

• at different customer voltage delivery levels,
• to customers of different sizes;
• to customers with different load patterns, or
• with different levels of service quality.

The second step is to set "standard prices" for wires services. Prices
should depend on the ways that serving different customers entails
differences in:

a) their incremental costs of supply, and

b) their total costs of supply.

Prices should also depend on the differences among customers in:

a) customers' willingness-to-pay for services, and

b) the intensity of competition for each customer's business.

Roughly speaking, the price of any service k sold to customer group i

should be approximately as follows:

E{ ICik *Q'ik } - a E{ Qik}

I1k (8)
E{ Q'ik }

Where E is the expectation operator, ICk is the incremental cost of the

service, Qik is the quantity of the service sold by the wires firm, Q'ik is the
rate at which the demand for the wires firm's service falls as price rises (i.e.,
the slope of the demand curve), and a is the "Ramsey" number. All
variables in equation (8) are particular to each service except for the Ramsey
number, which is common to all services. The Ramsey number is set so that
the revenues from all services and all customers just meets the wires firm's
revenue requirement. i Hence, equation (8) considers incremental costs
through the variable ICk, total costs through the variable a, and customers'
344 Pricing In Competitive Electricity Markets

willingness-to-pay and the intensity of competition through the demand

variables Qik and Q'ik.
Ramsey pricing is controversial because it leads to mark-ups, over
marginal costs, that are highest for those customers and those services for
which demands are least responsive to price - which are often those services
for which supply is least competitive. Ramsey pricing can lead to higher
percentage mark-ups for smaller customers than for larger customers, and to
substantial revenue recovery through fixed monthly fees rather than through
usage charges. Evolution of pricing in this direction will improve the
efficient use of distribution systems but will adversely impact some
customers and can raise questions of fairness.
The third step is to identify situations in which "special prices" rather
than "standard prices" are warranted. For example, wires firms will
sometimes face situations in which they are tempted to offer special deals -
such as "economic development rates" - to certain customers either to attract
them to the wires firm's service territory or to retain them. To an extent, the
same considerations that have governed utility-customer negotiations in the
past will govern them in the future: how much margin is at stake; how is the
local economy affected by the job impacts; and how large a price discount
does the customer really need? In a world of unbundled services, however,
additional complications arise from the fact that independent merchant firms
will provide power while the distribution wires firm provides transportation.
A part of the margin that is at stake can be the merchant's margin, not the
distribution firm's; and discriminatory behavior on the part of the
distribution firm may be more transparent and subject to controversy.

4. DEVELOPING AND PRICING CONNECTION

SERVICES

When opened to competition, distribution connection services will include

a large array of products that are tailored to provide different combinations
of equipment to serve customers' various needs for service reliability,
quality, and convenience. As examples, dual connection service can
improve reliability; special filters, capacitors, or reactors can enhance
quality, as can three-phase instead of one-phase service; and underground
service offers greater convenience (or aesthetic benefits) than overhead
service.
Competition will drive connection service prices toward the incremental
cost of each service. Competitors will be firms that have the relevant
engineering expertise and staff, including distribution utility firms and a host
Pricing in Competitive Electricity Markets 345

of construction firms. For the sake of public safety, all competing firms
must be required to meet minimum standards of service quality.
The incremental cost of providing connection services to a particular
customer consists of the capital, installation, and maintenance costs of the
particular facilities that serve the customer. These costs may be measured
according to equipment invoices and according to the labor time required for
installation and maintenance. For any existing facility, sunk capital and
installation costs should be measured at replacement value less economic
(not accounting) depreciation.
Instead of separately determining the costs of each individual facility, it
may, in many situations, be sufficiently accurate to determine typical costs
for each type and size of equipment used to provide connections services,
perhaps with additional cost differentiation for other factors (such as
regional location and load density) that significantly influence cost. These
typical cost estimates must be regularly updated.
When connection services are offered as product packages, the costs of a
package will generally equal the sum of the costs of the services that
comprise the package. For example, a "dual feeder service" package, which
improves the reliability of a customer's service by transporting power to the
customer on two feeder lines from two substations, may entail costs of lines,
poles, capacitors, and reactors. The cost of this service would equal the cost
of the facilities that are required to offer the service.
For connection services, prices equal to incremental costs will likely be
sufficient to allow full cost recovery, including a normal return on capital.
Depending upon the restrictions imposed by regulators, service providers
will set the prices of these services above incremental costs to the extent that
competition and customer willingness-to-pay allow. In general, service
providers will offer an array of standard connection services at standard
prices while offering discounts and tailored products as needed to meet
particular customer situations.

s. DEVELOPING AND PRICING CUSTOMER

SERVICES

When open to competition, distribution customer services will include a

modest array of options that can serve customers' various preferences for
certain kinds of convenience. Such conveniences can include: a) customers
being able to easily contact the distribution firm to resolve questions or
problems; and b) special billing arrangements that save the customer time in
understanding or paying their electricity bills.
346 Pricing In Competitive Electricity Markets

For distribution customer services, competitors are likely to include

merchant firms and non-electric utilities (such as gas utilities) that serve the
customer's area. These firms will have the customer databases, computer
software, and communications infrastructure required to handle customer
billing and communications needs.
Customer service costs can be distinguished according to service type
and service variation. "Service types" include billing and customer
communications. ii "Service variations" include service differences that may
affect cost, such as customer choice of billing option (pay by mail versus
automatic bank debit, standard versus aggregated billings) and customer
location.
There are three main methods for quantifying the per-customer costs of
customer services:

• Allocate financial costs to each service type and divide by the

appropriate number of customers.
• Allocate financial costs to each service type for a period of several
years and then, for each service type: a) determine a statistical
relationship between costs and numbers of relevant customers, with
due consideration for service variations; b) infer marginal customer
costs (preferably for each service variation) from the statistical
relationship; and c) allocate above-incremental costs among
customers (preferably according to relative willingness-to-pay).
• Assess the causal linkage between costs and customers on the basis of
individual activities such as metering, bill calculation, bill recording,
bill mailing, and remittance processing. This activity-based cost
assessment yields incremental costs that can be used to allocate total
accounting costs and to competitively price customer services.

Prices will reflect these per-customer costs.

6. PLANNING, OPERATIONS, AND SERVICE

QUALITY ISSUES

A significant challenge of deregulation is identifying incentive

mechanisms that will assure maintenance and even improvement in service
quality. For competitive services, this challenge is relatively minor:
competitors will be required to meet minimum standards if they wish to
participate in markets; and competition will weed out those competitors
whose services do not meet market standards in excess of the minimum
standards.
Pricing in Competitive Electricity Markets 347

The relatively harder challenge concerns the quality of non-competitive

wires services. There will be (and have been) cases in which it will be
profitable for wires firms to cut capital and operating costs at the risk of
service quality. The costs of that risk are borne by customers who suffer
outages or power quality deterioration but are virtually captive to the wires
monopolist.
A key role of distribution service regulation is to require that distribution
wires investment and operating procedures meet certain tests, and that
service quality meets certain standards. The basic test for investments and
operations is that the expected incremental benefit of any action should
exceed the expected incremental cost of that action. Because benefits
include those of reliable service and high power quality, this test would
include explicit consideration of the value of service quality to consumers.
With respect to service standards, regulators and wires firms need to
develop measures of service quality as well as mechanisms for quantifying
and recording the service quality that was actually achieved. Regulators and
wires firms also need to develop service quality targets, which should be
accompanied by performance incentives that penalize failures to meet
quality targets and perhaps reward performance that exceeds the targets.

7. CONCLUSIONS

The restructuring of the electric power industry will lead to at least some
unbundling of distribution services. There are three areas in which
distribution utilities will be most prone to competitive pressure; and this
pressure will drive reform in the ways that distribution services are packaged
and priced.
First, competition among merchant firms is driving those firms to grab
control of information about customers. Even if the provision of customer
services is not itself profitable, the customer information gained by
providing those services is so valuable to merchants that they will fight for
the proprietary right to that information. One can imagine independent
distribution utilities maintaining their monopolies over customer services
and freely providing the resultant customer information to all interested
merchants. But this approach seems unlikely to satisfy merchant firms who
are seeking market share and the large electricity consumers who want the
widest possible array of service alternatives.
Second, connection services are a natural candidate for widening
competition. Many suppliers can provide these services. Many customers
will want service variations that have not been traditionally been available.
Many other customers who have relatively inexpensive connection service
348 Pricing In Competitive Electricity Markets

needs will simply want to avoid subsidizing the connection services of

customers who have more expensive needs. A key technical challenge in
defining the boundaries of competition will be in distinguishing between
those facilities that serve identifiable groups of customers and those that
serve larger and more amorphous customer groups.
Third, even the final bastion of the distribution monopoly, the provision
of wires services, will face some competition. Some power market
observers believe that distributed resources threaten to undermine wires
monopolies: with photovoltaic cells on every roof, so the fantasy goes, wires
services will go the way of the horse and buggy. The more likely outcome is
that the potential competition from distributed resources will force some
rationalization of distribution wires rates. For example, traditional recovery
of fixed distribution costs through per-kWh charges encourages consumers
to install "uneconomic" distributed generation that has costs that exceed the
accompanying distribution wires cost reductions. This is an example of
"bad" distribution wires tariff structures inducing consumers to make
decisions that are bad for society. The solution is to reform distribution
wires tariffs so that they better reflect distribution wires costs.

NOTES

i More accurately, the Ramsey number is set to meet a profit target, which is equivalent to
meeting a revenue target when quantities sold (and hence costs) are fixed.

ii Metering is often considered to be a customer service; but because its physical and cost
characteristics are those of a connection service, we include it among connection services.
SECTION V

CASE STUDIES
Chapter 21

Pricing Throughout the Product Lifecycle: When

Mature Markets Meet Innovation

William LeBlanc
E-Source

Key words: Lifecycle; Marketing; Positioning; Pricing; Product Development; Profit

Margin.

Abstract: Pricing of products and services needs to change as products are

introduced, grow, and mature. In order to maximize profits, energy
service marketers must understand and work within the framework of
product life cycles.

1. TIMING IS EVERYTHING

As retail energy companies develop new products and services for

emerging markets, it is imperative for them to understand the unique
characteristics of each stage of a product lifecycle. As products move
through the phases of introduction, growth, maturity, and decline, marketing
strategies need to be altered accordingly. In particular, the pricing aspect of
the marketing mix takes on very different roles during each phase,
determining the level of profitability it will be possible to achieve.
The energy market poses a unique problem from a lifecycle management
perspective. Commodity energy markets can be considered as relatively
mature, but deregulation could throw them back into a growth phase, as we
witnessed in the restructuring of the telecommunications industry. In
addition, retail energy companies will be combining new products and
services that are in the introductory stage with mature commodity products,
creating pricing challenges not previously encountered in these markets.

A. Faruqui et al. (eds.), Pricing in Competitive Electricity Markets

© Kluwer Academic Publishers 2000
350 Pricing In Competitive Electricity Markets

This paper provides product managers with the concepts and tactics they
need to better manage pricing and marketing throughout product lifecycles.
It will consider a range of product and service categories, highlighting their
distinct characteristics and explaining the different tactics that will be critical
to successful marketing. The categories covered here include:

Products
• Technology-based
• Appearance-based
• Consumables
• Durables

Services
• Technology-Intensive
• Mass-customized
• Personalized

Product managers in competitive industries have been working within the

product lifecycle framework for many years. Retail energy companies can
learn from this experience and avoid many mistakes and financial losses.

2. PRICING AND MARKETING STRATEGIES FOR

THE FOUR LIFECYCLE STAGES

Most marketers are familiar with the four basic stages of a product's
lifecycle: introduction, growth, maturity, and decline. We have added a pre-
introduction stage identified as "research and development" (R&D). This
lifecycle concept is most useful when applied to a product category rather
than to individual products. For example, a video cassette recorder would
define a product category, but a VCR with four heads and improved sound
features would only be a twist on the basic product category.
Actively managing products and services based on their lifecycle stage
can maximize profitability levels. It also encourages marketers and product
developers to create families of products and to produce a continuous stream
of new products that will offer customers fresh choices. Intel is one
company that does an excellent job of making its own products obsolete.
While most companies try to squeeze every bit of profit out of their
successful products, Intel deliberately plans and creates products that will
put their earlier chips out to pasture.
Pricing in Competitive Electricity Markets 351

2.1 Stage 0: R&D

In this pre-introductory stage, pricing decisions may be far from the minds
of the product developers. However, from a positioning standpoint, this is
actually the most critical stage. Each product and service produced by a
company should follow its overall business and marketing strategy. For
example, Nike is a premium product company that thrives on the rapid
introduction of new products. Customers expect to see new fashions yearly;
they also expect to pay a high price for them. Selling the same basketball
shoe at a cut rate year after year just wouldn't fit with Nike's market
position.
Wal-Mart stores carry products that offer both price and cost advantages
to customers. It does not carry top-of-the-line or designer products. That
wouldn't fit Wal-Mart's low-price position in the marketplace. Ritz Carlton,
the luxury hotel group, spends whatever is necessary to hire and train
exceptional staff, but its customers expect to pay top dollar for superior
service. You won't see Ritz motor hotels along the U.S. Interstate freeway
network anytime soon.
Development teams should have a price level in mind when they are
creating new products and services. One of the most famous examples of
creating a product around a price point is Lee Iacocca's development of the
Ford Mustang. Ford market research determined that a large customer group
longed for the styling, but not necessarily the performance, of expensive
sports cars. So Ford designed a new car from the bottom up to meet this
customer need at a price point of $2,500, which this middle-class customer
group could afford in 1964. The Mustang turned out to be one of the most
successful and profitable cars of all time.
Compaq made a similar move when it created a computer for under
$1,000. That product broke open pent-up demand in a customer group that
couldn't fathom spending nearly $2,000 for a fully equipped machine. The
gamble was that customers would be willing to buy a system that wasn't
absolutely state-of-the-art if the price was right. The gamble paid off.

2.2 Stage 1: Introduction

The introductory stage of a product sets the tone for both the customer
base and potential competitors. Many product pioneers end up with arrows
in their backs during the introductory stage. When new products are
introduced to the world, substantial customer education is usually needed,
and all parts of the marketing mix must be right for products to succeed.
At the beginning of the introductory phase, most customers lack a
reference point for prices; without those comparison points, it will be
352 Pricing In Competitive Electricity Markets

difficult for them place a value on the product's benefits. For that reason,
new products and services are often targeted at customer segments that are
relatively price-insensitive. Sony's Discman initially retailed for
spectacularly high prices. Even so, some early adopters were willing to buy
them, helping to create awareness in the marketplace. As time progressed,
Sony dropped the price to gain more market share and to help hold off
competitors who were introducing their own models.
Introductory discount pricing is also common at this stage. That
approach is just the opposites of Sony's strategy, in those initial prices are
made deliberately low to get people to try the product. For example, pre-
paid phone cards are given away in hopes of getting customers hooked on a
new provider, and some banks offer free checking or free ATM use for a
year to new customers. However, marketers should try their best not to
discount prices in the introductory stage. Lowering prices signals that the
product might not be all that valuable, and attempting to raise prices at a
later date always creates barriers to adoption because customers have come
to expect discounts.

2.3 Stage 2: Growth

Companies with products in the growth stage are the ones throwing lavish
corporate parties and offering big bonuses to employees. The growth phase
is where the greatest profits are made. From a pricing standpoint, it is
critical for well-positioned products to maximize their return during this
stage, when customers are actualizing the value that a product or service
provides. Although competitors are typically joining the market in large
numbers throughout the growth stage, their products are likely to be
differentiated in some way, and there's usually room for many players to
succeed.
Marketers should be wary about their overall pricing objectives at this
stage. It is very tempting to lower prices to gain market share, rather than
maximizing profitability, but that will only throw the product into the
maturity stage more quickly. The key is to constantly innovate, rolling out
product improvements that will allow the premium-pricing phase to
continue. During the growth phase, many companies also create multiple
options, offering consumers choices ranging from high-priced, high-value
models down to more economical models.

2.4 Stage 3: Maturity

The maturity phase of the product lifecycle can also be called the price··
war stage. Intense competition is prevalent, and customers tend to shop on
Pricing in Competitive Electricity Markets 353

price because the products are no longer highly differentiated. All surviving
producers have made efficiency improvements, and there may be only two to
four significant players left, which hold some 80 percent of the market. For
example, markets for home appliances are largely mature. Each major
manufacturer has an economy, mainstream, and high-end brand, and
customers buy primarily on price. The airline industry also shows signs of
maturity, with competition centering on the price of a ticket.
Companies competing in mature markets are always trying to innovate
and move back into a growth phase, but usually their innovations are easily
matched by the competition, which winds up benefiting customers rather
than the sellers. When McDonald's has a promotion, for instance, you can
count on Burger King and Taco Bell following right behind. Computer
hardware is primarily a mature industry, even though it continues to evolve.
No single company has a highly differentiated product. Dell has been
successful due to its innovative distribution strategy, bypassing the
middleman and selling directly to customers.
At the mature stage, when pricing is the only differentiating factor,
controlling costs becomes a big issue. Advertising and branding are also
huge battlefields in mature markets. Given the lack of product
differentiation, sellers have found that bombarding customers with their
brand name will influence buying behaviors.
Although it is unusual for a mature market to move back into a growth
stage, it can happen if there are dramatic changes in technologies or industry
regulations. The telecommunications industry has experienced this kind of
phase shift. When the Bell system was broken up in 1984, it created a
market that looked like a mature, competitive market, marked by price wars
in abundance. But the lower prices were significant enough that they
spurred demand in a big way. In addition, a series of new technologies
spurred ongoing evolution of the entire business. The use of fax machines,
computer modems, e-mail, and other traffic across the Internet increased
telephone use tremendously. Cell phones were also introduced as a new
product category; videophones may soon be on the horizon.
Today's electricity market can be characterized as mature because it is
virtually impossible to differentiate electric service at present. Although
some forays are being made into selling "green" electricity and premium-
quality power, most of the activity in electricity product differentiation is
happening on the bundling side. Power marketers are trying to package
energy efficiency, energy information, metering, and other value-added
products with commodity purchases. Time will tell whether this bundling
approach will provide real value to customers or prove to be part of a
market-share grab by the larger power marketers.
354 Pricing In Competitive Electricity Markets

Companies can still make money in the mature stage, even though profits
are squeezed. Energy service providers are more likely to succeed if they
understand their costs in great detail, are able to change prices selectively
and quickly, possess excellent information on market segments, and can
distinguish a profitable sale from a loser.

2.5 Stage 4: Decline

The most important aspect of the decline stage is recognizing that market
failure is inevitable. Companies that hold on "one more year" are only
digging themselves a deeper hole. Discounting may not be a problem for
premium service providers, so the decline stage should be managed
carefully, moving valued customers from old products to new ones that are
in the growth stage. For neutral and lower-priced products, many companies
speed the movement of old inventory by offering discounted pricing.
Recently, Hunt Technologies, an automatic metering (AMR) company,
had difficulty obtaining the standard kilowatt-hour meters that have forP.1ed
the basis of its AMR retrofit. The mechanical meters appeared to be in the
decline stage, with key manufacturers lagging behind by three to six months
on orders. It was beginning to look like electronic meters would soon
replace mechanical meters as costs for the new meters dropped and their
functionality increased. It is possible that companies like GE have been
trying to phase out the mechanical meters they see as being in decline, yet
new add-on AMR applications may force these older meters back into
growth, at least for a few years.

3. PRODUCT AND SERVICE CATEGORIES:

LIFECYCLE VARIATIONS

Different categories of products and services have inherently different

lifecycle characteristics. Product managers can make better marketing
decisions if they take a close look at the issues that are unique to their
particular products or services.

3.1 Technology-Based Products

Electronic equipment and pharmaceuticals are prime examples of

technology-based products. These kinds of products are typified by intense
R&D costs. Such investments are risky, and only well-capitalized firms can
afford to take them on. Because of the high up-front costs, companies must
Pricing in Competitive Electricity Markets 355

achieve high profit margins throughout the product lifecycle in order to

survive and prosper.
The introductory stage can be quite lengthy, as customers are often
reluctant to try new "gadgets," and there may be constant pressure to lower
prices, because higher volumes can reduce unit production costs
substantially. Technology-based products can enjoy long growth stages,
during which new models and features may be added and performance may
improve as the technology evolves. Often, prices can remain high for the
high-end products while mass-market buyers enjoy older, lower-cost models.
The maturity stage is often short, as new categories of products replace old
ones.

3.2 Appearance-Based Products

Perrier water, Lexus automobiles, and many sporting goods are purchased
on the basis of appearance as well as functionality. Appearance-based
products have unique lifecycle characteristics. Products that are purchased
on the basis of exclusivity or prestige can often demand prices that exceed
what economists would consider their evident consumer value.
R&D for appearance-based products begins with creativity and depends
on branding to create customer demand. In this category, customers won't
be able to articulate to product developers what they want; instead,
innovative entrepreneurs must tell customers what they want. These
products have fast introductions and fast growth stages. They may then be
pulled by the supplier when it determines that the "aura" for a given type of
product has vanished.

3.3 Consumable Products

Consumable products are often staples, such as grocery foods, household

goods, and gasoline. They tend to be used up quickly, allowing the buyer to
make multiple purchasing decisions over the course of a year or so. Prices
for consumables are often quite low, making it relatively easy to conduct
market trials.
Branding becomes the most important means of gaining price premiums
for products within this category. Although producers try to increase sales
volume by providing "new and improved" versions or by selling variations
and twists on the primary product, consumables often linger in the maturity
stage for a long time.
356 Pricing In Competitive Electricity Markets

3.4 Durable Products

Appliances, furniture, garden equipment, and some automobiles can be

considered durables. The category applies to products that have relatively
long lives. Pricing is quite important for durables; buyers may spend a good
bit of time analyzing the value of their investments, since durables often cost
quite a lot. The maturity stage is quite long, and prices are ultimately very
competitive.

3.5 Technology-Intensive Services

Services have different lifecycle characteristics than products.

Technology-intensive services are growing rapidly in today's information
age. These services typically take some newer, more advanced approach to
their markets, creating new value, often within fairly mature industries. For
example, banking has been transformed with the advent of electronic
transfers and ATMs, and more change is coming as a result of the rise of
electronic commerce. Healthcare services are also evolving rapidly as new
high-tech machines make diagnosis and surgery more effective and efficient.
R&D stages for these kinds of services require substantial innovation.
Often, technologies exist, but they are not yet being utilized to their full
extent. Metering companies, for instance, are starting to use satellite
technology to read meters, and energy service providers can now manage
many buildings from a remote site through electronic links. These types of
services may pop up quickly as integrators put new packages together, but
the introductory stage can be quite lengthy. People may be slow to adapt to
new ways of doing business. For example, although electronic commerce is
convenient, it flies in the face of our culture of secure transactions.
Pricing for these services is based upon value-added or cost-avoided
characteristics. In healthcare, for instance, an ultrasound diagnosis may cost
less than an x-ray alternative. It may be possible to set the price for
ultrasound tests below the costs of x-rays and still provide significant profit
margins. Many other technology-intensive services become bundled,
functioning essentially as cost-reducers for the supplier (such as ATMs and
card scanners) or convenience enhancements for the customer (such as on-
the-spot receipts for car rentals or electronic airline tickets). In these cases,
the price of the services may be essentially zero to the customer.

3.6 Mass-Customization Services

Selling services to the masses is big business. Many of the largest

companies in the U.S. can be placed in this category. These companies have
Pricing in Competitive Electricity Markets 357

taken personalized services and made them available to the general

population at better prices, thanks to economies of scale and scope. This
category includes all kinds of chains, such as Starbucks, Taco Bell, Target,
Wal-Mart, Safeway, Texaco, H&R Block, SuperCuts, and Century 21.
Utilities also fall into this category.
Pricing in this category is usually very competitive. Companies
concentrate on efficient production and on keeping prices in line with the
competition. In the early introductory stages, the lines between personalized
service and mass-customization can be blurry. Some mass-customizers are
quite specialized and can command a premium, as Starbucks does. But they
will come under price pressure as soon as somebody copies their formula for
success - and more than one competitor may enter the fray.

3.7 Personalized Services

The services we use in our everyday lives - such as those provided by

doctors, lawyers, financial planners, and architects - are highly customized to
each individual. Virtually no R&D is involved for services in this category.
The introductory stage mainly involves gathering clients.
Although a premium is often charged for these personalized services,
they are still subject to the price pressures of mature markets: HMOs push
doctors' prices down, and RFPs subject engineers to price-only competition.
Personalized services are in constant flux, and innovation in the provision of
such services is most often the area with the greatest potential for growth.
Constant pressure to keep the service "customized" exists, to prevent buyers
from turning to mass marketers.

4. CONCLUSIONS

Marketers who understand the unique characteristics of product and

service lifecycles will have a competitive advantage over their competitors.
Product developers need to maintain a constant flow of fresh products, so
that as one goes into decline, it can be replaced by products in the growth
stage. Energy service providers face some unique challenges. Most
customers say that they want "good old reliable electricity" and claim that
they are not interested in bells and whistles. It is a marketing challenge to
alter this mind set and provide true value as the industry enters the new
century of energy competition.
358 Pricing In Competitive Electricity Markets

REFERENCES

LeBlanc, William J., and Fiebelkorn, Tammy. Product Life Cycle Management: Adapting
the Best Practices a/Other Industries. EPRI TR-108984. November, 1997.
Chapter 22

Residential TOU Price Response in the Presence of

Interactive Communication Equipment

Steven Braithwait
Laurits R. Christensen Associates, Inc.

Key words: Elasticity of Substitution; Electricity Demand; Electricity Pricing; Price

Elasticity; Time-of-Use Pricing.

Abstract: This paper examined customer price responsiveness to an innovative

residential TOU program in which customers face high critical prices during
periods of high wholesale power costs, and have the ability to use an
interactive communication device to pre-schedule certain major energy-using
devices. Customers facing this rate shifted substantial load from both peak.
and shoulder periods to lower priced off-peak. periods. Customer price
response under this innovative program exceeded by a substantial margin the
response found in previous TOU programs.

1. INTRODUCTION

How do residential customers respond to occasional market price signals

in the presence of an interactive communication system? This chapter
summarizes the findings from an innovative residential time-of-use (TOU)
rate program that provided each customer with such a system. The pilot
program, implemented by an east-coast utility in the summer of 1997, had
two key features that distinguished it from traditional TOU rates. First, the
communication system allowed the utility to send a "critical" price to
participating customers during periods of high-cost supply conditions.
Second, the system allowed customers to pre-schedule their response to both
the standard TOU prices and the receipt of a critical price signal by
indicating thermostat settings and possible circuit interruption at various

A. Faruqui et al. (eds.), Pricing in Competitive Electricity Markets

© Kluwer Academic Publishers 2000
360 Pricing In Competitive Electricity Markets

price levels. Analysis of customer load data confirmed that customer load
response under this innovative program exceeded by a substantial margin the
response found previously in typical TOU programs. Customers reduced
consumption during peak periods and some shoulder periods and increased
consumption during off-peak and certain shoulder periods. In addition,
estimation of customer demand models generated substantially larger
elasticities of substitution than those found in most previous studies of
traditional TOU programs, indicating strong customer price responsiveness.

2. BACKGROUND

Numerous pilot residential TOU programs have been implemented and

evaluated, revealing that customers in general respond to time-varying
prices. i However, traditional TOU rates have suffered from two fundamental
problems. First, TOU rates typically do not accurately reflect the actual cost
of providing power on a day-to-day and hour-to-hour basis. For example,
summer peak period prices are typically considerably higher than the
expected cost of power on most days of the summer; however, they fall short
of costs during the relatively few but important periods of tight system
conditions. Second, residential customers have limited capabilities to
schedule or modify their energy usage, particularly during daytime periods.
As a result, the potential benefits to be gained from residential TOU rates
have been found to be small relative to the additional costs of metering. ii
In contrast, the combination of a critical price in conjunction with a TOU
rate, plus the interactive communications equipment tested in this pilot
program provides several potential advantages. First, the ability to send a
relatively high critical price on the days of highest cost should allow lower
peak period prices than under a standard TOU rate, thus better reflecting
power costs. It should also reduce revenue losses due to customer response
to high peak-period prices on moderate-cost summer days. In addition, the
pre-scheduling feature of the system allows customers to respond to prices
without having to remember to take actions manually, and to respond
differently to the critical prices. Furthermore, the load relief provided by the
critical price comes precisely at the time needed by the utility.

2.1 Program Features

The variable rate consisted of a basic TOU rate with three price tiers
(peak, shoulder and off-peak), plus the possibility of a critical price that
could be sent during the peak period for a limited number of high-cost hours
of the summer. The program was designed with two alternative sets of price
Pricing in Competitive Electricity Markets 361

values for the three tiers, plus a $.50 IkWh critical price. The program was
offered through mail solicitation in two regions of the utility service area.
Volunteers were screened for certain dwelling and behavioral characteristics
(e.g., single-family home, presence of air conditioning, no plans to move in
the near future, compatible thermostats and wiring for the communications
equipment, etc.), and then selected for either the treatment group or a control
group. The treatment group was in tum divided equally into two groups,
each of which faced one of the two variable rates. A control group was
selected from the pool of volunteers to serve as a classical experimental
control group for purposes of estimating the load changes of the treatment
group in response to the variable rate. The communication equipment was
installed in the treatment group customers' homes during the spring of 1997,
while hourly interval recording devices were installed in the control group
customers' homes during the same period.
The specific tier prices for the two alternative TOU rates, and the hours
in which they apply are shown in Table l.iii

Table 1. Specific Tier Prices for two TOU rates.

Hours Price tier Rate 6173 ($IkWh) Rate 9122 ($IkWh)

1- 8 Off-peak. .065 .09

9 - 14 Shoulder .175 .125

15 - 18 Peak .30 .25

19 - 20 Shoulder .175 .125

21 - 24 Off-peak. .065 .09

Note that the peak to off-peak price ratios for the two different treatment
rate groups are 2.8 and 4.6 to 1, while the ratios for the critical price are 5.6
and 7.7 to 1. These price ratios are comparable to previous TOU tests that
have indicated that price ratios in the range of 4 to 1 or 5 to 1 are needed to
induce substantial response.

2.2 Analysis Approach

We undertook two types of analysis of the customer load and price data to
determine the extent of customer load response to variable prices. First, we
compared average usage patterns for treatment and control customers using
descriptive statistics and graphical methods. These provide an intuitively
362 Pricing In Competitive Electricity Markets

straightforward summary of customer response to TOU pricing under the

specific prices observed in the program. However, the load response
observed under one set of conditions cannot be easily extended or adjusted
to different price scenarios, customer characteristics, or weather scenarios.
For that reason, we also developed relative measures of customer price
response in the form of elasticities of substitution (e.g., between peak and
off-peak periods) derived from estimated models of customer demand.
These models relate customers' demand for electricity in different time
periods to the price in those periods and other important factors such as
customer and dwelling characteristics. Elasticities of substitution, which are
related to traditional price elasticities, represent the percentage change in the
ratio of usage between two time periods that occurs in response to a given
change in the ratio of prices for electricity during those periods.
We report elasticity estimates based on two alternative demand models.
One model, which we have used in a number of previous TOU evaluations,
assumed a constant elasticity of substitution (CES). We have found that the
CES form offers a useful compromise between theoretical attractiveness, and
simplicity and parsimony in parameters. The results shown in this chapter
represent the first level of a three-level CES model that we have used
previously to represent the substitutability of electricity between the TOU
time blocks of a given day, and between weekdays and weekends, as well as
any overall change in electricity consumption due to changes in the average
price of electricity. We have also estimated a model known as a generalized
Leontief (GL), which is a more complex but flexible model that allows for
differential rates of substitution between consumption during different time
periods. This more flexible form may be appropriate in the present case of
multiple prices and time periods.
The descriptive statistics are reported first, followed by the demand
model results.

3. DESCRIPTIVE STATISTICS
The results reported below compare energy usage by all or portions of the
treatment group to appropriate control group usage during the same period.
The control group usage serves as a proxy for the treatment group usage
pattern that would have occurred had they not faced the TOU prices and
made use of the load scheduling devices. The statistics and load curves
typically report averages across certain groups of customers, day types and
hours. For example, we may compare the average hourly usage of the entire
treatment group during the peak-price period on weekdays in July to the
comparable value for the control group. One sub-grouping of interest is the
Pricing in Competitive Electricity Markets 363

division of the treatment group into two groups that received different TOU
pnces.
Hourly load data for each of the treatment and control group customers
were collected for the period of June through September 1997. These data
were divided into three four-week "months" beginning in late June. Note
that all of the six critical price days occurred in the first "month." The
hourly load data were averaged over like hours separately for all weekends,
for all non-critical price weekdays, and for the critical price days. Also, in
averaging the load data across customers, we applied appropriate sample
weights for customers in each of four energy-use strata from which the
treatment and control group samples were selected.

3.1 Load Profiles

The following figures show average hourly loads for the treatment and
control groups for several day types. These figures highlight the differences
between their loads, which can be interpreted as the treatment group's load
response to the TOU prices. Figures 1 and 2 (below) show average hourly
loads for non-critical and critical price weekdays respectively, for the first
four weeks of the summer for which data were available. The load plots
show rather dramatic evidence of load shifting. The load reductions relative
to the control group are greatest during the peak period, averaging .53 kW,
or 26 percent, and somewhat less during the late-morning shoulder period. iv
The treatment group customers also appear to shift forward some
consumption into the last few hours of the morning off-peak period,
presumably by running discretionary appliances such as dishwashers, water
heaters and pool pumps at that time, and possibly by pre-cooling their houses
in anticipation of the higher-priced shoulder and peak periods. They also
appear to recover some of the usage foregone during the peak period in both
the two-hour early-evening shoulder period and the first few hours of the
following off-peak period.
In general, load response on the critical price weekdays (Figure 2)
follows the same pattern as the non-critical price days, although the load
levels and amount of response, particularly during the peak period, are
greater. V During the first hour of the peak period, the treatment group load
response was 1.24 kW, a nearly fifty percent reduction relative to the control
group. The load impact remained just below 1 kW during the next two
hours, falling somewhat during the last hour to .59 kW, presumably as
customers began to recover some of the foregone air conditioning load. The
treatment group usage also remained substantially higher than that of the
control group during the last two time blocks than it did on non-critical price
days.
 364 Pricing In Competitive Electricity Markets

3.00 ,--------------------------__-------------------------------------------- --,

2.50

.roo
,x --Control

Jt.,

• - -x' • - Treatment
x'
x,
'X.
,..
. ..z'

")C" 2:"

0.50

Off-Peak Sboulder Shoulder On·Peal

2 3 4 5 6 7 8 9 10 11 12 Il 14 15 16 17 18 19 20 21 22 23 24

Figure 1, Average Hourly Usage (kWhihr)-Treatment and Control Groups

Non Critical Weekdays, Month 1 (June 29-July 19, 1997)

400 r-------------------------------------------------------~

3.50

300

~ --Control
~ 250

~
i 2.00

,,'
't:
;
= x, .' •• -J:" -Treatment
~ 1.50
..,
~

100

0.50
Off-Peak Shoulder Shoulder Off-Peal

I 2 3 4 5 6 7 8 9 to II 12 13 14 15 16 17 18 19 20 21 22 23 24

Figure 2, Average Hourly Usage (kWhihr)-Treatment and Control Groups

Critical Weekdays, Month 1 (June 29-July 19, 1997)
 Pricing in Competitive Electricity Markets 365

3~ r----------------------------------------------------------------,

300

-.
I --Control

i • \'
\~

/ - • -Trc:BI~nl-R8te6173
x' • .e

- -x- - Treat~nl- Rate 9122

100

o.~

Off-Peak Shoulder Puk Shoulder Off-Peak

2 3 4 5 6 7 g 9 10 11 12 13 14 15 16 17 )8 19 20 21 22 23 24

Figure 3_ Average Hourly Usage (kWh/hr)-By Rate Group

Non Critical Weekdays, Month 1 (June 29-July 19, 1997)

2~ r------------------------------------------------------c

' .. i
~
J<'
2.00
.J<' -. i
")(-·s :•.

~
. \
--Control
~ 1.~
}f.' , .
j
t' .J<'
~
r. 1.00 • - .JC- - -Treatment
e X
~

'"
o.~

3 4 5 6 7 g 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

Figure 4. Average Hourly Usage (kWh/hr)-Treatment and Control Groups

Weekends, Month 1 (June 29-July 19, 1997)

366 Pricing In Competitive Electricity Markets

Figure 3 (above) shows separate load curves for the two treatment rate
groups on non-critical weekdays. The group that faced higher peak and
shoulder prices (Rate 6173) experienced approximately fifty percent larger
usage reductions during the peak (.64 kW, compared to .43 kW) and
shoulder (.33 kW, compared to .25 kW) periods than the group that faced
somewhat flatter prices. In addition, the group with the relatively higher
shoulder price continued to reduce usage during the late afternoon shoulder
period relative to the control group, while the other group's usage during that
period increased.
Finally, it is instructive to examine usage patterns on weekends and
holidays, during which the treatment group faced off-peak prices in all
hours. The average weekend loads shown in Figure 4 (above) for the
treatment and control groups appear quite similar, with slightly (not
significantly) lower levels for the treatment group.

Table 2. Average Daily Usage (kWh)

Weekdays
Difference
Month Control Treat Usage Percent
I 39.3 36.2 -3.1 -7.9%
2 31.2 29.3 -2.0 -6.3%
3 24.9 22.7 -2.2 -8.8%
Summer 30.4 28.3 -2.2 -7.1%

Weekends
Difference
Month Control Treat Usage Percent
I 35.9 34.6 -1.3 -3.7%
2 39.1 40.0 0.9 2.3%
3 28.7 27.7 -0.9 -3.3%
Summer 34.1 33.7 -0.3 -0.9%

All Days
Difference
Month Control Treat Usage Percent
I 40.3 38.1 -2.2 -5.5%
2 33.5 32.3 -1.2 -3.4%
3 26.1 24.3 -1.8 -6.8%
Summer 32.5 30.9 -1.5 -4.8%
Pricing in Competitive Electricity Markets 367

3.2 Total Daily Usage

In addition to the hourly load impacts of the TOU prices, it is of interest to

determine the effect on overall energy consumption. For example, to what
extent are the load reductions during the peak and shoulder periods made up
by load increases in the off-peak periods. Table 2 (above) provides average
daily consumption values by month and for the entire summer for the
treatment and control customers, by day type. Usage was greatest during the
first month, spanning parts of June and July, during which the weather was
the hottest of the summer. Averaged across all days of the summer, usage
for the treatment group declined by about five percent, although this
reduction is not statistically significant due to considerable variability in the
usage changes. Most of the usage reduction occurred on weekdays, where
the treatment group usage was found to be seven percent lower than control
group usage.

4. CUSTOMER DEMAND MODELS

Developing estimates of customers' price responsiveness in the form of

elasticities of substitution requires the specification of a particular functional
form for the underlying customer demand model and estimation of the
corresponding model parameters using data on loads and prices. In the case
of modeling customer demand for electricity by time of day, we have found
it useful to characterize a three-level model for allocating customers'
expenditures on electricity. The first level allocates weekday electricity
usage among time periods that are typically distinguished by different
energy prices. The second level allocates monthly usage between weekdays
and weekends. The third level determines the overall level of electricity
consumption out of total expenditures, or income.
This three-stage demand model can be characterized by the following
indirect utility function:

(1)

where Pp , Ps, Po are prices in the peak, shoulder and off-peak periods (critical
prices are averaged with the peak prices where appropriate), P w is an index
of weekday prices, Pe is the weekend price, PI is an index of overall
electricity prices, P g is a price index of all other goods, which is typically
normalized to a value of unity, and Y is household income. To implement
the model, we must specify a particular functional form for the price indexes
in (1). We first specify the model using the CES functional form.
368 Pricing In Competitive Electricity Markets

4.1 CES Model

As noted above, we have found that the CES form offers a useful
compromise between theoretical attractiveness, and simplicity and
parsimony in parameters. In particular, the three-stage model conveniently
captures in only a few parameters the substitutability of electricity between
the TOU time blocks of a given day, and between weekdays and weekends,
as well as any overall change in electricity consumption due to changes in
the average price of electricity.

Stage 1: Allocating Weekday Electricity Usage. In Stage 1, we specify the

weekday price index Pwas a CES functional form:

Pw = (L8 j Pj -PI) PI (2)

where j runs over the peak, shoulder, and off-peak periods, and l:~ = 1.
Applying Roy's Identity to (2), which represents indirect utility for the
separable commodity weekday electricity consumption, and taking log ratios
of the resulting demand equations relative to the off-peak period, yields the
following equations, which relate usage relative to the off-peak period as a
function of its relative price: vi

(3)

where Q i = In(b, I Do)' G 1 = 1 + PI' and K; and Pi represent kWh usage and
price in time period i. Note that under the standard non-TOU rate, where
prices are the same in all time periods, the ai parameters represent the
logarithm of the ratio of usage in period i relative to the off-peak period.

Stage 2: Weekday and Weekend Usage. In Stage 2, we also represent

the overall price index for total electricity usage, PI> as a CES functional
form:
Pricing in Competitive Electricity Markets 369

~ = (L f3j Pj -P2) P2 (4)

where j runs over the weekday and weekend price indexes, and L..f3J = 1.
Applying Roy's Identity to (4) yields the Stage 2 demand equations, written
in expenditure equation form:

(5)

where a w = In (j3 w/j3e), a 2 = 1 + P2 is the partial elasticity of substitution

between weekday and weekend consumption, and Ew and Ee are weekday
and weekend expenditures.

Stage 3: Overall Electricity Consumption. In Stage 3, we specify the

functional form of the overall indirect utility function V as semi-flex, an
extension of the CES form designed to allow non-proportionate changes in
overall electricity expenditures at different income levels:

(6)

where the 8 and p are parameters, Pg is a price index for all other
commodities (which can be normalized to unity), and Y is household
income. Applying Roy's Identity again yields the following demand
equation, written in expenditure form:

where

Y denotes household income,

Et denotes total electricity expenditures,
Eg denotes expenditures on non-electricity goods (set equal to Y -
E t ),
DTOU is an indicator variable which equals 1 if the household faces
TOU rates and = 0 otherwise, and
370 Pricing In Competitive Electricity Markets

The first level of the model, represented by equation 0), may include
several equations, depending upon how many distinct time periods the
analyst wishes to consider. In the present case, at least two equations are
needed to represent the ratios of peak and shoulder period usage to usage in
the off-peak period. Further refinements can distinguish between the
separate time periods of the day in which prices differ. The parameter OJ
measures the percentage reduction in the usage ratio between any two TOU-
price periods that accompanies each percentage increase in the
corresponding price ratio. Referred to as the elasticity of substitution, OJ
measures the sensitivity of the customer's weekday load shape to TOU
pncmg. In the second level, the parameter 0'2 measures the customer's
willingness to shift usage from weekdays to weekend days in response to
TOU pricing. Finally, in the third level, the parameter 0'3 measures customer
response to any overall change in electricity price under TOU rates. In
general, one would expect 0'3 to be a positive fraction, indicating that a 1
percent increase in the overall price of electricity would induce a less than 1
percent increase in electricity expenditures, due to a reduction in overall
electricity usage.

4.2 Generalized Leontief Model

A potential weakness of the CES model is the restriction of a constant

elasticity of substitution. Two factors suggest that this may be particularly
so in the present case. First, the multiple pricing periods offer the potential
for more complicated substitution possibilities than a simple shifting of peak
and shoulder to off-peak period usage. Second, the load profiles presented
earlier provide strong evidence that customers' usage patterns in the two
shoulder periods differ markedly, suggesting different substitution
elasticities between, for example, peak period usage and usage during each
of the two shoulder periods. An alternative demand model that has proven
useful in TOU studies is the more flexible generalized Leontief.

For the Stage 1 model we specify the following generalized Leontief

functional form:
Pricing in Competitive Electricity Markets 371

pw = [~ ~ 8P1/2p1l2]
L.JL.J I} I } '
(8)
;=1 j=1

where 8 IJ.. 8 JI.. and

to guarantee symmetry and linear homogeneity of Pw (which are two

restrictions needed to satisfy the theoretical requirements of a customer
demand model). Applying Roy's identity yields the following system of
equations:

where K; = usage in period i, and the periods are defined by price levels and
time periods.

4.3 Substitution Elasticities

The Stage 1 elasticities of substitution for the two alternative demand

models are presented in Table 3. Estimates were developed separately for
months one and two, where the first month included all of the critical price
days. The elasticity of substitution for the Stage 1 CES model was
approximately .30 in both months. This is among the highest values that we
have observed in previous analyses of residential TOU programs. Previous
estimates from a range of TOU programs have averaged approximately .17.
The most obvious potential reason for this large response is the presence of
the interactive communications equipment that gave customers the ability to
pre-schedule their air conditioner and certain appliance usage patterns
depending on the time period and price level.
The more flexible substitution elasticities for the GL model vary by time
period. In both months, the strongest load response is between the peak and
off-peak periods, with a value of approximately 040. There is also
substantial substitution between the shoulder and off-peak periods. The
substitution between peak and shoulder periods (.15) is nearly three times
greater during the first month, which included the critical prices, than in the
second month (.06). Finally, the customers facing the somewhat more
steeply priced Rate 6173 appear slightly more price responsive, particularly
372 Pricing In Competitive Electricity Markets

between the peak and off-peak periods, than the other group, which is
consistent with the load response statistics presented earlier.

Table 3. Elasticities of Substitution

Generalized Leontief
Month Time Periods CES Rate 6173 Rate 9122
1 Overall .306
Peak - shoulder .155 .166
Peak - off-peak .395 .356
Shoulder - off-peak .191 .187
2 Overall .295
Peak - shoulder .055 .060
Peak - off-peak .407 .366
Shoulder - off-peak .178 .176

5. CONCLUSIONS

This paper has examined customer price responsiveness to an innovative

residential TOU program in which customers face high critical prices during
periods of high wholesale power costs, and have the ability to use an
interactive communication device to pre-schedule certain major energy-
using devices. In brief, treatment customers facing this rate shifted
substantial load from both peak and shoulder periods to lower priced off-
peak periods. For example, peak period usage reductions on weekdays
during the hottest summer month averaged 26 percent, while reductions
during critical price periods approached 50 percent during some hours. In
addition, elasticities of substitution estimated with a three-stage constant
elasticity of substitution (CES) demand model ranged from .20 to .33. These
are among the highest values obtained in evaluations of previous residential
TOU rates, showing evidence of strong customer price responsiveness in this
unique program.
These findings have several implications for potential suppliers of retail
energy in a competitive electricity market. First, they suggest one way that
prices may be kept competitive, while at the same time protecting the
supplier from occasional unexpectedly high costs. Second, they indicate one
method that suppliers may use to expose portions of their customers' load to
market prices during those times that are most critical. Finally, they suggest
one type of differentiated service, including a bundled offer of the interactive
Pricing in Competitive Electricity Markets 373

communications equipment, that suppliers might use In particular market

niches.

REFERENCES

i See Laurits R. Christensen Associates, "DSM Customer Response, Volume I: Residential and
Commercial Reference Load Shapes and DSM Impacts," EPRI EM-5767, Section 5, 1988
for a survey ofTOU elasticities.

ri A third problem with TOU rates, which was not addressed in this project, has been the
relatively large fixed cost recovery component in both peak and off-peak TOU energy
prices, which implies a substantial difference between the TOU price and expected
marginal cost in the TOU periods. Thus, when customers reduce load during peak
periods, the utility's revenue falls by a greater amount than its cost; and any load increases
during off-peak periods are insufficient to make up those losses. A solution to this
problem of one-part TOU rates is a two-part design similar to that used in typical real-time
pricing programs, in which an access charge is used to collect the fixed costs, and the
energy prices are set closer to marginal costs.

iii The standard residential rate faced by control group customers was approximately $.12 for
the first 600 kWh per month, and $.153 for all additional usage.

iv All of the load impacts cited in this section are statistically significant at the 95 percent
confidence level.

v The critical price signal was sent out for different hours and different lengths of time on each
of the six critical -price days. For example, on four days the critical price was in effect for
only one hour, and on one each of the other two days the critical price was in effect for
either three or four hours.

vi Roy's identity is a standard result from economic theory of consumer demand that enables
the derivation of customer demand equations from a particular functional form for the
indirect utility function. For further details, see any standard quantitative micro-
economics textbook.
Chapter 23

Retail Pricing Tools to Meet Customer Needs

Christopher J. Holmes
UtiliCorp United

Key words: Load Shape Flexibility; Load Shapes; Natural Gas Pricing; Product Mix;
Profitability; Risk Aversion.

Abstract: This chapter describes how a process and software tool developed by EPR! has
been used to successfully design and implement a profitable, commodity based
retail energy product.

1. INTRODUCTION

This paper outlines a number of issues that present significant barriers to

the implementation of competitive pricing strategies to the electric
marketplace. These barriers are identified through analogy. By using a
previously deregulated marketplace, in this case natural gas, and applying
the competitive retail techniques a better understanding of the electricity
market can be developed.
As a starting point, what is discovered are several basic steps that need to
be accomplished before competitive retail pricing can begin. The following
three steps are critical:

1. Establishing the Market Price

./ Access to Markets
./ Verification & Customer Discovery

A. Faruqui et al. (eds.), Pricing in Competitive Electricity Markets

© Kluwer Academic Publishers 2000
376 Pricing In Competitive Electricity Markets

2. Customer Demand

./ Customer Demand - load shapes

./ Price Elasticity - competitor offering & supply alternatives
./ Flexibility of Use
./ A version to Risk

3. Costs of Delivery
./ Loading
./ Transfer Prices
./ Direct vs. Indirect

The establishment of market price remains an unsolved problem in

electricity markets in the Midwest. Wholesale trades are generally bilateral
and are not of a large volume compared to the regulated retail volumes. In
many cases developing retail pricing programs designed to work at the retail
level are constrained by the inability to find publicly traded electricity prices
that can be used by retail customers. Invariably what occurs is the utility is
required to use its actual costs. This presents two problems. First it
eliminates any incentive the incumbent utility may have to obtain
efficiencies and secondly, it opens the utility up to unfair competition by
conveying costs that can be used as target levels by competitors.
In the natural gas industry these problems have largely been solved. The
existence of commodity markets that can be accessed by large and small
retail customers, the existence of gas marketers, and the existence of
customer choice all provide customers and utilities the price information
they need.
The second category of information needs, the customer demand
information, is primarily what follows in this paper. A process will be
developed that combines the previous commodity price information with the
customer data to produce an expected level of profit, participation, and
uncertainty.
The final category, although seemingly inconsequential, can actually
produce a paralysis that is difficult to overcome. The tendency to want to
guarantee the recovery of overheads and the regulatory compact of the past
can lead organizations into a false sense of profitability. Loading and
transfer prices, when not market based, can produce a downward spiral of
profits as competitors undercut the incumbent's price. The only advantage
left is the incumbency and that won't last for long. Many electric utilities
have organized based on functional activity (i.e., generation, delivery, and
services). The natural tendency is to price these services at their embedded
Pricing in Competitive Electricity Markets 377

cost while ignoring the new marketplace. And when that happens, even the
best price and structure is doomed for failure.
These categories are brought up merely as points of consideration. More
thought needs to be devoted to these issues for those forward thinking
utilities that want to not only succeed but to thrive.

2. THE SITUATION

The company proposes to offer a modified interruptible program for

small volume gas customers that are currently regulated. These customers
are offered a tariff that is variable priced based on the price that the utility
purchases for natural gas each month. The company proposes to offer a
fixed price rath~r than a variable price. The only way to do that without
lengthy delays caused by regulatory intervention is to convert these
customers onto a transportation rate. This is a rate that allows customers to
choose their service provider. These two products will essentially be
identical with the exception of a fixed price option and the monthly customer
charge.

3. THE PROCESS

This analysis used a recently developed pricing tool called Product Mix,
developed by EPRI. The product is designed to simulate profitability for
alternative pricing products under condition of uncertainty and competition.
This tool has capabilities for modeling natural gas products.
The following results are driven by two basic assumptions: (1) the
customer's overall price response to energy products, (i.e., gas or electricity),
and (2) the customer's risk aversion for price uncertainty. This approach
will not provide the definitive answer to "optimal" margin design but rather
provide a process that can be used to move towards that optimal level.

4. EXPECTED RESULTS
Figure 1 provides an illustration of the impact of varying per unit margins
on overall profitability. Based on the assumptions used in this analysis, the
optimal per unit margin level is $0.18 IMMbtu, in addition to the margin of
the current small volume interruptible product. The increase in profit is due
378 Pricing In Competitive Electricity Markets

to higher margins driven by customer's preferences for fixed price rather

than variable price products.

Margin Analysis

100.50%.

100.00%

99.50%
~
c::
'§l
"
~ 99.00%
tf-

98.50%

98.00%

.~-rt'-7r.~~~,-n.~~~~rn.-TIr-~-'h-~,-nr~~~9
%

Incremental Markup

Figure 1. Impact Of Varying Per Unit Margins on Overall Profitability

This analysis is based on 333 PNG small volume interruptible customers,

distributed across several different customer types, including apartments,
industrial, churches and schools (See Table 1).

Table 1. Summary Impact from New Product Offering (based on $0. 18/MMbtu incremental
margin)
MMbtu Sales Incremental Margins Market Share

Base Case SVI 15,947,489 $0 100%

After Offering SVI 12,622,923 ($731,405) 79.5%

New Fixed Option 3,286,200 $1,224,335 21.5%

New Total SVI 15,909,123 $492,860 100%

The expected new product sign-up level is 21.5 percent. Sign-up

response that exceeds those expectations indicates that the assumed level of
Pricing in Competitive Electricity Markets 379

customer aversion to price uncertainty has been underestimated. Future

analyses will need to improve the accuracy of this assumption. Should sign-
up response fall below the 21.5 percent then customer level of price
uncertainty was overestimated. Future offerings will need to reduce this
level. Monitoring of the customer response will allow for more precise
margin optimization over subsequent offerings.
In addition, future offerings will require the updating of load weighted
cost of gas to ensure UCU is able to hedge the commitments being made for
these customers. It is recommended that open enrolment periods be
developed for these products that reflect the group of customers, and the
most up-to-date forward prices.

5. ACTUAL RESULTS

Customer sign-ups were 29.4 percent compared to the predicted level of

21.5 percent. This was likely a consequence of the lower then projected
mark-up level predicted by the product mix model. Actual incremental
margin generation was $300,000 compared to the predicted $492,000 based
on the higher mark-up. So, mark-ups were intentionally set lower than
predicted and penetration was higher, but margin generation was lower. The
initial estimate for the lower margin generation may be due to the lower
margin level. It may also be due to the actual size distribution of customers
that signed up for the program. It may be that smaller customers have the
need for more bill certainty and predominated the sign-up levels. This
assumption has not yet been tested.

6. CONCLUSIONS

Because of the uncertainty of the assumptions regarding the per unit margin
level, a more conservative level of $0.15 / MMbtu was used. This provides
additional certainty that customers will participate and should prove to be an
adequate test of this product offering and process for the first time.
This product should be positioned as a product that provide certainty with
energy costs, not bill reductions. The results of this analysis indicate that
customers may actually pay more for their gas service under the proposed
product offering than under the current offering. This is consistent with
market theory since customers should expect to pay more for price certainty.
Therefore, this product should be marketed in a way that emphasizes the
certainty aspect of the customer's bill, (i.e., the lower risk aspect of the
product, rather than the price or bill reduction).
380 Pricing In Competitive Electricity Markets

This example illustrates the process for moving towards a more optimal
level of retail pricing. It does that by setting a level of assumptions for
customer response, load patterns, and risk aversion to form a predicted
outcome. Deviations from that predicted outcome can be evaluated and
incorporated into subsequent offerings.
In this case the offering were based on twelve-month strips or hedge
periods. The expiration of those contracts will require that same analysis.
This will become an ongoing process.
Chapter 24

Pricing Options For The Baltic Electricity Market

Charles F. Zimmermann and Floyd Davis

Bechtel Consulting

Key words: Competition; Electricity Submarkets; Industry Privatization; Market Design;

Power Pooling; Power Sector Liberalization; Regional Cooperation.

Abstract: A comprehensive discussion of the Baltics as representative of many areas

where regional action is required to implement market-based electricity trading
due to the relatively small size of each individual country. Six submarkets are
considered in the context of a Baltic electricity market: (1) bilateral long-term
contracts, (2) a monthly spot market, (3) a daily spot market, (4) ancillary
services, (5) seasonal regulation and annual regulation, and (6) a futures
market. The specific characteristics of each submarket are identified and the
potential for increased competition is evaluated for each. In general, the case
is made that the market is to be favored over regulation and cost-based pricing;
however, limitations to near-term completion in submarkets (4) and (5) in the
Baltics are identified and the steps needed to manage these submarkets in the
absence of liberalization.

1. INTRODUCTION
Competitive generation markets are easiest to establish where there are a
large number of buyers and sellers. The electricity markets in most countries
that have introduced competitive generation markets, such as the United
Kingdom (U.K.), the United States (U.S.), Australia, New Zealand, and
Argentina meet this condition. There are a wide variety of circumstances,
but competition appears to require at least 30 TWh of annual demand.
Seventy-five percent of the countries of the world have lower demand levels
than this. While electricity demand from these countries represents less than
5 percent of the world total, they represent growth areas in the coming

A. Faruqui et al. (eds.), Pricing in Competitive Electricity Markets

© Kluwer Academic Publishers 2000
382 Pricing In Competitive Electricity Markets

decades. Yet, the level of demand is currently too low for these countries to
fully implement competitive models into their power sectors. Furthermore,
the development of regional markets of the size required to support
competition will require greater levels of international cooperation. In this
paper the ways in which the three Baltic countries can introduce greater
competition into their relatively small market is explored. While their
markets have their own unique characteristics, the conclusions are applicable
to many other countries as well.

2. ELECTRICITY SUPPLY AND DEMAND IN THE

BALTICS

Prior to 1991, the three Baltic states of Estonia, Latvia and Lithuania were
part of the Soviet Union and their power systems were developed as an
integral part of the Soviet Northwest Interconnected Power System.
Independence of the Baltic states and the subsequent collapse of the Soviet
Union were accompanied by an interruption of trading patterns and a
dramatic decrease in electricity demand. The three regional companies that
had served the Baltic states as part of the Soviet power system became
vertically integrated state-owned utilities for supply of electricity and district
heating. The generation and transmission equipment operated by these
utilities were designed and sized for a larger regional system with some 110
kV lines crossing back and forth over political boundaries.

Hydro
Nuclear Oil Shale (Estonia)
(Lithuania)
(Lithuania)

Figure 1. Generation Mix- 1998

Pricing in Competitive Electricity Markets 383

Figure 1 shows the generation mix for the three Baltic States for 1998.
The dominance of generation from the Ignalina nuclear plant and from
Estonian oil shale is apparent. Most of the conventional thermal generation
is from combined heat and power plants.
The complementary nature of the systems in the three countries and their
history as part of the same power system has reinforced cooperation in the
power sector and has allowed a degree of regional interdependence. The
region has had a surplus of generating capacity. Domestic demand has been
met with a system capacity factor well under 40 percent. i A regional
dispatch center is located in Riga. Estonia and Lithuania are net exporters
and Latvia is a net importer. Latvian hydro is a major source of load-
following capability. However, national priorities still affect regional
cooperation and operation of the regional system is still dependent upon
neighbors. These issues can be described as follows:

• The future of generation based on oil shale. Estonia is unique in its

heavy use of oil shale. A large portion of the output of six deep mines
and three open pit mines are used in the power sector. Its use has
become viewed by many as an issue of national security. The oil shale
mines and in the power plants contribute significantly to employment.
Estonia will want to protect the oil shale sector, for some transitional
period if not permanently.
• Sharing hydro resources. Hydro is likely to continue to have the
lowest cost of production. Latvia will not view it in its national 3interest
to let its hydro stations sell their generation at a "competitive price level"
if that results in tariff increases to Latvian domestic customers.
• The future of the Ignalina nuclear power plant. The plant provides
Lithuania with the bulk of its electricity needs. However, this two unit
plant has an RBMK design, as are those of Chernobyl, and there is
international pressure to shut down Unit 1 before 2005. While providing
low-cost electricity to the region, the price of its power arguably does
not reflect its true economic cost. Uncertainty about its future has
delayed restructuring decisions in Lithuania. The serious underfunding
of decommissioning and spent-fuel storage costs is a problem to be
faced.
• Continued dependence on Russia for primary fuel supply and
ancillary services. Russia is the sole source of natural gas and oil and is
the only source of RBMK nuclear fuel. The Russian system is one of
the primary sources of system frequency and voltage regulation and
spinning reserve. Nuclear fuel and ancillary service transactions are not
384 Pricing In Competitive Electricity Markets

transparent and have involved complex transactions that include

electricity exports to Belarus and transmission access to Kaliningrad.
While Russia is a critical trading partner, its contractual relationships do
not support market liberalization and transparency.

These issues affect both regional cooperation and power market

liberalization.

3. CURRENT STATUS OF MARKET REFORMS

The three Baltic countries have all made moves toward liberalization of
the electricity market. This has involved a separation of district heating
operations in Latvia and Lithuania and accounting separation of generation,
transmission and distribution operations. Ignalina is a separate entity and
Estonia has announced its intention of privatizing its two largest generating
facilities. Parts of the distribution system have been privatized in Estonia,
Latvia is seeking a strategic investor in its transmission and distribution
systems and Vattenfall owns about 10 percent of Lietuvos Energia, the
operator of the Lithanian transmission and distribution system and some its
non-nuclear generation capacity. Estonia also plans to privatize its oil shale
mining operations.
Estonia is in the first group of Central and Eastern European countries to
begin European Union (EU) accession talks and Latvia and Lithuania have
expressed a strong desire for EU membership. They are making changes in
their power sectors to respond to the EU Directive concerning common rules
for the internal market in electricity. All three countries have a regulator in
charge of electricity, natural gas and district heating pricing. Estonia also
regulates oil shale price. Electricity prices are adequate to cover financial
costs ii and retail prices have been rationalized to reflect the cost to individual
customer classes.
Third-party access to transmission is guaranteed by in Estonia and in
Latvia and transmission prices are published. The Lithuanian regulatory
commission is in the process of developing the methodology for
transmission and distribution pricing. At this time, the transmission system
operators (TSOs) in each of the three countries are part of vertically
integrated companies.
Pricing in Competitive Electricity Markets 385

4. FACTORS THAT FAVOR LIBERALIZATION

In additional to meeting the requirements for EU membership, there are

several possible developments that favor more rapid liberalization of the
power market in the Baltics. First, it is possible that the government of
Lithuania will decide that one unit of Ignalina will be shut down by 2005. iii
This would decrease market uncertainty and allow Independent Power Plant
(IPP) developers and power station owners to invest in generating capacity
expansions on the basis of a "partial shutdown" scenario for Ignalina.
Second, it is possible that the commercial terms of the Estlink Estonia-
Finland interconnection project will be announced soon. This would allow
Estonia-Finland electricity trade. It is technically possible for Estlink
capacity to be increased, to connect the Nordic market (Finland-Sweden-
Norway-Denmark) to the Baltic market.
Third, there are a number of possibilities for IPP development under
discussion. These include a gas-fired combined cycle station in the Kurzeme
region of Latvia, a gas-fired combined heat and power station in Tallinn and
a Kaliningrad combined heat and power station.
Fourth, it is possible that a Lithuania-Poland connection could be
established on a market-oriented basis, either by providing buyers and sellers
with access to transmission capacity or by creating a new company to
purchase electricity from a variety of sources on a competitive basis (for
example, in a daily spot market).
Finally, it is possible that in the next ten years Russia's power sector will
operate on an increasingly commercial basis. This would allow the natural
interdependency of the Russian and Baltic power systems to be a vehicle,
rather than a barrier, to market liberalization. The Russia-Finland, Russia-
Latvia, and Russia-Estonia interconnections could be used to provide the
Baltic countries with access to the Nordic market.
Among the possible developments supporting electricity market
liberalization, the one which gives the strongest incentive for competition
would probably be the announced shutdown of one unit of Ignalina on a
certain date. Although the Estlink project and most IPP projects require
detailed feasibility studies, the shutdown of an Ignalina unit could have a
broader impact on the market because it involves 1300 MW of base load
capacity and because it would have a major effect on the need for ancillary
services such as spinning reserve.
386 Pricing In Competitive Electricity Markets

5. OPTIONS FOR THE FUTURE

Each of the Baltic countries has made commitments to greater power

market liberalization. They are all committed to joining the EU and
participating in the internal electricity market and are seeking greater private
investment in their respective power sectors. However, there are currently
no transmission connections with any EU country and the time for all three
countries to be accepted to the EU could be well over a decade.
The goal of market liberalization is financially healthy, competitive
markets. Competition requires at least two, and preferably more parties.
Therefore, competitive generation markets will be difficult to achieve in any
single Baltic country. There are only two large plants in Estonia and
Ignalina supplies nearly 80 percent of Lithuanian demand. A competitive
market can only be thought of on a regional basis. Latvia being the main
buyer has begun such a market. Its power company receives monthly bids
(with prices and quantities) for energy supplied from Estonia, Lithuania and
Russia.
There is the option of setting up the framework for market liberalization,
creating a favorable market for investment and waiting for EU membership.
The alternative is a more aggressive pursuit of competition which can only
take place through regional institutions. The potential market should not be
limited to the Baltic states. Russia is already a participant and a goal of any
such regional market would be the establishment of cost-effective
interconnections with the Nordic countries and Poland.

6. THE NORDIC MODEL

Because Norway, Sweden, and Finland are nearby and have created the
first competitive international electricity market, it is useful to review the
Nordic experience. The following products exist in this market: iv

• Bilateral long-term trade. This type of trade is given priority in access

to the transmission interconnectors. Bilateral long-trade trade in the
Nordic countries accounts for about 80 percent of annual generation
in GWh.
• Bilateral short-term trade
• Spot markets, including a daily spot market operating on a regional
basis (Nord Pool) and intradaily spot markets operating on a national
basis
• Ancillary services markets. Regulation power and balance power are
traded separately in each Nordel country.
Pricing in Competitive Electricity Markets 387

• Futures markets and other financial trade.

Hydropower reservoirs in Norway, Sweden, and Finland are used for

seasonal exchange but there is no need to create a separate market for energy
storage service because the necessary "price signals" are given by futures
markets and are reflected in bilateral contracts. If there were only one or two
large reservoirs in the Nordel system it would probably be necessary for
these reservoirs to offer energy storage services to electricity market
participants.

7. DEFINITION OF POTENTIAL PRODUCTS IN A

BALTIC ELECTRICITY MARKET

A competitive regional market would require the development of a

number of products. There are at least six products and services, or groups
of products and services that should be considered separately:

• Bilateral long-term contracts. In this type of contract an entity owning

generating capacity negotiates a long-term contract with a purchaser (for
example, a distribution company or a large industrial consumer). "Long-
term" might be defined as more than one year in duration. The Baltic
countries have a long-term multilateral interconnection agreement but
they do not have long-term power sales contracts. v Typically a long-
term contract provides the buyer with a guarantee of a certain amount of
generating capacity, so that the buyer does not need to build that
generating capacity or purchase generating capacity from other sources.
In sales from hydro-electric generation the contract may also guarantee a
minimum amount of firm energy. Both capacity (in kWh) and energy
(in kWh) can be sold in a long-term contract.
• Monthly spot market. This market currently exists in Latvia to govern
imports. Monthly spot contracts are common in the international natural
gas industry but not in the electric industry. Spot markets in electricity
are normally operated on a daily basis, not monthly. In a bilateral
monthly contract the buyer purchases energy for the coming month; he
may also purchase the right to use a fixed amount of capacity in that
month. The terms and conditions of delivery are stated in one-year
contracts that are renewed every year.
• Daily spot market. A Power Exchange can be established to set up a
spot market in which producers submit offers to sell and purchasers
submit offers to buy. Prices are determined for each hour (or half-hour).
Two of the most important examples for the Baltics are the Nord Pool
388 Pricing In Competitive Electricity Markets

market (Norway-Sweden-Finland) and the Amsterdam Power Exchange

(Germany-Belgium-Netherlands). The only commodity being sold is
energy (in kWh) but it has a different price in each hour.
• Ancillary services. This is actually a group of services necessary to
maintain the reliability of electricity supply in the transmission system.
The services include: (1) guarantee of hot (spinning) reserves, (2)
guarantee of cold reserves, (3) frequency regulation, and (4) reactive
power regulation. The first three services are provided by power
stations, while reactive power regulation is normally provided by
transmission system operators with occasional assistance from power
stations if necessary.
• Seasonal and annual exchange. These are services that enable the
wholesale electricity market customer to reduce the annual cost of
electricity generation, when there is a large difference between peak and
off-peak generating costs. These services are not essential to the
operation of an electricity market, but they are desirable. In seasonal
exchanges, the service provider (typically a hydro station with a very
large reservoir capacity) receives energy in off-peak hours or in summer
and delivers it in peak hours or in winter. The wholesale market
customer is not obligated to use the energy storage on a daily or weekly
basis; the customer does not have to purchase peak energy every day or
every week. In annual exchanges the service provider receives energy
from the wholesale market customer in one year and delivers an equal
amount of energy in the following year. v;
• Futures market. Futures contracts are traded in a commodity
exchange, and the contract gives the purchaser the right to conduct a
certain transaction in the future. For example, an options contract may
give the purchaser an option to buy electricity in a certain time period at
a certain price per kWh. It is convenient to have a single Power
Exchange operate both the spot market and the futures market, but this is
not absolutely necessary. Futures contracts can be considered as a type
of financial instrument used to manage the risk of electricity price
fluctuations in the spot market. Without a spot market, there can be no
futures market. Nord Pool operates a futures market.

This is a list of products that can be provided by buyers and sellers other
than the transmission system operators (TSOs). The TSOs provide
transmission services to all of the other participants in the wholesale
electricity market. TSOs may also decide to offer exchange services to
distribution networks or neighboring TSOs. In an exchange agreement
electricity is delivered at one location and received at another location, and
the service provider does not charge a transmission tarife;;
Pricing in Competitive Electricity Markets 389

One of the difficulties with power sector liberalization is the technical

complexity of all these products and services. Simplification should be
encouraged for small markets such as exist in the Baltic states. However,
this has to balanced with providing enough products that the market can give
"price signals" needed to encourage the least-cost and reliable operation of
the system.

8. ASSESSMENT OF THE POTENTIAL FOR

COMPETITION IN EACH BALTIC SUBMARKET

Rather than making a very general statement that the electricity market
should be competitive, it is useful to examine each submarket.

• Bilateral long-term contracts. At the beginning of 1999, Eesti Energia

had a monopoly over export of oil shale generation in Estonia and
Lietuvos Energia had a monopoly over export of Lithuanian generation.
To assure adequate competition, either generating facilities would have
to be empowered to negotiate exports on their own or an additional
player would have to be part of the market. RAO EES Rossii, having a
monopoly over electricity exports from Russia, is one possibility,
although there would be difficulties in negotiating a "normal" bilateral
long-term contract with Russia with capadty payments and capacity
guarantees. It would, of course, be desirable to have a 1000 MW link to
Finland to enable Nordic market participants to compete in the Baltic
market.
Ideally the Lithuanian power station (1800 MW) and combined heat
and power stations in Vilnius, Kaunas and Riga should have an
opportunity to compete with Ignalina. The availability of heavy fuel oil
at low prices (or, in theory, natural gas at low prices) helps the
competitive position of these stations in spot markets but as long as there
is a surplus of generating capacity in the Baltic countries, these stations
will not be able to negotiate substantial capacity payments.
• Monthly spot market. There is already a monthly spot market operated
by Latvenergo, with three bidders - Eesti Energia, Lietuvos Energia, and
an offshore company representing RAO EES Rossii. Ideally there
should be more competitors (for example, separate bids by various
combined heat and power stations) but there is enough competition to
continue the operation of the monthly spot market. Because there are no
hourly prices, the participants in this market must cooperate with each
other by exchanging peak energy for off-peak energy - in effect, by
making an agreement to arrange settlements on the basis of monthly
390 Pricing In Competitive Electricity Markets

bidding and ignore the true cost of electricity generation in each hour.
This activity requires a combination of competition and cooperation
among power systems.
In theory, qualified customers could also participate in a monthly spot
market and purchase energy from the bidders who offer energy to the
monthly market. Latvenergo should not be the operator of such a market
if it were to expand to the entire region because of conflict of interest.
There will be a need for a daily spot market similar to that of Nord
Pool or most other exchanges. When such a market develops, there will
no longer be a need for a monthly spot market. It would not be very
easy for Nord Pool or other European electricity exchanges to set up a
market without hourly prices. Perhaps qualified customers will want to
set up their own electricity exchange - or electronic "bulletin board" - to
support a monthly spot market. In the longer term a daily spot market
will develop and when it does, the need for a monthly spot market will
emerge.
• Daily spot market. For a daily spot market to work efficiently (as a
complement to bilateral long-term contracts) there should be several
buyers and several sellers. The buyers may include distribution
companies, qualified customers, importers, and companies involved in
electricity marketing. Given recent trends in the Baltic countries it
should be possible to open up the market to qualified customers. During
peak hours when hourly electricity prices are high, the spot market will
also have enough sellers to make prices competitive. The main obstacle
to ensuring competition the daily spot market (with twenty-four hour
coverage) is the likelihood that only a small number of generators will
be able to compete during off-peak hours.
One approach to the lack of competition would be to establish market
rules in which off-peak energy is supplied only through the bilateral
contract market. The ideal solution would be to increase the size of the
Estlink connection to about 1,000 MW so that the Nordic electricity spot
market can be extended into the Baltics.
A Baltic spot market could exist side by side with price controls on
Latvian hydro generation (for example, a requirement for the hydro
stations to sell energy to the Latvian transmission system operator at
regulated prices) if this were necessary to gain Latvian participation.
• Ancillary services. The most important providers of frequency
regulation, voltage regulation, and spinning reserve are the Russian
power system, Latvia hydro stations and the Lithuanian pumped storage
station. To be precise, the Russian ancillary service providers are the
various provincial companies, such as Lenenergo and Mosenergo, that
are connected to the high voltage "ring" that connects the Baltic
Pricing in Competitive Electricity Markets 391

countries to Russia and Belarus, and the large regional power stations
connected to this ring. The ancillary services market in the Baltics is not
sufficiently transparent or sufficiently competitive to be operated on the
basis of competitive bidding. However, it should be possible for the
three transmission system operators (the high voltage networks of Eesti
Energia, Latvenergo, and Lietuvos Energia) and RAO EES Rossii to
negotiate contracts and methods of payment for ancillary services. To
simplify the management of ancillary services in the Baltic countries the
power systems might agree to keep the hydro and pumped storage under
the control of TSO, so that the TSOs would be the only electricity
market participants with authority to sign a contract with these stations.
An interesting proposal that has been discussed within Eesti Energia
is the idea of forming a single Baltic TSO - a high voltage network
company that would operate the high voltage network assets of Eesti
Energia, Latvenergo and Lietuvos Energia according to some type of
merger agreement. If an "Independent Transmission Operator" (ITO)
were formed according to this concept, the ITO could act as a "single
buyer" in the ancillary services market. The ITO could have exclusive
authority to negotiate contracts with the Daugava cascade, Riga hydro
station, and Kronius pumped storage station, and the ITO would provide
Latvian hydro-electric generation to the Latvian distribution network
company at regulated prices.
• Seasonal and annual exchange. The agreement between Latvenergo
and an offshore company representing RAO EES Rossii enables
Latvenergo to receive seasonal and annual exchange service from
Russia. Latvenergo may deliver surplus energy to Russia and receive it
at a later date when it is needed by Latvenergo. The contractual
arrangement is not transparent and apparently is currently not subject to
regulatory control, but it is our understanding that it is quite beneficial to
Latvenergo and its consumers. viii Indirectly the contract is probably
beneficial to Estonia and Lithuania; because anything that leads to
higher utilization of the Latvian high voltage network makes it possible
to set lower tariffs for use of that high voltage network.
As long as there is no futures market in electricity in the Baltic
countries, most electricity market participants will not have a serious
interest in information on the seasonal and annual energy exchange
offered by Russia. If a daily spot market were established, however, it
would be possible for a futures market to be established later. The
presence of a futures market would create intense interest in predicting
what is happening with regard to seasonal and annual energy exchange.
The future development of seasonal and annual exchange services in
the Baltics is a sensitive issue because there is a risk that Russia would
392 Pricing In Competitive Electricity Markets

not be able to provide these services in a transparent framework with

clear market rules and regulatory oversight. It would be a mistake to try
to liberalize this market too quickly, because the economic benefits to
Latvia and other Baltic countries would be lost. This is an area where
gradual reforms are needed.

• Futures market. Futures markets are very well established in the

Nordic countries as well as the u.K., the U.S. and Spain. Electricity
futures markets in Germany, Netherlands, and Belgium are likely to
develop rapidly in the next two years. As a result it is reasonable to
assume that there will be no problem attracting buyers and sellers to a
futures market for electricity in the Baltic countries, provided there is a
spot market and a set of market rules that enable physical delivery of the
electric energy in futures contracts. A competitive and transparent daily
spot market is the best foundation for a futures market.

9. CONCLUSIONS

The introduction of power sector competition has thus far been limited to
countries with relatively large power sectors. The Nordic countries have
demonstrated the benefits of international cooperation introducing
competition in power generation. This model has applications to the Baltic
region.
The three Baltic countries are set on a path of power market
liberalization, at least in part due to their desire for EU membership. The
development of a regional power market is the only way the Baltic states can
implement significant levels of competition among power generators prior to
either being fully integrated into the EU internal electricity market or
participating in Nord Pool. The development of such a market can grow
from the cooperative arrangements that already exist amongst the Baltic
countries. If the Baltic countries pursue this more aggressive path of power
market reform, there are two submarkets in which competition is not realistic
and liberalization should proceed gradually in order to preserve currently
favorable arrangements: (1) ancillary services, and (2) seasonal and annual
exchanges. In these submarkets it is perfectly reasonable to propose a
multilateral agreement among power systems.
Under this aggressive path of liberalization, all thermal power stations in
the Baltic countries, including IPPs that are proposed but not under
construction, should participate in bilateral long-term contracts and in a daily
spot market along the lines of the Nord Pool electricity market model.
Although it would not be possible to adopt the Nord Pool model completely,
Pricing in Competitive Electricity Markets 393

in the areas of ancillary services and seasonal and annual exchanges, there is
a need to have Baltic electricity market rules that support bilateral long-term
contracts and a daily spot market.

REFERENCES
i Exports reduce Lithuania's surplus of generating capacity. However. the Lithuanian
government's decision to export energy to Belarus has not been commercially successful
because Belarus lacks the ability to pay.

ii A notable exception is that the Ignalina tariff only covers a portion of future
decommissioning and spent-fuel storage costs.

iii This chapter was written on July 31 and does not reflect subsequent decisions by the
government of Lithuania.

iv Kurt Lindstrom, Director, International Relations, Finnish Power Grid, "Development of the
Spot Market in Finland, 1995-98 - Lessons for the Baltics." Proceedings of a conference
on Gradual Development Options of the Baltic Electricity Market, 7-9 December 1998,
Riga, Latvia.

v Possibly the first long-term contract in the Baltic countries will be a Power Purchase
Agreement such as the proposed contract for the sale of electric energy from the two large
oil shale-fired stations in Estonia.

vi There are several countries where power systems have hydro reservoirs that are large
enough to support annual energy exchanges, but it is unusual for annual exchanges to be
offered as a service to neighboring power systems.

vii Exchange agreements among North American natural gas pipelines are common.
Exchanges of very small amounts of energy and capacity exist in the electricity industry.
There are a few international borders (for example, Russia-Kazakhstan) where billing is
based on the net import or net export for the entire border without regard to power flows at
each interconnection; in this case the interconnection agreement is also an exchange
agreement.

viii The authors have no information whether Russia has negotiated similar agreements with
Eesti Energia and Lietuvos Energia. Because Latvia has substantial hydro generation,
Latvia has the greatest need for seasonal energy exchanges, and perhaps Latvia is the only
Baltic power system with this type of agreement.
SECTION VI

PRICING OF ENERGY SERVICES

Chapter 25

Value-Added Services in a Competitive Electric

Industry

Anne Selting
National Economic Research Associates

Key words: Australia; Customer Choice; England; Value-Added Services; Wales.

Abstract: Before the electric utility industry became competitive, consumers could not
choose who provided their electricity. Now that customers can choose their
electricity supplier, retailers of electricity must find a way to lure customers
away from traditional utility service. They can do so either by lowering the
price of electricity, or by adding products and services to basic electricity
service. These products and services-marketed in tandem with electricity-
are known as Value-Added Services (V AS).

This paper is an introduction to the role of VAS in newly competitive retail

electricity markets. It develops a framework for categorizing V AS, discusses
the economics of selling VAS to different segments of the retail market, and
reviews the role that customer choice and market design will have on the
development of VAS.

1. THE ROLE OF VALUE-ADDED SERVICES IN

COMPETITIVE RETAIL ELECTRICITY
MARKETS

Value-Added Services (VAS) are usually viewed as a minor feature of

competitive electricity markets. The thrust of industry reform has been to
lower the price of electricity by introducing competition into wholesale
generation markets. But V AS are a predictable by-product in any electricity
market that also introduces retail competition. Retail competition-giving

A. Faruqui et al. (eds.), Pricing in Competitive Electricity Markets

© Kluwer Academic Publishers 2000
398 Pricing In Competitive Electricity Markets

end-use customers a choice over who provides their electricity service-

provides three benefits to the market. First, retail competition encourages
retailers to pass through to their customers the lower generation costs
achieved through wholesale competition. Second, retail competition
provides customers with better pricing signals. And, finally, retail
competition stimulates the introduction of new products and services (V AS)
into markets that have lacked innovation under monopoly regulation. For
example, the deregulation of the telecommunications sector is, in part,
credited with encouraging the rapid increase of wireless and wireline
products and services.

Table 1. Types of Value-Added Services

Type of VAS Definition Examples

Complementary Energy Services that the retailer or utility • Metering

Services must provide in order for • Billing
customers to purchase electricity.
· Customer support

Enhanced Energy
Services
Improvements to basic electric
service that:
· Energy efficiency
products and services
1) Lower the customer's cost of
energy,
· Load management &
price forecasting
2) Increase efficiency of
· Risk management
operations, or
3) Improve upon basic electric
· Joint gas/power
management
service by offering additional • Green power
features.

Non-Energy Services i Products that are not related to

electricity consumption but can be
· Home security
monitoring
packaged with electricity service. • Equipment repair

Marketed based on convenience

·• Internet service
Loyalty/reward
(e.g., one bill for electricity and programs
other services). • Telecommunications
products
Pricing in Competitive Electricity Markets 399

2. TYPES OF VALUE -ADDED SERVICES

V AS is more of a catchall term than a precise description. Breaking V AS

into three types of categories helps delineate differences in the products and
services that a retailer may choose to bundle with basic electricity service.
The significant VAS classifications are: complementary energy services;
enhanced energy services; and non-energy services. Table 1 (above) defines
these classifications and provides examples.
Retailer interest in V AS is largely motivated by the economics of
electricity supply and demand. Buying electricity on the wholesale market
for re-sale to end-use customers is generally a high volume, low margin
business. For example, margins on wholesale sales in California are
reported to be approximately 1 mill per kilowatt-hourY EPRI estimates that
the gross margin for retailers who serve large retail customers is comparable
to the U.S. wholesale gas marketing margin of about 1 to 3 percent (EPRI,
1996). Thin margins are also a characteristic seen in more mature
competitive electricity markets such as Australia and England & Wales
(Electricity Supply Association of Australia, Ltd., 1999). While a viable
strategy for some retailers is to pursue a commodity-only business, other
retailers are offering V AS. By marketing VAS, the retailer is able to
differentiate energy offerings and charge a price for "commodity plus"
service. Table 2 (below) contrasts the strategy of selling VAS energy
against marketing commodity energy.

Table 2. V AS Retailing versus Commodity Retailing

V AS Retailing Commodity Retailing

• Heterogeneous product . Homogenous product

• Competition based on price and services, • Competition based on delivery of
premium pricing may apply wholesale energy at the lowest price;
• Customer value is in electricity service pricing at wholesale cost plus retailer
plus solutions to energy management margin
problems • Customer value is in the movement of
• Retailer profit is dependent on providing electrons
customers the V AS they want at a price • Retailer profit is driven by the ability of
they are willing to pay the retailer to arbitrage between the
wholesale and retail electricity markets
400 Pricing In Competitive Electricity Markets

3. MARKETING VALUE-ADDED SERVICES TO

LARGE CUSTOMERS

Customers with large energy demand are the most obvious source of
profit for retailers in a newly competitive electricity market. This is because
industrial and commercial customers account for a substantial portion of
total load in any retail market. iii Thus, by attracting a small number of large
accounts, a retailer can quickly gain significant market share from the
incumbent utility. In addition, relative to small customers, these customers
tend to have low average costs to serve.
The predominant V AS offering in the large customer market is enhanced
energy services, usually in the form of energy efficiency services. For
example, DukeSolutions, a subsidiary of Duke Energy, rejects the lowest-
cost retailing strategy. Instead, the company focuses on building industrial
and commercial relationships in which the retailer helps enhance the
customer's efficient use of energy. The company's key products are: 1)
Assistance with re-engineering manufacturing processes, 2) Monitoring
systems that analyze multi-site energy usage patterns, and 3) Relocation
services that assist customers in taking advantage of rate structures in
different regions.
Large customers appear to be willing to pay for products that decrease
total energy consumption, enable usage of lower-priced off-peak power, or
reduce the transaction costs of buying energy. Because U.S. experience with
retail competition is limited, some of the best evidence available regarding
large customer interest in VAS comes from more mature markets. England
& Wales opened competition to its largest customers beginning in 1990, and
the majority of Regional Electric Companies (RECs) offer a relatively wide
array of VAS.iv A sample of these VAS are provided in Table 3 (below).
Not all V AS are equally important to large customers. A survey of
industrial customer satisfaction in England & Wales asked customers to rank
the relative importance of various enhanced energy services to their
business. The results in Table 4 (below) indicate that the most vital V AS
were those which are most basic: easy access to consumption data, someone
to call in an emergency, and a qualified account executive.
Australia provides additional evidence of the VAS offered in more mature
retail markets. Three states -Victoria, New South Wales and Queensland -
currently allow retail competition for industrial and commercial accounts.
Each state is phasing in competition, beginning with the largest customer
classes. Victoria is the oldest competitive market, having opened retail
competition in early 1994. A recent survey in all three states asked
customers to indicate which V AS they found most important. These were:
Pricing in Competitive Electricity Markets 401

• Energy efficiency, energy audits and energy monitoring

• Information on energy consumption and charge details
• Technical support and advice on equipment maintenance
• Feedback on power failures and interruption reporting
• Market price information and price forecasting
• Power factor correction assistance
• Strategic alliance relationships
• Quick response to outages
• Assistance with negotiating with distributors
• Internet access to metering information

Table 3. VAS for Large Customers in England & Wales v

Exception The retailer notifies the customer if there are any deviations from
reporting expected consumption at a site. vi

Monitoring and Monitoring and targeting equipment pinpoints specific parts of an

targeting analysis industrial system that are causing energy surges.

Day-ahead price Day-ahead faxes of expected half-hourly prices for the following
reporting day. Allows firms to shift consumption to lower-priced periods.'ii

Monthly pool Analysis of pool performance and forecasts for the upcoming
report month.

Contract Retailer shadows a customer's chosen contract by comparing it to

shadowing other options that the customer did not select. Customers may have
an option in their contract to revert to the more cost -effective
contract.

Dedicated account Each industrial customer is served by an account executive whose

executive job it is to be familiar with the user's energy and business needs and
provide technical support.

Energy Energy efficiency programs for heating, cooling, lighting or

management manufacturing processes.

Contract Contract energy management services are limited to only the largest
management energy customers and entail upgrading and reworking the energy
management systems of a firm. It can also include management of
fuel supply contracts, O&M services and project financing.
402 Pricing In Competitive Electricity Markets

Table 4. Rating of V AS by Industrial Customers in England & Wales viii

Service 5 4 3 2 I Average
Crucial highly med. low not score
important important important important
Demand 43% 30% 9% 7% 11 % 3.86
Data
24-hr help 38% 30% 16% 12% 5% 3.84

Expert acct. 16% 27% 27% 23% 7% 3.23

mgr.
Load 11% 32% 25% 16% 16% 3.07
mgmt.
Energy 3% 37% 34% 9% 18% 2.98
Effie.
EDI 13% 18% 32% 22% 14% 2.94

Consulting 0% 18% 34% 28% 20% 2.51

While it is relatively early in the development of U.S. retail markets,

V AS appear to be an important component of retailer offerings. One
national retailer recently acknowledged that while it sells both commodity
energy and enhanced energy services, 90 percent of the company's revenues
are obtained from ten percent of its V AS business.
A 1998 study provides some insights into the VAS that are demanded by
large, non-industrial customers. Golove, et al. analyzed twenty solicitations
made by U.S. commercial and industrial businesses, government entities and
non-profit agencies. Four-fifths of the solicitations were from non-industrial
customers. The authors expected that commercial customers would largely
request "commodity only" service. Instead, the findings indicated that over
60 percent of the solicitants included requests for services that the authors
referred to as "commodity plus," such as tariff analysis, risk management,
infrastructure development, and energy efficiency services. About 65
percent of the solicitations requested assistance in ancillary services,
metering or billing. ix
Risk management is another V AS that some large customers demand. A
distinct feature of electric power is that economically viable storage does not
exist. This results in significant wholesale price volatility, particularly
during periods of scarcity. For example, it has been estimated that the
standard deviation of the percentage changes in the average daily price of
Pricing in Competitive Electricity Markets 403

electricity at Palo Verde, Arizona from August 1996 to October 1998 was
19.74 percent. In contrast, the standard deviation of the daily returns on the
S&P 500 index during October 1987 was 5.73 percent (Bessembinder and
Lemmon, 1999). This statistic underscores the potential for risk
management. A retailer may offer customers Contracts for Differences
(ctDs) or fixed price contracts and charge a risk premium for its services. x
Because V AS are most profitable in the large-customer segment of the
market, competition among retailers is keen. In the u.K., there are at least
twenty-one retailers that are licensed to serve large customers. In Australia,
a total of twenty-five retailers are active in the states that have introduced
competition to the industrial and commercial segments (Electricity Supply
Association of Australia, 1999). In the U.S., retailer entry into electricity
markets has also been significant, despite the fact that retailers can only
operate in the few states that have retail competition. At least five national
retailers are aggressively marketing to large customers in California. xi There
are at least ten retailers with sizable revenues that are marketing to large
customers on a national or regional scale. Despite complaints from retailers
that a low standard offer rate limits their opportunity to compete for
customers in Massachusetts, over twenty companies are licensed to serve
industrial customers in the state. xii And in Pennsylvania, nearly fifty retailers
are vying to win industrial and commercial accounts in PECO territory
alone. xiii
While retailers are uniquely placed to provide both commodity energy and
energy management services, a number of companies already specialize in
solving customer's energy-related problems. xiv In fact, national retailers
indicate that they consider their rivals not just other retailers but also
building and manufacturing control companies. xv Similarly, retailers are not
the only entities capable of providing risk management services. Banks,
insurance companies, credit-card companies and other entities that
understand commodity markets can offer risk management services. Unless
retailers are able to offer energy and risk management expertise at
competitive prices plus low-cost commodity energy, customers will have
incentives to unbundle these services from their electricity purchases.

4. MARKETING VALUE-ADDED SERVICES TO

SMALL CUSTOMERS

On a dollar-per-kilowatt basis, the cost to serve large customers is much

lower than the costs to serve residential and small commercial customers. In
particular, retailers serving small customers incur a range of costs to attract,
retain and serve customers. Some of these expenses include marketing,
404 Pricing In Competitive Electricity Markets

administration and customer account costs. xvi Wiser et al. (1998), for
example, estimates that the costs to simply acquire a residential customer
during the early years of competition are about $100 per customer. Because
of these high costs, it is difficult to develop service packages-VAS or
otherwise-that are attractive to customers and profitable for retailers.
The economics of serving small customers has, at least in the short run,
limited the number of retailers that serve residential customers. In
California, over 100 retailers initially registered with the California Public
Utilities Commission, a pre-requisite to entering the mass market. Today,
less than a dozen retailers are active in the residential market. Enron, one of
the first retailers to enter the California market, spent $10 million in 1998 on
marketing and advertising. The company signed up 30,000 customers out of
approximately 8 million residential accounts before it terminated service to
these accounts in April 1998. xvii Having spent approximately $333 per
customer on recruiting costs alone, the cOfllpany announced that the
California mass market was not profitable. xvlll Massachusetts has also
experienced the departure of prominent retailers. Xenergy, Inc., one of the
first energy service companies to offer competitive services, announced in
February 1999 that it would exit the Massachusetts market, in part due to
low utility standard offer service rates that make it difficult for retailers to
compete. xix
Retailers have yet to identify a successful combination of services that
attracts residential customers in numbers significant enough to profitably
serve this market. xx But surveys tend to indicate that residential and small
commercial consumers have interest in some VAS. For example, a national
survey found that nearly half of all customers polled said they would be
interested in new services and billing options from their electricity provider
(RKS Research and Consulting, 1999). The V AS that small customers said
they want included:

• Service guarantees that compensate customers for non-performance

• Customer flexibility over billing date
• One bill for all utility services
• Automatic outage notification
• Whole-house surge protection
• Time of use metering
• Energy auditing

While customers polled indicate that they want these services, they may
not be willing to pay the premiums that retailers must charge to profitably
serve this segment of the market. Until costs to serve small customers
Pricing in Competitive Electricity Markets 405

decline, there is likely to be a gap between customer interest in V AS and

retailer offerings of V AS.
The Internet is one vehicle that may reduce the high cost to serve small
customers and thereby introduce more robust VAS retailing. At least one
company has begun efforts to reduce costs by serving customers a modest
discount on electric service over the Internet. Utilty.com is a California-
based retailer that sells electricity and interacts with customers exclusively
over the Internet. Customers sign up on the Internet and receive their bills
and address customer service issues via e-mail. By avoiding the costs such
as a 24-hour customer call center, the average costs to serve small customers
are lowered. From the consumer's perspective, Utilty.com's package may
offer benefits beyond those available through standard utility service. xxi

5. THE ROLE OF PRICE IN CUSTOMER CHOICE

OF VALUE-ADDED SERVICES

The previous sections discussed the dynamics of selling V AS into retail

markets and the types of V AS that interest large and small customers.
However, it is also important to consider the potential of V AS within the
context of what drives customer choice of a retailer. Because V AS are
priced at a premium to commodity energy, a fundamental question is
whether customers want the lowest-priced commodity service or a VAS
package.
Research indicates that consumers are inclined to select a retail electricity
provider based largely on price. In the U.S. and England & Wales, the larger
the customer, the more pivotal is the price (EPRI, 1996 and 1998).xxii
Similar results can be gleaned from Australia. A November 1996 survey
asked business customers to indicate which of six attributes-ranging from
price to the business acumen of the retailer-mattered most in selecting a
retailer. Approximately 90 percent of respondents ranked price as their
single greatest consideration (Australian Chamber of Manufacturers, 1996).
The lesson in these results is that industrial or large commercial
customers will select an energy supplier based on the lowest cost of
electricity. If a retailer provides an energy efficiency package along with
commodity energy and this package has the lowest costs (on a net present
value basis), then evidence suggests that the customer will select this offer
over a higher-cost commodity-only contract. But the VAS package must be
the cheaper alternative relative to commodity-only options.
Some retailers have developed small customer strategies under the
assumption that this customer segment is more amenable to purchasing
higher cost V AS electricity packages. The bulk of survey research indicates
406 Pricing In Competitive Electricity Markets

that the higher the price discount, the more likely residential customers are to
leave utility service (see for example Cai et al.). For example, in England &
Wales, which recently completed phase-in of choice for 26 million
households and small businesses, approximately 83 percent of customers
who have switched providers indicate that price is their main consideration
(OFFER, 1999). But for small customers, non-price attributes of service,
particularly reliability, V AS, and the reputation of the supplier do matter. xxiii
Small customers' apparent willingness to consider non-price attributes of
service encouraged U.S. retailers to market green power at prices above
standard utility service. One of the reasons retailers targeted green power-
to the exclusion of other possible V AS-is that most research indicates that
U.S. households strongly support renewable power options. Farhar and
Houston (1995), for example, found that renewable energy and energy
efficiency programs were highly valued by a plurality of Americans.

100

• $90.89
• $85.89
• $00.89

70 • $70.14 ,
• sro.39 • $67.54 $65.39 • $66.1
• $61.93 • $63.29 I
00 • $58.70 • $00.77 • $00.89 !

"Sl.29 •$54.77

40

- PG&E Utility Service • GreenAlterrntive

Figure 1. Monthly Costs of Green Power Versus Standard Electric Service Line Break
Pacific Gas and Electric Territory

But on the green power front, survey responses have diverged

substantially from actual customer choices. The results in the California
market, for example, indicate that a majority of residential customers are not
willing to pay a premium to receive green power. Industry experts largely
believed that green power opportunities would have the best success rate in
California, a state known for its strict environmental standards and strong
support for renewable energy programs. But as of April 1999, about 89,268
customers, less than 1 percent of all residential customers, had changed
Pricing in Competitive Electricity Markets 407

suppliers. xxiv Figure 1 (above) benchmarks the monthly total bill for green
power service against the costs for basic utility service. xxv The majority of
the fifteen renewable energy offerings are between 7 and 56 percent above
the standard utility service offered by Pacific Gas & Electric. xxvi The
experience in Californiaxxvii is important as it underscores the importance of
price, even among smaller customers who were predicted to be receptive to
higher-priced VAS offerings.

6. THE ROLE OF RETAIL MARKET DESIGN IN

VALUE-ADDED SERVICE MARKETS

A final factor that influences VAS penetration in newly competitive

electricity markets is regulatory policy. During the phase-in of competition,
U.S. regulators have tended to support rate freezes and, for at least some
customers, rate reductions. For example, in California customers' rates are
frozen at their 1996 levels for roughly four years.xXviii In Pennsylvania,
restructuring legislation mandates that rates are frozen for fifty-four months
or until utilities recover all stranded costs. xxix Finally, in Massachusetts,
rates are capped at the standard offer value, which is set on an annual basis.
Customers that remain with the utility pay the same standard offer rate,
regardless of wholesale market prices. The manner in which these policy
objectives are achieved has significant implications for the success of V AS
in the early years of retail competition.
The result of these market rules is that customers that remain with the
utility are provided with "built in" risk management services as a component
of their existing service. xxx Given that the utilities are de facto providing risk
insurance through rate caps, the retail market for fixed price contracting and
other risk products during the transition is substantially weakened. xxxi
Rate reductions also have implications for V AS penetration. In addition
to caps on overall customer rates, many states have also introduced short-
term rate reductions for some or all customers. The amount and timing of
these reductions vary across states, but their impact on V AS markets is
uniformly negative. Rate reductions make a retailer's VAS products look
expensive because mandated rate reductions widen the price differential
between utility service and premium-priced VAS. xxxii
408 Pricing In Competitive Electricity Markets

7. THE FUTURE OF VALUE-ADDED SERVICES IS

UNCERTAIN

u.s. retail markets have not been open long enough to draw absolute
conclusions about the role of VAS in electricity competition. But evidence
suggests that large customers are willing to pay for VAS that increase energy
efficiency, simplify billing and payment, and allow companies to take
advantage of off-peak pricing. Large customer accounts have left utility
service for retail providers, even in markets in which the price discounts
(relative to utility service) are 10W.xxxiii This suggests that customers find
value in the services that retailers are providing. But large customers are
also price conscious and select a retailer whose package supplies energy at
the lowest overall cost. Because the revenue potential is strong, VAS
retailers that can also deliver low-cost commodity energy service will
continue to aggressively court this market segment.
Retailers have yet to tap the potential of the mass market. This speaks
not so much to the lack of strategic creativity of retailers but instead to the
likelihood that the economics of mass market retailing impose some
significant constraints on new entrants. Retailers with a national presence
have been tentative to enter these markets. The difficulty with offering
products to households is that they are largely untested in the market,xxxiv and
the lukewarm reception of renewable energy offerings has likely caused
retailers to reconsider their small customer strategies. xxxv However, there is
generally a lag between the introduction of competition and small customer
understanding of the choices available in the new market.xxxvi
Finally, the future role of VAS in competitive markets cannot be fully
assessed until the transition period ends and rate reductions and price caps
are removed. At present, these constraints are significantly affecting the
development of VAS markets, particularly in the area of risk management
services.

NOTES

i Non-energy VAS are not explored in detail here. These VAS generally consist of product
bundles that are offered to smaller customers. For example, a retailer may contract to
provide a household with electric service, cable and Internet access. Whether a retailer
can be successful in garnering business in non-traditional areas is largely a function of
three factors: 1) How well suited the services are to the retailer's core business, 2) How
price competitive the service is relative to unbundled options, and 3) How much customers
truly value and are willing to pay more for the convenience of bundled services. The most
successful VAS bundling to date is in England & Wales where British Gas provides
Pricing in Competitive Electricity Markets 409

combined electricity and gas service as a competllIve retailer. As of June 1999, the
company has acquired half of all competitive customers under 100 kW (OFFER, 1999).

ii Pacific Gas & Electric panelist at "Electric Industry Restructuring: A Research Conference,"
March 5,1999.

iii For example, in California there are over 9.7 million residential and small commercial
customers. These customers account for 42 percent of contestable load in the state. But
just 12.300 large customers account for nearly 55 percent of the electricity load in this
market. (The balance, about 3 percent, is agricultural load.)

iv Similar to US utilities, RECs are the incumbent utility that serves customers within a
designated franchise territory. But any REC may also be licensed to sell to competitive
services as a retailer outside its own territory.

v Source, MarketLine, 1996.

vi Exception notices help customers control energy usage and rectify equipment problems
before large electricity costs are incurred. At least one REC takes half-hourly
consumption data from the meter, compares the data to expected usage and highlights any
deviations over 10 percent. Other RECs provide monthly summaries of exceptions.

vii These reports are often provided in conjunction with triad warnings. In the U.K., a portion
of the customer's annual transmission charge is determined by his consumption during the
highest three half-hour load periods of the year. If a company can anticipate a triad period,
it can lower its transmission charge, sometimes substantially.

viii Source: MarketLine, 1996.

ix It is important to note that this study was based on what customers requested as part of the
bidding process. These requests may be very different from the services that customers
ultimately purchase.

x The value of fixed price contracting has not been demonstrated. Although multi-year
contracts between retailers and large customers were a feature at the beginning of the
California market, the long-term supply or demand for multi-year contracts is questionable
because of they are risky for both parties. The relevant hedging question for most large
consumers is whether average electricity prices will change enough from month to month
to make it worthwhile paying a premium for a fixed price contract of shorter (e.g. one
year) duration. Even annual fixed-price contracts may not last very long. Consumers may
soon learn not to buy them unless the fixed price is low, and retailers may find that they
can lose a lot of money competing to sell low-fixed-price contracts. It is possible that
fixed price contracting will be a low-margin, low-value added service.

xi Retailers with major industrial or commercial accounts include PG&E Energy Services,
New Energy Ventures, Enron, Sempra Energy, Edison International, and Commonwealth
Energy.
410 Pricing In Competitive Electricity Markets

xii Source: http://www.statc.ma.us/dpulrcstructJcompanv.htm

xiii Not all retailers operate in all regions of the state. PECO is provided here as an example as
it has some of the most robust activity to date. Source: http://www.electrichoice.com/c&i-
peco.htm!

xiv One example of new products and services being offered by entrenched rivals is Unity
Systems. The California company recently announced an Internet service that provides
commercial customers with temperature control plus on-site energy management
information including energy usage data, equipment run-time, outside air temperature and
energy rates.

xv For example, Exelon, a competitive affiliate of PECO Energy has been active in east-coast
markets and identifies its key competitors as being Enron, PG&E Energy Services, Duke
Energy Trading, Johnson Controls and Honeywell.

xvi For example, the retailer must send a bill to customers every month. If the cost of the bill
is $5, irrespective of customer size, a retailer who serves ten large customers will have
lower total billing costs than the retailer that serves 100 small customers. Assuming that
both retailers sell the same volume of energy, the retailer serving small customers will
have higher billing charges per each kilowatt sold.

xvii Enron continues to aggressively court large customer accounts.

xviii San Francisco Chronicle, April 23, 1998.

xix "Low Standard Offer Rates Drive Xenergy from Business," Electric Utility Week, February
8,1999, p. 5.

xx The exception to this may be Pennsylvania, where switch rates in the residential market
have been relatively robust. For example, in PECO territory, about 12 percent of
residential customers have left the utility. (Source:
http://www.state.pa.us/P A_Execl Attorney _ General/ConsumecAdvocate/elecomp/eleccha
rt.pdf). On closer inspection, however, it is not necessarily a winning retailer strategy that
accounts for high switching levels but instead a regulatory determined shopping credit that
has been set above the market price for power. Thus, the design of the market provides
retailers with a subsidy to sign up customers. This is distinctly different from the
dynamics that exist in a market in which competitors must legitimately woo customers
away from utility service by offering lower prices or better products.

xxi As of June 1999, the standard plan offers: 1) Five (5) percent off the customer's energy
component of the bill; 2) Bill payment with credit card or automatic withdrawal using the
web-based customer center; 3) Flexible billing days; 4) Balanced billing in which the
customer pays a flat, average rate each month; 5) Green power at no extra charge; 6)
Retailer donation of 5 percent of company profits to charities chosen by the customer; and
7) The ability for the customer to monitor current energy consumption online.
Pricing in Competitive Electricity Markets 411

xxli An EPRI cross-sectional study of twenty U.K. and Norwegian industrial customers
concluded that although the successful contract is a function of three factors-price,
services and contract length-price is the only real factor that influences the choice of
provider. Customers surveyed for the study indicated that VAS are at best only a marginal
component of their choice of supplier. As noted by a representative of Marks & Spencer,
the U.K.'s largest clothing, home furnishings and food retailer, the criteria for choosing a
power supplier are "price, price, and price-it's as simple as that" (EPRI, 1996, p. 31).

xxiii For example, a survey of households in England & Wales found that 25 percent of
residences indicate that the identity of their prospective retailer is the most important
factor in their decision to switch electricity service. Approximately 18 percent of
customers considered price to be the dominant driver. The same percentage of consumers
said that the products offered by the retailer are the main consideration
(PriceWaterhouseCoopers, 1998). Results in the US are similar (EPRI, 1998).

xxiv The number of customers purchasing green power is not tracked by state regulators, but it
is clear that even if green power retailers accounted for all the switching households,
customer interest is low.

xxv Based on average monthly consumption of 500 kWh per month. The analysis, provided by
the California Office of Ratepayer Advocates, assumes a baseline allowance of 324
kWh/month and a PX pool price of 2.4 centslkWh. (Customers who remain with the
utility are charged for energy based on the PX price.)

xxvi A total of eight retailers are currently providing 15 different green power packages. Prices
differ based on the "purity" of the green power (e.g., the ratio of system power to green
power) and the source of the green power (some resources such as wind are considered
"cleaner" than other green resources such as power supplied from large hydro facilities).
The two products that are priced close to utility service are the result of a recent effort by
Commonwealth Energy to discount green service. The California-based retailer conducted
focus groups and found that 90 percent of customers identified themselves as
environmentalists who support renewable energy. But when presented with premium-
priced green power options, the vast majority of customers indicated that they would not
pay more for renewable service. On the basis of these findings, Commonwealth Energy
recently revamped its marketing strategy and is offering green power at a discount relative
to utility service. One of its competitors, Cleen 'n Green, soon matched these discounts.
Customer discounts are possible in the short-term because of a state-sponsored subsidy for
renewable energy. The subsidy is scheduled to end in March 2002.

xxvli Other U.S. markets have also seen green power offerings. But retail competitIOn is
generally too new in these regions to draw strong conclusions. While Massachusetts
began retail competition at the same time as California, the standard offer rate has been set
below actual generation costs. Low standard offer rates have limited any residential
competition, making it difficult to draw conclusions about the relative popularity of green
versus non-green programs. Retailers are offering green power in Pennsylvania, but the
market has only been open five months.
412 Pricing In Competitive Electricity Markets

miii The rate freeze may end sooner if utilities recover their stranded costs before March
2001.

xx"' Specific rate freeze periods have been negotiated as part of utility settlements.

xxx For example, in California, customers who remain with the utility pay generation based on
the wholesale price of power in the PX. However, if the PX price exceeds the unbundled
generation rate, the utility, not the customer pays the difference. In Massachusetts,
customers who remain with the utility are charged a flat rate that does not vary with the
wholesale price of power.

xxxi The opportunities to sell risk management services are likely greatest in the large customer
market. Electricity bills are a small part of household expenditures and, moreover, the
energy component of the customer bill is generally less than one-third of the total bill. On
the other hand, small customers tend to be more risk averse. Few risk-based products are
available for small customers. One company, WeatherWise offers a WeatherProof bill
that allows small customers to pay a pre-determined, fixed bill for heating and cooling
needs, regardless of the weather. This product has only recently been available, and
customer demand cannot yet be assessed.

xxxu This is particularly true in Massachusetts, for example, where utilities must offer standard
offer service for seven years. In the early years of the market, the standard offer rates are
below the true market price for power. It has been estimated that the current standard offer
rate, which averages about 3.5 centslkWh, would need to be closer to 4 centslkWh to
enable retailers to compete in residential markets. The standard offer rate will not rise to
this level until 2002, when the standard offer rises to 4.2 cents/kWh. "A Year Later,
Residential Customers Still Looking for Deregulation Gains," Utility Spotlight, March 9,
1999, via ENERGY CENTRAL (http://www.energycentral.com)

xxxiii For example, in California most large customers are receiving discounts of 2 to 5 percent
off the generation component of their bill (Wiser et aI., 1998). Despite these low savings,
46 percent of California large commercial and industrial load has left the investor-owned
utility.

xxxiv Large customer appetite for energy efficiency, for example, was not unknown because of
the history of demand-side management programs in the US.

xxxv For example, Commonwealth Energy's introduction of discount green power followed its
initial attempts to match the utility's standard tariff price and then sell customers an energy
efficiency device called the Power Planner. (The Power Planner plugs into a home
appliance and regulates the amount of energy drawn by the appliance's electric motor.)
Commonwealth sells the device for around $50 and claims a reduction of energy
consumption of up to 20 percent. The California Energy Commission has challenged
these savings. While the company managed to sign up roughly half (45,000) of all
customers that switched in the California market after the first year, these penetration rates
are not strong and likely contributed to the company's shift in strategy, which it announced
in the spring of 1999.
Pricing in Competitive Electricity Markets 413

"",,,vi Kushler (1998) reviewed public awareness in states moving rapidly toward retail
competition and concluded, 'The available evidence suggests that the public knows very
little about electric industry restructuring." (p. 31). In California, focus groups conducted
with the general population six months before the scheduled start of the market found that
most customers "do not have a clear understanding of what is going to happen. Even those
who know that change is coming typically have a misperception of what will actually
happen (CPUC, 1997).

REFERENCES

Bessembinder, Hendrik and Michael Lemmon, "Equilibrium Pricing and Optimal Hedging
in Electricity Forward Markets," 1999 Power Electricity Industry Restructuring: A Research
Conference, Program on W?rkable Energy Regulation, University of California Energy
Institute, Berkeley, California, April, 1999.

Cai, Yongxin, Iraj Deilami and Kenneth Train, "Customer Retention in a Competitive
Power Market: Analysis of a 'Double-Bounded Plus Follow-ups Questionnaire'''. The Energy
Journal, Vol. 19, No.2 (1998).

Customer Feedback on Victoria's Competitive Electricity Market: A Report on the ACM

Survey of Contestable Electricity Customers, Australian Chamber of Manufacturers,
November 1996.

Electric Lead: Dynamism and Change in an Openly Competitive Electricity Market,

PriceWaterhouseCoopers Study, 1998.

"Survey of Electricity Customers in New South Wales, Victoria and Queensland,"

Electricity Supply Association of Australia, Ltd., 1999.

EPRI, Predicting Customer Choices Among Electricity Pricing Options, EPRI TR-
108864-V2, Palo Alto, California, April 1998.

EPRI, Is There Value in Value-Added Services? Evidence from Competitive Markets in

England, Wales and Norway, EPRI TR-106195, Palo Alto, California, February 1996.

Farhar B., and A. Houston, "Willingness to Pay for Electricity from Renewable Energy,"
1995, available at http://www.eren.doe.gov/greenpower/willing.html

Golove, William, Charles Goldman, and Steve Pickle. "Purchasing Power and Related
Services: A Window into Customer Preferences," The Electricity Journal, January/February
1998, pp. 17-23.
414 Pricing In Competitive Electricity Markets

Kushler, Martin "Restructuring and 'Customer Choice': Vox Populi or Dictum

Dictorium?" The Electricity Journal, January/February 1998, pp. 30-36.

MarketLine International, Ltd., Sales and Marketing in the U.K. Competitive Electricity
Market, London, U.K., 1996.

Office of Electricity Regulation (OFFER), Review of Domestic and Small Business

Electricity Supply Price Regulation: A Consultation Document, June 1999.

Wiser, Ryan, William Golove, and Steve Pickle, "California's Electric Market: What's in
it for the Customer?" Public Utilities Fortnightly, August 1998, pp. 38-45.

"Who's Who Among Energy Service Providers," Public Utilities Fortnightly, October 1,
1998, pp. 50-58.
Chapter 26

Measuring How Customers Value Electricity Service

Offers

Lisa Wood, Suzanne Gambin, and Patricia Garber

PHB Hagler Bailly, Inc., PHB Hagler Bailly, and EPRI

Key words: Customer Preferences; Market-Based Pricing; Market Research; Value-added

Services; Willingness-to-Pay.

Abstract: In a competitive environment, where retail customers are free to choose

electric service offers according to their preferences, cost-based pricing is
inappropriate. Market-based pricing involves understanding customer
preferences and offering products that match these preferences at prices that
provide value to customers. This paper presents the willingness to pay (WTP)
methodology as an approach for measuring how customers value electricity
service offers beyond commodity electricity. While the WTP methodology is
clearly a useful tool for designing and pricing electric service offers, we
suggest it also yields powerful information for the critical decisions that must
be made prior to the development of an offer--namely, to determine strategic
positioning in the marketplace.

1. INTRODUCTION

With the introduction of retail competition, electric utilities are shifting

from cost-based pricing to market-based pricing. In a competitive
environment, cost-based pricing is no longer appropriate because retail
customers will have different preferences for electricity service offers and
the associated willingness to pay (WTP) for these offers will vary. Market-
based pricing involves understanding these customer preferences and
offering products that match these preferences at prices that provide value to
customers.

A. Faruqui et al. (eds.), Pricing in Competitive Electricity Markets

© Kluwer Academic Publishers 2000
416 Pricing In Competitive Electricity Markets

In this paper, we present an approach for measuring how customers value

electricity service offers beyond commodity electricity. First, we provide a
brief overview of WTP, its relationship to customer value, and its usefulness
as a tool for measuring value for goods not (yet) traded in private markets.
Next, we discuss how to structure a research study to measure WTP and
customer preferences for an extensive set of hypothetical electricity service
offers. Finally, we offer insights on how to apply WTP findings to assess
market opportunities and determine strategic positioning in the marketplace.

2. THEORETICAL BACKGROUND OF WTP

According to welfare economics, the benefit to an individual of a service

or a product is defined as that individual's maximum willingness to pay
(WTP) for the service or product. To illustrate individual WTP for a bundle
of electricity services, assume that well being (also known as "utility")
depends on your income and services from electricity. Assume that an
electricity supplier offers services that move your overall well-being from a
specific electricity services (ES) state (e.g., say bare bones or commodity
electricity) to an improved state of electricity services (ES*). As shown in
Figure 1 (below), WTP is the maximum amount of money that you would
pay (Yo - Y 1) for the electricity service offer that moves you to the improved
state while remaining at the same level of overall "well-being" or utility. In
this example, WTP is a measure of how much an individual values a
particular bundle of electricity services. This varies by individual and
depends on willingness and ability to trade money for electricity services.
In a private market, this translates into individuals choosing electricity
service offers only when their WTP for the improved state is greater than or
equal to the price of the offer. Hence, in a private market, price is a lower
bound on an individual's WTP. At each price level, the number of
individuals that will choose a specific electricity service offer is equal to the
number whose maximum WTP is greater than or equal to the price.
Therefore, for most individuals that purchase a product, the price is less than
their maximum WTP (for some individuals, the price will be exactly equal to
their WTP).
For goods traded in private markets, WTP can be observed directly from
consumer purchase behavior. However, in the case of the evolving
deregulated electricity market in the U.S., a combination of factors - offers
that are not widely differentiated, unawareness on the part of consumers, and
low interest - make the observation of consumer purchase behavior difficult.
Hence, although some states have been deregulated, sufficient variation in
offers as well as low awareness makes it difficult to utilize this information
Pricing in Competitive Electricity Markets 417

to understand how consumers will value different electricity service bundles

in the longer term. Fortunately, indirect methods are available for estimating
WTP for goods not yet traded in private markets (e.g., hypothetical goods) or
goods not ever traded in private markets (e.g., public goods).

U = Utility ~_-U(ES*)

U(ES)

Yo Y =Income
I
Figure 1. WTP (YO-Y 1) is the Amount an Individual Would Pay to Move to an Improved
State of Electricity Services (ES *)

3. IS ELECTRICITY A COMMODITY?

Consumers of electricity as well as other products require solutions that

meet their needs at prices that are less than or equal to their WTP. In this
paper, we refer to all consumers of electricity - residential customers,
commercial customers, and industrial customers - as the retail market or
retail customers. While it is true that some retail customers will only choose
electricity services at the lowest possible price, this is only one segment of
the retail market. We refer to these customers as the "commodity market
segment." Recent research suggests interest in and WTP for other electricity
service offers. This implies the existence of other market segments beyond
the commodity market segment.
To-date, most of the offers available (in Pennsylvania and California)
have assumed the existence of only two segments - a commodity market
segment and a green segment. If other customers show a positive WTP for
418 Pricing In Competitive Electricity Markets

offers beyond commodity electricity or beyond green energy, what does this
imply for retail marketers? As demonstrated in other recently deregulated
industries (most notably airlines and financial services), companies can
differentiate among customers based on their preferences and charge a
premium to supply services that customers value. For electricity markets,
Selting and others refer to these services as value-added services (V AS).i
Selting defines three types of energy services - complementary energy
services such as billing and metering; enhanced energy services such as
energy efficient products and green energy; and non-energy services such as
security and telecommunications products.

Figure 2. Risk Free Supply

In a recent study sponsored by EPRI, in a focus group setting, we

organized a "power shopping" exercise for a group of commercial and
industrial (ell) customers and asked them to indicate the attractiveness of a
set of hypothetical electric service offers. Specifically, we asked customers
to allocate ten dots (or votes) to the electricity service offers that they liked
best; they could allocate as many of their dots as they wanted to any
particular offer and they did not have to allocate all ten dots. For illustration,
Pricing in Competitive Electricity Markets 419

Figure 2 (above) shows the poster descriptions for three of the offers that
customers reviewed and evaluated during this exercise. Table 1 (below)
presents the "headlines" or positioning statements of all offers reviewed
during the exercise.

Table 1. Hypothetical Energy Offers for Power Shopping Exercise

• People's • Learn how your • Supplying you Energy

Energy Energy with the Power of Options that
Powering Operations Information Suit Your
your Measure-Up Business
Community Needs

• Cleaning up • Personalized • Total Comfort • We Serve

our Comer Energy Supply Solutions Companies
of the Earth Service Like Yours

• EServe • Total Energy • Don't let your • Absolute

Solutions Load get out of Power
Shape

Energy • We'll keep the When only the • Energy

Services Lights on and most reliability, Services
Designed leave a little quality power Designed for
for Chief Extra Money in supply will do Financial
Engineers your Budget Managers

Of the sixteen offers, six were most appealing to ell customers.

Preferences differed somewhat across the customer types:

~ Small business customers allocated 65 percent of their votes to the

following five offers: low price (i.e., extra money in your budget),
customized energy options (i.e., suit business needs), green energy, total
energy solutions, and reliable power.

~ Large business customers allocated 61 percent of their votes to the

following five offers: customized energy options, low price, reliable
power, green energy, and risk-free energy.
420 Pricing In Competitive Electricity Markets

~ Industrial customers allocated 60 percent of their votes to the following

four offers: low price, reliable power, green energy, and risk-free energy.

As shown in Figure 3, across all customer types, 70 percent of all votes

(1,314 votes) were allocated to six offers: green energy, low price, reliable
energy, risk-free energy, customized energy options, and total energy
solutions.

12%
IIGreen Offer

CI Low Price Offer

~ Reliability Offer

II Risk-Free Offer

CI Energy Options Offer

ClTotal Energy Solutions

Offer
ClOther Offers

Figure 3. Customer Allocation of "Votes" Across Energy Offers

These "preferences" are based on a qualitative discussion group and a

small number of respondents but they very clearly indicate willingness to
"vote" for offers besides the commodity offer (i.e., the low price offer). The
results also indicate that while customers ARE concerned about price,
preferences for other energy services are evident
From this qualitative research, we conclude that, assuming a reasonably
competitive price, energy retailers could position their offers along several
dimensions in addition to low price - reliability, green, risk-free, customized
energy options, and total energy solutions. The next step is to quantitatively
measure customer preferences for alternative electricity offers and the
associated WTP values. We discuss an approach for valuing electric service
offers next.
Pricing in Competitive Electricity Markets 421

4. AN APPROACH FOR VALUING ELECTRICITY

SERVICE OFFERS

In a prior study, EPRI estimated WTP for a limited number of electricity

service offers that varied by type of electricity supplier, length of contract,
and price/rate structure. ii In a current study called ShareWars, EPRI has
designed a choice experiment to gather customer preference information on
a much broader range of possible electricity service offers. We are currently
in the process of collecting this information and are not in a position to
report on the actual WTP results. Below we describe the process of
designing a complex choice experiment to capture a wide range of possible
electricity service offers.
First, we drew on the findings from our qualitative research to identify an
array of value propositions that could be offered by electricity suppliers. We
assumed that a given supplier might offer products/services around one or (at
most) two value propositions. The value propositions that we identified are
the following:

~ Low Prices and Alternative Rate Structures

~ Green Energy
~ Superior Customer Service
~ Range of ValuelPremium Services
~ Community Presence

For each value proposition, we then defined a set of possible features or

attributes. We defined these sets of features or attributes associated with a
value proposition as a cluster. So that we could estimate WTP for all
attributes, we included the price of electricity in all of the clusters. Table 2
(below) shows the attributes for the five clusters.
For each of these attributes, we defined specific levels. For example, the
electricity supplier attribute included the following six levels:

1. Your local electric company

2. An affiliate of your local electric company
3. A neighboring electric company
4. A well-known electric company
5. A well-known energy company
6. An unfamiliar energy company
 422 Pricing In Competitive Electricity Markets

Table 2. List of Attributes by Cluster

Pricing/ Green Energy Customer Service Value/Premium Community

Cluster Services Cluster Presence Cluster
Alternative Cluster

Rate Cluster

· Electricity
supplier
· Electricity
supplier
· Electricity
supplier
· Electricity
supplier
· Electricity
supplier

· Contract
terms
· Rate plan
(price
· Rate plan
(price
· Rate plan
(price
· Rate plan
(price
variation) variation) variation) variation)

· Rate plan
(flat,
· Amount of
renewable
· Personalized
services
· Other fuels
offered
· Local
economic
seasonal, energy development
time-of-day,
and hourly)

· Sign-up
benefits
· Primary
renewable
· Customized
payment
· Other
energy-
· Local
voluntary
energy services related contributions
source services

· Customized
billing
· Power
outage
· Local
presence
services guarantees

· Information!
Web site
· Power
quality
services guarantees

As another example, for the rate plan variable in the pricing cluster, we
specified fixed, seasonal, time-of-day, and hourly prices. Then within each
price, we specified a range of levels. For the other clusters, we simply
varied a fixed rate.
We designed and implemented a choice-based conjoint study based on
the attributes and levels developed for each cluster. Conjoint analysis is a
technique for eliciting customer preferences for a set of "product profiles"
where a product (or in this case an electric service offer) is defined by a
specific set of attributes or features. In our choice-based conjoint study,
each question presented a set of four electric service offers and asked
respondents to choose the one that they prefer.
Pricing in Competitive Electricity Markets 423

Within each cluster, the choice experiment consisted of forty-eight choice

questions. We then designed twelve versions of a survey instrument that
consisted of four choice questions from each of the five-value proposition
clusters. This allowed us to cover the whole range of attributes and levels
across the clusters. The electric supplier and electric rate attributes were
common across all clusters and all choice questions. The resulting data will
allow us to estimate WTP for any attribute in the study and to simulate
market share for any combination of attributes within or across the five
clusters.

5. ESTIMATING WILLINGNESS TO PAY

We use responses to the choice questions to statistically estimate an

aggregate utility function for each of the primary customer segments
included in the study. The utility function quantifies the relationship
between each attribute-level and overall customer utility. Since WTP is
defined as the maximum amount of money a customer would pay to move
from one "state" (or service offer) to another, while remaining at the same
level of overall utility, the utility function is the basis for the WTP estimates.
For the purpose of illustration, a simplistic example of a WTP calculation
and the type of findings that might be generated using this methodology are
shown below. Note that the "~ Utility" parameters are coefficients
estimated in the overall utility function.

Example WTP Calculation: 1-Year Contract Term versus 3-Year Contract

Term*

WTP = [d Utility per d Contract Term] + [d Utility per d Rate]

Where: d Utility per d Contract Term = 0.55(1.Year) - 0.30(3.Year) = 0.25

And: d Utility per d Rate (¢ per kWh) = 0.75

and: WTP = 0.25 + 0.75 =0.33

In other words: Customers are willing to pay an extra 0.33 ¢ per kWh for a 1
year contract over a 3-year contract

*Data are hypothetical and presented for the purpose of illustration.

424 Pricing In Competitive Electricity Markets

The WTP calculation can be applied to a set of attributes in order to assess

how customers value a bundle of attributes or a complete service offer. For
example, Table 3 provides hypothetical calculations for two alternative
service offers compared to a base offer. First, a WTP estimate is calculated
for each level of the alternative offer that differs from the base. Then, the
level-specific WTP estimates are summed to arrive at a total WTP for the
service offer. Note that the final WTP estimates are expressed in cents per
kWh and are relative to the price of the base offer.
Recall the assumption that customers will choose an offer in the market
place when its price is less than or equal to their WTP. Thus, the
calculations demonstrated here represent, for the average customer, the
maximum price differential at which they would choose the offer. Actual
market prices will tend to be lower than WTP for most individuals.

Table 3. Hypothetical WTP Calculations for Alternative Electric Offers

Base Offer "Premium Service" Offer "Value Added" Offer

FEATURE FEATURE WTP FEATURE WTP

Local electric Local electric 0.00 Local service company 0.00

company company

No personalized Direct access to your 0.25 No personalized 0.00

services service representative services

No Web site Energy transactions 0.15 No Web site 0.00

Web site

No other fuels No other fuels offered 0.00 Natural gas, propane, 0.20
offered and oil also offered

No energy-related No other fuels offered 0.00 Warranties, 0.30

services maintenance contracts
on new equipment
Total WTP over Base: OAO Total WTP over Base: 0.50
Pricing in Competitive Electricity Markets 425

6. USING A WTP STUDY TO ASSESS MARKET

OPPORTUNITY AND STRATEGIC
POSITIONING

How should a retail energy company apply the WTP methodology to

develop its marketing strategy? Prior to development of an actual service
offer, the WTP methodology may be used to assess opportunities for overall
market positioning.
For instance, the methodology presented above is based on an aggregate
customer utility function and provides average or overall WTP estimates.
This approach can be replicated at the individual level. That is, individual
utility functions can be estimated to produce estimates of individual WTP.
Such customer-level preference data can be a great knowledge base for
exploring and understanding market segments. For instance, this
information can be used to answer questions such as:

~ How does this segment value a specific set of features differently than
the population at large?
~ What is the relative size of this segment?
~ What are the demographic/firmographic characteristics of this segment?
~ To what extent do these customers lie within the market segments we
already serve (geographically or otherwise)?
~ What do WTP estimates imply about our ability to take this offer to
market profitably?

After establishing an overall strategic pOSItIon in the market place, a

more focused WTP analysis can be used, in conjunction with other types of
assessments, to develop specific offers.

7. CONCLUSIONS

Recent research suggests consumer interest in electricity service offers

other than price-based offers. This implies the existence of other market
segments beyond the commodity market segment. The question is which
features are of most interest to non-commodity market segments and what
relative value is placed on these features?
We have presented the WTP methodology as a tool for answering these
questions. Specifically, the WTP methodology quantifies relative
preferences for possible service features and links these preferences to
market value (or the upper bound of market price). These preference data
426 Pricing In Competitive Electricity Markets

and value measures are key to exploring market segments and uncovering
customer value for non-commodity electric service offers.
While the WTP methodology is clearly a useful tool for designing
electric service offers, we suggest it also yields powerful information for the
critical decisions that must be made prior to the development of an
offer-namely, to determine strategic positioning in the marketplace.
Individual-level WTP analyses:

~ Yield information about the preferences of targeted market segments and

their relative size and characteristics, and
~ Establish the economic threshold at which a company must be able to
compete.

Such information provides critical insights to strategic decisions.

NOTES

i Selting, Anne. "Value-Added Services in a Competitive Electric Industry." See Chapter

25.

ii
EPRI, Predicting Customer Choices Among Electricity Pricing Options, Volume 2: Retail
Markets. EPRI TR-1 08864-V2. Final Report, November, 1998.
Chapter 27

Electricity Marketing: Is The Product The Price?

James Long, Bryan Scott, and Bernie Neenan

Central and South West Services Company (CSW), CSW and AXS Marketing, LLC.

Key words: Customer Preference; Flexible Pricing; Pricing Products; Real-Time Pricing
(RTP); SelectChoicesM ; Time-of-Use Pricing (TOU); ValueChoice sM .

Abstract: Utility pricing departments face a difficult challenge as retail markets are
restructured. They must maintain existing rate structures, which are based on
the utility's average costs, while customers' attention is more and more
focused on market-based prices. Customers will inevitably make the
comparison and demand that differences be rationalized. Customers are not
willing to wait for competition - they are demanding choices now. And, there
is no shortage of national marketers, hungry for brand recognition, who are
willing to supply those needs. How can utilities respond, when they are tied to
monolithic, cost-based rates that provide little room for diversity?

Central and South West Services has begun to develop and test a full line of
integrated pricing products under the ValueChoice sM brand. It is testing its
new products in selected retail markets. Experience with real-time pricing
(RTP) demonstrated that customers were willing to take on new risks if the
potential rewards were great enough. But, its appeal was limited by the
inherent price risks that most customers were simply unwilling to tolerate. To
meet the needs of smaller commercial customers, CSW developed a time-of-
use pricing (TOU) pricing product called SelectChoicesM . This allows
customers to choose from alternative TOU product options with different
combinations of peak hours and peak prices. Some product options provide
attractive prices for load growth, while others offer incentives to shift load off-
peak, while still others support complex behaviors that include both actions.

Many industrial customers who liked the RTP structure, but could not bear the
price risks, are finding SelectChoicesM an attractive alternative to their
conventional rate.

A. Faruqui et al. (eds.), Pricing in Competitive Electricity Markets

© Kluwer Academic Publishers 2000
428 Pricing In Competitive Electricity Markets

The RTP and SelectChoice sM pilots confirm that customers are eager to try
new ways of buying electricity, and that they are willing to undertake changing
their usage patterns in order to realize benefits.

1. INTRODUCTION

The retail electricity market in the U.S. is enormous and diverse.

Millions of customers are served by over 3,000 retail distributors that
operate a network of hundreds of thousands of miles of transmission and
distribution circuits and daily dispatch tens of thousands of generation units.
Annual retail billings exceed $200 billion and individual customer bills
range in size from a few hundred dollars a year to several million dollars a
year. The cost of supplying electricity to end-users varies substantially by
time of day and season of the year, as does the inherent value customers
realise from using electricity. However the coincidence between the two is
not exact.
One would expect that widely varying supply costs, combined with
extensive customer and end-use diversity, would support a robust
merchandising system with diverse product options that exhibit substantial
time differentiation in usage rates. This has not been the case. For decades,
a single monopoly supplier has provided its franchise customers service
under a single, mandated tariff structure utilizing rates that do not reflect the
market's inherently diverse temporal costs and usage values, but instead are
administratively set to recover embedded costs 'equitably'. Recognition of
differences in customers' value of electricity and corresponding preferences
for service diversity has been limited to categorizing them into broadly
defined service classes - residential, commercial, and industrial.
Assigning customers to one of a few service classes facilitates the rate
'engineering' process, which assigns cost to customers based on
standardized (and arbitrary) accounting conventions rather than on revealed
consumer preferences. This is largely because costs are easier to
characterize than customer preferences in the administrative environment
where rates are set. The regulatory rate setting process lacks the spontaneity
and creativity that characterize a competitive market required to set rates to
equate marginal usage values to marginal costs. Instead, it has
institutionalized monolithic rate structures as an unavoidable consequence of
fulfilling its main objective, ensuring that only those costs that are properly
incurred enter into the rate formulation, and that those costs are allocated
equitably among all customers who use the system.
Pricing in Competitive Electricity Markets 429

Utilities are concerned with recovery of costs, and prefer rates that limit
exposure to non-recovery over more diverse but risky tariff offerings. Since
simple mandatory rate structures meet these objectives, neither the utilities
that serve retail customers nor their regulators have had an incentive to
explore customer classifications that recognize the value received from
electric service.
This system has worked well until now because the primary objective of
all stakeholders has been to make electric service universally available while
taking advantage of technological opportunities for economies of scale in
generation. Questions of improved market efficiency and customer choice
have been largely relegated to the domain of academic inquiry. Setting rates
to equal marginal costs was considered a heroic but futile gesture, as the
resulting revenue stream would not follow incurred costs and destabilize the
industry, or worst, discourage investment and raise the overall cost of
maintaining universal supply. There appeared to be no way to bridge the
gulf between setting prices to reflect marginal costs and generating revenues
that met allowed rates of return, so administered rates set at embedded costs
became the accepted norm.
Why have customers tolerated this conformity at the expense of choice?
Perhaps because they have come to accept that monopoly provision of
electricity is necessary, and that lack of choice is the price they pay.
Fully integrated electricity markets characterized by bundled, embedded
cost-based rate have come under new scrutiny. Many argue that monopoly
franchises are no longer necessary and their very existence stands in the way
of realizing the full benefits stemming from system investments. Innovative
regulatory schemes have been proposed to break the bond between how
costs are incurred and how they are recovered. Wholesale markets are being
opened up to new entrants allowing non-utility investments in generator.
Early restructuring initiatives are focused on complete utility divestiture and,
in some cases, eliminate participation by the incumbent in the commodity
part of the business completely so that customers can get complete and
unbridled access to non-utility suppliers, as is the case in the natural gas and
telephone industries.

2. CONVENTIONAL RATE MAKING PRACTICES

WILL NOT MEET TODAY'S CHALLENGES

An enormous amount of attention has been focused on how the industry

should be restructured to allow entry of new suppliers while honoring
commitments to incumbents. The debate includes vigorous attention to the
divestiture dividend that customers will get - the one-time ratcheting down in
430 Pricing In Competitive Electricity Markets

costs. In this debate, little attention has been focused on what merchant
system will replace existing rates in equilibrium markets, what choices will
be available, and how customers will benefit. That is being left to the
invisible hand of the market place.
Nowhere is the gap between today's practices and tomorrow's
competitive market environment greater than in how electric services are
established and priced. The emphasis on embedded costs at the expense of
reflecting marginal costs is out of tune with competitive market practices.
Large, broadly defined service classes mask the underlying diversity of
customer needs. The lack of service options and service feature diversity
ignores the regular, temporal and spatial diversity in how customers use and
value electricity. If there are no choices, there is no choice.
Is drastic industry restructuring both a necessary and sufficient condition
for establishing electricity markets that provide customers with choices? Do
we need to throwaway the existing infrastructure so that customer
sovereignty can reign? Alternatively, can we reengineer the existing
ratemaking system so customers can have more choices faster, under the
existing industry structure?
To survive and prosper in a competitive electricity market, distributors
will have to overhaul their rate structures and introduce variety, diversity and
choice to their service offerings. In turn, regulators will have to do their part
by streamlining the process by which new electric pricing options are
reviewed and approved. Some of these new products will only serve to
transition the industry to a competitive market structure. Others will become
permanent fixtures. What is important is to recognize the changing politics
of electricity markets. Customers are demanding that they be recognized for
their diverse needs. Those who respond to these demands willingly and
proactively will be rewarded with newfound customer loyalty.
The utilities must revamp their retail electricity merchandising practices
in order to manage the transition to competitive markets and to sustain
prosperity once those market are full operational. Managing the transition to
competitive retail pricing is the most difficult part, as it requires maintaining
the old order, where embedded cost recovery is paramount, while
introducing aspects of competitive pricing, where marginal cost and product
diversity reign supreme, but bring with them a new risk/reward business
paradigm. It requires responding to customers' demands for diversity and
exercising flexibility in a regulatory environment that traditionally has
emphasized stability and conformity.
Finally, it requires encouraging out-of-the-box thinking and appending
new ideas and merchandising practices to conventional institutional
structures. How well today's utilities handle the transition to the new market
Pricing in Competitive Electricity Markets 431

structure, especially as it relates to the pricing of electricity, will go a long

way in determining how well they will prosper in competitive markets.

3. A HISTORY OF RESPONDING TO EXPRESSED

CUSTOMER NEEDS

Innovation in mechanizing electricity is not new to Central and South

West Services (CSW).i Over the past twenty years, its member utilities have
been adapting their traditional tariff offerings to promote economic
development and to meet the needs of an ever-changing marketplace. New
tariffs and special contract forms have been introduced to meet customer
needs that conventional tariffs simply could not accommodate, including
responding to competitive threats.

3.1 Natural Gas Tolling

In the early 1980's, rising natural gas prices, combined with pressures
from unregulated suppliers in Oklahoma and a faltering economy, raised
concerns about the efficacy of existing electric rates for industrial customers.
Those who were able to acquire and transport their own gas at lower prices
found building their own generating units increasing attractive, despite the
much higher initial capitalization costs compared to the large generating
units of Public Service of Oklahoma (PSO). PSO, one of the sister CSW
operating companies, introduced two programs to supplement existing rates
and avoid non-economic bypass. The first was the Industrial Gas to
Electricity Rider (IGER) which was replaced later by the Fuel Supply
Electricity Rider (FUSER).
Under IGER and FUSER, customers were allowed to purchase market
priced gas for delivery to PSO's power plants, a practice called tolling. In
return, their electricity rate was unbundled and fuel costs were removed, in
effect the customers swapped their gas for PSO's at the generating unit.
Many customers at that time were able to buy short-term gas supplies
cheaper than PSO, which was bound to long-term agreements. The
customers who engaged in tolling enjoyed lower costs because of their
participation in the supply process, and PSO was able to stimulate economic
growth and avoid encouraging uneconomic bypass. This proved to be a
useful lesson on the benefits of incorporating customer's interests into the
electricity pricing process.
432 Pricing In Competitive Electricity Markets

3.2 Time-of-Use (TOU) Pricing Services

In the mid 1980' s, the addition of a new base load generation plant to
meet growing demand caused the gap between average and marginal costs to
expand significantly. CSW responded by introducing time-of-use (TOU)
pricing structures into many of its commercial and industrial rates. These
TOU-based rates ensured the continued equitable collection of embedded
demand costs while keeping marginal usage costs more closely tied to actual
(marginal) fuel costs. These TOU services provided customers with
incentives to shift load from peak to off-peak hours and reduce their
electricity costs. Moreover, customers were offered attractive rates for load
growth during the off-peak hours that are free of demand charges. To this
day, some CSW customers exhibit the inverted load profile characteristic of
operations that were specifically organized around TOU pricing in order to
reduce electricity costs.

3.3 Real-Time Pricing Service

Recently, CSW has aggressively pursued the implementation of real-time

pricing (RTP) programs that introduce market-based pricing to supplement
its conventional tariff structures. Early experiences with RTP, beginning at
PSO in 1992 and later at all CSW operating companies, demonstrated that a
long established, but previously thought to be inaccessible, spot pricing
principle could successfully be put into practice in a retail setting. The RTP
model combined the equity of embedded cost recovery with the efficiency of
marginal cost pricing into a single product offering. In other words, it
offered a way out of the age-old compromise of trading off pricing
efficiency and equity.
RTP has been successful in its original adaptation and over time, as CSW
acquired experience, additional refinements were incorporated to improve
both customer satisfaction and enhance other stakeholders' benefits.
Moreover, RTP has evolved to become a highly flexible platform for
resolving long-standing pricing issues. For example, a variation of the
original RTP model was successfully adopted in 1994 by PSO to replace its
curtail able service in response to customer's demands for buy-through
prices. Now, CSW uses prices to induce beneficial customer load responses
rather impose them through mandatory curtailments.
More recently, the RTP platform has been used to design customized
solutions to meet special customer circumstances. When it became
necessary to respond to competitive threats, PSO turned in 1996 to a tailored
pricing product platform under the Contract Competitive Rate (CCR). The
CCR allows PSO to tailor, on a customer-specific basis, the RTP access
Pricing in Competitive Electricity Markets 433

charge to reflect the customer's supply alternative while ensuring that it

faces efficient and effective prices for usage.

3.4 The Challenge: Something for Everybody

The programs described above were developed in response to specific

threats to or to take advantage of the particular needs and capabilities of
customer segments. For the most part they were available to, or their appeal
was limited to, the largest customers, those over 1,000 kW maximum
demand. Moreover, due to the risks inherent in these programs, only a
subset of those that were eligible were able to generate significant benefits
from participation. These innovations demonstrated CSW's intent to expand
new choices to customers in how they purchase electricity, but they also
pointed out the need to be even more creative in designing new services so
that more customers would have choices.
Under RTP, customers face hourly prices that are posted one day ahead.
These prices vary from a level well below marginal usage costs under the
standard tariff to levels ten times the tariff energy rate. Market research
conducted by CSW revealed that while a wide range of commercial
customers understand and liked the two-part pricing structure used in RTP,
few can take advantage of it because of the price risks. They cannot dedicate
the time and resources required to develop price response strategies and
deploy them on a daily basis. These customers want product flexibility
combined with price security.
Others, including many residential customers said they wanted no daily
price variability at all. They realize value from the consumption of
electricity in well-established patterns. While some said they could
accommodate seasonal changes in the price they paid, others wants the
security and convenience of a fixed price of electricty, regardless of when
and how they used it.
The message was clear: RTP would not be enough to satisfy all customer
segments demands for product and pricing diversity. New pricing products
would be required to satiate customers' needs and fulfill the goal of
providing choices for every customer.

4. VALUECHOICEsM : CSW'S PLAN TO PROVIDE

CHOICES FOR ALL OF ITS CUSTOMERS

CSW has developed a new electricity-merchandising concept, called

ValueChoicesM , with the goal of providing all customers with alternatives to
434 Pricing In Competitive Electricity Markets

standard retail electricity rates. AXS Marketing developed the concept of a

complete line of electric pricing products, under its EMartSM brand, which
are offered as alternatives to existing rates. CSW has adapted that concept
and created the Value-Choice sM line of electricity pricing programs to help it
meet the diverse needs of all customers served by CSW's operating
. ii
compames.
The ValueChoice sM approach is simple in concept. Design products and
product features that better match customers' needs or wants and the cost of
supplying them The reward for doing so successfully is greater customer
satisfaction, higher customer loyalty and improved earnings in today's and
tomorrow's electricity markets. The challenge is to match customer's wants
and needs to the underlying supply costs in ways that benefit both the
customer and CSW, and that creates value relative to existing rates.
ValueChoice sM pricing products are based on four basic pricing platforms
that determine what features are used in the product and how those features
are combined to create value and establish usage prices. The platforms
represent different treatments of the underlying risks of supply. Figure 1
(below) describes the basic pricing scheme. The MarketChoice sM pricing
product uses the RTP model to provide customers virtual access to market-
based prices by incorporating embedded costs into an access charge and then
setting usage prices equal to contemporaneous marginal costs. Because
prices are delivered to customers either a day in advance, or in some cases
only an hour ahead, some of the inherent price volatility is transferred from
the utility to the customer. What does the customer get? It gets access to the
low prices (below the equivalent tariff purchase price) that prevail over 90
percent of the hours of the year. Customers who can mange risks get lower
prices than would otherwise be available.
The second in the line of pricing products is SelectChoice SM , which
utilizes a time-of-use pricing platform. As is the case with MarketChoice sM ,
the two part (access and energy) design means that SelectChoicesM
customers don't pay billing demand charges and they are offered incentives,
in the form of usage prices, to change how they use electricity. The big
difference between SelectChoice sM and MarketChoice sM is that
SelectChoice sM customers are not exposed to RTP's widely varying daily
usage prices. The time-of-use pricing platform converts hourly marginal
cost forecasts into established an annual schedule of peak and off-peak TOU
usage prices. The primary market for SelectChoice sM is expected to be
customers whose peak demand is between 50 and 1,000 kW.
Pricing in Competitive Electricity Markets 435

ValueChoice sM Pricing Products

Figure 1. Value Choice Energy Products

CSW calls the third product under the ValueChoicesM umbrella

CustomChoice sM . The costs of administering TOU pricing products exceed
its benefits, largely due to metering and data management costs, for many
customers. Especially for those who simply can not make changes large
enough to offset the incremental costs. However, these customers are as
vocal in their demands for new ways to buy electricity, especially if it
eliminates paying maximum demand charges.
CustomChoice sM offers customers the opportunity to pre-purchase
electricity in various sized blocks, much like telephone calling plans offers
blocks of monthly connect-time for a fixed price. A block of kWh can be
purchased to cover usage for a month, season, or an entire year. Each block
involves a fixed price for the block of energy, and provision for adjustments
for 'swings' in actual period usage above and below the purchased block
amount. This pricing product should appeal to smaller commercial
customers, who have predictable usage patterns or want to deploy control
strategies using a target bill amount. CustomChoicesM is also expected to be
popular among many residential customers, especially those seeking volume
discounts or who want to hedge against capricious changes in usage.
The final pricing product is called EasyChoice sM which, as its name
implies, is designed to appeal to appeal to residential and small commercial
customers who prize simplicity and security in their electricity purchasing
plan. Since most CSW customers already face some form of time-
differentiated pricing, in the form of time-differentiated demand or hours-use
rates (or, at a minimum, fuel charges are set dynamically), EasyChoice sM
offers a real alternative in how they buy electricity.
436 Pricing In Competitive Electricity Markets

The objective in designing ValueChoicesM is to develop a portfolio of

pricing products that fulfils virtually any customer's needs. This is a large
undertaking that will be accomplished in stages. Research has begun on
CustomChoicesM and EasyChoicesM , starting with a survey to identify
customer preferences for alternative product features. Currently CSW offers
MarketChoicesM products in all of its retail markets, and it is testing
SelectChoice sM in under pilot programs in most of the franchise markets.

4.1 SelectChoicesM Product Design

SelectChoice sM is designed to provide as an alternative to conventional

rates for small-to-medium commercial and light industrial customers who
are served under billed demand rates. Subscribers pay a fixed, individually
determined delivery charge (levied in twelve equal monthly payments) to
cover embedded generation, transmission and distribution demand costs and
a time-of-use price schedule determines the price of metered energy (kWh).iii
The SelectChoice sM design is unlike conventional TOU schedules in that
subscribers can choose from nine different time-of-use product options. The
product options involve different combinations of what and how many hours
of the day are designated as peak and in the level of the prices levied on peak
and off-peak usage (Figure 2).

[.low MemumHigll
Hours in
Peak kWh Price
'l1llbuts
6hoijis

Figure 2. Peak Price

SelectChoice sM offers customers three choices in the length of the peak

period. Each window is centered on the CSW system peak hour, which in
the summer is 5:00 P.M. The four-hour peak window is 3:00 P.M. to 7:00
P.M., the six-hour peak 2:00 P.M. to 8:00 P.M. and the eight-hour peak 3:00
Pricing in Competitive Electricity Markets 437

P.M. to 9:00 P.M. Each of these peak period windows applies to weekdays
of the summer months, June through September.
Why is it important to have multiple peak period definitions?
Conventional TOU rates offer only a single, and generally long, designation
of the daily peak period or window, despite the fact that customers' ability to
shift loads varies considerably. The SelectChoicesM peak windows are
designed to accommodate different levels of load shifting capabilities. The
narrow, four-hour peak window appeals to customers who want to shift load
and reduce their bill, but who are limited in their load management
flexibility, especially as it effects how many hours of the day they can
reduce usage. They can move important activities around a few hours, but
curtailing long periods would be too upsetting to the overall operation of the
plant. For such a customer, the four-hour TOU window makes participation
in TOU possible and beneficial, and opens up a new load management
market segment fort CSW.
One example from the pilot illustrates the possibility of such behavior. A
SelectChoicesM subscriber got its employees to agree to start earlier in the
summer so that they could shut down at 3:00 P.M. instead of at 5:00 P.M.
Then, by subscribing to product option 1, it reduced its base bill by shifting
load from the peak hours of 3:00 P.M. to 5:00 P.M. to the off-peak hours of
6:00 A.M. to 8:00 A.M. At the end of the summer, the subscriber found that
its employes enjoyed the new hours so much, because they had more
afternoon time for family and diversions, that it elected to adopt the new
work scheduel year-around. In this case, the benefits realized by the
customer exceeded those measured in bill savings.
Some other subscribers adjusted the timing of the operation of key prices
of electrical equipment in order to realize bill reductions. Those with a great
deal of processlbusiness flexibility were able to subscribe to a product option
with a six of eight hour peak and realize even greater savings.
Subscribers also can choose from among three peak and off-peak price
ratios. These pricing options accommodate the inherent diversity in
customers' timing, usage, and valuation of electricity. High peak to off-peak
price ratios are an incentive to shift load from the peak to lower bills. Low
peak to off-peak price ratios are appealing to customers who want to expand
load, but could not do so when facing maximum demand charges, since the
peak consumption 'penalty' is applied only to energy usage. Many
customers, including some who shifted load to save, used the low off-peak
prices, which are free of demand charges, to expand electricity usage and
enjoy greater value from their electricla service connection to CSW.
By offering three different price ratios, SelectChoicesM provides a span of
options that should cover most customer needs. Combining the three peak
window options and the three price ratios results in nine SelectChoicesM
438 Pricing In Competitive Electricity Markets

product options. From these nine possibilities, subscribers can select the
option that best fits their individual consumption profile and load
management capabilities.

4.2 SelectChoicesM Features

The basic pricing structure underlying SelectChoice sM involves fixed

monthly charges for delivery costs and variable energy charges based on
metered kWh usage. In addition, a program charge is used to collect costs
uniquely associated with SelectChoice sM . The SelectChoice sM formulation
is simple, straightforward, and easy for customers to understand (Table l).

Table 1. SelectChoice sM Formulation

Formula: Bill =Program Charge + Delivery Charge + Energy Charge

Where,
• Program Charge ($/month)- recovers incremental administration costs,
• Delivery Charge ($/month)- recovers allocated T&D costs and any residual generation
costs not recovered in the energy prices, and
• Energy Charge ($/kWh)- recovers marginal costs and collects a contribution to
embedded costs.

Design Features
./ No Metered Demand Charges
./ TOU Prices Fixed for Subscription Term (one year)
./ Choice of Nine TOU/Pricing Packages
./ No-Risk Satisfaction Guarantee

Table 2 provides a description of all the major SelectChoice sM product

features and describes the role each feature plays in balancing customer
needs with CSW's obligations and expectations. The table also provides a
brief description of the design challenge associated with each feature. A few
features merit additional discussion.
The use of a two-part structure requires establishing a load profile for
each customer that serves to establish its delivery charge. CSW had
extensive experience developing load profiles for larger customers in its
RTP program. Because the target market for RTP was larger, commercial
and industrial customer for whom extensive hourly load profile data were
available, developing customer-specific profiles was relatively easy.
 Pricing in Competitive Electricity Markets 439

Table 2. SelectChoicesM Base Product Features and Design Challenges

Base Product Role Design Challenge

Feature
Standard Tariff Establish revenue neutrality Tie delivery charge calculation to projected
with respect to the Otherwise rates for the annual contract period to
Applicable Tariff (OAT) prevent loss of base revenues. Must
anticipate rate changes or fuel adjustments
that will effect base rates during the contract
period.
Customer Establish revenue neutrality Develop fair procedures for establishing
Usage Profile with respect to base usage CUPs that balance customers' desire for
(CUP) under the OAT flexibility and concessions with the utility's
desire to prevent windfalls.
Program Recover program Determine what costs can be recovered
Charge administration costs through this fee without placing a constraint
on subscription.

Delivery Transparency with respect to Establish a minimum fixed monthly charge

Charge T&D charges that recovers non-variable transmission and
distribution costs.
Product Provide customers with Adapt the basic SelectChoicesM model to
Options choices prevailing circumstances to reflect
corporate pricing objectives.

TOUWindow Alternative daily TOU peak Balance utility's need to cover capacity cost
periods exposure with customers' desire for flexible
peak definitions.

TOU Usage Peak and off-peak prices Set prices to cover marginal energy supply
Prices and peak capacity cost exposure, and to
recover program costs and generate a
dividend to all stakeholders.
-
Metered Usage Actual metered and billed Establish contract subscription provisions
kWh under SelectChoice sM that minimize the costs of switching
customers to SelectChoicesM and meeting
their service needs.
No Risk Overcome customer Remove risks to subscriptions so
Guarantee skepticism of benefits of SelectChoicesM is attractive to customers
SelectChoice sM . without exposing utility to excessive
revenue losses.
440 Pricing In Competitive Electricity Markets

However, profile data are available for only a small percentage of the
target market customers for SelectChoicesM . For most, the only usage data
available for these customers are monthly billing data comprised of
aggregate monthly energy usage and maximum demand.
To provide load profiles for prospective subscribers to SelectChoice sM ,
CSW developed procedures for creating customer usage profiles (CUPs) for
the year prior to subscription. The CUP defines the aggregate peak and off-
peak kWh during the entire month. CSW developed CUPs by utilizing a
peak-proportioning methodology. Once a peak proportion was assigned to a
customer, the TOU peak and off-peak quantities were developed by
multiplying that proportion by the monthly billing kWh.
Using existing load research data and profiles gathered from external
resources, a library of proportional profiles was developed to characterize
different customers according to factors that would help identify their usage
profile; business type (SIC), size, number of shifts, and other factors. Each
prospective subscriber then was characterized according to these
classification variables, matched to a sample entity in the load library, and
then assigned the corresponding peak proportion.
Market research conducted during the design stage influenced how some
of features were established. For example, the delivery charge is applied
through a levelized monthly charge, rather than as individual monthly
amounts. The latter could vary considerably due to prices and month CUP
differences, because customers preferred that treatment because it would
simplify budgeting. The choice of the length of the alternative peak periods
was heavily influenced by how customers responded to alternative
configurations.
Market research also revealed that customers were skeptical of their
ability to benefit from SUbscription, at least initially, and wary of the
downside risks. Many customers reported that the prospect of paying more
under SelectChoice sM was enough to offset the prospective benefits, and
would prevent them from giving the program a trial. However, when
subscription was portrayed as accompanied by a No-Risk guaranteed, much
of this resistance to subscription was removed. Under the No-Risk
guarantee, if the customers pays more under SelectChoice sM than it would
have for its actual usage under the standard rate, then CSW would rebate it
difference. Subsequent experience in the pilot confirmed that this feature is
the most significant factor in getting many customers to subscriber to
SelectChoicesM .iv
Pricing in Competitive Electricity Markets 441

5. MARKETING SELECTCHOICE sM : A NEW

CHALLENGE FOR CSW

CSW recognizes that a new product like SelectChoice sM presents some

formidable marketing challenges in its initial introduction. The first job is to
motivate customers to become involved in electricity consumption decisions.
Next, the benefits of subscription must be quantified. Finally, the customer
must be convinced to subscribe and give SelectChoice sM a one-year trial.
This is a fonnidable task.
On the one hand, customers should welcome having choices. The mere
existence of choice provides customers with a means for measuring the value
they realize from electricity service. CSW's experience is that customer
satisfaction increases simply as the result of offering choices, even if the
customers choose not to exercise the choices. Many customers use the new
pricing option as a reference to measure the value they received from the
traditional rate, and in the process established newfound satisfaction with the
conventional rate.
On the other hand, customers are not accustomed to evaluating new
products, as they historically have had little choice in how they buy
electricity. Making a choice from among from nine SelectChoicesM options
may be somewhat imposing at first. Customers may be wary of the benefits,
or incapable of estimating what those benefits might be. If customers do not
understand today's rates, how can they compare them to the SelectChoice sM
product alternatives? The SelectChoicesM product offers customers choices,
but making the right choice may not be obvious for many of the potential
customers. The challenge develop delivery processes that make customers
aware of the new options, motivate them, by demonstrating the benefits of
subscription, to become more involved in how they use electricity, and then
assisting them in realizing those benefits by preparing and executing a load
management response strategy.

6. A PRODUCT -ORIENTED FOCUS

Successfully marketing SelectChoice sM required designing a new

merchandising system. CSW found through its market research that most
commercial customers want choices, and they do not find the nine product
options of SelectChoice sM too imposing. However, some encounter
difficulty making the benefit comparison between SelectChoicesM and their
current rate. Without good information that justifies changing services,
customers can be expected to exhibit status-quo bias and will opt to stay on
their existing rate. To meet the aggressive goals set for SelectChoicesM ,
442 Pricing In Competitive Electricity Markets

customers must be made aware of the benefits of SelectChoice sM , and they

must be convinced that the benefits are accessible to them. Accomplishing
this requires a highly coordinated and focused marketing and sales campaign
directed toward customers who are the most likely to benefit.
To ensure the proper direction and focus, all activities associated with the
merchandising and servicing of the SelectChoice sM product were managed
by the SelectChoice sM Product Manager. The product manager establishes
subscription and price response goals, in conjunction with the design team
and with the cooperation of fulfillment departments such as metering,
billing, and marketing, and is responsible for coordinating activities toward
the realization of those goals. Quarterly product profit statements are issued
that summarize the program's financial performance. All costs associated
with the programs operation are included, along with the cost of supplying
energy to subscribers. The use of a product manager to coordinate all
product implementation functions ensures that the high level of integration,
education, and co-operation required for success is achieved.
Targeting the right customers is important to success. CSW operating
companies serve over a quarter million customers on demand rates.
However, customers over 1,000 kW may find the guaranteed prices of
SelectChoice sM preferable to RTP prices supported by hedges and other
forms of price protection. Furthermore, customers smaller than 50 kW are
eligible for SelectChoice sM , and many may elect a trial subscription to gain a
greater understanding of how time differentiated prices match up to their
own usage profile and ability to adjust.

Target Segm ent for SelectC ho ice SM

Figure 3. Target Segment for SelectChoice SM

Pricing in Competitive Electricity Markets 443

For its initial marketing campaign, CSW targeted commercial and

industrial customers with loads between 50 and 1,000 kW for
SelectChoicesM . This segment represents all aspects of commercial activity
in the service territory, ranging from retail establishments to educational
institutions to a variety of light manufacturing operations. It is comprised of
a diverse set of customers, not all of which are necessarily capable of or
motivated to responding to SelectChoice sM prices.
Some will have the native ability to respond to prices like those of
SelectChoicesM , and will be able to realize benefits quickly. Others ration
electricity (either implicitly or explicitly) because they could not justify the
added demand cost of load expansions. SelectChoicesM provides them with
immediate opportunities to better utilize their facilities. Still others will be
approached at a time when they are making important decision about where
to expand their operations, or what energy source to use. SelectChoice sM
may well make the difference between the realization of new, profitable
loads in the operating companies or loss of market share. The challenge is to
find those customers, for whom SelectChoicesM offers the biggest benefits
for them and CSW, and then subscribe them as quickly and cost effectively
as possible.
The SelectChoicesM was initially targeted to customers who appeared
most likely to be able to benefit immediately upon subscription. It was
marketed through a combination of established personal relationships and
marketing channels and new, novel merchandising techniques. Prime
candidates include customers who already have installed load control
technologies, customers who demonstrated interest in RTP but who rejected
that option due to the perceived price risks, customers who have
opportunities to expand or switch fuel sources at better electricity prices, and
customers for whom choice and control are important and are anxious to try
new electricity purchasing plans.
Subsequent marketing efforts will utilize the experiences gained to better
quantify the benefits for specific customer segments. Testimonials from
early adopters will serve as a powerful marketing tool. To reach smaller
customers cost-effectively, CSW will develop mass-marketing campaigns
that appeal to decision drivers identified through program experience and
market research.

7. CUSTOMER RESPONSE TO SELECTCHOICEsM

In the spring of 1997, CSW conducted several focus groups with

customers to assess customer interest in SelectChoicesM and to test
acceptance of the base product design and features. The product options
444 Pricing In Competitive Electricity Markets

prices offered are listed in Table 3 (below). The response was

overwhelmingly positive. Approximately 80 percent of the respondents
surveyed reported that they were likely or almost certain to subscribe to a
pilot program, if one were to be offered by CSW. This clearly indicated that
commercial customers were ready for pricing alternatives, and that
SelectChoicesM was likely to fill that need.
The initial pilot, begun in the summer of 1998 in Oklahoma and parts of
Texas, has served over 300 customers. The customers range in size from a
small agricultural business « 10 kW) to a very large industrial business (> 1
MW). Customers are using SelectChoiceSM to increase electric purchases by
avoiding higher peak prices by shifting their usage to lower cost periods.
For example, a small industrial customer (500 kW) changed work shifts to
avoid the heat of the day and moved operations to lower cost periods.
Another customer replaced propane with electric heat in order to enjoy the
safety and security electricity offers. Generally, the trend has been that
SelectChoice sM subscribers increase purchases in lower cost periods,
resulting in lower average costs to customers, capacity cost savings for non-
participants, and profitable off-peak sales for CSW.

Table 3. SelectChoice SM Prices for the Initial Pilot Offering

Product Number of Peak Price Off Peak Price

Option Peak Hours

A 352 $.0792 .0412

B 352 $.0894 .0408

C 352 $.1015 .0403

D 528 $.0792 .0404

E 528 $.0894 .0397

F 528 $.1015 .0389

G 704 $.0792 .0395

H 704 $.0894 .0386

I 704 $.1015 .0375

Based on these positive early results, CSW elected to seek approval to

continue the pilot program and expand participation in Oklahoma. CSW
will continue to monitor results and will make the decision shortly when to
expand the program to other jurisdictions.
Pricing in Competitive Electricity Markets 445

8. FUTURE PLANS

CSW recognizes that its customers want choices in how they purchase
electricity. The experience with RTP and SelectChoicesM confirms that
properly designed pricing products induce customers to modify their usage
patterns in ways that benefit them and generates benefits to all customers
and to CSW. However, the more product diversity is needed to fulfill the
goal of providing alternatives to all customers. CSW plans to continue its
research and exploration of the full line of ValueChoicesM pricing products
in order to be able to offer its customers the opportunity to gain experience
with purchasing electricity under plans they are likely to encounter under
competitive market conditions.

NOTES

i CSW is comprised of four operating companies. Public Service of Oklahoma, Central Power
and Light, West Texas Utilities, and Southwestern Electric Power Company, who serve
customers in Oklahoma, Texas, Louisiana, and Arkansas.

ii AXS makes this electric merchandising system available under its EMartSM brand. CSW
chose to incorporate the concept into its Customer Choice and Control Program, adopting
the ValueChoice sM brand name.

iii Each customer is assigned a customer usage profile (CUP) that defines how much of its
historical load was used in the peak and off-peak periods of the peak season months June
through September. The total delivery charge is calculated by 'pricing-out' the CUP under
the standard rate (including demand charges and a forecast of fuel adjustments) and then
subtracting from that amount the cost of the CUP peak and off-peak energy amounts
priced out at the applicable SelectChoicesM product option prices. One-twelfth of this
amount is charged each month along with the actual metered peak and off-peak loads
times the corresponding product option prices. The only difference from the
RTP/MarketChoice sM access charge is that under the former the access charge calculation
is made monthly using the actual posted hourly prices. Here, the usage prices are known
in advance and therefore so is the delivery charge.

iv Customers who think that their load might decline during the year, for reasons not related to
SelectChoicesM pricing, weight that potential loss heavily in their decision of whether or
not to subscribe. The guarantee is applied against the entire year's aggregate billings and
is offered to first-time subscribers only.
Index

Access Demand, 307, 315, 318 Customer Services, 335

Advanced Metering, 153 Decrements, 295, 297
Allocation, 307,420 Demand Charge, 288, 307, 315, 318, 319
Anticipating Competitor Actions, 85 Demand Elasticities, 267
ARCH, 191, 193, 194 Demand Reduction, 153
Asset, 211 Demand Response, 153, 160,288,290
Australia, 10,26,62, 127, 149, 160,381, Demand-Charge Rates, 65
397,399,400,403,405,413 Demand-Side Bids, 267
Billing Systems, 47 Deregulation, 5, 7, 30, 184,412
Box-Cox, 191, 192, 193, 194, 195 Derivative(s), 30, 180189,211,229,295,
Brandeis, Louis, 65, 74 299
Break-Even Prices, 5, 17,22 Disco Pricing Strategies, 323
Building Blocks, 5, 10 Distribution, 150,208,244,245,246,
Bundling, 5, 27 315,323,324,331,333,335,336,338
Call, 211, 221, 222 Downside, 165, 177, 178
Capacity, 35, 37, 253, 307, 310, 317, 320 Earnings Stability, 323
Caseload, 307 Econometric, 191, 196, 263
CBL, 295, 296, 297, 300, 301, 303, 305, Economic Efficiency, 307
307,313,315,316 Elasticity of Substitution, 359
cm, 295, 298, 305 Electric Deregulation, 153
Class Responsibility, 307 Electric Rate Design, 65
Competition, 39, 40, 100, 103, 124, 125, Electric Utility History, 65
150, 151, 152,267,293,307,344, Electricity, 5, 12,30,39,46,49,54,70,
381,386,389,399 71,79,80,85,100,103,104,105,
Competition and Rate Design, 65 107, 110, 123, 124, 125, 127, 128,
Connection Services, 335 129, 130, 132, 150, 151, 152, 155,
Correlation, 165, 168, 191, 211, 239 165,169,170,249,267,271,293,
Co-Variance, 165 295,317,359,368,369,381,382,
Customer Choice, 25, 30, 47, 60, 61, 63, 384,387,392,393,397,399,403,
311,397,405,414,445 412,413,414,415,417,421,422,
Customer Preference(s), 413, 415 427 426,427,431

447
448 Index

Electricity Demand, 359 Load Management, 307

Electricity Price Forecasting, 249 Load Profiling, 47
Electricity Pricing, 49, 359, 413, 426 Load Shape Flexibility, 375
Electricity Submarkets, 381 Load Shapes, 373, 375
Energy Management, 307 Load-Following, 307
Energy Pricing, 153, 183 LOLH,307
Energy Service Provider, 5 Lowering Operating Costs, 323
England, 26, 62, 75, 103, 106, 107, 122, Marginal Cost, 65, 110,295,307
123, 127, 128, 129, 130, 132, 136, Market, 29, 31, 39, 41. 43, 47,50,51,54,
148,149,150,151,152,267,268, 59,80,85,86, 87, 88, 100, 103, 105,
273,293,397,399,400,401,402, 107, 117, 122, 123, 124, 125, 127,
405,406,408,411,413 129,148,150,151,152,153,155,
Financial Instruments, 21, 186, 187, 307 156,159,160,161,162,165,169,
Flexible Pricing, 427 173,187,197,198,200,201,204,
Forecasting, 136, 137, 138, 139, 140, 211,231,238,289,293,299,375,
150,249,263,267,269 378,381,384,387,393,401,407,
Forward, 5,10,17,18,152,197,198, 413,414,415,425,433,440
203,204,205,207,208,211,217, Market Clearing Price, 197, 200, 204
218,219,220,227,237,238,413 Market Design, 127, 381, 407
Forward Contract(s), 5, 217 307 Market Experiments, 85
Forward Curve Development, 197 Market Power, 100, 103,107,117,123,
Fuel, 171,201,211. 212, 227,431 124,148,151,153,156,159,160
Future,45,211, 333,379,386,408,445 Market Power Concentration, 103
Game Theory, 85, 90,99, 100 Market Research, 415
GARCH, 191, 193,194,195 Market Simulations, 85
Generation, 61,107,113, 115, 116, 120, Market-Based, 307
151,165,171,174,175,202,211, Market-Based Pricing, 47, 415
309, 382 Marketing, 90, 100, 349, 350, 400, 403,
Greeks, 165, 232 414,427,434,441
Guaranteed Price Contracts, 5 Monte-Carlo, 165, 169
Hedge, 211 Natural Gas Pricing, 375
Herfindahllndex, 103, 115 Neural Networks, 249, 263
Hopkinson Rates, 65 Non-Linear, 307
Hourly Capacity, 307, 317 Off-Peak, 237, 307
Incremental Costs, 335, 340 One-Part, 307,308, 314,316
Increments, 295 On-Peak,252,254,307
Industry Privatization, 381 Option, 198,211,220,221. 222, 227,
Information Technology, 47 313,378,444
Insull, Samuel, 65, 72, 74, 80, 81 Peaking, 307
International Markets, 127 PJM Prices, 249
Internet, 27, 29, 53, 55, 153, 154, 159, Positioning, 349, 425
162, 184, 189,353,398,401,405, Power Pooling, 381
408,410 Power Sector Liberalization, 381
Interruptible, 13, 15,307,313,331 Power Supply, 307, 311
Lifecyc1e, 349, 350, 354 Price Averaging, 307, 309, 311
Load Characteristics, 307 Price Caps, 5, 13, 19,305
Index 449

Price Competition, 5, 100 Retail, 5, II, 12, 13, 18,22,28,30, 54,

Price Elasticity. 201, 359, 376 85,95,100,128,165,172, 183, 186,
Price Floors, 5, 20 188,208,267,271,293,333,350,
Price Forecasting Methods, 249 375,397,407,426
Price Incentives, 307 Retail Energy Products and Services, 183
Price Response, 297, 307, 359 Retail Pricing, 13,85,267,375
Price Signals, 39, 161 Revenue Erosion, 307, 312
Price Spikes, 153 Revenue Retention, 323
Price Volatility. 127,238,239.240,241, Risk, 5, 9, 10, 13, 15, 16, 17,22,23,24,
242 25,29,165,167,168,177,180,211,
Pricing, 5,9,14,21,22,25,30,31,33, 214,215,216,232,234,235,239,
35,36,37,39,41,49,55,59,62,63, 295,303,307,311,313,375,376,
70, 152,267,293,295,307,310,311, 398,402,418,438,439,440
323,326,330,331,332,333,335, Risk Aversion, 375
341,343,349,350,356,357,381, Risk-Based Pricing, 5
413,422,427,432,438 Seasonal, 13, 14,257, 307, 388, 391
Pricing Options, 267, 293, 381,413,426 SelectChoicesM , 427, 428, 434, 436, 437,
Pricing Products, 427 438,439,440,441,442,443,444,445
Product Development, 349 Settlement Protocols, 47
Product Differentiation, 5, 100 Simulation, 165, 168, 175, 197
Product Mix, 23, 24, 25, 28,86,97, 100, Speculate, 211
375,377 Spot Price, 5, 13, 14, 130. 132, 137, 138,
Profit at Risk, 165, 176, 178 289,299
Profit Margin, 349 Spot Pricing, 267
Profitability, 111, 118, 119,375,378 State Rate Regulation, 65
Purchased Power, 307 Strategic Behavior, 127
Put,48,211,221,222 Structural Modeling, 197, 199
Rate Design, 307, 314, 315, 319 Technique, 191
Rate Options, 307, 312 Time Series Analysis, 127
Rationing, 307 Time-of-Day Rates (TaD), 65, 307, 309,
Real-Time Pricing (RTP), 13, 14, 15, 16, 312
27,30,62,295,296,298,300,301, Time-of-Use Pricing (TaU), 153,359,
302,303,304,305,307,308,309, 427
310,311,312,313,314,315,316, Transformation, 191
317,318,319,320,321,427,428, Two-Part, 295, 307, 308, 314, 315, 316
432,433,434,438,442,443,445, Unbundling, 53,307,333,336
Regional Cooperation, 381 Uncertainty Analysis, 197
Regulation and Rate Design, 65 Utilities, 45, 48,51,53,58,75,77,80,
Resource Use, 307 81,123,151,161,162,165,184,333,
Restructured Electricity Markets, 127, 357,404,414,429,445
267 Value, 5, 24, 27, 29,30,46, 100, 108,
Restructured Markets, 39 168, 180, 183,211,214,215,239,
Restructuring, 26, 30, 33, 52, 103, 150, 259,261,262,307,323,332,397,
409,413,414 398,399,400,403,405,407,408,
413,415,421,422,424,426,434,435
450 Index

Value of Service, 323 Wales, 103, 106, 107, 122, 123, 127, 128,
Value-Added Services, 5, 29, 397, 398, 129, 130, 132, 136, 148, 149, 150,
400,403,405,408,413,415,426 151,152,267,268,273,293,397,
ValueChoice sM , 427, 433, 434, 435, 436, 399,400,401,402,405,406,408,
445 411,413
VAR, 165, 166, 167, 168, 169, 170, 171, Weather Risk, 183, 185
172,173,174,175,176,180,181,182 Willingness-to-Pay, 415
Variance, 165 Win-Win, 295
Volatility, 165, 195,211,239,295 Wires Services, 335, 338, 340, 342, 343
Wright Rates, 65