This page intentionally left blank

Political Representation

Political representation lies at the core of modern politics. Democra-

cies, with their vast numbers of citizens, could not operate without
representative institutions. Yet relations between the democratic ideal
and the everyday practice of political representation have never been
well defined and remain the subject of vigorous debate among histori-
ans, political theorists, lawyers, and citizens. In this volume, an eminent
group of scholars moves forward the debates about political representa-
tion on a number of fronts. Drawing on insights from political science,
history, political theory, economics, and anthropology, the authors pro-
vide much-needed clarity to some of the most vexing questions about
political representation. They also reveal new and enlightening perspec-
tives on this fundamental political practice. Topics discussed include
representation before democracy, political parties, minorities, electoral
competition, and ideology. This volume is essential reading for anyone
interested in the ideal and the reality of political representation.

          is Sterling Professor of Political Science and Henry R.

Luce Director of the MacMillan Center for International and Area
Studies at Yale University.
      .       is John S. Saden Professor of Political Science and
Director of the Yale Program on Democracy at Yale University.
                is Professor of Political Science at Yale Uni-
versity and Professor at the Santa Fe Institute.
         .         is a doctoral candidate in the Department
of Political Science at Yale University.
Political Representation

Edited by
Ian Shapiro, Susan C. Stokes, Elisabeth Jean Wood,
and Alexander S. Kirshner
CAMBRIDGE UNIVERSITY PRESS
Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore,
São Paulo, Delhi, Dubai, Tokyo

Cambridge University Press

The Edinburgh Building, Cambridge CB2 8RU, UK

Published in the United States of America by Cambridge University Press, New York

www.cambridge.org
Information on this title: www.cambridge.org/9780521111270
© Cambridge University Press 2009

This publication is in copyright. Subject to statutory exception and to the

provision of relevant collective licensing agreements, no reproduction of any part
may take place without the written permission of Cambridge University Press.
First published in print format 2010

ISBN-13 978-0-511-76989-4 eBook (NetLibrary)

ISBN-13 978-0-521-11127-0 Hardback

ISBN-13 978-0-521-12865-0 Paperback

Cambridge University Press has no responsibility for the persistence or accuracy
of urls for external or third-party internet websites referred to in this publication,
and does not guarantee that any content on such websites is, or will remain,
accurate or appropriate.
 Contents

List of figures page vii

List of tables viii
List of contributors ix
Preface xi

Editors’ introduction 1
         ,       .       ,
                ,   
         .        

Part I. Representation before representative democracy

1. Hobbes’s theory of representation: anti-democratic or
proto-democratic? 15
           

2. Participation and representation before democracy:

petitions and addresses in premodern Britain 35
 

Part II. Theories of political representation

3. Varieties of public representation 61
 

4. Representative government and popular sovereignty 90

   

5. Making interest: on representation and

democratic legitimacy 111
                 

v
vi Contents

Part III. Representation and inherited injustice

6. Critical liberalism 139
  

7. Settlers and natives in North America 159

 

Part IV. What role for representative quotas?

8. Perverse consequences? The impact of quotas for
women on democratization in Africa 211
 

9. On quotas and qualifications for office 236

 

Part V. Preferences, persuasion, and democratic

representation
10. Electoral representation and the aristocratic thesis 271
    

11. Why does the Republican Party win half the votes? 304
 . 

12. The impact of electoral debate on public opinions: an

experimental investigation of the 2005 New York City
mayoral election 324
 ,
               ,         .     

13. Swing voters, core voters, and distributive politics 342

    .  

Index 358
 Figures

1.1 Hobbes’s basic theory of representation page 21

1.2 Hobbes’s full theory of representation 22
1.3 A Hobbesian theory of democratic representation 27
2.1 Number of petitions and signatures per session
Source: Taylor (1913) 43
3.1 Representers 63
3.2 Representation 65
11.1 PUNEs and observed policies in two pooled election years 316
11.2 Party-separation “hyperplanes” for the full model and the
two counterfactual elections 320

vii
 Tables

2.1 Petitions to parliament, 1785–1847 page 42

2.2 Public petitioning and addressing campaigns, 1640–1840 45
8.1 Women in national parliaments by region, 2005 213
8.2 Women’s representation in African countries 217
8.3 Highest ranking countries in terms of women’s
representation, 2003 218
9.1 Three general aims of qualifications with examples 250
9.2 A typology of justifications 253
9.3 Three forms of political equality 261
11.1 PUNEs and the decomposition of racism’s effect (δ0 = 1) 318
12.1 Summary statistics 329
12.2 Impacts of commitment to vote on voting 331
12.3 Summary statistics for commitment to vote treatment 332
12.4 Post-debate summary statistics 333
12.5 Impact of debate on opinions of candidates 336

viii
 Contributors

     .      is Assistant Professor of Political Science at

Marquette University.
    .   is Distinguished Professor of Political Science at the
University of California, San Diego.
            is the Carolyn S. G. Munro Professor of Political
Science and Senior Fellow of the Hoover Institution at Stanford
University, and is a Visiting Professor at New York University School
of Law.
            is Associate Professor of Political Science at Yale
University.
             is Professor of Political Science at the University of
the Witwatersrand.
                  is Associate Professor of Political Science
at Washington University in Saint Louis.
            is Professor of Political Science at the University of
Toronto.
         .         is a doctoral candidate in Political Science
at Yale University.
           is Professor of History at Warwick University.
              is Herbert Lehman Professor of Government
and Professor of Anthropology at Columbia University.
                   is Professor of Economics at Harvard
University.
            is the Laurence S. Rockefeller University Professor of
Politics and Human Values at Princeton University.

ix
x Contributors

             is Associate Professor of Political Science at

University of Washington in St. Louis.
     .       is Elizabeth S. and A. Varick Stout Professor of
Political Science and Economics at Yale University.
                 is Damon Wells Professor of International
Politics at Yale University.
            is Senior Lecturer in Political Theory at
Cambridge University and Staff Fellow in Politics at Trinity Hall.
          is Sterling Professor of Political Science at Yale
University.
      .       is John S. Saden Professor of Political Science at
Yale University.
               is Henry Kohn Assistant Professor of
Economics and Assistant Professor of Political Science at Yale
University.
                is Professor of Political Science at Yale
University.
 Preface

This volume is one of seven books to grow out of the Yale Political Science
Department’s initiative on Rethinking Political Order. It emerged from a
conference on “Representation and Popular Rule” that was held at Yale
in October of 2006. The other volumes in the series are Problems and
Methods in the Study of Politics, edited by Ian Shapiro, Rogers Smith, and
Tarek Masoud (Cambridge University Press, 2004); Rethinking Political
Institutions: The Art of the State, edited by Ian Shapiro, Steven Skowronek,
and Daniel Galvin (New York University Press, 2006); Identities, Affilia-
tions, and Allegiances, edited by Seyla Benhabib, Ian Shapiro, and Danilo
Petranovich (Cambridge University Press, 2007); Political Contingency:
Studying the Unexpected, the Accidental, and the Unforeseen, edited by Ian
Shapiro and Sonu Bedi (New York University Press, 2007); Divide and
Deal: The Politics of Distribution in Democracies, edited by Ian Shapiro,
Peter Swenson, and Daniela Donno (New York University Press, 2008);
and Order, Conflict, and Violence, edited by Stathis Kalyvas, Ian Shapiro,
and Tarek Masoud (Cambridge University Press, 2008).
We are pleased to record our gratitude to Yale University for the finan-
cial support that made this initiative possible. Thanks are also due to
two anonymous reviewers for their suggestions, and to John Haslam of
Cambridge University Press for shepherding the volume into print.

xi
 Editors’ introduction

Ian Shapiro, Susan C. Stokes, Elisabeth Jean Wood,

and Alexander S. Kirshner

Certainly, gentlemen, it ought to be the happiness and glory of a repre-

sentative to live in the strictest union, the closest correspondence, and
the most unreserved communication with his constituents. Their wishes
ought to have great weight with him; their opinion, high respect; their
business, unremitted attention. It is his duty to sacrifice his repose, his
pleasures, his satisfactions, to theirs; and above all, ever, and in all cases,
to prefer their interest to his own. But his unbiassed opinion, his mature
judgment, his enlightened conscience, he ought not to sacrifice to you,
to any man, or to any set of men living. These he does not derive from
your pleasure; no, nor from the law and the constitution. They are a
trust from Providence, for the abuse of which he is deeply answerable.
Your representative owes you, not his industry only, but his judgment;
and he betrays, instead of serving you, if he sacrifices it to your opinion.
Edmund Burke, 3 November 1774 (Burke 1999)

How should we reconcile the ideal that citizens in democracies aspire to

control their political representatives with the reality that elected repre-
sentatives have substantial leeway to act as they themselves see fit? This
leeway, which is sometimes described as the monitoring problem, is often
portrayed as a defect of the representative process – to be minimized if
it cannot be abolished. But, as Burke’s speech to the electors of Bristol
suggests, this is not the only possible view of the matter. Indeed, even
today few would embrace, in its starkest form, the suggestion that repre-
sentatives should be mere transmission belts for constituent preferences.
Burke’s formulation might seem like a quaint throwback to an era more
elitist than ours. Yet representatives who are excessively responsive to the
electorate often confront charges of pandering and flip-flopping.
Our goal in this volume is to explore the relations between repre-
sentation and democracy, better to understand the tensions between
them, and to help work toward an understanding of representation that
is both satisfying in its own right and comfortable with democratic
understandings of political legitimacy. This agenda leads our contrib-
utors to explore predemocratic conceptions of representation, various
contemporary theories of representation, the interactions between
1
2 Ian Shapiro et al.

representation and either replicating or altering inherited injustice

(including the effectiveness of, and justification for, representative quo-
tas), and the relations among preferences, persuasion, and democratic
representation. We do not claim to supply definitive answers to many of
the vexed issues raised in these areas. But our contributors move the dis-
cussion forward in ways those looking for distributive answers will have
to take into account.

I. Representation before representative democracy

We usually think of political representation as closely linked to modern
conceptions of democracy. This is reasonable, since the ancient Greek
view of democracy was based on the idea of ruling and being ruled in
turn – scarcely a representative notion. But, as David Runciman demon-
strates in our opening chapter, representation has an older pedigree than
modern democracy. Indeed, Runciman argues that Thomas Hobbes
developed an account of representation that opens up alternative per-
spectives on the monitoring problem. Leviathan might seem to be a
surprising point of departure here, because Hobbes vigorously rejected
democracy and mechanisms of democratic accountability. Contra Hanna
Pitkin’s celebrated account of Hobbes’s theory of representation, Runci-
man contends that Hobbes viewed authorization and representation as
two distinct and potentially separable activities: individuals authorize the
sovereign to act, and in so doing renounce almost all claim to repudiating
the sovereign or his actions. Yet the sovereign, on Runciman’s interpre-
tation, does not represent the individuals who have authorized him qua
individuals; he represents the collective body of them, the Common-
wealth.
Representative democracy requires that citizens retain the ability to
eject representatives. Yet even democrats, Runciman suggests, may rec-
ognize that government officials do not represent individual citizens, but
rather a collective entity – such as the people, the district, the state, or
some other. From this perspective, the concerns raised by discrepan-
cies between citizens’ preferences and the actions of their representatives
ought to be assuaged, in that the government does not represent the indi-
viduals to whom they are electorally accountable. Rather, the government
represents and acts for “the state” or the “the people.” Runciman argues
for the merits of Hobbes’s model in which publics authorize a govern-
ment to establish a newly created entity – e.g. the United States. The
model also fits with the iterative process by which representatives repre-
sent the state and citizens respond to the state’s actions. Though Hobbes
certainly would have resisted a democratic interpretation of his theory,
 Editors’ introduction 3

Runciman suggests that his account of representation may provide a basis

for a more persuasive model of democratic representation than theories
which treat representation as if it involves only representatives and the
represented.
Yet there is no denying that Hobbes offers us a sharply delimited con-
ception of the appropriate political role for citizens. In Chapter 2 Mark
Knights demonstrates that Hobbes’s sparse characterization of popular
engagement lived in some tension with the actual situation in predemo-
cratic Britain. To be sure, only a small proportion of British subjects
could vote in premodern parliamentary elections. Moreover, these elec-
tions were not intended to give “the People” control over the members
of parliament who represented them. Yet, as Knights illustrates, between
1785 and 1847 parliamentary elections were just one of many modes of
political representation available to the British public. Local represen-
tative bodies, associations, societies, unions, and petitions complicated
and checked the ideal of exclusive parliamentary representation. These
alternative, and in some cases competing, institutions provided an extra-
electoral means of communicating collective political opinion not only
to the public, but also to the political elite at Westminster. In an era of
comparatively limited suffrage these forums also stood as challenges to
parliament’s claim to represent the people.
Beginning with a brief overview of extra-parliamentary forms of rep-
resentation, Knights focuses on petitions and addresses. Because they
threatened parliament’s legitimacy, the use of the petitions sparked a
debate about the relative stature and authority of formal arenas of rep-
resentation, including parliament, and informal mechanisms, including
the petition and the address. Knights demonstrates that even in an
era in which formal institutions did not facilitate the responsiveness
of representatives, British subjects created informal mechanisms that
influenced parliament and more dramatically competed with Westmin-
ster’s formal representative monopoly. Knights’s analysis raises important
questions about the mechanisms linking informal and formal represen-
tative institutions, as well as about the conditions in which the relation-
ship between them is complementary and the conditions in which it is
competitive.

II. Theories of political representation

Many of us might agree with Runciman that democratic governments
represent corporate entities, but few would go all the way in severing the
link between representatives and the electorate. If there is a link, what
is it? Philip Pettit takes up this question in Chapter 3, distinguishing
4 Ian Shapiro et al.

between the various forms that democratic representation can take.

He divides representatives into four groups depending on whether they
represent as individuals or within a group, whether they act cooperatively
or uncooperatively, and whether, if they act cooperatively, the represen-
tation is corporate in nature (e.g. the British parliament). Similarly, on
his account, those who are represented may be individuals or groups,
and when they are groups, they may or may not be cooperative or cor-
porate. In addition to these distinctions, Pettit outlines three varieties of
relationship between representative and representee (the term which Pet-
tit, following seventeenth-century usage, deploys) which he describes as
simulative, enactive, and interpretative. Roughly speaking, these types of
representation correspond to standing for someone, acting for someone,
and speaking for someone.
Pettit acknowledges the attractiveness of enactive representation –
acting for – because of the hope it holds out of solving the monitoring
problem. Enactive representation corresponds with the idea of popular
control over political institutions. He insists, however, that in democra-
cies representatives must inevitably make decisions concerning matters
about which representees, either as individuals or as collectivities, have
no established preferences. Thus the actions of the representative, at least
from the perspective of enactive representation, will not track the wishes
of the represented. What is required, Pettit maintains, is for the represen-
tative to speak for the represented; the representative must engage in inter-
pretative representation. If this is not Burkean paternalism, it is surely
a step in that direction. This raises the question whether representatives
are free to fill in preferences where their constituents’ preferences are
missing or deemed by the representative to be incoherent or ill-formed,
and in what sense this form of representation is democratic. Pettit argues
that it can be considered democratic only if decisions are consistent with
a set of reasons that all of those represented would recognize as relevant
to answering the question at hand.
How are Pettit’s distinctions cashed out in actual politics? Pettit
presents ideal-typical models of the US and British governments, draw-
ing on his typology to illustrate how these systems can be characterized
and their differences demarcated. Washington prioritizes individual rep-
resentation of the nation (the President), of states (Senators), and of dis-
tricts (House members). By contrast, the highly disciplined Westminster
system prioritizes a corporate form of representation for the entire coun-
try (parliamentary government). Citing empirical findings that attribute
variation in legislative and political outcomes to these institutional dif-
ferences, Pettit concludes that the representational priority expressed by
 Editors’ introduction 5

a polity’s institutional set-up will profoundly shape key political relation-

ships and important substantive outcomes within that polity.
In Chapter 4 Bryan Garsten questions a central premise of Pettit’s
essay: that democratic representation should be valued because it offers
a possible solution to the monitoring problem. Garsten maintains that
the “purpose” of representative institutions is to “multiply and challenge
government claims to represent the people.” Garsten explores the liberal
and democratic logic of this conception of representative government by
examining the writings of Jean-Jacques Rousseau, Benjamin Constant,
and James Madison.
Garsten argues that although Rousseau was a committed defender of
popular sovereignty, he nonetheless rejected Pettit’s simulative represen-
tation. Fearing that representative governments would usurp the peo-
ple’s rightful claim to full political sovereignty, Rousseau favored indirect
government. Garsten offers a novel interpretation of the classic debate
between Constant and Rousseau, in which Constant did not merely reject
Rousseau’s conclusions. Indeed, they agree on basic questions about
representation. Garsten explains that Constant, like Rousseau, feared
the tyrannical outcomes that can arise when democratic governments
claim full political sovereignty. For Constant, as for Rousseau, institu-
tions should be established to undermine any single popular representa-
tive from claiming to represent the popular will.
Next Garsten explores, in the writings of James Madison and Alexan-
der Hamilton, the sentiment that political institutions should encour-
age debate about who rightfully speaks for the people. Garsten agrees
with them that true demagogues – rulers successfully claiming the sup-
port of the people – constitute a grave threat to representative legiti-
macy. By multiplying the number of government officials who can with
some plausibility claim to represent the people, representative institu-
tions encourage disputes and conversation about what the popular will
is and who may rightly claim popular authority. Our inherently multi-
vocal forms of government, Garsten argues, are therefore most successful
when cacophonous debate about who is the rightful representative of the
people drowns out the voice of any particular demagogue. That is how
ambition, in Madison’s words, is made to counteract ambition.
In Chapter 5 Clarissa Rile Hayward develops a different challenge
to the view that democratic representation is fundamentally about solv-
ing the monitoring problem. On her account this view fails to account
for the many ways in which the actions of representatives actually shape
their constituents’ interests, rather than simply reflecting preexisting
preferences and interests. Hayward uses the example of structural
6 Ian Shapiro et al.

inequalities – inequalities in resources and opportunities built into

law – to illustrate the shortcomings of conventional views of the monitor-
ing problem. Because common aggregative and deliberative approaches
to democracy theory overlook how political interests are shaped, pro-
ponents of these theories fail to recommend institutional reforms that
are likely to promote equality or inclusiveness in government. Hayward
makes a strong case that merely tinkering with representative forums is
unlikely to create incentives for representatives to challenge the struc-
turally entrenched inequities that shape their constituents’ lives.
Instead Hayward argues in favor of a contestatory ideal in which polit-
ical institutions are legitimate if they promote “free and equal struggles
over collective norms and principles.” Her contestatory approach does
not take interests for granted. Rather, institutions should encourage the
formation of interests that undermine both inequalities and fixed hier-
archies. She advances this case by exploring the example of political
jurisdictions in the United States. The growth in suburban municipali-
ties, at the expense of cities, deepened existing inequities (e.g. in public
schooling), limiting people’s ability to challenge policies that affected
them via the democratic process. An approach that understood inter-
ests as politically constructed would not only advocate for more effective
forms of representation within existing jurisdictions, but it would also
militate in support of changes in those jurisdictions. Such an approach,
Hayward contends, would put citizens on a more equal footing with one
another. It would also make it easier for the disenfranchised to confront
institutional barriers to political equality.

III. Representation and inherited injustice

The historical role of states in creating identities and fixing interests is
also at the heart of Courtney Jung’s discussion in Chapter 6, entitled
“Critical liberalism.” Jung argues that existing accounts of group rights,
including those that grapple with the appropriate extent and structure of
group representation, neglect the ways states create ethnic groups and
how states establish unjust institutional arrangements on the basis of
group distinctions. Recognizing this, she argues, provides us with greater
purchase on what the state owes to whom. Jung’s analysis falls into three
parts. First she challenges the assumption that groups and identities
which constitute society are natural – a perspective the author attributes
to theorists such as Will Kymlicka, John Rawls, and Ronald Dworkin.
Naturalizing groups and identities creates an impulse to privatize or pro-
tect ethnic groups, with the result that culture and democratic politics
are insulated from one another. As an alternative, Jung recommends that
 Editors’ introduction 7

we investigate where human attachments come from. In doing so, we

will not only trace how power has been exercised, but also identify the
historical obligations of states.
Like Hayward, Jung is particularly concerned with identifying and
reforming institutions that were shaped by, and in turn reproduce, struc-
tural injustice. Jung argues that claims for redistribution, special modes
of representation, or institutional reconfiguration should be responsive
to the specific conditions and policies that shaped the relevant ethnic
group. Jung’s chapter builds toward a conception of individual rights
that is consistent with her theory’s focus on politically constructed iden-
tities and structural injustice. She argues in favor of membership rights,
whereby individual rights accrue to members of structurally disadvan-
taged groups. Membership rights embody an attractive third way between
liberal rights regimes, geared toward preserving inequality on the basis of
an ostensible commitment to neutrality, and group rights regimes aimed
at shielding unjust group practices from liberal critique. Ultimately, Jung
contends that our normative accounts of democracy and representation
must address the challenge posed by the following question: “How should
democratic institutions process the political claims that arise to protect
and contest the exclusions and inclusions set in place by the modern state
itself?”
In Chapter 7, Mahmood Mamdani examines a stark instance of how
structural injustices become entrenched within and then reproduced by
representative institutions. The national system of representation of the
United States comprehends two versions of sovereignty. On the one hand,
in many cases the decisions of federal organs take precedence over the
decisions of lower bodies – such as local and state governments. This hier-
archy is consistent with traditional, unitary conceptions of sovereignty.
On the other hand, states are thought to retain some element of the
sovereign status they compromised when the United States was founded
in the eighteenth century. The representative institutions of the United
States – including both Houses of Congress and the presidency, whose
incumbent is elected by a state-based electoral college – reflect the inge-
nuity employed to create a single representative political system contain-
ing multiple permutations of political sovereignty.
Mamdani notes that, throughout the history of the United States,
Native Americans were forced to inhabit a perpetually fluid nether status
as wards of the United States – neither guaranteed rights as full citi-
zens, nor recognized as members of free communities within the United
States. Not only were Native Americans denied protection under the
Constitution, their independence and claims to land were systematically
undermined by the state. The singular democratic creativity employed to
8 Ian Shapiro et al.

reconcile federal and state sovereignty was never applied to shape institu-
tions that would recognize Native Americans as members of alternative
“sovereign” entities. In devastating contrast to the Founders’ remarkable
institutional imagination, Mamdani argues that the institutional and legal
relationship of the United States to Native Americans mimicked destruc-
tive regimes imposed on others by colonial Europe.

IV. What role for representative quotas?

If representation that reproduces structural injustice is infirm, what
should the remedy be? That question is taken up in the next two chap-
ters by authors who explore whether representative quotas and qualifi-
cations are a defensible strategy for reforming representative institutions.
In Chapter 8, Shireen Hassim analyzes the ways in which quotas for
women may paradoxically hinder their effective representation and sub-
stantive equality. Her focus is on South Africa, Uganda, and Rwanda,
where inherited structural injustice seems to offer compelling reasons to
embrace gender-based quotas so as to ensure a substantial presence of
women in their respective legislatures.
Yet South Africa, Uganda, and Rwanda are all effectively single-party-
dominant systems, even though the dominant political groups within
the countries established the gender-based quotas. Hassim identifies two
broad mechanisms through which quotas might undercut further democ-
ratization and the achievement of substantively egalitarian policies. First,
the women elected to these parliaments owe their position to the dom-
inant party. Consequently, they have relatively little political autonomy
and therefore often end up supporting the policy of the government
against the claims of women’s groups in civil society. Despite their lack
of political independence, the presence of female representatives in the
legislatures provides the dominant parties with a special claim to political
legitimacy. Second, quotas might have the effect of increasing the num-
ber of women in legislatures without concomitantly establishing a link
between those representatives and a political constituency intent on hold-
ing those representatives accountable for securing political outcomes that
are beneficial to women. Lacking an independent constituency, women
representatives are cogs within the larger party machinery. Though Has-
sim acknowledges that quotas might be a necessary transitional step for
achieving more democratic outcomes, she concludes that they are “not a
sufficient condition by any means.”
Complementing Hassim’s study, Andrew Rehfeld takes up the nor-
mative basis of electoral qualifications and quotas in Chapter 9.
While particular qualifications for elected office might be defended by
 Editors’ introduction 9

reference to past injustices or institutional pathologies, he argues, as a

general matter they ought to be presumed illegitimate.
Rehfeld defines a qualification as any rule or circumstance that has
the effect of creating different probabilities of success for individual can-
didates. Gender-based quotas fall within this definition of qualifications
because they effect the ability of individuals (typically men) to run for
office, as well as the ability of all individuals to choose whom they want to
govern them. There are, of course, some qualifications for office which
are essential to the process of democratic selection – i.e. the qualifications
determined by the voters themselves. Accordingly, Rehfeld distinguishes
between voter endogenous qualifications, which are not presumptively
illegitimate, and voter exogenous qualifications, which are.
Rehfeld then elaborates on the two democratic rights that all exoge-
nous qualifications for office violate: the right to be ruled by someone one
chooses and the right to run for office. The existence of a right to choose
one’s rulers is likely to be intuitive to most democrats, and the relation-
ship between this right and electoral qualifications is straightforward:
qualifications impinge on voter choice by assigning a greater probabil-
ity to a certain kind of candidate. But Rehfeld argues that the equal
right to pursue political office is also implied by the democratic princi-
ple of equality and by a democratic skepticism of aristocracy. While he
does not entirely foreclose the possibility that a polity might be justi-
fied in imposing exogenous qualifications for political office, he makes a
strong case that such action should have to overcome a strict justificatory
test. Ultimately, both Rehfeld’s and Hassim’s chapters are animated by
the shared intuition that the easy embrace of representative quotas and
qualifications as mechanisms of political empowerment gives short shrift
to the potential political and normative costs of this type of electoral
engineering.

V. Preferences, persuasion, and democratic representation

Are there other grounds for preferring indirect forms of representation,
if not Burke’s explicit paternalism? This question is taken up by John
Ferejohn and Frances Rosenbluth in Chapter 10. They investigate the
logical foundations of what they describe as the aristocratic thesis: that
indirect decision-making processes or government by an elite might gen-
erate better outcomes than direct democratic procedures. Ferejohn and
Rosenbluth ask whether representative government produces outcomes
that are different from those that direct democracy would generate, and
whether those outcomes are better or worse for the people. The authors
argue that an agency model – in which the voters are the principals and
10 Ian Shapiro et al.

representatives are the agents – captures central elements of the repre-

sentative dynamic. Still, they point out that such models fail to account
for important aspects of the voter–representative relationship: that voters
constitute a plural agent; that they lack important information about their
representatives; and that, unlike standard principals, voters are unable to
write better contracts for themselves – they can only choose to elect or
not elect. Each of these factors makes it harder for constituents to solve
the monitoring problem as conventionally conceived. They also imply
that representative outcomes will differ significantly from direct demo-
cratic outcomes. Ferejohn and Rosenbluth cite empirical findings that
the introduction of referendums or initiatives – instruments of direct
democracy – into representative democracy tends to push political out-
comes toward the preference of the median voter. These observations
reinforce the conclusion that representative and direct democratic out-
comes differ. Yet do they differ systematically? While Ferejohn and
Rosenbluth do not reject this possibility, they conclude that it remains an
open question whether representative systems produce better outcomes
than do direct ones.
John Roemer takes a different approach to the tension between voter
preferences and the outcome of representative institutions in Chapter 11.
The classic account of the relationship between voter preferences and pol-
icy outcomes was given by Anthony Downs. Downsian models do not
predict that a political party whose economic policies are in the interests
of a small minority of wealthy citizens will be electorally successful. Roe-
mer argues that the Republican Party is such a party, and that a more
complicated electoral model than Downs’s is needed to account for its
long-term electoral viability. After reviewing various explanations for this
counterintuitive outcome, Roemer contends that only three are likely
to have contributed to the long-term success of the Republican Party:
policy bundling (where parties combine less popular policies with more
popular ones); voters’ conceptions of distributive justice; and imper-
fect political representation (i.e. democratic institutions are organized in
such a way that individuals do not have an equal influence on political
outcomes).
Roemer then goes on to describe a four-dimensional model of party
policies and voter preferences that allows him to account for the following
facts: that the Democratic and Republican parties differ on more than
one issue, that there are extreme and moderate factions within both par-
ties, and that voters care about more than one issue (such as taxes, race,
and religion). Roemer uses the model to estimate the policy bundle effect,
which concerns the extent to which voters will support less-preferred
economic outcomes (e.g. low taxes) to achieve their preferred outcomes
with respect to race, and the anti-solidarity effect – how much they simply
 Editors’ introduction 11

believe that African-Americans abuse the social welfare state, and vote
against redistributive policies on that basis. To estimate the size of these
effects, Roemer ran counterfactual elections in which one or both effects
were removed from the calculations of the modeled voters. The esti-
mated effect of these race-based factors on the vote totals of the Repub-
lican Party in presidential elections was significant. The results suggest
that Republicans demonstrated significant political savvy by shifting the
political debate away from economic issues. Roemer’s finding also adds
weight to the increasingly familiar claim that racism has generated the
difference in social spending between the United States and advanced
European democracies.
Sendhil Mullainathan, Ebonya Washington, and Julia Azari also exam-
ine the mechanisms linking the political preferences of American voters
with their electoral choices in Chapter 12. In theory, televised polit-
ical debates provide relevant information that allows voters to make
choices reflecting their preferences. To investigate this potential effect,
the authors conducted a randomized experiment measuring the impact
of a 2005 New York City mayoral debate on likely voters’ opinions about
the mayoral candidates (Michael Bloomberg and Fernando Ferrer). Pre-
vious examinations of the effects of debates typically attempt to gauge
the influence of debates by conducting before-and-after viewer inter-
views. Such studies, however, suffer from familiar problems of potential
endogeneity – i.e. debate viewers may be systematically more or less
likely to have their views changed by a debate than the voting populace at
large. This study was not subject to the same problems, however, since
the participants were randomly assigned to watch a debate or a placebo
program (The NewsHour with Jim Lehrer). Using before-and-after tele-
phone surveys, the authors found that those watching the debate were
more likely to report that their opinions of the candidates had changed.
Despite the reported difference, those who watched the debate had the
same opinions as the control group when asked specific questions about
the candidates. This finding suggests both that the debates were uninfor-
mative and that voters seemed not to grasp how their own opinions were
influenced by the debate. Such a result raises serious questions about
the usefulness of political debates, about the mechanisms through which
they might actually influence political outcomes, and about how voters
incorporate new information.
John Roemer argued that voters are not only persuaded by parties’
economic policies but by their stance on other issues, such as those asso-
ciated with race. In Chapter 13 Gary Cox argues that models that focus
only on the attempt of parties to persuade voters miss a number of other
partisan activities which may be as important as persuasion in the pur-
suit of reelection. These activities mobilize reliable voters and coordinate
12 Ian Shapiro et al.

voters, thus serving to prevent, or at least impede, ideologically similar

start-up parties from challenging incumbent parties.
Cox’s chapter takes particular aim at the argument, made by Susan
Stokes, that core voters, who favor a party on partisan or programmatic
grounds, cannot credibly threaten to punish their party if it withholds
distributive rewards. The implication of Stokes’s argument is that parties
will focus on swing, not precommitted core voters. In contrast to Stokes,
Cox contends that core voters possess two ways of credibly threaten-
ing their preferred parties: by abstaining or by voting for a competitor
party on the same part of the political spectrum. These two defection
strategies in turn map onto two activities that parties must undertake to
limit defections: mobilization and coordination. Full models of parties’
political activities, Cox argues, should find that as long as mobilizing and
coordinating voters are seen as effective strategies for winning elections,
parties will shift their focus from swing to core voters.
Cox extends these insights into the legislative arena. Legislative parties
do not just convince swing legislators; they also mobilize and coordi-
nate their members. As in the electoral arena, core members or those
responsible for mobilizing and coordinating other members (e.g. whips
or committee chairmen) should be the chief recipients of pork from the
party. If Cox is correct, models which focus only on persuasion are neces-
sarily biased toward finding that parties focus on attracting swing voters,
rather than motivating core voters. The analytical payoff here is criti-
cal to understanding representation in partisan political systems. If our
accounts of party activity mischaracterize what parties do, then they will
necessarily mischaracterize whom they represent.
The debates among our contributors do not converge on a single
answer to the monitoring problem that was thrown into sharp relief by
Burke’s speech to the electors of Bristol. Rather they suggest that it is
in fact not a single problem. Different voters, indeed even different sup-
porters of a single representative, stand in very different relations to that
representative. And representatives often confront multiple, conflicting
imperatives vis-à-vis the same voters. Collectively, these authors build a
convincing case that we need a better understanding of these complex
relations in their multifarious parts before aspiring to develop any general
theory of representation.

BIBLIOGRAPHY
Burke, Edmund. 1999. “Speech to the Electors of Bristol,” in Selected Works of
Edmund Burke. Indianapolis, IN: Liberty Fund.
Part I

Representation before representative

democracy
1 Hobbes’s theory of representation:
anti-democratic or proto-democratic?

David Runciman

I. Hobbes, representation, and democracy

Thomas Hobbes occupies a deeply ambivalent position in the history
of modern conceptions of political representation. On the one hand, he
is often credited as the thinker who did most to make representation a
distinctively modern concept, by emancipating it from its medieval roots
and employing it to establish a clear identity for the state as a separate
entity in its own right (Skinner 1989, 2002). Hobbes used the idea of
representation to ground a secular conception of political authority, and
in doing so rescued the idea of political rule from various intractable the-
ological controversies.1 In this sense, Hobbes’s thought is foundational
for modern theories of representative government. But for all its apparent
modernity, Hobbes’s theory of representation suffers from one obvious
flaw when judged by the standards of contemporary politics: it appears
to be strikingly anti-democratic, and it is very hard to see how an anti-
democratic theory can also be viewed as foundational for the political
world we now inhabit.
This dual role occupied by Hobbes – both foundational but also oppo-
sitional in his perceived relationship to representative democracy – is best
captured by his treatment in what remains the most widely cited work
on the theory of political representation in English, Hanna Pitkin’s The
Concept of Representation (Pitkin 1967). There, Hobbes is treated first of
all as the thinker who did most to clarify the central feature of modern
political representation – that it is an “authorization” concept (or what
Pitkin goes on to call a “substantive” concept), in which political repre-
sentation emerges as a form of “acting for” rather than merely “standing
for.” But Pitkin also makes it clear that Hobbes’s theory of representa-
tion is not a viable theory, because it stands in direct opposition to the

1 For instance, it is one of the many striking features of Hobbes’s theory of representation
that his account of theological representation in Leviathan – where he discusses at length
the manner in which God is “represented” by the Trinity – follows from his account of
political representation, and not the other way around (see Hobbes 1996: 339–40).

15
16 David Runciman

democratic presupposition that individuals may object to what is being

done on their behalf by their political representatives. For Hobbes, a
sovereign representative, once authorized, cannot be constrained, and
therefore has absolute power to rule as he sees fit.2 Thus although this
may be an “authorization” concept, it is also far too authoritarian by our
current standards. So, Pitkin argues, Hobbes is clearly “wrong” about
representation, and a different account is needed.
In this chapter, I want to argue that it is possible to close the gap
between Hobbes’s apparently foundational role in the modern theory of
representation and our own democratic expectations of what representa-
tion must entail. I believe that Hobbes’s thought is not just foundational
in the sense that it marks a clean break with the medieval/theological past,
but also because it clearly points the way forward to the democratic forms
of representation that were to follow. By making sense of what is going
on inside Hobbes’s theory of representation, it is possible to establish
the ways in which a Hobbesian framework for representative government
remains the one in which we still have to operate.
In order to make this case, it is important to distinguish the approach
I am going to adopt from a number of other ways in which it might be
claimed that Hobbes is a more “democratic” thinker than he is usually
perceived. One way in which this might be done is to argue that Hobbes’s
own ideas were designed to be more democratic than commentators have
sometimes recognized. A version of this case has recently been made, for
example, by Richard Tuck, who has argued for a much fuller overlap
than is commonly perceived between Hobbes’s political thought and
both Aristotle’s conception of “extreme democracy” on the one hand and
Rousseau’s later democratic theory on the other (Tuck 2006). However,
though this might help to rescue Hobbes from undue pigeonholing as a
straightforward “anti-democrat,” it cannot do the same for his theory of
representation, because it places the focus on Hobbes’s earlier writings in
which his theory of representation is not fully articulated, at the expense
of Leviathan, where it is. Tuck’s case rests on treating De Cive (1642)
as a more significant work than Leviathan (1651), where Hobbes’s fire
is directed against the “democratical gentlemen” whom he blamed for
the catastrophe of the English Civil War. But it is no coincidence that
Leviathan is also where Hobbes firmly identifies representation as the
key to his theory of politics, which is something he does not do in his

2 Although Hobbes allows that the sovereign can either be an individual or an assembly
(this is one of the ways in which his thought is more democratic than is sometimes
supposed), for shorthand here I will treat the sovereign as always a single ruler (and as
male, though again Hobbes saw no reason why this should always be the case). This is
the shorthand Hobbes himself uses.
 Hobbes’s theory of representation 17

earlier works. For this reason, identifying Hobbes in historical terms as a

democrat appears to involve rescuing him from his own theory of political
representation, so that the gap between democracy and representation
remains.3
An alternative approach is to argue that Hobbes’s theory of representa-
tion, while not in itself properly democratic, exercised a strong influence
on the ideas of subsequent theorists of representation whom it is possi-
ble to think of as democratic theorists in their own right. The strongest
version of this argument is one that traces the influence of Hobbes’s
theory of representation on the political thought of the French Revolu-
tion, for example on that of the Abbé Sieyès (Hont 2005). But again, the
case is by no means straightforward, not least because Sieyès drew on
Hobbes in part because he was seeking to distance his own thought from
what he saw as the obvious perils of “crude” democratic theory.4 It is
also striking that some of the theorists who have drawn most heavily on
Hobbes’s theory of representation are often the ones that many contem-
porary democrats would be least comfortable identifying with. This is
particularly true of Carl Schmitt, whose self-declared “democratic” the-
ory of representation, with its emphasis on plebiscitary leadership and
extra-legal forms of political authorization, while echoing Hobbes in a
number of respects, hardly rescues Hobbes from the suspicion that he is
not a genuine democrat in anything but the most minimal sense of the
term (Schmitt 1996). Hobbes’s influence on Schmitt recapitulates rather
than resolves the problem.
The case I want to make here is not primarily a historical one, and I
do not wish to argue that either the earlier variations or the subsequent
impact of Hobbes’s theory of representation are what make his ideas
indispensable for democratic theory. Rather, I want to claim that the for-
mal content of his theory of representation in its best-known and most
clearly articulated setting – Leviathan – has been too narrowly under-
stood, by being too closely identified with his theory of authorization. In
this chapter, I give what I consider to be a fuller account of the formal
workings of Hobbes’s theory, in order to show how Hobbes’s account
allows a distinction to be drawn between representation and authoriza-
tion. With the aid of this distinction, I believe it is possible to establish
a much closer connection than is usually allowed between what Hobbes
has to say about representation and the functioning of representative

3 This is one of a number of reasons why Tuck’s case is not convincing; the others are laid
out in Hoekstra (2006).
4 “For I always maintain that France is not, and cannot be, a democracy . . . France is and
must be a single whole, subject throughout to a common legislation and a common
administration” (quoted in Forsyth 1987: 138).
18 David Runciman

democracy. I am not claiming that Hobbes drew a distinction between

authorization and representation in order to make his theory more demo-
cratic; if anything, the reverse is true, and Hobbes drew this distinction
because of what he saw as the pitfalls of a democratic conception of polit-
ical authority. What makes the connection between Hobbes’s theory of
representation and our own understanding of representative democracy
is that these pitfalls are still there, and it is only with the aid of a distinc-
tion of this kind that it is possible to avoid some of them. Hobbes’s wider
prejudices against democratic politics served in the end to make his the-
ory of representative government unworkable by our standards. In this
respect, he was wrong about democracy. But his understanding of what
a theory of representation requires to be workable at all was essentially
correct, and applies to democratic politics all the same.

II. Hobbes on representation and authorization

Pitkin bases her claim that Hobbes is a straightforwardly anti-democratic
thinker on his theory of authorization, as laid out in Leviathan. Viewed in
these terms, the case is a compelling one.5 For Hobbes, to be the “author”
of another person’s actions is to “own” whatever actions they perform;
in other words, it is to be bound by what they do (Hobbes 1996: 112).
Hobbes believed that it was possible to set limits on what one person
might do on the authority of another – for example, by authorizing them
to act only in certain ways or within certain limits – but that this was only
possible within the context of civil society, where contractual arrange-
ments of this kind could be enforced by the sovereign. By definition,
the act of authorizing the sovereign itself could not be subject to such
limitations, because there was no power in place to enforce them. The
ability of the sovereign to hold individuals to their contracts depended on
the fact that sovereign authority was unlimited, and therefore binding.
So Hobbes states that the individuals who authorize a sovereign to act on
their behalf must be the owners of anything and everything the sovereign
does in their name. Hobbes accepts that there are some circumstances
where this arrangement becomes absurd, as when the sovereign orders
the death of an individual subject: given that it is nonsensical for individ-
uals to command their own death, it cannot be plausible that they should
treat the command uttered by someone else as though it were their own
(ibid.: 151). They may, under such circumstances, resist, which would

5 Though Pitkin does not consider the historical background to this argument (as I also
do not here). For a full account of the polemical purposes of Hobbes’s theory of repre-
sentation in Leviathan, see Skinner (2005).
 Hobbes’s theory of representation 19

be to repudiate the authority of the sovereign. But this caveat is a very

slender basis on which to construct any significant limitation of sovereign
power, because all other subjects of the sovereign are still bound to obey
the command, so that the resistance of the individual freed from their
obligations will almost certainly be futile.
The fact that Hobbes allows individuals to disown the actions of their
sovereign when faced with death is not enough to make the theory in
any sense democratic, since democracy requires not simply that we can
disown the actions of our representatives but that we can constrain them
in some way. Hobbes’s sovereign remains unconstrained, because the
disowning of sovereign actions is either futile – when only a few people
are threatened – or redundant – as when the population at large is under
threat – because a sovereign who threatens the majority of his subjects
with death has already returned them to a state of nature, where political
authority is void. Pitkin is therefore right to argue that Hobbes’s account
of political authorization leaves no room for individuals to have a say in
what is being done on their behalf by their sovereign. But she is wrong to
suppose that this constitutes the whole of Hobbes’s account of political
representation. For what is striking about the version Hobbes gives in
Leviathan is that he does not on the whole use the language of repre-
sentation to describe the relationship between an individual subject and
their sovereign. He is careful to limit the description of that relationship
to the terminology of authorization and ownership. He talks about rep-
resentation primarily when he is discussing the relationship between the
sovereign and all his subjects, taken as a whole. Individual acts of autho-
rization give the sovereign the right, not merely to act on the authority of
each individual, but additionally to “Present the Person of them all (that
is, to be their Representative)” (ibid.: 121, italics in original). Though
this is a somewhat ambiguous sentence – the words “them all” might
refer either to a group of individuals qua individuals or to a corporate
entity – it is striking that Hobbes does not say that the sovereign repre-
sents the “persons” of them all. This makes a corporate reading more
plausible. In these terms, representation follows from the authorization
of the sovereign but is not strictly equivalent to it, because it is not the
individual authors of the sovereign who are represented by his actions
as individuals, but the commonwealth itself. When Hobbes talks about
the sovereign as “representative” in Leviathan (see, for example, ibid.:
128, 184), he emphasizes that what he represents are not the separate
persons of his individual subjects, even though these are the people who
individually “own” his actions; rather, what he represents is the state,
which is what their joint ownership of his actions has brought into being.
And it is crucial for Hobbes’s entire theory of representation that this
20 David Runciman

entity – the person of the state – does not have authority to act on its own
behalf, which means it does not have the authority to initiate its own acts
of representation.
It is true that Hobbes is not entirely consistent here, and this reading,
which emphasizes the corporate character of representation in Leviathan,
is hard to square with some other passages, where he suggests that
representation is indeed the representation of the sovereign’s individ-
ual subjects. So, for example, in Chapter XVI, Hobbes says of political
representation that “it is the Unity of the Representer, not the Unity
of the Represented, that maketh the Person One” (ibid.: 114). This
suggests that individual subjects, though disunited in their own per-
sons, nevertheless constitute the “represented.” Later on, Hobbes warns
that

where there is already erected a Soveraign Power, there can be no other Represen-
tative of the same people, but onely to certain particular ends, by the Soveraign
limited. For that were to erect two Soveraigns; and every man to have his person
represented by two Actors, that by opposing one another, must needs divide that
Power, which (if men will live in Peace), is indivisible. (ibid.: 130)

This seems to make it explicit that representation is the representation of

every man in “his [own] person,” rather than the representation of some
group person that transcends the individual. Nevertheless, it is important
to note that these two passages, in which Hobbes connects representation
with individual persons, are essentially accounts of what political repre-
sentation is not. Hobbes is saying that political representation is not the
representation of a united multitude; equally, it is not to allow individuals
to have more than one representative. Both of these ways of thinking
about representation are, in Hobbes’s eyes, inherently unstable. Indeed,
the second passage could be read in an even stronger way, as a warning
about what is liable to happen if you do collapse representation into the
representation of individuals – they are liable to consider themselves to
be represented by someone other than their authorized sovereign. That
is why it is so important for Hobbes to emphasize that representation
is representation of a single person of the state incapable of thinking in
these destabilizing terms.
If one distinguishes between Hobbes’s negative statements about polit-
ical representation – the ways in which it can go wrong – and his positive
statements – what needs to be true of it for it to be stable – then a pattern
emerges: the authorization of the sovereign by a “multitude” of natural
individuals makes the representation of the state possible, but is neither
equivalent nor reducible to it. Figure 1.1 makes clear that these can be
understood as distinct relationships.
 Hobbes’s theory of representation 21

“Sovereign”

Authorization

Representation

Ownership

“Multitude” “Commonwealth”

Figure 1.1 Hobbes’s basic theory of representation.

The obvious question to ask is why Hobbes might want to separate out
representation from authorization, given that the representative nature
of the sovereign does not change the basic obligations owed him by his
individual subjects. The answer, I think, is that Hobbes was conscious
that his bare theory of authorization was insufficient to construct an
account of the state, because it did not on its own generate for the state
its own collective identity. Even though it was part of the radicalism
of Hobbes’s theory to found it on individuals (in accordance with his
resoluto-compositive method), he recognized that representation could not
be a purely individualistic relationship. To leave the sovereign as merely
the representative of a series of individual “authors” is to reduce politics
to a series of personal relationships, each binding, but also each distinct,
with the result that the people as a whole remain a fragmented multi-
tude. For the people to assume a collective identity required that they be
represented as though they were a single person, not a series of separate
individuals. What Hobbes discovered in Leviathan was that the language
of representation enabled him to do this, notwithstanding the fact that
the authority of the sovereign continued to rest on the obligations of his
individual subjects. The concept of representation made it possible to
separate out the state from the multitude.
While it does not change the basic relationship between the individual
subject and the sovereign – and certainly does not democratize it – the
idea of the “sovereign representative” does at least raise the possibility
that there might be other significant relationships at work in Hobbes’s
22 David Runciman

“Sovereign”

Authorization

Personation

Representation

Ownership

Membership

“Multitude” “Commonwealth”

Dominion

Figure 1.2 Hobbes’s full theory of representation.

theory of politics beyond that of authorization: these include the relation-

ship between the sovereign and the commonwealth he represents, and
that between the multitude and the commonwealth that their authoriza-
tion of the sovereign brings into being. These are, as it were, the other two
sides of the triangle. Once a second side has been filled in, even if only
partially, it is no longer possible to argue that all political relationships
are merely replications of the one-sided relationship of authorization and
ownership between individual subject and sovereign. Hobbes appears to
recognize this. The account he offers in Leviathan goes some way to filling
in the other sides of the triangle. He does this by extending and enriching
some of the terminology that relates to the idea of the “sovereign repre-
sentative,” so that in the end Hobbes’s theory of representation includes
the notions of “personation,” “membership,” and “dominion” as well as
“authorization” and “ownership.” A fuller version of Hobbes’s theory of
representation is summarized in Figure 1.2, with a brief explanation of
the terms as follows:
Personation: Because for Hobbes representation derives from the con-
cept of the “person,” he sometimes talks of representation in terms of
“personation” – the wearing of a mask, or the inhabiting of a role (ibid.:
112) – so that the representation of the commonwealth makes it possi-
ble for Hobbes to describe the sovereign as “bearing the person of the
state.” But the language of personation does not fit the representation of a
 Hobbes’s theory of representation 23

multitude of individuals qua individuals, because it is impossible to see

how the sovereign could wear a multitude of masks simultaneously. An
actor can only wear one mask at a time, and though it is true that an actor
could wear a succession of different masks one after the other, it is hard
to reconcile this image with Hobbes’s understanding of politics; indeed,
this is just the sort of fragmentation of the political experience that his
theory of representation seems designed to avoid. To see the sovereign as
personating the commonwealth at least suggests that the occupant of the
“office” of sovereign representative is constrained by the requirement to
be a plausible performer of that particular role (Hobbes spells out at some
length what these requirements might be in Chapter XXX of Leviathan;
ibid.: 231–42). Hobbes is adamant that the commonwealth cannot itself
constrain the sovereign by its actions, because the commonwealth cannot
itself act except through its sovereign representative. But it can, at least
potentially, constrain by its implied presence, because even the represen-
tatives of things that have no independent existence are limited by the
need to keep up appearances. In the case of the sovereign, this means
acting in such as a way as to ensure that the subjects he represents appear
to be something more than a disunited multitude.
Hobbes also has to consider the question of how the multitude – that
is, the individual subjects – might relate to the person of the state. There
are in fact two questions here: Do they have any sort of relationship to the
state given their role in its creation? And do they have a relationship to
the state given the fact that they are ultimately responsible for what is done
in its name? One possibility is to answer both questions in the negative,
and to insist that the person of the state is simply reducible to the persons
of its individual members, so that no independent relationship between
them is conceivable. Hobbes might be thought to have good reasons
to choose this route, which closes down the possibility of individuals
forming a separate relationship with the political community than the
one that exists through their sovereign representative. Nevertheless, in
both cases, Hobbes offers the basis for an alternative answer, in the
following terms.
Dominion: In Chapter XVI of Leviathan, Hobbes makes it clear that
in the case of “incapable” entities, which are represented but unable
to authorize their own representation, authority lies with those who are
responsible for the thing in question, or who have brought it into being
(ibid.: 113). The state is just such an incapable entity, and those who
have brought it into being are the multitude, which at least suggests that
they have some kind of latent dominion over it, even if that dominion can
never subsequently be exercised because of the hold that the sovereign
exercises over them as individuals.
24 David Runciman

Membership: At the same time, Hobbes deploys the language of mem-

bership to describe the relationship between individual subjects and the
state to which they belong. For Hobbes, the language of membership
forms part of his image of the Leviathan as a kind of body with “mem-
bers” that enable it to function, as the arms, legs, etc. of the body allow
it to function also (ibid.: 168–9). Again, membership in this sense does
not imply any kind of independent control, but it does at least mean that
what is done in the name of the state depends on the proper functioning
of its individual members, joined together as part of a corporate body,
rather than simply responding as individuals jointly represented by the
same person.
All three of these additional ideas – personation, dominion, member-
ship – follow from the fact that Hobbes makes the sovereign the represen-
tative of the commonwealth, not of the multitude. Taken together, they
mean that commentators like Pitkin, by focusing exclusively on autho-
rization, offer an account of Hobbes’s theory of representation that is
much too narrow. But do any of them, either separately or jointly, serve
to undercut the force of Pitkin’s central claim that Hobbes’s account of
representation is inadequate because it does not allow anyone to object
to what the sovereign does in their name? The answer is no – at least,
not as Hobbes uses them. The representation of the person of the com-
monwealth serves to limit the ways in which the sovereign can act if he
is to fulfill his role, but it does not allow anyone to object to what the
sovereign does in that role. It simply sets tighter conditions for what will
constitute a failure of performance, and ultimately a failure of the state
itself. Personation imposes burdens on the sovereign, but does not grant
rights to anyone else, just as actors who play a part on the stage are
constrained in what they can plausibly do but are not open to objec-
tions from the character whose part they are playing. Dominion might
offer grounds for objection, but not in the case of the sovereign, who
once established exercises control over the forms of dominion enjoyed
by his subjects. Membership can be the basis for constraining the actions
undertaken by a representative of the group (the members qua mem-
bers may have certain rights over the determination of group outcomes),
but not Hobbesian membership, which is metaphorical, and designed
to illustrate that the members are conjoined with the sovereign in what-
ever he does. These are concepts that open up the possibility of a more
democratic conception of representation. But as Hobbes uses them, that
possibility is closed down.
Why is Hobbes so adamant that nowhere in the complex of relation-
ships between individuals, sovereign, and the state should there be the
basis for anyone to set limits to what the sovereign does, except for the
 Hobbes’s theory of representation 25

sovereign himself? There are two answers to this question. The first, and
most obvious, is that Hobbes was determined not to allow for the kinds
of competing claims to political authority that he believed opened the
door to civil war. This was the lesson he drew from the events of 1642
onwards: that to allow anyone to claim a “sufficient” basis to challenge
the judgment of the sovereign power was to leave the state vulnerable
to disintegration (to be “sufficient” in Hobbes’s terms meant to have
the requisite qualifications). The sovereign existed to close down poten-
tially debilitating differences of opinion among individuals, particularly
over what constituted a threat to their existence; allowing anyone to pass
judgment over the sovereign’s actions would merely open them up again,
and ultimately spell the end of the sovereign’s ability to act. For Hobbes,
this was the strongest argument in favor of his particular conception of
political representation. But if it is to be understood as a transhistorical
claim, then we can see that Hobbes was clearly wrong. We can now be
pretty confident, on the basis of the historical record, that states do not
open the door to anarchy if they allow for judgment to be passed on the
actions of “sovereign representatives” (or what we would now call gov-
ernments). Under conditions of relative material prosperity and social
stability, we can replace governments that we think have made the wrong
decisions without plunging the state into civil war.
But the second answer that Hobbes might give is more persuasive.
Hobbes’s reluctance to allow anyone the right to object to the way in
which they are being represented follows from the formal structure of his
theory. In a state in which government authority derives from individuals
but is exercised in the name of the people as a whole, it is difficult to
see where the right to object should reside. If it belongs with individuals,
then political life becomes fragmented, as it appears to allow individu-
als to exercise a right to opt in or out of collective judgment on their
own terms. This was Hobbes’s difficulty with resting ultimate political
judgment in the multitude. On the other hand, if the group as a whole
is allowed to object, then the individualistic basis of the theory is lost,
because the group becomes the source of authority of sovereign actions.
Hobbes wished to avoid this because he had straightforward intellectual
objections to bestowing political rights on shadowy collective entities that
had no natural basis. Hobbes was trying to strike a balance between his
sense that individuals could not be the final arbiters of political judgment
because that would destroy the collective life of the state, and his resis-
tance at the same time to the idea that the collective could act in its own
right, because that would trump the judgment of individuals. His theory
of representation is a way of trying to circumvent these difficulties. By
separating out representation from authorization, Hobbes is attempting
26 David Runciman

to retain a collective identity for the state without granting to the state
any distinctive powers of its own.

III. Hobbesian representation and

representative democracy
The problem of trying to strike a balance between the need to give the
state a collective identity and the wish to avoid giving the state priority
over the judgment of individuals is one that remains for democratic the-
ories of representation. It is hard to see how this balance can be achieved
without resort to the kind of formal structure provided by Hobbes. What
can be dispensed with from Hobbes’s account is the idea that authoriza-
tion must be a once-for-all event, rather than an ongoing process. But
what can be retained is the idea that those whom we authorize to act for
us act not in our name as individuals, but in the name of the state, though
it is as individuals that we pass judgment on their actions. A democratic
theory of representation needs to retain the capacity of individuals to
object to what is being done on their behalf by those with political power,
while denying to individuals the right to opt in and out of political life
as they see fit. If the first is lost, then democracy become authoritarian-
ism (as it does with Hobbes); if the second is allowed, then democracy
becomes chaos (as Hobbes feared). The solution, therefore, is to allow
individuals to pass judgment on what governments do in the name of the
state as a whole.
A version of how Hobbes’s theory might be adapted in this way is given
in Figure 1.3. It makes use of some of the terms drawn from Hobbes’s
full theory of representation, but freed from Hobbes’s unwillingness to
give them a democratic dimension. On this account, individual citizens
pass judgment (in elections and through alternative vehicles of public
opinion) over government action, and if necessary remove governments
they disapprove of, thereby constraining the capacity of governments to
act as they see fit. But governments, though authorized by the judg-
ment of individual citizens, do not represent them as individuals (except
when government politicians take up the cause of individual citizens, per-
haps in representations to a foreign government); rather, they represent
them as members of the state as a whole. In this sense, the govern-
ment personates the state, while individual citizens acting as a multitude
retain some dominion over what governments do in the state’s name.
As members of the state, individual citizens are bound by the actions
governments undertake on the state’s behalf; likewise, they have rights
as members which allow them to pass judgment over those actions. The
result is that political representation is an ongoing process that enables
 Hobbes’s theory of representation 27

“Government”

Authorisation/
judgment
Personation

Representation
Personal
representation

Membership
responsibilities

“Citizens” “State”

Membership
rights

Figure 1.3 A Hobbesian theory of democratic representation.

individuals to judge actions undertaken on behalf of a collective entity

to which they belong and for which they are ultimately responsible,
without assuming that they are personally represented in every decision
taken.
This account of political representation seems to capture an impor-
tant feature of how many representative democracies do in fact func-
tion in practice, though it does not hold for all forms of representative
democracy, and nor is it true that any representative democracy functions
like this at all times. Different systems place varying amounts of weight
on individualistic representation and on tradeoffs between the represen-
tatives of different constituencies within the state (something Hobbes
wanted to avoid at all costs). The US system in particular places greater
emphasis on the representation of individuals and of separate groups
within the state than do many other polities (see the discussion of the
difference between the “Washington” and “Westminster” models of rep-
resentation in Pettit’s chapter in this volume). But all states, including
the US, sometimes need the capacity to undertake actions in the name of
the people as a whole. When governments take certain sorts of collective
decisions (to go to war, to raise taxes, and so on), it does not make sense
to think of each individual member of the state being personally repre-
sented in those decisions; otherwise, it would be hard to see how to deal
28 David Runciman

with the fact that many individual members of the state will inevitably
object to decisions of this kind.
To put it in semi-Hobbesian terms, there is something implausible
about the idea that when government taxes an individual, it is repre-
senting that individual in doing so, since the chances are that the person
concerned would much prefer not to be taxed. But there is also something
unsatisfactory about the idea that when government taxes an individual,
it is representing some other group of individuals entirely, as that is likely
to make the politics of taxation a highly confrontational business (though
of course it often is highly confrontational, particularly in the US; see e.g.
Graetz and Shapiro 2005). Governments need to be able to act in the
name of the state, or of “the people” understood as a collective entity, not
a disparate collection of individuals. What makes this form of political
representation democratic is that a disparate collection of individuals can
still pass judgment over what governments do in the state’s name, either
through the ballot box or through other vehicles of public opinion, such
as pressure groups and opinion polls. This version of democracy, for all
its distance from a Hobbesian conception of politics, shares three key
features with Hobbes’s theory of representation. First, it separates out
authorization from representation: the individuals who give governments
the power to act are distinct from the collective entity in whose name
governments act. Second, it treats the state as an incapable entity (or
in Hobbes’s terms, a kind of “fiction”) that needs to be represented in
order to act at all.6 The only actors on this account are either individual
citizens or governments. The state cannot impose its own will on what
governments do or on how individuals judge the actions of their govern-
ments. But the state has a necessary presence in democratic political life,
because its representation is the means of giving a collective identity to
what would otherwise be just a multitude of individuals. Third, despite
separating out the state from its individual members, it makes one depen-
dent on the other – on this account, there is no state without individuals
to pass judgment on the actions of the state’s representatives, and to bear
the consequences of what the sovereign does in the state’s name.7
It is important to emphasize that this is not the only way of reconciling
a democratic presumption in favor of individual judgment with the col-
lective identity of groups. Different kinds of groups can square the circle
in a variety of different ways. This is another respect in which Hobbes
clearly got it wrong. He believed that all groups had to be modeled on

6 The details of this argument are given in Runciman (2000a) and Skinner (2002).
7 For a discussion of the different forms of collective responsibility for which a corporate
conception of political representation might allow, see Runciman (2007).
 Hobbes’s theory of representation 29

the representative structure of the state.8 But groups, particularly small,

cooperative groups, can acquire a collective identity in a variety of differ-
ent ways, without having recourse to Hobbesian forms of representation.
The authority of the group can be reduced to the authority of its individ-
ual members. Groups can establish their own identity through the joint
decisions of their members, either acting unanimously or by their joint
agreement to various judgment-aggregation procedures, none of which
need require the imposition of a representative to take decisions on their
behalf (Pettit 2003). Over time, these decisions may demonstrate the
consistency that we expect of any rational agent, so that it becomes pos-
sible to speak of the group having its own personality. Moreover, this
personality can still be reconciled with a presumption in favor of individ-
ual judgment by allowing individual members to exit the group if they
are unhappy with its collective decisions, or to remain members but to
absent themselves from personal responsibility for what the group does
in their name. Hobbes does not allow for any of this, because he does
not wish groups other than the state to develop a life of their own.
But the fact that Hobbes is much too restrictive in how he understands
group life in general does not mean that he is wrong about the particular
character of the state. States are different from other groups, because they
do not allow individuals to exit if they are unhappy, nor to absent them-
selves from responsibility for collective decisions (as Hobbes would be
the first to insist, states can only function if individuals who are unhappy
about tax-raising decisions taken in their name still have to pay their
taxes). Equally, the individual members of states, being a diverse mul-
titude, do not always display the kind of consistency in their judgments
over time that would enable them to be described as a collective person
in their own right. Public opinion is often fickle. This means that states
can only plausibly acquire their own personality through the attempts of
their representatives to give them a collective identity over time in a way
that is tolerated by public opinion, but is not predetermined by it. In that
respect, Hobbes was right to think that the personality of states depends
on their being represented as though they had such a personality, rather
than their actually having one.
Another way to put this is to say that Hobbes was doubtful about
the ability of the individual members of states to agree not merely that
they need a collective identity, but also what that identity should be. The

8 This was achieved either by copying the representative structure of the state with a single
sovereign, as in the case of families (Hobbes 1996: 162–3) or by acting on the authority
of the sovereign representative of the state, as in the case of business corporations,
universities, etc. (ibid.: 159–60).
30 David Runciman

difference can be captured by looking at what happens if you go the other

way around the triangle in Figure 1.3. This is perfectly possible – it could
be argued that what happens in political representation is that individ-
uals authorize a collective version of themselves to pass judgments on
the actions of representatives appointed by the group but representing
them as individuals. This would be a variant on a Lockean conception of
political representation: individuals would form themselves into some-
thing called a “society” or a “people,” who hire and fire governments
according to how well the government represents them as a diverse group
of individuals. Hobbes’s skepticism about such an arrangement derived
from his reluctance to rely on individuals agreeing about more than the
bare minimum that they needed to. For Hobbes, the bare minimum was
an agreement that government was necessary. But on this alternative
arrangement, before there can be government there has to be an agree-
ment about the collective identity of the state that is to be governed, and
above all about how it is to voice its objections to the way that it is being
governed.
It is true that states are often founded on an agreement of this kind, laid
out in the form of a constitutional settlement, something that Hobbes
explicitly rejects. Again, Hobbes was clearly wrong to imagine that consti-
tutional limitations on sovereign power would invariably serve as recipes
for disaster. But against that, it is also true that the process of drawing up
and ratifying constitutions accords more closely to a Hobbesian model of
political representation than to the alternative version. A constitutional
assembly that acts in the name of a people that cannot act for itself (by
definition, if it is thought that a people needs to be constituted in order
to act in the first place), and then has its actions ratified by a vote of
the population at large, fits reasonably well with the picture described in
Figure 1.3. Thus the non-Hobbesian version of popular representation
could be said to depend on a Hobbesian version to get it off ground.
The other thing to be said in favor of Hobbes’s account is that it allows
for the ways in which the state is represented, and for the ways in which
individuals can object to the representation of the state, to shape each
other over time. Because Hobbes believes it is not possible to determine
in advance any particular constraints on government action, his concep-
tual scheme of representation is more open-ended than one that requires
some agreement on the forms of collective action before popular repre-
sentation can take place. The collective identity of the state is fashioned
out of representation rather than being a precondition of it.
Thus Hobbes’s skepticism goes with a more creative conception of rep-
resentation than might be expected. Because Hobbes is so suspicious of
the idea that groups can have an identity before they are represented, he
 Hobbes’s theory of representation 31

has to allow that they can possess whatever identity their representatives
are able to give them. If Hobbes’s account is amended to include the
input of the state’s citizens in this process, then it is consistent with the
wide variety of ways in which the governments of democratic states can
have their actions conditioned by the judgment of their citizens, without
being predetermined by them. This is one of the clear merits of adapting
Hobbes’s idea of representation for democratic theory: it enables gov-
ernments and citizens to shape the ongoing life of the state, rather than
expecting them to be subordinate to it. Hobbes’s skepticism about the
possibility of large groups of disparate individuals fashioning a collec-
tive identity for themselves out of some kind of common will or group
mind has stood the test of time reasonably well, and is widely shared (see
Schumpeter 1994). It seems reasonable to accept Hobbes’s doubts about
the wisdom of giving group persons on the scale of nation-states priority
over the individuals by whom they are constituted. But Hobbes’s reluc-
tance to give up on the idea of states having their own personality at all
also has a lot to recommend it, since it is hard to know what states are if
they are nothing but the individuals who happen to inhabit them at a given
moment in time. Hobbes’s theory of representation is what enables him
to combine his well-founded doubts about transcendent claims to group
personality with his belief that politics must be something more than a
battleground of individual wills. In this respect, it is not as skeptical as
it might be, certainly compared to later theories of democracy that insist
democracy can be nothing more than the combined choices of individ-
ual citizens (Buchanan and Tullock, 1965).9 For all its anti-democratic
polemical force, Hobbes’s theory of representation is a measured one,
and that is why it remains an important resource for bridging some of
the gaps that still exist in the theory of democratic popular rule.

IV. Conclusion
The account of political representation Hobbes gives in Leviathan is
an unlikely vehicle for democratic politics. Hobbes himself was not a
democrat when he wrote Leviathan, and he believed that an idle faith in
democracy had helped to cause the English Civil War. But an idle faith
in democracy is never a good idea, and Hobbes was right to attempt to

9 Buchanan, in his “Marginal Notes on Reading Political Philosophy,” lists as authors

congenial to his and Tulloch’s approach the following: Plato, Hobbes, and Spinoza. But
he goes on: “Of these, and others within this tradition, only Spinoza’s work seems to have
much in common with our own” (Buchanan and Tullock 1965: 312). Hobbes’s theory of
representation is one reason for supporting this view of the essential lack of congeniality
between their work and Hobbes’s view of politics.
32 David Runciman

ground popular politics on something more solid, even if he was wrong

in many of the details. Moreover, his theory of representation was not
simply anti-democratic. In its fullest version, it also offers a means to
think about the complex relationships between individuals, governments,
and states without falling into some of the traps that come from trying
to oversimplify those relationships. Many theories of representation do
not avoid those traps, with the result that it is all too easy to conclude
that there is no satisfactory way of making sense of representation within
the terms of democratic theory (Przeworski 1999; Shapiro 2003). The
problem is that it is very hard to see how to reconcile the claims of
political representatives to take decisions on behalf of individuals with
the rights of individuals to judge how well they are being represented.
Hobbes offers a way around this difficulty by insisting that it is not simply
individuals who are being represented; instead, it is the state that is being
represented, and individuals who are involved in the process of making
the representation of the state work. This rescues political representation
from a constant and ultimately fruitless struggle to reconcile the apparent
independence of political representatives to act as they see fit with their
supposed dependence as representatives on the views of those whom they
represent (Pitkin 1967). On Hobbes’s account, political representatives
are independent by definition, because they do not purport to represent
the views of particular individuals, and the entity that they do represent –
the state – has no views of its own.
What Hobbes does not do is offer an explanation of how the views of
individuals might bear on the representation of the state. If this is added
to Hobbes’s account, as I have attempted here, it serves to limit the
independence of political representatives without requiring them simply
to represent the views of individuals. They act as representatives of the
state; individuals judge them on that. In some ways, this is still a pretty
minimal account of democracy, and as I have indicated, it describes the
workings of some polities better than others. It leaves out all the ways
that individuals might want their views taken account of, listened to, and
played out before important political decisions are made. It also leaves
out all the ways in which political representatives might take up the cause
of individual constituents, and seek to represent them. It does not take
account of all the groups other than the state in which individuals might
find a voice and by which they might be represented. But, though min-
imal, this account of political representation still has to be compared
with the alternatives. The difficulty that conventional theories of rep-
resentation have in squaring the circle of the simultaneous dependence
and independence of political representatives has led to some fairly stark
choices. Either democracy abandons the language of representation and
 Hobbes’s theory of representation 33

employs other terms to describe the relationship between governments

and electorates (for example, that of consumer sovereignty: see Shapiro
2003), or representation abandons the idea of national collective iden-
tity and looks for other outlets for the representation of individuals (for
example, the idea of “representation as responsiveness” in the interna-
tional arena; see Kuper 2004). Hobbes was trying to avoid the need for
choices of this kind. He believed that the idea of popular sovereignty
could be reconciled with a conception of representation without plac-
ing undue weight on the collective identity of a multitude of separate
individuals. His thought remains foundational because (for now, any-
way) states founded on the idea of popular sovereignty do not seem
to have been able to dispense with the language of representation and
replace it with something else. The problem has always been knowing
how to combine this dependence on the concept of representation with
the democratic conviction that individuals ought to be able to tell their
representatives what to do if it is going to make sense to talk about repre-
sentation at all. Hobbes’s triangle offers a way round having to square that
circle.

BIBLIOGRAPHY
Buchanan, James and Gordon Tullock. 1965. The Calculus of Consent. Logical
Foundations of Constitutional Democracy. Ann Arbor: University of Michigan
Press.
Forsyth, Murray. 1987. Reason and Revolution. The Political Thought of the Abbé
Sieyès. Leicester: Leicester University Press.
Graetz, Michael and Ian Shapiro. 2005. Death by a Thousand Cuts. The Fight over
Taxing Inherited Wealth. Princeton, NJ: Princeton University Press.
Hobbes, Thomas. 1996 [1651]. Leviathan, ed. Richard Tuck. Cambridge: Cam-
bridge University Press.
1998 [1642]. On the Citizen, ed. Richard Tuck and Michael Silverthorne.
Cambridge: Cambridge University Press.
Hoekstra, Kinch. 2006. “A Lion in the House: Hobbes and Democracy,” in
Rethinking the Foundations of Modern Political Thought, ed. Annabel Brett and
James Tully. Cambridge: Cambridge University Press, 191–219.
Hont, Istvan. 2005. Jealousy of Trade. International Competition and the Nation
State in Historical Perspective. Cambridge, MA: Belknap.
Kuper, Andrew. 2004. Democracy Beyond Borders. Justice and Representation in
Global Institutions. Oxford: Oxford University Press.
Locke, John. 1988 [1679]. Two Treatises of Government, ed. Peter Laslett. Cam-
bridge: Cambridge University Press.
Manin, Bernard. 1997. The Principles of Representative Government. Cambridge:
Cambridge University Press.
Pettit, Philip. 2003. “Groups with Minds of their Own,” in Socializing Meta-
physics, ed. F. Schmitt. New York: Rowman and Littlefield, 167–93.
34 David Runciman

Pitkin, Hanna. 1967. The Concept of Representation. Berkeley: University of

California Press.
Przeworski, Adam, 1999. “Minimalist Conceptions of Democracy,” in Democ-
racy’s Value, ed. Ian Shapiro and Casiano Hacker-Cordon. Cambridge: Cam-
bridge University Press, 23–50.
Przeworski, Adam, Susan Stokes, and Bernard Manin (eds.). 1999. Democracy,
Accountability and Representation. Cambridge: Cambridge University Press.
Runciman, David. 2000a. “What Kind of Person Is Hobbes’s State? A Reply to
Skinner.” Journal of Political Philosophy 8: 268–78.
2000b. “Is the State a Corporation?” Government and Opposition 35: 90–104.
2007. “The Paradox of Political Representation.” Journal of Political Philosophy
15: 93–117.
Schmitt, Carl. 1996 [1938]. The Leviathan in the State Theory of Thomas Hobbes.
Meaning and Failure of a Symbol, ed. George Schwab. Westport, CT: Green-
wood.
Schumpeter, Joseph. 1994 [1942]. Capitalism, Socialism and Democracy, ed.
Richard Swedberg. London: Routledge.
Shapiro, Ian. 2003. The State of Democratic Theory. Princeton, NJ: Princeton
University Press.
Skinner, Quentin. 1989. “The State,” in Political Innovation and Conceptual
Change, ed. T. Ball et al. Cambridge: Cambridge University Press, 90–131.
2002. “Hobbes and the Purely Artificial Person of the State,” in Visions of
Politics, vol. III. Cambridge: Cambridge University Press, 177–208.
2005. “Hobbes on Representation.” European Journal of Philosophy 13: 155–84.
Tuck, Richard. 2006. “Hobbes and Democracy,” in Rethinking the Foundations
of Modern Political Thought, ed. Annabel Brett and James Tully. Cambridge:
Cambridge University Press, 171–90.
2 Participation and representation before
democracy: petitions and addresses in
premodern Britain

Mark Knights

Participation and representation are often taken to be alternative forms

of popular rule. This chapter nevertheless suggests that, in predemo-
cractic Britain, they were not mutually exclusive, and investigates the
informal and participatory ways in which representation occurred out-
side the formal process of election. In particular it will focus on the mass
of petitions and addresses that were presented to both parliament and
crown from the seventeenth through to the nineteenth centuries. These
not only claimed to represent the people’s views but also procured their
active participation by obtaining their signatures and by stimulating pub-
lic debate. Indeed, analysis of the petitions and addresses, and of the
debates that they provoked, shows that there was often an ongoing, cre-
ative, and dynamic interaction between the represented and their elected
representatives.1 Representation did not end at the moment of parlia-
mentary electoral authorization; and public opinion operated before the
advent of democracy. This historical evidence should affect our under-
standing of the nature of representation and popular rule, both at a
theoretical and at a practical level.
One view of representation argues that the represented transfer power
to a representative and thus have little political role thereafter, being
deemed to consent to what is done in their name. This was essentially
the view of Thomas Hobbes in the seventeenth century, who argued
that the people transferred all their power to the representative person or
assembly, “the great Leviathan,” who was the authorized and exclusive
but also the “absolute” representative. The “people” became subsumed
within and by their representative sovereign (Hobbes 1651: especially
Part 1, Chapters 16, 22).2 Bernard Manin has argued that the system
I am grateful to the conference participants and to Mark Philp for helpful comments on
earlier drafts of this chapter.
1 Urbinati (2006) does not discuss petitioning but sees “participation and representation
not as alternative forms of democracy but as related forms” (p. 3) and stresses the active
and “informal expression of ‘popular will’” (p. 10).
2 Hobbes’s views are outlined in detail in David Runciman’s chapter.

35
36 Mark Knights

of electoral representation, which he suggests was devised in the seven-

teenth and eighteenth centuries (and which he sees as enduring), did very
much the same thing. He sees a representative Leviathan emerging as a
result of “the triumph of elections.” He argues that through the adoption
of elections the people consented to power rather than exercise it them-
selves and that elections were inherently aristocratic rather than popular.
Indeed, he argues, elections were chosen precisely because they were not
democratic, and that “representative government gives no institutional
role to the assembled people” (Manin 1997: 8).
To be sure, there is some evidence to support these positions. In Britain
in the mid-eighteenth century, elections were seen as securing the consent
of the people and they were very often aristocratic and exclusive in nature.
Claims were also made that parliament alone was the representative of
the people. Yet exclusive parliamentary representation was, especially
in the mid/late seventeenth to the early eighteenth century and again
in the late eighteenth to the early nineteenth century, more of an ideal
than a reality. As Hobbes himself recognized, the urban corporations
(who returned four-fifths of elected representatives) were places where
the people remained actively involved in governance and were thus, in
his eyes, “many lesser commonwealths in the bowels of a greater, like
wormes in the entrayles of a naturall man” (Hobbes 1651: chapter 29).3
The corporations were (chiefly) towns that had obtained royal charters,
of which there were only 38 in 1500, but 130 by 1600, 181 by 1640,
and 246 by 1833 (Withington 2005: 18; Sweet 1999: 33; Clark 2000:
Table 2). The charter created a legally constructed fictional body. “In this
guise, the freemen, burgesses and citizens who voluntarily participated
in this person could act collectively as a single body, so transcending
their individual lives and interests to form an entity that could sue at law
and be represented in parliament” (Withington 2005: 8).4 Through the
charter the freemen acquired rights of participation in trade and also in
governance. In essence the charter created a self-governing community,
albeit one subject ultimately to royal command and one which drew its
authority from the monarch. England was a commonwealth within a
monarchy.5

3 In 1683 Northampton’s loyal “address and petition” asserted that “corporations in gen-
eral have been the nurseries of sedition and schism . . . for want of necessary inspection
from above, they have degenerated by degrees from their allegiance to an imperial power
into the corruptions of a commonwealth” (London Gazette #1857).
4 See also Halliday (1998).
5 Anon. (1701b: 16) depicted the people not as something separate from authority but as
the practical constituents of it. In England, the author argued, “Tis hard to find a man
who has not sometime been call’d to bear office in his parish or borough.”
 Participation and representation before democracy 37

This insight has recently been developed by historians of the early mod-
ern state who argue that in predemocratic Britain it was not the case that
the people simply consented to power rather than exercise it themselves.
Rather, in the absence of a paid police force or indeed a paid bureau-
cracy, there was a system of negotiated power and an “unacknowledged
republic” of voluntary office-holders. Moreover, it is suggested, the sys-
tem of representation through parliamentary election was rather remote
from the “parish state” that most Britons encountered, in which partic-
ipation was prized. In this view, then, the British system was something
of a hybrid. It had tiers of representatives chosen by elections; but parlia-
mentary elections were only one part. There was a myriad of local posts
that were locally elected, and, moreover, a culture of citizenship that saw
office-holding and participation as civic duties, and active consent as a
civic right. Whilst there is some debate about how long this structure
lasted, one recent survey of governance points to it enduring well into
the nineteenth century (Collinson 1986–7; Braddick 2000; Braddick and
Walter 2001; Wrightson 1996; Goldie 2001; Hindle, 1999; Eastwood
1997 and 2002).6
The implication of this research is that elections did not simply
“triumph,” that the people continued to hold office and self-govern,
and that this participation was compatible (even if it was also at times in
tension) with electoral representation. To be sure, we should be careful
not to exaggerate how much participation occurred or how socially inclu-
sive the higher ranks of local office were, and we should note that there
was a good deal of hostility toward both popular rule and democracy.
But these caveats do not negate the fact that participation and repre-
sentation coexisted, and that both as an ideal and as practice the people
claimed a representative voice outside elections. The people did not sim-
ply authorize their representatives to act for them. Representation, as
it developed in seventeenth- and eighteenth-century Britain, was thus a
participatory process involving, on occasion, a good deal of direct polit-
ical activity. An electoral system “differing radically from democracy”
was thus not “invented” in the seventeenth and eighteenth centuries
(Manin 1997: 4). Citizens were not merely the source of political legit-
imacy through the consent they gave at elections, but possessed strong
notions of self-governance and more than one representative voice in a
system of overlapping but legitimate authorities. If that is true, the study
of the past may have implications for more modern forms of democratic

6 For a very different analysis see Clark (1985: 374–5), which minimizes the capacity for
popular politics in the period: “the common man was still presented with relatively simple
options of obedience and revolt.”
38 Mark Knights

representation outside election times. For there are representative insti-

tutions, such as the petition, that allow citizens an individual and corpo-
rate voice even while they have elected representatives. Such institutions
reflect a desire to escape the logic of binding authorization and may be
one way to strengthen accountability.

I. Petitioning as a form of representation

The informal means of representation – by which I mean forms of rep-
resentation that were not electoral, such as the press, petitions, clubs,
and associations – developed as a raft of linked practices that supple-
mented, overlaid, and in some cases claimed to replace, formal repre-
sentation. They could be associated with elections and they were also
used instrumentally in the campaign for a wider franchise; but they were
also representative practices in their own right, claiming to speak for the
people.
We might identify two important shifts in the context in which such
informal means of representation operated over the seventeenth and eigh-
teenth centuries. The first was the recognition of the public as a collective
force, of the public as an ideal representation of the people. I have argued
elsewhere that this occurred in the late seventeenth and early eighteenth
centuries as a result of a combination of frequent general elections, a
free press, an ideology of popular sovereignty, and the emergence of a
fiscal-military state and party politics (Knights 2005). A second shift is
identified by Charles Tilly as a transformation of popular politics, dated
as occurring in the mid-eighteenth century and culminating in 1832,
“from a significant reliance on patronage to autonomous claim-making
in national arenas,” a process that he calls “the development of mass
national politics.” This was accompanied by a “deep transformation” of
“the means by which ordinary people made collective claims on other
people, including power-holders and the state” (Tilly 1995: 13–14, 53).
Associations, clubs, societies, and unions were participatory, represen-
tative institutions that could rival the claims of the national parliament
to speak for the people. As T. M. Parssinen has noted, “The association
became, in the opinion of its extreme proponents, an anti-parliament.
They claimed that it represented the true wishes of the sovereign people
more fully than did the corrupt parliament at Westminster, and there-
fore deserved the people’s allegiance in an ultimate conflict between the
two bodies” (Parssinen 1973: 504). Equally alarming for the government
was the advent of political clubs with mass membership. For example,
in Sheffield in 1792 2,500 of the “lowest mechanics” were enrolled in
a political society formed after the founders had read Thomas Paine’s
 Participation and representation before democracy 39

Rights of Man (Stevenson 1989: 14). Over a hundred such radical soci-
eties were formed by the mid-1790s. Similarly the political unions of the
early 1830s grew out of the mass club that agitated for reform, such as the
Society for Constitutional Information and the Corresponding Society,
as well as the extra-parliamentary popular agitation of Wilkes and the
American Revolution. By 1832 there were more than a hundred political
unions, aimed at universal manhood suffrage, annual parliaments, and
the ballot. Self-consciously trying to unite the interests of “the lower and
middle classes of the people,” and a mixture of electors and unenfran-
chised in roughly equal numbers, they could boast 15–18,000 paying
members and their rallies attracted over 200,000. Claiming to represent
the people, they “played a vital role in seeing the Reform Bill [of 1832]
through parliament” (LoPatin 1999: 1, 11–12, 14).
But I want to focus on one very important informal representative insti-
tution: the petition and its associate, the address. Petitions and addresses
were ubiquitous in the processes of formal government (for example,
starting lawsuits or bringing grievances to the attention of local rep-
resentatives), as well as initiating or lobbying for and against pieces of
legislation;7 but they were also, from the sixteenth century, a key platform
through which to represent popular views on national issues of politics
and religion (Hoyle 2002). Petitions were often intended and timed to
influence representatives at Westminster; but they were also represen-
tative in a second and more important sense, since the act of signing
involved individuals in a collective action that claimed representative sta-
tus. Petitioners routinely claimed to speak for the people, to represent the
“will” or “feel the pulse” of the people, to show the “bent of the nation.”
They were the “voice of the country.” Of course, unlike an elected rep-
resentative, who had ongoing authority between elections, petitions had
only temporary representativeness; but they often provoked an iterative
process of petition and counter-petition so that an ongoing representative
contest ensued.
Petitions often created a dialogue with counter-petitioners, thereby
stimulating rival representative claims. Thus nineteen counties that peti-
tioned for “root and branch” reform of the church in 1640–1 provoked
twenty-nine counter-petitions loyal to the established form of worship,

7 This could be done collectively, as in 1718/19 against calicoes and 1763–6 against the
cider tax (Dickinson, 1994: 67–72). There has been no attempt to quantify this peti-
tioning over time, but for an indication of its growing size see Hoppit (1997: 18–21).
For studies of petitioning as a means of influencing legislation see Gauci (2001); Sweet
(2003); Handley (1990); Brewer (1988). My analysis of public petitions excludes those
presented to parliament about trade, though of course these were very significant in
number.
40 Mark Knights

claiming over 85,000 names. Petitions promoted between December

1641 and August 1642 came from thirty-eight of the forty counties
(Fletcher 1981: 91–3, 192; Maltby 1998: Appendix 1; Maltby 1999:
103–67). Such mass petitioning was new and it helped both to push
the country into civil war and to achieve something like a national cul-
ture. Third, Zaret suggests, it was the advent of printed petitions, after
the government lost control of the press in November 1641, that trans-
formed the power of petitioning as a weapon. He estimates that between
1640 and the restoration of the monarchy in 1660 about 500 petitions
were produced and that this created (unintentionally) a sense of public
opinion. Indeed he even suggests that print and petitioning in this period
created the origins of democratic culture (Zaret 1996 and 2000).
Since there was no franchise determining who could participate, sub-
scriptional activity was also open to sectors of the population who were
not participants in the process of formal representation. Thus appren-
tices and women could and did petition, the former being particularly
prominent in seventeenth-century London petitioning and addressing,
and the latter particularly important in later anti-slavery petitions.8 Sim-
ilarly, unenfranchised towns could, and increasingly did, find in petitions
a voice denied them by the formal representative system.9 Moreover,
despite the rhetorical humility customary in the form of a petition, many
mass petitions were also implicitly or explicitly threatening in their use
of numbers. And they mobilized and invoked the power of the people, as
those from the Levelers in the 1640s or from the parliamentary reformers
of the nineteenth century show. However, such representations were not
limited to radical groups. Addresses were used in very large numbers to
demonstrate public loyalty to any particular regime, monarch, or policy;
and in doing so loyalists often made conflicting and mutually exclusive
representative claims from the radicals. Such campaigns were frequently
national in scope (and colonies also joined the early eighteenth-century
ones). Indeed this and the widespread publication of them helped engen-
der a sense of the nation and of a national “public opinion.”
Despite their importance, there is no study of public petitions and
addresses over the premodern period as a whole. Indeed, it is not easy

8 For example, in 1683 the young men and apprentices of Bristol, Westminster and
London, and Hull addressed the crown (London Gazette #1863, 1866, 1874). For the
role of London apprentices see Harris (1987: 42–4, 174–7). For female petitioning in
the Civil War period see Higgins (1973). And for later eighteenth-century participation
see Colley (1996: 291–4).
9 This occurred in the later Stuart period. See, for example, Manchester, Leeds, and
Halifax presenting addresses in 1683 and 1701 (London Gazette #1859, 3761, 3762).
See also Phillips (1980: 615–16).
 Participation and representation before democracy 41

even to compile statistics about how many were presented, either to

parliament or to the crown, in order to quantify that importance. While
they are often discussed for particular historical moments, there have
been few attempts to look at them collectively over time or to relate
them to debates about participation and representation. Moreover, their
survival in state records is haphazard and many were burnt in the fire
that consumed the Houses of Parliament in 1834. This, together with
the fact that numbers were often inflated by promoters and disputed
by opponents, means that it is often difficult to verify precisely how
many signatures they attracted. Table 2.2 (later in the chapter) listing
public petitions and addresses 1640–1840 is therefore only a best guess,
given the current state of research, about the pattern and size of national
petitioning and addressing campaigns. But it does at least allow us to
demonstrate its truly vast scale, to chart its fluctuations, and to compare
campaigns over time.
From Table 2.2 it is clear that mass petitioning was most prolific in
the period 1640–1720 and again 1780–1850. Indeed, the quantity of
petitions and addresses in the earlier period is very impressive. Yet four
factors at the end of the eighteenth century and in the early nineteenth
served to expand petitioning campaigns into mass movements that aimed
to represent extra-parliamentary views and achieved a force that was dif-
ficult for any government to resist (Fraser 1961: 195–6). The first was
anti-slavery. Petitions were used to exert considerable pressure on MPs
whenever legislation was being considered to abolish slavery. In 1788
there were 100 petitions with about 60–75,000 signatures; but in 1792
this rose to 500 petitions with 350–400,000 signatures, and in 1814 anti-
slavery provoked 800 petitions with 750,000 signatures (Drescher, 1986:
93–4). A second factor was reform. In 1780 the Yorkshire Association,
led by Christopher Wyvill, did indeed petition for reform, and included
a demand for a “more equal representation of the people” (Black 1963:
49). Twenty-four counties and eleven cities and towns petitioned. But
Wyvill was not interested in mass subscriptions on the issue of reform.
They did come, however, in 1830–2, so much so that as a result parlia-
ment decided that no discussion whatever should accompany the presen-
tation of any petition; otherwise, it was claimed, no public business would
get done. Even with this restriction the Chartists continued to use peti-
tioning to agitate for universal representation, though more as a means
to maintain political momentum rather than through any belief that MPs
would take any immediate notice. The 1842 Chartist ‘Grand Petition’
had 3,317,702 signatures, was more than six miles in length and was
accompanied by a crowd of 50,000 people when it was carted to West-
minster. The Chartists gained 1.2 million signatures in 1839; 1.4 million
42 Mark Knights

Table 2.1 Petitions to parliament, 1785–1847

Average number of
Five-year period petitions per year

1785–90 176
1801–5 205
1811–15 899
1828–32 4,656
1833–7 7,436
1838–42 14,014
1843–7 16,397

in 1841; and 2 million in 1848, more than twice the size of the electorate.
The third factor was the French Revolution and the polarizations that it
produced in England. Thus in 1795 over 130,000 signatures were gath-
ered to petitions against the government’s repressive sedition bills, which
restricted the right to assemble and to speak freely (British Library, Add.
MS 27,808 f. 52, estimate by Francis Place). Fourth, these movements
tended to combine with other grievances – over Catholic emancipation,
the repeal of the Test and Corporation Acts, and so on. Petitioning was
thus “an integral part of the system of political representation” (Pickering
2001: 368, 373–4, 381–4).
The impact that these factors had on the amount of public petitioning
can be charted from official statistics (see Table 2.1) (Parliamentary
Papers 1852–3, vol. LXXXIII, p. 166, see http://gateway.proquest.com/
openurl?url_ver=Z39.88–2004&res_dat=xri:hcpp&rft_dat=xri:hcpp:
fulltext:1852–029528:2) (accessed September 12, 2008).
As Figure 2.1 illustrates, it was not until the twentieth century, once
mass democracy had been established, that petitioning faded as a means
of representation.

II. Misrepresentation
Petitioning always raised the issue of the relationship between informal
and formal means of representation, between passive consent to elected
representatives and active participation by the people in other more direct
forms of representation. The elected representatives in parliament usu-
ally resented or were hostile to such direct representation. Through-
out the premodern period the representativeness of public petitions and
addresses was highly controversial and often denied. This could be at
 Participation and representation before democracy 43

35,000
No. of petitions
30,000 per session to
25,000 nearest 1,000
20,000
15,000
10,000
5,000

1840 1850 1860 1870 1880 1890 1900 1910 1920 1930 1940 1950

Millions
7 No. of signatures to
petitions per session
6
to nearest 200,000
5
4
3
2
1

1840 1850 1860 1870 1880 1890 1900 1910 1920 1930 1940 1950

700
Average no. of
600
signatures per petition
500 to nearest 20
400
300
200
100

1840 1850 1860 1870 1880 1890 1900 1910 1920 1930 1940 1950

Figure 2.1 Number of petitions and signatures per session

Source: Taylor (1913).

the simple level of disputing the accuracy of the numbers claimed. It was
suggested that signatures had been invented and that those who signed
either did not understand what they were signing or had been misled
into doing so. Thus one tactic was to claim that the petition or address
was the product not of genuine popular opinion but of individuals or
small factions who sought to use the public to give color to their pretense
of representativeness. As historians we might share such worries, won-
dering how far petitions and addresses were representative of non-voters
or simply propaganda manufactured by government, the press, or those
who already ruled. Were they, as the earl of Lauderdale suggested to
Charles II, meaningless pieces of paper “fit for nothing but to wipe his
Royal A_” (Anon. 1702a: 20)? In fact we can demonstrate the degree of
44 Mark Knights

participatory representation that petitions and addresses afforded to

those outside the electorate. John Phillips found that over 50 percent
of those who signed petitions in the Middlesex election affair and over
economical reform were not voters in parliamentary elections. James
Bradley has found the percentage was as high as 81 percent at Poole and
80 percent at Southampton (Phillips 1980: 611; Bradley 1990: 144–5).
Clearly such signatories did not see themselves as being represented by
those entitled to vote.
However, rather than focus on the mass campaigns of the later eight-
eenth and early nineteenth century that provided the evidence for
Phillips and Bradley, I want to explore the intense contests over rep-
resentation that petitions and addresses provoked in the first age of party
politics, and hence to show how intrinsic such debates were to the for-
mation of Britain’s representative system. I propose to examine debates
in 1701–2, when a national debate took place about the relationship
between formal and informal representation. This occurred (within a
context of numerous petitioning and addressing campaigns charted in
Table 2.2) because in 1701 an unsigned petition, known as the “Legion
Memorial,” because it claimed to represent the grievances of two hun-
dred thousand “good People of England,” sought to put pressure on par-
liament to recommence war with France (Defoe 1701).10 It came hard
on the heels of a petition from a Kent grand jury, arguing along similar
lines, which was condemned by the House of Commons as “scandalous,
insolent and seditious, tending to destroy the constitution of parliament,
and to subvert the established government of this realm.” The presenters
of the Kent petition had been imprisoned by the House of Commons.
Legion nevertheless asserted that the people were parliament’s “masters”
and “superiors” and warned MPs “you are not above the people’s resent-
ments.” Legion claimed, moreover, that “it is the undoubted right of the
people of England, in case their representatives in parliament do not pro-
ceed according to their duty, and the people’s interest, to inform them
of their dislike, disown their actions and to direct them to such things
as they think fit, either by petition, address, proposal, memorial or any
other peaceable way.” Thus if parliament betrayed “the trust imposed
in them” it was “the undoubted right of the people of England to call
them to account for the same, and by convention, assembly or force may
proceed against them as traitors and betrayers of the country.” The tract

10 Its representative claim was denied by one author, since it bore “so little proportion
to the sentiments of the people and has so small an argument with either honesty or
truth, that not one of the good people of England, tho he boasts many thousands, can
be concern’d in it” (Anon. 1701a: 27).
 Participation and representation before democracy 45

Table 2.2 Public petitioning and addressing campaigns, 1640–1840a

Numbers of signatories
(max individual;
estimate of total
Date Occasion Number where known)

Dec. 1640 – July Root and Branch 19 4,488 max (Lancs.)

1641 In favor of Prayer Book 29 14,350 max (Somerset)
Dec. 1641 – Aug. Grievances (response to 46 30,000 max (Essex)
1642 Grand Remonstrance)
1646–9 Levelers 8 “many thousands”
(1647)
1658 Addresses for Richard 94
Cromwell’s Proclamation
as Protector
Dec. 1659 – June Petitions and addresses for a 29 23,500 max (London)
1660 free parliament and
restoration
Dec. 1679 – Jan. Petitions for parliament to 7 16,000 max (London)
1680 sit
Spring 1680 Addresses of abhorrence of 8
the petitions calling for
parliament
Summer 1681 Addresses of thanks for the 210 6,000 max (Northants),
king’s Declaration over 40,000 total
Jan. – Sept. 1682 Addresses against 199 “over 2,000” max
Shaftesbury’s Association (Derbys.)
Aug. – Nov. 1683 Addresses of abhorrence of 323 10,000 max (tin miners)
the Rye House Plot
Feb. – June 1685 Addresses of congratulation 361
on accession of James
1687 Addresses of thanks for the 197 800 max (Quakers)
declaration of indulgence
1688 Further addresses after 48
second declaration of
indulgence
Dec. 1694 – Apr. Addresses of condolence on 201
1695 death of Queen Mary
Nov. 1697 – Feb. Addresses of congratulation 260
1698 on peace
Oct. 1701 – Jan. Addresses against France’s 344 “near 5,000” max
1702 recognition of the (Northants)
pretender
Mar. – May 1702 Addresses of congratulation 400
on accession of Queen
Anne
(cont.)
46 Mark Knights

Table 2.2 (cont.)

Numbers of signatories
(max individual;
estimate of total
Date Occasion Number where known)

Feb. – Aug. 1704 Addresses of clerical thanks 25

for Queen’s bounty to
church
Aug. – Dec. 1704 Addresses of congratulation 265
on military victories
May. – Oct. 1706 Addresses of congratulation 303
on military victories
Mar. – Oct. 1707 Addresses of congratulation 213
on Union with Scotland
Mar. – July. 1708 Addresses against the 338
Jacobite invasion attempt
July – Oct. 1708 Addresses of congratulation 98 “several thousand”
on military victories (Westminster)
Dec. 1708 Addresses of condolence on 10
death of Prince Consort
Sept. – Dec. 1709 Addresses of congratulation 40
on military victories
1710 Tory addresses about 92 “many thousand”
dangers to state and (Westminster)
pressing for election
1710 Whig counter-addresses 15 533 max (Norwich)
June – Nov. 1712 Addresses of congratulation 291 “many thousand”
on imminent peace (Westminster)
Apr. – Sept. 1713 Addresses of congratulation 231 “many thousand”
on conclusion of peace (Westminster)
Sept. 1714 – May Addresses of congratulation 443
1715 on accession of King
George
July 1715 – Aug. Addresses against rebellion 320
1716 and thanks for its
suppression
Mar. – Nov. 1717 Addresses of congratulations 50
on victory
1745 Thanks for defeat of Jacobite Over 200
rebellion
1765 Anti-Stamp Act 20
May 1769 – Jan. Middlesex election affair 27 55–60,000 total
1770 (pro Wilkes)
Addresses against Wilkes 76
1775–6 Anti-aggressive policy 78 6,500 max (Lancs)
toward colonies address
supporting government
stance
 Participation and representation before democracy 47

Table 2.2 (cont.)

Numbers of signatories
(max individual;
estimate of total
Date Occasion Number where known)

1756 Addresses/instructions on 36
loss of Minorca
1779–80 Reform 37 60,000 total
1780 Anti-Catholic relief 50,000 max
1782–3 Economical reform (Wyvill) 35 20,000 total
1784 Addresses in support of Over 200 53,500 total
royal prerogative
(Fox–North)
1785 Economical reform (Wyvill) 14
1788 Petitions for abolition of Over 100 11,000 max
slave trade (Manchester),
60–75,000 total
1789 Addresses on king’s recovery Over 550
1792 Petitions against slave trade 500 350–400,000 total
1792–3 Addresses in favor of 386 11,000 max (Liverpool)
proclamation against
seditious writings and
activities
1793 Parliamentary reform 36
1795 Protests against sedition 95 131,000 total
laws
Jan. 1795 Petitions against war 17
Nov. – Dec. 1795 Anti-war petitions and 91 131,000 total
addresses
1795–6 Addresses re attempts to 1,800
assassinate the king
1797 Anti-war petitions >15
1800 Addresses re further attempt
on king’s life
1801 Addresses against war 30,000 max (Yorks)
1803 Addresses on renewal of war
1803 Addresses on discovery of
Despard Conspiracy
1807 Addresses supporting king’s
opposition to Catholic
relief
1807–8 Addresses for peace 150,000 from Yorks and
Lancs
1812–13 Addresses for peace Over 50,000 total
1809 Addresses on king’s 50th
birthday
1814 Addresses on centenary of
Hanoverian succession
(cont.)
48 Mark Knights

Table 2.2 (cont.)

Numbers of signatories
(max individual;
estimate of total
Date Occasion Number where known)

1814 Petitions against slavery 800 750,000 total

1816–17 Parliamentary reform 700 30,000 max
(Hampden clubs) (Manchester), 1m
total
1820 Pro Queen Caroline c. 800 (with over 70,000
female sigs)
1825 Anti-Catholic petitions over 400
1828 Anti-repeal of Test and 28
Corporation Acts
1829 Anti-Catholic petitions Over 81,000 max (Kent)
2,000
Pro-Catholic petitions 220
Oct. 1830–Apr. 1831 Parliamentary reform 3,000 21,000 max
(Birmingham)
May 1832 Parliamentary reform bill 290
1833 Petitions against slavery 5,000 1.5 million total
1839–43 Anti-Corn Law petitions Average (1.15m sigs) per
of parliamentary session
3,270

a The figures have been compiled from a large number of primary and secondary sources
too numerous to list here, and exclude those presented to parliament relating to economic
legislation.

concluded with the warning that “Englishmen are no more to be slaves

to parliament than to a king.”
The Kent petition and the Legion Memorial sparked an important
pamphlet debate about the nature of popular representation, since it
pitted a formal representative institution (parliament) against an informal
representative one (a petition). England’s Enemies Exposed attacked the
Legion Memorial’s claims on the grounds that as soon as electors made
their choice “the power and right of the elector devolves intirely upon
their representatives; and tho at convenient seasons and in decent words,
they may have leave to express their sentiments of things, yet they have
no power to compel their members to vote or act, but in conformity to
their own private judgments, tho the whole country, city or burrough
they represent were of a contrary opinion to that of their representatives”
(Anon. 1701d: 30). But against such views Legion also found some
ideological support. Jura Populi Anglicani, thought to have been written
 Participation and representation before democracy 49

by Whig leader John Somers, repeated the warning of a tyrannical House

of Commons and affirmed that “the House of Commons are not the
whole People of England’s Representatives.” The representative body
was, it was argued, not the Commons per se but the incorporated people:
“The Commons may be said to represent those Freeholders, Citizens and
Freemen who chose them; but what are they to the Body of the People,
who are represented in the Political State, and are entituled to all the
Benefits and Advantages of it?” The tract argued that the people did not
delegate power to their representatives “for them to do whatever they
please.” Rather, deploying Lockean arguments and defending a natural
right to petition, it argued that MPs were sent with a trust “which if
they should manifestly betray, the People, in whom the Power is more
perfectly and fully than in their Delegates, must have a Right to help and
preserve themselves” (Somers 1701: preface v, 51).
Jura Populi . . . Answer’d (1701) nevertheless refuted the idea that the
people were above parliament. MPs were “above being directed” and
sought to invalidate the voice of any one county as the voice of the
nation; rather MPs were only entrusted at a local level: “Each Particular
Member of the Honourable House of Commons has a Trust repos’d in
him, for the Preservation of the Rights which belong to each distinct
County, City or Borrough, and if the Inhabitants of the said Places think
themselves aggriev’d, they ought to Address themselves only to those
Knights, Citizens or Burgesses they have had the immediate Choice of”
and not petition parliament as a whole (Anon. 1701f: 70, 79.).11 More-
over, Jura Populi’s claim to popular sovereignty and the limited nature
of parliamentary representation infuriated a Tory-dominated Commons
which, on February 26, 1702, passed a resolution ‘that to assert that the
House of Commons is not the only representatives of the commons of
England tends to the subversion of the rights and privileges of the House
of Commons.”
The dispute, which had been hard to disentangle from the Tory
impeachments of Somers and other Whig leaders, seems also to have
provided an essential context for Jonathan Swift’s first political tract, A
Discourse of the Contests and Dissensions between the Nobles and the Commons
in Athens and Rome (1701). Swift took a slightly different tack and argued
that in recent years the nation had made “mighty leaps from prerogative
heights into the depths of popularity” and the cause of corruption was
ultimately political parties. Parties – which of course were themselves in
some part representative – made representatives into sheep who followed
the herd. As a result, Swift believed, the House of Commons had “lost

11 See also Anon. (1702a: 13).

50 Mark Knights

the universal favour of the numbers they represent.” Yet Swift never
confronted the ambiguities of the “people” as represented both in an
institutional form and as a social or political category with informal rep-
resentation. Thus on the one hand he argued that “Vox populi vox dei
ought to be understood of the universal bent and current of a people,
not of the bare majority of a few representatives; which is often procured
by little arts and great industry”; but on the other he talked in terms of
“an usurping populace” leading to tyranny (Swift 1701: 44–5, 47, 52,
54). This ambiguity was seized on by Charles Leslie, who asked “how
shall this universal bent and current of the people be known,” if not from
the people’s representatives? The only alternatives, he suggested, were
mob rule or polling “the whole nation,” both of which Leslie believed
to be absurd (Leslie 1703: supplement, 10). Daniel Defoe did suggest
an alternative to Leslie’s mob. Whilst supporting Swift’s assertion about
the dangers of partisanship, and upholding popular sovereignty, he made
it clear that when he was “speaking of the rights of the people” he was
to be understood as talking about “the freeholders” on the grounds that
“Property is the foundation of power” (Defoe 1702: 19).
Swift was answered in The History of the Last Parliament (1702). This
viewed the attempt to prioritize the collective rights of the people as a
partisan attempt by the embattled Whigs “to batter the Power of the
House of Commons, under pretence that they were not intrusted with
the whole Power of the Commons of England.” The manipulative new
Whigs, it was alleged, merely wanted “to deceive us in the Sense of
the People and . . . palm upon us Mercenary Noise and Clamour for the
general Voice and Consent of the People.” A “few Men, back’d by a
Faction” wanted to “exasperate the People against their Representatives,
and by Surprise, and false Representations to make ’em Accessary to the
Destruction of their own Rights and Authority” (Drake 1702a: 128, 131,
138).12 The author of this, James Drake, followed it with Some Necessary
Considerations Relating to Future Elections (1702), in which he further
argued that “the People ought to consider, that when they have made
the choice of their Representatives, they have parted with their Power
and lodg’d it in them for so long a time as they continue to be such,” an
argument that Rousseau was to make half a century later (Drake 1702b:
6).
The Kent petition and the Legion Memorial had not been the
only form of petition to raise the question of the nature of pop-
ular rule. Seventeen instructions presented to MPs after parliamen-
tary elections were printed collectively in The Electors’ Right Asserted

12 Written mostly by James Drake, but with some assistance from Anthony Hammond.
 Participation and representation before democracy 51

(Anon. 1701c).13 The tract countered claims that MPs were “left to
the absolute freedom of their own wills, to act without controul,” for any
MP carried the trust reposed in him by electors which he “ought carefully
and faithfully to discharge; or else how can he be properly called their
Representative?” Hence it was the “antient custom, continual usage and
undoubted right of the freeholders” and electors to deliver instructions
to their “delegates.” The point was made in the instructions themselves.
Gloucestershire thus took it “to be our undoubted Right, to consult with
and communicate our Advice and directions to you, now chosen our rep-
resentatives in Parliament, and from time to time by petition according
to law, to express our sentiments to the whole House of Commons.” The
Bristol electors agreed that “it is no doubt to us that we have a right to
direct our Representatives” (Anon. 1701c: 1–4, 11, 13).14 The role of
instructions to MPs, and hence of the representative nature of MPs, was
to reverberate for a very long time (Kelly 1984). Instructions were used
in 1681, 1701, 1715, 1733, 1753, 1769, 1774, 1783–4, and 1832. The
debate that Burke raised over precisely this issue in 1774 was thus part
of a much wider and longer one (Dickinson 1976).

III. Competing representative claims

The 1701–2 debates serve to highlight and introduce a number of points.
First, the boundaries between formal and informal representation were
contested very early in the emergence of representative government and
remained contested. Petitions and addresses were seen by many as a legit-
imate and representative voice of the people, but disputed by others
who asserted parliamentary sovereignty. Thus, on the one hand, parlia-
ment was championed as the genuine national voice. As one tract put it,
attacking the provincial petitions of 1701, “The interest of England is a
compound concern, and as solid judgments cannot be made of it, with-
out a due regard to every part of the island; which cannot be expected
from a better place than the house of commons, where are representatives
from every corner” (Anon. 1702b: 7).15 On the other, in predemocratic
Britain, legitimate representative authority was also vested in a number

13 See also Anon. (1701f). 14 Not all instructions made such explicit claims.
15 Anon. (1701e: 15, 47) argued that “their praying the House to have regard to the voice
of the people is nonsense; for every little faction lays claim to that appellation and have
wore it so threadbare that tis scandalous to make use of it, as appropriating it to a party;
for none can be truly called the people of England in a divided capacity; and they are
only whole and entire in their representatives in Parliament”; “the Kentish petition was
not the petition of the English nation but of a few giddy self-conceited people in a corner
of the kingdom.”
52 Mark Knights

of formal institutions besides parliament. Grand juries, for example, and

corporations had legal power to petition. As we have seen, local gover-
nance empowered mayors, aldermen, common councilors and freemen
with representative power. Problems occurred when these “local” rep-
resentatives disagreed with each other, as in 1681 when the Derbyshire
grand jurors formulated a Tory address at a time when the county had
elected two Whig MPs yet claimed themselves to be “the representa-
tives of the body of your majesties county” (London Gazette #1718). Or
in 1710 when Gloucestershire’s MPs reacted against an address from a
grand jury by promoting a rival text. As the Whig John Oldmixon showed
in a history of addresses in 1712 which lamented the clamor of the Tories,
the disparity between the voice of the people over time, or between the
voice of the addresses and the voice of the electorate, could lead to cyn-
icism about the apparent contradictoriness and giddiness of the people.
“We are sorry to see them changing like the wind, and fluctuating like
Tides,” he remarked (Oldmixon, 1709–10: 215). But such frustration
was itself a recognition of multiple representative voices. Informal rep-
resentation complicated the story of formal representation; and this was
inherent in the premodern British system.
Second, rival or contested petitions raised anxieties about misrep-
resentation. In 1710, for example, several tracts sought to expose the
fraudulent practices of loyalist addresses. The Principles and Designs of the
High-Church Party Discover’d (1710) revealed that in one county, Buck-
inghamshire, very few signatures had been obtained, even though it
claimed to show “the general bent of this county.” Moreover, it was
alleged, the high churchmen had deliberately used a populist slogan –
the church in danger – to gull the people to sign (Anon. 1710b: 7–8, 12).
Similarly The High-Church Mask pull’d off or Modern Addresses Anatomized.
Designed chiefly for the Information of the Common People (1710) noted the
range of misrepresentative practices: “menaces and promises used, false
stories invented, persons and actions misrepresented . . . violent sermons
preached to incense the vulgar. In populous places, through persuasion,
persons signed to rolls abstracted from the addresses, to which they were
afterwards affixed” (Anon. 1710a: 3). Informal representation, it was
alleged, was thus every bit as vulnerable to corruption as the electoral
system. Such malpractice was exposed by the press; yet the press was
also increasingly seen as part of the problem. England’s Enemies Exposed
(1701) thus attacked the Kent petition, Legion Memorial, and several
other Whig pamphlets as the product of a partisan propaganda machine.
The tract thus sought “to disabuse the nation, and recover the people
from the false notions and prejudicate prepossessions, imposed on them
 Participation and representation before democracy 53

by a turbulent malcontented and seditious party, who would lead them

blindfold into dangerous errors, set them at variance with their repre-
sentatives and put them out of love with parliaments” (Anon. 1701d:
dedication). Informal (and formal) representation relied on information
that was itself a possible partisan misrepresentation.
Third, informal representation helped to strengthen a key principle
of representative government, for competing representative claims led to
statements in many addresses that they represented the majority rather
than the whole. In other words the majority principle came to be embed-
ded not just through or in formal incorporations and elections but also
through petitions and addresses. Sandwich’s 1681 loyal address thus
came from “most of the jurats, most of the common council . . . and oth-
ers,” against the “faction” formed by the “chief magistrate and other
his adherents, who formerly have and still do use all possible arts and
endeavours to frustrate and stifle all discoveries of duty and loyalty”
(London Gazette #1735). The majority had thus to struggle against a
misrepresentative minority. Lancashire’s loyal address of 1713 insisted
that it was “the general, the almost unanimous voice of your people,
in great contempt of a clamorous but inconsiderable faction” (London
Gazette #5145). On occasion, however, it was recognized that the cor-
rupting influence of a faction had indeed prevailed so far that a minor-
ity had to be legitimized by reference to their loyalty rather than their
numbers.16

IV. Conclusion
This chapter has attempted to show the type of predemocratic partic-
ipatory representative society that existed in the seventeenth and early
eighteenth century, and again at the end of the eighteenth into the nine-
teenth century. The potentially highly participatory state offered many
representative moments and contained many overlapping institutions that
could each claim legitimately to represent the public. Non-electoral rep-
resentative institutions thus thrived in premodern Britain. One of the
most powerful and most widely deployed of these was the petition and
its close associate, the address. Through them both the franchised and
unenfranchised could find a voice, particularly after the advent of a free
press. Participation in campaigns reflected a popular frustration with the
notion of MPs as fully authorized representatives to whom they were
bound either by their own direct electoral consent or by their alleged

16 See, for example, Nottingham’s 1713 loyal address (London Gazette #5128).
54 Mark Knights

tacit consent to parliament’s national representative status. These con-

clusions suggest that Manin’s model of representation is overly schematic,
for the “triumph of elections” was never complete and often contested;
that electoral representation overlapped with a participatory structure of
local authority; that participation and representation could thus (albeit
sometimes uncomfortably) coexist; and that in Britain there were well-
established mechanisms for ensuring that it did. Debate about popular
rule over representatives was widespread in the predemocractic era when-
ever petitions and addresses were presented in any great number. Even
so, informal means of representation, particularly in an era of partisan
politics and a free press, carried with them their own set of anxieties
about misrepresentation and corruption; it was not just the electoral
system that was accused of distorting the voice of the people. Yet such
accusations themselves stimulated an important debate about the nature
of representation.
Finally, it is worth considering how these issues relate to more modern
concerns. There are some signs that the very old tradition of petitioning
might be being revived, perhaps as a modern parallel of the premodern
citizens who felt that their authorized representatives needed to be made
responsive to their concerns. There are now many electronic petition
sites, including one at Downing Street that in 2006 attracted 1.7 million
signatories to a single petition. A paper petition about threats to close
post offices recently collected over 4 million signatures, making it the
largest in British history.17 Whatever the issue, and in whatever form,
petitioning might be seen as an important institutional balance to the
idea that a representative has consensual authorization through elections
to declare the will of the people. As Hanna Pitkin put it in her seminal
study of representation, “in a representative government the governed
must be capable of action and judgment, capable of initiating govern-
ment activity, so that the government may be conceived as responding to
them . . . a representative government requires that there be machinery
for the expression of the wishes of the represented, and that government
respond to these wishes unless there are good reasons to the contrary”
(Pitkin 1967: 232–3). Elections may be the best form of such machin-
ery; but the premodern state had others that acted as important cogs
in it.

17 See http://petitions.pm.gov.uk/ (accessed January 10, 2007). The attention that the
e-petition received, perhaps because of its novelty, outstripped that given to the 4 million
strong petition in October 2006: one thousand sub-postmasters were on hand to present
a petition so large it covered the bottom level of a London double-decker bus. See
www.telegraph.co.uk/news/main.jhtml?xml=/news/2006/10/19/npost119.xml (accessed
September 12, 2008).
 Participation and representation before democracy 55

BIBLIOGRAPHY
Anon. 1701a. The Ballad or some scurrilous reflections in verse . . . answered.
London.
1701b. The Claims of the People, Essayed. London.
1701c. The Electors’ Right Asserted. London.
1701d. England’s Enemies Exposed. London.
1701e. The History of the Kentish Petition, Answer’d Paragraph by Paragraph.
London.
1701f. Jura Populi . . . Answer’d. London.
1701g. A Letter from Some Electors to one of their Representatives in Parliament
shewing the Electors Sentiments. London.
1702a. The Legionites Plot. London.
1702b. A Letter to a New Member of the Ensuing Parliament. London.
1710a. The High-Church Mask pull’d off or Modern Addresses Anatomized.
Designed chiefly for the Information of the Common People. London.
1710b. The Principles and Designs of the High-Church Party Discover’d. London.
Black, Eugene Charlton. The Association. British Extra-Parliamentary Political
Organisation, 1769–1793. Cambridge, MA: Harvard University Press.
Braddick, Michael J. 2000. State Formation in Early Modern England 1550–1700.
Cambridge: Cambridge University Press.
Braddick, Michael J. and John, Walter. 2001. Negotiating Power in Early Modern
Society: Order, Hierarchy and Subordination in Britain and Ireland. Cambridge:
Cambridge University Press.
Bradley, James. 1990. Religion, Revolution, and English Radicalism: Nonconformity
in Eighteenth-Century Politics and Society. Cambridge: Cambridge University
Press.
Brewer, John. 1988. The Sinews of Power: War, Money and the English State, 1688–
1783. Cambridge, MA: Harvard University Press.
Clark, J. C. D. 1985. English Society 1660–1832. Cambridge: Cambridge Univer-
sity Press.
Clark, Peter (ed.). 2000. Cambridge Urban History of Britain 1540–1840. Cam-
bridge: Cambridge University Press.
Colley, Linda. 1996. Britons: Forging the Nation 1707–1837. London: Vintage.
Collinson, Patrick. 1986–7. “The Monarchical Republic of Queen Elizabeth I.”
Bulletin of the John Rylands Library 69: 394–424.
Defoe, Daniel. 1701. Mr S_r, The Enclosed Memorial you are charg’d with. London.
1702. The Original Power of the Collective Body of the People of England Examined
and Asserted. London.
Dickinson, Harry T. 1976. “The Eighteenth Century Debate on the Sovereignty
of Parliament.” Transactions of the Royal Historical Society, 5th ser., 26: 189–
210.
1994. The Politics of the People in Eighteenth-Century Britain. Basingstoke:
Macmillan.
Drake, James. 1702a. The History of the Last Parliament. London.
1702b. Some Necessary Considerations Relating to Future Elections. London.
Drescher, Seymour. 1986. Capitalism and Anti-Slavery. Basingstoke: Macmillan.
56 Mark Knights

Eastwood, David. 1997. Government and Community in the English Provinces,

1700–1870. Basingstoke: Palgrave Macmillan.
2002. “Local Government and Local Society,” in A Companion to Eighteenth
Century Britain, ed. H. T. Dickinson. Oxford: Blackwell, 40–54.
Fletcher, Anthony. 1981. The Outbreak of the English Civil War. London: Edward
Arnold.
Fraser, Peter. 1961. “Public Petitioning and Parliament before 1832.” History
46: 195–211.
Gauci, Perri. 2001. The Politics of Trade. The Overseas Merchant in State and Society
1660–1720. Oxford: Oxford University Press.
Goldie, Mark. 2001. “The Unacknowledged Republic,” in The Politics of the
Excluded c. 1500–1850, ed. T. Harris. Basingstoke: Macmillan, 153–94.
Halliday, Paul. 1998. Dismembering the Body Politics: Partisan Politics in England’s
Towns, 1650–1730. Cambridge: Cambridge University Press.
Handley, Stuart. 1990. “Local Legislative Initiatives for Economic and Social
Development in Lancashire 1689–1731.” Parliamentary History 9: 231–84.
Harris, Tim. 1987. London Crowds in the Reign of Charles II. Propaganda and
Politics from the Restoration until the Exclusion Crisis. Cambridge: Cambridge
University Press.
Higgins, Patricia. 1973. “The Reactions of Women, with Special Reference to
Women Petitioners,” in Politics, Religion and the English Civil War, ed. Brian
Manning. London: Edward Arnold, 179–224.
Hindle, Steve. 1999. “Hierarchy and Community in the Elizabethan Parish: The
Swallowfield Articles of 1596.” Historical Journal 42 (3): 835–51.
Hobbes, Thomas. 1651. Leviathan, ed. Richard Tuck, 1996 edn. Cambridge:
Cambridge University Press.
Hoppit, Julian (ed.). 1997. Failed Legislation 1660–1800. London: Hambledon.
Hoyle, Richard. 2002. “Petitioning as Popular Politics in Sixteenth-Century Eng-
land.” Historical Research 75: 365–89.
Kelly, Paul. 1984. “Constituents’ Instructions to Members of Parliament in the
C18th,” in Party and Management in Parliament 1660–1784, ed. Clyve Jones.
Leicester: Leicester University Press, 103–67.
Knights, Mark. 2005. Representation and Misrepresentation in Later Stuart Britain:
Partisanship and Political Culture. Oxford: Oxford University Press.
Leslie, Charles. 1703. The New Association. Part II. London.
London Gazette. Various issues. London.
LoPatin, Nancy. 1999. Political Unions, Popular Politics and the Great Reform Act
of 1832. Basingstoke: Macmillan.
Maltby, Judith. 1998. Prayer Book and People in Elizabethan and Early Stuart
England. Cambridge: Cambridge University Press.
1999. “Petitions for Episcopacy and the Book of Common Prayer on the Eve
of the Civil War, 1641–1642,” in From Cranmer to Davidson: A Church of
England Miscellany, ed. Stephen Taylor (Church of England Record Society,
7). Woodbridge: Boydell and Brewer, 103–67.
Manin, Bernard. 1997. The Principles of Representative Government. Cambridge:
Cambridge University Press.
Oldmixon, John. 1709–10. The History of Addresses (2 parts). London.
 Participation and representation before democracy 57

Parliamentary Papers 1852–3, vol. LXXXIII. London: House of Commons

Parliamentary Papers Online. Proquest database.
Parssinen, T. M. 1973. “Association, Convention and Anti-Parliament in British
Radical Politics 1771–1848.” English Historical Review 88: 504–33.
Phillips, John. 1980. “Popular Politics in Unreformed England.” Journal of Mod-
ern History 52: 599–625.
Pickering, Paul. 2001. “And your Petitioners &c: Chartist Petitioning in Popular
Politics 1838–48.” English Historical Review 116 (2): 368–88.
Pitkin, Hanna. 1967. The Concept of Representation. Berkeley: University of
California Press.
Somers, Lord John. 1701. Jura Populi Anglicani. London.
Stevenson, John. 1989. “‘Paineites to a Man’?: The English Popular Radical
Societies in the 1790s.” Bulletin of the Society for the Study of Labour History
54 (3): 14–25.
Sweet, Rosemary. 1999. The English Town 1680–1840: Government, Society and
Culture. London: Longman.
2003. “Local Identities and a National Parliament, c. 1688–1835,” in Par-
liaments, Nations and Identities in Britain and Ireland, 1660–1850, ed. Julian
Hoppit. Manchester: Manchester University Press, 48–63.
Swift, Jonathan. 1701. A Discourse of the Contests and Dissensions between the Nobles
and the Commons in Athens and Rome. London.
Taylor, A. A. 1913. Statistics Relating to Sittings of the House, &c. Typescript
returns of the House of Commons.
Tilly, Charles. 1995. Popular Contention in Great Britain 1758–1834. Cambridge,
MA: Harvard University Press.
Urbinati, Nadia. 2006. Representative Democracy. Principles and Genealogies.
Chicago: University of Chicago Press.
Withington, Phil. 2005. The Politics of Commonwealth. Citizens and Freemen in
Early Modern England. Cambridge: Cambridge University Press.
Wrightson, Keith. 1996. “The Politics of the Parish in Early Modern England,”
in The Experience of Authority in Early Modern England, ed. Paul Griffiths,
Adam Fox and Steve Hindle. Basingstoke: Macmillan, 10–46.
Zaret, David. 1996. “Petitions and the ‘Invention’ of Public Opinion in the
English Revolution.” American Journal of Sociology 101: 1497–1555.
2000. Origins of Democratic Culture: Printing, Petitions and the Public Sphere in
Early Modern England. Princeton, NJ: Princeton University Press.
Part II

Theories of political representation

3 Varieties of public representation

Philip Pettit

I. Background and basics

Systems of representative government, I shall assume, are designed to
give control over government to the people. Far from being an alternative
to democracy, as some have taken them (Manin 1997), they embody an
institutional framework – or rather a family of frameworks – for realizing
the democratic ideal of giving kratos to the demos, power to the people.
The distinction between a participatory and a representative system is
not one between democracy proper and some faint approximation but a
distinction between rival proposals for the implementation of democracy.
My focus in this chapter is on representation in this democratic, pop-
ularly enabling sense. Thus the target of the chapter is narrower than
it might have been. As Hobbes in particular argues, the idea of repre-
sentation may be used, not just of representatives who are subject to
the continuing or periodic control of the people, but also of a hereditary,
absolute monarch. The defenders of parliament in the 1640s tried to give
its members a monopoly right on the use of the word (Skinner 2005),
but Hobbes argued against them that it was absurd that a monarch who
“had the sovereignty” over his subjects “from a descent of 600 years”
should not be “considered as their representative” (Hobbes 1994: 19.3).
His own view, to the contrary, was that “the King himself did . . . ever
represent the person of the people of England” (Hobbes 1990: 120).
But though my focus is narrower than Hobbes’s, it is broader than the
target to which many contemporary theorists give their attention. As will
appear, I use the notion of representation in such a way that any pub-
lic authorities, and any citizens who assume a legitimate role in public

I benefited enormously from discussion at the Yale conference where a first version of
the chapter was presented in November 2006; from discussion and commentary (by Ian
McMullin) at a seminar in Washington University, St. Louis, in April 2008; from exchanges
with Eric Beerbohm, Nate Kemp, Frank Lovett, Evan Oxman, and Andrew Rehfield; and
from the comments of two anonymous referees. I am grateful to Bryan Garsten for directing
me to expressions of the indicative ideal among anti-federalist writers.

61
62 Philip Pettit

discourse, may make a legitimate claim to represent the people (Richard-

son 2002). Others apply the notion, however, only to elected represen-
tatives – elected members of the legislature and, where relevant, of the
executive and the judiciary. The concept of electoral representation is
more tightly circumscribed than my looser concept of democratic rep-
resentation, but I think that it does not offer the same generality of
perspective.1 I hope that this chapter may help to vindicate that claim.
There are three factors in any relationship or system of representation,
whether in my sense of representation or in any other. First there are the
representatives or, using a seventeenth-century word, the representers; I
prefer this term since the other has an exclusively electoral connotation.
Second, there are the represented or, as I shall say, the representees. And,
third, there is the relationship that exists between those two parties: the
representation that is exercised by representers on behalf of representees.2

Representers
Representers may be individual agents or groups of individual agents.
And in the case where a group serves in this role, the members may each
act for their own ends, according to their own judgments, or they may
act on a shared intention to further this or that end. In this latter case the
individuals will each intend that together they promote the agreed end
and they will each do their bit for the promotion of the end, expecting that
others will play their parts too (Bratman 1999); they will be a cooperative
grouping, not a mere collection of individual agents.
The cooperative grouping that combines around a joint intention
comes itself in two forms. The members may act for shared goals, now
on this occasion, now on that, without ever forming a joint intention
governing their continuation over time. Or they may form the special,
shared intention that over time they together should constitute a corpo-
rate agent or agency: a body that simulates the performance of a single
agent with a single mind. The intention shared in this case will be that
they together cooperate in the organized pursuit of agreed ends accord-
ing to agreed judgments. The ends will usually be an evolving set of ends
selected under agreed procedures, and the judgments an evolving body
of judgments selected under agreed procedures (Pettit and Schweikard
2006). The distinctions are mapped in Figure 3.1.

1 For a very congenial and insightful account of representation in a broader sense than that
of electoral representation, see Rehfeld (2006).
2 On the emergence and development of the concept of representation in the late seven-
teenth and early eighteenth century, see Knights (2005).
 Varieties of public representation 63

Representers

Individual Group

Non-cooperative Cooperative

Non-corporate Corporate
Figure 3.1 Representers.

These distinctions are readily illustrated. The member of the legislature

who represents a certain constituency or the President who represents
the people as a whole exemplifies the individual representer. And the
legislature or the executive as a whole exemplifies the group representer.
The members of such a group may behave as a collection of independent
agents, each with their own brief, or as a cooperative grouping. And
if they behave as a cooperative grouping, then they may or may not
incorporate in the manner of a body with a coherent, evolving set of
goals and judgments.
The US Congress might be seen as an unincorporated, cooperative
grouping that does battle, now on this issue, now on that, looking in
each case to see if a more or less ad hoc majority can be assembled to
support a certain line. There is some concern with securing coherence
between the lines supported over time, of course, as well as coherence
with the Constitution. But this concern can take second place to the
other concerns of members – say, their concern to display their colors
back home. This will especially be so if in any case a presidential veto
is likely, or if the Supreme Court is expected to reject or reinterpret the
legislation.
The Westminster parliament might be seen, by contrast, as an incor-
porated, cooperative agency whose members assume a higher degree of
responsibility for legislating on coherent lines, at least within the time-
frame of a given parliament. The members acquiesce in a procedure
whereby a fixed majority will be established and that majority – say, a
64 Philip Pettit

single, incorporated party – will bring forward a coherent program of

legislation to be enacted after parliamentary discussion in the name of
the parliament as a whole.

Representees
So much for the possibility of variation on the side of the representers
in a democratic system. There is a corresponding degree of variation on
the side of the representees. The representee may be a single individual,
as when the member of a legislature takes up some cause on behalf of a
constituent. Or the representee may be a group. The group represented
may be cast as a mere collection, such as perhaps the electors in a given
district, or they may have a more cooperative aspect. And the cooperative
representee may be incorporated or unincorporated. In other words the
distinctions on the side of the represented may correspond to those that
we map on the side of the representing.
The loose pressure group or the ethnic minority that succeeds in find-
ing a spokesperson in the legislature or elsewhere will often be an unin-
corporated entity whose members are united around just a single issue.
But equally an entity with a corporate or quasi-corporate form of organi-
zation may figure as representee. The states that are represented by the
Senate of the US or Australia are surely entities of this kind.
The ultimate representee in any democratic system will be the people
as a whole: a body that might be taken as a mere collection but is usually
depicted as an incorporated entity. It is of the essence of a democratic
system that it is supposed to create a state that represents the people as
a whole, acting in their name on the international stage and in dealings
with individual citizens and groups of citizens (McLean 2004). Thus
John Rawls (1999) says that the government in a well-ordered society –
ideally, a liberal democratic society – will be “the representative and
effective agent of a people” (ibid.: 38) – “the political organization of
the people” (ibid.: 26). The disordered society where a small group
usurps power, like the unordered society where political organization
fails utterly, will be marked by the absence of precisely this system-wide
level of representation.

Representation
Not only are there variations of these kinds in who are represented
and who do the representing: that is, in the relata at either end of the
representative relationship. There are also variations in the nature of
the relationship. I argue that there are two fundamentally contrasting
 Varieties of public representation 65

Representation

Indicative (proxies) Responsive (deputies)

Directed (delegates) Interpretive (trustees)

Figure 3.2 Representation.

varieties of representation, indicative and responsive. Indicative repre-

senters stand for the representees in the sense of typifying or epitomizing
them; how they act is indicative of how the representees would act.3
Responsive representers act for or speak for the representees, playing the
part of an agent in relation to a principal; how they act is responsive to
how the representees would want them to act. Both sorts of representa-
tion, so I shall assume, have to be authorized by the representees.
Authorized indicative representers I describe as proxies, authorized
responsive representers as deputies. Deputies divide, in a traditional dis-
tinction, into delegates who are more or less explicitly directed by repre-
sentees and trustees who have interpretive discretion in determining how
to construe their representees. The distinctions appear in Figure 3.2.
Three metaphors have dominated the tradition of thinking about the
meaning of representation (Skinner 2005). One of these metaphors is
drawn from representation in the pictorial arts, and it maps onto indica-
tive representation. As the painting is indicative of how the subject of
the painting looks, so on this image should representers be indicative
of representees: they should be proxies. The other two metaphors are
drawn respectively from the courts and from the theater: from the way
in which an attorney represents a client, and an actor represents a char-
acter. These reflect the two different modes of responsive or deputy-
style representation: one directed, the other interpretive. As the attor-
ney acts under the explicit or implicit direction of a client, so should

3 Pitkin (1972) casts the indicative relationship as descriptive representation: as involving

nothing more than the sort of relationship that holds between statistical sample and a
population. But she misses the fact that a representee population may appoint a descrip-
tive representer with a view to having things done as it would do them, or want them
done; the possibility is nicely illustrated by the British Columbia citizens’ assembly that
I discuss later. It is only such controlled representation that I describe as indicative.
66 Philip Pettit

representers act under the explicit or implicit direction of their repre-

sentees: they should be delegates. As the actor constructively interprets
the mind of a character, so should representers interpret the mind of
representees: they should be trustees.
The following two sections review these two conceptions of representa-
tion, looking at how they might apply with different sorts of representers
and representees but focusing in particular on the representation of a
people. The fourth section then asks how these forms of representation
might be organized in a democracy, focusing on a particular problem that
is raised by responsive representation of the interpretive kind. In order to
illustrate the usefulness of the distinctions, the paper concludes with an
appendix that considers issues of representation in the presidential and
parliamentary systems that are exemplified respectively by Washington
and Westminster.

II. Indicative representation

The standard case

In the standard form of this first conception, the assumption is that there
are a number of representers, that the representees are the people as
a whole, and the indicative requirement is that the representers should
be a reliable or representative sample of the representees. They should
faithfully reproduce significant differences among the population, and
reproduce them in proportion to their realization within the community.
We are familiar nowadays with the idea of indicative representation in the
context of statistics. The random sample that is polled in the statistical
exercise is a representative sample in the sense of being a good indicator
of the population as a whole.
This conception of representation goes naturally with the pictorial or
figurative metaphor. As Quentin Skinner (2005: 163) argues, it appears in
those parliamentarian writers in England of the mid-seventeenth century
who look for a “speaking likeness” of the people in those who rule them,
“describing Parliament as a ‘representation’ – a picture or portrait – of
the body of the people.”
The relationship that is envisaged between the body of representa-
tives and the body of the people is certainly figurative, as in the likeness
between a portrait and the subject of the portrait. But it might best be
conceptualized as the relationship whereby a painting or sculpture, par-
ticularly one of an abstract kind, can exemplify a feature of the subject:
the mass of a body, the elegance of a movement, and so on. As a tailor’s
swatch contains pieces of cloth that exemplify the purportedly significant
 Varieties of public representation 67

features of various fabrics, so figurative works exemplify purportedly

significant aspects of an object or substance or situation or whatever
(Goodman 1969). And as figurative representations can play this part to
an aesthetic purpose, so politically representative bodies ought to serve
in a parallel, democratic role. They ought each to exemplify and indicate
the presumptively significant aspects of the people they represent, repro-
ducing salient variations amongst its members and in a proportion that
corresponds to their distribution in the population.
The idea of indicative representation figures early in democratic theory,
since it is the sort of representation that is achieved or is likely to be
achieved under the lottery system that was favored by the Athenians and
that also played an important part in later regimes such as those of the
Italian city-republics (Waley 1988; Hansen 1991). This lottery system
might be taken as a version of the technique of random sampling, but
random sampling put to use in the service of advancing goals espoused
by the people as a whole. While it may have been motivated by a desire
to have a regular turnover in the representer body, the important thing
from our viewpoint is that it would have ensured a degree of proportional
and indicative representation.
The indicative idea also appears in the jury system, as that was devel-
oped in medieval Europe (Abramson 1994). To be subjected to the judg-
ment of one’s peers, whether in determining that there is a legal case to
answer, or that one is legally liable, is to be exposed, not to a random
arbiter – a chance enemy, perhaps – but to a body that stands in for
the community as a whole. The idea is that the jurors should represent
a cross-section of the community, or at least of the fully enfranchised
members: in medieval Europe, the mainstream, propertied males.4
The ideal of indicative representation was entrenched in the thinking of
many of those associated with the American War of Independence and the
French Revolution. Melanchton Smith could write in 1788, in opposing
the American Constitution: “The idea that naturally suggests itself to our
minds, when we speak of representatives is, that they resemble those they
represent; they should be a true picture of the people” (Ketcham 2003:
342). Again, it was powerfully endorsed in a speech given by Mirabeau
to the French Constituent Assembly in January 1789, though he used
the image of a map rather than a picture to get it across. According to
this version of the model:

4 It is significant that a supporter of the anti-federalist cause in 1787 could complain that in
the enlarged United States there would not be a representative body in legislature or jury
“which possesses the same interests, feelings, opinions, and views the people themselves
would were they all assembled” (Ketcham 2003: 265).
68 Philip Pettit

a representative body is to the nation what a chart is for the physical configuration
of its soil: in all its parts, and as a whole, the representative body should at all
times present a reduced picture of the people – their opinions, aspirations, and
wishes, and that presentation should bear the relative proportion to the original
precisely as a map brings before us mountains and dales, rivers and lakes, forests
and plains, cities and towns. (Pitkin 1969, 77)

With the growth of electoral machinery, the indicative idea was natu-
rally applied to elections for the legislature, providing support for making
the electoral system more and more proportional (Mill 1964). Is it also
behind the practice of organizing the legislature around geographically
dispersed districts? It is hard to believe that it did not play some role in
justifying that practice but the evidence, according to Rehfeld (2005), is
against this hypothesis. Still, districting does induce a similarity in one
dimension – nowadays a fairly unimportant one – between the population
as a whole and the legislature that represents it.
The indicative idea survives in the continuing enthusiasm for pro-
portional representation and has been given new life in campaigns for
supplementing electoral representation with novel, statistically represen-
tative bodies. It is there in the general policy of organizing citizens’ juries
that would review various policy issues (Stewart, Kendall, and Coote
1994). It is present in the notion of the deliberative opinion poll that is
chosen as a random sample and then canvassed for its view on one or
another issue at two separate times: first, before members of the sample
make contact, and, second, after they come together to receive back-
ground information, to hear different points of view, and to debate the
right line to take on the issue under consideration (Fishkin 1997). A
particularly striking example appears in the citizens’ assembly that was
recently established in the Canadian province of British Columbia (Fung
and Fagotto 2006; Ferejohn 2007). A more or less representative sample
of 160 citizens was assembled and given the task, over much of 2004,
of reviewing the existing electoral system in the light of various hearings
and discussions, and making a recommendation on whether or not it
should be amended. The group recommended a change that then went
to referendum and won more than 50 percent support – just short of the
quota required to trigger a change.5
The similarity between representer and representee does not mean, in
itself, that the representee has any degree of control over the representer.

5 Since everything is a perfect indicator of itself, a limit case of indicative representation

is the participatory democracy where the whole population is present to vote, not just
a sample. Far from being cast as the contrast point for indicative representation, the
compulsory, participatory arrangement can be seen as a special case. The case is so
special, however, and so infeasible, that I ignore it in this discussion.
 Varieties of public representation 69

The connection between them is not one under which the representer
has to track the dispositions of the representee, and respond by acting in
conformity to those dispositions. Rather it is a connection under which
the representer is a good indicator or model of how the representee might
have been disposed to act or speak, if in the position of the representer.
Thus the people of British Columbia did not control the citizens’ assem-
bly but, given that the assembly was a microcosm of the people, it might
be thought to model the decision that the people as a whole would have
made had it been given the same information.
But while the indicator relationship does not necessarily give control
over representers to representees, it is possible for representees to exercise
control by selecting suitable indicators as their representers, by subjecting
them to constraints designed to ensure that they remain indicators, and
by deselecting them if they fail to act appropriately. Such authorized,
indicative representers, as indicated earlier, I describe as proxies of the
representees; they stand in for those who determine their selection.
Suppose that I see someone who thinks like me, as we say, and is a
good indicator in a certain domain of how I tend to respond on various
issues. While I do not directly control what he or she does, I do assume
a position of indirect control when I decide to authorize that person to
act in my stead, say on a committee. By appointing that person rather
than someone else, I make it more likely that the committee will act in a
congenial manner: act, as they would have been led to act, had I myself
been a member. And this indirect control can be further strengthened by
additional measures. I may put other constraints or incentives in place
that make it more likely that the person will simulate what I would have
done on the committee: for example, I might take steps to insulate the
person against special, warping motives. And I may retain the power of
de-selecting the person, should he or she, for whatever reason, not serve
my interests well. The person is my proxy, someone who takes my place,
with my authority.

The extended case

These observations about the control that is possible with indicative
representers should prompt us to recognize that while indicative repre-
sentation has traditionally been associated with the proportional repre-
sentation of groups, the model can also apply in very different contexts.
Consider the professional or quasi-professional public figures who are
appointed by government, often under strict regulations and often with
a (limited or unlimited) tenure that is independent of those who do the
appointing. Obvious examples will be judges and prosecutors, public
70 Philip Pettit

auditors and ombudsmen, agency heads and the members of public

commissions. These figures will serve as proxies for the public, being
appointed under public authorization because of the likelihood that they
will serve the public well.
This likelihood may not be based on a similarity of nature between
representers and representees, but it will be based on a similarity in
the tendency of their wishes. There will be a convergence between the
goals the representers are professionally constrained and motivated to
implement – at least ideally – and the goals that the public would want to
promote in the domain of their operation. The representers are recruited
to public service, plausibly, on the grounds that their interests, if things
go well, are likely to be indicative of the interests of the public. And they
can be subjected to constraints that make it even more likely that this will
be so. Such figures can be cast as proxy representers of the people on the
same grounds that certain proportional bodies can be cast as proxies.
Think of those who serve on a central bank like the Federal Reserve in
the US. Or think of those who are members of an election commission
of the kind that is given responsibility in many countries, at arm’s length
from parliament, to draw the boundaries of electoral districts. Absent
dependency on the elected officials who appoint them – absent reliance
on those officials for continuing tenure, for example – such individuals
may be moved by motives of professional reputation to act reliably in a
manner that answers to the interest of the people as a whole. And this
will be especially likely if their performance is subject to public check and
commentary. It is surely more likely that an independent electoral com-
mission would act in this way, for example, than that elected members
of a parliament or congress would do so, since elected representatives
would have special party interests in the shape of district boundaries
(Pettit 2004).
The idea of appointing professional figures was not unknown to tra-
ditional democratic theory. A good example is provided by the official
known as a “podesta,” who played a very important role in the life of
twelfth- and thirteenth-century Italian city-republics (Waley 1988). The
podesta, who might serve at different times in different cities, was an
outsider appointed for six months to a year in order to run a city’s affairs.
He was given a range of issues to control, was often forced to act without
having contact with special interest groups within the citizenry, and was
subjected to intense scrutiny at the end of his tenure. Since he might
hope to be appointed as podesta by different cities at different times, he
would also have been susceptible to reputational motives for doing his
job to the satisfaction of the citizenry. Such a figure would have been a
 Varieties of public representation 71

good example, in our terms, of an indicative representer: a proxy agent

who takes the place of those he represents.
But it is not just public appointees who can count as indicative rep-
resenters in the extended sense we are examining. For another sort of
example, think of the whistle-blowers who can expose abuses in public
life, or the “private attorneys general,” as they have been called (Rabkin
1998), who serve the public well by challenging certain laws before the
courts. The fact that the public or people give whistle-blowers protec-
tion, and the fact that they give private attorneys general a license to use
the courts as they do, means that these figures are authorized to act in
their characteristic manner. In the aggregate they serve the public well,
by most accounts, acting as they do on convergent if not always identi-
cal interests. Their authorization by the public means that like formally
appointed officials, they can be seen as representers who are employed in
their characteristic roles, precisely because of the indicative relationship
between their dispositions and the presumptive dispositions of the people
as a whole.

III. Responsive representation

Directed responsiveness
Where the indicative idea is that a microcosm or model of the people
should rule because it is likely to be a good indicator of how the people
as a whole would decide the issues that come before it, the responsive
idea takes the complementary line. In this conception the representer
tracks what the representee wants and responds with appropriate action,
playing a very different role from that of being a passive indicator of the
representee’s disposition.6 The judicial metaphor in which the lawyer or
attorney acts for the client offers an illustration of the idea in which the
representee explicitly or implicitly directs the representer. As the lawyer
tracks the client’s wishes, so the responsive representer is to track the
wishes of the representee.

6 I take indicative and responsive representers to be indicative and responsive only in a

positive mode. If an indicative agent pursues something, that indicates that it is something
that the principal would favor; but if the agent does not pursue something, that does not
indicate that it is something the principal would not favor. If a principal has an interest in
something, the responsive agent will not have an interest in it but if the principal does not
have an interest in something that does not mean that the responsive agent will not have
an interest in it. This assumption means that a responsive agent may not be indicative,
as an indicative agent may not be responsive.
72 Philip Pettit

Under the responsive conception of representation, the wishes of rep-

resentees exercise a degree of control over how representers act, and
they do this with the authorization of the representees themselves; the
representative process materializes, if not at their initiative, at least with
their knowledge and acquiescence. Indicative representers are recruited
or licensed to act because they are or can be made to be indicative.
Responsive representers are recruited or licensed to act because they are
or can be made to be responsive. Where I describe representers who are
authorized to serve in an indicative role as proxies, I describe represen-
ters who are authorized to serve in a responsive role as deputies. We shall
see that apart from directed deputies who answer to the metaphor of the
attorney, there are also interpretive deputies who answer better to the
metaphor of the actor.
Responsiveness among representers may materialize spontaneously, as
a result of a natural love or reverence or fear of the representees, or it may
be manufactured by contextual incentives or constraints that elicit suit-
able attentiveness. Let the representees form an intention that something
be done by the representers and, at least in the directed version of respon-
sive representation, the representers will become disposed to do it. The
representers may respond to instructions issued by the representees –
that is, to the manifestations of their wishes – or they may not need any
such instruction, being disposed to respond to the wishes that the rep-
resentees manifestly hold. Such explicitly or implicitly directed deputies
I describe, in a received term, as delegates of those they represent: they
are agents who serve as voiceboxes for the represented (Pitkin 1972; but
see Urbinati 2000).
The category of delegates is often taken to be illustrated – rightly or
wrongly, as we shall see – by those who serve in the legislature, or per-
haps the executive, in sensitivity to the wishes of the electorate.7 The
control that the people exercise over such public representers may take
an active, hands-on form, as when the representees impose suitable con-
straints on representers or give them explicit instructions. But it may
often be just virtual in character, constituting a sort of hands-off, arm’s-
length control (Pettit 2001); while this may also be possible with indica-
tive representers, it is particularly salient with responsive ones. Hands-off
control will materialize when the representers behave in a congenial fash-
ion and the representees stand by, ready to reveal relevant attitudes, or
impose relevant constraints, only in the event that the representers stop

7 For a state-of-the-art assessment of political, responsive representation, see Mansbridge

(2003). Mansbridge (forthcoming) defends congenial views on indicative representation,
as I learned when this paper was in press.
 Varieties of public representation 73

performing to their taste. In this case the representees ride herd on the
representers, to use an image from cowboy life. They intervene in the
affairs of the representers, eliciting suitable motivation or information,
only on a need-for-action basis.8
Might the control of responsive representers be increased by ensuring
that they are indicative as well as responsive? The hope that it can be
increased in this way motivates the drive for ensuring proportionality
among elected deputies. The case for proportional, electoral representa-
tion is made in John Stuart Mill’s classic work on representation (1964
[1861]), as indicated earlier, and it is reinforced by contemporary theo-
rists such as Anne Phillips (1995) and Thomas Christiano (1996). There
are many reasons to think that salient divisions among a represented pop-
ulation should be reflected in an elected house of representatives. But it is
not so clear that ensuring such proportionality will ensure the indicative
reliability that we might expect, for example, in the British Columbia
citizens’ assembly.
The reason is that no matter how much proportionality is imposed on a
parliament or congress, the fact that members are chosen by election from
among those who are ready to stand at the polls introduces a framework
of motivation that is going to undermine the prospect of parliament
operating as a reliable model or microcosm of the people. Those who
stand for election are not going to be typical of the society, and even
if they are typical, their performance in the legislature is bound to be
affected by a special motive: the interest in being elected and reelected.
In order to constitute a model of the people, legislators would need to
go about their business without a thought for anything but what by their
lights it is best to do. And no elected officials can be expected to do
that; they are bound to be subject to what from the point of view of the
indicative model is a deeply warping influence.
As suggested, there is a serious question as to whether public, respon-
sive representers really can serve as delegates who are explicitly or implic-
itly directed by their constituents. Directed, responsive representation
supposes that the mind of representees is made up and manifested to
representers. There are explicit or implicit directives available to express

8 The notion of control employed across the distinctions we have made is quite univocal.
We can say that one party, A, exercises control over another party, B, to the extent that
A raises the probability – robustly raises the probability (Lovett 2007: 712–13) – that B
will behave in a congenial manner above the level that that behavior might have had in
A’s absence (Dahl 1957; Pettit 2007). A will have such an effect if A actively elicits the
deputy or proxy performance sought. But A will also have this effect in the case where A
stands ready to intervene on a need-for-action basis. A will guard against the possibility
that B has a change of mind and by raising the probability that B will behave congenially
even in that case, A will raise the absolute probability of such behavior.
74 Philip Pettit

the mind of representers and the responsive representational job is to

track those directives. The problem that we should now note, however, is
that representees in the public worlds rarely provide representers with a
fully formed, directive mind. Thus responsive representation often has to
become interpretive or constructive in a manner that evokes the metaphor
of the actor rather than that of the attorney.

Interpretive responsiveness
Where attorneys have to be guided by the explicit or implicit directives of
clients, actors bear a much less constrained relationship to the scripted
characters that they are required to play. The actor interprets the char-
acter, taking the spare lines of the dramatist and giving them life in a
pattern of emphasis and presence that makes interpersonal sense; the
interpretation lets the character portrayed be understood as a person
amongst persons. Representation in this interpretive sense is described
by Hobbes (1994: 16) as personation: an act in which the representer
speaks with authority for another, in particular for another individual or
group of individuals.
Like any form of responsive representation, the interpretive sort that
answers to this metaphor has to be authorized by the representees, at
least if it has democratic credentials. Representees have to authorize the
representers to speak for them, even though they do not provide the
representers with the words to use. Where responsive representers on
the attorney model are voiceboxes of the people, as we might say, respon-
sive representers on the actor model are spokespersons. Where I describe
attorney-like deputies as delegates, I follow tradition in describing actor-
like deputies as trustees (Pitkin 1972).
The authorization that representees give interpretive deputies means
that those who speak for them do not report the mind of their con-
stituents, as a journalist or opinion survey might report that mind. Sub-
ject perhaps to certain provisos, the representees are deemed to think
what their authorized spokesperson – or body of spokespersons – says
that they think. The authorization is not just a prediction to the effect
that the representers will be a pretty good guide to what they think. It is
a guarantee that, at least within certain limits and under certain condi-
tions, they can be taken to be minded as the representing words portray
them as minded. They can be held to those words.
Responsive representation has to be interpretive with an individual
representee if that person is a minor – here authorization may be supplied
by a court – or if the person does not express his or her mind on some
particular issue where the representer needs to act. But the context in
 Varieties of public representation 75

which responsive representation is most clearly required to be interpretive

is one of political representation, where there are a number of individuals
in the position of representees, and the representer is forced, at whatever
level of grain, to ascribe a coherent, enactable set of attitudes to them –
a single mind.
Suppose that one or more representers are acting for an unincorporated
collection or multitude. Whatever control is exercised by the representees
in the multitude it had better allow the representers to act for a consis-
tent set of goals, according to a consistent set of judgments; otherwise
the representative actions are liable to undercut one another. Or if that
requirement is thought too strict, on the grounds that many agents fall
short of consistency, then the control certainly ought to allow the repre-
senters to be responsive to charges of inconsistency. They should not be
required to have to admit that what they hold or seek on behalf of their
representees is an inconsistent package; else they and their constituents
would be a laughing stock.
Might the representees impose an arrangement under which they con-
trol what the representers do in a goal-by-goal or judgment-by-judgment
way, and still hope to satisfy such a coherence constraint? It turns out that
they could not. And that suggests that the responsive representation of a
multitude is bound at a certain point to become interpretive. Those who
serve in such responsively representative roles will have to be spokesper-
sons, not voiceboxes: they will have to be deputies in the trustee mould,
not in the mould of the delegate.
The most obvious way in which a multitude might hope to control
the actions of its representers would be by forcing those agents always to
follow the majority vote of its members. But the “discursive dilemma”
shows why this will not work in general (Pettit 2001: ch. 5). The votes of
entirely consistent individuals on logically connected issues can generate
a set of inconsistent positions on those issues. Suppose that A, B, and
C want to establish a majority collective view on three issues: whether
p; whether q; and whether p&q. They might be judges in a court who
have to decide on whether a plaintiff in tort did harm (p), had a duty
of care (q) and so was liable for damages (p&q) (Kornhauser and Sager
1993). It is perfectly possible for A, B, and C each to have a consistent
set of views on these issues and yet for the group to be forced by majority
voting into endorsing an inconsistent set.
That will happen, for example, if the votes that the individuals submit
are as follows. A and B vote that p, C votes against; B and C vote that q, A
votes against; and so only B votes that p&q. A and C vote that not-p&q,
thereby giving a majority verdict that p, that q, and that not-p&q. The
following matrix should make the pattern vivid.
76 Philip Pettit

P? Q? P & Q?
A Yes No No
B Yes Yes Yes
C No Yes No
Majority vote Yes Yes No

This paradox is of general significance, since it turns out that no vot-

ing system is guaranteed to produce a consistent output from consistent
inputs if it is to work with all patterns of input and if, roughly, it is to
treat both individuals and issues even-handedly (List and Pettit 2002).
The lesson is that if the representers of a multitude are required to enact
a coherent mind, furthering a consistent set of goals according to a con-
sistent set of judgments, then they cannot be controlled by representees
on an issue-by-issue basis. Such control is liable to provide representers
with an inconsistent set of attitudes to enact: a set of attitudes like the
judgments that p, that q, and that not-p&q.
The lesson, put more positively, is that in a case like this responsive
representation has to be interpretive or constructive. It has to involve,
not the faithful reflection of a pre-given mind, but a constructive inter-
pretation in virtue of which the multitude is imputed a coherent mind or
mentality. When that occurs, the multitude will assume the status of a
corporate agent.
The idea that the interpretive representation of a multitude can trans-
form it into a corporate agent is present in Western thought from at
least the fourteenth century. In 1354, Albericus de Rosciate could say
that a collegial agent, although it is constituted out of many members,
is one by virtue of representation: collegium, licet constituatur ex pluribus,
est tamen unum per representationem (Eschmann 1946: 33, fn 145). The
theme dominates the work of legal theorists of the time, such as Barto-
lus of Sassoferrato and his pupil, Baldus de Ubaldis, who make much
of the way represented groups, in particular the represented people of a
city, could figure as corporate agents or persons (Woolf 1913; Canning
1983). Arguing that the populus liber, the free people of a city-republic,
is a corporate person, Baldus explains that this is because the council
represents the mind of that people: concilium representat mentem populi
(Canning 1987: 198).
The theme reappears in the writings of Thomas Hobbes in the sev-
enteenth century. He makes representation, a term that he uses as an
alternative to personation, central to the possibility of a group’s creating
and enacting a single mind. “A multitude of men are made one person,
when they are by one man, or one person, represented.” Where does
 Varieties of public representation 77

the unity come from? From the fact that the representing individual – or
body – will speak with one voice, thereby testifying to one mind in the
group: “it is the unity of the representer, not the unity of the represented,
that maketh the person one” (Hobbes 1994: 16.13).

IV. Representation and democracy

Varieties of democratic representation

Democracy, on the Lincolnian formula, is government of the people, by
the people and for the people. Representation has to have a place in any
democratic constitution insofar as it is only by courtesy of representers
that government can hope to be by the people; popular participation is
inevitably restricted to the occasional referendum.
There is obviously room in a democratic constitution for indicative
representation, of both the standard and the extended kind. The British
Columbia citizens’ assembly exemplifies a standard variety of indica-
tive representation that any democracy could usefully institute. And the
extended variety of indicative representation is bound to have a place
insofar as every plausible democracy requires some statutory officers and
bodies, such as judges, ombudsmen, and electoral commissioners, as well
as a possibility of contestation by public attorneys general and by other
informal invigilators of government.
Equally clearly, there has to be room in a democratic constitution for
responsive as well as indicative representation. In order for many reliably
indicative representers to be appointed to office, as statutory officers will
have to be appointed, there must already be elected and presumptively
responsive representers in place. But in any case there are quite inde-
pendent reasons why no plausibly democratic regime could operate on
the basis of indicative representation alone. Such representation requires
that people have a standing interest in decisions being made after a more
or less determinate pattern, as in requiring judges to adjudicate cases
according to established law. And in many cases there would be no such
interest available to guide indicative authorities. Thus popular repre-
sentation is bound to require a role for responsive as well as indicative
representation.
Democracy has to make room, then, both for proxies and for deputies.
This is not the place to investigate the best way of networking such
different kinds of representers in a satisfactory democratic dispensation,
nor the best way of frameworking their activities by suitable constitutional
and other constraints. Some of the relevant issues will be highlighted
78 Philip Pettit

by the discussion in the appendix, which constitutes the final section

of the chapter, on the operation of the Washington and Westminster
systems. But in the remainder of this section I discuss one serious problem
that may be raised about the democratic control of those responsive
representers, those deputies, that operate in an interpretive rather than a
directed mode. This is a telling issue since the argument at the end of the
last section suggests that all public deputies will have to be interpretive
trustees rather than directed delegates.

A problem
The medievals who spoke of the representation of a people – as I take
it, their interpretive representation – generally appear to have had an
intuitively democratic form of representation in mind, at least relative
to a citizenry of mainstream, propertied males. They looked to a form
of representation in which the individual people not only consented to
being represented but exercised an influence over what representers said
and did as a corporate spokesperson. They took it for granted that the
council of a city-republic would be appointed, at least in good part, by
voting and rotation – under the regimen ad populum – so that the city or
people could be described, in Bartolus’s words, as a sibi princeps, a prince
unto itself (Woolf 1913: 155–60, 180).
Hobbes went beyond this in arguing that the interpretive representation
whereby a people assumes the unity of a person might be practiced by a
monarch with absolute powers or by a committee of aristocrats: and this
with the authorization of members of the people. He did not think that
there could be democracy with representation by others, arguing that
democracy would require the people to self-represent, ruling themselves
in a committee-of-the-whole; this theme was later taken up by Rousseau
(1973). Mistakenly, in view of the discursive dilemma, he thought that
this self-representation could be achieved under a regime of majority
voting (Hobbes 1994: 16.15–17), as indeed did Locke (1960: Bk 2,
ch. 8.96) and Rousseau (1973: Bk 4, ch. 2).
The issue as to whether interpretive representation can be democra-
tized is important, given that interpretation is going to be unavoidable
when a multitude is responsively represented. We took responsive repre-
sentation, being a relationship of tracking the representees, to require the
control of those representees. The question here is whether such control
is available with interpretive representation as well as with responsive.
How might interpreters – authorized interpreters, as we can assume – be
subjected to the control of representees?
 Varieties of public representation 79

Participation in the committee-of-the whole, as envisaged in Hobbes’s

vision, is not a feasible way of imposing democratic control. Contem-
porary societies are just too big to allow for a participatory regime,
in particular the sort that would require members to adjust in face of
majority support for inconsistent positions. And neither is an electoral
regime going to provide an effective democratic discipline. Elected repre-
sentatives cannot be responsively controlled by the majority attitudes of
the population, since those attitudes may be incoherent. And if the only
restriction on those members springs from the desire to be reelected,
or from the fact of having to live under the laws proposed, still it will
leave them free to construct the mind of the people in a relatively uncon-
strained fashion. So is there any alternative or supplementary discipline
that might be imposed on interpretive representers?

A solution?
At the time of the civil rights disputes in the United States, as at many
other junctures in American history, the protagonists made rival claims
as to what was the way of thinking about racial relations that represented
the mind of the American people. Was it the much-vaunted heritage of
a state like Mississippi in which strict segregation had been enforced on
public transport, in public schools, and in other public amenities? Or was
it the message of equality and respect that had long been enshrined in
the Constitution and its amendments? The division on this issue created
a fault line that ran through the legislatures, the courts, the media, and
out onto the streets. Ultimately the constitutionally supported message
won out and became a theme around which citizens generally rallied. But
did this victory constitute a democratic breakthrough: a win for the right
interpretation of the people’s mind over the wrong interpretation? Or was
it a victory of what was morally right over what was morally wrong? Or
was it just a victory for the stronger over the weaker, the more numerous
over the less numerous?
I think that there is only one base on which the line taken in a case like
this can be justified as the right rather than the wrong interpretation of the
people’s mind. This is the base provided by considerations in the family
of what Rawls (1993, 1999) describes as public reasons. My own view
is that the victory in the civil rights disputes can be cast as a democratic
victory, so far as the line that emerged triumphant is the only line that
had the support of such public reasons.
Central to the idea of a democratic polity is the assumption that citizens
can debate with one another about what government and the state should
80 Philip Pettit

do from positions of relative equality: positions in which they can see

public action as action that they together license and support (Larmore
2003). So at any rate I assume. Given this assumption, it follows that
any proposal as to what government is to do should be supported by the
sorts of considerations that all can equally recognize as relevant, even if
they weight them differently. If one group can argue for a certain policy
only on the ground that it is good for its own members, or in line with
their particular view of the world, then that cannot be expected to pass
muster under the discipline of public debate – not, at least, unless the
policy can also be shown to benefit other groups or not to do them any
harm (Elster 1986a).
The existence of a political society in which members address one
another in arguing for this or that public policy ought, under the assump-
tion of equality, to generate a currency of public reasons. These will be
reasons that get established in the course of debate and argument as con-
siderations that are appropriately invoked in assessing public measures
and initiatives. If the members of the society cannot find a fund of con-
siderations on which to draw in argument about public policy – a fund
of considerations that all will take to be relevant, even while they weight
them differently – then the exchanges between them will degenerate into
power struggles and they will live in the precincts of civil war.
The assumption of equal positioning in public debate makes it
inevitable that certain core considerations are endorsed in any democratic
society: in particular, the claim of each member to equal respect and con-
cern. But there are many considerations that may receive endorsement
in one society and not in another. One society might endorse considera-
tions that support a system of more or less exclusively private ownership,
for example, another considerations that support a system that allows
for considerable public property. One society might foster considerations
that promote the development of certain cultural norms, another promul-
gate considerations that provide support for quite different conventions.
And so on.
The considerations acknowledged and valorized in any society will
typically leave issues of detailed policy underdetermined, of course; what
they rule out will be the policies that are unthinkable, certainly inde-
fensible, in the discourse of the public sphere. But even when there are
many candidate policies for dealing with a given issue, all of which pass
muster in public debate, the public reasons acknowledged in that debate
may support one or another procedure for resolving the tie. The sup-
ported procedure might involve a parliamentary vote, or reference to an
expert committee, or resort to referendum, or even the use of a citizens’
assembly.
 Varieties of public representation 81

As a democracy grows, then, we may expect that the trial and error
process whereby participants float considerations and find that they are
accepted or rejected will generate a currency of considerations that all
valorize, and valorize as a matter of common access. Those considerations
will provide the blocks out of which any case for changing or conserving
the way things are done will have to be built; they will indicate the
sorts of presumptions or premises on the basis of which arguments in
public policy are to be made. This observation is at the heart of Rawls’s
conception of public reasons, although he spells it out only occasionally.
In one version, he says, “the political culture of a democratic society that
has worked reasonably well over a considerable period of time normally
contains, at least implicitly, certain fundamental ideas from which it
is possible to work up a political conception of justice suitable for a
constitutional regime” (Rawls 2001: 34–5).
Returning to our original question, then, how can one argue that the
interpretations of the people’s mind that are selected and given authority
in public life really are the correct interpretations: the interpretations
that are controlled in an appropriate sense by the representees? The only
available base for arguing this in a large-scale democracy has to be that
they are the interpretations that emerge in a process where public reasons
are the primary selectional force. The interpretations may be uniquely
consistent with considerations that are given the status of public reasons.
But more likely they will be that particular set of interpretations, among
the sets that are compatible with public reason, that are selected under
procedures endorsed in public reason.
Go back, then, to the civil rights case. The argument of public fig-
ures in Mississippi and other southern states may have been in line
with some local traditions. But it seems clear that in the opinion of
the courts, the media, and the vast majority of US citizens those tra-
ditions conflicted with the requirements of the idea of equality as that
figured and still figures in American law and culture. It is doubtful if
the civil rights movement would have won the day had there not been a
widespread acceptance that practices of segregation were not supported
in the currency of public reasons that were accepted in the country. The
inconsistency with public reasons may not have been enough on its own
to bring about the changes that occurred in the 1950s and 1960s but,
arguably, it was an essential prerequisite for the success of the civil rights
movement.
It is time to sum up the overall argument of the chapter. We have seen
that representers and representees may be individuals or groups and, if
groups, that they may be mere collections, mere cooperatives, or full-scale
corporate entities. We have also seen that the relations of representation
82 Philip Pettit

come in two broad types, with representers figuring as indicative proxies

under the first variety, responsive deputies in the other. And, focusing
on issues of politics, we have argued that both sorts of representation are
bound to have a place in a democratic regime; that responsive represen-
tation is likely to require interpretive trustees, not directed delegates; and
that interpretive, responsive representation, however constructive, can be
democratically controlled under a discipline of public reason.
But are these ideas and distinctions of any use in considering empirical
systems? In order to vindicate their claim to significance I conclude with
observations on two ideal types of system, one associated with Washing-
ton, the other with Westminster. The distinctions we have made enable us
to mark a nice difference in representational priority between these two
models: inevitably, these two rather toytown models. And that difference
clearly matters.

V. Appendix: The Washington and Westminster systems9

Philip Pettit and Rory Pettit

The Washington model

The Washington system employs individual elected representers at a
number of sites. Broadly speaking, members of the House represent
districts, members of the Senate represent states, and the President rep-
resents the people as a whole. The representation at this level is responsive
in character, with members of the House being apparently committed to
the service of their districts, members of the Senate to the service of their
state, and the President to the service of the nation or union.
Representation is practiced in this system, however, not just at the
level of individuals, but also at the level where members of Congress
cooperate with one another and with the President and other members of
the administration. It is this group, Congress-cum-administration, that
ultimately gets various measures into law, performing a representative
function that parallels the representative work of the individuals in that
group. Where individuals may serve their different constituencies in a
responsive way, this group will be expected to serve the people as a whole
in a responsive manner. What it promulgates as law and policy is meant
to count, under suitable controls, as the law and policy supported by the
people.

9 This builds in some part on R. Pettit (2007).

 Varieties of public representation 83

Given the complexity of views in any district or state, the individual

representation of constituency is bound to be fundamentally interpre-
tive. The member of the House or the Senator will have to construct
the mind of the constituency he or she serves, operating as a trustee
rather than a delegate. Since the views of the people as a whole are
even more complex, the representation of the people by the Congress-
cum-administration is also bound to be interpretive. But how can the
Congress-cum-administration serve as an interpretive representer of the
people – a reliable trustee – if it is composed of individuals who often
have conflicting briefs? They have to serve as trustees both for their con-
stituency and for the nation or people as a whole.
Congress-cum-administration does not constitute a corporate entity
with a coherent set of goals and judgments. If it did, then that entity might
be a reliable system-wide trustee, imposing constraints on its members to
support suitable policies. Congress-cum-administration operates, rather,
as a loose aggregation that provides a majority, now for this compromise
and coalition, now for that. It has to get enough members on side in
order to create the requisite support for any bill that passes into law.
But the majority it musters in support of one bill may be different from
the majority it musters in support of another. And in each case it will
have to manufacture the majority by buying off various members with
favors that will play well with their electors back home. In operating on
this pattern, the Congress-cum-administration will be subject to various
constitutional constraints, but it is not clear that these provide anything
but the lightest level of regulative control.
In this system there is an inherent tension between two sorts of forces.
There are pressures on individual members of Congress to represent their
different constituencies responsively, on pain on not being reelected. And
at the same time there is pressure at the group or system level to generate
and maintain a body of law that has the coherence required, and the
fit with system-wide public reasons, to pass as the voice of the people
(Dworkin 1986).
These forces are quite likely to interact so as to produce laws that fail
to fit with one another, and perhaps even with the Constitution. The laws
themselves will often involve subclauses designed to pick up the support
of swing voters, and such complexity can make the demands of coherence
hard to identify. But even when those demands are obvious, the desire to
strike a popular, electorally helpful posture is liable to generate support
among members of Congress for laws that fail to fit with one another. In
that case the buck passes to the President, who may veto the law. And,
failing a presidential veto, the buck will pass to those members of the
public who may combine to challenge the law and to the courts that will
84 Philip Pettit

adjudicate that challenge. If there is a coherent body of law emerging at

the end of the process, that will be a product of this interplay of different
factors, not the work of any single corporate body.
This result can be summed up by saying that the Washington system
prioritizes the individual-level representation of constituencies at a pos-
sible cost to the group-level or system-level representation of the people
as a whole. It allows members of Congress to serve as trustees for their
constituencies, thereby raising a problem for how the administration-
cum-Congress can serve as a trustee for the people as a whole. It is
directed at ensuring not so much a representative government as a gov-
ernment of representatives – something that resembles what Edmund
Burke (1999) described as a congress of ambassadors.

The Westminster model

The reason why the Washington system of representation prioritizes
individual-level representation goes back to the fact that the executive
or administration is elected directly, not by members of the Congress.
This separation means that the fate of the administration does not depend
on how the members vote. And so those members are free to vote as they
wish, with party constraints serving only as a relatively light discipline, at
least under normal conditions. This marks the crucial difference between
the Washington and the Westminster regimes.
In the Westminster system the administration stands or falls with the
support of the individual members of parliament, in particular the lower
House. And that means that those parliamentary representatives who
have succeeded in electing an executive – the members of the winning
party or coalition – will have to stick together in order to maintain that
executive in power. In effect, the party or coalition will have to be orga-
nized so that its members are forced to cleave to an agreed party line, on
pain of losing party membership and the chance of being reelected. The
party or coalition will become a self-policing corporate entity.
Given this pressure, the behavior of parliamentary representatives can-
not be dictated by a desire to please those back in the constituency,
and to promote the prospects of reelection; they cannot serve as trustee
representatives for the constituency. If their party gets enough members
into parliament to be able to select the executive, then members will be
bound by the party whip to stick on most important matters to the party
line. And if their party fails to do this, then they will be bound by the
party whip to present a solid face of opposition to the party in power,
playing by the same hard-ball rules. This necessity is manifest to voters,
of course, so that legislators will be elected in great part on the grounds of
 Varieties of public representation 85

which executive they promise to support in office – usually, an executive

composed of members of their own party – not on the grounds of how
well they are likely to serve the local constituency.
In the Westminster system, then, the individual-level representation
of constituency gets put in a decidedly second place, since individual
members will tend to vote as their party votes and will be expected even
by those who elect them to vote that way. But if that is the weakness of
the system, the strength is that the parliament will operate very efficiently
to generate a body of legislation that can be expected to be internally
coherent, to cohere with established law and principle, and to be generally
responsive to the public reasons that carry weight among the people as a
whole. The legislative program for any parliament will have to be planned
by a corporate body: the party in power. This party will have had to ensure
the coherence of its legislative program. It will have had to advertise the
program as that which it would implement in office. And once in office it
will have to stick to the program, at least in general outline. Short of being
a minority or coalition government, it will have no excuse for departing
from it and any departure will be subject to a powerful challenge from
the opposition, the media, and the public.
We saw that the Washington system prioritizes the individual-level rep-
resentation of constituency at a cost to the system-level representation of
the people. These comments on the Westminster system show that things
there are almost exactly the other way around. The system prioritizes the
group-level representation of the people, giving parliament the cast of a
corporate body, albeit a body controlled by members of the party that
holds office. And it does this to the detriment of the individual-level rep-
resentation of constituency. In contrast to the Washington model, it is
directed at ensuring a representative government rather than a govern-
ment of representatives.

Significance
The difference of representational priority has enormous implications for
other aspects of the two systems, including implications that bear on the
claims of each to be a satisfactory democratic regime.10 The significance
of the difference in representational priority comes out in differences
of sensitivity to a variety of pressures. We can illustrate the point, some-
what speculatively, with reference to local pressures, expressive pressures,
lobby pressures and the pressure of public opinion.

10 Other assessments of the two systems tend not to focus on this issue of representational
priority: see Shugart and Carey (1992); Linz and Valenzuela (1994).
86 Philip Pettit

Local pressures Under the Washington system members of

Congress are bound to be influenced, for good or ill, by a concern for how
their votes will play back home. This localism may occasionally bring real
benefits to their district, as when members can secure legislative favors.
The members of a Westminster parliament will have little occasion to
think locally in this manner, since their votes will be controlled by the
party; local efforts will be restricted to providing some local advisory
services and to playing the part of a celebrity in local events.

Expressive pressures To act in a purely instrumental way is to act

out of a concern for the outcome of what one does; to act purely expres-
sively is to act out of a concern for the posture associated with acting in
that manner (Brennan and Lomasky 1992). The members of Congress
will often be in a situation where they cannot affect the outcome, even as
part of an effective majority, and where relevant they will be motivated to
take positions on an expressive basis. They will be tempted to grandstand
on issues like prostitution and drugs and crime, for example, focusing
on the symbolic rather than the substantive utility of doing so (Nozick
1994). The members of a Westminster parliament will be less exposed to
this pressure. The party in power will always be able to affect outcomes,
with the result that concern for outcome should balance if not eliminate
the concern for posture and thereby reduce expressive motives. Some-
thing similar may be true of the party out of power, for it will always have
to think about what to do if it wins office.

Lobby pressures The members of a Washington Congress, being

more or less free to vote as they will, are going to be subject to intense
lobby pressure for their vote, especially when the lobbies involved can
provide campaign finance for those who vote congenially. The members
of a Westminster legislature, not being free to vote against their party,
will not be subject to the same lobbying pressure, at least not outside the
ministry. Lobbies will have to buy over whole parties in order to have a
legislative impact, not just the swing voters on any issue.11

Public opinion The members of Congress will not necessarily

be responsive to a high level of public opinion in favor of a particular
measure. Whether a response is forthcoming will depend, first, on how far

11 The very fact that the party in power puts up a unified legislative program for the
duration of a parliament may facilitate the influence of an opinion-forming elite in the
process whereby that program is formed. And that particular danger may be relatively
absent under the Washington system.
 Varieties of public representation 87

that opinion generates local pressure on the representer to do something;

second, on how expressively beneficial or costly will be the posture of
support; third, on how far lobby pressures are silent or supportive on the
issue; and, fourth, on how far the transaction costs of getting a majority
together in support of the measure are manageable. But if public opinion
strongly supports a given measure, then the party in office under the
Westminster system can usually be expected to back the measure. It will
not be subject to the same local, expressive and lobbying concerns; there
will be low transaction costs involved, since it will already be organized as
a corporate entity; and, as an effective agent in power, it will be expected
to respond to public opinion and will run an electoral risk if it fails to
do so. The difference between the two models may help to explain why
the public support for gun control that generally emerges in the wake
of gun outrages routinely elicits a response in Westminster systems and
routinely fails to do so under the Washington model.
If these thoughts are on the right lines, then the representational pri-
ority adopted in a system of government is going to have an enormous
influence on how the system and the society as a whole works. Let the
system go toward Washington and it will activate local, expressive, and
lobby pressures and reduce the impact of aggregate public opinion. Let
it go toward Westminster and it will reduce local, expressive, and lobby
pressures and intensify the impact of public opinion.
These predictions should be qualified for variation in other factors,
of course, and need to be tested against empirical observations. But
assuming they are broadly on the right lines – and there is some evidence
that they are (Foweraker and Landman 2002) – they emphasize the
importance of the issue of representational priority. There are various
normative standpoints possible on that issue, of course, so that some
thinkers may favor the Washington model, some the Westminster. But
no one is likely to think that the issue is insignificant. It demonstrates, if
demonstration is needed, that the theory of representation matters.

BIBLIOGRAPHY
Abramson, J. 1994. We, the Jury: The Jury System and the Ideal of Democracy. New
York: Basic Books.
Bratman, M. 1999. Faces of Intention: Selected Essays on Intention and Agency.
Cambridge: Cambridge University Press.
Brennan, G. and L. Lomasky. 1992. Democracy and Decision.
Cambridge: Cambridge University Press.
Burke, E. 1999. “Speech to the Electors of Bristol,” in Selected Works of Edmund
Burke. Indianapolis, IN: Liberty Fund.
88 Philip Pettit

Canning, J. P. 1983. “Ideas of the State in Thirteenth and Fourteenth Cen-

tury Commentators on the Roman Law.” Transactions of the Royal Historical
Society 33: 1–27.
1987. The Political Thought of Baldus de Ubaldis. Cambridge: Cambridge Uni-
versity Press.
Christiano, T. 1996. The Rule of the Many: Fundamental Issues in Democratic
Theory. Boulder, CO: Westview Press.
Dahl, R. 1957. “The Concept of Power.” Behavioral Science 2: 201–15.
Dworkin, R. 1986. Law’s Empire. Cambridge, MA: Harvard University Press.
Elster, J. 1986a. “The Market and the Forum: Three Varieties of Political The-
ory,” in Foundations of Social Choice Theory, ed. J. Elster and A. Hilland.
Cambridge: Cambridge University Press, 103–30.
Eschmann, T. 1946. “Studies on the Notion of Society in St Thomas Aquinas:
St Thomas and the Decretal of Innocent IV Romana Ecclesia, Ceterum.”
Medieval Studies 8: 1–42.
Ferejohn, J. 2007. “The Citizens’ Assembly Model,” in Designing Deliberative
Democracy: The British Columbia Citizens’ Assembly, ed. H. Perse and M.
Warren. Cambridge: Cambridge University Press, 192–213.
Fishkin, J. S. 1997. The Voice of the People: Public Opinion and Democracy. New
Haven, CT: Yale University Press.
Foweraker, J. and T. Landman. 2002. “Constitutional Design and Democratic
Performance.” Democratization 9: 43–66.
Fung, A. and E. Fagotto. 2006. “The British Columbia Citizens’ Assembly.”
Kennedy School of Government, Harvard University. Cambridge, MA.
Goodman, N. 1969. Languages of Art. London: Oxford University Press.
Hansen, M. H. 1991. The Athenian Democracy in the Age of Demosthenes. Oxford:
Blackwell.
Hobbes, T. 1990. Behemoth or The Long Parliament, ed. F. Toennies. Chicago:
University of Chicago Press.
1994. Leviathan, ed. E. Curley. Indianapolis, IN: Hackett.
Ketcham, R. (ed.). 2003. The Anti-Federalist Papers. New York: Signet Classic.
Knights, M. 2005. Representation and Misrepresentation in Later Stuart Britain:
Partisanship and Political Culture. Oxford: Oxford University Press.
Kornhauser, L. A. and L. G. Sager. 1993. “The One and the Many: Adjudication
in Collegial Courts.” California Law Review 81: 1–59.
Larmore, C. 2003. “Public Reason,” in The Cambridge Companion to Rawls, ed.
S. Freeman. Cambridge: Cambridge University Press, 368–93.
Linz, J. J. and A. Valenzuela (eds.). 1994. The Failure of Presidential Democracy.
Baltimore, MD: Johns Hopkins University Press.
List, C. and P. Pettit. 2002. “Aggregating Sets of Judgments: An Impossibility
Result.” Economics and Philosophy 18: 89–110.
Locke, J. 1960. Two Treatises of Government. Cambridge: Cambridge University
Press.
Lovett, F. 2007. “Power,” in A Companion to Contemporary Political Philosophy,
ed. R. E. Goodin, P. Pettit, and T. Pogge. Oxford: Blackwell.
McLean, J. 2004. “Government to State: Globalization, Regulation, and Govern-
ments as Legal Persons.” Indiana Journal of Global Legal Studies 10: 173–97.
 Varieties of public representation 89

Manin, B. 1997. The Principles of Representative Government. Cambridge:

Cambridge University Press.
Mansbridge, J. 2003. “Rethinking Representation.” American Political Science
Review 97: 515–28.
forthcoming, “A ‘Selection Model’ of Political Representation,” Journal of
Political Philosophy.
Mill, J. S. 1964. Considerations on Representative Government. London: Everyman
Books.
Nozick, R. 1994. The Nature of Rationality. Princeton, NJ: Princeton University
Press.
Pettit, P. 2001. A Theory of Freedom: From the Psychology to the Politics of Agency.
Cambridge: Polity Press; New York: Oxford University Press.
2004. “Depoliticizing Democracy.” Ratio Juris 17: 52–65.
2007. “Republican Liberty: Three Axioms, Four Theorems,” in Republicanism
and Political Theory, ed. C. Laborde and J. Manor. Oxford: Blackwell, 102–
30.
Pettit, P. and D. Schweikard. 2006. “Joint Action and Group Agency.” Philosophy
of the Social Sciences 36: 18–39.
Pettit, R. 2007. “Reconceptualising Democratic Representation,” BA honors
thesis, Department of Philosophy, University of Sydney.
Phillips, A. 1995. The Politics of Presence. Oxford: Oxford University Press.
Pitkin, H. F. (ed.). 1969. Representation. New York: Atherton Press.
1972. The Concept of Representation. Berkeley: University of California Press.
Rabkin, J. A. 1998. “The Secret Life of the Private Attorney General.” Law and
Contemporary Problems 61: 179–203.
Rawls, J. 1993. Political Liberalism. New York: Columbia University Press.
1999. The Law of Peoples. Cambridge, MA: Harvard University Press.
2001. Justice as Fairness: A Restatement. Cambridge, MA: Harvard University
Press.
Rehfeld, A. 2005. The Concept of Constituency: Political Representation, Democratic
Legitimacy, and Institutional Design. Cambridge: Cambridge University Press.
2006. “Towards a General Theory of Political Representation.” Journal of
Politics 68: 1–21.
Richardson, H. 2002. Democratic Autonomy. New York: Oxford University Press.
Rousseau, J.-J. 1973. The Social Contract and Discourses. London: Dent.
Shugart, M. S. and J. M. Carey. 1992. Presidents and Assemblies: Constitutional
Design and Electoral Dynamics. Cambridge: Cambridge University Press.
Skinner, Q. 2005. “Hobbes on Representation.” European Journal of Philosophy
13: 155–84.
Stewart, J., E. Kendall, and A. Coote. 1994. Citizens’ Juries. London: Institute
of Public Policy Research.
Urbinati, N. 2000. “Representation as Advocacy: A Study of Democratic Delib-
eration.” Political Theory 28: 758–86.
Waley, D. 1988. The Italian City-Republics, 3rd edn. London: Longman.
Woolf, C. N. S. 1913. Bartolus of Sassoferrato. Cambridge: Cambridge University
Press.
4 Representative government and
popular sovereignty

Bryan Garsten

Are representative governments working well? The answer to that ques-

tion depends on what we think the purpose of representative government
is. Most research in political science presumes that the purpose of repre-
sentative government is to represent the will of the people by translating
popular sentiment or public interest into governmental policy. It there-
fore presumes that a good measure of the performance of representative
democracy, at least in its representative capacity, involves comparing
policy results with public opinion as it is or as it should be. The clas-
sic study of constituency influence in the House of Representatives by
Miller and Stokes, for example, focused on “the extent of policy agree-
ment between legislator and district” (Miller and Stokes 1963). More
recent work continues to investigate similar relations: Page and Shapiro
look for “congruence between changes in policy and changes in opinion”
and assume that “normative concepts of democracy” would mandate
something close to “direct democracy” (Page and Shapiro 1983). Stim-
son, Mackuen, and Erikson ask “whether the national system is efficient
in turning popular sentiment into policy” (Stimson et al. 1995). These
studies, and many more like them, presume a principle close to the one
that Bartels articulates clearly: “The appeal of representative democracy
hinges on the responsiveness of elected politicians to the preferences and
interests of their constituents” (Bartels 1991). Occasionally the notion
of responsiveness is examined in more depth (Manin 1997; Manin et al.
1999), but most of the political science literature simply presumes that
the purpose of representative government is to be an instrument of the
popular will.

The author would like to thank the following people for their comments and suggestions on
previous versions of this chapter: Costin Alamariu, Onur Bakiner, John Ferejohn, Hélène
Landemore, Karuna Mantena, David Mayhew, Paulina Ochoa Espejo, Pasquale Pasquino,
Melissa Schwartzberg, Annie Stilz, Susan Stokes, Elisabeth Wood, the members of the
Brown University Political Philosophy Colloquium, and the participants at the conference
on “Representation and Popular Rule” at Yale University, fall 2006.

90
 Representative government and popular sovereignty 91

In this chapter I do not deny the importance of responsive government;

Robert Dahl is surely right to suggest that the notion of responsiveness is
central to any understanding of modern democracy (Dahl 1971). But the
relationship between representative government and popular sovereignty
is more complicated than an emphasis on responsiveness alone would
suggest. In this chapter I want to indicate the nature of the problem and
put forward a view of the purpose of representative government that is
quite different from one that prioritizes responsiveness, a view drawn
largely from two of the early theorists of representative government in
France and the United States, Benjamin Constant and James Madison.
At the end of the chapter I will briefly suggest the sorts of empirical
research questions that might be investigated in light of this broader view
of representative government.
The broader view that emerges from considering these thinkers is, in
summary, this: a chief purpose of representative government is to multiply
and challenge governmental claims to represent the people. This goal is quite
different from asking the government as a whole to represent the popular
will as it can be found through any particular vote or poll. It is also differ-
ent from the goal of asking it to represent the popular will as it could be
imagined to emerge from a process of deliberation or from an indepen-
dent analysis of the public interest. In asking instead for a government to
multiply and challenge representative claims, representative government
aims (on this view) to provoke debate about precisely what the popu-
lar will is and thereby to prevent any one interpretation of the popular
will from claiming final authority. It aims to foster and institutionalize
popular impatience with our rulers, to both fuel and channel popular
grievances against those in power. For conceptual reasons to be men-
tioned later, it is always possible to claim that a government is not fully or
adequately representative. Representative government attempts to cap-
italize on this fact. By trying to insure that there are always competing
claims to represent the people present in government at the same time, it
makes any particular claim to fully represent the people implausible, and
it helps to combat the use of such claims to justify the concentration of
power. Counterintuitive as it sounds, a fundamental purpose of represen-
tative government, as Constant and Madison saw it, is to oppose popular
sovereignty in the sense that it is usually understood – to undermine the
idea that government can adequately represent the people.
To avoid misunderstanding, let me make one point clear at the outset:
the suspicion of government efforts to turn popular sovereignty into pol-
icy that lies at the heart of this conception of representative government
does not rest on a belief that the public is incompetent, unintelligent, or
lacking in virtue. It does not even rest on the idea that public opinion
92 Bryan Garsten

must be, in Madison’s words, “refined and enlarged”; it does not begin
from a sense that public sentiment must be filtered through the minds
of representatives or through a deliberative process before being allowed
access to political power. There may be some validity to these ideas.
Perhaps representatives sometimes are, by virtue of their competence
or their deliberations or the influence of their office, better decision-
makers than the people themselves. But this aristocratic point is not at
all the consideration that lies behind the argument for representative
government that I want to examine here. Too often, it is thought that
such doubts about the capacities of ordinary citizens are the only rea-
sons that one might have (aside from the practical impediments to gath-
ering large groups) for preferring representative government to direct
democracy. The aim of this chapter is to suggest that there are other
reasons, more fundamentally democratic reasons, for creating a form
of government that always remains at some distance from public opin-
ion, and whose claim to represent that opinion can never be made fully
convincing.
The aristocratic argument against popular sovereignty and direct
democracy is often traced to Edmund Burke, whose suspicion of popular
politics can be seen in his famous skepticism about the French Revolution
and whose statement to his own constituents has become the emblematic
articulation of the view that representatives should use their own judg-
ment rather than merely echoing the opinions of their constituents (Burke
2000, 2003). The view to which I want to draw attention is not drawn
from Burke. It is instead drawn from Burke’s enemies and critics – from
Rousseau, whose writings helped to inspire the French Revolution, and
from Benjamin Constant, who defended that Revolution against Burke’s
attack. Constant is often portrayed as a critic of Rousseau’s, since he
opposed the notion of popular sovereignty associated with Rousseau.
And since Constant was one of the first writers in any language to use
the word “liberal” in politics, and one of the first to outline a full account
of liberal, representative government, his relation to Rousseau is viewed
as emblematic of liberalism’s relation to democracy: liberalism, because
of its support for representative rather than direct democracy, is often
viewed as fundamentally undemocratic. The argument that I want to
suggest here draws the democratic Rousseau and the liberal Constant
much closer to one another, and so suggests that the phrase we use
to describe our own form of government, “liberal democracy,” is not
in fact an oxymoron. It is true that liberals such as Constant are sus-
picious of efforts to institutionalize popular sovereignty directly. But it
turns out that on this point Constant was merely following Rousseau.
If we can understand Rousseau’s reason for harboring this suspicion,
 Representative government and popular sovereignty 93

we will come closer to seeing the democratic argument for this liberal
position.

I. Rousseau on popular sovereignty

None of the canonical political theorists defended the sovereignty of the
people more insistently than Rousseau did. With the concept of the gen-
eral will he sought to capture the intuition that all citizens had to be
regarded as equal, that no one could impose his or her views on the rest,
and that the only legitimate source of authority in politics was the vote
of the people as a whole, where every individual had an equally weighted
vote. He thought that people’s authority could not be transferred or alien-
ated, even if the people wished it; nor could it be represented. Institutions
that claimed to hold the people’s sovereignty for them were, he wrote,
an inheritance from the feudal past and signs of political corruption. In
a healthy polity citizens would fly to the assemblies themselves rather
than allowing representatives to do the work of sovereignty for them
(Rousseau 1997). On the basis of such statements, political theorists
have regarded Rousseau as an inspiration for participatory democracy,
and populist politicians since Robespierre have quoted from Rousseau’s
Social Contract to advance their cause.
But the truth is that Rousseau, in spite of his radically democratic
understanding of sovereignty, did not advocate any form of direct demo-
cratic government in the Social Contract. In fact he warned against democ-
racy: “If there were a people of Gods, it would govern itself democrat-
ically. Such a perfect government is not suited to men,” he wrote. And
further:

In the strict sense of the term, a genuine Democracy never has existed, and
never will exist. It is against the natural order that the greater number govern
and the smaller number be governed. It is unimaginable that the people remain
constantly assembled to attend to public affairs, and it is readily evident that
it could not establish commissions to do so without the form of administration
changing. (ibid.: 3.4)

This passage might make it sound as if Rousseau’s primary reason for

warning against direct democratic government was practical. Later in
the book, however, he explicitly argued against the view that in large
modern states it is impossible for the people to gather together and vote
on important questions. He pointed out that Rome, no small polity,
routinely asked citizens to vote directly on matters of importance, and
he insisted that such regular meetings of the citizenry could be held in
modern times as well (ibid.: 3.12). Whether he was right does not matter.
94 Bryan Garsten

The point is that his reason for opposing direct democratic government
was not simply that it was impracticable to gather all the citizens in a
large state together.
Rousseau’s deeper reason rested on a theoretical distinction that he
regarded as crucial, a distinction to which he worried his readers would
not pay enough attention. This was the distinction between legislation and
execution. The legislative power was analogous to the faculty of the will in
an individual person; it was the power to decide what to do, and it was the
power that the sovereign had the authority to wield. The executive power,
on the other hand, was analogous to the physical power of the individual;
it carried out the decisions of the sovereign will and applied them to
particular situations. This was the power that government wielded. The
key point about these two powers, according to Rousseau, was that they
had to be kept distinct from one another in a polity. The reason was that
the sovereign will had to remain general in order to retain its unique legit-
imacy. Only legislative acts that were general in scope could be viewed
as equally touching all citizens. And only the fact that the general will
touched all citizens equally made it compatible with the freedom of all
citizens. For if submitting to the general will meant submitting to the
rule of particular other people, it would seem to require giving up one’s
freedom. Democratic sovereignty gained its special claim to legitimacy
from the fact that it did not require this, the fact that, as Rousseau put
it, “each, by giving himself to all, gives himself to no one” (ibid.: 1.6). If
submitting to the sovereign general will was different from submitting to
a ruler, as Rousseau insisted it was – if it was more democratic and more
compatible with freedom – that was only because the sovereign could
not mandate anything that was not general enough to affect all citizens
equally. But acts of government were necessarily particular. They affected
different people in different ways, and they reflected particular officials’
judgments about particular issues. Therefore, as soon as a sovereign
power engaged in executive acts of government, it would lose its claim
to be “no one” and would become someone in particular, a ruler with
distinct goals and interests. This meant that the sovereign could not gov-
ern. Only by remaining separate from most acts of government could the
sovereign popular will maintain its general character and thus preserve its
unique democratic legitimacy. Usurpation was Rousseau’s name for the
confusion of sovereignty and government; it occurred whenever the gov-
ernment claimed the authority that rightfully belonged to the sovereign.
According to Rousseau, usurpation could be minimized and delayed, but
it was difficult, if not impossible, to avoid; with time, it would eventually
bring the downfall of all governments, leading to tyranny or despotism
(ibid.: 3.10).
 Representative government and popular sovereignty 95

The problem with direct democratic government was that it asked the
same people to act as both sovereign and government. In theory one
person could play both roles without mixing them, but in practice it would
be difficult to keep the two roles separate in one’s mind. The people,
while they were acting as a government, would tend to claim sovereign
authority for their actions. And we might think, why not? The people
themselves were, after all, the sovereign; if the people in their capacity as
governors were usurping the authority of the sovereign, they were only
usurping that authority from themselves. But Rousseau would insist that
we consider this case more carefully: the party doing the usurping is “the
people” acting in pursuit of particular interests or considerations. The
party whose authority is being usurped is “the people” acting according
to the general will. Rousseau’s argument for the absolute sovereignty of
the democratic people applies only to the latter version of “the people.” If
this sovereign authority is replaced by a people acting instead according to
particular interests – either private interests or the interests of government
as a particular body in society – then its legitimacy disappears. Thus, to
say that a direct democracy has usurped the sovereign authority of the
people is another way of saying that the people have become corrupted
by private concerns, that they are no longer exercising their will in a way
that is general enough to be democratically legitimate. In the chapter on
democracy Rousseau therefore articulated his reason for being suspicious
of direct democratic government in this way:

It is not good that he who makes the laws execute them, nor that the body
of the people turn its attention away from general considerations, to devote it
to particular objects. Nothing is more dangerous than the influence of private
interests on public affairs, and abuse of the laws by Government is a lesser
evil than the corruption of the Lawgiver [the sovereign], which is the inevitable
consequence of particular considerations. (ibid.: 3.4)

Of course we might disagree with Rousseau’s view that a popular will

is only legitimately sovereign if it has a general or “uncorrupted” char-
acter. Then we would face the formidable task of explaining why a mere
tally of votes should have any particular moral legitimacy; why accepting
a majority vote as legitimate is not simply an example of might mak-
ing right. Perhaps there are non-Rousseauian ways to make the case for
democratic sovereignty, but this is beyond the scope of this chapter. Here,
I assume that Rousseau’s argument for popular sovereignty is powerful,
and I ask why that line of reasoning does not also make the case for
direct democratic government. Rousseau’s reason was that usurpation
was particularly likely in democratic governments, and that usurpation
destroyed the conditions under which a people’s will could rightfully be
96 Bryan Garsten

called sovereign. Therefore, even though Rousseau claimed that “the leg-
islative power belongs to the people and can belong only to it,” he just as
firmly insisted that the executive power “cannot belong to the generality
[of the people] in its Legislative or Sovereign capacity” (ibid.: 3.1). His
theory of sovereignty was democratic but his theory of government was
not.
Direct democracies were not the only governments that Rousseau
thought were likely to fall into usurpation, but they were especially ill-
equipped to resist the impulse because they include no institutional sepa-
ration of the legislative and executive roles. In other forms of government,
the people acting as sovereign could check or slow the process of usurpa-
tion by expressing its judgment about the actions of the government.
The most important practical recommendation of Book 3 of the Social
Contract was that every form of government should be subject to periodic
referenda in which all citizens could vote on two questions: whether the
present form of government should continue, and whether the present
office-holders should continue in office (ibid.: 3.12–14, 18). It is clear
enough how this would work in an elective aristocracy: the people as a
whole would vote on whether the few of them who had governmental
offices should retain power. But in a direct democracy the people as a
whole (as sovereign) would have to render judgment on themselves, the
same people as a whole (as government). The problem is not that this is
conceptually impossible; Rousseau mentioned that the British House of
Commons sometimes transformed itself into a committee of the whole
to discuss a matter and then transformed back into itself to hear the
recommendations of the committee, so the same group of individuals
performed two distinct roles (ibid.: 3.17). But Rousseau did not think
such an arrangement was to be recommended if one wanted to prevent,
or at least slow, the corruption of the people and the usurpation of its
sovereign authority. The periodic referenda that he recommended to pre-
serve the sovereignty of the people would work best if the people voting
in them were not evaluating their own performance. Direct democratic
government required every citizen to be judge in his own case, dividing
himself into sovereign citizen and governing citizen. Rousseau suggested
that usurpation was more easily avoided or delayed if this division were
institutionalized in a distinction between the people and its government.
Strikingly, he suggested that the sovereign people’s distance from the par-
ticular work of governing was precisely what enabled it to protect its own
sovereignty.
Rousseau’s famous opposition to “representation” must be under-
stood in light of what has just been discussed. The relationship between
sovereign and government was not one of “representation,” as he most
 Representative government and popular sovereignty 97

often used the term. Government did not represent the people in the
sense of bearing their sovereign authority. Government officials held their
power by “nothing but a commission, an office in which they, as mere
officers of the Sovereign, exercise in its name the power it has vested in
them, and which it can limit, modify, and resume” (ibid.: 3.1). Sovereign
authority always remained with the people and could not be transferred
to government; any effort by the governors to claim sovereign authority
for themselves, by saying that they represented the people, was nothing
other than a form of usurpation.1 The key point – and the counterintu-
itive one – was that the distinction between sovereign and government,
and the distance that it recommended between the people and the gov-
ernment, was wholly consistent with (and was in fact a consequence
of) the impossibility of “representation” in Rousseau’s understanding of
the word. In his terms, efforts to “represent” the sovereign people were
efforts to usurp their authority.
Popular sovereignty in Rousseau thus functioned not only in a positive
sense, as a way of thinking about what basic law the people might actively
authorize and legislate, but also – more fundamentally – in a negative
sense (Pasquino 2007). The negative function of popular sovereignty is
to remind us that governing institutions and officers are not sovereign.
Popular sovereignty understood in this way offers an argument to use
against would-be usurpers. The statement that “the people” and only
“the people” is sovereign is, precisely because of its abstract generality,
a rebuttal of any claim that a particular official or assembly might make
to fully represent or embody the popular will. This negative function
of popular sovereignty is itself institutionalized in the regular referenda
through which the people can reject the government and government
officials. While Rousseau opposed “representation,” the recommenda-
tions that he made about government were not so different than those at
the heart of our practice of representative democracy: he suggested that
governmental functions should be delegated and that periodic popular
elections should be held to judge the performance of the delegates.

1 Rousseau did say that acts of the government should be presumed to be consonant with
the sovereign general will so long as there was an opportunity for the sovereign to voice
opposition and it did not do so (ibid.: 2.1). But if the people did make its sovereign
will known, no governmental authority had any standing at all to contest that will: “The
instant the People is legitimately assembled as a Sovereign body, all jurisdiction of the
Government ceases, the executive power is suspended, and the person of the last Citizen
is as sacred and inviolable as that of the first Magistrate; because where the Represented
is, there is no longer a Representative” (ibid.: 3.14). In this passage Rousseau slipped into
using the language of representation to describe the relation of government to sovereign,
but the general point is clear: sovereign authority always remains with the people and is
never taken over by their governors.
98 Bryan Garsten

The most significant differences between what Rousseau suggested

and the basic structure of our liberal democracies are (a) that our regular
elections ask only the second of the two questions that Rousseau thought
should be put to the sovereign people – we ask whether office-holders
should be changed, but not whether the offices themselves, the constitu-
tional framework of government, should be changed; (b) that Rousseau
did not envision the role of political parties; and (c) that Rousseau did
not recommend the elaborate institutional checks and balances that we
have, such as separation of powers and federalism. These differences are
important, and so I do not claim that Rousseau had a modern theory of
representative government. But he did prefer an indirect system of gov-
ernment to a more direct form of democracy, and he did so for democratic
reasons. His suspicion of governmental claims to represent the popular
will arose from his desire to protect the sovereignty of the people against
potential usurpers. Many of the institutional arrangements promoted by
theorists after the democratic revolutions in France and America can
plausibly be viewed, I think, as aiming to do the same thing.

II. Liberal democracy

Benjamin Constant, after having watched the course of the French Rev-
olution from its hopeful beginnings through the Terror and eventually to
Napoleon’s coup d’état, argued that the remedies for usurpation found in
Book 3 of the Social Contract had been wholly ineffective. In fact, Con-
stant thought, Rousseau’s writings had been used as a justification for
new and more pernicious forms of despotism by politicians who claimed
to act in the name of the people. Rousseau had meant to give sovereignty
to the people, but politicians had quickly found that they could arrogate
that sovereignty to themselves precisely by claiming to represent it, just
as Rousseau had feared they would.
Constant thus began his major work on liberal politics, Principles of Pol-
itics Applicable to All Representative Governments, with a chapter on popular
sovereignty in which he repudiated part – but only part – of Rousseau’s
thought. He insisted, against Rousseau, that there was no such thing as
absolute sovereignty, not even if it was placed in the hands of the people
themselves. “When you establish that the sovereignty of the people is
unlimited, you create and toss at random into human society a degree
of power which is too large in itself, and which is bound to constitute
an evil, in whatever hands it is placed . . . There are weights too heavy
for the hand of man,” he wrote (Constant 1988). To limit sovereignty
he asserted that there was “a part of human existence which by
 Representative government and popular sovereignty 99

necessity remains individual and independent, and which is, by right,

outside any social competence” (ibid.). These parts of human existence
were protected by rights, and no government that interfered with them
was legitimate, no matter on what grounds it did so. Thus Constant
moved directly from his criticism of Rousseau to the fundamental com-
mitments of liberalism and to the notion of limited government.
Beneath this disagreement with Rousseau about whether popular
sovereignty was absolute, however, lay a more fundamental agreement
with him. Constant did not dispute the idea that popular sovereignty
was the only rightful source of political authority. In fact, he made the
argument even more succinctly than Rousseau had, insisting that the
only alternative to popular sovereignty was force, which was clearly ille-
gitimate. What Constant feared was not the idea of giving sovereignty
to the people, but the practical implication that particular individuals or
assemblies would claim the right to exercise that sovereignty on behalf
of the people. He acknowledged that Rousseau had feared precisely the
same thing:

Rousseau himself was appalled by [the] consequences [of his theory]. Horror-
struck at the immense social power which he had thus created, he did not know
into whose hands to commit such monstrous force, and he could find no other
protection against the danger inseparable from such sovereignty, than an expedi-
ent which made its exercise impossible. He declared that sovereignty could not
be alienated, delegated or represented. This was equivalent to declaring, in other
words, that it could not be exercised. It meant in practice destroying the principle
which he had just proclaimed. (ibid.)

Constant’s difference with Rousseau was therefore more one of strat-

egy than one of ultimate intention. He thought that Rousseau’s distinc-
tion between sovereignty and government, and the related claim that
sovereignty could not be represented, were arguments that were too
abstract to be successful in preventing governors from trying to usurp
sovereign authority. Once someone introduced a justification of absolute
sovereignty into the political universe, no set of arguments, no matter
how sophisticated, would succeed in preventing political actors from try-
ing to use that justification to support their own power. Sovereignty and
government could not be kept distinct (Constant 2003). The abstract
idea of a sovereign people tended to become concrete in the form of
demagogues claiming to rule in the name of the people. Popular leaders
became aspiring tyrants. In the first version of Principles he described this
process in a striking passage. Whenever the government wants to grab
power, he wrote,
100 Bryan Garsten

it quotes the imprescribable prerogative of the whole society . . . The government

can do nothing, it says, but the nation can do everything. And soon the nation
speaks. By this I mean that a few men, either low types or madmen, or hirelings,
or men consumed with remorse, or terror-struck, set themselves up as its instru-
ments at the same time as they silence it, and proclaim its omnipotence at the
same time as they menace it. In this way, by an easy and swift maneuver, the
government seizes the real and terrible power previously regarded as the abstract
right of the whole society. (ibid.)

Constant and Rousseau were thus united in their worry about govern-
ment misuse of the language of popular sovereignty. Constant, of course,
had the advantage of hindsight, having seen not only Robespierre use this
language but also, later, Napoleon, whose particular form of despotism
legitimated itself through elections and democratic justifications. Con-
stant devoted another work exclusively to describing Napoleon’s demo-
cratically tinged despotism, and the second term in the title he chose for
that work, “Conquest and Usurpation,” echoed Rousseau’s language of
“usurpation” (Constant 1988).
The real difference between Rousseau and Constant lay in their views
of how best to combat the danger that usurpation posed. The philosoph-
ical part of Constant’s solution, as we have seen, was simply to deny that
sovereignty could ever be absolute, to insist that the individual should be
afforded rights with which no authority could rightfully interfere. This
part of his solution is difficult to say much about, because Constant was
not clear about the grounding of these rights. But there is also another
part of his solution, the part that he wrote much more about, which
focused on institutional design. Constant’s specific proposals, which var-
ied over time, demonstrate an inability to settle on one best means of
carrying out his ideas. Early in his career he seems to have been more
republican, while later he embraced the idea of a constitutional monar-
chy. But what lay beneath all of his different institutional proposals was
a fundamental and unchanging desire to find ways of institutionalizing
resistance to centralizing and usurping authority. This is true even of his
later arguments on behalf of a constitutional monarchy. He justified a
monarch by arguing that it could be made into a “neutral” power that
could check and keep in place the various “active” powers of govern-
ment. When he defended the monarch’s power to dissolve representative
assemblies, he claimed that it was a means of checking the assembly.
Why did the assembly need checking? It is important not to misinter-
pret his point here (Holmes 1984). His worry about the assembly was
not that the people itself would act tyrannically through it. Unlike later
liberals such as Alexis de Tocqueville and John Stuart Mill, he was not
preoccupied with the danger that the people themselves would become
 Representative government and popular sovereignty 101

a tyrannical majority. Instead he suggested that it was the representatives

who often became dangerous once they were separated from the people:
“An assembly, the power of which is unlimited, is more dangerous than
the people” (Constant 1988). When he supported a monarch’s right
to dissolve an assembly, he viewed this monarchical right as a way of
defending the people against their own representatives: “The dissolution
of assemblies is by no means, as some have argued, an insult to the rights
of the people” (ibid.). On his plan, the monarch could not replace the
representatives’ policy with his own will, but could only send the matter
back for consideration by the next group of representatives to be elected.
In exercising this power, the monarch was contesting the claim of the
current representatives to be adequately representing the popular will. A
very similar argument had been used to justify the monarch’s “suspen-
sive” veto during the Revolution. Even Constant’s support for constitu-
tional monarchy, then, has a democratic justification: it was part of an
institutional plan to insure that no governmental power could advance a
claim to act in the name of the people without being subject to challenge
from another governmental power capable of making the same claim.
If usurpation could not be avoided, the best that could be done was to
multiply the sites of usurpation and set them against one another.
That Constant was not worried about the people so much as about
their representatives can also be seen in the fact that he opposed the Abbé
Sieyès’s system of indirect election for representatives. Sieyès had set up
an elaborate set of electoral colleges to filter public opinion in the course
of selecting delegates. Constant opposed the system and favored replacing
the electoral colleges with direct elections (ibid.). And in the earlier, more
theoretical version of Principles, he included an entire chapter devoted to
arguing against the view that representatives, or those in government,
were better suited to rule than the people themselves. The people who
say that “light has to come from elevated places” may be right when
speaking of uncivilized societies, he wrote, but their argument is wholly
out of place in modern civilized societies with an educated class of any
significant size. It was a mistake, he thought, to “attribute to governments
the superiority of enlightenment”: “We can reply to those who want to
subject the intelligence of the many to that of the few what a famous
Roman said to his son when the latter proposed to take a town, with the
sacrifice of three hundred soldiers. Would you care to be one of this three
hundred?” (Constant 2003).
Thus it is a serious misunderstanding of Constant’s liberalism to regard
his suspicion of government efforts to directly enact popular sovereignty
as a suspicion of the people itself. Far from favoring the rule of elites,
he was suspicious of those who claimed to represent the people, those
102 Bryan Garsten

who used the language of popular sovereignty to justify their own rule.
His targets were not the people as a whole but the individuals such as
Robespierre and Bonaparte who had usurped the people’s authority.
If we turn to Constant’s counterparts in the United States – those
theorists who had framed the Constitution in the aftermath of the Amer-
ican Revolution – we find that many of them had had concerns similar
to the ones that Constant voiced. James Madison, for example, har-
bored a similar wariness about the danger of popular demagogues, espe-
cially in light of activity in the state legislatures trying to respond to the
post-revolutionary debt crisis. And, like Constant, Madison thought the
solution to this problem was not to allow any one part of government
to become the sole institutional locus of popular sovereignty. Too often,
Madison’s defense of representative government is reduced to his famous
remark in Federalist 10 (Hamilton et al. 2003) about the need to “refine
and enlarge” public opinion. But this consideration does not explain his
stance on many of the provisions in the Constitution. The truth is that
he opposed giving sovereign authority even to the most “refined” ver-
sion of public opinion. He did not, for instance, think that the Senate
should have conclusive authority over the House of Representatives, even
though he thought that the quality of deliberations would be higher there.
Instead, he favored having two separate institutional efforts to represent
the popular will present in government at once. He also favored stagger-
ing elections in the Senate, so that the assembly would contain members
who were elected at different times, and who thus reflected different snap-
shots of the popular will taken at different times. Each group of elected
officials – each chamber of Congress, the winners from each election
cycle – has a plausible case to make that they represent the popular will,
and yet they may often disagree. What this means is that no group of rep-
resentatives can plausibly insist that they are the only representation of
the popular will; none can claim popular sovereignty without their claim
being contested by others with at least as plausible a claim (Ackerman
1991). Madison’s position cannot be captured by saying that he feared
the legislature or the people would become too powerful, because in the
1790s his worries were centered on the executive and he turned to pop-
ular electoral politics as a remedy (Ferejohn 2003). The common thread
in his various political positions over the years was not an opposition to
popular politics or the legislature’s representation of the people, but a
determination to resist allowing any one part of government, legislature
or executive, to claim sovereignty for itself.
Nor was Madison the only one of the American founders to take a posi-
tion similar to the one we have seen in Constant. Alexander Hamilton’s
defense of the Supreme Court’s power of judicial review in Federalist 78
 Representative government and popular sovereignty 103

uses an argument much like the one that Constant would use to support a
constitutional monarch’s right to dissolve the legislative assembly. Just as
Constant said that the monarch should be able to appeal over the heads
of the representatives to the people themselves, Hamilton suggested that
the Supreme Court’s ability to strike down legislation was designed to
prevent “legislative encroachments” and to insure that “the intention of
the people” was preferred to “the intention of their agents” (Hamilton
et al. 2003). And even on the other side of the political spectrum we find
Thomas Jefferson sometimes making similar arguments. Jefferson, the
American founder closest to Rousseau in spirit, argued in his Notes on
the State of Virginia that a concentration of power was despotic even if it
was found in the most representative branch, the legislature. In a passage
strikingly similar to what can be found in Constant’s writings, Jefferson
argued that despotism arose not from the fact that power was given to
one person rather than many, but from the fact that it was concentrated
in one institution, unchecked by others:

It will be no alleviation [to despotism], that these powers will be exercised by

a plurality of hands, and not by a single one. One hundred and seventy-three
despots would surely be as oppressive as one. Let those who doubt it, turn their
eyes on the republic of Venice. As little will it avail us, that they are chosen by
ourselves. An elective despotism was not the government we fought for. (Jefferson
1984)

Jefferson went on to advocate as a solution to this problem the sepa-

ration of powers and a system of checks and balances, which is why
Madison cited this passage at length in Federalist 48 when defending
the proposed Constitution. Across the political spectrum at the Amer-
ican founding, then, the theorists of representative government shared
the suspicion of representatives that Constant would make central to his
writings.
Of course the authors of the Constitution did not respond to this fear
by looking for an alternative to representation. They also did not take
Rousseau’s advice to institute regular referenda on the form of govern-
ment as a whole. Jefferson did propose something in this spirit in the
form of new constitutional conventions every generation, but Madison
and the framers rejected it for reasons set out in Federalist 49. Instead they
sought to deal with the problem of usurpation by multiplying the points of
representation within government and creating a contest between them.
It has been remarked that what distinguishes the system of representative
government instantiated in the Constitution from the “mixed regime”
described in classical sources such as Polybius is that in our system every
branch is ultimately responsible to the people. In the classical mixed
104 Bryan Garsten

regime, the people were represented by one of the powers, but the other
powers were supposed to represent the nobles or the rich or other orders
of society. In insisting that all three branches, and also all levels of gov-
ernment, local, state and federal, are in some sense representations of
“the people,” the American system aims to multiply the plausible claims
to represent the popular will. In Rousseau’s language, it multiplies the
sites of usurpation. This multiplication in turn challenges any particu-
lar branch’s claim that its representation is determinative. Perhaps, if all
three branches and all levels of government come to reflect a similar sen-
timent over a sustained period of time, then the government as a whole
can be said to represent the popular will in a particularly authoritative
way (Ackerman 1991). But at least in the ordinary course of politics such
agreement is not to be expected or even hoped for. In fact, the system
seems designed in part to resist registering any unified representation
of the popular will. By striving to have multiple representations of the
popular will present in government at the same time, the system aims to
encourage contestation about what precisely it is that the people want.
There is plenty of anecdotal evidence, at least, that the system achieves
this goal; no piece of rhetoric is more omnipresent in American politics
than the politician’s claim to represent what the American people want
or need – a claim made on every side of every issue by every sort of
politician.
From this perspective the various debates among political theorists
and philosophers about what exactly political representation is, and the
debates among political scientists about what sorts of institutions best
reflect the will of the people, can be seen in a different light. These
debates are precisely the kind that the system of representative govern-
ment is designed to encourage and accommodate. Those who claim
with Madison that public opinion must be refined and enlarged, such
as contemporary proponents of deliberative democracy, will find within
representative government a place for their views; but so will those who
argue primarily for the representation of special interests or of unrefined,
populist manifestations of public opinion. What no party will find sym-
pathy for in the theory of representative government that I am drawing
out here is the idea that its particular interpretation of public opinion
should be the final or authoritative interpretation. By locating the source
of sovereignty in an abstract entity, “the people,” whose voice can be
heard only through the various interpretations of its many spokespeo-
ple, representative government instigates constant debate about what the
popular will actually is. It calls for “interpretive representation” of the
kind that Philip Pettit describes in his contribution to this volume, but it
does not allow Rawlsian public reason or any other theory to dictate one
 Representative government and popular sovereignty 105

definitive interpretation. The constant contestation over how the popular

will should be interpreted gives political life in liberal democracies much
of its vitality and energy. It drives politicians to explore different interpre-
tations and to try to make their interpretations persuasive; representative
government understood in this way aims to foster a politics of persua-
sion (Garsten 2006). It also fuels the media, which helps to multiply and
amplify different iterations of the popular will, and which investigates any
claim to represent that will and exposes its problems. It is no accident
that the proponents of this vision of representative democracy, Constant
and Madison, were also great spokesmen for the importance of a free
press (Constant 1988; Madison 1888).
The view of representative government that I am describing takes
advantage of a peculiar feature of the concept of representation itself.
As Hanna Pitkin and others have noticed, the concept of representa-
tion seems to be structured in a way that makes its complete realization
impossible. A representation that is identical to the represented object in
every way is nothing other than the object itself, which is therefore not
re-presented but simply present. Representation implies the absence of
what is being represented (Pitkin 1967). Thus, political representation
of the sovereign people implies that the people are not actually present
themselves in government. The impossibility of fully and completely rep-
resenting the people’s will – the impossibility that Rousseau highlighted –
is therefore integral to the concept of representation itself. Representa-
tion properly understood requires a distinction between representatives
and the people. This is the distinction that demagogues aim to obscure
whenever they claim to fully represent the people; it is the distinction
that representative government, with its indirectness, aims to preserve;
and it is the distinction that Rousseau, with his warnings about mixing
sovereignty and government, wanted to protect. In this sense we can say
that both Rousseau and the theorists of liberal representative government
were responding to a feature inherent in the concept of representation
itself. Liberal democratic governments – representative governments –
are those that recognize the conceptual impossibility of adequately rep-
resenting the popular will.
Of course this is not the only interpretation of representative govern-
ment that could be drawn from the history of political thought. There
is a competing story that can be traced from Chapter 16 of Hobbes’s
Leviathan, through Sieyès, the architect of the first National Assembly
during the Revolution in France, and straight through to a commonly
invoked understanding of the state today. In this alternate story, the
sovereign power of the people is inert unless it is given coherence and
agency by being represented in one unified authority – a single ruler, a
106 Bryan Garsten

single assembly, or a government as a whole understood as having the

mission of responding to one representative will. Hobbes is the crucial
starting point for this story: he argued that a multitude could only be
considered as a people, in the sense of being one entity capable of action,
if it was represented in a single sovereign: “For it is the Unity of the Rep-
resenter, not the Unity of the Represented, that maketh the Person One”
(Hobbes 1996). And Sieyès, in spite of his many blueprints for institu-
tions that would divide the government in various ways, insisted with
Hobbes that a people or “nation” could not be a coherent entity except
through a single representation. He opposed any effort to appeal past the
representatives to the people themselves, viewing the suspensive royal
veto, for example, as a compromise on the principle of representation;
he insisted that the National Assembly was “the sole authorized inter-
preter of the general will” (Sieyès 1996). From the perspective outlined
earlier, we might say that when the representatives of the Third Estate,
inspired by Sieyès, gave themselves the name of “National Assembly” and
then proceeded to govern, they thereby joined the people (the Nation)
with the government (the State) in just the manner that Rousseau had
warned against. Robespierre would, once in government, find himself
unable to resist a similar act of usurpation. Insofar as we continue to
think of ourselves as living in “nation-states” and ask only that the state
reflect our will as accurately as possible, we treat our government as
a more complex version of the National Assembly.2 If we accept this
view of our governments, we implicitly accept the usurpation reflected
in the joining of nation and state. We may insist that the representative
nation-state nevertheless respects popular sovereignty, but we can do so
only by adopting a view of popular sovereignty closer to the account put
forth by Hobbes and Sieyès than to the democratic account we saw in
Rousseau.
Thus there are at least two conceptions of representative government
that might guide our thinking. One views representative government as
a “liberal democracy” in the sense that I have tried to invoke by joining
the democratic Rousseau with the liberal Constant; the other views this
form of government as a “nation-state” in the way just described. It is
true that liberal democracies are nation-states, and therefore that some
degree of usurpation is all but inevitable (as Rousseau himself acknowl-
edged). Still, the emphasis in the two views is quite different. The first
view of representative government accents the negative function of pop-
ular sovereignty and tries to multiply usurpations so as to weaken them,

2 If the term “nation” is understood precisely as Sieyès understood it, i.e. as a state, then
the phrase “nation-state” becomes a tautology, as Istvan Hont argues (Hont 2005).
 Representative government and popular sovereignty 107

while the second view pursues a governmental representation of popular

sovereignty and does not regard usurpation as a threat.
Today there are a host of ideas about where the authoritative version of
the popular will might lie. Some suggest that it can be found in a mandate
issued by a particular election or poll; others argue that it resides in the
Constitution as interpreted by the Supreme Court. Some argue that it
can be found whenever a particular consensus emerges about certain
issues and makes its way through various governmental institutions and
procedures over a sustained period of time (Ackerman 1991). Still others
look for the authoritative popular will in the views that a people might
express after being allowed to deliberate with adequate information and
fair procedures (Fishkin 1991). These are all plausible arguments about
where the popular will might be expressed in our system of representative
government. But insofar as they imply that one expression of the popular
will should be sovereign, they all implicitly adopt the second view of
representative government described above; they all view the government
as a nation-state trying to enact a singularly authoritative interpretation
of the popular will, rather than as a liberal democracy trying to encourage
the multiplication and contestation of claims to represent the people.

III. Research questions

In addition to asking which vision of representative government seems
most attractive, scholars should also ask which vision best reflects what
is actually going on in our system. What sort of empirical research ques-
tions could be formulated, in light of the issues raised here? We already
have research designed to test how responsive government as a whole, or
particular parts of it, is to public sentiment, and this line of investigation
should surely continue. But we could also try to test the extent to which
our system of representative government does what I have suggested
early liberal theorists wanted it to do: multiply and challenge govern-
mental efforts to represent the popular will. The institutional features
of the US system that I mentioned in this regard – staggered elections
with overlapping terms of office, the different chambers of Congress and
branches of government, and so on – are familiar features of the consti-
tutional system of “checks and balances” often mentioned in textbook
accounts of government. These textbook structural features of consti-
tutional government deserve serious consideration by political scientists
(Manin et al. 1999). Work finding that the Senate and House do not
differ much on measures of responsiveness (Stimson et al. 1995) begins
to approach such questions, but does not fully address them. The ques-
tion is not just how responsive these institutions are to public opinion
108 Bryan Garsten

on a uniform scale of responsiveness, but whether they are responding

differently to the same public opinion. For example, the House and Senate
seemed to respond differently to the proposed impeachment of President
Clinton. At the time this chapter was written those same chambers were
responding differently to President Bush’s proposed strategy for the war
in Iraq. Yet both houses claim to represent the popular will. Political
scientists should investigate what considerations explain these different
representations of the popular will. What is it about the different insti-
tutions that influences the kinds of claims they make, and the extent to
which those claims are accepted by others? To what extent is there in fact
a practice of conflict between these institutions? David Mayhew’s work
on congressional opposition to the president in the United States offers
an example of the kind of work that could be done in this vein (Mayhew
2001).
To see whether claims to represent the people are successfully being
multiplied and challenged, we would also have to investigate whether one
kind of claim, or one institutional source of claims, is consistently winning
out, and whether any one institution is effectively being granted final or
sovereign authority on many issues and over a long period of time. I do
not know what we would find. Some political commentators worry that
the growth of the executive threatens this multiplicity, while others would
point to the Court’s use of judicial review as the real threat. Empirical
and historical research could yield insights into such questions. We could
also look to see whether there is more diversity among governmental
interpretations of the popular will at particular times, and why. How
does the multiplication and contestation of claims change in response to
stimuli such as wars, economic disturbances, technological innovations,
and demographic trends?
Having a research agenda based on this broader understanding of
the purpose of representative government would not only provide use-
ful descriptive information, it would also be normatively desirable. As
things stand now, if any of the studies of responsiveness mentioned at the
beginning of this chapter were to find that a particular institution in gov-
ernment did accurately and reliably “translate” popular sentiment into
policy, there would be a strong tendency, I think, to conclude that that
institution should have final authority, that it was the legitimate bearer
of our sovereignty. If responsiveness is the purpose of representative gov-
ernment and the criterion of its legitimacy, there would be no reason
to deny this conclusion. I hope in this chapter to have indicated why it
is plausible to think that this conclusion might be wrong, and to think
so for democratic reasons. Popular sovereignty can be understood in a
negative sense, as Rousseau himself often seems to have understood it.
 Representative government and popular sovereignty 109

A government institutionalizes the negative sort of popular sovereignty

when it helps us to resist the ever-present temptation to grant final and
exclusive authority to the government, no matter how much we approve
of its interpretation of popular sentiment at any particular moment. What
popular sovereignty as a normative ideal does, on this theory, is ask us to
be on the lookout for usurpation and the demagogic appeals that accom-
pany it. To evaluate the health of representative democracy, we should
look not only at whether institutions are responding to our opinions, but
also at whether the structure of government succeeds in multiplying and
challenging claims to represent our opinions. Representative government
is working well when no claim to represent the people goes uncontested.

BIBLIOGRAPHY
Ackerman, Bruce A. 1991. We the People: Foundations. Cambridge, MA: Belknap
Press.
Bartels, Larry M. 1991. “Constituency Opinion and Congressional Policy Mak-
ing: The Reagan Defense Buildup.” American Political Science Review 85 (2):
457–74.
Burke, Edmund. 2000. “Speech to the Electors of Bristol,” in On Empire, Liberty
and Reform, ed. D. Bromwich. New Haven, CT: Yale University Press, 50–7.
2003. Reflections on the Revolution in France, ed. F. M. Turner and D. M.
McMahon. New Haven, CT: Yale University Press.
Constant, Benjamin. 1988. Political Writings, ed. B. Fontana. Cambridge:
Cambridge University Press.
2003. Principles of Politics Applicable to All Governments, trans. D. O’Keeffe, ed.
E. Hofmann. Indianapolis, IN: Liberty Fund.
Dahl, Robert Alan. 1971. Polyarchy: Participation and Opposition. New Haven,
CT: Yale University Press.
Ferejohn, John. 2003. “Madisonian Separation of Powers,” in James Madison:
The Theory and Practice of Republican Government. Stanford, CA: Stanford
University Press, 126–55.
Fishkin, James S. 1991. Democracy and Deliberation: New Directions for Democratic
Reform. New Haven, CT: Yale University Press.
Garsten, Bryan. 2006. Saving Persuasion: A Defense of Rhetoric and Judgment.
Cambridge, MA: Harvard University Press.
Hamilton, Alexander, James Madison, and John Jay. 2003. The Federalist Papers,
ed. C. L. Rossiter and C. R. Kesler. New York: Signet Classic.
Hobbes, Thomas. 1996. Leviathan, ed. R. Tuck, rev. student edn. Cambridge:
Cambridge University Press.
Holmes, Stephen. 1984. Benjamin Constant and the Making of Modern Liberalism.
New Haven, CT: Yale University Press.
Hont, Istvan. 2005. Jealousy of Trade: International Competition and the Nation-
State in Historical Perspective. Cambridge, MA: Harvard University Press.
Jefferson, Thomas. 1984. Notes on the State of Virginia, ed. M. D. Peterson. New
York: Library of America.
110 Bryan Garsten

Madison, James. 1888. “Virginia Resolutions,” in The Debates in the Several State
Conventions on the Adoption of the Federal Constitution as Recommended by the
General Convention at Philadelphia in 1787, ed. J. Elliot. New York: Burt
Franklin.
Manin, Bernard. 1997. The Principles of Representative Government. New York:
Cambridge University Press.
Manin, Bernard, Adam Przeworski, and Susan C. Stokes. 1999. “Introduction,”
in Democracy, Accountability, and Representation, ed. A. Przeworski, S. C.
Stokes, and B. Manin. Cambridge: Cambridge University Press.
Mayhew, David. 2001. “Congressional Opposition to the American Presidency:
An Inaugural Lecture Delivered before the University of Oxford on 27
November 2000.” Oxford: Oxford University Press.
Miller, Warren E., and Donald E. Stokes. 1963. “Constituency Influence in
Congress.” American Political Science Review 57: 45–56.
Page, Benjamin I., and Robert Y. Shapiro. 1983. “The Effects of Public Opinion
on Policy.” American Political Science Review 77 (1): 175–90.
Pasquino, Pasquale. 2007. “Constitution et pouvoir constituant: Le double corps
du peuple,” in Sieyès. Paris: Presses Universitaires de la Sorbonne, 13–24.
Pitkin, Hanna Fenichel. 1967. The Concept of Representation. Berkeley: University
of California Press.
Rousseau, Jean-Jacques. 1997. The Social Contract and Other Later Political Writ-
ings, ed. V. Gourevitch. Cambridge: Cambridge University Press.
Sieyès, Emmanuel Joseph. 1996. “Dire de l’abbé Sieyès sur la question du veto
royal,” in La Monarchie républicaine: la constitution de 1791, ed. F. Furet.
Paris: Fayard, 406–17.
Stimson, James A., Michael B. Mackuen, and Robert S. Erikson. 1995.
“Dynamic Representation.” American Political Science Review 89 (3): 543–
65.
5 Making interest: on representation and
democratic legitimacy

Clarissa Rile Hayward

Political representation has no necessary link to democratic legitimacy.

As Hanna Pitkin (among others) has argued, there are ways of “making
present” in a political sense people who are not literally present that stand
at odds with basic democratic principles, such as political equality and
inclusiveness in collective norm-making (Pitkin 1967: ch. 2).1 Repre-
sentation is often considered a means to promoting legitimacy, however,
when those who are subject to the norms representatives make first autho-
rize those who represent them, and then hold them accountable for their
choices and their actions. Such specifically democratic forms of represen-
tation often are understood to promote legitimacy in government under
conditions in which all cannot participate directly in norm-making, if and
to the extent that they enable and motivate representatives to track the
interests of the represented (Przeworski, Stokes, and Manin 1999: 2).
Call this understanding of the democratic value of representation “the
conventional view.” Although the conventional view does not provide the
only way to understand the link between representation and democratic
legitimacy, it provides one important way. In this chapter, I make trouble
for the conventional view. In so doing, I join other theorists who have
begun the project of rethinking this orthodox understanding of the nature
and the value of political representation. David Plotke, for instance, has
argued it is not the case that representation is a second-best alternative
to direct, participatory democracy: an unfortunate necessity that follows
from the complexities of modern political life (Plotke 1997).2 Bernard

This chapter was first presented at the fall 2006 meeting of the American Political Science
Association, the fall 2006 Yale Conference on Representation and Popular Rule, and the
spring 2007 Center for Interdisciplinary Law and Policy Studies workshop at Ohio State
University’s Mortiz College of Law. Thanks to all participants in these sessions, especially
Khalilah Brown-Dean and Jenny Mansbridge.
1 Hobbesian representation, which is characterized by a total lack of responsibility or
obligation on the part of the representative, is one example Pitkin (1967) discusses at
length.
2 Participation in representative government is, Plotke claims, itself an important and
democratically valuable form of political participation (Plotke 1997).

111
112 Clarissa Rile Hayward

Manin has challenged the conventional view’s implicit assumption that

the key institutions of modern political representation – elections – are
egalitarian and democratic (Manin 1997). Andrew Rehfeld has chal-
lenged the conventional view’s tendency to conflate the empirical fact of
representation with legitimacy, or with democratic representation (Rehfeld
2006).
In what follows, my own contribution to this project will center on
the role that interests play in representation. Drawing on constructivist
approaches to the study of democratic politics, which stress the endo-
geneity of political interests to political processes, I will argue that,
because the role of government with respect to interests is never sim-
ply to track them, but always also to help shape and reshape them,
legitimacy requires more than simply “making present” the interests
of all. It requires shaping political interests in democracy-promoting
ways.
To advance this argument, I begin by considering one of the prin-
cipal concerns about the conventional view that democratic theorists
have articulated in recent years: difficulties posed for it by the prob-
lem of structural inequality. After analyzing arguments in favor of what
some term “descriptive representation” as a solution to such difficul-
ties (Section I), I turn to the question of the relation between structural
inequalities and political interests. Illustrating with the example of racial
inequalities in the contemporary United States, I make the case that
political decisions and political actions that produce structural inequal-
ities produce corresponding political interests (Section II). Under such
conditions, political institutions – electoral constituencies, political juris-
dictions, and other institutions of representative democracy – promote
equality and inclusiveness in government, not by tracking all interests
equally, but by encouraging the formation of new political interests (Sec-
tion III).

I. Descriptive representation and political legitimacy

The conventional view, then, provides one way of thinking about the
value political representation holds for democratic politics. It provides
an account that runs into difficulties, however, in the face of deep and
enduring social hierarchies: hierarchies produced by what, in the follow-
ing discussion, I will refer to as structural inequalities. By “structural
inequalities” I mean asymmetries in access to resources and opportu-
nities and in the social capacity to act, which are built into structures
 On representation and democratic legitimacy 113

such as laws and other institutions.3 Structural inequalities are political

in origin. They are not simply the product of natural differences between
individuals or groups, although they are often justified by appeals to
allegedly natural differences. And structural inequalities are institution-
alized. They are not simply the product of ongoing acts of aggression
or discrimination, although they are often accompanied by and rein-
forced by such acts. Racial inequalities in the contemporary United States
are one example, and the principal example on which I draw in this
chapter.
Why do structural inequalities pose a problem for the conventional
normative view of representation? The principal reason is that those
whom they disadvantage often cannot, by virtue of their positions in the
hierarchies structural inequalities define, authorize representatives and/or
hold them to account. In some cases, structural inequalities formally or
informally exclude those whom they disadvantage from participation in
elections. In other cases, although the disadvantaged are not formally or
informally excluded, they form numerical minorities in every electoral
constituency, or in most. But even when neither condition obtains – when
those whom they disadvantage are permitted to vote and form a numerical
majority in some districts or in the population as a whole – structural
inequalities often translate into a lack of access to the organizational and
financial resources needed to back candidates, mount successful electoral
campaigns, and effectively hold representatives to account.
The result may be a male-dominated legislature in a patriarchal society.
It may be a legislature in a racialized society that is comprised mostly, or
even entirely, of members of dominant racial groups. It may be mostly, or
only, wealthy representatives in a legislature that governs a class-divided
society. If the disadvantaged cannot constrain such representatives to
“act for” them, or if such representatives do not understand the needs,
experiences, and/or perspectives of the disadvantaged (the worry is) rep-
resentative institutions will fail the test of democratic legitimacy.4 Politi-
cal representation will ensure only that government tracks the interests of
those who occupy privileged positions in extant hierarchies, rather than
the interests of all who are governed by the norms representatives make.

3 This definition is similar to that offered by Iris Young (2000: 98), according to whom,
“[s]tructural inequality consists in the relative constraints some people encounter in their
freedom and material well-being as the cumulative effect of the possibilities of their social
positions, as compared with others who in their social positions have more options or
easier access to benefits.”
4 I borrow the phrase “act for” from Pitkin, for whom it means, roughly, “act in the
interests of” (Pitkin 1967: ch. 6).
114 Clarissa Rile Hayward

One intuitively appealing response to this problem, and one which

several recent theorists have defended, centers on what Pitkin called
“descriptive representation” and what others have termed “the politics
of presence” or “self-representation” for the disadvantaged (Pitkin 1967:
81; Dovi 2002; Mansbridge 1999; Phillips 1995; Williams 1998; Young
1990: ch. 6).5 Stated at its most basic, the argument for descriptive rep-
resentation is that, in a political society that is both internally divided
and hierarchical, people who are disadvantaged by structural inequali-
ties should be represented by people who share their positions of dis-
advantage. Those who defend descriptive representation make the case
for devising institutional means (quotas, for instance, majority–minority
electoral districts, and/or various forms of proportional representation)
to ensure, or at least to make it highly likely, that people disadvantaged
by structural inequalities will be represented by people who share their
positions in extant hierarchies: in other words (and to paraphrase Jane
Mansbridge), to ensure that “blacks . . . represent blacks, and women . . .
represent women” (Mansbridge 1999).

Descriptive representation and advocacy

There are at least two analytically distinct reasons why a democrat might
endorse descriptive representation as a remedy for problems posed by
structural inequalities. The first centers on the motivations of represen-
tatives. If one assumes that people – not excluding people who hold
political office – are motivated principally to pursue their interests, then
one might regard electoral accountability alone as an insufficient guar-
antee that those who represent disadvantaged groups will act in ways
that advance their constituents’ interests. Self-interested representatives,
after all, regardless of whom they represent, may stand to gain more by
acting against their constituents’ interests while in office than they stand
to lose by failing to secure their votes at the end of term. For this reason,
one might make the case that, in addition to accountability, a further
check on representatives’ inclinations to pursue their self-interest at the
expense of the interests of their constituents is necessary. The interests
of representatives, one might argue, should be made to correspond as
closely as possible to the interests of the people they represent.
This line of reasoning is at least as old as James Mill’s Encyclopedia
Britannica article, “On Government” (Mill 1825). Here Mill makes the

5 Pitkin is critical of descriptive representation, which she suggests is most appropriate “in
contexts where the purpose of representation is to supply information about something
not actually present” (Pitkin 1967: ch. 4, here 81).
 On representation and democratic legitimacy 115

case that political elites will be motivated to make laws that serve the
interests of those who elect them – as opposed to laws that serve their
own interests as individuals, or the corporate interests of the governing
elite – if and only if they know that they, too, will be governed by the laws
they make (ibid.: 18). It is for this reason that Mill argues for frequent
elections and for relatively short terms of office. To be sure, he does
not advance an argument for descriptive representation: for the repre-
sentation of the interests of a particular class or subgroup of a political
society by members of that class or group. Indeed, he does not even think
it necessary that members of every class or subgroup be permitted to
vote.6 Still, if one extracts from Mill’s essay its elemental logic, it sup-
plies one set of premises from which a democrat might make the case for
descriptive representation. If people are motivated to advance their own
interests (this logic suggests), and if interests vary systematically with
position in social hierarchies, then in a hierarchical society the interests
of the disadvantaged are best represented by political elites who share
their positionality.
To adopt a view such as this one, it is worth underscoring, is not nec-
essarily to assume a complete homogeneity of interests among members
of the relevant structurally disadvantaged political group. Clearly, every
group defined by position in hierarchical relations of power (African-
Americans, for example, Latinos, women, poor people) is characterized
by an internal differentiation of interests, just so long as its members
are positioned in multiple power relations. The interests of middle-class
and professional African-Americans, to cite one well-known example,
depart in non-trivial ways from those of poor and working-class African-
Americans, or at least they do with respect to some political issues and
some proposed policies.
Still, to the extent that structural inequalities follow group lines (to the
extent that residential racial segregation persists at all income levels, for
example), and to the extent that important aspects of people’s lives – their
material well-being, their opportunities to develop their capacities and to

6 Mill does make the case for a relatively broad franchise, on the grounds that represen-
tatives will be motivated to serve all interests in the larger society if and only if those
interests are shared by the members of “the choosing body” (the electorate). But he
is notoriously lax in his application of this claim. No harm is done to the principle of
representation, he argues, if women are excluded from voting. No harm is done if people
who do not hold property are denied the franchise, or if everyone under the age of forty
is denied the vote. In such cases, the interests of those who are excluded from voting,
Mill asserts, are “included in the interests of other individuals” who are permitted to
vote: their husbands and their fathers, in the case of women and young people, and in
the case of the propertyless, their fellow citizens who own property, but only in small
amounts (Mill 1825: 21).
116 Clarissa Rile Hayward

achieve their aims and their ends – are shaped by their membership in dis-
advantaged groups, their political interests will vary systematically with
group membership. African-Americans, regardless of social class, are
subjected to racial discrimination in employment, in housing, in interac-
tions with the police, and in other contexts. What is more, due to their
historical and contemporary experiences of racial subjugation, they tend
to perceive their fates as linked to the fate of their racial group (Dawson
1994). African-Americans therefore tend to use racial group interest as
a proxy for individual interest when they are forming and acting upon
political preferences (ibid.).
Whenever patterns such as these obtain, a democrat might endorse
descriptive representation on the grounds that representatives whose
interests overlap substantially with those of their constituents make the
best advocates. Melissa Williams (1998: ch. 5) illustrates using the exam-
ple of civil rights legislation in the United States. If African-Americans
are represented mostly or only by whites in debates about, and in collec-
tive decision-making processes centered on, civil rights laws, then they
are represented by people whose interests are not affected by the pro-
posed legislation in the same ways their interests are. African-American
representatives, by contrast, who share their black constituents’ positions
in power relations constructed around race, make (to borrow Williams’s
language) more “trustworthy” advocates.

Descriptive representation and deliberation

Alternatively, democrats might endorse descriptive representation on the
grounds that representatives whose experiences and perspectives overlap
substantially with those of their disadvantaged constituents make the best
participants in deliberative processes that are aimed at discovering shared
or common interests (Williams 1998: chs. 4–5).7 Again, the idea that
representative institutions can be ordered in ways that promote common
interests is hardly new. Political philosophers from Burke to Madison to
(J. S.) Mill have conceived interests on a spectrum from the particular
to the public, and they have searched for institutional mechanisms to
promote the representation of the latter (Burke 1854–6; Madison 1969;
Mill 1975). In many instances, these thinkers’ motivations have been
conservative, in the sense that among their principal aims has been the
preservation of status quo hierarchies. For Madison, for instance, “the
first object of government” is the protection of the allegedly natural

7 The two reasons are analytically distinct, but not mutually exclusive (Williams 1998: chs.
4 and 5).
 On representation and democratic legitimacy 117

bases of those social hierarchies that are constructed around property

ownership and wealth. Promoting the public interest – a function which
Madison argues representative institutions are uniquely suited to fill –
involves controlling the effects of majority factions that have a particular
interest in overturning class-based hierarchies (Madison 1969).8
Nevertheless, a democrat might take this notion of the representation
of public interests, and she might employ it in a more egalitarian fash-
ion. She might argue, for instance, that at least in some cases, people
have shared interests in helping to alleviate the adverse effects structural
inequalities have on the disadvantaged. If such effects are harmful to the
political society as a whole – if they threaten national security or stability,
for example, or if they undermine ethical norms or moral principles that
are widely endorsed by members of the political society – then a democrat
might regard descriptive representation as offering important epistemic
advantages. All political actors view collective problems from perspec-
tives that are only partial. In a hierarchical society, the perspectives of
the dominant differ systematically from those of the disadvantaged. The
latter are best positioned to understand how, and to persuade others that,
structural inequalities undermine common interests. Hence, if only those
who are privileged by structural inequalities are present in legislatures,
and even if some – even if all – representatives are motivated to promote
the public good, they may not understand how best to do so (Williams
1998; Mansbridge 1999, 2003; Young 1990, 2000).
Authorization, what is more, is an insufficient guarantee of the effec-
tive representation of the disadvantaged in deliberative processes, since
much of what representatives do when they deliberate is not, and can-
not be, directed by constituents at the moment of electoral choice. Jane
Mansbridge offers the example of Carol Mosely-Braun’s opposition in
1993 to an amendment to a bill introduced by Jesse Helms that would
have extended the United Daughters of the Confederacy’s design patent
for their insignia, which featured the Confederate flag. Prior to Mosely-
Braun’s intervention, by Mansbridge’s telling, the design patent had not
been debated in state or national politics. It had not been discussed in
any electoral campaign. It was an issue that Mosely-Braun first brought
to the attention of her fellow senators and the public. According to Mans-
bridge, “[t]he most important reason for her action seems to have been
the particular sensibility, created by experience, that led her to notice the
Confederate flag and be offended by it” (Mansbridge 1999: 647).
The example illustrates a more general phenomenon, which was dis-
cussed by Pitkin and which has been emphasized by others since: the

8 Factions of propertyless people and debtors are two examples Madison cites.
118 Clarissa Rile Hayward

unavoidable autonomy of political representatives. A non-trivial part of

what representatives do while in office involves responding to changes
that occur in the political environment between elections. Representa-
tives identify collective problems that did not exist and/or that were not
on the table at the time of their campaigns. They respond to new conflicts,
new crises, new opportunities and possibilities. They develop, evaluate,
and debate about new proposals. Political representatives, when they
engage in deliberation with one another, engage in debates that cannot
be charted or plotted in advance. Hence descriptive representation (one
might argue), because it brings to deliberative fora people who share at
least some of the relevant experiences and perspectives of the disadvan-
taged, is crucial for democratic politics under conditions of structural
inequality.

II. The political construction of interests

As should be clear from the above, these two sets of reasons cor-
respond roughly to two approaches to theorizing democratic politics:
approaches typically identified as, respectively, “aggregative” and “delib-
erative” democracy. Because the benefits and the shortcomings of each
have been discussed at length elsewhere, I will not rehearse them here
(see, e.g., the essays collected in Macedo [1999] and Elster [1998]).
Nor will I engage in the kind of careful textual exegesis that would be
necessary to explore important distinctions within the bodies of demo-
cratic theory that are typically labeled “aggregative” and “deliberative.”
Instead, my account will be deliberately stylized. Its aim is to point up
broad differences in how political theorists might understand the role that
interests play, and the role that they should play in democratic politics.

Interests in aggregative and deliberative democratic theory

By the aggregative view, the interests that inform and motivate the
choices people make are relatively stable across political interactions.
The processes through which interests are formed, interpreted, and artic-
ulated are more or less exogenous to this model. The aggregative model
assumes, in addition, that the interests that matter the most for politics
are typically people’s particular interests: what it is that serves their good
as individuals and as members of subgroups of their political society,
and/or what individuals and group members believe to serve their good
(their preferences). Political actors are assumed to pursue their interests
even if – and by some accounts especially if – those are at odds with
the interests of other individuals and subgroups. Democratic politics is
 On representation and democratic legitimacy 119

a matter of fairly aggregating conflicting particular interests, in order to

arrive at a collective choice that is legitimate, because it is responsive –
and equally so – to all whom it affects (see Dahl 1989: chs. 6–9).
For example (and to continue with the case of race in the contemporary
United States), because African-Americans are disproportionately con-
centrated in urban school districts with overcrowded, underfunded, and
underperforming public schools, while people constructed as “white”
are disproportionately suburban and wealthy, and have disproportionate
access to resource-rich, high-performing schools, a theorist of aggregative
democracy likely would expect most blacks to act in ways that promote
their interests in the improvement of urban education and/or the redis-
tribution of educational resources and opportunities. She would expect
most whites, by contrast, to act in ways that promote their interests in
keeping resources in their local public schools. The principal advan-
tage of descriptive representation, by this view, is that it would ensure
that blacks’ particular interests were represented whenever choices about
education (or other collective choices that affect them) were made.
As the example makes clear, however, on the aggregative model there
are real limits to descriptive representation as a solution to the problem of
the chronic underrepresentation of the interests of African-Americans,
or for that matter of any subset of a collectivity that forms a numer-
ical minority. If the purpose of descriptive representation is to ensure
governmental responsiveness to the interests of all, including those who
are disadvantaged by structural inequalities, and if political actors are
motivated principally to promote their particular interests, then descrip-
tive representation cannot ensure governmental responsiveness to dis-
advantaged minorities whose interests vary systematically from those of
dominant majority groups. African-Americans are a numerical minority
in the US, and in every state of the US. Thus even if they are repre-
sented in proportion to their numbers in state and federal legislatures,
and even if their representatives advocate effectively for their interests,
those representatives themselves will remain in the minority.
Hence the appeal to some theorists of descriptive representation of a
more deliberative model of democracy, which assumes that the interests
that shape and motivate the political choices people make are relatively
fluid over the course of political interactions. The deliberative model
assumes, in particular, that people’s perceptions of what it is that serves
their good can – and that in some non-trivial subset of cases they do –
change in response to the reasoned exchange of opinions and argu-
ments. The deliberative model assumes, further, that although people
often are motivated to pursue their particular interests, they also can
be motivated to pursue common or collective interests, even when such
120 Clarissa Rile Hayward

action is at odds with their particular interests as they understand them

at the outset of deliberation. Democratic politics, by this view, is not
principally a matter of aggregating particular interests. It is also – and
from a normative perspective it is most importantly – a matter of encour-
aging open, inclusive, and egalitarian forms of rational argumentation,
with a view to discovering shared or common interests. Representation
promotes legitimacy in government, not by tracking any interests, but
by tracking people’s post-deliberative interests, or their interests as they
(would) understand them after subjecting them to free, equal, and public
rational argumentation (Benhabib 1996; Habermas 1990, 1996; Manin
1987).
In the case of racial disparities in access to educational resources, a
deliberative democrat might emphasize the importance of enabling urban
blacks to persuade suburban whites that their shared interests in having
an educated citizenry, and/or their common interests in promoting the
principle of equal opportunity, outweigh whites’ particular interests in
keeping local schools relatively resource-rich. The principal advantage
of descriptive representation, by this view, is that it gives voice to the
perspectives and the claims of those who are disadvantaged by extant
hierarchies. It thus enables them to persuade their more privileged inter-
locutors to reconceptualize their political interests and/or to sacrifice their
particular interests for the common good.
Again, the example makes clear the limits to descriptive representation
as a solution to the problem of the underrepresentation of those who
are disadvantaged by structural inequalities. Even if black representa-
tives understand the perspectives of their black constituents, even if they
share (some of) their relevant political problems and experiences (if they
know what it means to struggle with a failing school system, for exam-
ple), they will not necessarily be able to persuade the representatives
of whites to act in ways that promote educational redistribution. To be
sure, the privileged sometimes sacrifice their particular good to advance
what they understand to be the shared or the common good. More often
than not, however, those who are privileged by extant power relations
pursue their particular interests in maintaining their privilege: in paying
relatively low taxes for a top-notch public education, to continue with the
present example, or in reducing the competition their children will face,
down the road, for positions in elite universities. Even if black represen-
tatives advance compelling claims about the advantages of educational
redistribution for American society as a whole, even if they advance philo-
sophically sophisticated arguments based on widely endorsed moral or
ethical principles, it seems pollyannaish to hope or to expect that the
representatives of suburban whites will heed their claims.
 On representation and democratic legitimacy 121

Making interests
These shortcomings notwithstanding, both the aggregative and the delib-
erative approaches offer important insights about the role that interests
play in democratic politics. Theorists of aggregative democracy rightly
underscore that among the most reliable predictors of how political
agents act are their particular interests. Even many deliberative democrats
grant this point. Many grant that, even if there are instances in legisla-
tures and in other political settings when actors with conflicting particular
interests discover new common interests, much of politics remains a mat-
ter of deciding what to do in the face of conflicting interests, and many
collective decisions are coercive in the sense that they further the inter-
ests of some of the individuals and groups they affect, but not others
(Mansbridge 1996).
Interests are important motivators of political action, and theorists of
aggregative democracy are right to emphasize this is the case. At the
same time, deliberative democrats are right to underscore that interests
are not fixed or static: that much of what happens in democratic politics
is not simply a matter of responding to people’s interests and preferences
(by advancing them, or by failing to) but also of shaping what it is that
people want and/or what it is that serves their good.
Typically, deliberative democrats emphasize only one way in which
political action shapes interests: they stress that reasoned argumentation
changes how people understand or interpret their interests and/or how
they weigh them against competing claims. But if an actor’s political inter-
ests are her interests only relative to some particular set(s) of collective
norms (if I have an interest in maximizing my income and my wealth,
for example, only relative to the norms of a capitalist economy), then
any political action that shapes relevant institutions (the legal construc-
tion of property rights, for instance) also, unavoidably, shapes political
interests.9 Political action that produces structural inequalities – collec-
tive choices that result in institutional norms that define unequal access
to resources and opportunities and to the social capacity to act – tends to
produce corresponding sets of political interests. For those whom struc-
tural inequalities privilege, it tends to produce interests in exploiting

9 “Shapes” is important, since institutions do not determine interests. Faced with the norms
of a capitalist economy, I might nonetheless reject the aims of income- and wealth-
maximization. I might, for instance, adopt religious ascetic ends and define my interests
in terms of my own spiritual development. If so, however, either these interests are not
political interests (perhaps I pursue self-denial solipsistically, wholly indifferent to how
my society structures its economy) or capitalist norms do shape my interests, by helping
define the worldly practices and the desires I reject and resist.
122 Clarissa Rile Hayward

and perpetuating their privilege. For those whom structural inequalities

disadvantage, it tends to produce interests in escaping, if possible, and if
not in minimizing or managing the effects of disadvantage. If democrats
devise ways to represent interests such as these – to “make present” politi-
cal interests which themselves are the products of structural inequalities –
the likely result is the perpetuation of hierarchy.
Consider again the interests that many black Americans share in the
redistribution of educational and other opportunities and resources. By
producing both race-based residential segregation and also a highly dif-
ferentiated and decentralized system of schooling, political actions taken
by local municipal officials and by state and federal legislatures (among
other actors) helped to construct and maintain these interests. At the
same time, they constructed a set of “white” interests that center not
only on reaping the benefits of an inegalitarian educational system, but
more generally on maintaining relatively high levels of local public service
provision with minimal tax expenditure.
“White” interests in perpetuating racial privilege, it is worth underscor-
ing, are neither natural nor necessary. Indeed, had the collective norms
that helped to produce the interests of today’s suburban whites been sub-
jected at the point of decision to the forms of rational argumentation that
deliberative democrats recommend, many of them likely would not have
withstood that test. The decisions that helped produce racial segregation,
for example (such as collective decisions in the early part of the twentieth
century to engage in racial zoning, or decisions through mid-century to
permit and to enforce racially restrictive covenants), were informed by
claims about racial differences that were, at best, unsubstantiated. These
included, perhaps most notably, the claim that African-Americans, by
virtue of their race, pose a threat to property values.10
Although people constructed as white, if they believed the racist claims
in circulation at that time, may have believed that racial segregation
served their interests, no doubt for many of them it did not. Pro-
cesses of racial segregation constrained not only the housing choices
of African-Americans, but also the housing choices of many so-called

10 This claim was advanced especially forcefully by prominent members of the real estate
industry. Frederick Babcock, for example, in his influential Valuation of Real Estate
asserted, “Among the traits and characteristics of people which influence land values,
racial heritage and tendencies seem to be of paramount importance. The aspirations,
energies, and abilities of various groups in the composition of the population will largely
determine the extent to which they develop the potential value of land” (Babcock 1932:
86). Rose Helper (1969), who systematically surveyed mid-century real estate text-
books, appraising manuals, and National Association of Real Estate Boards (NAREB)
publications, found that almost all endorsed some version of this claim about the relation
between race and property values.
 On representation and democratic legitimacy 123

“ethnic whites.” In many instances, the fear of racialized others that

racist claims incited prompted working-class people to abandon neigh-
borhoods in which they had invested substantial resources and had estab-
lished important networks (Sugrue 1996).
Once these claims had informed collective decisions that in turn
helped produce structural inequalities, however, the result was a new
“white” interest in racial exclusivity. A case in point involves decisions
made by New Deal housing agencies such as the Home Owners Loan
Corporation (HOLC) and the Federal Housing Administration (FHA),
which are widely credited with nationalizing the practice of redlining
(Jackson 1985). The latter agency defined racial segregation in the neigh-
borhood surrounding a property as a necessary condition for that prop-
erty to receive a favorable rating for a federal loan guarantee. It specif-
ically recommended the use of racially restrictive covenants to promote
segregation.11 In the postwar years it channeled the vast majority of state-
backed mortgages to racially exclusive white suburbs.12
The FHA mortgage program affected people’s interests profoundly. It
favorably affected the interests of many working- and middle-class white
Americans by dramatically lowering their down-payments and interest
rates and by extending their amortization periods, thus making it possible
for them to purchase and to own private homes. It adversely affected the
interests of many African-Americans to whom it denied this important
opportunity to accumulate wealth. But, because it extended loan guaran-
tees to white home buyers on the condition that their homes be sited in racially
exclusive enclaves, the FHA not only differentially affected interests, but
also helped to create a new constellation of race-based political interests.
A parallel argument could be developed about processes of disinvest-
ment from urban public schools and from other institutions in the ghet-
toized sections of America’s cities. By channeling resources away from
black ghettos and to newer cities and suburbs, as well as to the cen-
tral business districts and the gentrifying residential sections of older
cities, state and federal legislatures, local redevelopment authorities, and

11 In its 1938 Underwriting Manual, the FHA advised that neighborhood ratings
should reflect the presence of “Adverse Influences,” which it defined to include
“incompatible racial and social groups.” “Recorded restrictive covenants,” the man-
ual asserted, “should strengthen and supplement zoning ordinances [and] . . . should
include . . . prohibition of the occupancy of properties except by the race for which they
are intended” (FHA 1938: par. 937).
12 From the start of the FHA program though the early 1960s, a period during which
this agency insured mortgages on close to a third of new housing in the United States,
African-Americans received less than 2 percent of these state-insured mortgages. Even
these went disproportionately to racially segregated neighborhoods in the South (Squires
1994).
124 Clarissa Rile Hayward

powerful private actors such as builders and developers, made collective

decisions that did not obviously serve the interests of all middle-class
people, or of all people constructed as white (Baxandall and Ewen 2000;
Frieden and Sagalyn 1990; Gelfand 1975; Hirsch 1998). Urban disin-
vestment exacerbated non-trivial collective problems, among them job-
lessness, crime, victimization, and a host of social problems associated
with concentrated poverty (Massey and Denton 1993; Wilson 1987).
Again, such actions differentially affected interests. Urban blacks, for
instance, were made to bear disproportionately the financial, social, and
psychological costs of “slum clearance” (Fullilove 2004; Teaford 1990,
2000). But they also created new political interests: interests that corre-
sponded to the structural inequalities they defined. For many poor and
working-class African-Americans, they created interests in managing or
minimizing the effects of the collective problems they localized in so-
called “slums.” For the many whites whom they privileged, by contrast,
they created interests in keeping the effects of collective problems at bay,
by zoning them out of their (urban and suburban) enclaves.

III. Reconstructing interests

The example points to what I want to suggest is a larger problem with the
conventional view of representation’s democratic value: it is insufficiently
constructivist about political interests. Representative institutions do not
promote political equality and inclusiveness in government by enabling
and motivating representatives to track the interests of those they repre-
sent, if and when political interests themselves reflect structural inequal-
ities. If representative institutions are to be democracy-promoting under
conditions of deep and enduring hierarchy, they need to do more than
simply “make present” the interests of all. They need to encourage the
formation of new political interests. In particular, they need to encourage
among those who are disadvantaged by extant hierarchies new opposi-
tional interests, and among those who are privileged by extant hierarchies,
new interests in addressing – rather than simply in avoiding the effects of –
collective problems.

Interests in contestatory democratic theory

If the arguments for descriptive representation sketched above (Section
II) correspond, respectively, to aggregative and deliberative understand-
ings of what democracy is and what it requires, this more construc-
tivist approach is informed by an account of democratic politics typically
labeled “contestatory” (Connolly 1991; Mouffe 2000; Tully 1999, 2000,
 On representation and democratic legitimacy 125

2002; Wolin 1996). Legitimacy in government, by the contestatory view,

is not a matter of the fair aggregation of particular interests. Nor is it
a matter of the realization, or the approximation, of rational consensus
about shared interests. Instead, legitimacy is the product of inclusive and
egalitarian political contests, which center on ethical and moral princi-
ples, as well as on collective decisions. In contradistinction to the delib-
erative approach, by this view the orientation of participants in political
contest is not mutual agreement, but success. When people engage in
political struggles, that is to say, their aim is to win. In contradistinction
to the aggregative model, by the contestatory view, political struggle has a
status that is not merely empirical, but also normative. Politics not only in
fact does, but also should involve contests over both principles and norms.
What lends success-oriented contestation this normative status? Con-
testatory democrats have offered multiple answers to that question. Some
have suggested that contestation is of value for democracy because it is
instrumental in promoting other democratic goods, such as a willing-
ness to engage respectfully and non-violently with political opponents.13
Others have suggested that the political changes that result from
contestation – the destabilization of institutions and the repeated revision
of collective norms – themselves hold democratic value.14
The most compelling reasons, however, are Arendtian in their logic
(Arendt 1958: 176–7; Pitkin 1981; Honig 1995). If human agency and
human plurality place what is sometimes called “collective autonomy”
out of reach – if, that is to say, democratic political actors never attain
sovereignty over the effects of their actions, much less over the social
constraints that delimit their fields of possible action – then democratic
freedom cannot be equivalent to the independence of action, or to non-
interference with action, or to what Patchen Markell has called “masterful
agency” with respect to one’s action (Markell 2003; Hayward 2000).
Instead, democratic freedom is best understood as a capability that is
both social and relative: a politically produced capacity to participate in

13 See, for example, Connolly (1991). This particular line of reasoning makes contestatory
democracy less distinctive from deliberative democracy than do others, however. Delib-
erative democrats often argue for some forms of contestation on the Millian grounds that
they are instrumental to promoting other democratic goods: namely, the discovery of
rational and reasonable interests. This line of reasoning is also largely unsubstantiated.
There is little reason to believe that contestation over norms and principles encourages
mutual respect across lines of difference. For a persuasive critique along these lines, see
Deveaux (1999).
14 See, for example, Wolin (1996). Again, this line of reasoning is less than persuasive.
It is unclear why the destabilization of institutions and the revision of collective norms
are, in and of themselves, democratic goods. For a persuasive critique of this line of
reasoning, see Markell (2006).
126 Clarissa Rile Hayward

those struggles through which the norms that govern action are made and
remade. Legitimate government, by this view, is realized just to the extent
that people who stand to one another in relations of interdependence and
mutual vulnerability are enabled – and constrained – to participate in free
and equal contests over the terms of those relations.15
This chapter is not the place for a comprehensive or a systematic
discussion of the differences that separate contestatory from aggregative,
and from deliberative theories of democracy.16 But one difference that,
for present purposes, is worth highlighting is the value for democratic
politics that this approach accords to political interests. By the aggregative
view, the principal reason interests should influence political outcomes
is that the agents those outcomes affect “have” them – or more precisely:
the agents those outcomes affect believe they have them and, other things
equal, are likely the best judges of whether they do.17 That it is in my
interest to do x or to have y, by this view, is an objective fact about me as
a political agent (or, at least, very likely a fact about me, given my belief
that it is). Hence, if I am deserving of inclusion in some democratic
process because the outcomes of that process affect me, then the process
should be structured such that my interests relative to its outcomes are
tracked.
By the deliberative view, by contrast, interests should influence out-
comes, not simply because the agents whom those outcomes affect believe
they have them, but because – and only if – their beliefs about their inter-
ests have withstood, or in principle could withstand, the test of inter-
subjective reason. It might be the case that I believe it is in my interest

15 By including the phrase “and constrained,” I do not mean to suggest that people should
never be able to opt out of participating in democratic struggles or contests, but rather
that they should not be able to opt out of engaging their political opponents while
continuing to act in ways that affect them. This point is developed below.
16 As in the previous section, my aim here is not textual exegesis. Instead, it is an explication
of a particular alternative understanding of what makes democratic politics legitimate,
and how representative institutions might contribute to that end. Hence my argument is
not necessarily one that all theorists who self-identify as contestatory democrats would
endorse. Nor is it necessarily one that all who self-identify as aggregative or deliberative
democrats would reject. I draw in this section on the work of James Tully, for example,
who presents his understanding of contestatory democracy as a version of deliberative
democracy. More generally, I would include in this category readings of Habermas that
emphasize the importance of the revisability of both the outcomes of and the rules that
structure discourse, and that de-emphasize consensus as a political ideal.
17 Thus, for Dahl, the “principle of the equal consideration of interests” (which states that
“the good or interests of each person must be given equal consideration”), combined
with the “presumption of personal autonomy” (the claim that, absent “a compelling
showing to the contrary everyone should be assumed to be the best judge of his or
her own good or interests”), justify the use of democratic procedures for determining
collective norms (Dahl 1989: 85, 100).
 On representation and democratic legitimacy 127

to do x, but that I would believe it is in my interest to do not-x if I

had perfect information, complete autonomy, and an unconstrained set
of alternatives from which to choose. If so, my belief about my interest
is irrational (Elster 1997). By the same token, it may be the case that
it is in my interest as a private individual to do x, but that I also have
a public interest in doing not-x, which I share with other actors whose
interests are affected by the relevant norm. If so, my interest in doing x is
unreasonable.18 Democratic processes, the deliberative claim is, should
be structured such that they track only those interests that are rational
and reasonable.
On this question of the democratic value of political interests, the
contestatory view is at once more demanding than the aggregative, and
less demanding than the deliberative democratic position. If interests are
contingent in the sense that they are always the product of some partic-
ular configuration of institutional norms, then it cannot be the case that
interests should influence outcomes just because those affected “have”
them. The privileged might have an interest in maintaining their privilege
by naturalizing or otherwise depoliticizing its foundation. But if so, that
does not mean that that interest should influence political outcomes.
Still, interests can be of democratic value without being fully rational
and fully reasonable in the strong (Rawlsian) sense of that term.19 The
disadvantaged might have interests that conflict with those of the privi-
leged, for example, and irreducibly so – there may be a case of a zero-
sum conflict of interests among those affected by a norm or a proposed
norm – and yet it might be the case that their interests should influ-
ence outcomes. The relevant question is neither which interests people
happen to have, nor whether the interests people have are fully rational
and reasonable. Instead, it is whether political interests promote – or
whether they inhibit – free and equal struggles over collective norms and
principles. When power relations are hierarchical, therefore, and when
hierarchies are relatively enduring because they are the product of
inequalities that are structural in form, then the interests that should
affect outcomes include interests in destabilizing inequalities and in
subverting entrenched hierarchies. Political institutions, including insti-
tutions of political representation, should encourage the formation of
interests such as these.

18 Or, more precisely, it would be unreasonable for me to pursue my interest in doing x, if

and to the extent that the choice whether to x is a political choice which affects not only
me, but also others. Were I to pursue my interest in doing x, I would, in Elster’s terms,
employ behavior appropriate to the market in the political forum (Elster 1997).
19 “Reasonable” in the Rawlsian sense means, roughly, concerned to live with others on
terms they might accept (Rawls 1993: 48–54).
128 Clarissa Rile Hayward

Contestatory democracy and representative institutions

Thus the standard complaint against contestatory democratic theories –
that, in political contests, those advantaged by extant hierarchies always
win – is misplaced. The privileged often win not by engaging in free
and equal political contests with their opponents, but rather by avoid-
ing such contests altogether. Still, a different complaint carries weight.
Theorists of contestatory democracy – unlike theorists of deliberative
and aggregative democracy – have had relatively little to say about the
legitimacy, or about the comparative democratic value, of different polit-
ical institutions.20 This is the case in no small part due to the emphasis
many place on the democratic value of transgressing institutional forms.
If democracy is (to borrow Sheldon Wolin’s language) “fugitive” – if it
appears only during those “rebellious” episodes when ordinary people
use collective power to undo some entrenched political system or some
established political order – then whatever new order they create – what-
ever institutions, whatever laws, whatever rules and procedural norms –
will be, by definition, not democratic, but the end or the edge of democ-
racy (Wolin 1996). Thus, for Wolin, “Institutionalization marks the
attenuation of democracy: leaders begin to appear, hierarchies develop,
experts of one kind or another cluster around the centers of decision;
order, procedure, and precedent displace a more spontaneous politics”
(ibid.: 39). Democracy, by this view, is a “moment” or an “ethos” more
than a political form or a set of institutional arrangements (Wolin 1996;
Connolly 1991).
If one conceives the normative justification for democratic contestation
in more Arendtian terms, however – if one sees the value of political
contest, not in transgression or rebelliousness per se, so much as in the
capacity it creates for the agentive and egalitarian making and remaking
of relations of power – then one can articulate guidelines for developing
and for evaluating political institutions, including institutions of political
representation. One important such guideline is that institutions should
constrain those agents who (would) act in ways that affect collective
norms to engage politically the other agents those norms (would) affect.
Such engagement need not be direct. It can be mediated by parties, by
elections, and by other institutions of representative democracy. But, at
the limit, no single participant in a relation of power, and no subset of
participants in such a relation, should be enabled to act unilaterally to
affect its terms.
This guideline points up what, from a contestatory democratic per-
spective, is the principal problem with despotical rule. The problem is

20 One exception is James Tully. See especially Tully (1995).

 On representation and democratic legitimacy 129

not that the despot decides and acts in ways with which her subjects do
not (or would not) agree, much less that her opinions and theirs are not
properly aggregated. Instead, the problem is that the despot works on
her subjects like an artisan on material. Rather than acting in a way that
interacts with, that is responsive to, the actions of the others her action
affects, the despot extends into politics the logic of what Arendt calls
fabrication (Arendt 1958: Part IV). Even if the decisions that she makes
are decisions with which her subjects (might) agree, even if the decisions
that she makes do not prevent her subjects from behaving as they wish
to behave, the rule of the despot undermines democratic freedom (Pettit
1997).
Such unfreedom, of course, can obtain absent despotical rule. Such
unfreedom can obtain even absent rule by any identifiable agent who
“works” on the fields of action of other actors. Deeply entrenched norms
and institutions can subject people to forms of impersonal power that
are as far removed from their influence as the dictates of a tyrant. A
second guideline, then, is that institutions should render the terms of
power relations in principle open to political challenge and revision. To
say so is not to imply that constant change is necessary or desirable. But
change must be possible. To the extent that the terms of power relations
are removed from the reach of some or all participants (to the extent, for
example, that they are naturalized or sacralized), democratic freedom is
abridged.
Consider again the example of power relations structured around
“race” in American cities and suburbs. One important set of institu-
tional norms that helps set the terms for the management of these power
relations consists in norms defining electoral constituencies and political
jurisdictions: institutions (such as local school districts and other special-
purpose governments, as well as general-purpose governments such as
urban and suburban municipalities) that determine whom representa-
tives represent, and in the making of which collective norms.
Local constituencies and political jurisdictions in the United States
are exceedingly fragmented: a trait that, in the context of racial and
socioeconomic residential segregation, enables the privileged to act in
ways that profoundly affect the disadvantaged, but without engaging
them politically. In 2002 (the year of the most recent census of American
governments), there were nearly 88,000 local governments in the United
States. Of these, nearly 39,000 were general-purpose governments.21

21 The exact figures are 87,849 and 38,971, respectively (US Bureau of the Census 2002).
In the Pittsburgh metropolitan area (the most politically fragmented of America’s major
metropolitan areas), 412 general-purpose governments averaged less than 6,000 resi-
dents each (Dreier, Mollenkopf, and Swanstrom 2004: 45).
130 Clarissa Rile Hayward

This level of political fragmentation is a relatively recent development in

the US. Through much of the nineteenth century, most American cities
expanded by annexing their surrounding territories. Beginning in the
early decades of the last century, however – just when the black population
surged in the urban north, and when the mass production of automobiles
and tract housing opened the suburbs to the middle and the lower-middle
classes – state legislatures throughout the nation shifted from a pattern
of enabling urban annexation to one of legally restricting annexation,
permitting suburban municipalities to incorporate, and granting them
significant governance authority (Briffault 1990).
By mid-century, most American suburbs could engage in a fairly wide
range of autonomous political decision-making. Most could make land
use decisions autonomously. Most could pass exclusionary zoning ordi-
nances, for example, if there was a high enough level of local public
support. Most could set admissions policies for, and they could raise
taxes to fund, public local schools. More generally, most American sub-
urbs could exclude would-be residents who lacked the resources needed
to gain entrance, and they could raise and spend taxes to fund services
which they made available to residents only. The privileged thus were
enabled to retreat to relatively homogeneous political jurisdictions where
they could form constituencies and elect representatives whose deci-
sions would affect the disadvantaged, but where the disadvantaged were
excluded from authorizing those representatives and from holding them
accountable.
In the US, local constituencies and jurisdictions are not only frag-
mented; they are also territorially based. The latter trait, as Richard
Thompson Ford has argued, serves both to naturalize these institutions
and to depoliticize them (Ford 1999). Territoriality, Ford’s claim is,
encourages people to conceive constituencies and jurisdictions in one
(or in both) of two (logically inconsistent) ways: as the natural product
of organic communities, which predate the exercise of state power, or
as politically neutral tools through which states administer policies and
laws which treat all people equally. In the US, Ford argues, territorial
jurisdictions work to legitimize race-based social hierarchies of the sort
sketched in Section II, above: hierarchies that likely would be widely per-
ceived as illegitimate if not perpetuated through the legal mechanism of
the territorial jurisdiction.
It is in large part due to the fragmentation and the territorial grounding
of political jurisdictions that many urban blacks have political interests
in managing educational and other inequalities (for example, by work-
ing locally to increase property taxes and/or to shift public expenditures
toward schooling) or in minimizing them (by working through state courts
 On representation and democratic legitimacy 131

to press for school finance reform, for example), rather than in eradicating
these inequalities altogether (by instituting cross-district desegregation
plans, for instance, or by eliminating residency requirements for public
school attendance). It is not in their political interests to pursue the latter
objectives for the simple reason that these courses are next to impossi-
ble to chart within the extant structure of representative institutions. By
the same token, it is because local constituencies and jurisdictions enable
them to do so that many suburban whites have interests in avoiding the
effects of failing schools (for example, by passing zoning ordinances that
effectively prevent poor and working-class people from living in their
municipalities, thus keeping student need relatively low and the prop-
erty tax base relatively high) rather than in addressing these and related
collective problems.
In a context like this one, changing political interests requires chang-
ing the definition of electoral constituencies and political jurisdictions.
There are multiple ways to do so. Some theorists have made a case for
a substantial centralization of governance functions to the metropoli-
tan and/or the regional level (Young 1990 ch. 8). Others have argued
for periodically redrawing municipal and other political jurisdictional
boundaries, with a view to promoting political equality and inclusive-
ness in government (Briffault 1996). Still other theorists have proposed
various forms of cross-jurisdictional voting, so that, for instance, city
dwellers and residents of older “inner-ring” suburbs would be permitted
to vote in the elections of neighboring affluent suburban municipalities
whose decisions significantly affected them (Ford 1994). Each approach
has its practical strengths and limitations (Hayward 2003), but the basic
principle motivating each is the same. Representative institutions should
ensure that it is in the interests of all to address problems that are, in a
causal sense, collective. They should constrain people who are privileged
by extant hierarchies to engage politically those whom their decisions
and actions affect, and they should grant those who are disadvantaged by
structural inequalities the collective power to change, rather than simply
to negotiate, those inequalities.

IV. Conclusion
In this chapter, I have argued against the conventional view, according
to which representative institutions promote democratic legitimacy by
enabling and motivating representatives to track the interests of the rep-
resented. Under conditions of structural inequality, I have argued, legit-
imacy requires that representative and other political institutions shape
interests in democracy-promoting ways. Specifically, it requires that they
132 Clarissa Rile Hayward

encourage among the privileged interests in engaging in political contests

over collective problems and collective decisions, and among the disad-
vantaged oppositional interests in subverting entrenched hierarchies.
What it means to encourage the formation of such political interests,
and which specific institutional changes that requires, will vary from case
to case. In some cases, descriptive representation may be one important
step. But for the reasons sketched above, descriptive representation often
will not be enough. The challenge, then, is to restructure representative
institutions so that they do not simply track the interests of all, but rather
change political interests in ways that promote democratic inclusiveness
and political equality.
In many cases, changing interests will require enabling the formation
of new alliances and new solidarities among groups that, individually,
form minorities in the relevant population, and that are differently dis-
advantaged by extant hierarchies. More generally speaking, the aim is
not to aggregate the interests of the governed, whatever those interests
happen to be. Nor is it to ensure that collective norms promote the inter-
ests of all, because they are the product of the rational agreement of all.
Instead, the aim is to ensure that those who stand in relations of power
to one another engage in free and equal contests over the terms of those
relations. To the extent that they engage one another through political
representatives (whom they elect and hold accountable), representative
institutions should help make and remake political interests in ways that
promote such contests.

BIBLIOGRAPHY
Arendt, Hannah. 1958. The Human Condition. Chicago: University of Chicago
Press.
Babcock, Frederick Morrison. 1932. The Valuation of Real Estate. New York:
McGraw-Hill.
Baxandall, Rosalyn and Elizabeth Ewen. 2000. Picture Windows: How the Suburbs
Happened. New York: Basic Books.
Benhabib, Seyla. 1996. “Toward a Deliberative Model of Democratic Legit-
imacy,” in Democracy and Difference, ed. Seyla Benhabib. Princeton, NJ:
Princeton University Press, 67–94.
Briffault, Richard. 1990. “Our Localism, Part II – Localism and Legal Theory.”
Columbia Law Review 90 (March): 346–454.
1996. “The Local Government Boundary Problem in Metropolitan Areas.”
Stanford Law Review 48 (May): 1115–71.
Burke, Edmund. 1854–6. “Speech to the Electors of Bristol,” in The Works of the
Right Honorable Edmund Burke, 6 vols., ed. Edmund Burke. London: Henry
G. Bohn, vol. IV.
 On representation and democratic legitimacy 133

Connolly, William. 1991. Identity\Difference: Democratic Negotiations of Political

Paradox. Ithaca, NY: Cornell University Press.
Dahl, Robert. 1989. Democracy and its Critics. New Haven, CT: Yale University
Press.
Dawson, Michael. 1994. Behind the Mule: Race and Class in African-American
Politics. Princeton, NJ: Princeton University Press.
Deveaux, Monique. 1999. “Agonism and Pluralism.” Philosophy and Social Crit-
icism 25 (4): 1–22.
Dovi, Suzanne. 2002. “Preferable Descriptive Representatives: Will Just Any
Woman, Black, or Latino Do?” American Political Science Review 96 (4):
729–43.
Dreier, Peter, John Mollenkopf, and Todd Swanstrom. 2004. Place Matters:
Metropolitics for the Twenty-First Century. Lawrence, KS: University Press
of Kansas.
Elster, Jon. 1997. “The Market and the Forum: Three Varieties of Political
Theory,” in Deliberative Democracy: Essays on Reason and Politics, ed. James
Bohman and William Rehg. Cambridge, MA: MIT Press, 3–34.
(ed.). 1998. Deliberative Democracy. Cambridge: Cambridge University Press.
Federal Housing Agency (FHA). 1938. Underwriting Manual, Underwriting and
Valuation Procedure under Title II of the National Housing Act. Washington,
DC: US Government Printing Office, para. 937.
Ford, Richard Thompson. 1994. “The Boundaries of Race: Political Geography
in Legal Analysis.” Harvard Law Review 107 (June): 1843–1921.
1999. “Law’s Territory (A History of Jurisdiction).” Michigan Law Review 97
(4): 843–930.
Frieden, Bernard and Lynne Sagalyn. 1990. Downtown, Inc.: How America
Rebuilds Cities. Cambridge, MA: MIT Press.
Fullilove, Mindy Thompson. 2004. Root Shock: How Tearing Up City Neighbor-
hoods Hurts America, and What We Can Do About It. New York: Ballantine
Books.
Gelfand, Mark. 1975. A Nation of Cities: The Federal Government and Urban
America, 1933–1965. New York: Oxford University Press.
Habermas, Jürgen. 1990. “Discourse Ethics: Notes on a Program of Philosoph-
ical Justification,” in Moral Consciousness and Communicative Action, trans.
Christian Lenhardt and Hierry Weber Nicholsen. Cambridge, MA: MIT
Press, 116–94.
1996. “Three Normative Models of Democracy,” in Democracy and Differ-
ence, ed. Seyla Benhabib. Princeton, NJ: Princeton University Press, 21–
30.
Hayward, Clarissa. 2000. De-facing Power. Cambridge: Cambridge University
Press.
2003. “The Difference States Make: Democracy, Identity, and the American
City.” American Political Science Review 97 (4): 501–14.
Helper, Rose. 1969. Racial Policies and Practices of Real Estate Brokers. Minneapo-
lis: University of Minnesota Press.
Hirsch, Arnold. 1998. Making the Second Ghetto: Race and Housing in Chicago,
1940–1960, second edn. Chicago: University of Chicago Press.
134 Clarissa Rile Hayward

Honig, Bonnie. 1995. “Toward an Agonistic Feminism: Hannah Arendt and the
Politics of Identity,” in Feminist Interpretations of Hannah Arendt, ed. Bonnie
Honig. University Park: Pennsylvania State University Press, 135–66.
Jackson, Kenneth. 1985. Crabgrass Frontier: The Suburbanization of America. New
York: Oxford University Press.
Macedo, Stephen (ed.). 1999. Deliberative Politics: Essays on Democracy and Dis-
agreement. New York: Oxford University Press.
Madison, James. 1969. “Federalist 10,” in The Federalist Papers, ed. Clinton
Rossiter. New York: New American Library of World Literature, 77–89.
Manin, Bernard. 1987. “On Legitimacy and Political Deliberation.” Political
Theory 15 (3): 338–68.
1997. The Principles of Representative Government. Cambridge: Cambridge Uni-
versity Press.
Mansbridge, Jane. 1996. “Using Power/Fighting Power: The Polity,” in Democ-
racy and Difference, ed. Seyla Benhabib. Princeton, NJ: Princeton University
Press, 46–66.
1999. “Should Blacks Represent Blacks and Women Represent Women? A
Contingent ‘Yes.’” Journal of Politics 61 (3): 628–57.
2003. “Rethinking Representation.” American Political Science Review 97 (4):
515–28.
Markell, Patchen. 2003. Bound by Recognition. Princeton, NJ: Princeton Univer-
sity Press.
2006. “The Rule of the People: Arendt, Archē, and Democracy.” American
Political Science Review 100 (1): 1–14.
Massey, Douglas and Nancy Denton. 1993. American Apartheid: Segregation and
the Making of the Underclass. Cambridge, MA: Harvard University Press.
Mill, John Stuart. 1825. Essays on Government, Jurisprudence, Liberty of the Press,
and Law of Nations. New York: Kelley.
1975. Considerations on Representative Government, in John Stuart Mill: Three
Essays. Oxford: Oxford University Press.
Mouffe, Chantal. 2000. The Democratic Paradox. London: Verso.
Pettit, Phillip. 1997. Republicanism: A Theory of Freedom and Government. Oxford:
Oxford University Press.
Phillips, Anne. 1995. The Politics of Presence. Oxford: Clarendon Press.
Pitkin, Hanna Fenichel. 1967. The Concept of Representation. Berkeley: University
of California Press.
1981. “Justice: On Relating Private and Public.” Political Theory 9 (3): 340–1.
Plotke, David. 1997. “Representation is Democracy.” Constellations 4 (1): 19–34.
Przeworski, Adam, Carol Stokes, and Bernard Manin. 1999. Democracy, Account-
ability, and Representation. Cambridge: Cambridge University Press.
Rawls, John. 1993. Political Liberalism. New York: Columbia University Press.
Rehfield, Andrew. 2006. “Towards a General Theory of Political Representa-
tion.” Journal of Politics, 68 (1): 1–21.
Squires, Gregory. 1994. “Community Reinvestment: The Privatization of Fair
Lending Law Enforcement,” in Residential Apartheid: The American Legacy,
ed. Robert Bullard, J. Eugene Grigsby, III, and Charles Lee. Los Angeles,
CA: CAAS, 257–86.
 On representation and democratic legitimacy 135

Sugrue, Thomas. 1996. The Origins of the Urban Crisis: Race and Inequality in
Postwar Detroit. Princeton, NJ: Princeton University Press.
Teaford, Jon. 1990. The Rough Road to Renaissance: Urban Revitalization in Amer-
ica, 1940–1985. Baltimore, MD: Johns Hopkins University Press.
2000. “Urban Renewal and Its Aftermath.” Housing Policy Debate 11 (2): 443–
65.
Tully, James. 1995. Strange Multiplicity: Constitutionalism in an Age of Diversity.
New York: Cambridge University Press.
1999. “The Agonic Freedom of Citizens.” Economy and Society 28 (2): 161–82.
2000. “Struggles over Recognition and Distribution.” Constellations 7 (4): 469–
82.
2002. “The Unfreedom of the Moderns in Comparison to their Ideals of
Constitutional Democracy.” Modern Law Review 65 (March): 204–28.
US Bureau of the Census. 2002. “Preliminary Report No. 1: The
2002 Census of Governments.” Available at: http://ftp2.census.gov/govs/
cog/2002COGprelim_report.pdf (accessed June 8, 2009).
Williams, Melissa. 1998. Voice, Trust, and Memory: Marginalized Groups and the
Failings of Liberal Representation. Princeton, NJ: Princeton University Press.
Wilson, William Julius. 1987. The Truly Disadvantaged: The Inner City, the Under-
class, and Public Policy. Chicago: University of Chicago Press.
Wolin, Sheldon. 1996. “Fugitive Democracy,” in Democracy and Difference, ed.
Seyla Benhabib. Princeton, NJ: Princeton University Press, 31–455.
Young, Iris Marion. 1990. Justice and the Politics of Difference. Princeton, NJ:
Princeton University Press.
2000. Inclusion and Democracy. Oxford: Oxford University Press.
Part III

Representation and inherited injustice

6 Critical liberalism

Courtney Jung

Representative democracies struggle with the demands of ethnic minori-

ties. In the last decade of the twentieth century, these struggles were
exacerbated by the rise of new cultural claims and claimants. Democrati-
zation engendered discussions about citizenship and membership, which
raised questions of national identity and belonging. The Soviet Union
disintegrated as its component units exercised their rights to national self-
determination; the European Union attempted to develop new standards
for the protection of minorities; and indigenous rights were enshrined in
international law. The demands of ethnic minorities punctuate the polit-
ical landscape in rural and urban areas, in developed and developing
countries, in the East and in the West, reshaping the terms of political
debate, and raising new points of political contestation.
Liberal democrats worry about this political turn. They worry about
the extent to which democracies can accommodate the competing claims
of ethnic groups, and they worry about their obligations to do so. They
are concerned in particular because they believe that demands for bilin-
gual education, autonomy, and cultural respect represent fundamental
demands for the recognition of human identity, and that they therefore
engage a deeper level of commitment than other claims. As such they
impose greater obligations on the democratic state. They also pose a
greater threat, as democratic institutions struggle to process and accom-
modate the deeply held but incompatible cultural commitments of their
citizens.
Critical liberalism responds to these concerns with a theory of obliga-
tion that is rooted in the structural origins of groups rather than in the
cultural difference of groups. States are obligated to engage and respond
to the claims of indigenous peoples, other ethnic minorities, African-
Americans, women, immigrants, and so on, not because of who they are,
or because they may command human attachment, but as a result of the
historical record of exclusion and selective inclusion through which they
have been constituted. The obligation of states lies in the fact that states
themselves have forged social groups, and the identities they anchor, by
139
140 Courtney Jung

using such markers as cultural practices, alleged phenotypical traits, bio-

logical sex, property ownership, and wealth to organize access to power
and delimit the boundaries of citizenship.
In the first section of the chapter I develop a structural theory of
obligation. In the second and third sections, I draw out the norma-
tive implications of such a theory. In the second section I argue that a
constructivist theory of identity formation demands that the normative
standing of particular claims be established through the framework of
structural injustice rather than cultural difference. In the third section I
introduce the category of membership rights – rights that recognize that
social groups are constructed and nevertheless acknowledge that they
often play a central role in determining life chances. I draw examples
from the politics of indigenous rights to illustrate some of my arguments.

I. A structural theory of obligation

Normative political theorists of democracy and multicultural citizen-
ship have largely ignored the political and strategic context in which
cultural claims are formulated and advanced. Such theorists have been
content, for the most part, to begin from the premise that people feel
strong attachments to their cultural groups (Rawls 1985, 1993; Dworkin
1978; Kymlicka 1995), that human identity is in fact constituted by cul-
tural group membership (Sandel 1982; Galston 1995), and that human
authenticity arises in dialogue with members of one’s own cultural group
(Taylor 1994). They assume that demands for bilingual education, local-
level self-government, territorial autonomy, and public support for the
cultural practices of minority groups are the political expression of a uni-
versal human need for the affirmation and preservation of the cultural
groups to which we belong.1
Will Kymlicka best expresses the general consensus that there is no
real point in trying to explain the cultural attachments of individuals. As
he says, the bond between individuals and their cultural groups is simply
“a fact . . . whose origins lie deep in the human condition, tied up with
the way humans as cultural creatures need to make sense of their world,
and that a full explanation would involve aspects of psychology, sociology,
linguistics, the philosophy of mind, and even neurology” (Kymlicka 1995:
90). Such attachments, Kymlicka and others insist, importantly shape
human identity. By making an appeal to neurology, Kymlicka suggests
they may even be biologically hard-wired. The challenge for democratic

1 Seyla Benhabib’s The Claims of Culture (2002) is a notable exception to this trend. She
focuses attention instead on the ways in which culture is deployed for political purposes.
 Critical liberalism 141

theorists, he and others believe, is not where they come from, but how
to deal with them.
Against this backdrop, claims for recognition take on the air of a reflex-
ive reaction against the homogenizing threats of modernity and the fre-
quency with which supposedly insular and intact cultural units come into
contact with one another, with the penetrating reach of the modern state,
and with neo-liberal economic policies (Yashar 1998, 2005). Demands
for cultural preservation seem to stem from a protective instinct in defense
of the familiar, local, and particular against the threat of Nike, Eminem,
and English.
This initial assumption is not neutral, and it has played a crucial
role in framing the terms of debate over what democratic states owe
to ethnic minorities. The contemporary normative discussion over mul-
ticultural citizenship is divided between those who insist that democracy
requires the privatization of cultural commitments, and those who believe
that democracies have an obligation to protect, or preserve, the cultural
groups in which their citizens have membership. Both impulses, to priva-
tize or to protect the ethnic group – to insulate democratic politics from
culture, or to insulate culture from democratic politics – are rooted in
assumptions that make a fetish of culture.
Privatizers worry that religious and cultural commitments anchor what
is most fundamental in human identity, and that such convictions are
therefore not amenable to the normal give-and-take of negotiation and
moderation that representative democracy requires. Claims that rest on
belief or tradition threaten to tear apart the public sphere. Because they
cannot be moderated, they should instead be excluded from public delib-
eration, leaving citizens free to express, preserve, and exercise their preex-
isting religious and cultural commitments in private. Democratic society
will only be maintained if it is protected from debate between competing
conceptions of the whole truth (Rawls 1971, 1985; Macedo 1995). If the
public sphere is to remain neutral, and equally available to everyone, it
must be limited to reasoning that everyone could understand and agree
to. Typically, privatizers invoke groups that are unsympathetic, such as
Christian fundamentalists, Nazis, or skinheads, and invite us to imag-
ine what would happen if their viewpoints were to achieve dominance
through the democratic process (Macedo 1995).2
Protectionists agree that religion and culture represent fundamental
human commitments, but they argue that this is a reason to protect,
not to privatize, the collective units in which citizens hold membership.
They argue for the recognition of group rights, though they sometimes

2 Stephen Macedo (1995) writes about the Christian right, for example.
142 Courtney Jung

disagree over which groups and which rights the state should pro-
tect. They maintain that respect for the individual entails respect for
his culture, and that cultural and religious minorities should therefore
be protected from majoritarian politics that threaten the capacity of a
culture to reproduce itself (Taylor 1994). They argue that democracy
requires respect for difference (Galston 1995), and that individual free-
dom depends on access to one’s own societal culture (Kymlicka 1995).
Protectionists tend to invoke sympathetic groups, such as indigenous
people and the Amish, and to suggest that the laissez-faire character of
privatizing liberalism threatens to extinguish the traditions and lifestyles
of these ancient and vulnerable communities to the detriment of all
mankind.
Theorists on both sides make a compelling case by invoking fear
of tyranny of the majority. Privatizers raise the possibility that hate-
mongering skinheads might take over local governments in some parts
of the United States, which is a genuine concern. Protectionists invoke
the threat to the survival of indigenous languages and traditions, which
is palpable almost everywhere. Each of them is correct in thinking that
the procedural character of democracy may threaten minority interests
and empower dangerous majorities. Their instinct to preclude such out-
comes by limiting democracy has nevertheless left them vulnerable to
compelling criticism.
By denying cultural groups the grounds to contest politics, privatization
has the effect of naturalizing the cultural status quo, leaving little viable
choice to minorities but assimilation. And it is not only that others are
forced to assimilate to liberalism, as Macedo claims when he argues the
case for liberalism with a spine (Macedo 1995). They are also forced to
assimilate to the dominant language, religion, culture, etc., all of which,
absent contestation, will be naturalized as part of the backdrop of social
exchange. As critics contend, privatizing cultural commitments really
does limit the capacity of minority groups to navigate and prolong many
of their own cultural practices (Taylor 1994; Kymlicka 1995).
On the other hand, by arguing for the right to collective self-
government and cultural autonomy, which is also an exemption from
democratic politics, protectionists have come under fire for depriving
group members of the political means to contest the norms and bound-
aries of their own cultural and religious groups. Unless we can assume
the complete internal coherence of the cultural group, or that the cultural
group is literally an organic unit, collective rights that draw their legiti-
macy from cultural norms alone run the risk of violating individual rights.
Recognition is always partial, privileging those who hold power within
 Critical liberalism 143

the cultural group (Okin 1999), and freezing the cultural boundaries and
practices that are salient at the moment of recognition (Appiah 1994).
Ironically, this often gives the state more leverage to adjudicate what
counts as tradition, insisting, for example, that only pre-contact prac-
tices warrant special exemptions, or that only those people who continue
to live on their ancestral lands qualify as genuinely indigenous.
The deepest criticisms of each side stem from the paradox that, despite
their different commitments, both privatizers and protectionists sup-
port the exclusion, or exemption, of cultural groups from democratic
politics – either to protect the supposed neutrality of the state, or to
protect the supposed integrity of the cultural group (Jung 2001). The
common impulse to exclude derives in turn from the fact that, whether
they recommend privatization or protection, normative theorists employ
the same assumptions regarding the crucial role of culture in constituting
individual identity. Culture should be protected or privatized precisely,
they believe, because demands for cultural recognition arise from a uni-
versal and transhistorical human need for cultural group recognition.
Most people feel a strong attachment to their cultural groups and prac-
tices, they say, and such attachments have normative standing. From
such attachments flow obligations.
Nevertheless, there is a prior level of politics that has been obscured
by a fixation on the fact of cultural difference. The debate over minority
rights is often conducted in a register that leaves many fundamental
questions of power, access, politics, and exclusion largely off the table. It
does so mainly because it has neglected questions about where cultural
claims come from, and why they are salient in contemporary political
discourse. Although many theorists concede that ethnic identities are
constructed, they draw no analytical connection between the origins of
demands for cultural group recognition, on the one hand, and the just
disposition of such demands, on the other (Barry 2001; Carens 2000;
Dworkin 1978; Galston 1995; Kymlicka 1995; Macedo 1995; Rawls
1971, 1985, 1993; Taylor 1994). Theorists who have actually confronted
this question insist that the fact that social groups are constructed has
no bearing on how they are managed, or what they are owed, by liberal
democratic institutions (Walzer 1983: 31; Kymlicka 1995).
Such theorists have attempted to bypass the significance of construc-
tivism for normative theory by saying “yes, ethnic groups may be con-
structed, but regardless of where they come from, or how they change,
human beings are fundamentally attached to, and indeed constituted by,
the (constructed) cultural groups they are born into.” Indeed this is a
reasonable response to the “weak-constructivist” claim that a group’s
144 Courtney Jung

culture is shifting and subject to reinvention and that traditions are

often transformed by such outside influences as colonial and postcolonial
governments. It is certainly true, and has been amply shown, that many
of the traditions that are portrayed as ancient are in fact new. But this cri-
tique issues a feeble challenge to the existing corpus of literature on mul-
ticulturalism and democracy, which takes no position on where culture
comes from, focusing instead on the extent to which human well-being
depends on access to one’s own societal culture, regardless of where such
culture comes from or how it may change.
The “strong-constructivist” argument is different, however, and offers
grounds for a fundamental reconsideration of the line of thinking that
has driven the debate over multiculturalism and democracy. The strong-
constructivist argument holds that it is human attachment itself that is
constructed. It is not that people do not have fundamental attachments –
of course they do. Instead, as Akeel Bilgrami has argued, religious or
cultural identities are not fundamental because they are the spontaneous
expression of “the subject’s irreducible interest in the definition of itself”
(Bilgrami 1992: 831). They are fundamental because of the way (or to the
extent) they have developed in particular historical and cultural contexts
(ibid. 1992: 831, 833).
The likelihood that cultural practices and traditions will rise to the
level of “ethnicity,” generating boundaries with social significance and
the weight to invest a distinction between “us” and “them” with authority
and commonsense meaning, is contingent. The extent to which human
beings identify themselves as members of ethnic groups depends on the
extent to which cultural practices have been vested with public and social
significance. Whether individuals perceive their identities to be most
fundamentally mediated by ethnicity, or by gender, race, or class, or
even by such personal attributes as intelligence, profession, or marital
status, reflects structural conditions rather than law-like psychological
traits. Like race, class, and gender, culture develops political resonance
when it has been used as a marker of selective inclusion and exclusion by
states organizing access to power.
One implication of the strong-constructivist argument for liberal
democratic theory is that it lodges responsibility for the relative salience
of cultural group membership in the state. It draws a direct connec-
tion between how groups are formed and what they are owed. Move-
ments form around gender, sexuality, race, and culture because these
social categories have regulated political access and the organization of
power, not because they express people’s deepest sense of themselves.
This does not make them any less potentially real or valuable to human
 Critical liberalism 145

experience, but it does mean that the ostensible reality, importance, and
value of culture depend not on cultural practices alone but on the social
and structural forces that lend them weight in our lives.3
Tracing the origins of political salience is relevant not only as an exer-
cise in exposing the way power operates, but also as a way of establishing
the historical responsibility of states. The real normative torque of cul-
tural politics is anchored in the structural location that certain groups
occupy as a result of historical discrimination and exclusion. Groups
that are constituted through exclusion and oppression, that have come to
shape political identity because their members have suffered in common,
and that occupy a position of structural inferiority, have grounds for
redress that lie in their own particular history. The legitimacy and scope
of ethnic group claims lie in the structural location of the group, not
in differences of tradition and practice, and not in the extent of human
attachment to such traditions.
The appropriate question for normative political theory therefore is
not “How can democratic governments accommodate the preexisting
and deeply held cultural commitments of their citizens?” but instead
“How should democratic institutions process the political claims that
arise to protect and contest the exclusions and inclusions set in place
by the modern state itself?” Taking seriously the historically and politi-
cally constituted character of ethnicity leads neither to privatization nor
to protection but to a transformative engagement that sustains a more
critical conception of liberalism.

II. Structural injustice

Cultural norms and practices generate ethnic boundaries and sentiment,
the capacity to bear attachment, and relevance as political identities
through the way that such markers as language, tradition, alleged phe-
notypical traits, and cultural norms have been used by states to mark
the boundaries of citizenship. Human beings identify themselves as
members of ethnic groups, and perceive cultural practices as relevant

3 A structural account of identity formation does not erase the possibility of agency. Human
beings rework, challenge, and revalue their categories of membership, shaping the mean-
ing and contours of their own identities. Nevertheless, the categories of membership they
revalue are those that have been publicly marked, and the project of cultural retrieval does
not take place in isolation – culture is deployed, reworked, and invoked in engagement
with politics. The fact that human beings are also at work in the construction of their
own identities does not diminish the relevance of focusing on the structural dimension
of identity to establish the normative and political standing of oppressed groups.
146 Courtney Jung

markers of identity, to the extent that culture has been used to organize
society. Ethnicity is the deeply contextualized result of a particular histor-
ical experience, not a universal category of human political organization.
The most direct implication of constructivist theories of identity for-
mation for normative frameworks of multiculturalism and democracy is
that universal solutions are mismatched to the task of adjudicating the
claims of cultures and the responsibilities of democratic states. This fol-
lows directly from the insight that ethnicity arises as a result of structural
configurations that are particular to different societies and historical expe-
riences. What ethnic groups are owed by liberal democratic governments
depends on the historical conditions that have given rise to particular
ethnic groups.
As a result, claims should be judged on the basis of the existence of
historic and ongoing injustice, and not on the degree to which a group
is culturally distinct. Jeff Spinner-Halev has called this type of injustice,
which has roots in the past but persists to the present, enduring injus-
tice. Enduring injustice, he says, has both historical and contemporary
components (2007: 3, 7). Spinner-Halev argues that injustices endure
because they are beyond the scope of liberal conceptions of justice –
they cannot be remedied through individual rights or a fair system of
distributive justice (ibid.: 7).
The structural injustice framework focuses attention instead on the
causes of historical and ongoing injustice. Structural injustices are those
with systemic roots that persistently, over the course of generations, struc-
ture access and power to the disadvantage of people who have some trait,
such as biological sex or skin color, in common, with the effect of gen-
erating a sense of common identity. They are woven into the fabric of a
society’s political institutions and laws, through its norms and mores, and
into the very texture of everyday life. Structural injustices persist across
generations because they have structural systemic roots in the history of
liberal democracy.
The cause of such injustice is therefore the same as its normative force.
Injustices that flow from laws and policies put in place by the state – struc-
tural injustices – offer particular normative standing to those who have
suffered from them, and whose “groups” have been shaped by them. On
this account, persistent injustice is caused by the failure of actually exist-
ing liberal democratic history, rather than the failure of liberal democratic
theory, as Spinner-Halev contends. A critical conception of liberalism,
which foregrounds this history, can offer a remedy.
The fact that constructivism forces normative theory to turn for
answers to the particular condition of structural injustice (or privilege)
that is the condition of the group’s existence has a number of corollary
 Critical liberalism 147

implications. There are four reasons that the claims of ethnic groups are
more appropriately legitimated through the structural injustice frame-
work than through the cultural difference framework.
First, by resting group claims in particular historical experiences, criti-
cal liberalism is able to cut through orthodox group distinctions, offering
a theory of multiculturalism and democracy that has something to say
about race, gender, and class. It allows normative theorists to use the
same standard – the particular set of structural relations that has consti-
tuted the group – to contemplate the moral and political standing of all
groups. Ethnic groups need not be treated in a category apart from race
or class groups.
This is useful because it sidesteps the classificatory reification that con-
fuses our thinking on these questions. There is a poor fit between actual
groups, such as indigenous peoples and African-Americans, and the cat-
egories that are used to determine how they should be treated by liberal
democratic institutions, such as race, class, and ethnicity. Over the last
five hundred years, indigenous people in Mexico have been reclassified
by race, class, and ethnicity. Each designation located them in a distinct
policy space, with important implications for their experience of political
exclusion and incorporation, and for the scope and substance of their
political claims. But the fact that a single population has been serially
“raced,” “classed,” and “ethnicized” should alert us to the ways in which
these categories are deployed by states, in line with dominant ideolo-
gies of human classification, to render their populations legible (Jung
2008). Such categories, which have been used by states to classify their
populations, most obviously through censuses, are reified when they are
redeployed to establish the proper standing of groups and to evaluate the
legitimacy of the claims they make.
Second, following the same logic in the other direction, critical liberal-
ism is also able to distinguish among cultural groups, affording them dif-
ferent standing. Insisting that the legitimacy of ethnic group claims rests
on structural conditions and historical obligations – and not on cultural
difference or attachment – allows critical liberalism to parse similarities
and differences among groups in a way that is more satisfying than the
recommendations that multiculturalists have been backed into by their
commitment to cultural distinctiveness. Part of the reason that theoreti-
cal debates over cultural group rights sometimes seem lodged at too high
a level of abstraction is that, by locating rights to self-determination and
territorial autonomy in cultural difference, multiculturalists give the same
grounds to Kurds and other victims of genocide as they do to Serbs or
to Afrikaners. All groups can claim cultural difference, and such claims
give all groups the same moral standing.
148 Courtney Jung

Yet it somehow does not ring true that all groups that stake a claim to
cultural distinctiveness are equally legitimate, or should have the same
normative standing. This is a point that some multiculturalists have rec-
ognized, but the logic of their own arguments gives them very little room
to distinguish among cultural groups. Will Kymlicka, for instance, makes
a case for according different rights to First Nations, national minorities,
and immigrants, based primarily on the degree to which they maintain
intact societal cultures and whether or not they have chosen to live in
their current location (Kymlicka 1995).
Yet this argument follows on from his main point, which is that liberal
democracies have a responsibility to protect the groups he calls societal
cultures because individual freedom, and the capacity to make mean-
ingful choices, relies on access to one’s own culture. This is a powerful
claim, and it is a universal claim. The subsequent distinction Kymlicka
draws among cultural groups is counterintuitive against this backdrop. If
freedom – which he says is the primary commitment of a liberal demo-
cratic society – depends on maintaining cultural group membership, how
can such societies accord greater cultural protection to some groups than
others? If basic freedom itself is at stake, liberal democracies cannot be
governed by pragmatic decisions that protect the individual freedom of
First Nations people but not of immigrant minorities.
I suspect instead that our instinct to try to distinguish among claims,
and to accord different status to Afrikaner demands for a Volkstaat than
we do to Kurdish demands for a homeland, depends not on the degree
to which each group maintains an intact societal culture (the Afrikaner
language and culture is both distinctive and robust), but on the degree
to which the group has been constituted through discrimination and
oppression, or privilege. Viewed through the prism of structural injus-
tice, groups in structurally superior positions, which stand in structural
relations of power over other groups, have weaker standing, and com-
mand less normative legitimacy, than groups in structurally inferior posi-
tions, even when both can claim cultural distinctiveness. Although the
distinction Kymlicka draws between groups seems surprising from the
perspective of cultural distinctiveness, it makes sense through a structural
injustice framework.
The third way in which the appeal to structural injustice advances pol-
itics beyond an appeal to communal integrity is that it expands the arena
of the politics of recognition. Ethnic minorities do not only demand cul-
tural protection and bilingual education. They have also been subjected
to a raft of injustices that cut in the other direction, namely inadequate
integration, racism, exclusion from full citizenship, insufficient access to
healthcare, welfare, and social service benefits, and poor or non-existent
 Critical liberalism 149

educational opportunities. With this litany of structural injustices behind

them, they can make claims for representation, redistribution, and resti-
tution, and not only for recognition.
Furthermore, from the perspective of political organization, cultural
groups may make a strategic error when they anchor their political claims
in cultural difference. By locating culture at the center of political voice,
indigenous peoples and others who would mobilize ethnic difference risk
being backed into demands for cultural protection and the preservation
of tradition, which do not issue a fundamental challenge to many of the
structural relations that are the condition of their disadvantaged posi-
tion. The claims of culture issue a feeble challenge to the neo-liberal
economic order for instance, or to the structural location of indigenous
populations at the bottom of that order. Some states may even be eager to
confer self-government on their indigenous populations, using territorial
autonomy to abjure state responsibility for providing development and
social services.
Finally, this type of claim centers the issue of the historical recollec-
tion of injustice, which is important to people whose identity has been
constituted by such injustice. As Waldron says, “it is a well known char-
acteristic of great injustice that those who suffer it go to their deaths with
the conviction that these things must not be forgotten” (Waldron 1992:
5). The descendants of those who suffer such injustice are normally just
as committed to keeping the memory alive, insisting that the fact that
such injustice occurred be publicly acknowledged and made part of the
public record.
The political significance of memory should not be underestimated.
It is widely acknowledged that the indigenous rights movement in Latin
America was precipitated through mobilization against planned celebra-
tions of the five-hundredth anniversary of the “Discovery of the Amer-
icas” in 1992. Indigenous people throughout Latin America rose up
to insist that their own governments, and the government of Spain,
acknowledge the violence and oppression that the so-called discovery
had entailed. They defiantly renamed the anniversary “Five Hundred
Years of Oppression.”
As Waldron goes on to say, “reparations may symbolize a society’s
undertaking not to forget or deny that a particular injustice took place,
and to respect and help sustain a dignified sense of identity-in-memory
for the people affected” (ibid.: 6). To forget is considered immoral, a sign
of disrespect toward those who suffered and died, and of contempt toward
the present community. A society offers respect to the community by
acknowledging, mourning, and seeking to redress, the historic injustices
that have been perpetrated against it.
150 Courtney Jung

Beyond maintaining “identity-in-memory” for those who suffered, the

retrieval of injustice establishes the present identity of groups against
whom harm was historically perpetrated. In international and domestic
forums alike, indigenous peoples establish their standing through a cat-
alogue of the harms they have suffered. As Melissa Williams has argued,
“collective memory may surface as a response to discriminatory treat-
ment, creating a solidarity and group self-consciousness that would not
otherwise exist” (1998: 186).
Moreover, as Williams shows, the framework of historic and ongoing
injustice offers “a schema for identifying groups that have a strong claim
to recognition within political institutions” (ibid.: 176). As she says,
“the claims of marginalized ascriptive groups to special representation in
politics are legitimized by a clear and strong connection between present
inequality and the kinds of past discrimination that were sanctioned by
dominant social groups and often enforced by the state” (ibid.: 177).
There are both objective and subjective criteria by which such claims can
be measured. The objective criterion is the historical record: “evidence
provided by law, documents, and other generally accepted sources that
the discrimination actually occurred.” The subjective criterion, however,
is provided by memory – “the meaning the past has for members of
those groups who were the targets of discrimination and oppression”
(ibid.: 177).
Constructivist theories of identity formation generate the deep nor-
mative implication that developing a universal theory of multicultural
citizenship is a mistaken enterprise. In every case, how democratic insti-
tutions should process the claims of culture is a political matter, rooted in
a particular logic and history of political incorporation and exclusion. It
is a matter that is therefore more appropriately, and fruitfully, addressed
through the framework of compensatory justice than through the frame-
work of communal attachment, identity, and cultural group difference.
The structural injustice framework seeks to transform the future by offer-
ing restitution for a past that has actually constituted the social groups
that play a principal role in shaping life chances.4
The framework of structural injustice establishes a direct line of respon-
sibility between the actions and the obligations of states. States have obli-
gations toward oppressed populations, not because they have distinctive
cultural practices, but because the state has discriminated against them

4 In his excellent book Postcolonial Liberalism, Duncan Ivison also makes an argument for
considering indigenous claims through the prism of historic and ongoing injustice. He
nevertheless comes to this argument from a completely different route. Melissa Williams
also includes the concept of historic injustice as part of her argument for legislative
representation for historically disadvantaged groups (Ivison 2002; Williams 1998).
 Critical liberalism 151

on the basis of those distinct cultural practices. If the state has not done
so, it can reasonably claim that the fact that a group has distinct cultural
practices generates no responsibility on the part of the state. But since
modern states have been built on conceptions of nationhood that often
invoke cultural (and racial) markers to legitimate state boundaries, most
contemporary states cannot claim that they have not discriminated on
the basis of cultural difference, and the structural injustice framework
therefore establishes a clear contemporary line of obligation that links
“ethnic” politics, including indigenous politics, to the Westphalian (and
postcolonial) state system. In so doing, it attempts to sidestep the nor-
mative instabilities that are produced by claims to standing that rest on
“identity.”

III. Membership rights

Membership rights flow from a constructivist account of identity forma-
tion along three streams. First, through membership rights, states offer
redress for structural injustices that have harmed individuals because they
are members of low-status groups. Second, membership rights stretch
“identity” politics beyond culture, to include demands for redistribu-
tion and representation as well as recognition. Third, membership rights
offer a framework for forging oppositional and independent political chal-
lenges that may be able to break out of the narrow political space offered
by culture and identity. Membership rights ground a deeper and poten-
tially more challenging engagement between the state and its citizens
than either individual or collective rights.
The category of membership rights tries to respond to the question
of what kind of rights are compatible with the fact that groups are
constructed and, more specifically, with the fact that they have been
constructed precisely through the unequal allocation of rights, represen-
tation, and citizenship. Neither individual nor collective rights respond
appropriately to the fact that states have played an important role in
constructing contemporary boundaries of ethnic group membership by
using cultural practices, language, and tradition to mark the boundaries
of exclusion and selective inclusion. Membership rights are differentiated
individual rights that take seriously both the constructed character and
the social weight of groups.
The liberal commitment to individual rights is rooted in a universal
framework of equal treatment. For many liberals, justice and equality are
anchored in a conception of neutrality that requires that every individual
be treated equally. As Joseph Carens has pointed out, this is an impor-
tant ideal, and one which “provides grounds for criticizing policies and
152 Courtney Jung

practices that exclude or marginalize distinct groups of citizens” (Carens

2000: 178).
But, given the legacy of differentiation and discrimination that struc-
tures all contemporary societies in different ways, equal treatment will
not be sufficient to provide all individuals with equal opportunities. An
undifferentiated conception of citizenship treats everyone as if they are in
fact equal, whereas states themselves have created differences of access,
recognition, and prestige, as well as wealth, education, and services, with
lasting implications. Given the central role they play in determining life
chances, the constructed character of groups does not imply that they
have no social or political character, or that they carry no normative
weight. Individual rights, which treat everyone equally, respond inade-
quately to injustices that have been suffered as a result of putative group
membership.
Nevertheless, against the backdrop of constructed histories, the nor-
mative weight of ethnic groups does not extend to collective rights to
“self”-government that rely on organic conceptions of the group. It is no
coincidence that demands for collective rights – rights that accrue to the
group as a whole rather than to individual members of the group – rest
on essentialized notions of group history, boundaries, and identity. They
must, because there must be an identifiable unit to which (or whom) a
right accrues. As Charles Taylor has pointed out, the Herderian notion
of a cultural group, as a Volk (singular) that should be true to itself (sin-
gular) underpins the very idea of a collective right (Taylor 1994: 31).
If a cultural group is as likely to be internally divided, to have shifting
boundaries, and to be of varying levels of importance to different mem-
bers of the group, then it follows logically that a collective right will be
captured by some powerful segment of the group – that is, by individual
members of the group (Okin 1999). Such groups are not organic units,
and they need therefore to justify their authority through the principles of
representation and human rights, in the same way that states are obliged
to establish the legitimacy of their rule. Rights that treat a collective as a
single unit are logically incoherent in the face of the fluid, constructed,
and multiple dimensions of the collective units in which people claim
membership.
Membership rights try therefore to steer a path between individual
and collective conceptions of rights in a way that is consistent with the
constructed character of groups. Differentiated individual rights are sen-
sitive to the structural conditions that constitute individual human beings
and that afford them differential access to most of the things that matter
in contemporary societies. When the rights of some citizens have been
distorted and degraded by discrimination and oppression directed at
 Critical liberalism 153

members of particular groups, such members can legitimately demand

membership rights that offer all of the normal entitlements of citizenship
along with some distinctive entitlements.5 Such rights retain the non-
negotiable concern of liberalism for the well-being of individuals, while
nevertheless acknowledging that individuals are embedded in groups or
categories that have (negatively) affected their life chances, but also,
importantly, offer opportunities for political agency.
Membership rights nevertheless endorse a conception of differenti-
ated individual rights that is justified on the basis of structural rather
than cultural difference. Some structural differences have culture as their
organizing principle – people suffer discrimination and exclusion because
they have different cultural practices. But others suffer because they are
Black, or poor, or gay, or women. The principle behind membership
rights is that the liberal democratic state owes differentiated rights that
are responsive to the demands of members of each of these groups for
the same reason – because the state has played a central role in ordering
society in such a way that people who have the same phenotypical traits,
biological sex, sexual orientation, or cultural practices might perceive
such markers as a binding identity.
In so doing, membership rights attempt to move beyond the contro-
versy over individual versus collective rights that has fixed debates over
multiculturalism and democracy and narrowed the scope of such poli-
tics. The attempt to reconcile concern for the well-being of individuals
with a desire to be responsive to the demands of groups has dominated
the normative theoretical agenda. But this controversy has limited our
attention to one particular category of rights, namely third-generation
cultural rights, obscuring the demands that many groups also make for
first- and second-generation rights, and crowding out a broader analysis
of how such rights work with third-generation rights.
Membership rights bundle civil and political, social and economic,
and cultural rights, refocusing attention on the premise, which is taken
for granted in the field of human rights law, that the three generations
of rights are interdependent and mutually supportive (Scott and Mack-
lem 1992). A focus on collective/cultural rights, to the exclusion of civil

5 The idea of differentiated individual rights is well entrenched, both in actually existing
constitutions and in political theory. Joseph Carens (2000) explains, for example, that the
concept of “citizen-plus” status was coined in the 1966 Hawthorn–Tremblay Report that
sought to reconcile Canadian citizenship with Indian status. As he says, “Clearly the goal
here was to achieve some sort of evenhanded justice, granting distinctive entitlements
out of a sense that this was a fair way to respond to the history and circumstances of
Indians in Canada” (ibid.: 187). What I argue here is that it is this conception of rights
that is conceptually matched to a constructivist account of identity and group formation.
154 Courtney Jung

and political and social and economic rights, will actually undermine
the possibility of securing a robust cultural framework. In the face of
poverty, discrimination, and hopelessness, culture can disintegrate, fur-
ther marginalizing people who have been constituted by exclusion and by
their inability to shape the terms of their inclusion. Such groups should
not have to make a choice between demands for recognition, redistribu-
tion, and representation (Fraser 2000).
Unlike a cultural difference framework, which legitimates the demands
of ethnic groups for such cultural rights as language preservation and cul-
tural autonomy, a structural injustice framework justifies extending the
political reach of so-called ethnic groups to demands for redistribution
and participation as well as some forms of cultural protection. A struc-
tural injustice framework therefore covers more political ground than a
cultural difference framework. Using structural injustice, activists can
insist on rights to political representation, the protection of civil liber-
ties, individual freedoms, protection from discrimination, and social and
economic welfare, that may not be covered by demands for cultural pro-
tection and recognition, or even by autonomy and self-government.
Membership rights therefore have a strategic effect on the scope of
opposition politics. Such rights expose groups constituted through struc-
tural exclusion to alliances among groups, whereas, as many people have
argued, some forms of a politics of identity spell the demise of alliances
across groups and the twilight of common dreams (Gitlin 1995). Because
they do not derive from cultural specificity, membership rights have a
universal logic, notwithstanding the fact that they accord special rights
to members of some groups and not others. The grounds of struc-
tural injustice on which they tender such rights establish the basis for
alliances among all sorts of groups (women, indigenous peoples, African-
Americans) that have suffered such injustice. Groups that anchor their
demands in cultural difference may instead turn inward, lacking grounds
to make common cause with potential allies.
Logically, the proposition that membership rights should include first-
and second-generation rights, and not only third-generation rights, fol-
lows from the account advanced here of the structural foundations of
group identity. That is, the blinkered fixation with collective and cul-
tural rights that has focused normative debates on multiculturalism and
democracy, as well as many activists who mobilize so-called identity
groups, follows from the assumption that what relevantly distinguishes
such groups is cultural difference. If instead the legitimating principle
of group standing is structural injustice, the political demands of groups
may extend beyond culture.
 Critical liberalism 155

Finally, rights may also play an important role in opening up the polit-
ical space of oppositional politics. Where such rights are enshrined in
constitutions, they recast oppositional politics as a demand that the state
fulfill its own promise. As such, rights can structure politics as an imma-
nent critique, harnessing the prevailing de jure ideology against the de
facto inequalities it sustains. Rights offer a way into the political realm
because they expose the gap between promise and fulfillment, offering a
powerful discursive strategy and point of intervention to activists who are
seeking to engage the state. Through rights, activists politicize the con-
tradiction between the dominant ideology and their reality (Crenshaw
1995: 111).
Individual rights have nevertheless been criticized as atomizing. Such
rights may divert politics into the courtroom, and sacrifice collective
political organization to individual legal strategies (Tushnet 1984; Gabel
and Kennedy 1984). The focus of attention on the individual may under-
mine collective solidarity, and encourage the pursuit of private happiness
over the pursuit of the common good.
Collective rights may also divert attention from the public political
sphere, redrawing the boundaries of the polity around minority groups,
and directing attention inward. One of the goals of collective rights is to
establish the priority of subgroup boundaries, and to reduce the points
of interaction and dependence between group members and the national
society and institutions that threaten their culture and have rendered
their subordinate status. Collective rights are not designed primarily
to foster transformative engagement between group members and state
institutions.
Membership rights may engender such engagement. Such rights offer
standing to excluded peoples, generating incentives for collective political
solidarity. Indigenous activists, for example, played an important role in
developing and entrenching the concept of indigenous rights in interna-
tional law. But indigenous rights – which have the scope of membership
rights rather than collective rights – have in turn fostered the forma-
tion of an international indigenous rights movement. Whereas individual
rights may fail to motivate collective action, membership rights furnish a
framework that is distinctly political.
At the same time as they generate incentives for collective solidarity,
membership rights should also sustain a framework for creative reappro-
priations of identity, and for challenges to state-imposed categories of
membership. Membership rights should have more room than collec-
tive rights for multiple and overlapping memberships, for moving among
memberships, and especially for the rejection of membership. Whereas
156 Courtney Jung

collective rights can serve to reify existing categories, boundaries, and

hierarchies, membership rights are designed to adapt to shifts in struc-
ture and social meaning.

IV. Conclusion
The argument for justifying political claims through an appeal to struc-
tural injustice follows directly from a structural account of the origins of
social groups. The obligations of states lie not in the fact that people have
distinct cultural practices but in the fact that states have used such cultural
practices to establish the boundaries of inclusion and exclusion. States
have also used such markers as race, gender, and class to establish the
boundaries of citizenship, which extends the obligations of states beyond
“ethnic” groups alone to categories of people who have been “raced,”
“gendered,” and “classed.” A structural injustice framework establishes
the link not only between the state and the groups it has constituted, but
also among groups that have been similarly constituted, notwithstanding
the different markers that have been employed to denote irreconcilable
difference. A structural theory of obligation links state responsibility to
state actions.
Membership rights respond to structural injustices. They arise to the
extent that groups have been constructed through the unequal allocation
of rights and representation, precisely because group membership there-
fore plays an important role in determining life chances. Membership
rights are differentiated individual rights that are assigned to individuals
whose group membership has operated as a source of exclusion, dis-
crimination, or oppression. They are designed to redress such structural
injustices, to put individuals on an equal footing in a robust, and not
merely formal, sense. To this end, membership rights bundle civil and
political, social and economic, and cultural rights.
Critical liberalism proposes a reformulation of the paradigm that has
guided both normative theorizing and political activism around the issue
of ethnic minority rights, offering a normative framework that transforms
the scope of so-called identity politics and provides an alternative theory
of state obligation toward ethnic and other dispossessed categories.
In general, however, a critical conception of liberalism has less to say
about which solutions will be negotiated in particular instances – whether
the appropriate response to social group demands is affirmative action,
representative quotas, or self-government – and more to say about the
way in which political dialogue is framed. A more critical conception
of liberalism is aimed at shaping democratic deliberation in ways that
take responsibility for the past, and reorienting the rights framework at
 Critical liberalism 157

the core of liberalism. These framing constraints make some settlements

more likely than others, and they will affect the scope of such solutions as
self-government. But just which solutions will be negotiated in particular
instances is still an explicitly open question, to be determined by the
historical context that has shaped the social group, and by politics.

BIBLIOGRAPHY
Appiah, Anthony. 1994. “Identity, Authenticity, Survival: Multicultural Societies
and Social Reproduction,” in Multiculturalism, ed. Amy Gutmann. Prince-
ton, NJ: Princeton University Press, 149–63.
Barry, Brian. 2001. Culture and Equality: An Egalitarian Critique of Multicultural-
ism. Cambridge, MA: Harvard University Press.
Benhabib, Seyla. 2002. The Claims of Culture: Equality and Diversity in the Global
Era. Princeton, NJ: Princeton University Press.
Bilgrami, Akeel. 1992. “What is a Muslim?: Fundamental Commitment and
Cultural Identity.” Critical Inquiry 18 (4), Summer: 821–42.
Carens, Joseph. 2000. Culture, Citizenship, and Community: A Contextual Explo-
ration of Justice as Evenhandedness. Oxford: Oxford University Press.
Crenshaw, Kimberlé. 1995. “Race, Reform, and Retrenchment: Transformation
and Legitimation in Antidiscrimination Law,” in Critical Race Theory, ed.
Kimberlé Crenshaw, Neil Gotanda, Gary Peller, and Kendall Thomas. New
York: The Free Press.
Dworkin, Ronald. 1978. “Liberalism,” in Public and Private Morality, ed. Stuart
Hampshire. Cambridge: Cambridge University Press.
1985. A Matter of Principle. London: Harvard University Press.
Fraser, Nancy. 2000. “Rethinking Recognition.” New Left Review 3 (May–June):
107–20.
Gabel, Peter, and Duncan Kennedy. 1984. “Roll over Beethoven.” Stanford Law
Review, I (29): 1–55.
Galston, William. 1995. “Two Concepts of Liberalism.” Ethics 105 (3 April):
516–34.
Gitlin, Todd. 1995. The Twilight of Common Dreams: Why America Is Wracked by
Culture Wars. New York: Holt.
Ivison, Duncan. 2002. Postcolonial Liberalism. Cambridge: Cambridge University
Press.
Jung, Courtney. 2001. “The Burden of Culture and the Limits of Liberal Respon-
sibility.”Constellations 8 (2): 219–35.
2008. The Moral Force of Indigenous Politics: Critical Liberalism and the Zapatistas.
Cambridge: Cambridge University Press.
Kukathas, Chandran. 1992. “Are There Any Cultural Rights?” Political Theory
20 (February): 105–39.
Kymlicka, Will. 1995. Multicultural Citizenship. Oxford: Oxford University Press.
Macedo, Stephen. 1995. “Liberal Civic Education and Religious Fundamental-
ism: The Case of God v. John Rawls?” Ethics 105: 468–96.
Macklem, Patrick. 2001. Indigenous Difference and the Constitution of Canada.
Toronto: University of Toronto Press.
158 Courtney Jung

Okin, Susan. 1999. Is Multiculturalism Bad for Women? Princeton, NJ: Princeton
University Press.
Rawls, John. 1971. A Theory of Justice. London: Oxford University Press.
1985. “Justice as Fairness: Political not Metaphysical.” Philosophy and Public
Affairs 14 (3): 223–51.
1993. Political Liberalism. New York: Columbia University Press.
Sandel, Michael. 1982. Liberalism and the Limits of Justice. Cambridge: Cam-
bridge University Press.
Scott, Craig and Patrick Macklem. 1992. “Constitutional Ropes of Sand or
Justiciable Guarantees?” University of Pennsylvania Law Review 141: 1–148.
Spinner-Halev, Jeff. 2007. “From Historic to Enduring Injustice,” Political Theory
35 (5): 574–97.
Taylor, Charles. 1994. “The Politics of Recognition,” in Amy Gutmann, ed.
Multiculturalism. Princeton, NJ: Princeton University Press.
Tushnet, Mark. 1984. “An Essay on Rights.” Texas Law Review 62: 1363–1403.
Waldron, Jeremy. 1992. “Superseding Historic Injustice.” Ethics 103: 4–28.
Walzer, Michael. 1983. Spheres of Justice. New York: Basic Books.
Williams, Melissa. 1998. Voice, Trust, and Memory: Marginalized Groups and the
Failings of Liberal Representation. Princeton, NJ: Princeton University Press.
Winbush, Raymond, A. (ed.). 2003. Should America Pay? Slavery and the Raging
Debate on Reparations. New York: HarperCollins.
Yashar, Deborah. 1998. “Contesting Citizenship: Indigenous Movements and
Democracy in Latin America.” Comparative Politics 31 (1): 23–42.
1999. “Democracy, Indigenous Movements, and the Postliberal Challenge in
Latin America.” World Politics 52 (1): 76–104.
2005. “Citizenship Regimes, the State, and Ethnic Cleavages,” in Contesting
Citizenship in Latin America. Cambridge: Cambridge University Press, 31–
53.
Young, Iris Marion. 1990. Justice and the Politics of Difference. Princeton, NJ:
Princeton University Press.
7 Settlers and natives in North America

Mahmood Mamdani

I. A colonized population
The legal position of Native Americans under US law has been the subject
of two key debates. The first revolves around state versus federal claims
to sovereignty over Native Americans. The second is over the relationship
of Indians to the political community that is the United States. The first
debate has been resolved in favor of federal sovereignty. The second has
revolved around two issues – tribal autonomy and individual citizenship
rights – and has yet to reach a satisfactory resolution. Tribal autonomy is
subject to federal plenary powers; at the same time, Native Americans are
a special class of citizens: both citizens and wards of the United States,
they are without any constitutionally guaranteed rights.
State sovereignty was historically the preferred shield for settlers want-
ing to advance local interests. States vied in the courts for full jurisdic-
tion over Native Americans and African-Americans residing within their
boundaries and, in extreme cases, went to war over it. State v. Forman
(1835) was the first case to confirm state criminal jurisdiction over Indian
tribes in direct contradiction to Supreme Court decisions to the contrary
(Harring 1994: 42). By the 1880s, most states with significant Indian
populations asserted such criminal jurisdiction and simply exercised it.
Some, such as California, did so expressly by statute; others, such as
Wisconsin and New Mexico, did so with support from judicial opin-
ions under existing state law (ibid.: 49–50). Finally, in 1953, Congress
enacted Public Law 280 authorizing states to assume criminal and civil
jurisdiction over Indian reservations within their borders, with or with-
out the consent of the tribes involved.1 None of this, however, should
detract from the fact that whatever powers states enjoyed over Indian

1 In 1963, the State of Washington assumed full jurisdiction over consenting Indian tribes
but only partial jurisdiction over non-consenting tribes. The assumption of partial juris-
diction depended on three factors: the status of the land on which the regulated conduct
occurred, the subject matter of the regulated conduct, and the race of the person involved
in the activity, Indian or non-Indian (Newton 1983–4: 281).

159
160 Mahmood Mamdani

communities were a result of federal delegation; whether during the colo-

nial era or after, the Indian question has been an exclusive prerogative of
the federal government and not the states.
This much is clear from an examination of how the crown and the US
Constitution dealt with key aspects of Indian policy: land, Indian affairs,
trade, and treaties and wars. Every colony the crown established in North
America had one legal characteristic in common – the soil granted to the
colonies was subject to the unextinguished right of occupancy of the
Indian tribes. The crown, however, retained the title to all lands occu-
pied by those tribes and the exclusive right to extinguish the tribal right
of occupancy. The Proclamation of 1763 both limited Indian Territory
to west of Mississippi and put it beyond the reach of settlers. The United
States built on this tradition after 1776. Native Americans have never
been accorded the legal status of full title by the US government; the
land they held was put in a different conceptual category, called vari-
ously “Indian title” or “right of occupancy” (Castile and Bee 1992: 63).
Their land rights have been more qualified, akin to usufruct (use) rights
of African peasants that obtained under forms of communal tenure insti-
tuted during the British colonial era.
Indian affairs refers to contacts between Indians and settlers. Initially
regulated by the individual colonies, the crown began to remove control
of Indian affairs from the colonists in 1755 and vested two British super-
intendents with full responsibility for regulating political relations with
the Indians. In matters of trade, a plan was proposed in 1764 to remove
Indian trade from the control of the colonies and put it into the hands of
officers appointed by the crown. Though never formally adopted, it was
used by the British superintendents until 1768, when the regulation of
trade was returned to the colonies. The provisions on Indian affairs and
trade were reproduced in the final copy of the “Articles of Confedera-
tion,” accepted by the Continental Congress in 1777, which contained
the provision that Congress had the sole and exclusive right and power
of “regulating the trade and managing all affairs with the Indians not
members of any of the states,” followed by an ambiguous qualification –
“provided that the legislative right of any state within its own limits be
not infringed or violated.” The Articles also provided that no state could
make treaties or alliances or engage in war with the Indians without the
consent of Congress, unless actually invaded or in great danger of being
invaded (Johnson 1973: 981–2).
The provisions in the Articles of Confederation were carried into the
US Constitution, different articles in which gave the federal government
the unqualified right to extinguish the tribal right of occupancy and estab-
lish control over Indian “trade” and “affairs,” including making treaties
 Settlers and natives in North America 161

or war. It is worth underlining the different ways in which the Constitu-

tion treated the Native American and the African slave. Unlike the chattel
slave who “was subject to the internal laws of the US without his consent
or participation,” Indians were classified in law as “free persons,” but
they were excluded from representation, and thus were not citizens. Not
only was the Indian “not a member of the new national community,”
he was seen as “a threat” who “had to be controlled” (ibid.: 985–6).
Unlike persons of African origin who were to be mastered, Indians were
enemy peoples. In Federalist 24, Alexander Hamilton described “the sav-
age tribes on our western frontier” as natural enemies of the US and
natural allies of the British and Spanish, and cited the presence of the
Indian tribes as reason for keeping a national defense force. In Feder-
alist 25 Hamilton raised the specter of the territories of Britain, Spain,
and the Indian tribes encircling the Union from Maine to Georgia as a
common danger (ibid.: 984). The 1866 Civil Rights Act that guaranteed
“the same rights” to citizens explicitly excluded Indians from citizenship
(ibid.: 217). Even when Indians did become citizens, starting after the
First World War, they were not considered citizens by birth, and their
citizenship rights remained qualified by earlier historical limitations. As
we shall see, Indians – unlike other citizens of the United States – were
defined as both citizens and wards. As Johnson says, “though Indian
affairs are not strictly ‘foreign’ affairs, they are in a constitutional sense
‘external’ affairs.” That power over Indian affairs is “to be exercised as
an attribute of national sovereignty” means that “authority rests in the
federal government even in the absence of an affirmative constitutional
grant” (ibid.: 987). The Indian was thus different from the Black person
in politics, and therefore in law. The Black person was marked by indi-
vidual servitude, but Indians were colonized as peoples comprising so
many tribes and nations. The specificity of the Indian question was fully
elaborated by Chief Justice John Marshall in a set of Supreme Court
rulings over a decade, 1823 to 1832.

Marshall’s verdict
Supreme Court opinion subordinating Native Americans to the juris-
diction of the federal government is set out in three seminal opinions,
rendered over a decade by Chief Justice John Marshall, in three crucial
cases, Johnson v. M’Intosh (1823), Cherokee Nation v. Georgia (1831), and
Worcester v. Georgia (1832). Marshall was a member of the founding gen-
eration, and developed a legal model of Indian rights that relied upon the
same basic language used by the Constitution to define US Indian policy.
Both a jurist and a historian, he was loath to confuse legal and historical
162 Mahmood Mamdani

judgment. While upholding the legal order, his judgments often provide
acute and rare insights into the relationship between law, history, and
politics. Marshall’s single most important contribution was to underline
federal recognition of Indian nations as “domestic dependent nations.”
By so doing, he both incorporated Native American communities within
the scheme of constitutional federalism and defined the limits of doing so.
Was the Marshallian doctrine of “domestic dependent nations” a victory
for those who sought autonomy within the United States, or did it mark –
however gracefully and skillfully – the gravestone of Indian autonomy and
Indian rights? Two views are on offer. One has been advanced by Charles
Wilkinson who sees the Marshall trilogy as having “conceived a model
that can be described broadly as calling for largely autonomous tribal
governments subject to an overriding federal authority but essentially
free of federal control” (Wilkinson 1987: 24; Williams 2005: 228). A
lesser degree of conviction is conveyed by the Native American legal the-
orist Robert Williams Jr. who has argued that Marshall built a “model of
inferior and diminished Indian rights” (Williams 2005: 48–9).
It is worth going through each judgment in the Marshallian trilogy, for
together they laid the foundation stone of federal Indian policy for the
next century and a half. The first of the opinions, Johnson v. M’Intosh
(1823), gets to the heart of the matter distinguishing Native Americans
from American settlers and states that Native Americans are a conquered
people. The significance of Johnson is that it is the opinion in which
Chief Justice Marshall unquestioningly incorporated the European colo-
nial era’s “doctrine of discovery” as the keystone for defining the legal
position of Native Americans under US law. This is how Marshall’s argu-
ment proceeded.
On the discovery of this immense continent, the great nations of Europe were
eager to appropriate to themselves so much of it as they could respectively
acquire . . . But, as they were all in pursuit of nearly the same object, it was
necessary, in order to avoid conflicting settlements, and consequent war with
each other, to establish a principle, which all should acknowledge as the law
by which the right of acquisition, which they all asserted, should be regulated
as between themselves. This principle was, that discovery gave title to the gov-
ernment by whose subjects, or by whose authority, it was made, against all
other European governments, which title might be consummated by possession.
(Johnson v. M’Intosh 1823: 572–3)

True, Indians continued to possess the title of occupancy, but this

was inferior to the right of domain that discovery gave to a European
power and which was “an exclusive right to extinguish the Indian title
of occupancy, either by purchase or by conquest” (ibid.: 587). Marshall
was under no illusion that the doctrine of discovery was “opposed to
 Settlers and natives in North America 163

natural right, and to the usages of civilized nations,” but he argued that
it was “indispensable to that system under which the country has been
settled.” Thus, he concluded, in as unequivocal a confirmation as one is
likely to come across in legal texts, that politics trumps law:

Conquest gives a title which the Courts of the conqueror cannot deny, whatever
the private and speculative opinions of individuals may be, respecting the original
justice of the claim which has been successfully asserted. The British government,
which was then our government, and whose rights have passed to the United
States, asserted title to all the lands occupied by Indians, within the chartered
limits of the British colonies . . . The title to a vast portion of the lands we now
hold, originates in them. It is not for the Courts of this country to question the
validity of this title, or to sustain one which is incompatible with it. (ibid.: 588–9)

Loath to paint necessity as virtue, Marshall nonetheless acknowl-

edged the dark side of the history of right: that might creates right. Can
one speak of Native Americans possessing “rights” in such an arrange-
ment, no matter how qualified, as “inferior” or “diminished”? Marshall
answered this question in Cherokee Nation v. Georgia (1831). He began
by affirming that the Cherokees were indeed a state, “a distinct political
society, separated from others, capable of managing its own affairs and
governing itself” (ibid.: 14–15). But he doubted whether these tribes can
be “denominated foreign nations” for they are territorially “a part of the
United States,” “within the jurisdictional limits of the United States,”
and “under the protection of the United States.” He suggested they be
more appropriately designated as “domestic dependent nations”:

They occupy a territory to which we assert a title independent of their will,

which must take effect in point of possession when their right of possession
ceases. Meanwhile they are in a state of pupilage. Their relation to the United
States resembles that of a ward to his guardian.

And then he proceeded to define the relationship between ward and

guardian:

They look to our government for protection; rely upon its kindness and its power;
appeal to it for relief to their wants; and address the president as their great father.
They and their country are considered by foreign nations, as well as by ourselves,
as being so completely under the sovereignty and dominion of the United States,
that any attempt to acquire their lands, or to form a political connexion with
them, would be considered by all as an invasion of our territory, and an act of
hostility. (ibid.: 16)

A ward is like a child in law. A ward holds no property right, only a right
of use. A ward cannot bring independent action in courts of the United
164 Mahmood Mamdani

States (Williams 2005: 61).2 Without property rights and the right to
maintain an action in court, Indians were not citizens; rather than rights-
bearing members of the political community called the United States,
they lived on the territory of the United States on sufferance of its citizens
and their representative political institutions, particularly Congress. So
Marshall concluded his judgment:

If it be true that the Cherokee nation have rights, this is not the tribunal in which
those rights are to be asserted. If it be true that wrongs have been inflicted, and
that still greater are to be apprehended, this is not the tribunal which can redress
the past or prevent the future. (ibid.: 18)

The conundrum that Native Americans faced can be glimpsed from

the key facts surrounding Worcester v. Georgia (1832). The case originated
from an attempt by the state of Georgia to extinguish the tribe as a
distinct, self-governing society within its borders. Georgia’s solution was
to strip the Cherokee of their tribal citizenship and lands, and thereby
to integrate them as a free people but under white racial dictatorship in
the state. The result would be to place individual Cherokees in the same
position as free Blacks – subject to the onerous, racially discriminatory
legal regime imposed by Georgia on all “free persons of color” within
its sovereign borders. Section 15 of the Act passed by the legislature
of Georgia on December 19, 1829 spelt out the consequences in no
uncertain terms:

And be it further enacted, that no Indian or descendant of any Indian, residing

within the Creek or Cherokee nation of Indians, shall be deemed a competent
witness in any court of this state to which a white person may be a party, except
such white person resides within the said nation. (Worcester v. Georgia 1832: 7)

As legally designated second-class citizens of color, they would be

unable to vote, unable to serve in the state militia, and unable to send
their children to Georgia’s public schools under the racial apartheid laws

2 Marshall then went on speculate as to why the founders did not provide constitutional
rights for Native Americans:
In considering the subject, the habits and usages of the Indians, in their intercourse with
their white neighbors, ought not to be entirely disregarded. At the time the constitution
was framed, the idea of appealing to an American court of justice for an assertion of right
or a redress of wrong, had perhaps never entered the mind of an Indian or of his tribe.
Their appeal was to the tomahawk, or to the government. This was well understood by
the statesmen who framed the constitution of the United States, and might furnish some
reason for omitting to enumerate them among the parties who might sue in the courts of
the nation.
This speculation can only be excused as an attempted rationalization.
 Settlers and natives in North America 165

that would apply to the Cherokees under state jurisdiction. In addi-

tion, the “Cherokee codes” would enable the state to take control over
the immensely valuable Indian lands within its borders and make them
available to Georgia’s white citizen farmers and plantation owners. So the
choice for the Cherokee was between a colonial subjugation to the federal
government and racial subjugation to settlers in the form of a second-class
citizenship in the state of Georgia. In mid-nineteenth-century United
States, the former was the lot of Indians and the latter of Blacks.
It is also worth noting that the Court failed to arrive at a unanimous
verdict as regards the legal status of the Cherokee within the United
States; instead, the justices divided three ways, offering three different
interpretations. Marshall’s position, that the Cherokee nation was not
a foreign nation but a domestic dependent nation, was shared by only
Justice McLean. Justices Thompson and Story believed that the Chero-
kee were indeed a foreign nation, even if defeated. The two remaining
justices, Baldwin and Johnson, believed that the Cherokees were a con-
quered people with no status as a nation at all, either foreign or domestic
(Harring 1994: 30–1). The interesting fact is that none of the justices
were willing to think outside the box, beyond the historical fact of con-
quest, to contemplate either their coexistence or their integration as a
free people within the political community called the United States. I
shall return to this point after considering the third case.
In his third and final opinion on the subject, Worcester v. Georgia (1832),
Marshall returned to the question of power and to its claims, first to the
“doctrine of discovery,” and then to ask whether the right that followed
upon discovery was an effect of a superiority of civilization or of brute
force. He began by reflecting on the world in 1491: “America, separated
from Europe by a wide ocean, was inhabited by a distinct people, divided
into separate nations, independent of each other and of the rest of the
world, having institutions of their own, and governing themselves by their
own laws.” He then dismissed the idea that discovery could be the basis
of a claim of dominion:

It is difficult to comprehend the proposition, that the inhabitants of either quarter

of the globe could have rightful original claims of dominion over the inhabitants
of the other, or over the lands they occupied; or that the discovery of either by the
other should give the discoverer rights in the country discovered, which annulled
the pre-existing rights of its ancient possessors. (Worcester v. Georgia 1832: 33)

He next posed a rhetorical question, as to whether the right of domain

could be the effect of superiority in civilization: “Or has nature, or
the great Creator of all things, conferred these rights over hunters and
166 Mahmood Mamdani

fishermen, on agriculturists and manufacturers?” Finally, he concluded

that the claim of dominion lay in the naked fact of conquest:
But power, war, conquest, give rights, which, after possession, are conceded
by the world; and which can never be controverted by those on whom they
descend. We proceed, then, to the actual state of things, having glanced at their
origin; because holding it in our recollection might shed some light on existing
pretensions. (ibid.: 33–4)

Worcester went on to reflect on the nature of that power: first, on who

had the right to exercise it and, second, on whether there were any
limits to that exercise. On the first question, Marshall was unequivocal,
holding that it is the federal government, and not the individual states,
that possesses the exclusive right to exercise control over Indian affairs
(Williams 2005: 63). But on the second question he equivocated, leaving
plenty of room for further interpretation. To begin with, Marshall defined
the limits of the “principle of discovery”: if it “shut out the right of
competition among those who had agreed to it,” it could not “annul
the previous rights of those who had not agreed to it.” These were “the
rights of those already in possession, either as aboriginal occupants, or
as occupants by virtue of a discovery made before the memory of man.”
What were these rights of the original occupants? According to Marshall,
they were the right to sell – but only to the discoverer; they had the right
of occupancy, but subject to the superior right of domain that belonged
to the discoverer (Worcester v. Georgia 1832: 34–6).
Worcester has been regarded by many as a landmark victory for Indian
rights. But none of these rights is “inalienable,” and none is protected
by a constitutional guarantee. While its principle of federal supremacy
in Indian affairs theoretically immunizes tribal Indians from different
forms of state encroachments on tribal rights and interests (Williams
2005: 69), its focus on the relation between the same Native Americans
and the federal government does no more than affirm a colonial rela-
tionship between the United States and Native Americans.3 The notion
of tribal sovereignty was not without substance: in placing treaty Indians
outside the domain of state law, the Supreme Court gave them effective
protection against state jurisdiction. But in doing so, the Supreme Court
placed these same Indians squarely within the domain of federal law, at
the mercy of Congress, and without any protection. At the heart of this
relationship is a lack of constitutional definition and protection of Indian
rights, a situation that continues today and that explains the shifts and

3 Williams notes that his response to an 1828 address made by his close friend and colleague
on the court, Justice Joseph Story, suggests that Marshall himself was keenly aware of
the Indian’s essential fate as perpetual colonial subject under US control (2005: 226–7).
 Settlers and natives in North America 167

turns in the legal position of Indian communities – as summed up in the

doctrine of tribal sovereignty – in the United States over the past two
centuries and more.

II. Tribal sovereignty

Indian sovereignty historically rested on two independent pillars: control
over land as the principal means of subsistence, and the community’s
right to organize and administer its own affairs. Two turning points mark
the step-by-step erosion of the economic and political ground that sup-
ported these pillars: the 1763 Royal Proclamation that for the first time
referred to a portion – and not the whole – of North American territory
as Indian Country, and the 1871 pronouncement by the US Congress
that no more treaties would be entered into between the United States
and Native American communities. Both formally acknowledged pro-
cesses that had been in motion for some time: the loss of land in the first
instance, and the loss of political independence in the second.
The term Indian Country4 was first used in the famous Royal Procla-
mation of 1763, issued by King George III. Prior to that there was no
reference to a place called Indian Country, since all attempts to access
land in North America began with the recognition that it belonged to
Indians (Wilkinson 1987: 89). Three types of practices predominated
in the early days. The first was that of renting, as when Maine settlers
rented land from Indians. The other two involved the purchase of land
from Indians, either by individual settlers (as when Dutch and English
colonies made grants of Indian land to settlers and then expected them
to purchase it individually from Indians – that is, if the land had not
been confiscated from Indians in war) or by the colony concerned (as
English successors to the Dutch did in the Middle Atlantic colonies, when
the colony bought the land wholesale from the Indians, then granted it
piecemeal to settlers, thereby obviating the need for individual settlers or
groups of settlers to bargain with the Indians). This third type of practice
was adopted by the English crown when it took control of Indian affairs
from some of the colonies, and then by the USA on the formation of the
confederation and then the Union. Eventually, all three – the colonies,
the crown and the US – refused to recognize Indian claim to Indian
lands beyond simple possession or occupancy. And yet the fact was that
all three had in the past paid money to the Indians in exchange for a

4 From this point of view, Indian tribes are viewed as foreign nations and their aggregated
reservation areas are technically known as “the Indian Country,” with each reservation
being “Indian Country.”
168 Mahmood Mamdani

formal quitting of their claim. Did this not, at least implicitly, acknowl-
edge Indian ownership of land? In time the United States Supreme Court,
trying to resolve the inconsistencies of theory and practice, referred to
the purchase from Indians as the purchase merely of their “right of occu-
pancy,” as distinct from a right of domain or ownership (Macleod 1928:
203–4).
It is the distinction between the right of use and the right of domain
over the same land that became the cutting edge of a process which
acknowledged Indian right to use land but denied their sovereignty over
it. Tribes were free to sell land, but only to the nation said to be their
protector. For example, Article 2 of the treaty of the US with the tribes
of the territory northwest of the Ohio, of January 9, 1789, guaranteed
them the possession of their lands forever, provided that “the said nations, or
either of them, shall not be at liberty to sell or dispose of the same, or any
part thereof to any sovereign power, except the United States; nor to the
subjects or citizens of any subject power, nor to the subjects or citizens
of the United States” (ibid.: 440). A renewal of the treaty of October 22,
1784 with Iroquois tribes, except the Mohawks, fixed “a boundary line
between the Six Nations and the territory of the United States,” once
again, “forever.” Similarly, according to the Hopewell treaty of November
28, 1785, the Cherokees “do acknowledge themselves to be under the
protection of the United States of America, and of no other sovereign
whatsoever” (ibid.: 440–1).
A second limitation on tribal sovereignty concerned criminal jurisdic-
tion. So the treaty of January 1789, with the Delaware and other tribes
northwest of the Ohio, stipulated that the United States should extend
its criminal jurisdiction over them, applying the criminal law of the ter-
ritory or the state which surrounded the tribal domain. Practically that
same limitation was contained in the Creek treaty of 1790 (ibid.: 442).
A further limitation, as in the 1805 treaty with the Piankeshaw, went
with a change in language from “protection” to “care and patronage” to
describe the relation of the US with the tribe, a sure sign that the status
of the tribe had been degraded from a protectorate to a community of
“wards” (ibid.: 443).
During the treaty era, tribal sovereignty referred to the right of use
granted tribes, usually in perpetuity, by treaties. Even though the treaty
era ended in 1871, Section 2079 of the Revised Statutes, March 3, 1871,
which declared the end of treaty-making, also guaranteed the sanctity
of all previous treaties, as in the following clause – “provided, further,
that nothing herein contained shall be construed to invalidate or impair
the obligation of any treaty heretofore lawfully made or ratified with any
such Indian nation or tribe.” But the provision that the statute was not
to be retroactive in its application created an anomaly. The 1871 statute
 Settlers and natives in North America 169

created wards of Indians, but the non-retroactive provision recognized

the independence of treaty Indians, however qualified. Thus the Com-
missioner of Indian Affairs complained to the President two years later
(1873) of “the anomalous relation of the Indian tribes to the government,
which requires that they be treated as sovereign powers and as wards at
one and the same time” (ibid.: 537).
The doctrine of “tribal sovereignty” persisted because it was suscep-
tible to multiple uses. To get an idea of the range of uses, we can look
at the extremes. When it comes to asserting federal over state jurisdic-
tion, the notion of “tribal sovereignty” had real bite; but when it came to
asserting Native American “sovereignty” in their relation with the federal
government, its significance was more rhetorical than real. Without a
consideration of the multiple uses to which it has been put, it is diffi-
cult to understand why the doctrine of “tribal sovereignty” continues to
survive in the language of the law to this day.
Thus, federal courts have continued to use the doctrine of tribal
sovereignty to check local white interests, as when Justice Hugo Black
held in Williams v. Lee (1959) that the Navajo possessed “the right to
govern themselves,” a right that the state of Arizona could not infringe
by enforcing state civil law on the reservation (Harring 1994: 289). But
rather than give real content to “sovereignty,” the doctrine did the reverse:
it defined in perpetuity the boundaries of a conquered, colonized, com-
munity – the “tribe” – in its relations with the federal government.
The fact was that Indian tribes were no longer sovereign; the notion
of “tribal sovereignty” referred to a continuing practice of internal self-
administration, rather than any real independence of administration from
external forces. In this context, “tribal sovereignty” defined more a de
facto state of affairs – which continued only so long as it was tolerated by
those in power – than a statement of constitutionally enforced principles.
At the heart of this de facto state of affairs was a regime of legal
dualism that was defined racially (as in a colonial setup) rather than ter-
ritorially (as it would be in a federal setup).5 The Intercourse Act of 1834

5 Alaska was the one place where a significant number of Native Americans lived under
the same law as whites. For this there were historical reasons. Alaska was bought in
bulk from Russia and there were no treaties with Native American communities. Natives
constituted a substantial majority and there were few whites. American courts came to
Alaska only in 1884 and in the nineteenth century, Alaskan natives lived under the same
law as Alaskan whites (Harring 1994: 207). There is also the case of the Pueblos. In US
v. Joseph (1876), the court decided that the Pueblos had a settled (i.e. civilized) domestic
existence and were therefore not subject to the same laws passed for the protection and
civilization of “wild tribes.” As in Alaska, the laws of New Mexico Territory applied to
the Pueblo just as they applied to other people in the state. This was overruled in US v.
Sandoval (1913). And finally there were the Indians of Oklahoma, a case we shall look
at later in some detail (Harring 1994: 45–6).
170 Mahmood Mamdani

provides that the general laws of the US will apply to crimes committed
any place in the US, except in “Indian Country” where it will apply only
“where a non-Indian is the criminal or victim, not where the affair is
between Indians” (Macleod 1928: 535). The relationship was defined by
Justice Taney in United States v. Rogers (1846): tribes in Indian Country
may have authority over Indians within their lands, but the United States
has authority over whites, as well as the right to determine who belongs
to which racial category. Therefore, two criminal jurisdictions coexisted
side by side in Indian Country, with the United States exercising the
power to determine who fell under which jurisdiction (Harring 1994:
61). With federal district courts having criminal jurisdiction over whites
and over Indians who committed crimes against whites, a jurisdictional
void was created. On the one hand, Indian courts lacked jurisdiction over
non-Indians in Indian Country; on the other, federal courts lacked juris-
diction over crimes committed by Indians against Indians. Non-Indians
came to Indian Country for several reasons. Given several hundred years
of intermarriage and intertribal migration, there were not only many
non-Indians integrated as Indians, but also many persons born of inter-
marriages known as “mixed bloods.” If Indian was defined as a racial
status, as it was by US courts, the definition of Indian would be up to US
authorities; but for Native Americans who saw Indian as more a political
than a racial identity, as membership in a political community, this group
were Indians.6 The last category of non-Indians similarly attracted were
criminals who found refuge in Indian Country because there was often
no US court within a hundred miles or more (Harring 1994: 63–5, 166).
That federal courts lacked jurisdiction over crimes committed by Indi-
ans against Indians – an aspect of a residual “Indian sovereignty” – was
highlighted by Crow Dog (1881). The details of the case were as follows:
on August 5, 1881, Kan-gi-shun-ca (Crow Dog) shot to death Sin-ta-
ga-le-Scka (Spotted Tail) on the Great Sioux Reservation in Dakota
territory. Because the two men had taken up radically opposed posi-
tions in the debate over how to relate to white power, the killing had
political significance. In the ongoing debate on the reservation, Spotted
Tail had called for a realistic recognition of the overwhelming nature of
white power and thus favored a qualified embrace of “civilization” as
identified with that power, whereas Crow Dog had stood for the outright

6 Take, for example, the facts in US v. Rogers (1846) where Rogers, a white man, claimed
that he had been adopted into the “Cherokee tribe of Indians, and having married a
Cherokee Indian woman” under tribal law, had renounced his US citizenship. His novel
legal argument to the court was that, politically speaking, he was an Indian and therefore
exempt from the US criminal jurisdiction over non-Indians in Indian Territory (Williams
2005: 73).
 Settlers and natives in North America 171

repudiation of white power in defense of Indian “tradition.” The families

met and settled the matter, but police hunted down Crow Dog and state
courts convicted him of murder. In December 1883, the US Supreme
Court reversed the conviction, holding that the United States had no
criminal jurisdiction over Indian tribes in “Indian Country,” because the
tribes, inherently sovereign, retained the right to administer their own
law as an element of that sovereignty (Harring 1994: 1). But even in this
landmark case, Indian sovereignty was not defined as the hallmark of
Indian independence from external constraint, but as a sign of the spe-
cific nature of its subordination to the external power that was the United
States, and thus of its alienation from the political community called the
United States. This is how Justice Mathews defined the relationship of
Indians to the United States:
They were nevertheless to be subject to the laws of the United States, not in the
sense of citizens, but as they had always been, as wards, subject to a guardian;
not as individuals, constituted members of the political community of the United
States, with a voice in the selection of representatives and in the framing of the
laws, but as a dependent community who were in a state of pupilage, advancing
from the condition of a savage tribe to that of a people who, through the discipline
of labor, and by education, it was hoped might become a self-supporting and self-
governing society. (Ex Parte Crow Dog 1883: 568–9; Williams 2005: 77)

Plenary powers
Congress responded to Crow Dog (1883) by defining the status of Indians
as wards of the United States. The result was to repudiate, in deed if not
in word, any claim to Indian “tribal sovereignty.”7 The congressional
response extended federal law to tribal Indians under the Major Crimes
Act (1884).8 Two years later, the Supreme Court recognized in United

7 But the tendency among those who write of federal Indian law is to see a divide between
cases that affirm Indian sovereignty and those that repudiate it. This is how Wilkinson
assesses the historical significance of Crow Dog: “A line of cases affirming sovereignty
runs through Worcester, Crow Dog, and Talton v. Mayes 163 US 376 (1895) while an
opposing line of cases denying that sovereignty and giving the United States ‘plenary
powers’ over the Indian tribes begins with Kagama and Lone Wolf and dominates Indian
law in the first half of the twentieth century” (Wilkinson 1987: 24; Harring 1994: 9).
8 As we shall see, this was followed by the imposition of a repressive system of admin-
istrative justice, under the authority of the Bureau of Indian Affairs, for lesser offenses
on reservations. The IRA proposal was for the Indian agent to serve as judge in courts
without juries, with a right of appeal only to the commissioner of Indian affairs, and with
few due process protections (a second-class citizenship at best) (Harring 1994: 134). An
Act of Congress of March 3, 1885, arrogated to the US the right to punish the reser-
vation Indians for crimes committed against one another, and reserved to state courts
the right to punish the crimes of Indian against Indian off-reservation, and the crimes of
non-Indian against non-Indian on-reservation (Macleod 1928: 537–8).
172 Mahmood Mamdani

States v. Kagama (1886) how very fundamentally Congress had changed

the terms on which Indian communities could govern themselves. That
sanction is known as the plenary power doctrine. This new doctrine put
the tribes completely under the control of Congress and the US political
process, and legally justified Congress’s assertion of total power over
the tribes (Harring 1994: 101). Even if accented by an unmistakable
arrogance of power, one is struck by the extent to which the language of
Kagama marks a continuation of the language used by Justice Mathews
in Crow Dog:
These Indians are the wards of the nation. They are communities dependent on
the United States. Dependent largely for their daily food. Dependent for their
political rights . . . From their very weakness and helplessness . . . there arises the
duty of protection. And with it the power . . . The power of the General Govern-
ment over these remnants of a race once powerful, now weak and diminished in
numbers, is necessary for their protection, as well as to the safety of those among
whom they dwell. (Harring 1994: 142, 147–8)9

In spite of congressional plenary powers, the notion of “Indian

sovereignty” continues to be an important point of reference in fed-
eral Indian law. Wilkinson argues that it is the scholarship of Felix Cohen
during the 1940s that played a cardinal role in preserving the doctrine
of tribal sovereignty for modern courts. Cohen retrieved and reinforced
Marshall. As late as US v. Wheeler (1978), the Supreme Court distin-
guished “Indian tribes from both cities and federal territories, which
derive their power from the states and the federal government respec-
tively” and “found that Indian tribes possess a third sovereignty in the
US” (Wilkinson 1987: 61). And Wilkinson concurs: “Perhaps the most
basic principle of all Indian law, supported by a host of decisions here-
inafter analyzed, is the principle that those powers which are lawfully
invested in an Indian tribe are not, in general, delegated powers granted
by express Acts of Congress, but rather inherent powers of a limited
sovereignty which has never been extinguished” (ibid.: 57–8). Both the
courts and the writers who shared this judgment were right in acknowl-
edging the distinctive historical origin of Indian sovereignty, and that it
predated the United States, but their argument did not fully recognize
the historical developments that had eroded this claim. Thus, history (and
origin) hardly illuminated the scope of Indian sovereignty in the present.
For the contemporary reality is that there is no limit to how far Congress
may breach this historical sovereignty. Even today if tribes have a right to
define the tax regime or adjudicate conflicts within reservations, the fact

9 For full quotes from the court’s judgment re the extra-constitutional powers of
Congress – thus not subject to judicial review – over Indians, see Williams (2005: 81–3).
 Settlers and natives in North America 173

is that this right is subject to sufferance by Congress, and can be repealed

by Congress at will, and without review by the Supreme Court – a fact
that Wilkinson himself is compelled to recognize.10
More recently, the Supreme Court has been less willing to tolerate
ambiguity on this question. Robert Williams has outlined the debate in his
recent book on the Rehnquist court. Two points of view can be discerned
in recent Supreme Court decisions. On the one hand is Chief Justice
Rehnquist’s outright dismissal of the notion of “Indian sovereignty” in
Oliphant (1978) as untenable, either legally or politically:

Indians are within the geographical limits of the United States. The soil and
people within these limits are under the control of the Government of the United
States, or of the States of the Union. There exists in the broad domain of
sovereignty but these two. There may be cities, counties, and other organized
bodies with limited legislative functions, but they . . . exist in subordination to
one or the other of these. (Williams 2005: 110)

Chief Justice Rehnquist further reiterated this point of view in United

States v. Lara (2004) when he held that Congress, under the Constitu-
tion’s plenary grants of authority over Indian affairs, has the power “to
enact legislation that both restricts and, in turn, relaxes those restrictions
on tribal sovereign authority” (ibid.: 150). Justice Thomas, in contrast,
began by stating his express disagreement with the Lara majority’s hold-
ing “that the Constitution grants to Congress plenary power to calibrate
the metes and bounds of tribal sovereignty.” In Justice Thomas’s view,
“the tribes either are or are not separate sovereigns, and our federal
Indian law cases untenably hold both positions simultaneously”:

It seems to me that much of the confusion reflected in our precedent arises from
two largely incompatible and doubtful assumptions. First, Congress (rather than
some other part of the Federal Government) can regulate virtually every aspect
of the tribes without rendering tribal sovereignty a nullity. Second, the Indian
tribes retain inherent sovereignty to enforce their criminal laws against their own
members . . . It is quite arguably the essence of sovereignty not to exist merely at
the whim of an external government. (cited in Williams 2005: 159)

While conceding that states can have their sovereignty diminished by

Congress, Thomas explained that the states of the Union are part of a

10 According to Wilkinson: “Wheeler dealt with the source of tribal powers, not with the
difficult question relating to their scope.” Even if the Supreme Court ruled in Menominee
Tribe of Indians v. United States (1968) that the federal government may terminate its
recognition of a tribe but that does not terminate its existence, until the tribe does so,
this was not a recognition of sovereignty but identity: “Indians are properly viewed not
as members of a race but as citizens of a government with whom the United States has
a special government-to-government relationship” (Wilkinson 1987: 62, 73, 75–6, 86).
174 Mahmood Mamdani

“constitutional framework” that recognizes their sovereignty and “specif-

ically grants Congress authority to legislate with respect to them.” The
tribes, according to Thomas, are not part of the constitutional order
and their sovereignty is in no way guaranteed by the Constitution (ibid.:
159). Rather than be treated as a fait accompli, argues Justice Thomas,
this most unsatisfactory situation demands a clear resolution.
What, then, is the difference between “Indian sovereignty” and the two
forms of sovereignty recognized in the Constitution of the United States:
federal and state? The first is historical: that Indian sovereignty in the
Americas preceded that of all settler states. The second is contemporary:
the very establishment of the United States, and consolidation of settler
power, extinguished Indian sovereignty. The fact is Indian communities
continue to exist in a colonial relationship to the United States, and this
is why the notion of Indian sovereignty finds neither recognition, nor
definition, nor protection in the US Constitution, whether as originally
drafted or as amended thereafter.
This brings us back to that great jurist, Chief Justice Marshall, and
his deliberations on the colonial conquest of Native America. In his
reflections on the Indian question, Marshall tried to think of a future
outside the box, the settler conquest of the native:
The title by conquest is acquired and maintained by force. The conqueror pre-
scribes its limits. Humanity, however, acting on public opinion, has established,
as a general rule, that the conquered shall not be wantonly oppressed, and that
their condition shall remain as eligible as is compatible with the objects of the
conquest. Most usually, they are incorporated with the victorious nation, and
become subjects or citizens of the government with which they are connected.
The new and old members of the society mingle with each other; the distinction
between them is gradually lost, and they make one people. Where this incorpora-
tion is practicable, humanity demands, and a wise policy requires, that the rights
of the conquered to property should remain unimpaired; that the new subjects
should be governed as equitably as the old, and that confidence in their security
should gradually banish the painful sense of being separated from their ancient
connexions, and united by force to strangers. (Johnson v. M’Intosh 1823: 589)

The thought is bold: conquest must not only draw the curtain on
the independence of a previously existing political community, it must
also open the possibility of creating a new political community, where
conqueror and conquered, whether joined or separate, shall be as equal
members. But this is where Marshall hesitated, uncertain that this prin-
ciple of “civilized” society could apply to natives in the Americas.
But the tribes of Indians inhabiting this country were fierce savages, whose occu-
pation was war, and whose subsistence was drawn chiefly from the forest. To
leave them in possession of their country, was to leave the country a wilderness;
 Settlers and natives in North America 175

to govern them as a distinct people, was impossible, because they were as brave
and as high spirited as they were fierce, and were ready to repel by arms every
attempt on their independence. (ibid.: 590)

Thus, argued Marshall, “that law which regulates, and ought to regu-
late in general, the relations between the conqueror and conquered, was
incapable of application to a people under such circumstances” (ibid.:
591). So Marshall concluded by reconciling himself to political reality
as it exists, even if it “be opposed to natural right, and to the usages of
civilized nations”:

However extravagant the pretension of converting the discovery of an inhabited

country into conquest may appear; if the principle has been asserted in the first
instance, and afterwards sustained; if a country has been acquired and held under
it; if the property of the great mass of the community originates in it, it becomes
the law of the land, and cannot be questioned . . . However this restriction may
be opposed to natural right, and to the usages of civilized nations, yet, if it be
indispensable to that system under which the country has been settled, and be
adapted to the actual condition of the two people, it may, perhaps, be supported
by reason, and certainly cannot be rejected by Courts of justice. (ibid.: 591–2)

There is a limit to the justice courts can dispense; that limit is defined
by the political power. Courts do not establish rights, power does! It was
to Marshall’s credit and testimony to his moral courage that he tried to
think through the colonial question. But that he failed to think of a future
where the native would be as free as the settler, indeed where native and
settler would cease to be opposed political identities, was his failing. If the
strength set him apart from contemporaries, he shared the failing with
them. The challenge for successive generations, including for us, is to
reflect on the question Marshall had the courage to pose, the colonial sta-
tus of Native Americans, so as to craft a future he failed to contemplate –
a world of political equality between erstwhile natives and settlers.

III. Separate and unequal futures

A common thread that runs through the history of settler power in the
United States is the separate and unequal futures for settlers and natives.
Within this broad framework there have been a number of shifts. The
settler enterprise began with the promise of separate representation for
Indians in the US Congress, then moved on to the idea of two states, one
white, the other Indian, on each side of the Mississippi, except that the
native state would be a protectorate of the settler state. When westward
expansion in the nineteenth century destroyed that vision, settlers crafted
other, but more cruel, futures for natives, all within an explicitly colonial
176 Mahmood Mamdani

context. One of these was forced assimilation, the other compulsory

segregation.

Creating Indian states

In early treaties with Indian communities – sometimes called nations,
other times tribes – both the English crown and the United States offered
these communities independent representation in Congress. On the face
of it, these seemed offers to coexist in political equality under a common
political roof, made to all Indian tribes that believed in peaceful coexis-
tence with whites. Later, the terms of the offer changed to a contradictory
mix of equality and protection – and finally, to a form of colonial pater-
nalism. The first treaty of the US, then a confederation, was made with
the Delaware, in September 1778. Article 6 invited the Delaware nation
“to form a state” by joining with “other tribes, friendly to the interest
of the United States,” and provided that “the Delaware nation shall be
the heads” of the state “and have a representation in Congress.”11 The
Hopewell Treaty of November 28, 1785, with the Cherokees contained
a provision acknowledging the right of the Cherokee nation “to send a
deputy of their choice, whenever they think fit, to Congress.”12
During the second phase of its relations with Indian communities,
the US deliberated on, and sometimes publicly considered, a two-state
policy. It began as a promise to those tribes willing to move east of the
Mississippi. In his last annual message to Congress in 1825, President
Monroe proposed that after their removal west of the Mississippi, the
tribes could be confederated into a protectorate of the United States and
used to extend the influence and trade of the United States westward to
the Pacific (Macleod 1928: 446). In 1834, once again, the United States
Congress considered a proposal for a permanent homeland for Indians,
with a governor and a council of tribal representatives, that might ulti-
mately become a state of the Union, but then dismissed the idea (Kiernan
1978: 40–1). Indian removal to the west followed a dual approach: semi-
captivity for Indians who resisted assimilation and promises of statehood
for those who came on board the “civilizing mission.”

11 This is how the full provision reads in Article 6: “The parties further agree, that other
tribes, friendly to the interest of the United States, may be invited to form a state, whereof
the Delaware nation shall be the heads, and have a representation in Congress.” See
Worcester v. Georgia (1832), p. 39.
12 This is how Article XII of the Treaty read: “That the Indians may have full confidence
in the justice of the United States, respecting their interests, they shall have a right to
send a deputy of their choice, whenever they think fit, to Congress.” See, “Treaty of
Hopewell with the Cherokees, November 28, 1785,” in Prucha (1976: 8).
 Settlers and natives in North America 177

With the Civil War and the Reconstruction era that followed, the pen-
dulum swung yet again between two contrasting proposals. The single-
state solution envisioned Indians as becoming part of the United States
as a political community; the two-state solution was premised on the
establishment of a separate state for Indians. The former initiative was
part of the 1867–9 Southern constitutional conventions that swept the
Reconstruction South in the wake of the Civil War. These constitutions
established the South’s first state-funded systems of compulsory educa-
tion and the first systems of providing poor relief. All guaranteed civil and
political rights for Blacks. It is in the context of what a Texas paper called
“the equal rights revolution” that Florida granted Seminole Indians two
representatives in the state legislature (Foner 1988: 319–20). The latter
was given a fresh lease on life with the new treaty offered to the Five
Civilized Tribes which had already moved west of the Mississippi.
Largely abandoned by the Union and courted by the Confederacy, the
Five Civilized Tribes had followed the leadership of a slave-owning elite
and joined the war effort of the South. After the Civil War, retribution
was to follow. Based on guidelines from the Secretary of the Interior, the
Commissioner of Indian Affairs offered new treaty arrangements to the
Indians. One of six key stipulations he outlined called for the formation
of “one consolidated government after the plan proposed by the Senate
of the United States, in a bill for organizing the Indian territory” pro-
vided “other arrangement be made.” The Harlan Bill (named after the
Secretary of the Interior) called for the formation of an Indian state on
terms different from those enjoyed by all other states. These were the
terms of a Protectorate. As spelt out by the Commissioner, this would
be “a regular territory of the United States, with a governor appointed
by the President having an absolute veto over the legislative council and
acting ex officio as superintendent of Indian affairs.” At the same time,
“tribal citizens would elect a council, and the Territory thus constituted
would send a delegate to Congress.” When the Indian nations refused
to accede to such an arrangement, they were forced to sell large parts of
their land to the US “for use by other Indians, thus dividing the Indian
Territory roughly in two; the Five Civilized Tribes retained the eastern
half, while the remainder was to be used as homes for western tribes who
were to be settled on reservations as part of the government’s policy of
consolidation. Slavery was prohibited, and the freedmen were granted
certain rights within the nations” (Prucha 1976: 375–6; Kiernan 1978:
80–1). The boundaries of Indian Territory shrank further: from conti-
nental US in 1491, to west of the Mississippi after Indian Removal, to
the future state of Oklahoma during the Civil War, to its eastern part
thereafter. Each time the promise of autonomy lasted only until the next
rise in the settler tide.
178 Mahmood Mamdani

Once the treaties of 1866 had been signed, Ely S. Parker, Commis-
sioner of Indian Affairs, urged that action be taken to organize the general
council spoken of in the treaties, and suggested: “The next progressive
step would be a territorial form of government, followed by their admis-
sion into the Union as a state” (Prucha 1976: 377–8). The first annual
report of the Board of Indian Commissioners, issued that same year, on
April 10, 1869, spelt out a combined policy for “civilized” and “unciv-
ilized” Indians, whereby both could be brought together in the Indian
Territory that would become a state of the Union:
The policy of collecting Indian tribes upon small reservations contiguous to each
other, and within the limits of a large reservation, eventually to become a state of the
Union, and of which the small reservations will probably be the counties, seems to
be the best that can be devised. Many tribes may thus be collected in the present
Indian territory. (ibid.: 107–8, emphasis mine)

The next time a state solution was proposed for Native Americans was
when Congress appointed the Dawes Commission on March 3, 1893.
The three commissioners were authorized to negotiate with the Five Civ-
ilized Tribes (the Cherokee, the Choctaw, the Chickasaw, the Muskogee
[or Creek], and the Seminole) “for the purpose of the extinguishment
of the national or tribal title to any lands within that Territory now held
by any and all of such nations or tribes,” so as “to enable the ultimate
creation of a state or states of the Union which shall embrace the lands
within said Indian Territory” (Commission to the Five Civilized Tribes
(Dawes Commission), March 3, 1893, ibid.: 189). On the face of it,
the Commission’s mandate was to extinguish and replace “national or
tribal titles” with the “creation of a state or states of the Union.” But
we shall see that the real object of pursuit was the rich agricultural and
mineral lands in Indian Territory highly coveted by whites. According to
the mandate, as we shall see, the dissolution of national or tribal govern-
ments was to go alongside another dissolution – that of national or tribal
land – so as to allot titles to individual Indians and thereby to declare
the remaining land as surplus and secure it under US custody. The move
would shatter the hopes of the Five Civilized Tribes who, more than any
other group of Indians, had embraced the promise of “civilization,” and
tried their best to adapt to the settler way of life. It would also be the last
time the US would offer any Indian community the mirage of separate
statehood.

IV. Forced assimilation

The experience with “civilization” proved to be equally disillusioning.
There is little to suggest Indian resistance to new ways so long as their
 Settlers and natives in North America 179

freedom was not in question. The contention between “civilization” and

“tradition,” with one set up as the alternative to the other, arose under
conditions where civilization seemed to go along with loss of indepen-
dence and tradition was a safeguard of independence. As we shall see
with the Five Civilized Tribes in Indian Territory in eastern Oklahoma,
their experience with the “civilizing mission” was contradictory and dis-
enchanting. Though openness to civilization was initially demanded of
them as a condition for autonomy within the United States, the actual
results of “civilization” were given as reasons for ruthlessly undercutting
that autonomy.
Nothing in the early accounts of Native Americans suggests that they
shunned European immigrants. This is why we need to begin by distin-
guishing voluntary integration from forced assimilation. Macleod writes:

The Indians were always eager to have their children educated by Europeans.
Indian parents would trustfully turn children over to some trader, whom they
came to admire for his apparent wisdom and wealth, to be taken to the coast and
to be trained according to European ways. Within a specified length of time the
young apprentice was to be returned to his parents. The white guardian would
instead ship the boy or girl to the slave market, and, in time, inform the Indian
parents that their child had died. (Macleod 1928: 303)

Until King Philip’s War in 1674, white and Indian children sat together
in the same classrooms in the schools of the New England colonies. Until
the Indian wars of 1675, men such as Peter Fontaine and Colonel Byrd
in Virginia openly advocated intermarriage of the races as the best way
of building a common future, and so did Lawson of the Carolinas until
the disastrous Indian war of 1715 (ibid.: 374–5).
But even later assimilation did not always have to be forced on reluctant
Native American communities. Several consented, and for good reason.
The fate of those who resisted could not be ignored by those bearing
witness. In a context where force was increasingly the arbiter between
settler and native, those who were convinced of the superiority of the
force of the settler were keen to look for ways of becoming reconciled to
it. Leading that search were the Cherokee.
The Five Civilized Tribes moved to the west, most in the 1830s and the
Seminole in the 1850s, where they constituted themselves as “nations”
and proceeded to make changes under federal pressure. The Cherokee
and the Choctaw most readily adapted to the creation of new legal and
political institutions parallel with those in the United States. The con-
test between “civilization” and “tradition,” the former a synonym for the
white man’s way and the latter for the Indian way, triggered a protracted
debate in their ranks. The Cherokee, for example, built a modern prison,
whereas the Creek maintained that a prison was not fitting punishment
180 Mahmood Mamdani

for an Indian. Of all the nations of the Indian Territory, the Creek and
the Seminole most successfully excluded whites and tried to preserve
their traditional ways (Harring 1994: 58–9). The debate between “civi-
lization” and “tradition” had the most tragic outcome among the Creek:
this nation of perhaps ten thousand people had five civil wars between
1860 and 1908 – roughly one per decade – and on several occasions
had parallel governments, one representing those calling for adaptation
to “civilization” and the other those championing “tradition” (Harring
1994: 73–4, 76, 78–9, 82–92, 284).
Although the Treaties of Removal had included strong protection of
Indian lands and Indian autonomy, they also contained language that
would justify strong federal intervention when the times so demanded.
Take, for example, the Choctaw Treaty of 1830:

The United States under a grant specially to be made by the President of the US
shall cause to be conveyed to the Choctaw nation a tract of country west of the
Mississippi River, in fee simple to them and their descendents, to inure to them while
they shall exist as a nation and live on it . . .
The government and people of the United States are hereby obliged to secure
to the said Choctaw Nation of the Red People the jurisdiction and the government of
all the persons and property that may be within their limits west, so that no Territory or
State shall ever have a right to pass laws for the government of the Choctaw nation
or the Red People and their descendents; and that no part of the land granted
them shall ever be embraced in any Territory or State; but the US shall forever
secure said Choctaw Nation from, and against, all laws except such as from time
to time be enacted in their own National Councils, not inconsistent with the Consti-
tution, Treaties, and Laws of the United States; and except such as may, and which
have been enacted by Congress, to the extent that Congress under the Constitution are
required to exercise a legislation over Indian Affairs. (cited in Prucha 1976: 374–5)

Two kinds of guarantees were given in strong language. The first con-
cerned land, and the second persons and their possessions. Land was to
be granted “in fee simple” and not simply as a right of occupancy, and
more or less in perpetuity, “while they shall exist as a nation and live on
it”; the treaty also secured to the Choctaw nation “the jurisdiction and
the government of all the persons and property that may be within their
limits”; but these guarantees were subject “to the extent that Congress
under the Constitution are required to exercise a legislation over Indian
Affairs.” Here, then, was the clause that the Supreme Court would later
use to justify the right of congressional rule by decree over Indian tribes.
Creek treaties were compromised from the outset (Harring 1994: 65–
6, 69–70). To begin with, the federal government reserved the right to
decide who was a Creek and who was not. Given the historic alliance
between Creeks and Blacks, the federal government demanded the right
 Settlers and natives in North America 181

to decide whether a Black admitted into the Indian nation – or the off-
spring of a union between a Creek and a Black – was an Indian or not
(ibid.: 68ff, 97ff).
In the period that followed the Civil War, congressional rule by decree
tightened the squeeze so much that both the right of tribal property and
tribal property itself tended to go under as the era of removal gave way to
that of reservations, an institution that combined compulsory segregation
with forced assimilation.

The Dawes Act

At the heart of forced assimilation13 was a program of forced decol-
lectivization as a way of forcibly appropriating Indian lands. “Common
property and civilization cannot coexist,” a Commissioner for Indian
Affairs laid down succinctly in 1838: only separate holdings could breathe
enterprise, morality, and progress into Indians (Kiernan 1978: 38). Not
surprisingly, the debate on Indian rights got particularly hot when it came
to their right to sell off land. Every speculator who hoped to benefit from
the sale of Indian land championed the right of Indians to sell their land
(Williams 2005).
Probably the most ambitious attempt to force assimilation on Indian
tribes took place towards the end of the nineteenth century. Its legislative
instrument was the Dawes Act (the General Allotment Act) of 1887
and, following it, the Curtis Act of 1897. The two Acts applied to the
two halves of what was later to become the state of Oklahoma. The
earlier Dawes Act applied to the western part which was designated
Oklahoma Territory, and where the same reservation status pertained and
the same laws applied as those that governed Indian communities outside
Oklahoma. The eastern part was known as Indian Territory. There, the
legal situation was defined by early treaties and legal precedents that
explicitly pertained only to groups removed as part of the Indian Removal
Act. The Five Civilized Tribes in eastern Oklahoma were consistently
excepted from federal laws, such as the Major Crimes Act and the Dawes
Act, which had been applied to the Plains Indian tribes that were later
settled on reservations in the western part of the state. Also excepted were
the Osages, Miami, Peoria, and Sacs and Foxes in the Indian Territory,

13 On the process of assimilation, read Prucha (1984: 609–757); for primary documents,
Prucha (1973); see also Hoxie (2001) cited in Harring (1994: 13). On Locke: read
Pagden (1998) and Macpherson (1964). On Burke, critic of the Royal Proclamation of
1763, which imposed British authority to the rear of the American settlers as a barrier
to their westward expansion, which he saw as the fulfillment of a divine plan, see Hall
(2005: 74–7).
182 Mahmood Mamdani

the Seneca Indians in New York, and the strip of Sioux lands in Nebraska
(Prucha 1976: 255).
The exception was not popular among two groups of whites with a
direct interest in Indian Territory: on the one hand, reformers who had
long called for “the Americanization of all the Indians”; on the other, a
rapidly growing population of white farmers from the surrounding states
of Arkansas, Texas, and Kansas, who coveted “rich acres only partially
used by the Indians” (ibid.: 373). The exception turned out to be short-
lived; ten years after the Dawes Act, its provisions were embedded in a
new Act, the Curtis Act, and applied to the “civilized” tribes in Indian
Territory (ibid.: 96–7). Together, these Acts authorized the President
of the US to break up the land of any reservation into separate farms,
and confer citizenship alongside. Proceeding on the assumption that “it
is doubtful whether any high degree of civilization is possible without
individual ownership of land,” the stated objective of the Act was “to
individualize the Indian problem and thus lead more rapidly to a disin-
tegration of tribal relations.”
The Dawes Commission issued its report on November 20, 1894 (cited
ibid.: 190–5). The report began by narrating how the United States had
contracted different treaties and patents and “conveyed to the several
tribes the country originally known as the ‘Indian Territory’,” but how
a part of this same Territory had been ceded to the United States by
the same tribes for white settlement (and, we may add, for the establish-
ment of reservations for other Indian tribes being pushed west), and how
the remaining part was subject to stipulations contained in the treaties,
mainly that “these tribes were to hold this territory for the use and enjoy-
ment of all Indians belonging to their respective tribes,” and that “their
laws should not conflict with the Constitution of the United States.”
The report argued that these stipulations had been violated partly due to
the force of changing circumstances: “These stipulations naturally grew
out of the situation of the country at the time they were made, and of
the character of the Indians with whom they were made. The present
growth of the country and its present relations to this territory were not
thought of or even dreamed of by either party when they entered into
these stipulations.” This is why, though the project “seemed successful”
for a few years, the conditions that made for this success had “undergone
so complete a change” that success soon turned into failure. The report
concluded that the failure was “largely the result of outside considera-
tions for which neither is responsible and of the influence of forces which
neither can control.”
What, then, were these conditions that had undergone so complete
a change and that were not susceptible to control? It turns out they
 Settlers and natives in North America 183

were precisely the “progress” and the “civilization” which the set-
tler population and the United States had always preached to the
native population; the only difference was that, as reality rather than
promise, these changes were evident in their full-blown contradictoriness.
The Commission’s report detailed these developments, starting with
railways:

These governments consented to the construction of a number of railways

through the Territory, and thereby consented that they bring into the Terri-
tory all that is necessary in the building and operation of such railroads – the
necessary depots, stations, and the inevitable towns which their traffic was sure
to build up, and the large building which white men alone could develop and
which these railroads were sure to stimulate and make profitable.

The report invited whites to come and take employment in the Territory
(“they have, by their laws, invited men from the border states to become
their employees in the Territory, receiving into their treasuries a monthly
tax for the privilege of such employment”); encouraged the development
of lucrative commercial crops such as cotton (“In some sections of the Ter-
ritory the production of cotton has proved so feasible and profitable that
white men have been permitted to come in by thousands and cultivate it
and build trading marts and populous towns for the successful operation
of this branch of trade alone”), and other resources such as “vast and rich
deposits of coal” along with “large and valuable plants for mining coal”
and “vast pine forests.” Furthermore, “towns of considerable importance
have been built by white persons under leases obtained from Indians”; as
a result, “permanent improvements of great value have thus been built by
white citizens of the United States, induced and encouraged thereto by
the tribal governments themselves, and have become immovable fixtures
which cannot be taken away” (all cited ibid., italics mine).
The Commission then detailed the downside of progress, from “cor-
ruption of the grossest kind” to outright “fraud” making for “large for-
tunes,” a trend that none could check for “the courts of justice have
become helpless and paralyzed” in a context where “violence, robbery
and murder are almost of daily occurrence” resulting in nothing less than
“a reign of terror” in Indian Territory. The conclusion notwithstanding,
not only did the details of this charge sheet read like symptoms of “civi-
lization,” they seemed no different from what was happening throughout
the western territories in the throes of a gold rush.
In time-honored imperial fashion, the Commission took up the cudgels
on behalf of the poor and oppressed in Indian Territory. The first of these
were “the full-bloods who do not speak the English language” (italics
mine):
184 Mahmood Mamdani

The governments have fallen into the hands of a few able and energetic Indian
citizens, nearly all mixed blood and adopted whites, who have so administered
their affairs and have enacted such laws that they are enabled to appropriate to
their own exclusive use almost the entire property of the Territory of any kind
that can be rendered profitable and available.
In one of these tribes, whose whole territory consists of but 3,040,000 acres
of land, within the last few years laws have been enacted under the operation
of which 61 citizens have appropriated to themselves and are now holding for
pasturage and cultivation 1,237,000 acres. This comprises the arable and greater
part of the valuable grazing lands belonging to that tribe. The remainder of
that people, largely the full-bloods who do not speak the English language, are
excluded from the enjoyment of any portion of this land, and many of them
occupy the poor and hilly country where they get a scanty living from such
portions as they are able to turn to any account. (ibid.)

The second was the group of those who had before been slaves in the
tribes and who were to be adopted into the tribes and granted equal
rights by provision of the treaties when they were reaffirmed in 1866.
Their condition was the worst among the Chickasaws, among whom
“they are shut out of the schools of the tribe, and from their courts,
and are granted no privileges of occupancy of any part of the land for a
home, and are helplessly exposed to the hostilities of the citizen Indian
and the personal animosity of the former master.” As for the freedmen
among the Choctaw and the Cherokee, “although they have been adopted
according to the requirements of the treaties,” their condition was far
from satisfactory: “They are yet very far from the enjoyment of all the
rights, privileges and immunities to which they are entitled under the
treaties.” Finally, there was the group of poor whites: “thousands of white
children in this territory who are almost wholly without the means of
education, and are consequently growing up with no fitting preparation
for useful citizenship.”
Clearly, the record of the Indian tribes was being measured against
rhetorical standards rather than real practice among those said to be
custodians of “civilization.” When it came to its final recommendation,
the Commission pulled no punches:

These tribal governments have wholly perverted their high trusts, and it is the
plain duty of the United States to enforce the trust it has so created and recover
for its original uses the domain and all the gains derived from the perversions
of the trust or discharge the trustees . . . They have demonstrated the incapacity
to so govern themselves, and no higher duty can rest upon the Government that
granted this authority than to revoke it when it has so lamentably failed.

Having set the terms of the Commission with precisely these objectives
in mind, the federal government complied without hesitation. On July 1,
 Settlers and natives in North America 185

1898, the national governments of the Five Civilized Tribes were stripped
of their powers. The tribes of the Indian Territory were brought under
the full jurisdiction of the territorial courts of Oklahoma. There were
to be no tribal lands and no “Indian Territory” (Harring 1994: 71–3).
Having been promised that the embrace of “civilization” would secure
them autonomy, “civilization” was now advanced as the reason why they
could no longer remain autonomous.14
Congress moved in against the Five Tribes with the full force of plenary
powers. In 1906, Congress denied the legislatures of the Five Civilized
Tribes the right to meet more than thirty days per year, and their legisla-
tive action was made subject to veto by the President of the United States.
Legislative intervention even extended to federal power over tribal money.
Statutes provided that money due to tribes from tribal assets could be
appropriated at the discretion of Congress (Cohen 1942). Congress even
claimed “the ultimate authority to determine who was a tribal member
for purposes of distributing property, annuities and trust money, and
how that money was spent.” Finally, Congress gave to itself the right to
authorize the consolidation of tribes, no matter what their history or eth-
nology (Newton 1983–4: 223–4). In their generality, these powers were
no different from the powers assumed by the colonial governor of the
South African province of Natal over the Zulu tribes towards the end of
the nineteenth century (Mamdani 1996).
But the leaders of the Five Civilized Tribes persisted; their goal was
to preserve their identity within the federal system by promoting sepa-
rate statehood for Indian Territory. Officials of the Creek, Cherokee and
Choctaw (supported by the Chickasaw as well) met in a joint conven-
tion of tribes at Eufaula in the Creek Nation on November 18, 1902, and
adopted a statement against union with Oklahoma Territory. At the same
time, individual tribes also made separate protests.15 Three years later,
in the summer of 1905, new agitation for a separate state developed, and
strong arguments were presented. Indians pointed to the Atoka Agree-
ment of 1897 which spoke of the preparation of Indian nation lands “for
admission as a state of the Union.” In July the heads of four of the five
tribes issued a call for a constitutional convention to meet on August

14 Even against all odds, these native communities survived. Take the example of the
Creek. “The Curtis Act attempted to outlaw their entire tribal existence. ‘Final Rolls’
were made for the abolishment of the Creek Nation. And the Muscogee people were to
become citizens without a Native State . . . And yet the Creek nation survived.” In 1979,
the Creek Nation ratified a new tribal constitution organized under the provisions of
the Oklahoma Indian Welfare Act (Strickland 1985–6: 746–7).
15 The Eufaula resolutions are printed in Congressional Record, 36: 567. For action by the
various tribes, see Senate Document No. 143, 59-1, serial 4912.
186 Mahmood Mamdani

21 at Muskogee. All residents of Indian Territory were invited to vote

for delegates, and many whites cooperated with the Indians. The con-
vention drew up a constitution for a proposed state of Sequoyah, and
in an election of November 7 the document was ratified by a vote of
56,000 to 9,000 (though not more than half the qualified voters went
to the polls).16 Once again, Congress blocked any further development.
Although bills for admission of the separate state were introduced in
both Houses, no action was taken, and the whole movement simply died.
According to a special census of 1907, the population of the new state
stood at 1,414,177, slightly less than half of which was in the former
Indian Territory. Of this total, there were 101,228 persons in the Five
Civilized Tribes, including intermarried whites and the freedmen, and
15,603 other Indians (Prucha 1984: 756–7).
The Dawes and Curtis Acts gave the appropriate authorities in reser-
vations the power to grant to deserving Indians a “trust patent” to his
allotment of land, whereby the land title was held in trust by the US for
twenty-five years, after which the premises were conveyed to the Indian
by fee simple patent. The Act initially delegated to the Bureau of Indian
Affairs (BIA) authority to allot 160 acres of tribal land to each head of
household and 40 acres to each minor, and was soon amended to provide
for allotments of 80 acres of agricultural land, or 160 acres of grazing
land, to each tribal member. The May 8, 1906, Burke Act provided
that the allotted Indian might become a citizen only when he became an
unrestricted landowner at the end of twenty-five years (Macleod 1928:
539). Forced allotment, according to the Act, was a deliberate means of
destroying “communistic” tribal culture and forcing Indians to assim-
ilate into the mainstream of US life as farmers and ranchers (Harring
1994: 143). Surplus land was then sold to whites. This is how the land
in Indian possession fell by more than half over the next half-century
(Kiernan 1978: 103).
The Dawes and Curtis Acts had several consequences, but four are
of particular concern for our inquiry. The first was a drastic reduction
in land under Indian control. Indian landholdings were reduced from
138 million to 52 million acres. More than 26 million acres of allotted
land were transferred out of Indian hands after they passed out of trust,
by fair means and foul (Wilkinson 1987: 8–9, 20). Sixty million of the
86 million acres lost to Indians by the allotment regime were due to the
surplus lands facet of the 1887 Act (ibid.: 19–20). By 1985, according
to records of land in the Old Indian Territory at the Bureau of Indian

16 See Senate Document No. 143, 59-1, serial 4912; the memorial on p. I-27, and the
Constitution on pp. 47–8.
 Settlers and natives in North America 187

Affairs, only 65,000 acres remained in tribal hands and less than a million
acres in individual Indian hands (Strickland 1985–6: 733).
Second, instead of turning Indians into farmers, the Dawes Act led to
leasing of land. By 1898, 112,000 of the 140,000 allotted acres had been
leased. This involved two-thirds of the men with allotted land (Prucha
1984: 873). Third, through the sale and lease of land, the Dawes Act
opened up reservations to non-Indian settlement. Finally, as the own-
ership and population of reservations changed, “the BIA moved in as
the real government” (Wilkinson 1987: 21). Ironically, the Dawes Act
turned Indians from collective to individual wards and the paternalism of
the federal government grew instead of diminishing, “until the bureau-
cracy of the Indian Service dominated every aspect of the Indians’ lives”
(Prucha 1984: 759).
The fact is that the grant of individual titles did not erode the power of
federal agents to incarcerate Indians for infractions of reservation rules
and to enforce their “civilization program” with military power: “School
masters continued to separate children from their parents. Religious orga-
nizations continued to operate with a level of federal support that clearly
violated the First Amendment of the US Constitution. Authorities con-
tinued to break up unauthorized religious activities and destroy sacred
objects. Officials could even ‘withhold rations’ from tribal members who
opposed them” (Castile and Bee 1992: 38).
Not even the allotment law’s grant of citizenship to Indians taking up
individual landholdings ended federal supervision over native communi-
ties. In 1910 the Supreme Court spoke for the entire bureaucracy when
it declared that “Congress, in pursuance of the long-established policy of
the Government, has the right to determine for itself when the guardian-
ship which has been maintained over the Indian shall cease. It is for that
body, and not for the courts, to determine when the true interests of
the Indian require his release from such condition” (US Supreme Court
1910, cited in Castile and Bee 1992: 39). The simple fact was that, cit-
izens or not, Indians were wards of Congress and only Congress could
terminate that colonial relationship.

V. Compulsory segregation
Towards the end of the twentieth century, the Native American writer
Vine Deloria Jr. reflected on the sharply contrasting historical experiences
of Native and African-Americans (Deloria 1988). He explained that the
difference arose from two different objects coveted by “civilization”: if the
Indian was an exotic “wild animal” who had to be tamed by “civilization,”
the Black person was a “beast of burden” who had to be kept in his place.
188 Mahmood Mamdani

The white man adopted two basic approaches . . . He systematically exc-

luded blacks from all programmes, policies, social events, and economic
schemes . . . With the Indian the process was simply reversed . . . Indians
were . . . subjected to the most intense pressure to become white . . . The white
man forbade the black to enter his own social and economic system and at the
same time force-fed the Indian what he was denying the black. Yet the white man
demanded that the black conform to white standards and insisted that the Indian
don feathers and beads periodically to perform for him.

The difference, according to Deloria, was this: unlike Black Amer-

icans who bore the brunt of compulsory segregation, Native Ameri-
cans confronted the demand that they assimilate. The two responses
were shaped by two different situations and contrasting objects of
desire. Whereas Black Americans signified labor, Native Americans sig-
nified land. Whereas compulsory segregation went alongside a regime of
coerced labor, forced assimilation began with the disbanding of native
communities and the fragmentation of communal land into individual
agricultural plots and the designation of the commons as surplus land
that could be sold cheaply to land-hungry settlers. One of the most enthu-
siastic champions of assimilation was Thomas Jefferson, who called for a
program to civilize Indians by making Christians of them, getting them
to take up farming, and teaching them literacy (Hall 2005: 432).
But Deloria was only partly right. The difference between the expe-
rience of the African-American and that of the Native American is rel-
ative. The fact is that the Native American experience was not single
but differentiated: whereas some Native American communities were
forcibly dismantled, others were just as forcibly confined to reservations.
Both lost land, but with different outcomes. Native Americans experi-
enced both forced assimilation and forced segregation. The institution
that combined forced assimilation and compulsory segregation in one
was the reservation.
As the reservation era gathered pace, forced assimilation and compul-
sory segregation turned out to be like Siamese twins, more complemen-
tary halves of a single policy than alternatives. This became clear as the
nineteenth century drew to a close, and the Bureau of Indian Affairs
took control of reservations. The turning point was when the House of
Representatives decreed in a rider to the Appropriations Act of 1871
that henceforth “[n]o Indian nation or tribe within the territory of the
United States shall be acknowledged or recognized as an independent
nation, tribe, or power with whom the United States may contract by
treaty.” Henceforth, Indian law became more a matter of domestic law,
and Indians were regarded as subjects to be governed, rather than foreign
 Settlers and natives in North America 189

nationals.17 The formal end of treaty-making went alongside “the con-

scious intention thereby to denigrate the power of the chiefs” (Prucha
1976: 329). Not only did treaties with Indians cease to receive consti-
tutionally backed protection, the federal government even abandoned
a rhetorical commitment to obtaining Indian consent. The era of con-
quest was over and the era of peace was dawning. If conquest had been
associated with a wholesale slaughter of Indians, peace was associated
with forced appropriation and semi-captivity. Semi-captivity, in turn,
was identified with a set of institutions. The first was the reservation; the
second was the BIA.
The Peace Policy was elaborated during the post-Civil War tenure of
President Grant. Congress established the Board of Indian Commission-
ers on April 10, 1869. When he met the new Board on May 27, 1869,
Grant delineated the main lines of his Indian policy: to abandon the
treaty system, replace it with reservations, and secure Indian titles to
“such lands as they could be induced to cultivate” (ibid.: 36). The Board
began its tenure with an all-round critique of government Indian policy,
issued in its first annual report of November 23, 1869: that the treatment
of Indians had been “unjust and iniquitous beyond the power of words
to express,” that the history of the government’s relations with the tribes
had been “a shameful record of broken treaties and unfulfilled promises,”
and that the primary blame for Indian wars lay with frontiersmen. “Para-
doxical as it may seem,” the commissioners acknowledged, “the white
man has been the chief obstacle in the way of Indian civilization” (Prucha
1973: 38).
That this bit of history was meant for the archives rather than as
guidelines for future policy became clear when the President’s Secretary
of the Interior, Columbus Delano, spelt out the fourfold objectives of his
“peace policy” in a statement in 1873. The first was to segregate Indians
on reservations where they could be kept away from frontier settlements.
The second was forcibly to assimilate them, by “all needed severity,” so
as “to punish them for their outrages according to their merits, thereby
teaching that it is better to follow the advice of the government, live
upon reservations and become civilized, than to continue their native
habits and practices.” The third was to ensure that funds for procuring
Indian supplies were properly spent. And the fourth was to secure the
cooperation of religious organizations to procure “competent, upright,
faithful, moral and religious” agents to run the reservations, as well as

17 VII Newton, 206 Act of March 3, 1871, ch. 120, #1, 16 Stat. 566 (codified at 25 USC
#71 (1976)).
190 Mahmood Mamdani

churches and schools meant to bring them “the comforts and benefits of
a Christian civilization” (Prucha 1976: 31–2).

Reservations
As an institution designed to hold a subject population captive, the reser-
vation had a long history, starting with seventeenth-century Ireland and
the Puritan colonies in North America. The modern reservation, how-
ever, came into being only in the second half of the nineteenth century.
Its inauguration signaled the final defeat of Native Americans in North
America.
To install a reservation system required more or less the complete sub-
jugation of natives. When Queen Elizabeth’s nobles received lands and
governing powers in Ireland, the condition was that they conquer the
natives in the assigned territories. Many concluded that it would be nec-
essary to exterminate the Irish. But this was not possible “since neither
the Crown not any feudal lord was wealthy enough to finance the work
of extermination” (Macleod 1928: 162–3). The alternative to extermi-
nation, Chichester suggested in 1607 (also the year of the settlement of
Jamestown, Virginia), was a reservation system for the leaderless natives.
King James used this plan in 1600 in his final scheme for the colonization
of Ulster. According to the plan, the natives of the region, divided into
six British counties, were ordered to leave the country immediately or to
gather together on reservations. But they were also left with little choice
in the matter, given that they had nowhere else to go, and that the order
called for death for any person found outside reservations after a certain
date (ibid.: 163–4).
The reservation system was born of a compromise: the king wanted
the tribal system of government utterly and completely abolished,
but the bishops counseled compromise to avoid bloodshed. Not only were
the strategies pursued by Puritans in North America remarkably similar
to those executed by English Protestants in Ireland, early Indian treaties
were also remarkably like the treaties with Celtic tribes. The tribes were
required to submit to the crown’s sovereignty, and stop internecine war-
fare, but they were allowed to retain their tribal organization for adminis-
trative purposes. Three different kinds of prohibition were enforced. The
first provided for a virtual ban on the sale or transportation of liquor,
permitting only home brew, save permission for the chiefs to import a
little good whisky or wine for their personal use. The second targeted
specific cultural practices identified as barbaric: trial marriage, a native
custom, was forbidden; also the tribes, hitherto Catholics – first Celtic,
then Roman – were required to submit in religion to the established
 Settlers and natives in North America 191

episcopal church, to build churches and support a clergy. Oliver

Cromwell suppressed religious freedom from 1651 to 1660 as rigorously
as did his fellow Puritans in America. Anglicans, Jews, and Catholics
were forbidden to worship except in Presbyterian or Congregationalist
churches. When Cromwell led his army into Ireland, he was determined
to wipe out both the tribal system and the native religion, thereby forcing
natives to submit both to English rule and to Congregationalist reli-
gious belief and organization (ibid.: 170). Finally, there was the strictest
prohibition against the cultural and intellectual leadership of the tribes,
combined with provisions to create an alternative intellectual leadership.
So the bards or seannachies were forbidden to sing to remind the tribes-
men of their ancient, primitive glory, and the chiefs of their age-long
genealogies. The chiefs were even forbidden to support them, and the
bards were to be doomed to starvation or manual labor – and silence.
At the same time, the eldest children of the chiefs and nobles were to
be educated in English schools, under the direction of the crown but at
the expense of their parents (ibid.: 167–8). From the outset, compulsory
segregation went hand in hand with forced assimilation.
The first reservations in the English colonies were made up of sur-
vivors from among the conquered Pequot. These Christianized Indians
were called the “Praying Indians.” They were allocated a bounded piece
of land as their domicile under colonial jurisdiction. The first of these
reservations was the “town” of Natick. Over time, these reservations
became like magnetic islands attracting freed slaves seeking refuge. The
freedmen intermarried, acquired land, and settled down, in the process
bringing accumulated skills in European agriculture to the Indian popu-
lation (ibid.: 389, 392).
The modern American reservation system dates from the middle of
the nineteenth century when William Medill, Commissioner of Indian
Affairs, renounced the concept of a barrier state in his annual report
of 1848 and suggested two reservations (called colonies) for the tribes
standing in the way of expansion: “one north, on the dead waters of
the Mississippi and the other south, on the western borders of Missouri
and Arkansas” (Trennert 1975: 30). In other words, the reservation system
emerged as a successor to Indian Removal and an alternative to a Native Amer-
ican state. The following year, on March 3, 1849, the Senate approved
the creation of the Department of the Interior to which it transferred
Indian Affairs. There ensued a division of labor between the army and
the Indian Office, the former retaining responsibility for the hostile tribes
and the latter for the subdued tribes (ibid.: 41–2). The first of the agents
was sent out in 1849, and a single consolidated Indian Department was
created by the federal government in 1851 (ibid.: 57–60).
192 Mahmood Mamdani

Gold-rush California became a state of the Union in 1850, and soon

afterwards became an implementation site for the untested policy of
reservations. Federal officials first proposed reserves for the Golden State
in a series of eighteen treaties negotiated in 1851 and 1852 with numer-
ous bands and villages of natives. These agreements assigned Indians to
specific territories in the state, promised a subsidy of food and livestock,
and offered instruction and supervision for at least five years. When the
US Senate refused to ratify the treaties on July 8, 1852, the Office of
Indian Affairs proceeded to found reservations in a manner that would
not require Senate ratification. It appointed Edward F. Beale, a promi-
nent frontier adventurer and office-holder, as superintendent of Indian
affairs in the state, and instructed him to establish a “system of colo-
nization” for the purpose of “making useful, our present worthless, and
troublesome Indian population.” Beale and his successors focused atten-
tion away from the coastal strip, where Hispanic settlement had either
eliminated or absorbed most natives, to the great interior valley and the
northern coast, where contact with Indians had been limited before the
1840s. This is where the first three experimental reserves – El Tejon,
Nome Lackee and Fresno farm – were first started (Findlay 1992: 15–
17). Six more reservations or farms founded by agents of the Office of
Indian Affairs followed in 1852–5, and yet another six in 1856–70; but
only three survived up to 1871 (ibid.: 18). In an analysis of the early
California reservations, John M. Findlay identifies three factors as shap-
ing their demise: starvation, violence by encroaching white settlers, and
incompetence or corruption on the part of Indian agents. “The institu-
tion was too weak to provide them with food and clothing, too weak to
protect them from white attacks, and too weak to make them stay . . . If
Indians tried to drive off or kill the white man’s stock, they were threat-
ened, beaten or shot.” Without adequate security and subsistence, most
Indians who went to California’s first three reservations “either died there
or departed,” whereas the rest avoided the reservations altogether (ibid.:
23).
As officials looked for familiar models with which to identify this new
institution, “reserves were variously likened to farms, fortresses, asylums,
schools, final refuges, and Spanish missions.” Even if the context made
for a shift of meanings, one model, that of the asylum, endured more
than any other (ibid.: 14). If anything, the reservation was designed
as a movable asylum. Since Indians were marched from one site to
another, early reservation policy “assumed not only a high degree of
impermanence but also a substantial amount of coercion.” Indian agents
looked to the military both for an example of how to organize life within
the reservation and for assistance in forcing the Indians to do as they
 Settlers and natives in North America 193

were told. But the role of the military was inconsistent with the reserva-
tion’s reform and welfare functions (ibid.: 21–2). To facilitate this role,
California denied civil and political rights to Indians as soon as it was
admitted to the Union in 1850 – and authorized and financed military
campaigns against Indians, as part of a larger removal effort (ibid.: 25).
Findlay says that early reservations suffered the same fate as asylums
which had been established to rehabilitate deviants and regenerate sta-
bility, but by the late 1850s were “losing their reformist mission and
becoming storage bins for society’s outcasts.” In another decade, by the
end of the Civil War, reservations too had become places to segregate
Indian “misfits” from the rest of society, even if they had been designed
for an opposite purpose – as a means to incorporate Native Americans
into the mainstream (ibid.: 19). In 1855, US Army Captain E. D. Keyes
praised the superintendent at Nome Lackee because, instead of holding
expectations about the Indians’ ultimate survival and acculturation, he
reportedly understood that his chief purpose “must be to deprive the Red
Man of his power to do mischief” during the few years he had remain-
ing on earth. “We ought . . . to act,” Keyes wrote, “on the determined
certainty that the aborigines of this Country will soon become extinct”
(ibid.: 20). In this vision, domestication was not seen as an alternative,
but as a transit facility to extermination. The early California reservation,
which seemed like an early version of the industrial concentration camp,
was discredited and supplanted by a new generation of smaller reserves
and rancherias during the late nineteenth and early twentieth centuries
(ibid.: 28).
Not all settlers were in favor of a reservation policy. Some, as in Texas,
preferred a more immediate final solution: extermination. When the fed-
eral government proposed the establishment of a reservation in Texas as
a way of separating the races and securing peace, newspapers such as the
Democratic Telegraph and Texas Register were bitterly opposed and called
for a war of extermination. Some Texan residents called for the removal
of all Indians from Texas. The Texas legislature approved a reservation
policy on February 16, 1854, but no sooner were reservations estab-
lished than “armed parties invaded the reservations and began a system-
atic slaughter of the tribesmen.” Finally the government relented, and in
1859 authorized the complete removal of all Texas tribes north of the Red
River (Trennert 1975: 68, 77, 92–3). Reservations were eventually set
up elsewhere, in 1854 (Missouri), 1867 (Cheyenne and Arapaho), and
1868 (Sioux). The policy was generalized in the context of the “Indian
Wars” on the western frontier between 1860 and 1890, when the US
“decided to put a full reservation system into operation by defeating or
starving the tribes into compliance” (ibid.: 158, 191, 196).
194 Mahmood Mamdani

The federal policy of confining Indian tribes to remote reservations

was initiated by Abraham Lincoln during the Civil War, and was inten-
sified by President Grant under what he claimed was a Peace Policy.
In his 1869 inaugural address, Grant promised to pursue policies that
would “civilize” Native Americans and lead to their “ultimate” citizen-
ship. As preparation for the policy turnaround, he fired all existing Indian
agents and hired new ones, chosen by Quakers, and later also by other
Christian denominations. Thus began the most extensive federal-level
collaboration between the state and churches in the history of the United
States – one that made “repudiation of native religion and ways of life,
and acceptance of middle-class American Christianity with its attendant
customs, official prerequisites for admission to US citizenship” (Smith
1997: 318–19).
Eric Foner has noted that, when completed in 1869, the transcontinen-
tal railroad “expanded the national market, facilitated the penetration of
capital in the West, and heralded the final doom of the Plains Indians.” In
this west “lay millions of acres of fertile and mineral-rich land roamed by
immense buffalo herds that provided food, clothing, and shelter for a pop-
ulation of perhaps a quarter of a million Indians, many of them members
of eastern tribes forced inland two centuries before from the east coast,
and moved again earlier in the nineteenth century to open the Old North-
west and Southwest to white farmers and planters.” This jackpot was now
targeted by Grant’s Peace Policy, to be executed by military and civil-
ian officials, “nearly all” of whom “shared a common presumption: that
the federal government should persuade or coerce the Plains Indians to
exchange their religion, communal form of property, and ‘nomadic’ way
of life for Christian worship and settled agriculture on federally super-
vised reservations. In a word, they should surrender most of their land and
cease to be Indians.” Towards this end, generals employed methods they
had perfected during the Civil War, destroying the “the infrastructure of
the Indian economy,” even if that meant destroying the buffalo, and abro-
gating the treaty system that dealt with Indians as independent nations.
This step was “strongly supported by railroad corporations, which found
tribal sovereignty an obstacle to construction, and by Radical Republi-
cans, to whom the traditional system seemed a form of local autonomy
incompatible with the uniform nationality born of the Civil War.” By
the time Grant left office, the world of the Plains Indian had been shat-
tered by a combination of the railroad, farms, ranches, Indian Wars, and
reservations. The population of Plains Indians, hitherto, estimated at a
quarter of a million, had diminished drastically, while the settler popu-
lation of the Middle Border states (Minnesota, the Dakotas, Nebraska
and Kansas) had grown from 300,000 in 1860 to well over two million in
 Settlers and natives in North America 195

twenty years. On the ashes of the Plains Indians thus arose, in the words
of Eric Foner, “a new agricultural empire” (Foner 1988: 21, 462–3).
The reservation policy, in the words of Kenneth D. McRae, “con-
demned the Indian and the métis to a non-competitive, sheltered, subsi-
dized, and essentially neglected and uneducated existence on reservations
of his own” (McRea 1964: 264). But they were not places “of his own.”
Rather, constructed for him and her, reservations were sites where the
Indian was not only detained (“neglected”) but also made the butt of
a whole array of coercive measures designed to subdue (“civilize”) him
and her into full obedience (“citizenship”).

The Bureau of Indian Affairs

The reservation was presented to white society both as a way of segre-
gating the races, and as a hothouse for the accelerated “civilization” and
assimilation of Indians. The Indian agent was to be the key catalyst in this
process. This is how the Indian Office’s regulations for the Indian service
defined the role of an agent: “The chief duty of an agent is to induce his
Indians to labor in civilized pursuits. To attain this end every possible
influence should be brought to bear, and in proportion as it is attained,
other things being equal, an agent’s administration is successful or unsuc-
cessful” (Prucha 1984: 643, 645). Created to administer the assimilation
of the Indian tribes into the American nation, the BIA turned out to
be the instrument of their forced segregation into semi-captivity. BIA
agents ruled the reservations without any legally imposed restraint dur-
ing the plenary power era, which went from 1877 to the 1930s (Harring
1994: 4, 56). Harring says BIA’s claim that the tribes were not ready for
the full responsibilities of US citizenship rationalized the near-absolute
administrative powers exercised by its agents (ibid.: 175).
The BIA intervened in the daily lives of Indian peoples through two
extra-legal institutions: the Indian police created in 1879, and the Courts
of Indian Offenses established in 1883. The Indian police wore the uni-
form of army privates and had their roots in the army’s Indian scouts
(ibid.: 1982). Under the command of the agent, the Indian police func-
tioned as “quasi-military forces” that were “a substitute for army control
of the reservations” (Prucha 1976: 201). The strongest objections against
having a standing police force under the control of the agent came from
the chairman of the House Committee on Indian Affairs in 1880:

This provision turns him [the Indian] over, bound hand and foot, to the agents.
These men had authority before almost without restriction, except as they are
restricted by the want of physical force. Now we give them eight hundred men
196 Mahmood Mamdani

armed and equipped and thus the fullest authority is allowed with fearful power
to execute not known laws, but the will of the agent. (ibid.: 207–8)

The Indian police were soon joined by Courts of Indian Offenses,

introduced by Secretary of the Interior Henry M. Teller, who in Decem-
ber 1882 called attention to several cultural degradations: “the continu-
ance of old heathenish dances” which he insisted led to a war spirit and
demoralized the young, the practice of polygamy, and the influence of the
medicine men who kept children from attending school and promoted
heathenish customs. Together, said Teller, these constituted “a great hin-
drance to the civilization of the Indian.” Not surprisingly, the jurisdiction
of the courts set up by Teller included all three: dances, polygamy, and
medicine men (Prucha 1984: 646–8). This is how the “Rules for Indian
Courts” defined these offenses:
Dance, etc – any Indian who shall engage in the sun dance, scalp dance, or
war dance, or any other similar feast, so-called, shall be deemed guilty of an
offense . . .
Practices of Medicine Men – Any Indian who shall engage in the practices
of so-called medicine men, or who shall resort to any artifice or device to keep
the Indians of the reservations from adopting and following civilized habits and
pursuits, or shall adopt any means to prevent the attendance of children at school,
or shall use any arts of a conjurer to prevent Indians from abandoning their
barbarous rites and customs, shall be deemed guilty of an offense . . .
Misdemeanors – And provided further, that if an Indian refuses or neglects to
adopt habits of industry or to engage in civilized pursuits or employments, but
habitually spends his time in idleness and loafing, he shall be deemed a vagrant
and guilty of a misdemeanor. (cited in Strickland 1985–6: 728)

The judges were given specific jurisdiction over these practices, and
other misdemeanors. Composed of the first three officers in rank of the
police force, or, if the agent did not approve them, of three other per-
sons selected by the agent, the courts were a further “extension of the
agent’s authority.” By 1900, approximately two-thirds of the agencies
had courts. The exceptions were the “civilized” tribes with tribal govern-
ments: the Five Civilized Tribes, the Indians of New York, the Osages,
the Pueblos and the Eastern Cherokees (Prucha 1976: 208–9). For the
tribes considered “uncivilized,” First Amendment rights were drastically
curtailed. Indian religions were banned upon threat of criminal prose-
cution in the Courts of Indian Offenses set up by the Bureau of Indian
Affairs (Newton 1983–4: 227).
Whereas every reservation had a police force, not every reservation
had a court system. Many agents never bothered to create them. But
where they were created, the judges were appointed by and served at
the pleasure of the agents. Occasionally they were police officers, but
 Settlers and natives in North America 197

often they were influential chiefs. The degree of apparent assimilation

was a major criterion in the selection of judges, and agents’ reports are
replete with descriptions of judges as “Christian,” “wearing white man’s
clothing,” or “monogamous.” The agents placed this legal process in the
hands of progressive Sioux, factionalizing the Sioux legal system. These
judges did not set out to implement “Indian law” but rather a set of BIA
rules. The rules for Indian courts were revised twice, in 1891 and 1904,
but the substance remained that of the original 1883 rules (Harring 1994:
186, 191).
With or without the assistance of courts, agents were determined
to demonize Indian culture. Herring concludes from a reading of BIA
annual reports that the struggle to outlaw traditional dancing was long
and drawn out (ibid.: 274). Illustrative of the BIA’s views is an obser-
vation in the BIA annual report of 1883: “It was found that the longer
continuance of certain old heathen and barbarous customs, such as the
sun-dance, scalp-dance, war-dance, polygamy & c., were operating as a
serious hindrance to the efforts of the Government for the civilization of
the Indians” (ibid.: 186).
In practice, assimilation turned into a culture war against practices
that distinguished Indian tribes from white neighbors, such as long hair,
painted faces, and Indian dress. Commissioner William A. Jones pointed
out “a few customs among the Indians which, it is believed, should be
modified or discontinued.” The first was the wearing of long hair by men,
which the Commissioner believed was “not in keeping with the advance-
ment they are making, or will soon be expected to make, in civilization.”
“Certainly all the younger men should wear short hair,” Jones directed,
“and it is believed that by tact, perseverance, firmness, and withdrawal
of supplies the superintendent can induce all to comply with this order.”
Commissioner Jones also directed action against painting the face and
against dances and Indian feasts, which “were simply subterfuges to cover
degrading acts and to disguise immoral purposes.” Finally, superinten-
dents and agents were directed to encourage the wearing of citizen’s
dress instead of Indian costumes and blankets (Prucha 1984: 764). If
Indians employed by the Indian service refused to comply, they could be
discharged and their supplies stopped. If they became obstreperous, “a
short confinement in the guard house at hard labor, with shorn locks,
should furnish a cure.”
Successive commissioners upped the ante, calling for the renaming
of Indians. Commissioner Thomas J. Morgan argued in a circular of
March 19, 1890 that there will be “needless confusion” and “consid-
erable ultimate loss to the Indians if no attempt is made to have the
different members of a family known by the same family name on the
198 Mahmood Mamdani

records and by general reputation.” He contended that “among other

customs of the white people it is becoming important that the Indi-
ans adopt that in regard to names.” He condemned the translation of
Indian names into English, a practice which often resulted in “awkward
and uncouth” names. He was also against “the habit of adopting sobri-
quets given to Indians such as ‘Tobacco,’ ‘Mogul,’ ‘Tom,’ ‘Pete,’ etc.
by which they become generally known.” He suggested that “unusually
long and difficult” Indian names be shortened arbitrarily, and authorized
the substitution of English names for Indian ones too difficult for whites
to pronounce and the introduction of Christian given names before the
surnames (ibid.: 673–4).
In reservations that lacked functioning Courts of Indian Offenses –
which was in most – the agent simply jailed whoever he wanted to on
his personal authority or he asked the military to do so. The military
believed it had authority over Indians because they remained within the
United States after their defeat in the Indian wars as a conquered people
subject to military discipline (Harring 1994: 193). Harring says there is
no question that many thousands of Indians were so detained for offenses
ranging from murder to resisting allotment.
The full brunt of the effort of reeducation was directed toward Indian
children who were shipped away from the reservation or brought together
at reservation schools. The philosophy was most simply expressed by
Richard Pratt, the founder of Carlisle School: “Kill the Indian and save
the man” (Strickland 1985–6: 729). Vaunted as a key benefit of “civiliza-
tion,” as it was at the World’s Columbian Exposition in 1893, the regime
at school led to “heartbreaking experiences that drove Native children to
commit suicide and to run away repeatedly” (Huhndorf 2001: 201).
In reality, the agent ran an occupation regime. Besides wielding a police
force and influencing courts where they existed, he had powers to control
the ebb and flow of resources to the reservation. When tribal councils
were dissolved in 1906 as a correlative to forced individualization of
property, his grip on power was further tightened (Kiernan 1978: 103).
“By 1915, the agent, now called a superintendent, had control of all
welfare services, and anyone who found fault with his doings could be
set down as a malcontent” (ibid.: 101). More often than not, reservations
appeared as “mammoth poorhouses rather than nurseries of civilization”
(Prucha 1976: 223).
The Board of Indian Commissioners boasted in 1901 that not only was
the object of the Indian Office to “make all Indians self-supporting, self-
respecting, and useful citizens of the United States,” but also that unlike
other divisions of the government, “the Indian Bureau should always
aim at its own speedy discontinuance” (Report of the Board of Indian
 Settlers and natives in North America 199

Commissioners, 1901, pp. 4–5; 1906, pp. 8–9, cited in Prucha 1984,
779). But the trend, as Prucha showed, was to the contrary: the employ-
ees in the Indian Office increased from 115 in 1900 to 262 in 1920, and
the communications received increased from 62,691 to 261,486 over the
same period (ibid.: 781). In practice, the reservation was more the site
of social decay and social crisis than progress. Forced segregation was
perhaps the most important factor fostering this crisis. Forced isolation
robbed the Indians’ way of life of both vitality and meaning, reducing
it to a set of disconnected rituals without the possibility of innovation.
Not only were the buffalo and the wide stretches of free territory gone, it
was also the case that social organization, built up through generations to
protect fundamental interests, no longer met crucial problems, but was
preserved as if in a museum. To be sure, there were substitutes, such as
agricultural life and Christianity, but they were both unfamiliar and not
always relevant. In this spiritual and social vacuum came the rapid devel-
opment of movements such as the Ghost Dance religion, “engendered by
the wish for the old security and distaste of the white man’s civilization,
and eagerly sought and accepted by one after another of the distraught
plains tribes” (Macleod 1928: 530).
The first major critic of forced assimilation was the applied anthropol-
ogist John Collier, one of the founders of the Interamerican Indigenista
Institute (whose first executive director was Manuel Gamio) (Hall 2005:
223–4). Collier called for an end to “monopolistic and autocratic control
over person and property by a single Bureau of the Federal Government”
and for protection of Indian property rights, for respect of “elemen-
tary rights guaranteed to other Americans by the Constitution or long-
established tradition,” and for the use of cultural pride and “native social
endowments and institutions” in the education of the Indians (Prucha
1984: 806).
Taking office as Commissioner of Indian Affairs on April 21, 1933,
Collier modernized Indian Affairs into a version of British-style indirect
rule in tropical Africa. To do so, he brought in a flock of anthropologists
to shape policies and programs. Collier tried to reconfigure Indians as a
tribal government with a tribal economy, even if the group in question
was in practice organized into bands and villages (ibid.: 1010).
The centerpiece of Collier’s policy was the Indian Reorganization Act
(IRA), an ambitious package of interrelated reforms designed to halt and
reverse the destruction of Native American communities. The title of the
bill he introduced was self-explanatory:

A bill to grant the Indians living under Federal tutelage the freedom to organize
for the purposes of local self-government and economic enterprise; to provide for
200 Mahmood Mamdani

the necessary training of Indians in administrative and economic affairs; to con-

serve and develop Indian lands; and to promote the more effective administration
of justice in matters affecting Indian tribes and communities by establishing a
Federal Court of Indian Affairs. (ibid.: 957)

Collier made it clear that while a tribe that rejected the Act would not
be discriminated against, there were sure to be consequences born of
official neglect: “it will merely drift to the rear of the great advance open
to the Indian race. It will stand still and will probably continue to lose its
lands while those who accept the Act, its benefits and responsibilities can
preserve and increase their lands and will move forward” (ibid.: 964). He
set up an Indian Organization Division within the Bureau to reorganize
tribes according to newly adopted constitutions, bylaws, and corporate
charters. Within ten years, ninety-three tribes, bands, and Indian
communities had adopted constitutions and bylaws and seventy-three
were granted charters (ibid.: 967–8). The Act authorized preferential
hiring of Indians in the Indian Service. The number of permanent
Indian employees in the service rose from a few hundred in 1933 to
4,682 in 1940. These included eight superintendents, 251 professionals,
935 clerical workers and approximately 3,475 in other skilled jobs
(ibid.: 992). Of all Bureau employees 78 percent were Indians or Alaska
natives by 1980. Within a decade, Collier had successfully created a
combination of self-government and a comprehensive administrative
paternalism (ibid.: 1086).
To stem the continuing erosion of the Indian land-base, all further
allotments were prohibited; the trust status on existing holdings was
extended indefinitely and $2 million a year was appropriated for the
purchase of additional land. At the same time, to help them regain a
measure of autonomy and self-sufficiency, tribes were encouraged to
organize their own governments and to establish business corporations,
with a revolving loan fund of $10 million set up to provide capital for
plant and equipment. But this vision of tribal autonomy was promptly
undermined by making every important decision subject to approval by
the Secretary of the Interior (Wilson 1998: 347–8; Churchill 2001: 292).
The Act’s first major critic was Scudder Mekeel, director of the Applied
Anthropology Unit in the Office of Indian Affairs, 1935–7. In a broad-
ranging critique, Mekeel focused on two salient aspects of the Act. First,
that “it closely resembles the British policy of ‘indirect rule’ in that the
native political and social organization is strengthened by utilizing it
for administrative purposes.” The bureaucracy may be Indianized, but
Mekeel felt the strengthened bureaucracy would only have the effect of
furthering a trend begun under the old Indian Service: “Practically all
tribes, aside from certain groups in the Southwest, have seen their native
 Settlers and natives in North America 201

form of government disintegrate and disappear under the bureaucracy of

the Indian Service.” Even if “many ‘Tribal Councils’ have been organized
under Indian Office sponsorship,” they “have been almost completely
controlled by the reservation superintendent.” Second, he pointed out
that “the Indian Reorganization Act continues the policy of segregating
the Indian population from the white – a policy inherent in the reservation
system itself” (Mekeel 1944: 209–17). Here, too, Mekeel was spot on.
After Collier left office in 1945 and Harold Ickes (his strong backer)
resigned as Secretary of the Interior in 1946, the policy pendulum swung
back to forced assimilation as the executive branch of the government
joined Congress in a massive drive to assimilate Indians once and for
all (Prucha 1984: 1013). President Truman appointed Dillon S. Myer,
who had been in charge of the wartime internment and resettlement of
Japanese Americans, as Commissioner of Indian Affairs. Myer’s forceful
methods included interfering in tribal elections, selling land without the
tribe’s consent, supporting a bill which would have allowed Indian Service
employees to carry guns and arrest Native Americans without warrant,
and removing a rare BIA superintendent who had upheld Indian rights
(Wilson 1998: 362).
The thrust of the Truman administration’s Indian policy was known as
the Termination Policy. Adopted in August 1953, it declared Congress’s
intention to make Native Americans “subject to the same laws and enti-
tled to the same privileges and responsibilities as . . . other citizens of the
United States” by “freeing” them from “all Federal supervision and con-
trol.” A few days later, Congress passed Public Law 280, which extended
state control over all Indians – with a few specified exceptions – in five
states, and empowered other states to enact similar legislation. From
1954 to 1960 some fourteen recognized tribes with reservations were
terminated, often without their consent. Most were small, impoverished
communities who had little idea of what was happening to them, but a
few – most notably the Menominee of Wisconsin and the Klamath of
Oregon – were large, wealthy tribes with considerable natural resources,
who fought the government’s decision for some years (ibid.: 361–2).
The Termination Policy was repudiated by the Nixon administration in
1970.
The longer-run impact of Collier and the 1934 Indian Reorganization
Act was to move federal policy on reservations, even if fitfully, “away
from seeing them as temporary impounding points for ultimate assimila-
tion and toward a recognition of their status as permanent homelands,”
thus aiming to enhance their economic development and stabilize their
internal organization as communities. It is for this reason that, notwith-
standing the brief resurgence of assimilationist and abolitionist sentiment
epitomized by the “termination” resolution of 1953, most observers are
202 Mahmood Mamdani

agreed that Collier’s influence on Indian policy survives even half a cen-
tury after he left office. The ultimate legislative expression of this policy
is the Indian Self-Determination Act of 1975, which seeks both to make
the Native American peoples entirely self-administering and to elimi-
nate their dependent status as administered communities (Castile and
Bee 1992: 5–6). As George Pierre Castile and Robert L. Bee observe in
their introduction to State and Reservation: “New explanations are needed
why the reservation system and federal–Indian relations have seemingly
become permanent” (ibid.: 8).

VI. Citizen wards: citizens without rights

If Marshall could see no future for Native American communities beyond
a perpetual colonial condition, Supreme Court justices such as Taney
explored the possibility of individual futures beyond colonial subjugation.
Taney wrote in Dred Scott (1856) that even though Indian tribes had not
historically formed a part of the colonial communities and had never
“amalgamated” with them, and though the Indians were “uncivilized,”
“the course of events” had brought them “under subjection” to the white
race, “and it has been found necessary, for their sake as well as our own,
to regard them as in a state of pupilage, and to legislate to a certain extent
over them and over the territories they occupy.” Following a trajectory
well laid by the Marshall model’s guardian–ward principle, Taney noted
that Indians could only become naturalized as citizens under the laws
of the United States by abandoning their ties to their savage tribes and
taking up “abode among the white population” (Williams 2005: 17–18,
72). The alternative Taney offered would today be termed ethnocide, the
erasure of their group identity in favor of assimilation into white society
as individual citizens of the United States.
Citizenship was first thought of with reference to the “civilized” tribes,
and as an alternative to the promised two-state solution. In its very first
annual report of 1869, the Board of Indian Commissioners urged citi-
zenship for Indians considered “civilized” – the Five Civilized Tribes in
the Indian Territory – and “wards of the government” as the appropriate
legal status for “uncivilized Indians” (Prucha 1976: 40). A law confer-
ring citizenship on the “civilized” tribes was passed in 1901. But when a
Select Committee of the Senate was sent to investigate conditions in the
Indian Territory in connection with the Act of 1906, it remarked on how
this law seemed not to have taken any effect:

Yet notwithstanding this express legislative naturalization, Congress in its sub-

sequent legislation, and the Department of the Interior, acting under such
 Settlers and natives in North America 203

legislation, has apparently ignored entirely this established citizenship, and in

nearly every instance has treated the questions arising within the Five Civilized
Tribes as though no such act had ever been passed and as though the Indians
were still in the broadest sense wards of the Government. (Prucha 1984: 899)

Even as Woodrow Wilson paid homage to the right of self-

determination globally, the end of his first term coincided with a for-
mal Supreme Court ruling that confirmed the second-class citizenship
of Indians, and the fact that the terms of Indian citizenship were not
a matter of fundamental right but would continue to be determined at
the pleasure of Congress. In United v. Nice, Justice Van Devanter wrote
that citizenship was “not incompatible with tribal existence or continued
guardianship, and so may be conferred without completely emancipating
the Indians or placing them beyond the reach of congressional regulations
adopted for their protection” (cited in Smith 1997: 463).
The formal grant of citizenship to Native Americans took place over
two decades, from 1901 to 1924. The first to be granted citizenship were
members of the Five Civilized Tribes, who together counted for more
than one third of persons officially termed “Indian” in the US. Roughly
another third were naturalized between 1907 and 1921: in 1907, all
remaining Indians in Oklahoma; in 1919, all Indians who had served in
the US army in the First World War; and in 1921, all remaining members
of the Osage tribe. The last third of Indians were naturalized with the
Act of Congress of June 2, 1924, which provided “that all non-citizen
Indians born within the territorial limits of the United States be, and
they are hereby declared to be, citizens of the United States” (Macleod
1928: 540; Amar 2006: 439).
As citizens, Indians form a category of their own. At least three different
types of limitations set them apart from other native-born American
citizens. The first and central limitation to Indian citizenship is that
Indians remain legal wards of the US government. This fact was spelt
out by the Eighth Circuit Court in a 1912 case: “The word ‘Indian’ in the
statute describes a person of Indian blood. The word ‘citizen’ describes
a political status” (Newton 1983–4: 228). Based on the ward status of
the Indian, there were three types of Indian citizens, signifying three
different stages in the process of assimilation. The least assimilated are
Indians with a tribal membership. They partake of the collective resources
of the tribe as members of the tribe. Without allotment of individual
plots of land, they are known as the “unallotted” Indians. The “allotted”
Indians, those who have received individual plots of land from what
was previously tribal property, are further divided into two categories:
“restricted” and “unrestricted.” The second group of Indian citizens
204 Mahmood Mamdani

comprise the “allotted” but “restricted” Indians: these are Indians who have
received their allotment of tribal land, but their land title is held in trust by
the US government for a period of twenty-five years. The third group of
Indian citizens is that of Indians who are “allotted” and “unrestricted”:
with an unrestricted allotment, only they are full citizens of the US.
These were the “unrestricted” Indians. The first two groups of Indians
are as wards of the federal government: the first group as tribal wards,
and the second group as individual wards. In 1920, for example, there
were 240,000 restricted Indians. Of these 125,000 were unallotted, and
115,000 allotted (Prucha 1976: 543–4). The arrangement was formally
sanctioned by the Supreme Court which held that “citizenship is not
incompatible with tribal existence or continued guardianship, and so may
be conferred without completely emancipating the Indians or placing
them beyond the reach of Congress and regulations adopted for their
protection” (Macleod 1928: 540–1). The ward status means that Indians
are subject to limitations as Indians – no matter the degree of their
assimilation in white society or their citizen status. To take an example,
individual Indians can be subjected to liquor laws, whether or not they
have severed tribal relations and whether or not they are on a reservation.
In Hallowell v. United States [221 US 317 (1911)], the defendant “had
been active in county and state governments as judge, county attorney,
county assessor, and director of a public school district,” but “he was
still subject to the law because of his status as an Indian.”
Second, unlike other persons born in the United States, Indians are not
citizens by birth. The Court held in Elk v. Wilkins [112 US 94 (1884)] that
the tribal Indian owed allegiance to the tribe from birth and therefore was
not born “subject to the jurisdiction” of the United States (Johnson 1973:
993). Citizenship did not necessarily mean enfranchisement. Indians
continued to be prohibited from voting in a number of western states,
and not until 1948 did Arizona and New Mexico extend the franchise to
them (Prucha 1984: 794). Even then, only “unrestricted” Indians had the
right to vote. By the time Macleod wrote in 1928, only some 29,000 or so
had qualified as voters (Macleod 1928: 541). The rest comprised Indian
Country, the name given to settler America’s first colony, comprising
200 reservations, with resident agents in about ninety of them. “Today
the official ‘Indian Country’ comprises the aggregate of reservation lands
unallotted and of allotted lands the title to which is still held in trust by
the government” (ibid.: 541).
Third, Indians are the only citizens without rights constitutionally
guaranteed and enshrined in the Bill of Rights. Just as citizenship for
Indians came to have a lax meaning – not considered incompatible
with federal powers of guardianship, nor with restriction on the right to
 Settlers and natives in North America 205

alienate property – so did civil rights come to be defined in a loose and

informal way when Congress passed the Indian Civil Rights Act of 1968
(ICRA). Instead of the full application of the Bill of Rights as contained
in the Constitution, Title 1 of the Act speaks of the protection of a select
and specific set of individual rights. Title II selectively incorporates por-
tions of the Constitution as a Bill of Rights for Indians. Interestingly,
“the Act omits some constitutional guarantees including the prohibition
against laws respecting the establishment of religion and the right to
indictment by a grand jury.” It then limits other guarantees: “the right to
counsel is only available at one’s own expense and the right to jury trial
is limited to those accused of crimes punishable by imprisonment.” Title
III directs the Secretary of the Interior to draft and recommend to the
Congress a model code governing Courts of Indian Offenses. The code
is to “assure that any individual being tried for an offense by a Court
of Indian Offenses shall have the same rights, privileges, and immuni-
ties under the United States Constitution as would be guaranteed any
citizen of the United States being tried in a Federal Court for any sim-
ilar offense.” But the code is not binding on tribal courts; it is only a
recommendation (Johnson 1973: 998–9).
That the Act lacked an enforcement mechanism became clear when the
Supreme Court in Santa Clara Pueblo v. Martinez (1978) bluntly denied
that it carried any federal remedies, finding that only tribal forums are
available to enforce the ICRA. Just as with citizenship, here too we have
the anomaly of a federal civil rights act that provides limited federal
remedies to enforce it. Application of the full Bill of Rights, argued
Prucha, would upset traditional governing practices in Indian Country,
including the quasi-theocracies that ran some of them (Prucha 1984:
1107). Tradition, in this case, referred to a colonial mode of governance,
not to the historical mode of governance in pre-Columbian America. At
its heart is the fact that all Indians are as wards of Congress. The trust
relationship means that they possess no constitutionally guaranteed rights
in relation to Congress; put differently, the enforceable rights Indians
do possess are against the executive branch, a fact that has driven the
narrowly conceived struggle for Indian rights in recent times (Newton
1983–4: 233).
Ironically, the colonial paternalism that is the essence of the guardian–
ward relationship between the federal government and Indian commu-
nities has been defended in the language of self-determination by both
sides. Even the resurgence of militancy in the decade of the 1960s that
began with the American Indian Chicago conference in 1961 and con-
cluded with the formation of the American Indian Movement in 1968
in Minneapolis did not change this substantially. Take, for example, the
206 Mahmood Mamdani

“policy for the future” as summed up by the American Indian Policy

Review Commission – established in the aftermath of Indian demonstra-
tions at the Bureau of Indian Affairs in 1972 and at Wounded Knee in
1973 – in its final report of May 17, 1977. The report begins with an
elaboration of the dual “Foundations of Federal Indian Law”:

1. That Indian tribes are sovereign political bodies, having the power to deter-
mine their own membership and power to enact laws and enforce them within
the boundaries of their reservations, and
2. that the relationship which exists between the tribes and the United States
is premised on a special trust that must govern the conduct of the stronger
toward the weaker. (Prucha 1984: 1115, 1117–18)

The report proceeds to define “the Trust Responsibility” at the heart of

the relationship between the federal government and the Indian commu-
nity, presenting a total colonial paternalism as some kind of a warm and
comforting wraparound administrative blanket:

1. The trust responsibility to American Indians extends from the protection

and enhancement of Indian trust resources and tribal self-government to the
provision of economic and social programs necessary to raise the standard of
living and social well being of the Indian people to a level comparable to the
non-Indian society.
2. The trust responsibility extends through tribe to the Indian member, whether
on or off the reservation.
3. The trust responsibility applies to all United States agencies and instrumental-
ities, not just those charged specifically with administration of Indian affairs.

How difficult it has been to think of alternatives to colonial paternalism

is evident from the fact that even Indian activists have been unable to
come up with a solution that goes beyond a reformed version of that
same colonial relationship.

BIBLIOGRAPHY
Amar, Akhil Reed. 2006. America’s Constitution: A Biography. New York: Random
House.
Castile, George Pierre and Robert L. Bee (eds.). 1992. State and Reservation:
New Perspectives on Federal Indian Policy. Tucson and London: University of
Arizona Press.
Cherokee Nation v. Georgia. 1831. 30 US 1; 8 L. Ed. 25; 1831 US LEXIS 337.
Decided, pp. 16.
Churchill, Ward. 2001. A Little Matter of Genocide: Holocaust and Denial in the
Americas, 1492 to the Present. San Francisco: City Lights Publishers.
 Settlers and natives in North America 207

Cohen, Felix. 1942. Handbook of Federal Indian Law. Washington DC: Depart-
ment of the Interior.
Congressional Record, XXXVI: 567.
Deloria, Vine, Jr. 1988.“The Red and the Black,” in Custer Died for your Sins: An
Indian Manifesto. Norman: University of Oklahoma Press, 168–96.
Dred Scott v. Sandford. 1856. 60 US 393.
Ex Parte Crow Dog. 1883. 109 US 557.
Findlay, John M. 1992. “An Elusive Institution: The Birth of Indian Reservations
in Gold Rush California,” in State and Reservation: New Perspectives on Federal
Indian Policy, ed. George Pierre Castile and Robert L. Bee. Tucson and
London: University of Arizona Press, 13–37.
Foner, Eric. 1988. Reconstruction: America’s Unfinished Revolution. 1863–1877.
New York: Harper and Row.
Hall, Anthony. 2005. The American Empire and the Fourth World: The Bowl With
One Spoon, vol. I. Montreal: McGill-Queens University Press.
Harring, Sidney L. 1994. Crow Dog’s Case: American Indian Sovereignty, Tribal
Law, and United States Law in the Nineteenth Century. Cambridge: Cambridge
University Press.
Hoxie, Frederick E. 2001. A Final Promise: The Campaign to Assimilate the Indians,
1880–1920. Lincoln: University of Nebraska Press.
Huhndorf, Shari M. 2001. Going Native: Indians in the American Cultural Imagi-
nation. Ithaca, NY: Cornell University Press.
Johnson, Kenneth W. 1973. “Sovereignty, Citizenship and the Indian.” Arizona
Law Review 15 (4): 974–1003.
Johnson v. M’Intosh (1823) 21 US 543.
Kiernan, V. G. 1978. America: The New Imperialism, From White Settlement to
World Hegemony. London: Zed Press.
Macleod, William Christie. 1928. The American Indian Frontier. London: Kegan
Paul.
Macpherson, C. B. 1964. The Political Theory of Possessive Individualism. Oxford:
Oxford University Press.
McRae, Kenneth D. 1964. “The Structure of Canadian History,” in The Founding
of New Societies, ed. Louis Hartz. New York: Harcourt Brace, 219–62.
Mamdani, Mahmood. 1996. Citizen and Subject. Princeton, NJ: Princeton Uni-
versity Press.
Mekeel, Scudder. 1944. “An Appraisal of the Indian Reorganization Act.” Amer-
ican Anthropologist 46 (2): 209–17.
Newton, Nell Jessup. 1983–4. “Federal Power over Indians: Its Sources, Scope
and Limitations.” University of Pennsylvania Law Review 132 (2): 195–288.
Pagden, Anthony. 1998. Lords of all the World. New Haven, CT: Yale University
Press.
Prucha, Francis Paul. 1973. Americanizing the American Indian. Lincoln: Univer-
sity of Nebraska Press.
1976. American Indian Policy in Crisis: Christian Reformers and the Indian, 1865–
1900. Norman: University of Oklahoma Press.
1984. The Great Father, the United States Government and the American Indians,
vol. II. Lincoln: University of Nebraska Press.
208 Mahmood Mamdani

Smith, Rogers M. 1997. Civic Ideals: Conflicting Visions of Citizenship in U.S.

History. New Haven, CT: Yale University Press.
Strickland, Rennard. 1985–6. “Genocide-at-Law: An Historic and Contempo-
rary View of the Native American Experience.” University of Kansas Law
Review 34: 713–55.
Talton v. Mayes. 1895. 163 US 376.
Trennert, Jr., Robert A. 1975. Alternative to Extinction: Federal Indian Policy and
the Beginnings of the Reservation System, 1846–51. Philadelphia, PA: Temple
University Press.
US v. Joseph. 1876. 94 US 614.
US v. Rogers. 1846. 45 US 567.
US v. Sandoval. 1913. 231 US 28.
Wilkinson, Charles F. 1987. American Indians, Time, and the Law: Native Societies
in a Modern Constitutional Democracy. New Haven, CT: Yale University Press.
Williams, Jr., Robert A. 2005. Like a Loaded Weapon: The Rehnquist Court, Indian
Rights, and the Legal History of Racism in America. Minneapolis: University
of Minnesota Press.
Wilson, James. 1998. The Earth Shall Weep: A History of Native America. New
York: Grove Press.
Worcester v. Georgia. 1832. 31 US 515.
Part IV

What role for representative quotas?

8 Perverse consequences? The impact
of quotas for women on democratization
in Africa

Shireen Hassim

I. Introduction
The emergence of a global political movement agitating for the adop-
tion of gender quotas – what Drude Dahlerup (2005) has described as a
“quota fever” – has reinvigorated feminist theorization of representation.1
The fundamental questions of representation have been posed again:
who do elected officials represent; does increased representation improve
marginalized groups’ access to decision-making institutions; and under
what conditions does increased representation lead to improved policy
outcomes for marginal groups? Two sorts of disillusionment have coin-
cided in global political discussions about women and representation.
On the one hand, feminists in older democracies despair about the slow
progress of women in elected bodies. On the other, in new democracies,
feminists are skeptical of the willingness of pro-democracy movements
to effectively address the interests of women. A remarkable consensus
has been forged between these two groups in favor of new mechanisms
to increase women’s representation, most notably the use of some form
of gender quota.
Debates on quotas have advanced in two distinctive phases thus far.
In the first phase, feminists were concerned with justifying the idea that
special mechanisms were needed to redress the democratic deficit in rep-
resentation. Most opponents of quotas do indeed focus on normative
liberal objections, whether these relate to the relative privileges of groups
over individuals, or the value of equal opportunities, or the relative weight
of women’s claims over those of other groups. I deal very briefly with
this set of normative objections, arguing that many of these have been
effectively countered by democratically based feminist arguments. One
An early version of the arguments in this chapter was first elaborated in Shireen Hassim,
“Women, Parliaments and Gender Equality in Africa: Exploring the Virtuous Circle of
Representation,” in Bauer and Britton (2006).
1 See the special issue on representation in Politics and Gender 2 (4) (December 2006); also
International Feminist Journal of Politics 10 (3) (2008).

211
212 Shireen Hassim

outcome of the first phase of outlining the problem of representation is

that substantial expectations were raised as to what quotas might be able
to achieve with regard to deepening liberal democracy at least. For many
advocates of gender quotas, there is a causal relation between greater
descriptive representation and substantive, or interest, representation, a
relationship I have described elsewhere as a “virtuous circle of repre-
sentation” (Hassim 2006a). This assumption is equally widely held by
women’s movements, leading to a global campaign for equal representa-
tion. Building on this assumption, in the second phase of research, the
majority of researchers focused attention on analyzing the kinds of mech-
anisms that could be employed to ensure that women gained fair access
to the processes and institutions of representation – or, to use the new
parlance, to “fast-track” representation (Krook 2006).
In this chapter, rather than debate the efficacy of different types of
quotas, I suggest that the foundational questions with regard to repre-
sentation are worth attention with the benefit of some years of experience
with the implementation of quotas in Africa. First, is the expectation that
quotas lead to increased democratization of the institutions of political
representation (political parties and legislatures in particular) borne out
by empirical evidence in new democracies? Second, do quotas facilitate
the representation of women’s interests and if so, which interests are more
likely to be advanced? What evidence is there that increasing women’s
representation enhances the extent of democracy in political systems?
I argue that the use of the fast-track model of women’s representation
opens new concerns about whether quotas are indeed more likely to
advance feminism’s democratic aims. I draw attention to two factors gen-
erally neglected in the quota literature: first, political cultures of respon-
siveness and accountability of elected officials, and second institutional
maturity. Drawing on research on women and parliaments in Africa, I
argue that expectations that quotas will democratize political systems in
ways that slower incremental changes might not is questionable. Quotas,
I argue, can have profoundly negative effects on deepening democracy
when adopted in contexts where the key institutions of democracy and
human rights are weakly developed or absent, and where elected political
actors are weakly accountable to electoral constituencies. Quotas may
fast-track women’s representation but they do not fast-track equality or
democracy.

II. The persistent deficit in representation

Women’s underrepresentation in elected office is depressingly similar
in political systems around the world. Even in advanced industrial
 The impact of quotas for women on democratization 213

Table 8.1 Women in national parliaments by

region, 2005

Percentage
Region women

Nordic countries 40.1

Americas 18.6
Europe (OSCE excl. Nordic) 16.8
Asia 15.0
Sub-Saharan Africa 14.8
Pacific 13.2
Arab states 6.5

Note: From Inter-Parliamentary Union (IPU) (2005).

OSCE = Organization for Security and Cooperation in
Europe.

countries, the deficit in representation remains large despite decades

(in some cases over a century) of women’s suffrage. As Table 8.1 shows,
if one factors out the Nordic countries, Europe and the Americas have
made as few strides in including women in legislatures as Asia, where
there are supposedly greater social barriers to women’s participation
in public life. Although early explanations of women’s inclusion in the
public sphere emphasized modernization as the crucial determinant –
levels of socioeconomic development, women’s participation in the paid
labor force and the erosion of traditional attitudes to women’s public
roles (Matland 1998) – by the twenty-first century it is apparent that
the link between socioeconomic development and political representa-
tion is tenuous. With respect to women’s political representation (but
not other aspects of gender equality), barriers and biases in institutions
(in electoral and party systems especially) and resistant political cul-
tures appear to override the gains made in socioeconomic status. In
these contexts, there is a growing view that quotas might create a desir-
able exogenous “shock” and introduce rupture in institutions (Baldez
2006). The proponents of quotas focus on the mechanism as a remedy
for political exclusion in advanced democracies, but they assume the
existence of a range of democratic conditions: respect for freedom of
speech and association, well-institutionalized party and legislative sys-
tems and horizontal and vertical guarantees of accountability of elected
representatives.
Feminists studying “new democracies” – those that emerged in the
transition from various forms of authoritarianism in Africa, East and
Central Europe, Asia and Latin America from the 1980s – express equally
214 Shireen Hassim

strong disillusionment with liberal democracy. Transitions offered unique

opportunities, both ideological and practical, to redesign institutions to
facilitate inclusion. Transitions also created space to debate the under-
lying values of the “new” society and to reshape constitutional and
political principles in ways that might accommodate oppressed and
marginal groups (Waylen 2007). The broader acceptance of gender
equality in international human rights discourse facilitated the inclusion
of some kind of gesture in this direction by the new political leader-
ship (even if only rhetorical). These processes of institutional design and
constitutional-ideological debate propelled a stronger expectation that
inclusion of women in decision-making would indeed advance gender
equality.
The global consensus on the need to increase women’s representation
was underpinned in many countries by significant levels of women’s par-
ticipation in pro-democracy movements – indeed, in many Latin Amer-
ican and African countries women’s organizations ensured the survival
of these movements in contexts of extreme authoritarianism. One note-
worthy lesson from women’s involvement in such movements was that
liberation movements that depended on women in their struggles against
authoritarianism were nevertheless quite likely to abandon their support
for women’s liberation once they came to power. The demand for quo-
tas was, therefore, in many cases a demand for a guarantee that history
would not repeat itself.
The demand for quotas (which may be reserved seats, candidate
quotas, or intraparty quotas) as a mechanism for breaking through
entrenched structural and cultural barriers to women’s entry into elected
office is thus a consequence of a frustration with liberal democracy itself.
Various claims are made in relation to what quotas may achieve. Quota
demands are based on the assumption that increasing the representation
of women is a democratic good in itself. This seems fairly uncontestable;
women’s consistent underrepresentation in legislatures undoubtedly con-
stitutes a democratic deficit. The now extensive research on women and
parliamentary politics demonstrates that at the very least the rapid inclu-
sion of substantial numbers of women (usually through quotas) does
change perceptions of women’s role in politics and may even shift gen-
dered modes of deliberation within parliament. In both old and new
democracies, similar arguments are made for the democratic importance
of increasing women’s representation. Women’s claims to representa-
tion entail the claim to equal inclusion in the nation-state. This claim,
in its most minimal form, draws attention to the denial of suffrage
to women as well as to the gap between women’s distribution among
voters and their presence as elected members of legislatures. Women’s
 The impact of quotas for women on democratization 215

underrepresentation is undoubtedly a function of systematic historical

discrimination against and marginalization of women as a group.
Demands for inclusion can signal different understandings of the body
politic: for some feminists, equality means women taking their place
in the modernist, liberal democratic state as it is currently constituted.
For others, inclusion is a necessary but insufficient step in the process of
redistributing resources in a more equitable manner. For the latter group,
demands for inclusion emphasize the representation of women as a group
and the recognition of women’s historical disadvantages. Where minimal-
ist claims for inclusion see equality of access and level playing fields as
sufficient redress of the democratic deficit, maximalist claims demand
more active intervention in electoral systems and institutional designs to
guarantee women’s inclusion. In either case, there is a consensus that
representation matters, that the pattern of women’s underrepresentation
devalues the quality of citizenship, and that interventions are needed to
facilitate the election of more women in legislatures. Feminists promoted
quotas as mechanisms for democratizing both political parties (by giving
women guaranteed access to candidate selection processes) and legisla-
tures (by forcing open space for the inclusion of new legislative and policy
agendas).
A significant core of the proponents of quotas make a further, more
powerful claim that increasing the number of women in decision-making
institutions increases women’s influence over decision-making in regard
to national budgets, policy priorities, and the ideological direction of
government policies, skewing these in ways that redress inequalities.
For example, the Women’s Environment and Development Organiza-
tion (WEDO), the NGO spearheading the global campaign for parity in
parliaments, argues that increasing women’s representation will increase
the likelihood that issues such as childcare, violence against women,
and unpaid labor will become policy priorities (Beaman et al. 2007). In
many cases, it can indeed be shown that women parliamentarians intro-
duce new policy issues onto the agenda. Jane Mansbridge argues that
“descriptive representation by gender improves substantive outcomes for
women in every polity for which we have a measure” (Mansbridge 2005:
622). The idea of a “virtuous circle of representation” also underpins
many women’s movements’ views that achieving higher levels of rep-
resentation will extend political activism into the state and allow them
policy leverage. An influential team of researchers attached to the Interna-
tional Institute for Democracy and Electoral Assistance (IDEA) sweeps
away theoretical objections to quotas and argues for the importance of
“fast-tracking” women’s representation, saying “women’s impatience is
growing worldwide” and that the “incrementalist” Scandinavian route
216 Shireen Hassim

to the empowerment of women can no longer be (if it ever was) the

strategic model (Dahlerup and Friedenvall 2005; Dahlerup 2006). In
part their view is justified; if the assumed relationship between mod-
ernization and women’s representation is patently false, there seems lit-
tle reason why women should wait for the “correct conditions” to be
achieved.
The quota literature lists a number of institutional factors that under-
pin the success of quotas. These include an electoral system based on
proportional representation with a closed list, “placement mandates” for
ensuring that women are placed in winnable positions on the list (for
example the “zebra” or “zipper” model proposes that every alternate
candidate on a list is a woman), multimember districts, and sanctions
for the non-enforcement of quotas (Dahlerup 2005). These institutional
reforms accompany virtually all campaigns for quotas.
Nevertheless, for those concerned with deep democracy the con-
cern is how to ensure that descriptive representation will have virtuous
effects, that is, that it will increase the effective representation of women’s
interests. Based on experiences in Scandinavian parliaments, the notion
emerged that increases in representation exhibit an impact on interest
representation when the numbers of women in legislatures reaches a
critical mass of 30 percent of the legislature. This is based on an argu-
ment by Drude Dahlerup (1988) that representation of a marginal group
shifts from token representation to influential representation when this
critical mass has been achieved. Many studies of European democracies
do indeed show that policy concerns relating to care-work and women’s
rights are more likely to be introduced when women attain a signifi-
cant proportion among decision-makers. The idea of a critical mass has
become a kind of magic number, seen in demands for 30 percent repre-
sentation of women in global campaigns, and as a target set in various
national and transnational feminist documents. For example, the Bei-
jing Conference set a target of 30 percent of women in parliaments by
2005, and this target was subsequently adopted by the Southern African
Development Community (SADC). The targets have strategic value in
that they constitute a measurable standard by which government com-
mitments to gender equality can be assessed. They have had some effect,
though perhaps not as spectacular as is often claimed: in 1995 women
accounted for 11.3 percent of members of parliament worldwide; by
2005 this number had increased to close to 16 percent (see Table 8.2)
(Ballington 2005: 25).
Quotas undoubtedly address the claim to inclusion and the impor-
tance of descriptive representation in dramatic fashion. They instan-
taneously impose a significant number of new actors onto institutions,
 The impact of quotas for women on democratization 217

Table 8.2 Women’s representation in African countries

Year quota % women in national

Country Type of quota introduced legislature, 2003

Burkina Faso Voluntary party 2002 11.7

Botswana No quota – 11.1 (2004)
Cameroon Voluntary party 1996 8.9
Côte d’Ivoire Voluntary party – 8.5
Djibouti Legislative 2002 10.8
Malawi No quota – 13.9 (2004)
Mali Voluntary party – 10.2
Mauritius No quota 17.1 (2005)
Mozambique Voluntary party 1994 30.0
Namibia Voluntary party 1997/9 29.2
South Africa Voluntary party 1994 29.8
Tunisia Voluntary party – 11.5
Djibouti Legislative 2002 10.8
Eritrea Constitutional 1995 22.0
Kenya Constitutional 1997 7.1
Morocco Legislative 2002 10.8
Niger Legislative 2000 1.2
Rwanda Constitutional 2003 48.8
Somalia Constitutional 2001 10.0
Sudan Legislative 2000 9.7
Tanzania Constitutional 2000 22.3
(reserved seats)
Uganda Constitutional 1989 24.7

Note: Adapted from Tripp (2006); Dahlerup and Friedenvall (2005); and
Dahlerup (2006).

disrupt male monopolies, and may have downstream effects on the gender
composition of party elites through the “contagion” effect (Matland and
Studlar 1996). There is considerable evidence to show that institutional
trust and legitimacy are enhanced by greater descriptive representation
(Htun 2005). There are also studies that demonstrate the democratiz-
ing effects of women’s inclusion on the everyday features of institutions
(Bauer and Britton 2006). Proponents of quotas argue that the mere
presence of women in parliaments shifts “the patriarchal demeanor of
political institutions” (www.gender.ac.za/50/50, accessed September 20,
2006) and forces institutions to recognize women. Indeed women parlia-
mentarians in South Africa are fond of retelling what happened in 1994
when the number of women MPs rose from 14 to 111. There was only
one small toilet for women in the vicinity of the general assembly; the air-
conditioned temperature of the House was set at a level that, in the words
218 Shireen Hassim

Table 8.3 Highest ranking countries in

terms of women’s representation, 2003

Percentage
Rank Country women

1 Rwanda 48.8
2 Sweden 45.0
3 Denmark 38.0
4 Finland 36.5
5 Norway 36.4
6 Costa Rica 35.1
7 Iceland 34.9
8 Netherlands 34.0
9 Germany 32.2
10 Argentina 30.7
11 Mozambique 30.0
12 South Africa 29.9
13 Seychelles 29.4

Note: From Ballington (2004).

of one MP, was “comfortable for overweight men in suits”; and House
sitting times were organized on the assumption that children were cared
for by stay-at-home wives. Mansbridge makes an even stronger claim
for descriptive representation in arguing that descriptive representatives
have greater advantages in communicating their constituents’ interests
with other representatives – “they can speak with authenticity and be
believed” (Mansbridge 2005).
While there is political convergence among feminists on the impor-
tance of representation, a considerable reserve of skepticism about quo-
tas remains in mainstream political science. In Western liberal democ-
racies, and particularly in the United States, the idea of quotas raises a
series of normative objections. The first objection is procedural. Quotas
are opposed because they place external restrictions on voters’ rights to
choose their representatives. This is a weak objection, in my view. Quota
supporters may counter this objection by pointing out that political par-
ties are in any case gatekeepers to political office, and that decisions about
the composition of candidate slates frequently employ “informal quotas”
in relation to regional, occupational, or ethnic criteria.
Stronger objections relate to the argument that quotas promote essen-
tialist and homogenized views of social groups, reinforce stereotypes, and
rigidify identities. A closely related objection is that quotas are based on
the problematic assumption that descriptive representation is in some
 The impact of quotas for women on democratization 219

sense better than other forms of representation – that is, that women are
better represented by women. In responding to these objections, support-
ers of the minimalist approach to quotas frequently argue that support
for more women in legislatures does not necessarily imply that their role,
once elected, is to specifically represent women. As Krook (2006) points
out, these are sex quotas, not feminist quotas. Rather, the concern is to
include new actors and perspectives so that the link between maleness
and political office is irrevocably broken. To the extent that feminists
are concerned with the representation of women’s interests, it is argued,
this will be advanced by using the access created by quotas to build on
other forms of political action, including alliances with male democrats,
to shift policy agendas. Others, of a maximalist persuasion, argue that
historical discrimination does create a minimal set of group interests
among women that transcend race, class, ethnic, and other differences
(Phillips 1995). These include the elimination of overtly discriminatory
practices (that is, ensuring the implementation of commitments to gen-
der equality) as well as attention to the specific needs of women arising
out of gendered divisions of labor. These ought be to be understood as
“shared experiences” rather than as a set of essentialized identities or
interests.
Although not necessarily based on essentialist understandings of
women, quota demands do tend to emphasize the creation of collective
identity; they rest on the successful articulation of women’s group-based
interest in entering arenas of power. Quotas are a means of achieving
recognition; indeed they are best understood as a form of symbolic pol-
itics. Quota advocates are agnostic on the issue of the outcomes of rep-
resentation (e.g. Dahlerup and Friedenvall 2005; Nanivadekar 2006);
they are willing to concede that once representatives begin to claim that
they represent women’s interests the notion of women as a homogeneous
group will be abandoned. This will be the inevitable result of address-
ing interests, as the resource claims of some women based on their class
and/or race disadvantages may come into conflict with the interests of
other women, or require privileging the building of alliances with other
social actors. Yet of course quotas “make” women a group as much as
they presuppose group existence. Seen from the social movement per-
spective of framing, quotas are “a constitutive act of social definition
that can have important consequences” (Brubaker 2005) that are both
positive and negative, albeit in the latter case unintended. In this sense
quotas are not neutral in relation to the politics of identity and it is equally
likely that women representatives elected through a quota-based system
will resist addressing those interests of women that stem from their class
and/or race positions. In other words, it is conceivable that the manner in
220 Shireen Hassim

which feminist claims are advanced can itself constitute the boundaries
of what kinds of interests are addressed by elected officials.
This outline of the key arguments for quotas and their objections
still leaves unresolved a wide range of questions about whether quotas
will facilitate the democratization of political systems and enhance the
substantive representation of women. The growing political consensus
among feminist activists that quotas are the “magic key” that will unlock
the door to political power has produced, with a few notable exceptions,
superficial analyses of the effects of quotas on democratic process and
democratic culture. Proponents of quotas are to some extent justified
in making arguments that stop short of predicting the policy outcomes
of increasing women’s representation; the arguments for democratizing
access are strong enough. This should not be good enough for those
seeking to advance democracy though. The supporters of the fast-track
approach to representation imply that there are no inherent benefits to
incremental processes of changing representation imbalances. In other
words, the process matters less than the numerical outcome. As I will
argue, though, in African countries it is apparent that short-circuiting
the political process of creating strong constituencies and defining policy
agendas has negative consequences for the extent to which democratic
culture takes root in new democracies.
For the most part, quotas are seen as at best improving substantive
representation and at worst having little or no effect on the political
system. Yet looking at quotas in African countries – traditionally seen as
the success stories of the quota strategy – opens new sets of considerations
for democratic debates and suggests that in certain circumstances quotas
can indeed have harmful effects on processes of democratization.

III. Do quotas democratize political institutions? The impact

of electoral systems in Africa
Transitions from various forms of non-democratic rule in African coun-
tries opened new opportunities for women’s movements to make claims
on the political system. The wave of political liberalization that swept
through the continent in the 1990s ushered in liberal democracy with
its attendant features of multiparty competition, regular elections, and
the removal of restrictions on the media, political organization and civil
society mobilization. No fewer than forty-eight African countries held
multiparty elections in the 1990s, most of these for the first time in
decades.
Why would men allow women to take seats in legislatures when
they might themselves be displaced? One reason is that in these new
 The impact of quotas for women on democratization 221

democracies, elections conducted under new rules and with new demar-
cations of voting districts allowed women to enter the political process
without directly displacing male incumbents (although competition for a
limited number of seats does imply that the use of quotas will reduce the
opportunities for aspiring male politicians). Where the political landscape
is dominated by a single party, as in South Africa, Rwanda and Uganda,
quotas can be a relatively costless strategy that has little impact on elec-
tion outcomes. By way of contrast, in one of Africa’s oldest and most
stable democracies, Botswana, quotas are resisted by political parties,
partly because they would alter the gender composition in a “winner-
takes-all system,” i.e. some men would lose their seats. Similarly, stable
Mauritius has not adopted quotas – in the absence of violent political
conflict, it has retained the first-past-the-post electoral system with no
plan to change the electoral system.
Designing the appropriate electoral system is key to the acceptance
of quotas. Electoral systems play a key role in determining the nature
of the relationship between elected representatives, political parties, and
constituencies. Proportional representation (PR) systems are now com-
monly assumed to be the most favorable for women, and closed list PR is
particularly effective in increasing women’s representation. But the sys-
tem carries costs that are less frequently detailed in discussions of quotas.
On the one hand, a PR system allows progressive parties to bypass cus-
tomary and cultural objections to women’s election – no small factor in
societies where conservative religious forces dominate civil society. On
the other hand, it also allows parties to establish mechanisms of control
over elected leaders and exacerbate party paternalism. PR systems breed
loyalty to party rather than constituency, and this tendency is intensified
in political systems where the conditions for full and free contestation
between different interest groups are limited. Tripp (2006) argues that
in Uganda allegiances to the National Resistance Movement (NRM) at
times hamper the ability of women MPs to support legislation favored
by the women’s movement.2 Similar concerns have been expressed in
the South African case, where women MPs have found it difficult to
establish a set of priorities for feminist intervention (Hassim 2006b). A
strong dominant party with a centralist political culture results in women
MPs being more likely to believe that policy-making is legitimately the
responsibility of party elites.
The ability of women representatives to mobilize within their parties
and their willingness to challenge party hierarchies is an important deter-
minant of the extent to which women will be effectively represented, yet

2 See also Tamale (1999).

222 Shireen Hassim

individual women MPs often find it difficult to develop the confidence

and political base from which to push for gender equality platforms.
The argument for a critical mass suggests that at the tipping point of
30 percent women MPs can begin to develop significant mechanisms of
support. In many countries parliamentary women’s caucuses have been
mooted as a strategy for setting priorities and building support and confi-
dence among women MPs. However, in Uganda and South Africa these
have not been as successful as feminists anticipated, precisely because of
the electoral system. Women are, after all, elected to represent the party
and not women. The challenge is therefore whether and how women’s
gender interests can be articulated in a way that is distinct from their
party interests and identities. Uganda is an intriguing example in this
respect. There, women are elected to seats reserved for women by an
electoral college made up of women and men councilors, but rather than
unambiguously representing women, they are required to represent the
district as a whole. Unlike other special groups (soldiers, youth, workers,
and people with disabilities), women are not elected members of their
group, arguably because the relative strength of the women’s movement
might result in women MPs who are hostile to the NRM leadership
(Tamale 1999; Tripp 2006). (Independent women’s organizations have
been key opponents of Museveni’s attempts to lift term limits on the
President, for example.) This means that women MPs remain beholden
to the movement as the primary political force and that the election of
women MPs who might challenge movement policies in parliament is
as unlikely as in a multiparty PR system. As Tripp (2006) argues, this
has led women’s movement activists to argue for a change in the electoral
process so that the women’s seats are elected by universal adult suffrage –
unsurprisingly, a move opposed by the President.
In both Uganda and Rwanda, women are constructed as a group that
stands outside ethnic categories. This suits certain explanations of polit-
ical conflict in Africa, which see ethnic groups as either fixed and pri-
mordial, or purely as colonial constructs used instrumentally to divide
African populations. Both, as Newbury and Newbury (1999) point out,
are inadequate depictions of ethnicity and ethnic mobilization. Summa-
rizing an extensive body of scholarship on ethnicity in Africa, they remind
us that individual identities need to be separated from corporate concepts
of identity; that ethnicity as an identity is not coterminous with ethnic
mobilization as a political force; and that ethnic identities are histori-
cal products that are neither essentialist nor easily politically fabricated
from “outside.” In Rwanda and Uganda, ethnic mobilization was closely
tied to resource contestations by the political class; yet at the local level
regional and clan affiliations were more dominant than corporate ethnic
 The impact of quotas for women on democratization 223

identities. “Ethnicity served as a language through which fears and ambi-

tions were expressed, but it was not ethnicity that ‘caused’ the violence”
(ibid.: 316). During the Rwandan genocide, “connection to the state
apparatus seems to have been the common characteristic of local leader-
ship in the massacres” (ibid.: 314). It is outside the scope of this chapter
to look at all the underlying causes for and forms of mobilization of vio-
lence in Rwanda and Uganda. Of particular interest here, though, is the
way in which one form of group identity and representation is treated as
illegitimate, and the other as acceptable and even progressive. It could be
argued that in both countries women’s representation provided a kind of
alibi for the progressive, “democratic” nature of new governments that
at their core nevertheless remained authoritarian, and are increasingly
so. And indeed, as some of the examples discussed above suggest, some
women MPs colluded in the politics of silencing opponents of the ruling
parties on the grounds that they represented an “illegitimate” base.
A different set of objections relates to the ways that certain types of quo-
tas might reinforce rather than challenge the social biases against women
in politics. One effect that has been documented by Indian and Ugandan
scholars is the emergence of a hierarchy of legitimacy: “quota candidates”
are seen as token appointments – less effective and less accountable rep-
resentatives than those elected on “open” tickets. In India, for example,
women who stood for the reserved seats in the panchayats were accused
of “fronting” for their husbands, fathers, and brothers. For women politi-
cians, the effect is to intensify their second-class status – in other words,
while quotas may facilitate women’s election to office, they have the
consequence of reinforcing existing patriarchal biases. Special seats for
women at local government level in Uganda have had similar responses.
Women occupying “special” seats are bypassed by citizens who go to
“real” representatives with their problems (Tamale 2003; Tripp 2000;
Ahikire 2003). Tamale argues that support for women’s agendas in fact
can act as a disqualification for broader political support, thus limiting
women’s attempts to be seen as legitimate representatives of a wider
constituency.
Senegal, with its mixed electoral system, also offers an interesting
counterpoint to the conventional PR system. Although more women
are elected to parliament through PR than through the plurality sys-
tem, Creevey (2006) points out that an increasing number of women
are elected through the plurality system. She shows the role that polit-
ical leadership and a modernizing party can play in shifting attitudes
toward women – showing what possibilities there are for feminist pol-
itics even in constituency-based systems. Similarly, Longman (2006)
writes that in Rwanda it would be erroneous to attribute the massive gains
224 Shireen Hassim

in gender representation to quotas alone, as the number of women in

non-reserved elected seats has also increased dramatically – with women
winning nearly half of all the seats in the national assembly in the 2003
elections (against a guaranteed 30 percent of seats). Longman (ibid.)
points out that women won offices where quotas did not apply, as in the
election of judges. The involvement of women in previous governments,
and particularly in the democracy movement, appears to account for the
rapid increases in the number of women in elected seats.
Strong support for women’s rights and women’s political represen-
tation from the president of the leading party, sometimes against the
views of other male leaders, is a double-edged sword. Both Uganda and
Rwanda, and possibly also South Africa, show the potentially ugly side
of presidentialism and single-party dominance, with rising authoritar-
ian tendencies in the ruling party unable to be checked by women MPs
(Tripp 2000; Longman 2006; Hassim 2006a). Indeed, Longman argues
that the increase in women’s representation in parliament “serves more
as an instrument of legitimizing and preserving RPF [Rwandan Patri-
otic Front] power” (Longman 2006: 147). In Uganda and Rwanda, the
extension of patronage to women by the dominant party has been highly
instrumental in building up a core of loyal, “progressive” cadres. In
Uganda, women are the central actors in the parliamentary NRM cau-
cus that ensures the NRM leadership’s support. In several cases, female
NRM loyalists acted to limit the participation and voice of NRM oppo-
nents, and “those who contested the outcomes of various [electoral] races
watched helplessly as the courts dismissed one petition after another in
favour of NRM loyalists” (Tripp 2006: 119, citing Tamale 1999). Tripp
quotes a woman opposition leader explaining that many NRM women
MPs “will rather go by the position of the government because they look
to the government as the only agency that has brought them where they
are. So in other words, we are now turning our affirmative action into
patronage” (Tripp 2000: 119). President Museveni himself has congrat-
ulated women MPs for helping him defeat the “multipartyists.” On the
flip side, women (like men) who go against the President may see the
end of their political careers, even if they are not in parliament. Winnie
Byanyima, a leading Ugandan feminist and MP married to opposition
presidential contender Kiiza Besigye, was forced to exit local politics and
take up a post in the African Union after both she and Besigye were
accused of treason, and Besigye was accused of rape.
In Rwanda, the head of a women’s organization who “questioned
the effectiveness of the country’s female legislators in solving women’s
problems, likening them to flowers which look good but do little
else . . . was condemned and threatened. Shortly afterward, she fled the
 The impact of quotas for women on democratization 225

country” (Lacey 2005). The RPF pushed through new legislation to

create a Forum of Political Parties that would have the power to deter-
mine who was fit to stand as an MP. Critics of the government were
deemed “divisionist” and hence unfit to stand (Longman 2006: 147).
Women MPs stood by while Kagame rewrote the Constitution to make
parliament in effect a one-party institution.
In South Africa, political culture and oppositional movements are
strong enough to allow women politicians to criticize and question both
the government and the President. There is enormous room for dissent in
the system more broadly and no threat of imprisonment or torture. Even
so, room for dissent within the African National Congress (ANC) is more
limited. Leading feminist MP Pregs Govender, who had headed up the
parliamentary committee responsible for the removal of discriminatory
legislation and the promulgation of pro-choice legislation, left parliament
in 2002 after clashing with the President and party leadership over an
arms deal and the government’s policy on HIV/AIDS. Since her depar-
ture, the parliamentary committee on women has been virtually defunct,
despite the large number of women MPs and a substantial proportion of
ministers who are women (including the Deputy President).
These tendencies suggest that, in order to be successful, democratic
women’s movements would have to move quickly to buttress the openings
allowed in transitional periods to build strong movements outside par-
liament that will sustain women’s rapid representational gains through
quotas. Women cannot rely on political parties as their only vehicles
for representation. While women’s movements do well to build relation-
ships with political parties, their success in representational politics also
depends on their degree of autonomy from those parties. This ensures
that women are not only mobilized for their votes but as electoral con-
stituencies – that is, constituencies with clearly articulated policy inter-
ests. I will return to this point below.

IV. Quotas and procedural democracy

The argument that quotas will at the minimum impact positively on pro-
cedural democracy bears interrogation. African experiences of quotas
implicitly reinforce the importance of linking representation to account-
ability. Formally, this accountability operates both internally (within
political parties) as well as externally (accountability to constituen-
cies). In most African democracies, however, political parties are poorly
institutionalized and operate in a highly centralized fashion. Opposi-
tion parties tend to be weak and political contestation seldom revolves
around debating the content of political manifestos. Rather, histories of
226 Shireen Hassim

struggle against colonialism and accusations and counter-accusations of

corruption – fiscal or electoral – are the stuff of party contestation. In
conditions where the basic principles of free and fair elections are con-
stantly violated by the party in government, quotas can achieve very little
unless they are backed up by strong campaigns to change the rules of
the game. In some cases, they can even harm struggles for procedural
democracy. For example, Ugandan feminists have argued that Musev-
eni’s acceptance of reserved seats for women might have helped give
women seats in parliament but it was based on an attempt to create a
solid and reliable block of women supporters of the NRM in govern-
ment. At crucial moments, this dependent bloc of women MPs observed
silently as Museveni repressed opposition to the NRM.
Of course, women can benefit from multipartyism, even if women’s
organizations have to affiliate to a party in order to have a political
voice. In mass parties that have emerged from nationalist movements,
for example South Africa’s ANC or Zimbabwe’s ZANU-PF, national
women’s organizations operate as the women’s wing of the party. The
valid vehicle for political voice is through the nationalist party, based on
the assumption that the nation is a singular homogeneous entity with
one legitimate representative party. Support for gender quotas in many
African countries was won by the women’s movement aligned with the
dominant party and has reinforced the hegemony of that dominant party.
If opposition parties were to develop their own internal women’s wings,
however, the landscape could change. This would entail an acceptance
that political constituencies are not pre-given, or the moral preserve of
one particular movement/party, but that there is a socially and politi-
cally constructed relationship between citizens and political parties that
needs to be constantly nurtured. Quotas work best in multiparty systems
where several parties are well institutionalized, have clear relationships
of accountability for their political actions and represent a diversity of
interests. In these cases, we see evidence of the contagion effect, where
constituencies of women can push parties into supporting gender equality
because of the threat that they can vote for other parties. In Sweden, for
example, feminists successfully advanced the argument that they would
switch their allegiances from the Social Democratic Party if women’s
views on party policies (such as nuclear weapons) were not taken into
account. Without the threat of electoral retaliation, strategies that focus
exclusively on descriptive representation make limited contributions to
democratization.
However, the ability to democratize political parties, enforce account-
ability and clean up the electoral game is dependent on the strength of
women’s organizations in civil society and the development of strong
 The impact of quotas for women on democratization 227

relationships between women legislators and their supporters. The rela-

tionship between women parliamentarians and the women’s movement
is as difficult as the relationship between women activists and leaders
within political parties. In some respects there are organic political ties
within these spheres; many women parliamentarians cut their political
teeth in women’s organizations. As a result, the relationship between
party/parliament and civil society can often be fluid, as women activists
can move back and forth between state and civil society. This fluidity may
enhance policy influence but can also have negative impacts in countries
with small political elites. Close personal and political relationships can
breed a sense of loyalty to comrades that undermines criticism. Yet even
if political elites were a much larger segment of the population, there are
inherent tensions in the relationship between feminists in politics and
those in civil society. Anne Summers (1986) has characterized this as the
tension between missionaries (activists in civil society) and mandarins
(politicians) with each expecting relationships of support and account-
ability that may be hard to fulfill. Accountability relationships depend
on quota mechanisms working in tandem with strategies to strengthen
women’s organizations outside the state.

V. The impact of quotas on the articulation of interests

In the previous section, I addressed the claim that, at a minimum, quo-
tas have a positive effect on democratizing political procedures. I have
shown that this can only be achieved if quota campaigns are tied to other
processes of party and system reform. What about the stronger claim
made by Mansbridge (2005) that descriptive representation is causally
related to substantive representation? Here too the evidence from African
countries is mixed. The demands for institutional inclusion can galvanize
and reinforce women’s movements – although the extent to which this
will happen depends on whether strong relationships can be maintained
between women in parliament and women’s organizations in civil society.
If inclusion in the strong sense is to be advanced (that is, if the citi-
zenship claim is framed in terms of women’s access to decision-making
as part of a strategy to redress inequalities), then the crucial factor is the
existence of strong associations that have the capacity to articulate the
particular interests of their constituencies, to mobilize those constituen-
cies in defense of those interests, and to develop independent strate-
gies to achieve their aims. A strong social movement requires a degree
of political autonomy in order to retain its relative power within any
alliance. Although in general the importance of robust civil societies is
widely supported in Africa, there is far less tolerance of independent
228 Shireen Hassim

feminist organizations. While autonomy was highly valued in West-

ern women’s movements – and in many was seen to be a condition of
existence – in postcolonial countries autonomy is less highly valued.
Because women’s political activism in postcolonial contexts has been
enabled by larger struggles against colonial and class oppression, the
result is a more highly developed politics of alliance rather than auton-
omy. In many African countries, women’s movements have been weakly
developed, and feminism as a mobilizing ideology has been delegitimated.
Quota proponents tend to neglect these “thicker” understandings of par-
ticipation as an activity that encompasses collective mobilization, through
associational vehicles over which women have control. Indeed, it can be
argued that quotas are likely to be supported by women’s movements
that recognize their organizational weaknesses – that is, they can be seen
as a compensatory strategy.
Quotas must be distinguished from effective participation, where the
emphasis is on more effective interest articulation and representation –
i.e. making the “voice” of particular constituencies of women louder in
processes of policy-making (Goetz and Hassim 2003). Supporters of quo-
tas might argue that they have the strategic value of fast-tracking a “crit-
ical mass” of women into parliament who will then demonstrate/model
the value of women’s participation. As I have argued in the previous
section, however, mere numerical presence does not translate into voice,
nor does amplified voice “automatically strengthen the moral and social
claims of the powerless on the powerful and produce better accountabil-
ity to that group” (ibid.: 34). Not only presidential or party control, but
also institutional norms and procedures and the nature of processes of
deliberation can undermine the extent and impact of women’s voice in
the public sphere. The more subtle patterns by which power hierarchies
are upheld, even when new groups are included in institutions, are both
hard to make visible as well as hard to change. Studies of parliament
have shown that women’s participation as individuals in institutions of
decision-making is generally limited – women are less likely to participate
in parliamentary debates than men and are more likely to feel intimidated
by the demands of public speaking, even when the women elected are
highly skilled professionals (Norris and Lovenduski 1996).
A stronger understanding of representation locates interest articulation
and interest representation through parties and civil society organizations
as central, and is more clearly tied to a politics of transformation. Inter-
est articulation has a checkered history in African governments, as it has
been associated most closely with the idea of national machineries for
women, which have treated women’s interests as narrow and “special”
and in need of patronage. Creating a set of specialized institutions for
 The impact of quotas for women on democratization 229

the consideration of gender shifted the issues of gender inequality out of

the realm of politics and into the technical realm of policy-making. As
Banaszak, Beckwith, and Rucht (2003: 6) point out, this is increasingly
a problem with national machineries around the world: “women’s move-
ments have been presented with an increasingly depoliticized and remote
set of policy-making agencies at the national level . . . The relocation of
responsibility to non-elected state bodies eventually reduces social move-
ment influence.” In the administration, gender equality concerns have
fallen hostage to a range of institutional hierarchies and systemic block-
ages that are hard to deal with from outside the bureaucracy. In these
contexts, interest articulation becomes not just de-politicized but also
anti-political, to the extent that women’s policy demands are shunted into
bureaucratic alleys.
To be sure, state actors are central to the process of articulating inter-
ests. But bottom-up politics, in which different constituencies of women
identify and define policy demands which can be advanced as part of a
horizontal coalition of interests (for example, poor urban women linking
up with housing movements), as well as through policy-focused alliances
with (or where necessary pressure on) women parliamentarians, is a cru-
cial corollary. The emphasis on building a constituency-based politics
recognizes that not all women have the same interests, that women’s orga-
nizations might conceive of their interests differently from women politi-
cians, and that strong organizations of women at the local level enhance
the virtuous circle of representation. In Uganda, women’s organizations
have been partially successful in pursuing such strategies, although their
effectiveness is highly constrained by the authoritarian political culture
in which they are operating, where building alliances outside of Musev-
eni’s approved network can impact negatively on parliamentary politics.
Similarly, it can be argued that in South Africa women who are elected
into parliament are increasingly distant from movements of poor peo-
ple. For example, witness the political (and constitutional) contestation
between the Treatment Action Campaign and the Health Minister (and
ANC Women’s League leader) Manto Tshabalala Msimang over access
to retroviral treatment for people who are HIV-positive. As Sylvia Tamale
has argued, male African politicians are more willing to allow descrip-
tive representation where women MPs symbolically “stand for” women
rather than being representatives who “act for” women (Tamale 2003).
Consequently there is a tendency for quotas to produce an increasingly
elite-based body of representatives. This kind of politics may bring certain
sections of the women’s movement into contestation with elite and party-
oriented members, as it is a form of politics that is likely to take a more
confrontational approach over party manifestos and state policies. The
230 Shireen Hassim

outcome of such alliances may be a marginalization of these actors from

the state and political parties.
I would argue that recognition through quotas is a deceptively easy
strategy. A much more transformative demand for representation (that
is, that women need access to decision-making in order to advance a
project of eliminating gender hierarchies of power) is reduced to a simple
mechanism. Although it is generally accepted in feminist literature that
a combination of factors is responsible for women’s increased access to
political office – the nature of the political system and the organization
of political competition, the nature of civil society and especially of the
feminist lobby within it, and the nature and power of the state – all too
often actual political strategies are collapsed into a demand for a quota.
This is not surprising; it is without doubt more difficult to reshape the
nature of the political system.
This is not to suggest that there have been no gains from the increase
in the number of women in African parliaments. Perhaps the most visible
gain has been to remove many of the most overt examples of discrimina-
tory legislation from the statute books. In some new democracies, women
MPs can point to the introduction of legislation that erodes women’s for-
mal inequality. For example, in South Africa, the right to abortion, legal
and institutional protection for victims of domestic violence, and the
inclusion of quotas for women in economic affirmative action legisla-
tion and policies are all claimed by women MPs as the product of their
presence in the parliament. I scanned the literature to find examples of
women’s impact on legislation that went beyond changes in electoral laws
and the introduction of quotas at different levels of decision-making –
I have found few examples other than in South Africa, where there is a
relatively well-developed women’s movement. Critics could point, on the
other hand, to the failure of women MPs to stand firm on broader issues
of democracy and human rights and – in Uganda and South Africa, for
example – to the failure of political presence in cases where the interests
of women have come into contestation with those of men. Even demands
for formal equality can be muted when they come up against intense
mobilization of competing groups in society. For example, key women
MPs in Uganda did not support the amendments to the 1998 Land Act
that would increase women’s rights to property ownership (a watered-
down right for spousal co-ownership, at that). Despite intense lobbying
by women activists outside parliament, Museveni and his women loyal-
ists in parliament opposed extending land ownership rights to women,
eventually using the technical rules of parliament to prevent the key
clause from being read into the Act. According to Museveni, “when I
 The impact of quotas for women on democratization 231

learnt that the Bill was empowering the newly-married women to share
the properties of the husbands, I smelt a disaster” (Tripp 2006: 125).
Key women MPs, including Vice-President Speciosa Kazibwe, opposed
women’s ownership of land on the grounds that access was the issue, not
control. Kazibwe held a meeting of women MPs with the specific aim of
ensuring that they remained loyal to the President.
In South Africa, the case of the Communal Land Rights Act (champi-
oned by a woman minister) dramatically showed up the limits of women’s
representation, as the Act failed to extend full rights of land ownership
and control to women living under the jurisdiction of traditional lead-
ers. In this case, the ANC was subjected to conflicting pressures from a
constituency hostile to women’s interests and, moreover, it was a con-
stituency that posed a perceived electoral threat. While women MPs
might see themselves as organically linked to women’s organizations in
civil society, they placed considerable emphasis on intraparty debates and
were aware of their need to survive within the party. They voted with the
party leadership on the Act, despite protests from women’s organizations
in civil society. Without an electoral constituency to back them up, the
reality was that women activists remained hostage to the goodwill and
support of the party leadership.
These examples show the difficulty of making claims for the impor-
tance of quotas when one goes beyond very thin versions of formal equal-
ity. Throughout the continent, women’s movements have historically
struggled for forms of liberation that included but were not restricted
to the achievement of equal rights and opportunities. This is not to deny
the importance of formal political and civil rights, especially in contexts
where there is tremendous resistance to the idea of women’s equality. Yet
poor women’s movements have consistently held out for an understand-
ing of liberation that would include economic transformation and (to a
lesser extent) the social reordering of gender relations. These expectations
of women’s organizations have specific implications for social policy, as it
would require that resources be directed in such a way that they serve not
only to address the needs of the poorest women, but also become part of
an incremental process of enhancing the recognition of women’s person-
hood and their full participation in political and economic processes. This
would entail considerable debate about the relationship between the pub-
lic and private sphere, the cultural recognition of women’s unpaid work,
and at the very least a public debate about how to negotiate women’s
individual rights against those of the communities to which they belong.
On these measures, there is little to show as yet for women’s increased
representation.
232 Shireen Hassim

If the particular content of gender equality ought to be left to women’s

organizations themselves to define, in their particular socioeconomic
and cultural contexts, the argument for the collective mobilization of
women and the building of electoral constituencies as the undergirding of
representational politics is strengthened. Disaggregating representation
from the debate about desirable policy directions and policy outcomes –
however strategically useful such a strategy might be in the short term –
carries the danger that political party elites (albeit with a gender-neutral
face) will continue to determine the content of policies.

VI. Conclusion
This chapter argues that while there is a strong democratic case for the
inclusion of women in institutions of political representation through the
use of affirmative action mechanisms, these mechanisms do not neces-
sarily advance feminist ambitions for deep democracy. Arguments for
quotas tend to neglect the context in which quotas are adopted, and
assume that the institutional conditions for effective representation are
easily built by new entrants to political structures. This neglect stems
from assumptions that multiparty, liberal democracies are globally sim-
ilar. However, as these examples from African countries suggest, the
weaknesses of party systems and of mechanisms of accountability of rep-
resentatives to voters may in certain conditions be exacerbated by the
adoption of quotas The political conditions that facilitate the adoption
of quotas in many new democracies – party systems that are only thinly
institutionalized, centralist political cultures, and elite control of poli-
tics – may limit the effectiveness of quotas, and at times have negative
consequences for deepening democracy.
I argue that quota mechanisms can result in demands for representa-
tion being disembedded from processes of constituency formation. This
impacts significantly on feminist ambitions to shift the nature of gender
power relations. Rapid inclusion into political institutions through mech-
anisms such as quotas has many advantages. However, if these are not
tied to processes of extra-parliamentary collective mobilization in a free
political environment on the one hand, and broad public deliberation on
the principles and priorities underpinning legislative and policy propos-
als on the other, the impact on more transformative feminist ambitions
is limited. As far as the impact of increased representation of women in
legislatures is concerned, this has to be analyzed in the broader context of
the political system. There is little value in promoting women’s represen-
tation in authoritarian systems: at best, a substantial amount of political
energy is expended for little substantial gain; at worst, women simply
 The impact of quotas for women on democratization 233

become collaborators in the process of restricting the political rights of

oppositional groups.

BIBLIOGRAPHY
Ahikire, Josephine. 2003. “Gender Equity and Local Democracy in Contempo-
rary Uganda: Addressing the Challenge of Women’s Political Effectiveness
in Local Government,” in No Shortcuts to Power: African Women in Politics
and Policymaking, ed. Anne Marie Goetz and Shireen Hassim. London: Zed
Books, 213–39.
Baldez, Lisa. 2006. “The Pros and Cons of Gender Quota Laws: What Happens
when you Kick Men Out and Let Women In?” Politics and Gender 2 (1):
102–9.
Ballington, Julie (ed.). 2004. The Implementation of Quotas: African Experiences.
Stockholm: International IDEA.
2005. “Introduction,” in International IDEA Handbook: Women in Parliament:
Beyond Numbers, ed. Julie Ballington and Azza Karam. Stockholm: Interna-
tional IDEA, 23–30.
Banaszak, Lee Ann, Karen Beckwith, and Dieter Rucht (eds.). 2003. Women’s
Movements Facing the Reconfigured State. New York: Cambridge University
Press.
Bauer, Gretchen and Hannah Britton (eds.). 2006. Women in African Parliaments.
Boulder, CO: Lynne Rienner.
Beaman, Lori, Esther Duflo, Rohini Pande, and Petia Topolova. 2007. Women
Politicians, Gender Bias and Policy-Making in Rural India. State of the World’s
Children 2007 Background Paper, UNICEF.
Brubaker, Rogers. 2005. “Ethnicity Without Groups,” in Remaking Modernity:
Politics, History, and Sociology, ed. Julia Adams, Elisabeth S. Clemens, and
Ann Shola Orloff. Durham, NC: Duke University Press, 470–92.
Creevey, Lucy. 2006. “Senegal: Contending with Religious Constraints,” in
Women in African Parliaments, ed. Gretchen Bauer and Hannah E. Britton.
Boulder, CO: Lynne Rienner, 215–45.
Dahlerup, Drude. 1988. “From a Small to a Large Minority: Women in Scandi-
navian Politics.” Scandinavian Political Studies 11 (4): 275–98.
2005. “Increasing Women’s Political Representation: New Trends in Gen-
der Quotas,” in International IDEA Handbook: Women in Parliament: Beyond
Numbers, ed. Julie Ballington and Azza Karam. Stockholm: International
IDEA, 141–53.
2006. “The Story of the Theory of Critical Mass.” Politics and Gender 2 (4):
511–22.
Dahlerup, Drude and Lenita Friedenvall. 2005. “Quotas as a Fast-Track to Equal
Representation for Women.” International Feminist Journal of Politics 7 (1):
26–48.
Goetz, Anne Marie and Shireen Hassim (eds.). 2003. No Shortcuts to Power:
African Women in Politics and Policy Making. New York: Zed Books.
Hassim, Shireen. 2006a. “Women, Parliaments and Gender Equality in Africa:
Exploring the Virtuous Circle of Representation,” in Women in African
234 Shireen Hassim

Parliaments, ed. Gretchen Bauer and Hannah Britton. Boulder, CO: Lynne
Rienner, 246–66.
2006b. Women’s Organizations and Democracy in South Africa: Contesting Author-
ity. Madison: University of Wisconsin Press.
Htun, Mala. 2005. “Women, Political Parties and Electoral Systems in Latin
America,” in International IDEA Handbook: Women in Parliament: Beyond
Numbers, ed. Julie Ballington and Azza Karam. Stockholm: International
IDEA, 112–21.
Inter-Parliamentary Union (IPU). 2005. “Women in National Parliaments: Situ-
ation as of 31 January 2005.” Available at: www.ipu.org (accessed September
3, 2006).
Krook, Mona Lena. 2006. “Gender Quotas, Norms and Politics.” Politics and
Gender 2 (1): 110–18.
Lacey, Marc. 2005. “Women’s Voices Rise as Rwanda Reinvents Itself.” New
York Times, February 26, 2005.
Longman, Timothy. 2006. “Rwanda: Achieving Equality or Serving an Author-
itarian State?” in Women in African Parliaments, ed. Gretchen Bauer and
Hannah E. Britton. Boulder, CO: Lynne Rienner, 210–33.
Mansbridge, Jane. 2005. “Quota Problems: Combating the Dangers of Essen-
tialism.” Politics and Gender 1 (4): 622–37.
Matland, Richard E. 1998. “Women’s Representation in National Legislatures:
Developed and Developing Countries.” Legislative Studies Quarterly 23 (1):
109–25.
Matland, Richard E. and Donley T. Studlar. 1996. “The Contagion of Women
Candidates in Single-Member District and Proportional Representation Sys-
tems: Canada and Norway.” Journal of Politics 58: 707–33.
Nanivadekar, Medha. 2006. “Are Quotas a Good Idea? The Indian Experience
with Reserved Seats for Women.” Politics and Gender 2 (1): 119–28.
Newbury, Catharine and David Newbury. 1999. “A Catholic Mass in Kigali:
Contested Views of the Genocide and Ethnicity in Rwanda.” Canadian Jour-
nal of African Studies / Revue Canadienne des Études Africaines 33 (2/3): 292–
328. Special issue on French-Speaking Central Africa: Political Dynamics
of Identities and Representations.
Norris, Pippa and Joni Lovenduski (eds.). 1996. Women in Party Politics. New
York: Cambridge University Press.
Phillips, Anne. 1995. The Politics of Presence. London: Polity Press.
Sawer, Marian. 2002. “Representation of Women: Meaning and Make-Believe,”
in Women, Politics and Change, ed. Karen Ross. Oxford: Oxford University
Press, 5–17.
Summers, Anne. 1986. “Mandarins or Missionaries: Women in the Federal
Bureaucracy,” in Australian Women: New Feminist Perspectives, ed. Norma
Grieve and Alisa Burns. Melbourne: Oxford University Press.
Tamale, Sylvia. 1999. When Hens Begin to Crow. Kampala: Fountain Publishers.
2003. “Introducing Quotas in Africa: Discourse and Legal Reform in
Uganda.” Paper presented at International Institute for Democracy
and Electoral Assistance (IDEA) / Electoral Institute of Southern
Africa (EISA) / Southern African Development Community (SADC)
 The impact of quotas for women on democratization 235

Parliamentary Forum Conference, Pretoria, South Africa, November 11–13,

2003.
Tripp, Aili Mari. 2000. Women and Politics in Uganda. Kampala: Fountain
Publishers.
2006. “Uganda: Agents of Change for Women’s Advancement?” in Women in
African Parliaments, ed. Gretchen Bauer and Hannah E. Britton. Boulder,
CO: Lynne Rienner, 158–89.
Waylen, Georgina. 2007. Engendering Transitions: Mobilization, Institutions and
Gender Outcomes. Oxford: Oxford University Press.
9 On quotas and qualifications for office

Andrew Rehfeld

I. Introduction
In every existing democracy, qualifications for office restrict who may run
for various elected offices. Most frequently, these qualifications require
that the candidate be older than a certain minimal age, that she be a
citizen of the nation in which she is running, and that she be a mem-
ber of the party or resident of the district that she stands to represent.
Increasingly, qualifications for office are being crafted based on gender,
race, and ethnicity. In the last twenty years, dozens of democratic govern-
ments around the globe have established secured seats in the legislature
or on party lists for women (Baldez 2004; Krook 2006). And the effect of
other rules and circumstances – such as the need for large sums of money
to mount a campaign in the United States – has created virtual qualifi-
cations for office, stacking the deck in favor of some kinds of candidates
at the expense of others.
The use of qualifications is hardly new, nor is the desire to use them
to achieve desirable results. For example, on December 4, 1820, in a
debate on alterations to the Massachusetts State Constitution, Daniel
Webster defended the inclusion of a “profession of belief” in Christian
principles as a qualification for office (Webster 1903). As a general mat-
ter, Webster argued, the use of qualifications for office was consistent
with principles of representative government that depended on voters
themselves making good principled judgments about the types of people
who should represent them. Voters used their own standards to filter out

For comments on earlier versions of this argument I thank Corey Bretschneider, Randy
Calvert, David Estlund, Susan Hyde, Jack Knight, Mona Lena Krook, Charles Larmore,
Frank Lovett, Andrew Martin, Larry May, Robert Reich, Margo Schlanger, Ian Shapiro,
Anna Stilz, Susan Stokes, Daniel Weinstock, Elisabeth Wood, and participants in the
Washington University Law School faculty seminar, the Brown University political philos-
ophy workshop, the Yale University conference on Representation and Popular Rule, and
the 2008 annual meeting of the Association for Political Theory.

236
 On quotas and qualifications for office 237

all sorts of undesirable types, and rightly so.1 More specifically, Webster
argued, voters regularly elected those with a strong Christian faith so
that requiring a profession of belief would not substantively change any
voting outcome (a claim that was most probably true at the time). A
constitution, he argued, should not be prevented from institutionalizing
the very distinctions that voters use, and would like to use, as a sort of
insurance policy for their own considered judgments (ibid.).
Presuming Webster was right that requiring a profession of belief would
not alter any intended election results (and leaving aside the discrimina-
tion that this particular qualification involved) what good reasons are
there to object to externalizing voter judgments in the forms of qualifi-
cations for office? Certainly, qualifications make these judgments inflexi-
ble; no matter how good a judgment may appear to voters at a particular
time, legal qualifications make revision difficult, if not impossible, when
the need or desire arises (Manin 1997; Schwartzberg 2007). Yet if we
presume that voters have a right, and perhaps even an obligation, to use
value-based preferences when voting by choosing “better” candidates
over “worse” ones; if we presume that individual voters ought to use
internal constraints such as “must have good policy ideas” or “must be
competent,” or “must have a good character” to guide their own choices;
and if we acknowledge that such preferences and constraints tend to be
robust over time; what principled reasons are there to object to formal-
izing these preferences into democratic constitutions?
The goals of this chapter are to explain how qualifications for office
operate conceptually; to describe their unique relationship to the under-
lying principles of representative government; and to explain why their
use always violates two presumptive democratic rights: that of citizens to
run for office, and that of voters to choose whomever they wish to rule
them. The goal here is neither to defend nor assail their use, but rather
to precisely specify the conceptual bounds of these institutions, and the
necessary normative tradeoffs they entail.
In large part, the present project is meant to clear the ground for
the normative and empirical analysis of these institutions, something
of particular importance given the explosion of gender quota laws2
in emerging and established democracies over the last two decades
(Baldez 2004; Krook 2006). The empirical literature now emerging by
and large addresses the effects of these laws on women’s representation,

1 The filtration argument has been a staple of theories of representative government for
over 300 years. For early variants of this filtration argument see Harrington (1992),
Hume (1987), and Madison (1961). For a more recent account see Manin (1997).
2 I outline a defense of treating gender quota laws as a qualification in Section II below.
238 Andrew Rehfeld

or details the robust and differentiated political discourse that surrounds

their adoption or defeat (Kittilson 2006; Krook 2009; Opello 2006).
The normative literature has treated this material indirectly, as a means
of securing a more completely defended principle of justice. Thus theo-
rists have argued for secured seats (creating a qualification for that seat)
as a way of achieving descriptive representation, correcting past harms,
or promoting greater justice and deliberative legitimacy going forward
(James 2004; Young 2000; Mansbridge 1999; Williams 1998; Phillips
1995; Young 1990).
In both empirical and normative literatures, the institution of a qual-
ification for office itself is poorly described and unclearly articulated.
Nowhere are the conceptual problems clearly delineated, nor are the
democratic potential and pitfalls that accompany their use specified. In
large part this is understandable and typical of how the study of politics
often (and justifiably) proceeds: particular political problems are identi-
fied and struggled with, and only later are the conceptual and normative
underpinnings developed and clarified. If successful, such clarification
can then feed back into the study of politics by seeing and developing
new ways to solve these practical problems. The fact that such an analy-
sis has not yet been done, despite the attention to the expanding use of
qualifications around the world, is thus hardly surprising.
Although I will not make the argument here, my own view is that
qualifications may only legitimately be used to solve collective action
problems brought about by institutional limitations, and not to limit
democratic options by an appeal to substantive justice. I state, without
defending, this position because it deviates significantly from the way
qualifications are usually justified. In many empirical cases, qualifications
have been justified by reference to principles of substantive justice or a
commitment to a certain view of democratic deliberation. A commitment
to justice, for example, is sometimes said to require that individuals of
a certain sort, say African-Americans in the United States, be given
representation in a legislature as a form of reparation for past harms. A
commitment to democratic deliberation, for example, is sometimes said
to require descriptive representation in a legislature in order to ensure that
a diversity of opinions, ways of thinking, or mere “presence” of that group
is made real (Phillips 1995). And of course there are many other kinds
of arguments that might support the use of quotas and qualifications:
from increasing sociological legitimacy to maintaining stable political
institutions.3

3 Related arguments may be found in Phillips (1995), Kymlicka (1996), Williams (1998),
Mansbridge (1999) and James (2004). This literature generally fails to distinguish
between “group representation,” in which a group is given the power to elect its own
 On quotas and qualifications for office 239

But qualifications to limit voter choice in the service of principles of

justice – whether to ensure that representatives profess religious belief or
to ensure that representatives have a certain set of descriptive character-
istics – violate the very principles upon which representative government
is endorsable as a form of democratic self-rule. Qualifications entail high
democratic costs because they violate two presumptive democratic rights:
the right of citizens to run for any office that stands to make and enforce
law over them; and the right of citizens to choose whomever they want
to fill those offices.4 Furthermore those committed to democratic princi-
ples should view the use of qualifications for office with suspicion based
on past experience: historically, they have been motivated by democratic
distrust, and used to exclude vast portions of the population from office
due to the “questionable” judgment of the demos. In the past, qualifica-
tions excluded those without property, the non-white, and the female.
Today, the excluded groups are different, and the exclusion is motivated
by a desire to act justly by including those who were previously excluded.
But the arguments are structurally the same: we need quotas and qual-
ifications today, it is said, because the judgment of the demos cannot be
trusted (they only vote for men after all).
Whether the two presumptive democratic rights are worth violating and
whether quotas can be secured without democratic distrust is our ulti-
mate concern. For surely many anti-democratic rules that restrict voter
choice are defended to promote democratic values (or more precisely
in order to promote a certain kind of robust democracy). The answer
to these substantive questions can only be addressed by first laying out
clearly what tradeoffs are involved.
In this chapter I present a systematic way of thinking about qualifica-
tions for office as any rule or circumstance that has the effect of creating
different probabilities of success for individual candidates prior to elec-
tion. In Section II I provide a conceptual road map distinguishing qualifi-
cations and quotas from claims to group representation, define the terms,
illustrate three variants of these institutions, and delineate between three
kinds of justifications for their use. I conclude the conceptual section
by distinguishing between qualifications such as “conforming to voter
judgment” (i.e. winning an election) that are endorsable by foundational
principles of representative government, and qualifications such as “pro-
fession of religious belief” that do not reference voter judgment at all. In

representative, and “legislative presence,” in which legislative seats are secured for mem-
bers of a particular group. Quotas and qualifications for office secure legislative presence,
but not necessarily group representation; similarly giving a group the right to elect its own
representative may secure group representation but not necessarily legislative presence.
I take this up in Section II.1.1 below.
4 I will defend these as core democratic rights later in the chapter.
240 Andrew Rehfeld

Section III I lay out the necessary democratic costs of any qualification for
office, explaining why they always constitute a presumptive violation of
two core democratic rights: the rights of citizens to run for office and the
rights of voters to choose whomever they want to rule them. I conclude
by suggesting (but not defending) that qualifications for office should
only be used in conditions that amount to democratic self-limiting rather
than imposition of substantive justice that forecloses the exercise of voter
judgment at election time.

II. A conceptual roadmap to qualifications for office

In this section I offer a number of conceptual clarifications to guide the
normative discussion that remains. In Section II.1 I propose treating
qualifications for office as rules that create differential probabilities of
candidates’ success prior to voter judgment. In Section II.2 I offer a
typology of justifications for qualifications drawn from empirical cases.
And in Section II.3 I differentiate between different sorts of justifications.

II.1 Defining “qualifications for office”

Traditionally a “qualification for office” has been any rule that limits
legal eligibility for an office to a specific set of people. This eligible set
may be a subset of all citizens (e.g. “citizens over thirty-five years old”);
it may include non-citizens (e.g. “anyone over the age of thirty-five years
old”); it may be identical to all citizens (e.g. “any citizen of this nation”);
or limited to a single person (“the first male heir of the king”). The
hallmark of a qualification for office is simply that it has the effect of
limiting who may run for office and, by extension, limiting the set of
candidates for whom a voter may vote. Framed in this way, quotas that
set aside a certain number of seats in the legislature or on party lists
(without setting up restrictions on who may vote for those seats) are a
type of qualification for office. Examples of these include the parité laws
in France that require party lists to have a certain percentage of women
on them, and rules in Rwanda that require 30 percent of all legislative
seats to be held by women. These rules create a gender qualification for
those seats.
In this section I make three formal distinctions. First, I argue that
qualifications (and quotas) should be distinguished from group repre-
sentation (II.1.1). Second, they should be framed in terms of probability
assessments rather than as “legal eligibility” (II.l.2). And they should
be framed in reference to the effect of rules and circumstances on can-
didates, rather than the language of a rule itself (II.1.3). Finally, I will
 On quotas and qualifications for office 241

differentiate between qualifications that emerge from voter preferences

and those that do not (II.1.4).

II.1.1 Qualifications versus group representation Qualifications for

office make more likely what has been called, following Anne Phillips,
“legislative presence” for a member of a particular group, whether the
group is “citizens over thirty-five years old” or “women” (Philips 1995).
Legislative presence for a particular group is secured when there is a
member of that group in the legislature regardless of who elects that mem-
ber to office. Thus, we can secure the legislative presence of women, that
is, ensure that there are women in office, simply by establishing qual-
ifications that require women to be elected without changing anything
about the makeup of the electoral constituency that elects that mem-
ber. By contrast, we can precisely describe “group representation” as a
case of substantive representation, where a group’s interests are substan-
tively represented by a representative’s activity in office, regardless of the
descriptive characteristics of that representative.
Frequently “group representation” is secured by defining electoral con-
stituencies around members of a particular group. To use the American
example again, drawing electoral districts that create African-American
majorities secures the substantive representation of blacks to the extent
that the elected representative is responsive to the interests of those black
voters. But “group representation” (as in the substantive representation
of a group) may also be achieved when a group’s interests are pursued by
a representative “virtually,” even in cases where the group does not for-
mally elect a representative: as a group, poor Americans are represented
to the extent that US Senator Ted Kennedy defends their interests in
Congress. This “virtual” representation may be inadequate to secure a
group’s interests over time. Nor does it have a democratic basis, since the
group being represented has no authority over the representative himself.
These considerations point at the limits of virtual versus actual represen-
tation, but do not affect the “groupness” of the representation itself: the
poor in this case are represented to the extent their interests as a group
are pursued in the legislature.5
In much of the literature the term “group representation” blurs the
distinction between using group membership as the basis of an electoral
constituency (e.g. an African-American electoral district) and having a

5 For a defense of the underlying conception of representation on which this depends,

see Rehfeld (2006). For the inadequacy of the traditional distinctions upon which this
relies, including those by Hanna Pitkin and Jane Mansbridge see Rehfeld (2009), Pitkin
(1967), Mansbridge (2003).
242 Andrew Rehfeld

member of the group be a member of the legislature (e.g. having an

African-American representative). Because of this, “group representa-
tion” is sometimes described as giving a group electoral autonomy over
their representative, or at other times described as securing seats in a leg-
islature for members of that group. By precisely delineating the bounds of
these two concepts we can thus more precisely describe and explain some
confusions that arise when one, but not the other, of these two is secured.
So, for example, when white voters elect a conservative African-American
(like J. C. Watts, the Republican former member of the US Congress from
the state of Oklahoma), the legislative presence of African-Americans is
secured (by definition). Conversely, when a white man is elected mayor
of a predominantly black American city (like Martin O’Malley, former
mayor of Baltimore, MD), we can say that blacks there achieved group
representation, but not legislative presence.
Some readers may want to object to the claim that Watts counts as
achieving “legislative presence” for blacks at all, because Watts does not
speak as the majority of blacks would, nor perhaps is he representative of
American blacks. (A related example is that of Clarence Thomas on the
US Supreme Court.) Such a claim generally confuses group representa-
tion with legislative presence itself. We may all agree that Watts does not
defend African-American interests but we cannot seriously claim he does
not provide black presence in the legislature since he is African-American
himself. More plausibly, when critics say that Watts does not achieve “leg-
islative presence” for blacks (despite being an African-American himself)
it is only because critics do not want legislative presence of blacks. Rather
they mean legislative presence of certain kinds of African-Americans and
Watts is not one of those kinds. Thus, making clear the kinds of presence
that one wants, and making clear the modes of group representation, we
can more precisely explain what is going on here.
The distinction between “group representation” and “legislative pres-
ence” is an important one, but in practice these two concepts admittedly
blur and overlap. For example, sometimes legislative presence may be best
achieved by giving a group the right to elect its own representative, which
may well be the best way of making sure a member of that group gets
elected. Similarly, group representation may be best achieved by secur-
ing legislative presence for a member of that group. This is particularly
true when it is difficult or undesirable to define an electoral constituency
by reference to a particular group characteristic. The defense of securing
seats for women as a way of promoting their interests (i.e. using legislative
presence to secure group representation) is often made on the imprac-
ticality or impossibility of defining “women-only” electoral constituen-
cies. But the fact that in practice legislative presence may secure group
 On quotas and qualifications for office 243

representation or vice versa should not lead us into blurring the under-
lying conceptual distinctions between presence, constituency definition,
and group representation.6

II.1.2 Framing qualifications as probabilities of success Traditionally,

qualifications for office have been treated as necessary legal conditions:
either a candidate meets a stated qualification and he can run for office,
or a candidate does not meet this standard and is legally forbidden to
run. But we might express qualifications in more functional, probabilis-
tic terms rather than as a dichotomous legal permission. In this broader
framing, traditional qualifications are rules relative to which some candi-
dates have a zero probability (p = 0) of success in an upcoming election,
while others have a non-zero (p > 0) probability of success. The tradi-
tional framing of qualifications as legal permissions and exclusions can
then be translated into these terms: candidates that meet a qualification
are legally permitted to run, and thus have a non-zero probability of suc-
cess, relative to that rule, in the election; candidates that fail to meet the
qualification are forbidden from running and thus have a zero probability
of success (again relative to that rule) in the election.
(What does “relative to that rule” add to the account? First, it acknowl-
edges that there are many rules and circumstances that affect someone’s
probability of success. In any particular case we want to isolate a given
rule or circumstance independent of these other factors, to the extent
such isolation is possible. Second, “relative to that rule” acknowledges
that these rules work in so far as the rule is in fact followed. A candi-
date may be legally forbidden from running for office by some rule, but
still fake his eligibility, legally challenge the rule, or take arms to coerce
authorities to let him run. Thus the probability of success is only relative
to that particular rule, since there is always some probability that the rule
may not be followed or something else would intervene.)
Once seen in probabilistic terms, qualifications for office can be gener-
alized to rules that assign non-equal probabilities of success to individual
candidates. In the traditional case there is a binary assignment of proba-
bilities: the rule assigns some individuals a p = 0, and to other it assigns
a p > 0, where p takes the same value for all individuals in each of the
two groups. But there are other rules that have the effect of assigning dif-
ferent non-zero probabilities of success across candidates. Indeed, from

6 Mansbridge distinguishes between flexible and inflexible ways of securing group rep-
resentation, treating quotas as an inflexible way. But I think such a view fails to see
the conceptual limits of “constituency,” as well as “qualifications.” For an extended
treatment of “constituency,” see Rehfeld (2005).
244 Andrew Rehfeld

the standpoint of democratic theory, in which we presumably want to be

sensitive to inequalities in political power, it is the assignment of different
probabilities of success to each individual that should be of concern to
us rather than one particular case in which some have a probability of
“0” of winning, and others have a probability “p >0” of winning. Put
differently, the traditional dichotomous treatment of qualifications (p =
0, p > 0) is simply a special case in which rules give some candidates
a greater probability of winning than others. We will thus treat qualifi-
cations for office more expansively as any rule whose effect is to create
differential probabilities of success across a set of individual candidates even in
cases where all individuals are legally permitted to run (i.e. even in cases
where no individual has a p = 0 relative to that rule).
With this treatment we can see why some rules that do not follow
the traditional form are properly described as “virtual” qualifications
for office. Consider the case of gerrymandered districts in the United
States in which 70 percent of all voters are registered members of the
Democratic Party, or a district in which 70 percent of all voters are
African-American. Candidates running in these districts are far more
likely to win if they share the quality upon which the constituency was
defined, be it party ID, race, and so on. Treating qualifications as the
assignment of different probabilities of success, regardless of what value
p actually takes on, explains why in these districts “race” or “political
party” forms a very real if implicit qualification for office: the probability
of winning in a majority–minority African-American district is greater if
one is black than if one is not, even though one need not be an African-
American to win. To use another case, there is no formal rule with the
effect of requiring candidates to be wealthy (or be good fundraisers) in
order to successfully compete for office or forbidding those who are neither
of these to run. But given other rules that make it almost impossible to win
without access to funds, we may treat access to wealth as a qualification
for office, since those without it are much less likely to win than those
with it. So long as we limit the term “qualification” to denote legal
permissibility or exclusion, we cannot explain why, in these cases, race
and wealth indeed are qualifications for office. Instead we say that these
are qualifications for office because they distribute probabilities of success
differentially across candidates.

II.1.3 Qualifications as the effects of rules and circumstances Tradition-

ally, qualifications for office have been treated as explicit rules, such as
an age or residency requirement, that set legal permissions (p > 0) and
exclusions (p = 0) to run for office. In addition to framing them in terms
of differential probabilities of success, I want to broaden the treatment in
 On quotas and qualifications for office 245

two other ways: by considering the effects of rules, rather than the rules
themselves; and by considering circumstances as well as rules. I will take
each in turn.
First, we should consider the effect of a rule, rather than the rule itself,
as what constitutes a qualification for office. If the effect of two different
laws or rules is the same, and if that effect is to differentially distribute the
probabilities of success in an election, then both rules will be treated as a
qualification for office no matter what the language of the rule actually is.
This is consistent with other features of election law, and in fact governs
the US Justice Department’s oversight of electoral districts.
As an example, consider these two proposed voting rules:
“Women with young children may not vote.”

and
“No strollers are allowed within one mile of a polling booth.”

Only the first rule explicitly forbids women with young children from
voting. But the second creates a de facto, if not absolute, limit and
should rightly be treated as a voting rights infringement. Less fanciful
and more heinous examples are readily available throughout American
history, particularly in the attempt to disenfranchise African-Americans.
The point is, we should look to the effects of rules, rather than their
explicit language, to judge whether differential probabilities of success
are created and thus whether the rule counts as a qualification for
office.
Once we see that it is the effects of rules that matter, but not the explicit
rules themselves, we can justify the expansion of our treatment from
rules to circumstances as well. Here “circumstances” refer to constraints
that affect the probabilities of candidate success. These might include
circumstances personal to a candidate – the health of their family – or
more broadly social facts like the number of opponents they face in the
election (the more candidates there are in an election the less likely any
particular one of them is to win, all things considered7 ). Here, as in the
case of rules, qualifications for office refer only to those circumstances
that assign unequal probabilities of success to different candidates. Thus,
the fact that all candidates faced the same diminished probability of suc-
cess the more candidates there are that run does not form a qualification

7 Recall that we consider the effect of the rule by itself and not other subsequent reactions
to the rule. So, all things considered, the more candidates run for office the less likely
any one of them is to win, even though it may be the case that an individual may benefit
from the entry into a race of a polarizing candidate.
246 Andrew Rehfeld

for office. By contrast, access to money would be a qualification for office

since it creates differential probabilities of success.8
Focusing on “circumstances” is a way to acknowledge structural
inequalities that create differential probabilities of success for different
candidates, even in the absence of formal rules. I leave open the possibility
that “circumstances” may well be reducible to a set of very specific rules
that form these circumstances – in this case perhaps treating political
speech as commercial speech or the rules of the free market. Circum-
stances thus allow us to consider more fully the set of social facts that
have the effect of creating unequal probabilities of success for different
candidates.
Finally, I note that although conceptually we need not differentiate
between qualifications that emerge from circumstances, and qualifica-
tions that emerge from rules, as a normative matter we may well care
about their etiology. For various reasons we may want only to regulate
formal rules whose effect is to form qualifications for office, but not
attempt to intervene in circumstances that are hard to pin down even
though they have a particular effect. But the fact that there may be good
normative reasons for dealing with some but not all qualifications for
office in a certain way does not explain what should count conceptu-
ally as a “qualification.” Thus we will leave open the possibility that
there may be normative reasons to treat the different etiologies of qual-
ifications differently without presuming that conceptual separation in
advance.
In sum, then, I define qualifications for office in this way:

Qualifications for office = Rules or circumstances that have

the effect of differentially distributing probabilities of success
to candidates for office.

II.1.4 Voter-endogenous and voter-exogenous qualifications for office Qual-

ifications for office as just stated refer to any rule or circumstance that
differentially affects the probability of a candidate’s success. Although we
normally think of qualifications as external to voters, voters themselves
regularly constrain their own choices to a set of individuals possessing
features they take to be important. Some voters will only vote for mem-
bers of a particular party; others will choose between a few parties; still

8 I call “access to money” a qualification instead of “money” itself because each candidate’s
probabilities of success are equally changed by each additional dollar they have. Money
itself thus does not create differential probabilities of success; rather “access to money”
must form the qualification. I acknowledge that this is not as precise as it could be, but
it is precise enough to proceed with the present account.
 On quotas and qualifications for office 247

others make their choice based on the personal characteristics of the

candidates. These internal judgments differentially distribute probabil-
ities of success among candidates no more or less than external choice
constraints would if written to correspond with the internal judgments
of a particular voter.
The fact that individual voters use their judgment to select candi-
dates (rather than choose randomly) means that “corresponding to voter
preferences” should itself be counted as a qualification for office, since
candidates are not equally likely to so correspond. This should not be
particularly controversial since “winning an election” is about the most
readily used “qualification for office” around and even conforms to the
traditional use of legal permission, in that it assigns those who meet that
standard (i.e. who win the election) a p = 1, and all others a p = 0. But it
allows us to make an important distinction between voter-exogenous and
voter-endogenous qualifications that will be of significance to normative
analysis.

Let voter-endogenous qualifications for office denote qualifications emerging from

the exercise of voter judgment.
Let voter-exogenous qualifications for office refer to all non-voter-endogenous
qualifications for office.

Since representative government is often defended on the ability of vot-

ers to promote into office “better” representatives than would otherwise
be selected were random selection (Manin 1997) or direct democracy
(Mill 1991 [1861]; Madison 1961) employed, we should distinguish
between voter-endogenous and voter-exogenous qualifications for office
and treat them differently. Indeed, while there are normative reasons
to be concerned about the kinds of judgments that voters make, we
have different reasons to be concerned about rules and circumstances
external to voters that create differential probabilities of success among
candidates. And the reason we should treat voter judgment differently
from exogenous factors has to do with the presumptions of democratic
representative government. Representative government is built on the
presumption that voters will indeed make non-arbitrary judgments built
upon non-random preferences, and thus systematically favor some kinds
of candidates and disfavor others. This is not only a fact of representative
government, it is arguably why representative government is preferable
to direct forms. By contrast, we should be extremely suspicious of voter-
exogenous qualifications because they restrict democratic options while
voter-endogenous qualifications are more difficult to assail on democratic
principles. At the very least, we should expect different sorts of arguments
248 Andrew Rehfeld

concerning voter-exogenous interventions and voter-endogenous inter-

ventions.
Of course, different individuals cast votes for numerous reasons. But
the point here is that we can differentiate constraints internal to voters
(their preferences, judgments about their interests or the interests of the
nation) from those externally applied to voter choice prior to the exercise
of their judgment. These external constraints are distinctive precisely
because they operate external to voters and prior to the exercise of their
judgment. The aristocratic nature of elections that Manin was right to
emphasize can thus be institutionalized in two different ways, by internal
and external constraints.
I will say more about voter-endogenous and voter-exogenous con-
straints below in Section III.2. But notice that relying on constraints
imposed by voters to select “proper” candidates depends on trusting vot-
ers to make good choices and to allow for flexibility when necessary. The
US founders, for example, thought that candidates for political office
ought to be property owners of a certain level. But rather than making
this requirement rigid as law, they opted to rely on voters’ internal judg-
ment that the founders believed would have the same effect of promoting
only prominent property owners to elective office. Further, by relying on
voter judgment rather than external constraints, the proposed system was
less rigid and allowed for flexible response by voters if the circumstances
warranted it.9
If the advantage of the propertied classes is assured by a statistically proven
regularity of electoral behavior, the system offers a measure of flexibility: cir-
cumstances may arise where the effect [i.e. property ownership] does not obtain,
because an exceptional concern overrides voters’ ordinary inclination toward
“conspicuous” candidates. The situation is different if legislative position is
reserved by law to the higher social classes, because the law is by definition
rigid. (Manin 1997: 126)

Relying only on internal constraints maintains flexibility.

II.2 A typology of qualifications

Having mapped out the conceptual space surrounding qualifications for
office, I turn to a typology of qualifications that differentiates three dis-
tinct types of ends for which formal qualifications (i.e. explicit rules) have
been used. Note that formal qualifications (like most rules) use proxies

9 The flexibility of the American system is in sharp contrast to the English and French
systems of the times which used qualifications more robustly (Manin 1997: 126–9). For
more on the value of revisability in constitutional democracies see Schwartzberg (2007).
 On quotas and qualifications for office 249

for the desired end. For example, age qualifications are brute devices that
are meant to secure “political maturity” – for lack of a better term – of
legislators. Few people care that a legislator is of a certain age, but many
of them care that they have some minimal maturity or life experience.
Since age is closely related to maturity of all kinds, age qualifications are
used because they increase the likelihood that legislators will have polit-
ical maturity. Here the age requirement stands as a proxy for “political
maturity” or simply “maturity” that corresponds systematically to age.
The point is that “being age X” is not itself the desideratum of the rule;
rather the rule is a proxy for the desideratum of political maturity.
Consider some other examples in which proxies are used to achieve a
substantive end. Property ownership was proposed to secure a variety of
aims including independent judgment and having a stake in society. Gen-
der and race requirements today are used to secure diverse perspectives
within the legislature, as well as provide a remedy for past discrimina-
tion and harm. Term limits are supposed to protect representatives from
becoming too entrenched. And ethnic and subnational group member-
ship qualifications are often seen as the only way to guarantee stability
and/or give a population a sense of participation within the whole. With-
out presuming the correctness of the argument (that is, the proxy used
may or may not achieve these ends), qualifications are most often used in
order to aim at a purported end, rather than being the end in themselves.
Historically, formal qualifications have been used to secure three dif-
ferent kinds of ends: (i) to secure the same minimal competency for
each member of a legislature; (ii) to ensure correspondence between an
individual representative and her constituents; or (iii) to create legisla-
tures that are composed of different kinds of people. These three aims –
competency, correspondence, and composition – sometimes overlap but they
are conceptually distinct. I will take each in turn. (A summary of this
discussion appears in Table 9.1.)
First, qualifications have historically been used to increase the proba-
bility that each candidate will have minimal competency. By “competency”
I mean: (i) an acquirable trait; that is (ii) purportedly necessary for the
job; and (iii) applies to everyone who is a member of the body (say, a
legislature). A minimal age requirement in a legislature, for example, is
precisely of this kind: it is an acquirable trait that is purportedly necessary
for the job10 and everyone in the legislature is required to have it. Reli-
gious oaths for office might also be considered under this category since
they were acquirable, purportedly necessary for the job, and applicable

10 Or in light of what we just said, “political maturity” is required for the job; minimal ages
act as proxies for this.
250 Andrew Rehfeld

Table 9.1 Three general aims of qualifications with examples

Purported general aim

of qualification Description of aim Specific examples of aim

Competency: Competencies are: Age qualifications;

To set minimal (i) features that any individual religious oaths; term
competence of all can acquire; limits
representatives (ii) purportedly necessary for the
job; and
(iii) applicable to all who serve.
Correspondence: Correspondence qualifications Residency requirements;
To maintain ensure that office-holders share party membership
correspondence some characteristic feature(s) requirements;
between used to define their electoral citizenship
representatives and constituency. requirements.
their electoral
constituents
Composition: Composition qualifications are: Group membership;
To craft how the (i) seat specific; and gender quota laws;
legislature (or other (ii) determined with an eye to the district
body) looks. membership of the gerrymandering.
whole body.

to all who served. Literacy tests and qualifying examinations would be

further examples of this kind of qualification.
A second kind of qualification aims to increase the probability that
officer-holders correspond to a particular group, most often to their par-
ticular electoral constituencies. The most familiar of these types of qualifi-
cations are residency requirements, in which representatives are required
to live in the state (frequently in the electoral district) of the voters that
vote for them. Correspondence between representatives and their elec-
toral constituents is also secured in proportional representation systems
in which electoral constituencies are defined by the votes that voters cast
at each election (Rehfeld 2005). And here, correspondence between,
say, the Labor constituency and its representatives is guaranteed by the
requirement that those who offer themselves up as Labor candidates in
fact support Labor policies. It is this obvious fact (that party candidates
support the party platforms for which they stand) that guarantees cor-
respondence even in these systems. The correspondence requirement
is perhaps most readily seen in cases of professional representation in
which, for example, pediatricians elect pediatricians, psychiatrists elect
psychiatrists, etc., for representation on their own professional councils.
Importantly, “correspondence” does not specify the competency of a
 On quotas and qualifications for office 251

representative, but indicates an identity between a representative and her

constituents on some particular dimension.
In addition to securing minimal competencies for all office-holders, and
to ensuring that all office-holders correspond to some feature of their
electoral constituents, qualifications for office can also be used to shape
the composition of the legislature when considered as a whole. In this case,
qualifications for office can increase the likelihood that a legislature (or
other body) looks a certain way by establishing different rules for each seat
in the legislature. For example, gender and racial quotas are often said
to be important to secure deliberative or descriptive diversity within the
legislature – what we have called “legislative presence.” Unlike competence
and correspondence, which could apply to all offices, qualifications that
alter the composition of a legislature differ for each seat or group of seats
in the same corporate body. So, for example, if we wanted more gender
diversity in the legislature we might set aside certain seats for men and
certain seats for women by creating different qualifications for each seat.
Similarly, we might specify that a certain percentage of seats be set aside
for ethnic groups whether to protect their rights or for other reasons.
These three kinds of qualifications are conceptually distinct, and the
typology is meant to delineate and describe how these institutions are
used, and have been used historically. Importantly, they are not mutually
exclusive and whether they are used to ensure competence, correspondence,
or composition may depend on other political factors. One might use a
residency qualification that ensures correspondence between voters and
representatives in order to secure a certain kind of composition of the
legislature. Or, if electoral constituencies were defined by gender, then
a qualification for office to ensure correspondence between constituen-
cies and their representatives would also ensure that the legislature was
composed in a certain way. Whether this result is the primary aim, or a
secondary consequence, will depend on how the qualification itself is
conceived based on one of these categories. Further, the normative jus-
tification for qualifications for office may indeed rest as much on which
of these three aims the qualification is meant to secure, as it is on what
the consequences are of securing that aim.

II.3 Kinds of justifications

In II.1 I offered a conceptual framework for qualifications including a
reconceptualization of what they were; in Section II.2 I offered a typology
of the three ends qualifications for office are used to secure. I turn now
to a final typology about the kinds of justifications that are used for
252 Andrew Rehfeld

qualifications for office. These will be critical to assessing the normative

legitimacy of their use.
Why take up “justifications” for qualifications at all? At a formal level,
justifications for these rules are not about qualifications per se but about
the ways we justify any rule to each other. Here I presume, rather than
defend, what is the dominant, if not universally agreed upon, account
of political legitimacy in which democratic institutions are normatively
legitimate in so far as they are justifiable to all reasonable citizens.11
In plain language this means that a government has a right to make
and enforce its laws when they are justifiable to citizens over whom
they govern.12 The ultimate question is whether any qualification can be
justifiable to democratic citizens on this basis.
Allowing that a range of specific justifications for qualifications for
office have in practice been offered for their use, at a general level of
abstraction there have been three substantive justifications for them: jus-
tice (they aim at purportedly just ends); democratic distrust (they limit
the harms that voters can cause); public choice (they solve irresolvable
conflicts of public choice). I will take each in turn. (A summary of this
section appears below in Table 9.2.)
First, qualifications may be justified by a principle of justice; i.e. the
purportedly just ends at which they aim. In the case of contemporary
gender qualifications, for example, ensuring that the legislature’s composi-
tion includes men and women is purportedly just for a set of reasons (e.g.
because a legislature without many women is ipso facto unjust; because
ensuring gender parity is presumed to secure fairer laws and policies;
because of the role modeling effects of seeing all kinds of people in the
legislature, etc.). When qualifications are rejected, they are also rejected
by reference to substantive principles of justice that they purportedly vio-
late: e.g. the rejection of property or literacy tests or racial requirements
to run for political office.
A second kind of justification for these rules is based on democratic
distrust. Following a robust tradition dating at least back to Plato, the
use of formal qualifications in England and America in the seventeenth

11 This is a simplification of a vast and important literature on justification. The contem-

porary touchstone is of course Rawls (1993). What such justification means in practice
is developed in a masterly way in MacGilvray (2004). See also the discussion in Rehfeld
(2005) that provides a more complete summary of this.
12 As stated, this is a presumption; I am not defending this claim here. Famously, the
meaning of “reasonable” is contentious, and whether or not individuals actually have
to be reasonable for a government to be legitimate raises some problems of consent. It
would seem that a government might be legitimate so long as it was endorsed by “rea-
sonable” people who gave their hypothetical consent even if none existed and therefore
none consented to the current regime.
 On quotas and qualifications for office 253

Table 9.2 A typology of justifications

Kind of justification Description Examples

Principles of justice The end towards which a Gender qualifications that create
qualification aims is better policies; secure seats for
defended by the ethnic groups as a remedy for past
substantive justice it harms.
brings about.
Democratic distrust Qualifications limit the bad Property qualifications that kept the
judgment that voters can poor from running lest they dupe
make. voters; gender and racial
qualifications that are imposed
because voters are racist and sexist.
Problems of choice Qualifications can resolve Racial gerrymanders in cases where
decisional conflicts that voters both want to vote for
emerge from any system “someone like me” and want
of rules. representatives that look like the
nation as whole.

and eighteenth centuries made this explicit: motivated by spite and envy
rather than love of the whole, the masses could not be trusted to elect
leaders who would pursue the public good. Thus, for example, were
higher property qualifications for office justified: those with property
purportedly had a stake in society and without the use of qualifications
they were unlikely to be elected (Lerner 1995). The large electoral dis-
trict was similarly intended to create a de facto qualification for office
because it was designed in large part to increase the likelihood that men
of good reputation would be elected (Madison 1961; Harrington 1992;
Manin 1997; Rehfeld 2005). Again, the use of these qualifications was
purportedly justified to protect the nation from outcomes resulting from
the venality or stupidity of voters.
To the extent that prejudice explains why women are not elected, the
contemporary use of gender quotas appears to be motivated in large part
by the same sense of democratic distrust that animated earlier campaigns.
Voters today no longer lack “moral virtue”; today they are merely “racist”
or “sexist,” or unjustly refuse to acknowledge the claims of the oppressed.
Voters are no longer said to be stupid or uneducated; rather they are said
not to know what their own interests are or ought to be. These are
precisely the same kinds of arguments that were made to justify other
sorts of now rejected qualifications for office.
A third justification is based on the limits that emerge from the use of
voting rules themselves. As has been well developed, institutional prob-
lems, rather than ignorance or malice, can prevent desired outcomes from
254 Andrew Rehfeld

being achieved. For example, imagine all voters in the United States used
the following principle of “racial correspondence” to decide their vote: “I
will vote for someone who’s the same race as I am.”13 And imagine that
the US is racially integrated at the level of the enormous congressional
district.14 In such a case, most electoral constituencies might contain a
majority of whites and the legislature would be entirely white even if all
voters would prefer a racially diverse legislature. Whites might endorse this
principle of racial correspondence for the same reasons that non-whites
do (because they believe race is a good proxy for whether a candidate
shares a citizen’s perspectives and interests), yet sincerely lament the
fact that there are no blacks elected to Congress. Electoral gerrymanders
are premised upon this very principle and supporters of them are often
explicit about this in their endorsement.15
Some gender quota laws may be designed (and justified) to remedy
similar sorts of institutional limits. Even absent venal discrimination at
the time of voting (in this and the last case), if individual candidates are
selected simply using the decision rule “I will vote for whomever has
the most experience” or “I will vote for whomever is likely to be the
most effective politician given ongoing sexism (that I lament but which
nevertheless exists),” then men will almost always be elected even as we
all lament the fact that there are no female representatives.16 The only
way around that is to impose external constraints (and in this case such

13 When whites use “racial correspondence” as a principle of selection they are often
accused of racism or merely excluding black voices; when blacks use it they are lauded
for trying get their perspective in the legislature. I suspect voters in each group are
motivated by a combination of “keeping out” the other and “getting in” their own.
In either case, if the principle of racial similarity is a good one, there seems to be no
prima facie reason to restrict its use by any individual. The problem is only a problem
collectively. Unless, of course, one thinks that using racial correspondence itself is not
in fact a good principle on which to select political representatives.
14 The population of the United States is roughly 300,000,000 people. Although each
congressional district does not have precisely the same population owing to differences
in state population and in apportionment rules, the size of each district is roughly the
same: 690,000 individuals per electoral district. The territorial scope will of course
differ given population density. The territorially large state of Montana has but a single
member of the US House of Representatives, the same number that can be found in a
relatively small area of Manhattan.
15 In law, gerrymanders are justified based on the inability of blacks to elect whomever
they want, and not on whether blacks in fact are in the legislature. And yet in the United
States the test of whether a community has been able to elect one of their own has often
been based on how successful blacks have been in getting into office. Thus however else
racial gerrymanders are justified, they do appeal to the election of blacks themselves,
and not merely to giving the black community more autonomy to choose their own
representatives.
16 Indeed when parité was passed in France, it was difficult to recruit women to run as
candidates because there were relatively few of them with any political experience.
 On quotas and qualifications for office 255

constraints would simultaneously bring about the conditions by which

these particular internal constraints would fail to be salient).
To sum up here, qualifications for office need not be based on anxieties
about the masses. When justified as a corrective for the presumed incom-
petence of voters they are explicitly anti-democratic. But when justified
for their potential to solve conflicts of democratic values or collective
action problems they are no more or less democratic than any other such
provision of preemptive restraints in the law. But appreciating qualifica-
tions for office in this light, rather than as a reflection of anti-democratic
sentiments, demonstrates that they need not be anti-democratic, and that
they need not fail the test of public justifiability. And, as I will argue later,
it is necessary for qualifications to be justified based on the solution to
institutional conflict, rather than as an anti-democratic corrective, for
them to be reasonably democratic at all.17

III. The normative implications of qualifications for office

Having delineated the conceptual space surrounding qualifications for
office, I turn now to a discussion of the normative implications of their
use. As defined, qualifications for office differentially alter the probability
that a potential candidate will win an election. Further, in cases where
they set a p = 0 for some candidates they limit a voter’s choice set by
excluding these candidates. Both of these constitute the prima facie lim-
its of two core democratic rights: a right to run for office and a right
to vote for an unconstrained choice of candidates. The goal of this sec-
tion is to explain why qualifications so limit these rights. (What follows
merely describes the normative terrain and explains the normative stakes
involved. While it suggests some ways that qualifications might be justi-
fied, owing to the limits of this chapter it does not offer a full defense of
these positions.)
The two rights can be illustrated by considering again Daniel Web-
ster’s argument about transforming the internal judgments of voters (who
wanted candidates to profess Christian beliefs) into an external institu-
tional constraint (that required the profession of belief as a condition
of office-holding). Webster’s proposal was problematic for three reasons.
First, any formalized rule creates a rigid rule in place of flexible voter
judgment. Second, qualifications for office also abrogate a right to offer

17 It may be that qualifications are also necessary for justice. And indeed, I think sometimes
considerations of justice may limit democratic institutions, as checks on the majority
do. This presumes that “democratic” and “just” as categories do not precisely overlap,
and that it is possible to have at least an unjust democracy (whether or not a “just non-
democracy”), though I acknowledge this is a disputed point. See for example Shapiro
(1996).
256 Andrew Rehfeld

oneself as a contender to rule; Webster failed to distinguish the right to

serve from the right to run. Third, to argue that in a democracy people
can enact any precommitment in law it wants so long as it is consistent
with what they would do anyway is an overstatement. Qualifications for
office do not merely stand to reshape democratic institutions, they stand
to unmake democracy: imagine that only those with some individual’s
DNA are legally permitted to run; the result is an elective dictatorship or
monarchy, not a recognizable democracy.
To state this more formally, qualifications are problematic because
they limit or violate two presumptive democratic rights: (i) citizens have
a right to run for offices whose occupants make and enforce the law
that governs them; and (ii) voters have a right to choose whomever they
wish to rule them. As I said above, I presume that legitimate democraties
begin with a strong presumption in favor of these rights forming an
institutional default position of any purportedly legitimate democratic
government. Justifying qualifications for office will thus mean explaining
the conditions under which violations of these rights are acceptable. And
that requires first specifying in greater detail what these rights amount to.
In this section I will argue that an equal right to run is institutionalized,
first, in the legal permission to run for office, and, second, in the absence
of formal rules that alter the external probability of success for one kind
of candidate over another kind. A right to run for office is only minimally
secured first by the legal permission to run, and further presumes the
absence of institutional rules that differentially distribute the probabilities
of success among different kinds of candidates. In this way the right to
run for office parallels the treatment of “equal voting rights”: the equal
right to vote has meant not merely a legal permission to cast a ballot, but
also the equal probability of a vote to affect the outcome of an election,
as well as the absence of other rules that would encourage or suppress
voting by certain kinds of voters. Understood in this way, qualifications
for office thus always violate this core, presumptive right. I develop this
in two sections, first more generally in terms of political equality (III.1)
and then in a specification of an equal right to run (III.2).

III.1 Political equality

I presume at a minimum that any reasonably democratic government will
secure these three political rights for all citizens:
(1) The right to vote: citizens have a right to vote of equal value in choosing
among candidates for office;
(2) The right to an unconstrained choice set: citizens have a right to be
ruled by anyone they choose (i.e. to an unrestricted choice set of candi-
dates);
 On quotas and qualifications for office 257

(3) The right to run for office: citizens have a right to compete for the offices
that govern them.
The view that these are core democratic rights is treated here as a
presumptive, non-exhaustive, conditional view, rather than a view that is
metaphysically justified or ontologically defended. By this I mean three
things. First, these rights are presumptive rather than necessary: for a
polity to be reasonably democratic it must either secure these rights for
its citizens or it must justify a deviation from them. Second, these rights
are not exhaustive: I presume that there are other political rights that
must be guaranteed by any reasonable conception of democracy – rights
that govern influence over or access to power, rights of expression, etc. –
though I do think that these three are more fundamental or democrati-
cally prior to those. Finally, these rights are conditional in the sense that if
one values democracy then one presumptively gives these rights a promi-
nent place. By contrast, an unconditional defense of these rights would
explain whether they or democracy were primary; whether they were
endorsable for detached or dependent reasons, as Dworkin has described
it; and perhaps whether they derived from natural law, social norms and
agreement, or God’s will.18 Though these are important questions to any
full account of democratic rights, they need not concern us here. Instead,
I presume these rights have a primacy but not necessity for anyone who
endorses democracy, for whatever reasons they do: these rights derive
from a conception of democracy as equal distribution of power, but they
nevertheless may be abrogated if there are good reasons to do so.
There are a few things to notice about these core rights. First, while
they are rights that purportedly accrue to citizens, they leave unaddressed
important questions of membership – of who gets to be a “citizen.” This
is at least consistent with the historical development of these rights:
arguments for universal suffrage, for example, were premised on what
citizenship entailed, and not on who ought to be a citizen per se.19
Second, the first two rights are interrelated: the right to vote is mean-
ingless without a sufficiently expansive choice set: that is, without the

18 Dworkin (2000: 186–7). Though I presume these are detached, they may also be depen-
dent. In fact, Dworkin uses these terms in slightly different ways: dependent views
depend on across-the-board measures of equality; detached views of democracy are
premised only on the equal distribution of power. Thus do I say, “among other things.”
I find Pettit’s version of this tradeoff of principle and consequences to be clearer, but I
use Dworkin’s here because it speaks more directly to issues of political equality (Pettit
1997).
19 As Keyssar (2000) has documented, there have been and continue to be many arguments
for the extension of suffrage to non-citizen residents or workers of a nation who might
not be eligible to receive full rights of citizenship. Here I am only considering why these
three rights should correspond to citizenship, and I leave unaddressed whether others
should also be afforded them.
258 Andrew Rehfeld

right to choose. It is unlikely that any system in which only one candidate
were legally permitted to run (say, Hussein or Castro) would be a plau-
sible democracy at all (let alone normatively legitimate) even if universal
suffrage were guaranteed. The right to vote must be paired with a right
to an unconstrained choice for office as a starting point for any plausi-
ble theory of self-rule. Of course, the fact that a completely constrained
choice set is not reasonably democratic leaves open the question of how
unconstrained a voter’s choice set can be. Maybe legitimate democracies
need merely to guarantee voters some sufficient but not unlimited choice
for each office. If so, the task would be to specify what this sufficiency
requirement would amount to – perhaps it is sufficient to have a limited
choice between candidates who hold differing views, or who are members
of different groups.20 But the point here is to establish a default position.
Limits on unconstrained choice, rather than unlimited choice need to be
justified.
A third and longer clarification is needed concerning the relationship
between the right to run for office and the right of voters to have an
unconstrained choice set at election time. Although a truly unconstrained
choice set necessarily entails a universal right to run for office,21 the
former is conceptually distinct from the latter. In fact, each concerns
different citizenship roles. A voter’s right to an unconstrained choice set
is derived from the citizen as voter: as I said above, the unconstrained set
gives meaning to the right to vote. By contrast, the right to run for office
concerns the ability of citizens to offer themselves as potential rulers.
While both are forms of political participation we should not necessarily
presume, nor do we usually presume, that all forms of participation are
justified by the same sorts of arguments. Given that there will likely be
justifiable limitations on a voter’s choice set and that these limits will likely
restrict which citizens may run for office,22 we should work separately to
establish these rights on two different bases.
I want to raise three issues concerning the right to run, explaining:
(i) why we should presume such a right; (ii) its distinction as a mid-
point between conceptions of ancient and modern equality; and (iii) why

20 I remind readers that many other conditions would have to be met for a government to
be democratically legitimate, including, I presume, the necessity for robust free debate
between candidates.
21 It is logically impossible for a voter’s choice set to include every human being who wants
to run and to forbid some human beings from running despite wanting to.
22 This is only a probable limit. We could well imagine that the only constraint on voters
is to vote for fellow citizens. In such a case their choice set is radically limited (from “all
human beings” to “all citizens of this state”), without limiting at all a citizen’s right to
run themselves.
 On quotas and qualifications for office 259

the right to run provides a rights-based articulation of the principle of

distinction that is a hallmark of representative government.
(i) First, a citizen’s right to run for office emerges from a consideration
of democratic political equality, of what it means to equally partici-
pate in a democratic polity. The intuition here is that politically equal
citizens should all have an equal right to offer themselves up for selec-
tion, even if we cannot reasonably demand an equal right to be cho-
sen (for reasons to be developed momentarily). The intuition can be
generated by imagining that only one small group – say white men –
are legally permitted to offer themselves for selection. We would
rightly reject such a provision as anti-democratic and presumptively
wrong. This right to run for office is broadly endorsed by a range of
democratic theorists, even though none of them has specified what
it means in practice or in any other systematic way23 (Dahl 1991:
222; Beitz 1989; Dworkin 2000; Christiano 1996; Rawls 1971).
(ii) Second, it is useful to note that a right to run for office is also not
expressible in the distinction between ancient and modern forms
made familiar by Constant and expanded more recently by Manin.24
In this first distinction, ancient political equality meant that each
citizen had an equal right to serve in office; modern political equality
means that each citizen has the equal right to choose who serves in
office. Modern political equality has thus emphasized the rights of
citizens to choose their rulers, to participate as voters (deliberate about
candidates and issues; have equality of influence over who represents
them, etc.). Midway between the ancient and modern ends, midway
between the equal right to serve and the equal right to choose, stands a
third form of political equality: the equal right to run for the offices
that govern us.
(iii) Framed in this way, a political right to run for office is an institutional
method of tempering the aristocratic tendency of government. With

23 Dahl comes closest to justifying a right to run for office with a discussion of why each
citizen has a right to fully participate in ruling the polity (Dahl 1991). But this leads
only to a right of equal participation in citizen deliberation and voting and never actually
establishes a right to run for office as differentiated from these other citizen activities.
24 Constant (1988), Manin (1997). There are other ways of conceiving of political equality,
and most promising are the distinctions introduced by Beitz: equality of the individual
vote, equality of (group) electoral success, and equality of legislative outcome. As Beitz
rightly observes, these three may not be mutually achievable. As Lublin has shown,
for example, ensuring group electoral success may come at the cost of worse legislative
outcomes for that group. Similarly, equal voting shares may not produce group electoral
success. In a sense my questions arise from the other side of the coin: ensuring these
kinds of political equalities (particularly of group electoral success) often means a cost
of individual political rights intimately related to core democratic presumptions (Beitz
1989; Lublin 1997).
260 Andrew Rehfeld

Manin, in pure democracies the right to rule is reflected in the insti-

tution of rotation in office, whereby any citizen who wishes stands
an equal chance to serve, their selection being overseen by the use of
lot. Indeed, corresponding also to Aristotle’s argument in The Poli-
tics, a polity can claim to rule itself in part so long as its citizens have
a reasonable expectation to rule and be ruled in turn. In contrast,
citizens of representative government cannot reasonably expect to
rule at any time. As Manin argues, the distinction is what Madison
and Sieyès explicitly argued for, and we might add it is endorsed by
Guizot and Mill in the nineteenth century and in admittedly differ-
ent forms by both Schumpeter and Dahl in the twentieth century.
The exclusion found in contemporary representative government
means that most citizens are more likely to be hit by lightning than
ever to serve in their national legislature (Rehfeld 2005).
At the same time as securing that these presumptive rights move rep-
resentative governments in a democratic direction, the possibility that
they might be abridged in order to enhance democracy allows institu-
tional designers to embrace publicly the aristocratic formulation and
help self-consciously to shape, rather than be shaped by, the exclusion
of representative government. Here, we can now be sensitive to whether
qualifications for office create a de facto governing elite that, as Manin
has illustrated, ensures that being a millionaire is a virtual qualification
for office. Rather, it is to be far more explicit and public about the kind
of aristocracy we are forming, in an attempt to make it more democratic
in spirit if not in nature.25
The discussion in this section can now be summarized in Table 9.3,
which outlines these three principles of political equality (column 1), lists
types of regimes that are commensurate with that principle (column 2),
and provides examples of regimes that fit (column 3). As the table shows,
modern political equality in the Constant/Manin articulation is incom-
plete because dictatorships, and tyrannies more generally, can guarantee
an equal right of voters to choose who rules.26 What is needed, as I said
above, is a provision about the real choice set they face which requires
that citizens be permitted to run, and that they stand a chance of winning
that is institutionally equal to others who are running, an issue to which
we now return.

25 Justifications for qualifications thus explicitly conform to the prescriptions of Repub-

licanism as articulated by Philip Pettit (1997). For one recent example of this, see
McCormick (2006).
26 Similarly, Dworkin notes the inadequacy of equal voting power: “In totalitarian dicta-
torships private citizens have equal political power: none” (Dworkin 2000: 191).
 On quotas and qualifications for office 261

Table 9.3 Three forms of political equality

Commensurate with
Principle of political equality these kinds of regimes Examples

Equal right to rule Democracies Fourth-century Athens

(i.e., ancient political equality)
Equal right to select those Dictatorships; Present-day Syria; Iraq
who rule representative systems; 1979–2003; United
(i.e., modern political equality) democracies States; France; India;
fourth-century Athens
Equal right to run for Democracies; US; France; India;
office/right to choose representative systems fourth-century Athens.
whomever one wants.

III.2 What an equal right to run for office entails

The right to an unconstrained choice set is relatively straightforward: cit-
izens should have the right to choose whomever or whatever they want to
fill the offices that will make and enforce the laws that govern them. This
right extends even to choices that are bafflingly stupid: for example, the
right to choose a fichus tree for their President, a choice no more or less
sensible than the choice of a newborn or merely an adult with decreased
or marginal capacity for rational thought. The right to an unconstrained
choice set is a right of self-determination, even to the extreme. Some
might think that completely undifferentiated choices would not be cov-
ered by this right since it is a right to choose someone with the capacity to
rule, but fichus trees (along with babies and incompetents) have no rul-
ing capacities.27 While I acknowledge the apparent force of this problem,
it is ultimately question-begging: there is no way to create a criterion
of “meaningful choice” or “proper objects of choice” without limiting a
voter’s choice set in the same way that “only women for this seat” or “only
Jews for this one” does. Now it may be that the criterion “meaningful”
or “proper” is less democratically problematic than some other (say, the
securing of descriptive representation). But that is precisely what has to
be worked out.28
What, then, does “an equal right to run for office” mean? At a mini-
mum, a right to run for office appears to be dichotomous: either one is
permitted to run or one is forbidden to run. By this view, an equal right to

27 I thank David Estlund for sharing this objection.

28 I thank Estlund and members of the Brown political philosophy workshop for their
comments on a version of this argument presented there in October 2006.
262 Andrew Rehfeld

run for all citizens would be realized when every citizen had the permis-
sion to run. But treating this right as dichotomous would separate it from
the treatment of other political rights that are considered by reference to
their enjoyment and not mere permission. To use a close analogy, it would
not be sufficient to grant every citizen the permission to vote without
considering whether other systematic or institutional factors restricted
the enjoyment of that right. The equality we care about refers to citizens
having an equal ability to actually cast a vote. Equal voting rights, and
equal rights to run, constitute two parts – permission and enjoyment. A
similar approach should apply to this right.
What then does it mean to equalize the enjoyment of a right to run
for office? For one thing, it does not mean that all citizens have an equal
probability of success, for this would make an equal right to run identical
to an equal right to serve. Further, we must find an articulation of this
right that does not impinge on a voter’s ability to use their own internal
judgments to determine for whom to vote, even though these rules (such
as “vote for the most competent”) may radically reduce the probability
of success for many candidates.
I want to propose that the enjoyment of an equal right to run means
equalizing the effect of external constraints on each candidate’s probabil-
ity of success; or, in terms we developed in Section II, a set of institutions
in which no voter-exogenous qualifications for office existed. Institutional
rules should treat citizens equally who choose to run or compensate for
external social factors that leave citizens unequally situated relative to
these external constraints. In more formal terms, an equal right to run for
office means that voter-exogenous probabilities of success must be equal
prior to being filtered, unequally, by internal voter judgment.
To put this more formally, then, an equal right to run for office entails
the following:
Equal Right to Run for Office = For every elective office with authority over
any individual citizen or with the authority to vote on laws that will govern the
citizen:
i) the citizen must be legally permitted to run for office;
and
ii) the citizen’s voter-exogenous probability of success in her run for office must
be equal to any other citizen’s.

As I said above, the first clause expresses a necessary but insufficient

condition: an equal right to run for office must at least entail the per-
mission to run. The key provision is the second clause, particularly its
specification of a “voter-exogenous probability of success.” I will treat
this now at some length.
 On quotas and qualifications for office 263

In any given election, many factors will affect a candidate’s probabil-

ity of success. Taller, conventionally attractive, and articulate candidates
will stand a better chance than shorter, inarticulate, and conventionally
unattractive candidates. Candidates whose views are closer to a majority
of voters may be more likely to win than those whose views are farther
from a majority of voters. And if you live in a district that has a majority
of African-Americans, being black increases your probability of success.
Of these kinds of factors, we can distinguish those that result primarily
from voter judgment (i.e. internal voter constraints) versus those that
result primarily from rules that are external to voter judgment (even
if they are written with voter judgments in mind). So, for example, of
the factors I have just listed, all except gerrymandering are factors that
are endogenous to voters – the reason that taller, articulate, and con-
ventionally attractive candidates have a better probability of winning is
due almost entirely to voter preference for these things. By contrast, the
reason that African-Americans stand a better probability of winning in a
black electoral district is a combination of voter-endogenous and voter-
exogenous factors: (i) voters prefer to vote for someone of their own race
(voter-endogenous); and (ii) electoral laws group similar races together
(voter-exogenous). So, “voter-exogenous probability of success” simply
means the probability of success given the rules that shape electoral
contests.
Focusing on exogenous factors means that the use by voters of internal
systematic rules (such as “only vote for Republicans”) to choose candi-
dates does not constitute a presumptive violation of a citizen’s right to run
for office. And this corresponds to our notion of what modern represen-
tative government entails vis-à-vis democratic exclusion and distinction,
and now makes clear the importance of that initial discussion. As we saw
above, representative government is a distinctive form because it allows
for such distinctions to be made. And often these distinctions (such as
“vote for the most competent”) are advisable, should be encouraged, and
arguably justify representative government as preferable to direct democ-
racies. But even when voters use bad rules to decide for whom to vote –
rules that are unjust or ill informed – it seems inappropriate to describe
this activity as a violation of another citizen’s right to run for office. When
a voter is a racist, or sexist, or merely votes with an eye to his own good
despite the harm it will cause the whole, it expresses the venality, igno-
rance, or simply unjustness of voters, but it is not an obvious violation of
anyone’s right to run. So an equal right to run for office is thus consistent
with formal systematic rules that voters may use internally to limit their
own choice set in exercising their judgment about for whom to vote. An
equal right to run is violated only by systematic rules exogenous to voter
264 Andrew Rehfeld

judgment that alter the probability of one kind of candidate winning

a contest. Put simply, qualifications for office always violate the equal
right to run for office at the heart of any plausible account of democratic
legitimacy. However, they may at times be used to counterbalance other
voter-exogenous qualifications.
The distinction here maps closely onto the conceptual distinction
between justice and legitimacy: conceptually, a citizen may rely on bad
judgment to choose for whom to vote without that choice being obviously
illegitimate for the use of that bad judgment. This could happen, for
example, if the legitimacy of an outcome was merely a question of con-
sensual procedures and nothing else. Of course, one may, with Dworkin
and Shapiro, believe that in practice and substance these two concepts
are intertwined. But conceptually I take them to denote very different
things. Legitimacy here specifies the conditions under which the state
has a moral right to make and enforce its rules. Justice specifies the kinds
of rules and political behavior (including voting) that are morally good.
And while one may in fact be required for the other, this is a contingent
fact not a necessary one.

IV. Conclusion
Quotas and qualifications for office stand not merely to structure the
legislature but to undo the democratic nature of any polity. Qualifications
for office may secure some principles of justice at a direct cost to two
presumptive political rights at the core of democratic institutions: that
voters have a right to elect whomever they want to rule them; and that
citizens have a political right to offer themselves to rule. To illustrate the
problem in extremis, we might endorse the qualification that limits all
offices to those with the DNA of Smith, Jones, and Cooper. Although this
might be a just policy (Smith, Jones, and Cooper are really good, smart,
and decent people), it is insufficiently democratic to be endorsable. Do
current qualifications that restrict in a less dramatic but no less limiting
way, whose intention is purportedly different from earlier exclusionary
institutions, violate similar principles or not?
The purpose of this chapter was to clear the conceptual and normative
ground required to answer such a question: to provide a conceptually
precise definition of qualifications for office, explain how they connect
to foundational principles of representative government, and articulate
the basic problems that qualifications for office pose for democratic
theory. As I have argued, qualifications are usefully conceptualized as
the assignment of differential probabilities of success among candidates.
As a normative matter, we should be concerned with voter-exogenous
 On quotas and qualifications for office 265

qualifications because they always violate two presumptive democratic

rights: a voter’s right to an unrestricted choice set for voting; and a citi-
zen’s equal right to run for any office that makes and enforces laws over
them. As I have argued, I do not presume that these are inviolable rights:
legitimate democracies may well want to violate these rights in pursuit
of other normative ends.29 But framing these institutions by reference to
their rights violations is a first step in figuring out precisely when their
abrogation is justified.
The starting point of any subsequent analysis should be a presumption
against the use of qualifications in favor instead of unlimited constraints
for voters, and for citizens to run for office. Because the right to run for
office, and the right to vote for whomever one wants, are core democratic
rights – political rights that accrue to a citizen on account of their being
a citizen of a liberal democracy – then qualifications for office must be
justified not merely by reference to their purported pragmatic benefits
(to secure voice, political maturity, correspondence with a constituency,
etc.), but against the costs to democratic legitimacy that are involved in
their use at any time. To put it in the more familiar terms of the right
to vote, even if limiting voting rights to an elite class of citizens would
produce better and more just outcomes, this alone would not justify
limiting the vote to that elite group.30 If qualifications are to be used, we
need instead to justify them against their democratic costs.
The next step is to consider the conditions under which such rights
violations are justified, and to do so in a manner that is consistent with
democratic principles. In the international debate about qualifications
these normative costs have been underemphasized. Instead, qualifica-
tions and quotas are defended by reference either to derision of voter
judgment (voters are said to be sexist, racist, or stupid) or to principles
of justice (e.g. correcting for past oppression; combatting current voter
discrimination; better law). Such violations may be justified for these rea-
sons. Doing so entails a preference for justice (as we now conceive it) over

29 I have argued elsewhere that equal voting rights are of trivial value when framed conse-
quentially (“one-person-one-vote”; “make votes count”), as many advocates of alterna-
tive voting systems do (Rehfeld 2005: 192–7).
30 But see Estlund (2007) for the oppposite view. Some might argue that universal suffrage
is itself pragmatically recommendable. That may be true and may be an additional
reason to grant universal suffrage. But it is incidental to the point: the right to vote is
guaranteed by reference to a non-instrumental commitment to political equality and
not simply because of the policy outcomes it may or may not produce. I acknowledge
here that this appears to commit me to a certain non-consequentialist account of rights.
I would say this is an appearance only – I think it is consequentially useful to act as
if these rights were non-consequentialist, and this is ultimately why I think such an
account and framework are useful. I do acknowledge the familiar bag of worms this
opens, and cannot adequately address it in the context of this chapter.
266 Andrew Rehfeld

democracy. Or qualifications may be defensible to correct for inescapable

tradeoffs involved in public choice. Whether these tradeoffs are worth it
depends on having a clear picture of how qualifications function and their
precise democratic costs.

BIBLIOGRAPHY
Baldez, Lisa. 2004. “Elected Bodies: The Gender Quota Law for Legislative
Candidates in Mexico.” Legislative Studies Quarterly 29 (May): 231–58.
Beitz, Charles R. 1989. Political Equality: An Essay in Democratic Theory. Prince-
ton, NJ: Princeton University Press.
Christiano, Thomas. 1996. Rule of the Many. Boulder, CO: Westview Press.
Constant, Benjamin. 1988. “The Liberty of the Ancients Compared with that
of the Moderns,” in Benjamin Constant: Political Writings, ed. Biancamaria
Fontana. Cambridge: Cambridge University Press, 308–28.
Dahl, Robert. 1991. Democracy and its Critics. New Haven, CT: Yale University
Press.
Dahlerup, Drude (ed.). 2006. Women, Quotas and Politics. New York: Routledge.
Dworkin, Ronald. 2000. Sovereign Virtue: The Theory and Practice of Equality.
Cambridge, MA: Harvard University Press.
Estlund, David. 2007. Democratic Authority: A Philosophical Framework. Prince-
ton, NJ: Princeton University Press.
Goodin, Robert. 1995. Utilitarianism as a Public Philosophy. New York: Cam-
bridge University Press.
Guizot, François. 2002. The History of the Origins of Representative Government in
Europe, trans. Aurelian Craiutu. Indianapolis, IN: Liberty Press.
Gutmann, Amy, and Dennis Thompson. 1996. Democracy and Disagreement.
Cambridge, MA: Belknap Press.
Harrington, James. 1992 [1656]. The Commonwealth of Oceana. New York:
Cambridge University Press.
Holmes, Stephen. 1995. Passions and Constraint. Chicago: University of Chicago
Press.
Hume, David. 1987. “Idea of a Perfect Commonwealth,” in Essays: Moral Political
and Literary, ed. Eugene F. Miller. Indianapolis, IN: Liberty Fund, 512–29.
James, Michael. 2004. Deliberative Democracy and the Plural Polity. Lawrence:
University of Kansas Press.
Keyssar, Alexander. 2000. The Right to Vote: The Contested History of Democracy
in the United States, with a New Afterword. New York: Basic Books.
Kittilson, Miki Caul. 2006. Challenging Parties, Changing Parliaments: Women
and Elected Office in Contemporary Western Europe. Columbus: Ohio State
University Press.
Krook, Mona. 2006. “Reforming Representation: The Diffusion of Candidate
Gender Quotas Worldwide.” Politics and Gender 2 (3): 303–27.
2009. Quotas for Women in Politics: Gender and Candidate Selection Reform World-
wide. New York: Oxford University Press.
Kymlicka, Will. 1996. Multicultural Citizenship. New York: Oxford University
Press.
 On quotas and qualifications for office 267

Lerner, Ralph. 1995. “Giving Voice.” Paper presented at the Giornate Atlantiche
di Storia Constituzionale, Laboratorio di Storia Constituzionale “Antoine
Barnave,” Università degli Studi di Macerata.
Lublin, David. 1997. The Paradox of Representation. Princeton, NJ: Princeton
University Press.
McCormick, John. 2006. “Contain the Wealthy and Patrol the Magistrates:
Restoring Elite Accountability to Popular Government.” American Political
Science Review 100 (2): 147–63.
MacGilvray, Eric. 2004. Reconstructing Public Reason. New York: Harvard Uni-
versity Press.
Madison, James. 1961 [1787–8]. The Federalist, ed. Jacob E. Cooke. Middletown,
CT: Wesleyan University Press.
Manin, Bernard. 1997. The Principles of Representative Government. New York:
Cambridge University Press.
Mansbridge, Jane J. 1999. “Should Blacks Represent Blacks And Women Rep-
resent Women? A Contingent ‘Yes.’” Journal of Politics 61 (3): 628–57.
2003. “Rethinking Representation.” American Political Science Review 97 (4):
515–28.
Mill, John Stuart. 1991 [1861]. Considerations on Representative Government.
Amherst: Prometheus Books.
Opello, K. A. R. 2006. Gender Quotas, Parity Reform, and Political Parties in France.
Lanham, MD: Lexington Books.
Pettit, Philip. 1997. Republicanism: A Theory of Freedom and Government. Oxford:
Clarendon Press.
Phillips, Anne. 1995. The Politics of Presence. New York: Oxford University Press.
Phillips, Herbert. 1921. The Development of the Residential Qualification for Repre-
sentatives in Colonial Legislatures. Cincinnati, OH: Abingdon Press.
Pitkin, Hanna Fenichel. 1967. The Concept of Representation. Berkeley: University
of California Press.
Rakove, Jack. 1996. Original Meanings: Politics and Ideas in the Making of the
Constitution. New York: Alfred A. Knopf.
Rawls, John. 1971. A Theory of Justice. Cambridge, MA: Harvard University
Press.
1993. Political Liberalism. New York: Columbia University Press.
Rehfeld, Andrew. 2002. “Representation,” in Dictionary of American History.
New York: Charles Scribner’s Sons.
2005. The Concept of Constituency: Political Representation, Democratic Legitimacy
and Institutional Design. New York: Cambridge University Press.
2006. “Towards a General Theory of Political Representation.” Journal of
Politics 68 (1): 1–21.
2008. “Extremism in Moderation: A Response to my Critics.” Polity 40 (2):
254–71.
2009. “Representation Rethought: Trustees, Delegates and Gyroscopes in the
Study of Political Representation and Democracy.” American Political Science
Review 102 (2): 214–30.
Schwartzberg, Melissa. 2007. Democracy and Legal Change. New York: Cam-
bridge University Press.
268 Andrew Rehfeld

Shapiro, Ian. 1996. Democracy’s Place. Ithaca, NY: Cornell University Press.
Smith, Rogers. 1996. Civic Ideals. New Haven, CT: Yale University Press.
Waldron, Jeremy. 2002. God, Locke, and Equality: Christian Foundations in Locke’s
Political Thought. Cambridge: Cambridge University Press.
Webster, Daniel. 1903 [1820]. “Testimony Before the Massachusetts Constitu-
tional Convention,” in The Writings and Speeches of Daniel Webster. Boston:
Little, Brown: vol. III, 3–7.
Williams, Melissa S. 1998. Voice, Trust, and Memory. Princeton, NJ: Princeton
University Press.
Young, Iris Marion. 1990. Justice and the Politics of Difference. Princeton, NJ:
Princeton University Press.
2000. Inclusion and Democracy. Oxford: Oxford University Press.
Part V

Preferences, persuasion, and democratic

representation
10 Electoral representation and the
aristocratic thesis

John Ferejohn and Frances Rosenbluth

I. Introduction
A modern democracy is conducted by a small body of elected officials
who make the laws and control the state. This fact has been taken by
some to mean that modern democracy is really a kind of elite rule. In
its strong form, this claim implies that modern democratic governments
pursue the interests of an elite or aristocracy; the weak form claims no
more than the evident fact that government is run by a relatively narrow
class of people and leaves open the question of whose interests may be
served by this arrangement. Strong form elite theorists do not necessarily
reject the possibility that the people may have some influence in picking
and choosing which parts of the elite class control government. But they
typically stress how weak and ineffective such controls are, especially
when it comes to getting the elite to pursue public interests. If public
interests are served, on the strong account, they are served gratuitously,
by leaders who happen to be publically motivated, and not because of
any strong incentive leaders may have to govern for the people.
Versions of the elitist view were stated by Schumpeter years ago and
have been developed in various ways by Przeworski, Manin, and Dunn.
The emphases of these theories vary quite a bit but all concur in seeing
elections as devices for picking an elite and not an instrument by which
the people exercise real control over these leaders. All of them reject the
democratic idea that government is, in any intelligible sense, “by” the
people. The core idea shared by these thinkers – that a government of
elected representatives is necessarily aristocratic in some sense – is quite
ancient, dating back at least to Aristotle. And the common conclusion
of this view is that electoral democracy is really nothing at all like direct
democracy as was practiced, for example, by the Athenians.
Two typical complaints about Athenian-style direct democracy are
often run together but they are independent of one another. One,
attributable to Thucydides, is that democratic rule is turbulent and unsta-
ble. Emblematic of this worry is the example he gives of the Athenian

271
272 John Ferejohn and Frances Rosenbluth

assembly deciding to put an uppity city to the sword and then, a few hours
later, reversing the decision, dispatching a fast ship to stop the slaughter.
Another example was the rash decision to attack Syracuse. The idea is
that there is something about direct rule by a large assembly that makes
it vulnerable to demagogic oratory and other kinds of deliberative fail-
ures. Madison summarized this worry when he remarked that even if the
Athenian assembly was made up of copies of Socrates it would still have
been a mob.
The second complaint, due to Aristotle, is that direct democracy is rule
by the poor who are more numerous, and it systematically reflects their
class interest. Democracy’s defect on this account is not that it is willful
or arbitrary but that it is biased to prefer one part of the city. These two
complaints, while they may be held by the same person (as they were in
the case of Madison, for example), would seem to have different kinds of
remedies.
For the first complaint – turbulence, unsteadiness, emotionality – the
best remedy according to Kant (and Montesquieu and Madison among
others) was to ensure that government is indirect or conducted by repre-
sentatives, rather than by the sovereign people themselves.1 Many writers
hoped that in an indirect or representative government power would be
exercised by a better sort of person – by those skilled in leadership, or
at least by educated people, or people who specialized in government.
But many of them thought that, even if none of that was true, indi-
rect government would minimize or eliminate appeals to the public that
might corrode governmental stability and rule of law. Arguably, changes
in Athenian government in the fourth century provide an example: while
every citizen could attend the assembly or be selected to most execu-
tive offices, certain kinds of vital decisions (military and financial) were
reserved to elected magistrates. In this respect even democratic Athens
came to practice a degree of indirect government.
The second complaint may require a different sort of remedy. Assum-
ing the poor are actually a majority, Aristotle thought that either the
electoral franchise ought to be restricted, or else that the other inter-
ests in the city be given some voice or check on what policies should be
pursued.2 Essentially, each of these is a strategy of representation: in the
first case, by disenfranchising some of the poor, political representation

1 The sovereign we are speaking of here is, of course, the people, but Kant saw that the
argument also implied that a monarchy, if it is to avoid despotism, ought to work through
representatives rather than through the direct imposition of the will of the monarch.
2 There is a third option. The majority may be motivated to pursue common rather than
class interests but it seems unrealistic to rely on that motivation in designing political
institutions of a large heterogeneous polity.
 Election and the aristocratic thesis 273

will tend to weigh the interests of the wealthy more heavily. The second
strategy gives institutional roles to people with wealth that permit them
to project and protect their interests; in this sense the wealthy are over-
represented in the corridors of power. This second strategy amounts to
a kind of mixed government strategy and is exemplified by the system
of institutional checks of the kind that Polybius admired in the Roman
constitution.
The strategy of mixed government has been especially attractive in
modern democratic conditions where franchise restrictions are generally
hard to justify. Aristotle himself proposed both strategies at various points
in the Politics, sometimes favoring a restriction on citizenship, and some-
times favoring a kind of mixed democracy. And many other classical and
modern writers have done so as well. Interestingly, one of the institutions
that Aristotle thought could help moderate democracy into a good form
of government (politeia) was election. So for Aristotle it seems that the
two evils of democracy – turbulence and class bias – may yield to the
same remedy.
While the idea of a government conducted by elected representatives
has become commonplace in the modern world, it has two distinct fea-
tures that Madison, Rousseau, and many others have noticed. First, if
representatives are drawn from an elite class, they may pursue the inter-
ests of that elite, at least to some extent. This is, of course, exactly what
those who favored mixed government anticipated and endorsed. Second,
and more problematically, whether or not representatives come from a
distinct social class, they may themselves constitute a class of a certain
kind and would be expected to some extent to pursue their own interests
at least some of the time. If they do, then there is a natural sense in which
we could understand representatives to be a kind of oligarchy, united in
pursuing their own interests rather that those of the public. Nowadays
we call this a problem of agency, and it seems to us to form an alternative
basis for the aristocratic hypothesis.
Political representation is prone to a distinctive class of agency prob-
lems for several reasons. In ordinary agency relations the concern is to
design a “contract” – or incentive system – for agents who want to pursue
private remuneration or leisure. This is ordinarily done by establishing
a system of rewards and punishments for the agent which are condi-
tioned on some measure of her actions. But the actions of political agents
are very hard to observe, and elections, the typical way of disciplining
political agents, are a crude and imperfect way to control officials; they
happen infrequently and they can usually only punish or reward officials
by withholding or awarding office. Second, representatives have a wide
range of possible motivations: like ordinary agents they may desire to
274 John Ferejohn and Frances Rosenbluth

increase their own wealth and leisure but they may also desire to pur-
sue ideological goals. Finally, there is a reluctance on the part of many
people to see elected representatives as employees hired to pursue the
interests of the public. Elected officials of course foster this view, claim-
ing not to be mere servants but to be leaders who symbolize the dignity
of the public and who ought to be trusted to exercise their own vision
and resourcefulness.3 Such deferential attitudes towards elected officials
may further limit the kinds of electoral “contracts” that are available for
selecting and incentivizing leaders; and they may limit the willingness of
voters to enforce such contracts by punishing poor behavior. For these
reasons, elected representatives usually have a great deal of latitude to
pursue their own goals. On the surface, then, there appears be a great
opportunity for a kind of aristocratic rule – rule by elected officials –
under the cloak of nominally democratic procedures.
There are, however, countervailing forces. Because elective office offers
so many attractive options, we expect there to be intense competition for
it. In equilibrium, we expect the value of office to be dissipated in this
competition; indeed, in some models, one would expect aspirants to
willingly incur in campaign costs the full expected value of the office,
discounted by their chance of winning it. Or they might promise post-
election benefits that would reduce their (private) value of winning office.
Aspiring office-holders might be expected to provide benefits to electors
such as by bribing them with favors prior to the election, as was described
in Lewis Namier’s (1929) wonderful study of eighteenth-century
England. Mark Kishlansky (1981) has documented such activities in the
previous century, as has Gordon Wood (1991) during the early years of
the American republic. More generally, we expect campaigns to be based
on promises of future rewards for voters, sometimes private, sometimes
public. Such practices will not always benefit voters, but they will tend to
erode any gains from office that might otherwise accrue to the elected offi-
cials. Rather, policies are likely be chosen that will please those who have
effective control over access to office such as contributors and activists.

3 Perhaps this phenomenon is not confined to political agents. Potential agents have a
generic incentive to reduce interagent competition, and to some extent high-status agents
(usually labeled “professionals”) have succeeded in getting state assistance in doing this
by imposing entry restrictions in the form of licensing or training requirements. Such
professional requests are always, more or less plausibly, based on concerns to protect
the health or safety of the public (i.e. the potential principals), but they invariably result
in conceding more freedom of action for agents. Often part of the price for getting the
public to go along with such regulations is a more or less credible commitment on the
part of the profession to self-regulate by means of a code of conduct or something of that
sort.
 Election and the aristocratic thesis 275

The point is that competition can limit the capacity of the elected officials
to rule in their own interest.
The thrust of our argument is not, on balance, very supportive of any
strong form of the aristocratic hypothesis. While political agents have a
great deal of scope for autonomy while in office, they are unlikely to gain
much from it. The potential agency gains are liable to be competed away
by other aspirants to public office.4 Instead, political agency is likely to
favor those in society – voters and contributors – who have a comparative
advantage in monitoring the actions of representatives. Insofar as there
is a political aristocracy, it is to be found in those privileged principals
rather than among the agents.

II. Democracy
Definitions of democracy are inherently controversial because any defini-
tion involves value judgments. One kind of definition focuses on partic-
ular institutions and how they function. In the classical period the focus
was on the use of lottery to choose officials, and direct rule by popular
assemblies. More recently, many have followed Dahl (1971) in requiring
well-functioning competitive elections as the defining feature. Another
approach emphasizes democracy’s connection with equality and requires
that everyone have an equal opportunity to rule. (Aristotle’s notion was
that democracy is a system in which everyone takes turns ruling and
being ruled.) As we want to leave open the possibility that a government
of elected officials may (or may not) be democratic, our definition of
democracy focuses on whose interests guide governmental policy. We
claim that a government is democratic if its policies reliably track the
interests of a majority, and that this tracking is accomplished through the
agency of the citizens.
In a representative government, our definition requires that govern-
ment policy generally follows the public’s opinion about what its inter-
ests are, at least at times before elections. Political leaders may to some
extent shape or persuade the public, but if they fail at this, leaders may
have to give up on a policy altogether – at least eventually. Other back-
ground conditions must of course be satisfied as well so that the desires
of the majority may reasonably be counted as expressing the views of
a large part of the people as a whole. The specifics here seem histori-
cally variable but nowadays we would insist that both the franchise and

4 This suggests that those who have good outside opportunities have little reason to choose
a career as a representative, so that selection into the political occupation may be drawn
from among the others.
276 John Ferejohn and Frances Rosenbluth

eligibility for office be open to (virtually) every adult, that information cir-
culate quite freely, that elections are frequent, and that votes be counted
equally.
Admittedly this definition is weaker than the idea that the people rule
themselves directly, as the Athenians were supposed to have done when
they determined policy in their popular assembly or decided verdicts in
their popular courts. But, as we emphasized above, even the Athenians
eventually chose to leave the most important matters to elected officials,
and they also relied extensively on mechanisms of legal accountability to
ensure that, generally speaking, all officials (however they were chosen)
had reason to act for the general interest.5 Our definition is not the mere
requirement that policy track the interests of a majority or of the people
as a whole. That might as well be accomplished by a wise, poll-taking,
and benevolent monarch. We insist that the best account of why policy
tracks majority or popular interests must be that voters have some way
of motivating their leaders to take their opinions as to their interests into
account. This implies that leaders have reason to pay attention to what
the voters actually want (their opinions) and not only what is thought
to be best for them (their interests). Democracy can be a poor form of
government if citizens tend to have defective views about what is in their
own interests and if its policies therefore do not reflect their real interests.
This is, in fact, what Schumpeter and Aristotle thought.6
How leaders will respond to public opinion depends on the structure of
the democracy in question. Elected officials may, in some systems, be held
personally accountable before voters in single-member districts (or open-
list systems), or be collectively responsible to them as members of a party.
Either way, as long as voters have a real opportunity to demand such an
account at regular intervals, we would say the system is democratic. This
opportunity must be real, of course. Elected officials or parties must face
a prospect of losing office and this “reality” test implies that elections

5 The use of mechanisms of accountability in Athens was ubiquitous and far more per-
vasive than anything in the modern world. For one thing, the terms of office were very
short – only a year – and there was a requirement that officials be examined before assum-
ing office, and especially upon leaving it. Then too, officials could be impeached before
the assembly or before the courts. And anyone carrying out a public function, whether
elected as an official or selected by lot, or simply acting on his own initiative to push
a proposal before the assembly, was subject to legal liability before the courts. Finally,
there was always the possibility that someone (any citizen) could simply be ostracized for
no stated reason at all, though as far as we know the reason almost always amounted to
getting “too big for one’s britches.”
6 This was also Churchill’s view when he remarked in 1945 upon being voted out of office,
“In my country the people can do as they like, although it often happens that they don’t
like what they have done” (Gilbert 2007: 864).
 Election and the aristocratic thesis 277

must generally be competitive.7 And what that requires depends on the

institutional structure in various ways. But if those conditions are fulfilled,
it seems to us that a system could be democratic while being governed
by a small class of potential aspirants for office. That small class might
well be described as an aristocracy in some circumstances, such as if
membership in it were determined by heredity or wealth or education.
What is critical is that each leader or party be genuinely vulnerable to the
judgments of the electorate.
We think it is a ubiquitous fact about modern democracies that posi-
tions of leadership are precarious. While journalists often complain of an
insulated and protected political class, the fact is that public officials are
always in danger of losing their jobs, and they know it. If you want to
be, or remain, in public office you need public approval, and no matter
how far you have debased yourself in its pursuit, you can never be sure
that you have enough to survive the next test. Fear of losing office is
an existential fact of political life. We need only recall a few prominent
political careers, such as that of Richard Nixon, to see how realistic that
fear is, and to imagine the lengths to which leaders might go to cope with
it. American scholars of congressional elections have long observed that
nearly every congressman has, at one time or another, suffered defeat
or at least a close call that she did not expect. From the perspective of
governing officials in a democracy, then, public life is characteristically a
treacherous business. Perhaps this is less so today than for the Athenians:
Themistocles, the creator of the fleet that destroyed the Persian armies
and brought about the possibility of a vast Athenian empire, was ostra-
cized from the city. And even Pericles himself suffered the loss of office.
The point is that, in a democracy, political leaders can never forget that
they serve at the pleasure of a fickle and unpredictable master.
This “democratic condition” implies that even if politicians as a class
share some interests in common, which of course they do, it is exceedingly
dangerous for them to try to pursue these interests if that pursuit would
risk support among the people. Moreover, even if elected and appointed
officials share some interests, the democratic condition forces them to
compete with each other for scarce political rewards, and this gives them
reason to undercut and double-cross each other. Elected officials may be
described as an aristocracy in virtue of possessing or seeming to possess

7 It is too much to ask that every election be genuinely competitive; rather what is required
is that control of government be competitive among parties or leaders. Therefore, as is
the case in all democracies, the seats of many backbench members of parliament may be
quite securely held even if control of government is insecure.
278 John Ferejohn and Frances Rosenbluth

rare and valuable skills, but they are likely to have a very hard time
converting these advantages into personal or class gains. That is not to
say they cannot succeed, but success in pursuing private interests is not
guaranteed, and from a certain perspective looks doubtful.

III. Representation
As our purpose is to analyze a feature of representative government rather
than studying representation in general, we shall not follow Pitkin’s
(1967) lead in trying to unravel the concept of representation.8 We
shall work instead with a stipulative definition that we think illuminates
the “core” aspects of political representation. Specifically, we take what
Pitkin calls a formalistic view and focus on the relationships of authoriza-
tion and accountability to try to understand the constraints on official
representatives and ask whether or in what sense representatives can be
characterized as an elite. There are two separate ideas here: a person is
authorized to act for a group by being given authority prior to taking
action. Moreover, a person is accountable for her actions on behalf of a
group if she can be rewarded or punished for those actions if they fail to
respond to the group’s interests (by some entity, not necessarily by the
group itself).
The authorization/accountability view significantly narrows our focus,
in that it marginalizes certain normative claims that may work within
practices of representation, such as the idea that a representative is sup-
posed to mirror the represented. But such views are only marginalized
as potential definitions. There may still be reasons for a representative to
try to, or claim to, mirror her constituents in some way or other.9 Doing
so may help her to get selected and to hold onto office, or even to make
wise decisions. And it may make her constituents more inclined to take
her words and acts as authoritative for them.
Election is one way that a representative may be authorized. Repre-
sentatives could, alternatively, be appointed as Senators are in Canada,
or selected by lotteries as in classical Athens, or inherit their office as do
members of the House of Lords. Such officials are rightly regarded as
representatives as long as it is accepted that those chosen owe duties to

8 We take the analysis in her book to demonstrate the extreme difficulty of such an endeavor.
In any case, if it is true that representation is essentially contested, any such effort must
fail.
9 In fact, the agency view can illuminate such appeals by showing when they might be
persuasive. If potential agents can differ in the degree to which they resemble principals,
perhaps in the sense of sharing their preferences, there is a competitive reason for each
of them to be seen as similar to the principal as possible.
 Election and the aristocratic thesis 279

advance certain interests of the represented in appropriate contexts.10 To

be a representative is, in this sense, to be selected according to a conven-
tionally accepted scheme and thereby to be embedded in a certain kind
of normative relationship with a group.11 This normative relationship
may be to represent the desires or opinions of the group (to act as its
ambassador or delegate) or to represent its interests (to act as a trustee).
A representative may be accountable for his failure to fulfill his nor-
mative obligation to represent. Duties to represent might be enforced, if
they are enforced at all, directly by those to whom the duties are owed –
in which case the representative is made accountable before the repre-
sented – or they may be enforced by whoever has the authority to select
the representative. For example, while a representative may have a duty
to pursue the common interests of the people, she may be selected by
voters in a particular geographic district. This is another form of political
accountability. As before, the representative is or is not reelected accord-
ing to the will of a majority. Alternatively, as in the case of fiduciary
duties, duties might be enforced by a third party or by courts of law,
according to some legal or normative standard.12 We call this an instance

10 Regarding representation as a normative relation of a certain kind has the additional

advantage of making sense of “informal” claims to represent, such as when someone
claims to speak for a group or subset of the population. Such claims remain informal
because there is no accepted convention for conferring authority, and because informal
representatives are not entitled to exercise state authority.
11 We are employing an interpretive construct here in order to facilitate institutional com-
parisons. If a people has a set of practices that can best be explained by positing their
possession of a unified concept that supports the idea, then they actually have something
like that concept even if they have no single word for it. For example, the Athenians had
an extraordinary range of practices of holding public figures to account. They required
magistrates (whether selected by lottery or election) to give an ex post account of their
official behavior before a court (and sometimes in the assembly as well). They exposed
anyone proposing decrees or laws to public law prosecutions. And they exposed everyone
to ostracism (and this must have been most dangerous for public leaders). Moreover,
the content of these interrogations – the questions posed – appeared to relate mostly to
whether the official had acted in ways that served the common interests of the city as
expressed in its laws. That the magistrates and other public figures owed duties to act in
the interest of the city, and that, in addition, these duties were enforced through making
them accountable to the people, suggests that the magistrates were expected to act as
representatives. This is obviously a kind of functional definition and is situation-specific.
12 Legal accountability is one mode of enforcement, and traces of it remain in impeach-
ment procedures in modern polities. The Athenian requirement was not only legal but
also popular or political since the courts were popular institutions without professional
judges or lawyers. Modern systems tend to rely mostly on political rather than legal
accountability, a practice the Athenians had largely abandoned by the end of the fourth
century. Impeachment and criminal sanction remain supplementary forms of account-
ability relations as well, even if they are used rarely in practice. We emphasize that duties
may not be enforceable at all: duties can sensibly exist that cannot be enforced. Moral
duties are an example. Where duties cannot be enforced or enforced very well, it is
important that the people select representatives who are likely to have a kind of moral
280 John Ferejohn and Frances Rosenbluth

of “legal” accountability. When an agent is politically accountable to the

represented, the representatives are free to decide to retain her or not,
according to their will. When an agent is legally accountable before a
tribunal, the agent may be removed only for cause: for failing to perform
up to some legal or normative standard. Persuasive reasons have to be
given to the tribunal to justify the removal.
Traces of legal accountability remain in the impeachment procedures
in modern polities, and in the fact that political officials are not com-
pletely immune to prosecution for criminal acts. But mostly modern
governments rely on political accountability to select and de-select rep-
resentatives. We can appreciate this by contrasting it with Athenian prac-
tices. Democratic Athens required its officials and other public actors to
account for their actions in legal as well as political forums and, after 403
BC the balance was heavily toward legal rather than political account-
ability. Their courts were themselves popular institutions without pro-
fessional judges or lawyers, so the difference between the two forms
is perhaps less sharply etched than in modern polities.13 But they did
retain some distinctly political forms, permitting impeachment before
the assembly and ostracism as well.14
We emphasize that some duties may not be enforceable at all. Moral
duties often are not. Indeed, we think representatives have moral duties
to constituents even if, and especially if, their actions cannot be observed
by their constituents. Where duties to represent cannot be enforced or
enforced very well, the people have reason to select representatives who
are likely to have some kind of internal motivation to pursue their inter-
ests. Perhaps this is best done by choosing representatives whose private
interest and/or ideology are correlated somehow with the public interest
they are to serve. But perhaps there is a residual need for public officials
to have moral motivations.
The expression “representative” government is usually used to refer
to a government made up of representatives selected according to

motivation to pursue their interests. Perhaps this is best done by choosing representa-
tives whose private interest is correlated somehow with the public interest they are to
serve.
13 Still, there was a distinction in that legal judgments are supposed to be based on reasons
(of law and fact), whereas political judgments can be more or less arbitrary.
14 The emblematic case of this is the story of the poor and nearly illiterate farmer struggling
to spell a name on a pottery shard (ostraca). The philosopher Aristedes (who was
generally known as Aristedes the Just) asked if he could help: “Whose name are you
trying to write?” The farmer replied “Aristedes.” “But why?” asked the philosopher.
The farmer snorted in reply, “I am sick and tired of hearing of this ‘Aristedes’ always
being called ‘the Just’.” Ostracism was often aimed not so much at those who abused
power or popular trust, it seems, but simply at the famous who had grown perhaps too
big for the city.
 Election and the aristocratic thesis 281

conventionally accepted procedures for allocating offices. It is often used

descriptively, though there are some who think that calling a government
representative is a kind of praise. When “representative” is used as a qual-
ifier of “democracy,” as in the expression “representative democracy,”
the normative terrain is more divided: some people take it to indicate
a kind of qualification or even a negation of the real thing (genuine
democracy), made necessary by the scale and complexity of the mod-
ern nation-state.15 Others regard representative government as a kind of
improvement over direct democracy, either by making democracy more
attractive by moderating its democratic aspect, or by making it more
rational or deliberative (in Madison’s expression: representatives were
expected to “refine and enlarge” public opinion).
Everyone agrees that representatives, as political agents, are supposed
to act in the interests of the people in settings where they are authorized
to act and, as a means of achieving this, to be answerable to them in some
way. But citizens have only meager means for disciplining their represen-
tatives. Representatives’ actions are hard to observe; the occasions for
demanding an account are few, and are shared by a heterogeneous group
of people. For that reason, elected officials do not seem to be on a very
tight leash. Why is this? Perhaps we think of political officials differently
than we do other kinds of agents: as leaders rather than servants. Because
of this we may want to provide them with a fair amount of autonomy to
take the kinds of actions that might best advance our interests. So we
may not insist, as we might in other agency relations, on being able to
audit their actions closely.
These “distortions” in the agency relation remind us that political
agency may be distinctive in permitting political agents a lot of slack or
autonomy. Elite theorists, starting with Aristotle, think that giving leaders
the chance to take independent action can increase the prospects for
beneficial policy choices and stability in government. The distinctive slack
in political agency is what makes room for the aristocratic hypothesis.
Those with more popular sympathies worry that slack will allow official
corruption or the capture of government by private interests. Discussions
of political agency are marked therefore by controversy over how much
scope should be permitted to an agent and how agency duties are best
enforced or at least encouraged.

15 Kant and Madison and legions of modern writers have thought representation con-
stitutes an improvement over direct democracy because it introduces non-democratic
elements in a kind of mixed government, which moderates populist impulses and pro-
duces reasonable or moderate policy. Others have argued that genuinely democratic rule
must be representative for other reasons: for example, representatives are better able to
deliberate effectively and to pursue common interests than the people would be.
282 John Ferejohn and Frances Rosenbluth

IV. Political agency

We argue that, suitably modified, the agency model – a version of what
Pitkin has called the authorization/accountability model – captures the
central aspects of formal representation.16 We hire an official by means
of an election and expect her to employ her legal authority to pursue
our interests in government. We hire a private lawyer for similar reasons:
we expect that lawyer to advance the interests we entrust to her within
the confines of legality and professional norms. In both cases we retain
some degree of control over our representative both ex ante and ex post:
before the fact we make it known what we expect, sometimes implicitly
by voting for the candidate who makes the most attractive offer; and after
the fact we can punish or reward her for her performance. In both cases,
our representative is expected to act in our interest, to be accountable to
us, to warrant our trust, and to explain and justify her behavior when she
appears to fall short.17 And she runs risks of being fired or not paid (or
sued) if she fails to fulfill our expectations.18
We do not deny, of course, that the word “representation” is used in
other ways to describe other activities: as metaphor or synecdoche, as
a theory of meaning or a concept of mind, or as a particular way of
drawing or painting. Some thinkers have sought to put some of these
other meanings on a par with the idea of representation as authoriza-
tion, arguing that political representation must make sense of these other
usages. Within political life, people sometimes make claims that some-
one can speak for another only insofar as he is “like” the other in some
way. Sometimes this is applied to elected representatives and sometimes
to competitors for leadership. Such claims may be persuasive in various
contexts – usually they are aimed to persuade – but they are essentially
normative arguments aimed at influencing action, or perhaps undercut-
ting claimed authority, rather than conceptual claims. Moreover, such

16 Pitkin’s criticism of this view is that it focuses too narrowly on aspects of representation
– that the agent has been given authorization to take certain actions – to the exclusion
of others: she lists several including “having one’s actions attributed to another” and
“having the right to command another” (Pitkin 1967: 51). Neither of these is a part
of the agency view at all as far as we can see, and neither seems necessarily a part of
representation either, unless the principal somehow authorizes them. This may or may
not occur.
17 The best ethnographic account of this relationship, as it appears to the representative,
is still Richard Fenno (1978).
18 This does not imply that agents can easily be removed for non-performance. The typical
situation in agency relations is, indeed, that some non-performance occurs in equilib-
rium. And we do not deny that punishment is harder in political settings, but it is hard
in many non-political settings as well. The difference is not a matter of degree, and
depends on specific features of the agency relationship, political or non-political.
 Election and the aristocratic thesis 283

arguments are not confined to the political realm: you might want to
have as a doctor someone of your own age or gender. But those wants
count only as normative reasons for choosing between doctors. In the
end, the actual picking remains the crucial thing: until you actually do
that you do not have a doctor at all. And once you have picked a doctor
she will be authorized to make certain choices on your behalf, whether
or not she happens to share your gender or age.
We used the expression “suitably modified” to signify that the kind of
agency relations found in public life may differ from agency relations in
the private sphere. In what we shall call the standard model, the agency
“contract” provides ongoing incentives for the agent by attaching rewards
or punishments to her actions. Such contracts also serve to screen or
filter potential agents by making the role of agent more or less valuable
to different kinds of people. There are really two distinct ideas here:
agents, having different preferences than the principal, need to be given
incentives to take actions that are in the principal’s interest. And, some
potential agents are better than others for the principal in some way:
they may be more competent, more ethical, or have preferences more
like those of the principal. These two ideas correspond to the “moral
hazard” and “adverse selection” perspectives. In the first, the focus is on
controlling an agent’s actions while she is in office; in the second it is on
selecting the right kind of agent in the first place.
In either case, there will normally be more slack in political agency rela-
tions than in other kinds of agency relations. One reason for this is that
political principals tend to be collectivities rather than individuals. This
requires that there be some way of making collective decisions – elections,
for example – for resolving disputes among principals who may disagree
about what they want their common agent to do. Virtually every way of
making such decisions creates problems of collective action of various
kinds and opportunities for agents to exploit.19 Second, political agency
“contracts” are typically crude, in that it is not only costly or impossi-
ble for voters to observe most activities done by their representatives,

19 Both of these features of political agency are explored in Ferejohn (1986), in the con-
text of a very simple model. In that paper, the principal’s only way of controlling the
policymaker was the possibility of firing her. Even so, as long as the policy space is
one-dimensional so that there is a median voter, the principal could exert some degree
of control on the agent. But in higher dimensions the agent is basically uncontrollable
unless the principals could somehow agree to judge the agent using a one-dimensional
performance criterion. In effect, only if the principals can solve the collective action
problem and act as a kind of “person” can they hope to get the agent to pursue or
represent their interests. That model was extreme in limiting the tools available to the
principals to control the agent, but the logic of the situation will carry over to much
more complex settings.
284 John Ferejohn and Frances Rosenbluth

but it is also impossible to make rewards and punishments sensitively

contingent even on observed differences. Voter opportunities for express-
ing disapproval are limited to infrequent elections and perhaps by some
legal institutions that permit impeachment or criminal accusation.20 So,
for various reasons, we are unable to treat officials differently depending
on the actions they take or the outcomes that result. Finally, political
agency relations may be subject to relatively weak competitive pressures.
Normally we would expect that a person would refrain from hiring an
agent whose behavior she has little control over. But, as we pointed out
earlier, a citizen cannot really refuse to enter the relationship if its terms
seem unattractive. Such a refusal requires costly efforts at coordination.
In market settings one expects that competition among agents will lead
to the creation of more attractive agency contracts – that is, contracts
that can be conditioned on a finer description of events and that are
more effective in motivating and screening agents. But this process is
often blocked in the political sector (and it is occasionally blocked in
markets as well). Why is this? Political agency relationships differ from
ordinary agency relations in a way that does not really seem to track
any formal feature of agency models. When someone hires an agent,
one might think that a kind of hierarchical relationship is established:
the agent is an employee, a hireling, whose duty it is to do what the
principal wants and whose compensation depends on performance in
some way. Think of hiring a person to mow your lawn or fix your car
or sell your home. This model breaks down a bit in the case of high-
status professions such as surgeons, but for reasons that seems to have
nothing to do with the agency relation. No doubt surgeons have skills
and information that permit them to do things for us that we cannot
do for ourselves; and, in the course of surgery, our ability to monitor the
physician’s actions is probably compromised. And maybe in certain areas
of surgery, the important skills are rare and there is little competition
among potential agents. Anyway, someone who needs surgical services
probably is in a hurry, which places further limits on competition. But
these circumstances are not special to surgeons. Think of having your
car break down on an isolated road with only one garage in the vicinity.
Remember, a couple of centuries ago people thought of surgeons as on
a par with barbers and mechanics.

20 In market settings one expects that competition among agents can improve this sit-
uation. But even there, competition may not eliminate problems. The reason we see
doctors and lawyers and other professionals adopting ethical codes and organizational
modes of enforcing them is precisely because agency relations are imperfectly policed
by incentive contracts, and potential agents seek efficiency gains through self-policing
or governmental regulation.
 Election and the aristocratic thesis 285

Perhaps officials in a democracy enjoy an especially high status because,

though they may be our employees, they are expected to “lead” us, to
direct or command our actions. Why would that matter? Possibly, it is a
matter of psychology: it may be difficult to accept orders from a subordi-
nate, so people may psychologically “elevate” political agents, projecting
onto them some undeserved superiority. Another answer is to say that
elected officials are not “our” servants as collections of individuals but
are servants of our collective interest. We elect them not to subordi-
nate ourselves but because we think they have special political skills that
make them better able to discern what is in our collective interest than
we, as disparate individuals and groups, would be. These seem plausible
enough answers, but dangerous ones, and we doubt that robust democ-
racies could concede so much to their leaders. Another possibility, one
that seems more consonant with a democratic culture, is that elected
leaders enjoy a kind of democratic deference out of respect for elections:
that representatives who have been chosen in orderly and fair elections
may be due a special degree of respect that is traceable not to them as
individuals but to the people.
The belief that political leaders are somehow above individual citizens
may explain why we are reluctant to demand more effective agency con-
tracts concerning them. This may account for why elected officials enjoy
relatively long terms, why popular recall is rarely available, and why bind-
ing instructions to elected officials are almost never countenanced. To
restrict the terms or employment conditions of our elected officials may
be to express a kind of contempt and not only for them as officials, but
also for the people who elected them. To be sure, these are all variables
rather than constants because there is nothing that dictates the appropri-
ate length of a term, for example. Therefore, if we are right, hierarchical
beliefs are probably variable as well. The Athenians, for example, per-
mitted officials to serve only a year and required them to undergo ex ante
oaths as well as stringent ex post audits, and subjected them to a number
of other ongoing checks on their official performance. Of course many of
their officials were ordinary citizens chosen by lottery, and people may
have thought they had to be kept on a short leash. But their elected
officials were constrained in the same way.
It seems significant that most of the restrictions on the agency contract
are not in any sense “natural,” but are actually imposed on us by our polit-
ical agents. Political manipulation rather than deference may explain the
distance at which politicians hold voters at bay. In competitive settings
one would expect agents who offer poor contract terms to disappear, so
one expects agency contracts to be attractive. But typically the restric-
tions on political agency are imposed collectively and not individually.
286 John Ferejohn and Frances Rosenbluth

Sometimes they are placed in the constitution: the Bonn Constitution for-
bids binding instructions on representatives and the American Constitu-
tion has been interpreted to resist term limitations, to take two examples.
More often, agents set the rules by statutes or chamber rules that govern
how observable their actions are and the extent to which their rewards
and punishments can be made to depend on them. These practices limit
the extent to which contract terms can be subjected to competitive pres-
sures that might force them to be more favorable to principals. Sometimes
competitors do try to offer new and more attractive terms – such an offer
was the basis of the “Contract with America,” with its pledge of term
limits and the like – but such offers are neither credible nor, if they are
accepted by gullible voters, actually implemented.

V. Representative government and aristocratic rule

Madison thought that part of the genius of American government was
that it was indirect: the sovereign power, the people, had no actual role in
government (Federalist 63). Governing is done by delegates who obtain
grants of power from voters, through election, and are entitled for a
period of time and within certain limits to exercise it on behalf of the
people. There is no doubt that Madison thought this feature of the new
Constitution – the fact that the people had no direct role in government –
was not an unhappy compromise forced upon it by the size or extent of
the new republic, but an essential feature of a good form of popular
government.21

21 He voiced apprehension of a direct appeal to the people in Federalist 49 as well, where he

explained that such appeals would undermine the development of popular veneration
for the new Constitution. That Madison worried about direct popular involvement does
not make him anti-democratic in the way that term is used nowadays. His preference
was explicitly for a republican government in which all authority was drawn directly
or indirectly from the people. He thought, however, that direct popular involvement
in government would undermine the possibility of stable republican government by
unleashing passionate appeals to transient majority sentiment. In fact Madison went
much further than this in seeking to limit the indirect influence of popular majorities in
government. He thought the more dependent a government was on the people, the more
dangerous it was, and the more need there was to control and check its power. The state
legislatures were dangerous to liberty precisely because of their popular proximity, and
the federal House of Representatives similarly dangerous, though less so, for the same
reason. For that reason the legislative branch, which was necessarily close to the people,
posed a powerful threat to republican rule: Madison said that the legislative “vortex”
was the chronic source of turbulence, irrationality and danger to stability. His central
arguments for establishing a national government (in Federalist 10), for federalism (in
Federalist 45–46), for checks and balances (Federalist 51), all aimed at restraining popular
influence. Finally, in what seems the direct expression of a desire for a kind of aristocratic
rule, albeit of the republican kind associated with Rome, he argued that by adopting large
 Election and the aristocratic thesis 287

Madison was not alone in thinking that a government of representa-

tives was to be preferred to a direct democracy. Kant also thought that
democratic government – by which he meant a direct democracy – was
necessarily will-driven and for that reason despotic.22 “Every form of
government that is not representative is properly speaking without form,
because one and same person can no more be at one and same time the
legislator and executive of his will (than the universal proposition can
serve as the major premise in a syllogism at the same time as be the sub-
sumption of the particular under it in the minor premise)” (Kant 1970
[1795]: 114). A government of representatives was much more likely to
be able to act under the control of reason and therefore to be moderate
or temperate.
Kant’s reasons for preferring representative government have nothing
to do with representatives being in some sense “better” or more virtuous
than the people. Rather, his argument is based on the defining feature
of indirect rule: the person who takes action – the representative – is
not the one source of valuation or preference. He is an executive who
acts on the interests, passions, preferences, or opinions of others rather
than on his own valuations and he is, for that reason, better able to
act rationally. This is a kind of separation-of-powers argument in which
the passions and emotions of the people are acted on, and disciplined
by, their representatives. One sees in it echoes of Plato’s division of the
soul (Plato 1987), where the rational part controls the appetitive and
passionate parts. In that respect, it also echoes Montesquieu’s idea that
the separation of legislative (evaluative) and executive powers is neces-
sary to avoid tyrannical rule, by which he meant rule by passions rather
than reason. But Kant argues here for a vertical (between people and
their representatives) rather than a horizontal (between departments of
government) separation of powers.
Several Greek writers had already expressed similar misgivings about
direct democracy. Thucydides chronicled the passionate and turbulent
politics of democratic Athens during the Peloponnesian War which
he attributed partly to the effects of democratic rule. Xenophon,

constituencies for the House of Representatives one could hope that elections would
tend to choose virtuous leaders who would be likely to “refine and enlarge” on public
opinion.
22 “[D]emocracy, in the proper sense of the term, is necessarily a despotism” (Kant 1970:
114). By “proper” democracy Kant meant direct democracy, and he went on to specify
what was defective about direct rule (whether democratic or not). Kant, like Madison,
favored a popular component in government but insisted that the authority of the people
be exercised only through representatives. He went on to criticize the ancient republics –
surely referring to Rome and possibly to Athens as well – for failing to understand this
and therefore degenerating into despotism.
288 John Ferejohn and Frances Rosenbluth

condemning the popular reaction to an attempt to introduce legal pro-

cess into the famous trial of the generals, quoted an incensed member
of the assembly as saying, “It is outrageous to say that the people cannot
do whatever they want.” Unmediated or direct democracy was seen as
impulsive and subject to bouts of irrationality. Indeed, there is reason to
think that even ordinary Athenians thought there was reason to temper
or moderate direct rule: after the democracy was restored at the end of
the fifth century, the Athenians embraced various judicial institutions
that could limit and control the actions taken by the assembly.23
Aristotle’s normative views were complex. He did not reject democracy
as such but sought ways to ensure that a democratic government would
act moderately. In Book VI of Politics, for example, he expressed admira-
tion for an agrarian democracy in which most of the people lived outside
the city and were too busy farming to take much interest in politics, and
would therefore prefer to use elections to select their leaders (rather than
choosing them by lottery as the Athenians did).24 He had already defined
democracy as a kind of government in which the citizens actually rule in
some direct sense, whether they take turns ruling as in a good constitu-
tion, or rule at the same time as in Athens. He did not expect either kind
of direct rule in an agrarian democracy. Elections would be employed
to choose magistrates and would select only certain kinds of people into
office. Possibly these would be especially able or wealthy which might
conduce to a better government. In any case, electing the able or the
wealthy would recruit them into public service rather than leaving them
outside of government where they could cause trouble. We imagine that
he thought it a saving feature of Athenian democratic practice that some
of its officials – the most important ones – were chosen by election rather
than by lottery.25 That, at least, permitted the selection of officials with

23 Even in the heyday of direct democracy in the fifth century, Athenian government
had important elements of mixture that incorporated the rich and upper middle-class
governmental institutions. Election of the generals was an important kind of mixing
– the generals tended to come from those who had military or naval equipment and
experience as far as we know – as was the extension of elections to choose financial
officers.
24 Aristotle (1962: Book VI, ch. 4).
25 While we present the argument from “competence” here it is by no means clear that that
is the reason the Athenians used elections to pick generals throughout their history, and
to pick financial magistrates in the fourth century. It may also be that elections allowed
the state to harness potentially dangerous people with private armies. The powerful
families that produced Themistocles, Cleisthenes, Pericles, and others would have been
a danger to the state if they were not in its service. Indeed, Themistocles ended his life
in exile. The oligarchic coups of 411 and 404 were led by such people who may have
been eclipsed or threatened by parvenus. And the creation of elected financial offices
may have been an attempt to enlist wealth-holders in the service of the state rather than
any recognition of competence or superiority.
 Election and the aristocratic thesis 289

the ability to make reasoned policy decisions in complex and dangerous

policy domains, though it did not guarantee that they would.
Later thinkers, such as Schumpeter, have put a somewhat different
spin on their preference for a government of representatives. Schum-
peter (1942) thought that ordinary people, being unfamiliar with political
life and issues of public policy, would be incompetent to make reason-
able political choices. Public officials may or may not be talented in
this respect but their day-to-day experience would at least lead them to
acquire some skills relevant to governing. So he argued that the people
ought to confine their involvement in politics to choosing a government
and perhaps removing it later if they did not like the results. While he did
not argue this directly, he probably thought that the comparative advan-
tage of elected elites partly explains why it is that modern democracy
takes the representative form that it does.
We imagine a representative government can become a system of elite
rule in two different ways. Michels (1915) claimed in his “iron law of
oligarchy” that any organization will devolve ruling authority on a small
set of people. Michels thought elite rule could be an emergent feature of
any organization irrespective of the beliefs or values of the organization’s
members or of the mechanisms for choosing leaders. Schumpeter appears
to have shared a similar view, since his arguments do not seem to turn
directly on the use of election itself, though we admit it is less than clear
from his writings. On this view we would expect to see elite rule within
the lottery-selected Athenian boule as well as in an elected assembly.
Against this view, we may contrast theories that claim that an organi-
zation is elite-dominated if it employs a method of leadership selection
that tends to choose leaders who have special qualities. That is to say,
certain selection processes may “recognize” certain people with special
qualities relevant for leadership. Madison and Aristotle entertained theo-
ries of this kind and so does, with qualifications, Bernard Manin (1997).
Aristotle and Manin thought that the electoral mechanism was a device
for elite selection. Madison thought that elections in large districts would
perform in this way.
Manin appears to accept the idea that elected representatives will gov-
ern differently than ordinary citizens would if they were rotated into
office. It is not that he thinks that representatives are likely to be a better
sort of person. Nor does he claim, like Schumpeter, that elected rep-
resentatives are likely to be more competent in matters of policy and
government. Manin argues that it is simply the use of elections as the
device for selecting leaders that produces aristocratic tendencies, locat-
ing its critical feature in the tendency of voters to make choices based on
certain kinds of considerations. He sought “to deduce the inegalitarian
and aristocratic effects [directly] from an abstract analysis of election”
290 John Ferejohn and Frances Rosenbluth

(ibid.: 135). By doing this he hoped to drive home the conclusion that
these effects are intrinsic to the electoral mechanism itself.26 However, he
acknowledged that his deductive argument depended on certain empir-
ical (contingent) assumptions – essentially that voters will tend to elect
those they think are superior in some relevant way: “The dynamics of
choice and cognitive constraints usually leads to the election of repre-
sentatives perceived as superior to those who elect them” (ibid.: 145).27
So he could not reach all the way to the analytic conclusion he sought.28
Still he thought these contingencies were likely to be satisfied in most
circumstances of modern democracy.
We agree. The most that can be said is that voters tend to choose those
whom they believe to possess some valuable characteristic for political
leadership, and they may well be wrong both in the value of the charac-
teristic and who are likely to possess it. Second, whether this constitutes
an aristocratic tendency seems to depend on the idea that the charac-
teristics in question are somehow valuable or attractive in some wider
sense. Third, we have to think that these characteristics are fixed prior to
election and not conferred by the fact of being elected. After all, if elite
characteristics were automatically attributed to those who happen to be
elected, election would simply constitute an aristocracy. It may be true
that marks of superiority are not too mutable, at least not deliberately
so, in a variety of circumstances, but we doubt that that it is anything
like universally true. Moreover, to say that contingent factors matter is to

26 Manin presents his argument as a kind of completion of a project begun by Aristotle,

showing that elections are an aristocratic device and are, for this reason, incompatible
with democracy. In the case of Aristotle we are not yet persuaded. Aristotle could be
read as speaking merely of tendencies or statistical regularities – that lottery tends to
be employed in democracies and election in oligarchies – or else as making something
like a constitutive claim – that something is not a democracy if it uses elections for
important offices. But Aristotle plainly thought that Athens was democratic and it
employed election for significant offices. Moreover, the constitutive claim seems even
harder to defend in light of Politics, VI, 4 where he discusses (agrarian) democracies that
employ elections. In any case, both the constitutive claim (that democracies necessarily
employ lottery) and the statistical claim are distinct from the claim that we think Manin
wants to make: that democracies have good reason to employ lottery and that they will
therefore tend to make that choice. This seems a causal claim and Manin’s “deduction”
is an attempt to elucidate the causal elements of that story. And, of course, it is no defect
of a causal argument that contingencies play an essential part.
27 And even here he qualifies the sense of “superiority” as being only in respect to relevant
qualities for government. There is little sense that the elected are more virtuous in any
moral sense or even more competent except with respect to political skills – or, rather,
reputed political skills.
28 The four assumptions are these: the unequal treatment of candidates by voters which
says that voters may choose candidates in an arbitrary fashion; distinction of candidates
due to the circumstance of choice, which says that voters choose candidates thought
to have some mark of superiority; that certain persons are more salient or visible than
others; and that it is costly to secure public recognition. We agree, of course, that these
factors make it unlikely that as a matter of fact election will treat everyone equally.
 Election and the aristocratic thesis 291

leave open which factors actually matter and how likely they are to arise
in practice. Perhaps other contingent factors operate to defeat the effects
of the contingent factors that Manin cites.
In any case, the qualities of the candidates may not be as important
for the aristocratic hypothesis as what the governmental officials actually
do and what policies they produce. Thucydides’ criticism of Athens was
based at least partly on his judgment that disastrous or unstable policies
were chosen and other writers seem largely to agree. No doubt Thu-
cydides thought that there was some connection between the quality of
the officials and their policies, but after all it was Pericles who led Athens
into the war with Sparta (whose qualities Thucydides would not have
doubted), and Alcibiades (another highly capable leader, even if he was
a flawed character) who convinced the Athenians to undertake the dis-
astrous Sicilian expedition. One has to be open to the possibility that
that connection between capable leaders and good policy is uncertain.
Why not, instead, ask the policy question directly? Doing this may indeed
support an alternative version of the aristocratic hypothesis: the idea is
that, even if elections do not produce better officials, the policies pro-
duced tend to support the public interest or perhaps the interest of some
elite.

VI. Policies in representative v. direct democracy

Athenian democracy had three characteristic institutional features. The
first was what is now called direct democracy: any citizen who wanted
to could attend the assembly and vote or speak on any issue. Second,
every citizen was eligible to be chosen by lottery to serve in the mag-
istracies that governed the city. Third, every citizen was eligible to be
selected a juror, again by lottery, in the very powerful court system. In
these respects everyone could take (equal) turns in ruling the city. In
this section we want to compare direct democracy – where the people
choose policies by some kind of popular referendum – with representative
democracy.
Since the appearance of Downs’s (1957) work half a century ago,
many positive political theorists have shared the intuition that majority
rule tends to perform similarly in different institutional settings, indepen-
dently of whether people are choosing between policies or candidates.29

29 This intuition is supported by the median voter theorem which is thought of as a general
tendency of majority rule in a one-dimensional setting. However, to be accurate, the
statement needs to be qualified by saying it represents a kind of intuition. After all,
it is easy to create game theoretic specifications in which majority rules are extremely
sensitive to institutional details in the sense of departing from the preferred outcome of
the median voter.
292 John Ferejohn and Frances Rosenbluth

If additional institutional structure matters, it is only because these addi-

tional details introduce some kind of bias that constrains the majoritar-
ian property. In a one-dimensional world this intuition is supported in
median voter models, in which the preferences of the median voter deter-
mine the outcome, assuming that distorting institutional frictions, such
as monopoly agenda control, are absent.
For example, in two-candidate competition over a one-dimensional
policy space, under very weak assumptions, both candidates “converge”
on the position of the median voter. And, in an open-agenda legislature
(where anyone is free to make proposals), the legislative outcome will
be at the same point. The median voter theorem may not hold in some
settings, such as when someone has agenda control or if candidates are
insufficiently motivated to seek office or are not somehow free to pick
winning platforms. Except for imperfections or frictions of this sort, the
intuition is that the majoritarian aspect of the institution will dominate.
At least in some special cases results of this kind extend to higher
dimensions, though many complications arise that define such exten-
sions, since in many models equilibria frequently do not exist. But when
an equilibrium does exist in a “legislative game” in which people are
free to propose alternative policies under an open agenda, then a two-
candidate equilibrium exists as well in which both candidates propose the
legislative outcome. So there is at least some support for the majoritarian
intuition even in this unpromising setting.
In the same spirit, we can sensibly ask the comparative institutional
question posed in the introduction: when will an elective/representative
democracy produce similar policies to a direct democracy? Let us assume,
to begin, a one-dimensional policy space and direct democratic rule with
an open agenda, as in the Athenian ekklesia, where everyone is free to
propose whatever they want. Such an institution would tend to select
policy at the preferred position of the median voter. Assume, now, the
simplest possible representative government: one person is elected who
then is to set policy, and this person is elected in a two-person election
(we leave details aside for now). As argued above, the two candidates
will have reason to promise to implement the preferred policy of the
median voter. In this simple case, therefore, the Downsian intuition is
satisfied. One can imagine extending this argument to a multidistrict
setting if, for example, the districts are ideologically identical to the voting
population so that each representative will advocate policies at the median
position of her constituency. In that case the representatives will all have
preferred policies at the population median. It is easy to see that this
result could be further extended (a bit) by permitting constituencies to
vary as long as the median legislator is located at the population median.
 Election and the aristocratic thesis 293

And, again in somewhat special cases, one could devise extensions to

higher dimensions.
Ideas of this kind have been pursued further in two-stage models,
which take account of how representatives would campaign for legislative
office. When standing for election, a legislative candidate cannot plausibly
claim that she alone would set policy if elected. Rather, she could only
argue that she will play a more or less predictable part in producing
legislation. Such models contemplate that the voters form beliefs as to
the consequences of electing one representative rather than another –
in effect, voters are assumed to have a model of the legislative process
that allows them to assess the policy consequences of electing one or
another candidate – and then vote based on a full assessment of the
policy effects of their vote. One such model was developed by Austen-
Smith and Banks (1988) in which they establish a connection between
the preferred policy of the median voter in the population – which we
could assume would be chosen in a direct democracy – and equilibrium
in the representative democracy game. Their results depend on a number
of other quite restrictive assumptions about the structure of elections and
about the kind of electoral competition that takes place. But their basic
finding is that equilibrium policy is related to the position of the median
voter.
These arguments sidestep problems of political agency which we have
discussed earlier by assuming that representatives will do as they promise
after the election. If that assumption is relaxed it is not clear how con-
strained representatives are by having to compete for election. It is easy to
imagine models in which it is a powerful constraint: a world where a single
elected official wants to stay in office, faces frequent elections, and can
choose among policies contained in a one-dimensional space, for exam-
ple. In that world, if information about the official’s actions is sufficiently
available to voters, and sufficiently easy to interpret, it might be hard for
the official to stray far from the preferred policy of the median voter. But
these informational assumptions are stringent, as is the notion that the
space of policy has such a simple structure. And, as we depart from dras-
tically simplifying assumptions of this model, by permitting numerous
politicians to make decisions, allowing the policy space to be complex
and to map uncertainly into a multidimensional space of outcomes, the
reelection constraint facing the representatives probably diminishes in
its force. Once account is taken of these considerations, the theoretical
perspective adopted here, therefore, seems unlikely to support the idea
that representative and direct democracy will generally produce the same
or similar policy outcomes. So we turn, for now, to examine empirical
evidence that may bear on this issue.
294 John Ferejohn and Frances Rosenbluth

To examine the question of policy convergence, some authors have

asked about the effects of introducing direct democratic mechanisms
– the referendum or the initiative – into representative democracies.
While the resulting literature is somewhat diverse, the basic finding so far
is that the introduction of such devices tends to push policies in the direc-
tion of the median voter (Frey 1994; Gerber 1999; Matsusaka 2004;
Funk and Gathmann 2005). This effect sometimes is direct: a popular
initiative is enacted that moves policy in the direction of the median on
that issue. And sometimes it is indirect in that the legislature in initia-
tive states may tend to produce policies nearer the median, presumably
out of worry that failure to do so would provoke popular policymaking.
Indeed, sometimes one can actually observe such thinking at work where
the legislature enacts a law after seeing initiative petitions being gathered;
or where the legislature puts a referendum on the same ballot as the ini-
tiative in an effort to undercut support for the initiative (Gerber, Lupia,
McCubbins, and Kiewiet 2001).
These phenomena suggest that states without the initiative tend to
produce systematically different policies than states with such an option.
In effect they lend support to the idea that representative democracy
chooses different policies than direct democracy would. Of course, the
comparison is very imperfect as we are forced to make an extrapolation
from those polities permitting popular initiatives to a polity that makes
all its policies in this way. That is a very long extrapolation indeed, and
so evidence of this kind can only be suggestive.
In any case, from a normative viewpoint it is not very reassuring to
learn that our government will tend to do whatever the median voter
happens to want. After all, the median voter is, like the rest of us, a private
person occupied with making a living or raising a family and probably
not very familiar with the world of policy. As Schumpeter worried, her
policy preferences are likely to be influenced by superficial beliefs and
emotional responses rather than being well thought out or consistent.
There is reason to think, both theoretically and empirically, that direct
and indirect democracies tend to choose different policies, except in
the special circumstances we have outlined. This difference seems to be
systematic rather than random: direct democracy, as far as we can see,
tends to pick out median voter outcomes and representative governments
depart from that in some direction or another. And the difference seems
regularly related to how purely representative the government is: partial
systems, which include some institutions of direct democracy, exhibit
policies closer to the position of the median voter. Finally, we do not
have any idea whether these differences are properly traceable to the use
of elections as such or to the specifics of how elections are implemented
 Election and the aristocratic thesis 295

(e.g. the absence of rotation, term limits, instructions, and the presence
of many incumbent-favoring features).
The evidence seems weaker when it comes to the question of whether
the policies of representative government favor the interests of an elite
rather than the general interest. Madison, for example, would have
expected systematic differences of the kind we discuss but thought that,
in the right constitutional setting, these policies may simply have been
better at advancing the common good than the median voter’s preferred
policy. But there is some evidence on this issue and it will be discussed
in the conclusion.

VII. Election or lottery

Bernard Manin reminds us that until the eighteenth century, educated
people thought that lottery was the selection mechanism most compatible
with democracy. The assumption was that officials chosen by lottery
would be more likely to pursue policies that most citizens (who were poor)
would want, whereas elected officials would tend to choose systematically
different policies. We saw in the previous section that recent empirical
evidence suggests that elected leaders probably do choose differently than
the people themselves when asked to choose policies directly. But would
elected officials choose differently than officials selected by lottery? We
do not know a way to examine this issue empirically, so we shall resort
to theory.
Let us start with Aristotle’s political ontology which divides the city
into wealth classes – sometimes he speaks only of the rich and the poor,
but sometimes he refers to a middle class (mesoi) – and assumes that
each class will necessarily seek to pursue its own interest. And, in his
two-class model, assuming that the poor are much more numerous than
the rich, a democratic government will tend to be a government of the
poor. Aristotle had two ideas for moderating this tendency: one was to
impose a property requirement for citizenship that was sufficiently high
that the new democracy would tend to choose moderate policies for the
whole city (including those who were not permitted to be citizens). Such a
strategy could work, of course, only if there were a middle class: otherwise
franchise restrictions would have no effect or else would shift authority
directly from the poor to the rich. The Athenians, who we assume had a
large middle class, attempted to achieve something like this following the
oligarchic coup of 411, when there was an effort to reduce the number
of citizens to 5,000, but this effort collapsed quickly and the democracy
was restored.
296 John Ferejohn and Frances Rosenbluth

Aristotle’s second idea was more robust to social circumstances and

could work, in principle, even if the city was strictly divided between the
rich and the poor. It was to devise institutions that would permit the rich
to take part in ruling the city along with the poor. It is in this context
that he saw election as a device by which the rich could participate in
government, while lottery would tend to select the poor. If some offices
were chosen according to each principle, that would implement a kind of
balance in government. Overall policy would then be a kind of moderate
compromise that was likely, he thought, to approximate, to the common
interest, an interest that neither class would pursue if it monopolized
the government. He recognized, of course, that neither class would be
satisfied and thought that, in his own day, the establishment of democracy
(rule by the poor) was more or less inevitable, partly because there were
so many poor people in every city. But he thought it was partly inevitable
too because of the kind of deep egalitarianism of Greek popular culture
at the time.
Aristotle framed the aristocratic hypothesis by comparing elections
and lottery as a device for picking officials. We can make this comparison
by using citizen candidate models, in which any citizen is free to stand
for election, and so election, like lottery, is simply a device for selecting
leaders. The question posed is this: who (or which kinds of people) will
be selected as a representative out of the population, and which will tend
to be chosen by lottery? Citizen candidate models allow us to examine
the aristocratic hypothesis in a particularly precise manner since every
other feature of the city is held constant in the comparison. Of course this
precision is won within the context of very simple and stylized models.
To begin, it seems plausible to think that lottery will pick a person
whose characteristic is an unbiased estimate of the mean of the population
distribution. And these models also allow us to ask whether, or when,
election by itself will produce officials who might plausibly be regarded as
an elite in some way. We do not offer a definition of “elite” at this point,
leaving the idea to emerge from the specific model (hopefully) intuitively.
In many of the early models, voters had complete information about
each other’s characteristics and each had to decide whether to stand for
election. This can be seen as a very simple kind of agency model in
which voters have no ex post control over the behavior of an official and
can achieve their ends only by selecting agents with the right kind of
(internal) motivation. In the simplest two-period model, plurality rule
is used to elect a single official (Osborne and Slivinski 1996; Besley
and Coate 1997) who, after being elected in the first period, sets policy
in the second. Citizens differ only in their policy preferences in a one-
dimensional policy space and these are commonly known. Any voter is
 Election and the aristocratic thesis 297

free to declare her candidacy in period 1 and declared candidates (voters)

pay an entry cost. Once entry decisions are made, the election takes place
under a plurality rule with ties broken by a fair random device. In the
second (policymaking) period the official will necessarily choose her most
preferred policy. Any first-period promise to do otherwise would not be
credible. The fact that promises are not credible is what permits voters
to form preferences between the declared candidates and to decide how
to vote.
In this setting, depending on entry costs and the benefits of winning
office, only a small number of citizens will declare candidacy (how many
depends on the cost of entry), and each will receive the same number
of votes. If entry costs are high enough, there will be two-candidate
equilibria, and we confine our discussion to those. In a two-candidate
equilibrium, policy positions of the leading vote-getters will be moder-
ate (not “too far” from the median) and will be symmetrically situated
around the position of the median voter (in the sense that the median
voter must not prefer one candidate over the other). The intuition is
that their positions are close enough to the median that no one near the
median would find it worthwhile to enter the election, but sufficiently
separated that no one at either extreme would want to enter either. Many
pairs of candidates can actually satisfy this condition so there are many
such equilibria, and in each one the leading vote-getters must receive the
same number of votes. And the policies of the leaders will be moderate
(not too far from the median) but separated from each other.30
It is a little hard to see how this kind of result could support the aristo-
cratic hypothesis, for two reasons. First, the set of electable candidates is
different from the kind of voter likely to be picked in a lottery. A lottery
pick in this model has an expected value at the mean of the distribution
of the population over the issue space. If we think of the electoral elite as
the set of all voters who could be office-holders in some equilibrium, this
elite will be symmetrically arranged around the position of the median
voter. So if the mean and the median are identical – as they would be in
a symmetric voter distribution – there is a sense in which election and
lottery picks would both be centered on the mean of the population dis-
tribution. Roemer (2001) has shown that the set of electable candidates
is of the same dimensionality as the original issue space, so it is, in this
respect, a large subset of the population, and this set is centered at the

30 The trick in these models is accomplished by the assumption that candidates cannot
commit to a policy different from their ideal point. So candidates can merely decide
whether or not to enter and not which position to offer. This may be sensible in a single
election but is less so when elections take place over time. And it is less sensible where
there is incomplete information.
298 John Ferejohn and Frances Rosenbluth

mean of the population. While it is true that the set of people who could
be picked by lottery is even larger (it includes everyone), the most likely
lottery choices are close to the mean/median in the symmetric setting
and resemble in policy terms those who are electable. In this setting it is
hard to see how the electables can be seen as an elite in any interesting
sense.
There is a natural interpretation of this model that is still more dam-
aging to the aristocracy hypothesis. Assume that voters have identical
preferences over consumption of a single commodity, but differ only in
their initial wealth levels. The collective decision to be made is to set a
uniform wealth tax, with the proceeds to be distributed on an equal per
capita basis, perhaps with some attenuation. In this model policy pref-
erences are wholly determined by initial wealth, with the richer voters
preferring lower tax rates and the poorer ones higher taxes. In a two-
candidate equilibrium, the candidates will be located near the median,
with one a bit richer than the median voter and the other a bit poorer. In
each equilibrium, the expected policy outcome is at the median. More-
over, if we assume that the wealth distribution is skewed in the way that
wealth distributions generally are – so that the median income is less than
the mean – in each equilibrium the candidates will be two relatively poor
citizens (relative to the mean of the distribution). But, in this model, the
lottery pick is centered at the mean and so will tend to be richer than the
citizen elected. Again, it is hard to see policy as manifesting any kind of
elite bias; what bias there is seems to be in another direction. Moreover,
Roemer’s argument shows that the set of people who could be the elite
will be large, as well as relatively poor.
Something like an aristocratic tendency may be observed in a slightly
more complex model of the political economy.31 Suppose a two-period
model in which capital investment is productive and that people have
identical utility functions but different wealth endowments. In the first
period each person has to decide how much to invest, which will yield a
second-period return, and to consume what remains. Assume the govern-
ment has an option to set the second-period tax rate and to redistribute
the proceeds on an equal per capita basis. In the second period, the
investment decisions have been made and the government could tax pro-
ductive returns in a non-distortionary manner. But if it is expected to
set second-period tax rates too high it will discourage investment, lead-
ing to low levels of redistribution. So the government (i.e. the median

31 The model in this section is a simplification of one in Chapter 7 of Persson and Tabellini
(2002). In their more complex settings they reach similar general conclusions.
 Election and the aristocratic thesis 299

voter) would want to commit to optimally low second-period taxes. Let

us assume for illustrative purposes that the median voter has no wealth
endowment so that all of her income comes from redistribution. Then
she needs to assure high-income voters, credibly, that second-period taxes
will not be set too high. One way to do this might be to give the power
to set second-period tax rates to a voter with sufficient wealth to want
to set the rate at the optimal level.32 One might be tempted to say that
in this model an “aristocrat” is given the power to make policy in her
own interest. Of course that is true, but the policy she chooses is the best
policy from the standpoint of the median voter, who is poor. While there
is a kind of aristocracy in this model, the aristocrats are forced by a kind
of invisible hand to do the bidding of the median voter.
In view of these results, it is not clear that anything like an aristocratic
tendency can be generated in simple settings of the kind that have been
studied in these models, but we do not want to belabor results of this
kind that are set in very sparse environments. Citizen-candidate models
are of potential interest because they allow us to focus on how elections
select candidates. Citizen-candidate models suggest that elections gener-
ate something like a weak median voter outcome even if voters lack the
power to monitor and punish their behavior in office. But these models
abstract away the strategic complexity introduced by multiple principals
and multiple agents, which we take to be the core problem of political
agency.
In a somewhat more complex model where information is not complete
about some relevant characteristic which we may call “competence,”
voters would like to choose highly competent candidates but it is hard to
identify them. Given the difficulty of measuring competence, one would
expect that if office is valuable, people with low levels of competence
will try to imitate more highly competent types. And, assuming that
the electoral mechanism can create only very imperfect incentives for
screening candidates (as we have argued), some of these imitators will
be expected to succeed in equilibrium. Moreover, which people find it
worthwhile to become candidates for office depends importantly on their
other opportunities. If good political leaders are those that have skills with
a high market value and poor ones do not, then one expects talented
leaders to avoid running for office and the resulting pool of potential

32 Doing this entails another credibility problem: how can the median voter commit to not
taking back the power to set policy in the second period? One could imagine various
“solutions” but none is really fully satisfying. Perhaps the policymaker could be given the
control of the military. This is a kind of neo-Hobbesian solution but it is still vulnerable
to ex post revolution.
300 John Ferejohn and Frances Rosenbluth

candidates to be mediocre. If that assumption is right, then the election

mechanism, taken as a whole, will be likely to select inferior candidates –
candidates who voters would agree are inferior if they could directly
observe their attributes. Elections will, in short, select the mediocre and
not the aristocratic, if that term is used to refer to high-quality potential
officials.
Of course we are not claiming that such a sweeping and dismal conclu-
sion is actually warranted. Possibly the talented will sometimes choose to
run out of a desire to serve. Or perhaps political skills are not marketable
and competent political leaders have poor outside prospects. In any case
it seems possible that the important qualities can be learned after election
and that, even if elections have a tendency to choose those with meager
inherited gifts, those who take office will tend to form a professional elite
and get better with practice. Thus it seems possible that under some
circumstances elections may select different and better candidates. But
the circumstances in which this may happen seem fairly special.

VIII. Conclusions
The Romans thought that elections could not be trusted to produce or
reproduce the kind of elite that they thought was central to the stable
and conservative rule of the Republic. They fretted mightily, as far as can
be inferred from their practices, over the details of the voting procedures
that were employed in their lawmaking assemblies (comitia). They were
careful to assemble voting classes so that the rich and well-armed had
greater voting weight, to manage tribal membership, and to ensure that
the poor of Rome were “packed and stacked” into a small subset of
tribes. And they tried to ensure that the order and procedure of voting
were such that the elite would get the results they wanted (Staveley 1972).
As far as we know, usually this worked as expected. But there were several
notorious exceptions where laws were enacted over the objections of the
elites.
In modern times, things are arranged less bluntly. But probably they
may still be arranged in ways that permit the reliable reproduction of an
elected elite of some kind. Almost everywhere the representatives get to
choose the details of the electoral system by which they get and hold their
jobs, and to change it if they think it is working unacceptably. They have
nearly universally avoided term limitations and instructions from their
constitutions and usually too the institutions of direct democracy. Almost
everywhere they have succeeded in gaining or keeping working conditions
that make them hard to monitor and control even by the most attentive
 Election and the aristocratic thesis 301

voters and interest groups.33 These are “aristocratic” achievements of a

sort. But are they traceable to elections as such or to many additional
restrictions on the way that election is employed? More importantly, is
there any reason to think that the benefits of their achievements are kept
by political incumbents?
Political office is valuable precisely because of these (slack produc-
ing) achievements, and so one would expect this value to be the object
of intense competition. Competition for office makes political aspirants
dependent on the favor of those who can help them to get it and keep it.
Sometimes this forces dependence on parties or on others who can help
mobilize voters. More recently, however, the dependence seems to run in
the direction of those with disposable wealth. Martin Gilens (2001) and
Larry Bartels (2002) have presented evidence that seems, on the face
of things, to support this view. They argue that representatives tend to
respond disproportionately to those from upper income groups who have
greater access to the tools of political influence, and that policy exhibits
a kind of class bias as a result. To the extent that representatives are
forced to compete away the privileges that they would want to reserve
to themselves, rendering them beholden to those who control the means
to put them in office and keep them there, the elected are a strange and
dependent kind of aristocracy.

BIBLIOGRAPHY
Aristotle. 1962. The Politics of Aristotle, ed. and trans. Ernest Barker. New York:
Galaxy Books.
Austen-Smith, David, and Jeffrey Banks. 1988. “Elections, Coalitions, and Leg-
islative Outcomes.” American Political Science Review 82 (2): 405–22.
Bartels, Larry. 2002. “Economic Inequality and Partisan Politics.” Paper pre-
sented at the Annual Meeting of the American Political Science Association,
September, Boston.
Besley, Timothy. 2006. Principled Agents? The Political Economy of Good Govern-
ment. New York: Oxford University Press.
Besley, Timothy, and Stephen Coate. 1997. “An Economic Model of Represen-
tative Democracy.” Quarterly Journal of Economics 112 (1): 85–114.
Dahl, Robert. 1971. Polyarchy: Participation and Opposition. New Haven, CT:
Yale University Press.

33 For a qualification to this argument see John Ferejohn (1999). That paper argues that
under certain conditions elected officials may voluntarily increase the degree to which
they may be monitored in order to induce people to support a larger public sector. In
effect, this argument, like others in this chapter, cuts against any strong form of the
aristocratic hypothesis.
302 John Ferejohn and Frances Rosenbluth

Disch, Lisa. 2006. “Rethinking Re-presentation.” Paper prepared for delivery at

the Annual Meeting of the American Political Science Association, Septem-
ber, Philadelphia, PA.
Downs, Anthony. 1957. An Economic Theory of Democracy. New York: Harper
and Row.
Dunn, John. 2005. Setting the People Free: The Story of Democracy. London:
Atlantic Books.
Fenno, Richard. 1978. Home Style. Boston: Little, Brown.
Ferejohn, John. 1986. “Incumbent Performance and Electoral Control.” Public
Choice 50: 5–25.
1999. “Accountability and Authority: Toward a Political Theory of Electoral
Accountability,” in Democracy, Accountability, and Representation, ed. Adam
Przeworski, Susan Stokes, and Bernard Manin. New York: Cambridge Uni-
versity Press, 131–53.
Frey, Bruno. 1994. “Direct Democracy: Politico-Economic Lessons from Swiss
Experience. American Economic Review 84 (2): 338–42.
Funk, Patricia, and Christina Gathmann. 2005. “Estimating the Effect of Direct
Democracy on Policy Outcomes: Preferences Matter.” Working paper, Stan-
ford University Department of Economics.
Gerber, Elisabeth. 1999. The Populist Paradox: Interest Group Influence and the
Promise of Direct Legislation. Princeton, NJ: Princeton University Press.
Gerber, Elisabeth, Arthur Lupia, Mathew McCubbins, and Roderick Kiewiet.
2001. Stealing the Initiative: How State Government Responds to Direct Democ-
racy. Upper Saddle River, NJ: Prentice Hall.
Gilbert, Martin. 2007. Churchill: A Life. New York: Henry Holt.
Gilens, Martin. 2001. “Political Ignorance and Collective Policy Preferences.”
American Political Science Review 95 (2): 379–96.
Hamilton, Alexander, John Jay, and James Madison. 1961. The Federalist, ed.
Jacob E. Cooke. Middletown, CT: Wesleyan University Press.
Kant, Immanuel. 1970 [1795]. “Perpetual Peace,” in Kant’s Political Writings,
ed. Han Reiss. New York: Cambridge University Press.
Kishlansky, Mark. 1981. Parliamentary Selection. New York: Cambridge Univer-
sity Press.
Manin, Bernard. 1997. The Principles of Representative Government. New York:
Cambridge University Press.
Matsusaka, John. 2004. For the Many or the Few: The Initiative, Public Policy, and
American Democracy. Chicago: University of Chicago Press.
Michels, Robert. 1915 [1911]. Political Parties: A Sociological Study of the Oli-
garchical Tendencies of Modern Democracy, English trans. Eden Paul and Cedar
Paul. New York: The Free Press.
Montesquieu, Charles de Secondat. 2002. The Spirit of the Laws. Amherst, NY:
Prometheus Books.
Namier, Lewis. 1929. The Structure of Politics at the Ascension of George III.
London: Macmillan.
Osborne, M. J. and A. Slivinski. 1996. “A Model of Political Competition with
Citizen Candidates.” Quarterly Journal of Economics 111 (1): 65–96.
 Election and the aristocratic thesis 303

Persson, Torsten, and Guido Tabellini. 2002. Macroeconomic Policy, Credibility

and Politics. London: Routledge.
Pitkin, Hanna. 1967. The Concept of Representation. Berkeley: University of
California Press.
Plato. 1987. The Republic. London: Penguin Classics.
Rousseau, Jean-Jacques. 1968. The Social Contract. London: Penguin Books.
Schumpeter, Joseph. 1942. Capitalism, Socialism, and Democracy. New York:
Harper Colophon Books.
Staveley, E. S. 1972. Greek and Roman Voting and Elections. Ithaca, NY: Cornell
University Press.
Wood, Gordon. 1991. The Radicalism of the American Revolution. New York:
A. A. Knopf.
11 Why does the Republican Party
win half the votes?

John E. Roemer

I. Introduction
How can a political party whose economic policies are in the interests of
only a small fraction of the richest citizens maintain a sizable vote share
in a democracy with full enfranchisement? The prime example of this
puzzle today occurs in the United States. My aim in this chapter is to
outline the possible answers to the question and then to present in some
detail a study of one of the possibilities. I cannot claim, however, to know
the answer.
One might, first of all, challenge my presumption that the economic
policies advanced by the US Republican Party are indeed only in the
interests of a small fraction of citizens at the top of the wealth distribu-
tion. Indeed, by the standards of one hundred years ago, the US has a
progressive economic policy. About 31 percent of the national product is
collected in taxes (at the federal, state, and local levels), and these taxes
are used predominantly for transfer payments and expenditure on public
goods. There is universal public education through age seventeen, and
many states provide publicly financed tertiary education with modest
private co-payments. Although the United States remains unique in the
degree of private financing of its health services, nevertheless approxi-
mately 50 percent of health expenditures are public. There is a universal
publicly financed pension system which is redistributive.
Yet the Republican Party seems intent on rolling back many of these
practices or institutions, and it has maintained roughly one-half the vote
share in national elections, despite its open attack on them. The Bush
administration passed the most regressive tax rollback in many years,
and has refused to mollify its anti-tax posture despite a large growth in
the fiscal deficit, entailed in part by military expenditures in Iraq and
Afghanistan, but also by virtue of its not matching tax cuts with cuts
in public expenditure. It has tried, unsuccessfully, to privatize social
security, a move which would have increased debt financing even more
(because of the pay-as-you-go nature of social security finance), and also

304
 Why does the Republican Party win half the votes? 305

would have reduced dramatically the redistributive nature of social secu-

rity. It has vociferously supported repeal of the estate tax, a repeal which
would again increase the deficit, and would be only in the clear economic
interests of fewer than 1 percent of American dynasties. These policies
are only the most obvious ones. Somewhat less obviously sectarian is the
Republican Party’s position on global warming and energy policy. Given
current scientific consensus of the effect of industrial production on
global warming, it is almost certainly objectively the case that strong mea-
sures to reduce energy consumption (such as a stiff gasoline tax) are in the
interest of the vast majority of citizens – I say, however, this policy is less
obviously sectarian because it would entail short-run costs for citizens.
I will without further argument assume that Republican Party eco-
nomic policy is only in the interest of a small fraction of the wealthiest
citizens, and proceed to a discussion of how the party is able to main-
tain its large vote share. I believe there are six categories of possible
explanation:
r policy bundling;
r imperfect political representation;
r voters’ conceptions of distributive justice;
r voter misunderstanding of “how the economy works,” that is, of the
mapping from policies to outcomes;
r voters’ incorrect beliefs about the probability of becoming rich;
r manipulative policy framing.
Let me give examples of these phenomena. By policy bundling, I mean
that the Republican Party maintains its large vote share by bundling its
unpopular economic policy with other policies that do attract voters. In
the period after the Civil Rights Movement until the 1990s these were
policies on race issues, and in the last decade they have been policies of
patriotism and “family values”: prayer in schools, and opposition to gay
marriage and abortion.
By imperfect representation I mean the fact, if it is one, that legisla-
tors do not represent constituent interests in proportion to their num-
bers, but are unduly influenced by their wealthy constituents, or other
wealthy institutions, particularly corporations. Among the six categories
of explanation, this is the one that is most easily conceived to be a failure
of democracy.
The third explanation is, of course, related to the view that voters
behave sociotropically – they vote not primarily according to self-interest,
but according to what they think will be good for society – a behavior
which is not identical to, but similar to, voting according to justice.
Perhaps the most prominent example of the fourth category is the
relationship between taxation and “effort,” by which I mean the effort of
306 John E. Roemer

economic agents to become educated, to work hard, to innovate, or to

invest. Republicans argue that effort elasticities with respect to taxation
are high, while Democrats tend to argue that they are low. Thus, “supply
side economics” is the view that lowering taxes will not only increase
the size of GNP, but will also increase government revenue. The most
prominent expositors of this view were Ronald Reagan and his economist
Arthur Laffer. (Lee and Roemer [2006] recently calculated that the max-
imum of the Laffer curve for the US economy in the Reagan period was
over 50 percent – meaning that income tax revenues would have risen at
least to the point of 50 percent marginal tax rates, contrary to Laffer’s
claim.)
The fifth category of explanation really requires two suppositions: that
voters believe that the probability of their or their children becoming
rich is much greater than it actually is, and that voters are sufficiently risk
loving that they are willing to take the gamble – keeping redistribution at a
low level to their present (relatively poor) selves, against the possible gain
of being lightly taxed as their possibly rich future selves (or dynasties).
Clearly, America as the “land of opportunity” is an important motivating
folk tale for this explanation.
By the sixth explanation I mean that regressive economic policies are
framed as being consonant with desirable personal characteristics – indi-
vidualism, self-reliance, and so on.
My own current view is that the power of these explanations is roughly
ordered in the way I have listed them: I believe the first three explanations
are the most important. I cannot say, however, that I have proof. I have
done some work on policy bundling, and some on imperfect representa-
tion (Roemer 2006); I will describe the former below. Otherwise, I am
relying on a view about political fundamentals that I summarize in the
next section.
There is a cognitive problem for the social scientist. One chooses to
work on one of these explanations – say, policy bundling – because one
has a prior belief that it is important. If one gets positive results, that
view is reinforced: one updates the prior belief in a positive direction.
One can, however, find researchers who advocate the primacy of each of
these explanations. I do not feel that political science currently has a full
grasp of the problem.

II. What are the fundamentals?

My own view is that we should be guided by looking at the fundamen-
tals. Just as economists seek to explain the values of economic variables
over the long term by the fundamentals of preferences, technology, and
 Why does the Republican Party win half the votes? 307

distribution of endowments, so political scientists should be guided by

political fundamentals – preferences of voters on policy issues and the
nature of democratic institutions that mediate between voter prefer-
ences and political outcomes. My assumption is that the phenomenon of
Republican Party strength in the United States is not a flash in the pan;
it is a long-term phenomenon, and hence we should search for its funda-
mental determinants. I am therefore not attracted to the sixth explana-
tion, offered above, that the Republicans have succeeded in manipulative
policy framing. Over the long period, they should be no better at this than
Democrats, and so even if such manipulation is politically important, it
cannot be the kind of fundamental explanation that interests me. I must
underscore the point, however, that this is a methodological presump-
tion on my part: perhaps I have too short a view of what constitutes the
long run, and one hundred years from now Republican power will, in
retrospect, be seen to have been a flash in the pan. (See Smith [2007]
for a serious piece of analysis that places more importance than I do on
manipulative policy framing.)
On the other hand, I do not have much sympathy for the view that the
only true fundamentals are the economic interests of voters. Marxists of
yore would say that a voter who is swayed by, for example, religion or
the church in his political choices, is a victim of false consciousness –
the implication being that only political behavior which advances one’s
economic interests is true consciousness. I will take the preferences of
voters as fundamentals – not their true economic interests. I am not
willing to take economic interests as more fundamental than other kinds
of values and preferences that people possess. Whether they are more
fundamental in determining political outcomes is an empirical question;
it should not be postulated. Indeed there may well be a presumption
against the necessary primacy of self-centered economic interests, espe-
cially as economic development makes survival a less pressing concern.
Non-economic preferences may be like a luxury good, becoming more
salient as income rises.
I place less importance on voters’ misunderstanding of the mapping
from policies to outcomes than I used to (see Roemer 1994), and for
the same reason that I place relatively little importance on framing. It
is certainly in the interests of the wealthy, and of the Republican Party,
if it is the primary representative of that stratum, to convince voters
that effort elasticities are high – but if the Democratic Party represents
the poorer half of society, it should be just as convincing in arguing
that effort elasticities are low. In addition, we might expect that in the
long run the more convincing arguments on this question can be given
by the side that is right, and Lindert’s important book (2004) gives
308 John E. Roemer

convincing historical evidence that the low-elasticity side is right. Lindert

argues that the high-tax welfare states have paid essentially no price in
terms of lower productivity for their policies. (These countries have lower
GDPs per capita than the US because of the labor–leisure choice their
citizenries have made, not because of lower productivity. Surely choosing
more leisure is a possible response to high taxation. But we do not hear
conservatives arguing to the American polity that they should vote for low
taxes because otherwise they will want to spend more time on vacation
with their families.)
One of the most persuasive studies that focuses upon the mapping from
policies to outcomes is Hirschman (1991), who writes that reactionaries
have exploited three main rhetorical strategies: to argue that progressive
reforms are either perverse (they will have consequences which are the
opposite of what is intended), futile, or will put into jeopardy other more
valuable social characteristics. But again one must ask: if the claims
of perversity, futility, and jeopardy are false, which Hirschman himself
believes, then why should progressives not have the edge in the argument?
How does one explain the fact that in the United States today each side
has the support of about half of the polity?
Another proposal of recent vintage is due to Piketty (1995). In his
model, voters are uncertain about the importance of effort versus luck
in producing income, where income is a random variable. They update
their beliefs about the true importance of effort based upon their own
experiences. Eventually, a society will converge to having a distribution
of beliefs on this question. Thus people come to have either a “right-
wing” view that effort is important for producing income and luck is
relatively unimportant, or a “left-wing” view that luck is more important
than effort. Based upon these views, voters advocate little or much redis-
tribution. Importantly, two societies with the same distribution of initial
beliefs can end up with very different political equilibria.
I have suggested, conditionally, and surely controversially, that the
Democratic Party represents the poorer half of society. Now in a represen-
tative democracy, it would: that is, in a representative two-party system
the natural outcome is that the polity partitions into two elements, each
of which is represented by one party, and so if the Republicans represent
the wealthy, then the Democrats should represent the rest.1 One of the
main ways in which imperfect representation could be an explanation of
the question of my title is that the Democrats fail to do this. I believe that

1 I take it as an axiom of (good) representation that every citizen is represented by at

least one political party, and that citizens of different types are represented by parties in
proportion to their numbers (whatever that turns out to mean).
 Why does the Republican Party win half the votes? 309

this is a substantial possibility, although I will not address it seriously in

this chapter.
I did not list as one of the possible explanations that Republicans have
offered candidates who are or seem to be more competent or personally
attractive than Democrats. There is a line of research in contemporary
formal political science which focuses upon voter perceptions of candi-
date competence or attractiveness, or “valence.” I am unmoved by this
work: I grant such questions may be important in particular elections,
but they are not fundamental, and hence not of concern to someone who
is interested in long-term political trends. Over the long run, no party’s
candidates should have a monopoly on these randomly distributed char-
acteristics.
Against my view, you could take a “quantum-mechanical” stance,
or perhaps I should say a “catastrophe-theoretic” or “chaos-theoretic”
stance. To wit, you might say: the outcomes of some particular elections,
like that of the US President, may be determined by non-fundamentals
(like the butterfly ballot in some Florida precincts, or George W. Bush’s
homey veneer), and these may set the political dynamic on a path quite
different from what it otherwise would have been, thus engendering a
very different future than otherwise would have occurred. I cannot say
that I understand the laws of motion of political economy sufficiently
well to argue that this is wrong – only that I take a methodologically
conservative stance, which is, if you will, “Newtonian.” The catastrophe
theory view, indeed, challenges the whole idea of fundamentals. If we
come to believe it is true, we will have to give up much of what we think
of as political science and become more like anthropologists.
What about false beliefs? Can we attribute the quite significant support
for repeal of the estate tax to voters’ false beliefs in the probability of
becoming rich, and their high degrees of risk tolerance? (Perhaps the loss
of revenue from repealing the estate tax is sufficiently small that voting for
its repeal is like going to a gambling casino – people understand that the
odds are against them, but the expected losses are sufficiently small that
people behave in a risk-loving manner.) Behavioral economics certainly
teaches us to treat cognitive errors (in this case, misunderstanding the
probabilities of becoming rich) seriously. But in this case I am more
persuaded by the view that the stakes are small (from the viewpoint of
government revenue) and that, because of this, a particularly dedicated
group of “death tax” opponents has succeeded in framing repeal in an
attractive way. (See Graetz and Shapiro [2005] for a detailed study.)
Before moving on, I must mention one non-fundamental explana-
tion for Republican ideological power that I find quite persuasive, but
that I did not list: that American conservatives have been prescient in
310 John E. Roemer

establishing conservative think-tanks that have succeeded, by patient and

careful work, in disseminating conservative and reactionary ideas among
the public. A persuasive advocate of this view is Mark A. Smith, who has
studied the possibility in two books (Smith 2000, 2007). Conservative
American foundations, particularly the Olin Foundation, have funded
think-tanks (the Manhattan Institute, Hudson Institute, Heritage Foun-
dation, Cato Institute, American Enterprise Institute, Hoover Institute)
which hired intellectuals to produce and disseminate broadly conser-
vative and libertarian ideas. While social science in the academy has
been predominantly liberal, professors have made little effort to reach
the public as the think-tanks have. The atrophy of American unions,
which were formerly to some extent schools for solidaristic ideology, has
left a vacuum which conservative intellectuals have filled via the think-
tank conduit. Ironically, perhaps, the right understood the importance
of ideas and theory, while the left did not. I say this explanation is of a
non-fundamental sort because there seems to be no reason why the right
should have a long-term advantage in employing this strategy. But it may
be an important explanation of the present American political landscape.

III. Policy bundling

The research I describe in this section has been published in Lee and
Roemer (2006), and is described in even more detail in Roemer, Lee,
and Van der Straeten (2007), so I limit myself here to a non-technical
summary.
We study the extent to which voters’ views on the race issue in the
United States influence the politically determined equilibrium values on
the tax issue. We analyze presidential elections over the period 1976–
92; we postulate a spatial model to describe these elections, in which
the policy space is two-dimensional – one dimension concerns taxa-
tion and redistribution, the second concerns the race issue. By the race
issue, we mean policies specifically concerning the race question: school
integration, busing, affirmative action, racial justice issues, and the like.
Denote the space of policies by T, which consists of ordered pairs (t, r)
where t is a party’s position on the tax issue, and r is the party’s position
on the race issue. Voters are assumed to have preferences over these two
issues, which are represented by utility functions with a parametric form
to be given below. Each voter is of a type specified by two numbers – his
ideal policy on the race issue (ρ) and his annual wage rate (w). Thus the
space of voter types is a space H of ordered pairs (w, ρ) endowed with a
probability measure F.
 Why does the Republican Party win half the votes? 311

The utility function of a voter of type (w, ρ) over policies (t, r) is

postulated to be of the form:
log y(t, 0.20)
v(t, r ; w, ρ) = ϕ(t, w) − α(r − ρ)2 + (δ0 − δ2 ρ) . (1)
log y(t, 0.80)
The first term, ϕ(t, w), is the voter–worker’s self-interested economic
utility, gotten from maximizing a traditional Cobb–Douglas utility func-
tion over labor and income, given by log((1 − t)wL + b(t)) + β log(λ −
L), where (1 − t)wL + b(t) will be the worker’s after-tax income under
an affine tax system with marginal rate t and lump-sum payment b(t),
should the worker choose to work L fraction of year, and (λ − L) is the
worker’s leisure time. The second term, −α(r − ρ)2 , is the voter’s utility
from the race policy r, which is assumed to be a Euclidean loss increasing
as the policy departs from his ideal policy, ρ. The number y(t, π ) is the
after-tax income of the worker whose wage is at the π th quantile of the
wage distribution, if the tax policy is t. Thus the last term is a measure
of the degree of equality in the distribution of after-tax income, for as
t increases, redistribution increases, and this term increases. Thus it is
assumed that all voters care about equality to the extent that δ0 is positive,
but that more racist voters (who have a larger value of ρ) care less about
equality than less racist voters – to the extent that the parameter δ2 is
positive. The parameters of the utility function that are assumed to be
the same across all voters are thus (β, λ, α, δ0 , δ2 ) These parameters will
be estimated from the data.
Two comments are in order. In actual fact, we modeled households
with two adults (man and woman), and we made the utility of each
member a function of family income and the leisure of each member;
we estimated male and female parameters separately. We also allowed
the two household members to have different preferences on the race
issue. I am eclipsing these complications here for the sake of expositional
simplicity. Second, we originally postulated that the coefficient in the
last term was of the form (δ0 − δ1 w − δ2 ρ), allowing for the possibility
that higher-wage workers would care less about equality than lower-wage
ones: but we found this was not the case, and so we set the parameter
δ1 = 0.
Thus political parties in presidential elections are assumed to take posi-
tions on the tax question and the race question, and voters are assumed
to choose between parties based upon their preferences on these issues.
Before describing the concept of political equilibrium, I describe our
conceptualization of the ways in which preferences on the race issue will
influence parties’ positions on the tax issue. There are two of these effects,
which we denote the anti-solidarity and policy bundle effects.
312 John E. Roemer

The anti-solidarity effect is a direct one: voters who believe that

minorities – and blacks in particular – exploit the welfare state may
vote for lower tax rates (than otherwise) because they wish to reduce the
degree of redistribution to the undeserving. Many writers have noticed
this effect: we mention Luttmer (2001), and Alesina, Glaeser, and Sac-
erdote (2001). Indeed, Alesina et al. perform cross-country regressions
in which they show that the smaller the size of the welfare state, the larger
the size of the poor minority in the country. This is taken to verify, or to
be consistent with, what we call the anti-solidarity effect.
The policy bundle effect is indirect and more subtle. Suppose there
is a sizable fraction of relatively poor, white voters, who are racist. On
grounds of economic self-interest, they should vote for high tax rates and
redistribution. But if the race issue is of sufficient importance to them,
they would vote for a party that proposes a low tax rate, if that party is
sufficiently conservative on the race issue (i.e. proposes a large r). This
would be the case if α and ρ were sufficiently large. Thus, if the mass
of voter types so described (small w, large ρ) is large, a party, say the
Republicans, could maintain a large vote share by proposing a racist,
low-tax platform.
Of course, whether the group of voters just described is larger than the
group of high-wage, racially liberal voters is an empirical question. The
policy bundle effect can also make it possible for Democrats to attract
such voters, while maintaining a high-tax policy, which such voters might
dislike.
The so-called Southern strategy of the Republican Party since the
Goldwater years is, indeed, consistent with the view that the Republicans
have exploited the policy bundle effect. That strategy has been to appeal
to racism and law-and-order preferences – and to the extent that such
appeals attract votes, the Republican Party can be less compromising on
the issue of redistribution.
Our equilibrium model will enable us to estimate the anti-solidarity
and policy bundle effects on tax policy over the period in question in the
US.
I next describe our concept of political equilibrium. The Downsian
model, and its famous median-voter theorem, is of no use in this situation,
because that model only possesses equilibria (essentially) when the policy
space is unidimensional. Our policy space is two-dimensional: indeed,
that is the essence of our problem – the policy bundle effect could not
exist if taxation were the only issue (although the anti-solidarity effect still
could). We use a concept of equilibrium that I have called party unanimity
Nash equilibrium (PUNE); these equilibria exist even when policy spaces
are multidimensional.
 Why does the Republican Party win half the votes? 313

The critical way in which PUNE differs from other strategies to deal
with multidimensional political competition is that, rather than com-
plexifying the Nash concept of competition, it complexifies the concept
of what a player is. Formally, a player in a game is a payoff function
defined on the cross-product of the players’ strategy spaces. We propose,
in contrast, that a political party is not a single payoff function, but rather
a set of three payoff functions, held by different factions among the party’s
entrepreneurs. To wit, the entrepreneurs in a party have different goals –
some are Downsian, in desiring to propose a policy that will maximize the
party’s probability of victory, or in another formulation, its vote share;
some are Militant, in desiring not to compromise, but to propose a pol-
icy as close as possible to the ideal policy of the party’s constituency;
and some are Reformist, in desiring to maximize the expected utility, on
average, of the party’s constituency, with the expectation taken over the
two states in which the party, or its opponent, wins. We propose that
these three factions within a party bargain with each other, while facing
the proposal of the opponent party. A PUNE is a pair of policies each of
which is a bargaining outcome in one party, given the policy of the other
party. Thus, the Nash notion of best response equilibrium is preserved,
but a player’s best response is not the maximizer of its (single) payoff
function, given the other player’s strategy, but rather the consequence of
bargaining among three payoff functions, given the other player’s strategy.
This equilibrium concept was first used in Roemer (1998, 1999).
Indeed, we need not be terribly sophisticated about the nature of bar-
gaining: we employ only the assumption that bargainers reach a Pareto-
efficient solution at the bargaining table – that is, that in their response
to the opposition party, there should be no policy that increases (weakly)
the payoffs of all three factions over the policy that they propose.
The fortunate fact is that, even with multidimensional policy spaces,
equilibria of this kind typically exist. Indeed, it turns out that, from
the mathematical viewpoint, the Reformists are gratuitous! Thus the
set of equilibria is not changed if we conceptualize parties as consisting
only of Opportunists and Militants. Which equilibrium in this set will in
fact be observed depends upon the relative strength of the factions in
each party at the bargaining table. We do not attempt to estimate these
relative strengths: we will simply take a party’s position to be its average
platform over the entire set of PUNEs. We justify this approach in the
cited publications.
In summary, I make four points. First, we view the bargaining
described as taking place among professional party entrepreneurs – those
who set policies – not among voters. Parties are viewed as organizations
run by politicians, which in an imperfect way represent constituencies.
314 John E. Roemer

Do not think of these entrepreneurs as being voters of a particular type.

They are professionals who are adopting different protocols to further
their careers – one protocol is to maximize the probability of winning,
another is to appeal to constituents. Among these entrepreneurs, the
Reformists might be viewed as the most selfless ones, who are looking
out for the good of the constituency. But that, too, could just be a good
protocol for living a long political life.
The second point is that the description of party factions as bargaining
with each other is a formalization of something which may be much
more amorphous, namely, that among a party’s entrepreneurs, there are
competing tendencies – maximizing victory probability and representing
constituents. I do not assume there are literal negotiations taking place.
The idea that is captured by the bargaining metaphor is that the aims
of representing constituencies, and of winning elections, are often very
different, and it is more appealing to represent these aims as a competition
between payoff functions rather than as the maximization of a single payoff
function.
The third point is that the memberships or constituencies (these two
words are synonyms in the model) of the parties are endogenous. A
PUNE is a partition of the polity into two elements each of which is rep-
resented by a party in the sense described, and a pair of platforms, one
for each party, such that the entrepreneurs of the parties (at least the
Militants and Reformists) represent their constituencies in the manner
described, and each member of each constituency prefers its party’s pol-
icy to the opposition’s. Thus, in equilibrium, every voter is satisfied (at
least to the first order) with her party’s policy, in the sense that she will
vote for it against the opposition party.2
Fourth, in modeling the average utility function of a party’s con-
stituency, we do not take a simple average of the voter types assigned to
the party, weighted by the probability distribution F. Rather, we attempt
to capture the apparent fact that politicians represent more faithfully the
preferences of the richer voters in their constituencies. Bartels (2002) has
estimated this phenomenon, and we have devised a system for weight-
ing voters’ representation in the payoff function of the Militants using
his data. Indeed, we found we got better estimates of observed policies
by doing this than by taking a simple unbiased average of voter types
associated with the parties.

2 Where does “probability of victory” come in? We assume that there is some macropolitical
uncertainty in the elections, either because of party uncertainty over voter preferences,
or voter misperception of policies, or shocks that may occur between the statement of
party policies and elections.
 Why does the Republican Party win half the votes? 315

I next describe the estimation of the data of the model. We used two
data sets, the Panel Study on Income Dynamics (PSID) and the Amer-
ican National Election Studies (ANES). The PSID is used to estimate
the economic parameters β and λ. The ANES is used to estimate, in
each election year, the joint distribution of types, F: that is, of wages
and racial views. We also used the ANES to estimate the positions of
presidential candidates on the issues in each election year, based on what
voters’ perceptions of these positions were. We then chose the remaining
parameters α, δ 0 , and δ 2 to give a good fit of the model to the data – that
is, so that the calculated PUNEs are close to the observed party positions
in those elections.
I will not summarize the details of this estimation, except to describe
how we estimated voters’ positions on the race issue, the parameter ρ.
We used, as a first-order approximation, the voter’s answer to the survey
question: “Do you think the government gives [too little, about right,
too much] aid to blacks?” The respondent gives an answer on a seven-
point scale from 1 (far too little help) to 7 (far too much help). To a first
approximation, we identify these answers as going from anti-racist (1) to
racist (7). However, this identification is problematical, for a voter might
answer “7” not because she is racist but because she is libertarian, and
believes the government gives too much help to everyone. So we needed
to sterilize the answer to the “aid to blacks” question of its libertarian
factor.
To do so we ran a factor analysis on a set of ten questions from the
ANES: we located four orthogonal factors of political ideology, which we
named racism, libertarianism, compassion for the poor, and feminism.
(The names come from the nature of the questions upon which these four
factors load highly.) We were then able to extract the libertarian compo-
nent from the “aid to blacks” question: it turned out that that component
was rather small. Indeed, “racism” is about three times as important a
determinant of the answer to “aid to blacks” as libertarianism, a fact of
independent interest. Thus we constructed a measure of racism as the
“racism-induced response to the aid to blacks” question. This gave us
the value of ρ for the individual.
Next, we calculated the equilibria of the model for various choices of
the parameters that are not directly estimated, the vector (α, δ0 , δ2 ). For
each election year, we chose values for this vector that give us a good
estimate of the observed platforms for that year. Figure 11.1 shows the
equilibria for two of the election years. The plane of the figure is (t, r).
Each PUNE consists of a pair of policies, one small black point (Repub-
lican policy) and one small gray point (Democratic policy). The large
dots are the observed policies for that year (actually, as you observe, we
316 John E. Roemer

1976–80 1984–88

Figure 11.1 PUNEs and observed policies in two pooled election years
(dark points are Democratic, light points are Republican).

pooled data for pairs of consecutive election years). The model appears
to do quite well in capturing the observed policies. Denote the average
PUNE policies in this election for the Democratic and Republican Parties
(t D , r D ) and (t R, r R).
Before proceeding to the next step, examine the utility function in eqn.
(1), and note that the critical parameter for the anti-solidarity effect is δ2 :
it is this parameter, along with the individual’s value of ρ, that determines
the extent to which racism reduces the desire for redistribution. In like
manner, it is the value of the parameter α, along with ρ, that determines
the policy bundle effect.
We now move to the heart of our project, the estimation of the anti-
solidarity and policy bundle effects. We run a counterfactual election in
which we set the race policies of both parties to be identical – it does not
matter what policy value we choose. This will be an election on just one
dimension: tax policy. In this election, voters will have no reason to vote
for the Republican Party because they prefer its race policy: so there will
be no policy bundle effect in this election. Equivalently, we could set α
equal to zero, because the race policies of the parties will have no effect
in this election on voter choice. However, the anti-solidarity effect will
still be active: a voter who thinks blacks exploit the welfare state will still
have reason to vote for lower tax rates. Denote the average tax policies3
in the equilibria of this election t ID and t IR. We consequently estimate the

3 In point of fact, we used a method of averaging which weighted equilibria according to

their expected frequencies; the reader is referred to the cited paper for clarification.
 Why does the Republican Party win half the votes? 317

policy bundle effects on the D and R party policies as the differences

t ID − t D and t IR − t R.
We next ran a second counterfactual election: this time, we not only set
the racial policies of both parties to be the same, but we set δ 2 = 0, thus
annihilating the anti-solidarity effect. Denote the average equilibrium
(tax) policies in these elections by t IDI and t IRI . These policies are purged
of both the policy bundle and anti-solidarity effects. We thus estimate the
total effect of racism on tax policy for the two parties, as t IDI − t D and tIRI −
t R. In addition, the equilibria come with vote shares for the two parties, so
we estimate the expected effects as the vote-weighted averages of the effects
for the two parties. This is an ad hoc way of modeling the thought that
implemented policy will be some average of the platforms of the parties,
an average reflecting the vote shares of the parties due to legislative
implementation.
Table 11.1 summarizes the results of these computations: as I men-
tioned, we pooled data for adjacent election years, so we performed
four experiments. We estimated that the expected income tax rate would
have been between 11 percentage points (in the 1980–4 experiment) and
18 percentage points (in the 1976–80 experiment) higher, had racism
not been a political factor. The decomposition of this effect between
anti-solidarity (ASE) and policy bundle (PBE) varies between the two
parties and across the years: we may say that the PBE is significant, so
the Southern strategy appears to have been a good one for a Republican
Party whose central focus was maintaining low tax rates. Indeed, if our
estimates are to be credited, then racism would appear to explain the dif-
ference between the size of the public sector in the US and the advanced
European democracies.
I conclude this section with a discussion of Figure 11.2. The three
panels in this figure present the estimated density function of voter types:
thus the plane of the domain in these figures is the type space (w, ρ).
The two “mountains” of the density function represent white and black
voters: all black voters are coded as non-racist (a value ρ = 1). In the “full
model” panel, the “hyperplane” illustrated separates the space of types
into voters who form the constituencies of the R (above the hyperplane)
and D (below the hyperplane) parties. Now examine panel (b): here, the
hyperplane again partitions the type space into R and D voters, but in the
first counterfactual election. Panel (c) presents the partition into party
members in the second counterfactual election. Of course, in the sec-
ond counterfactual election, the racial issue has entirely disappeared, so
the only relevant dimension of a person’s type is his wage: therefore, we
have in panel (c) purely “class politics”: that is, a party partition which
is completely characterized by a pivotal wage rate. Now the interesting
Table 11.1 PUNEs and the decomposition of racism’s effect (δ0 = 1)

1976–80 Full First counterfactual Second counterfactual Total effect PBE ASE PBE (%) ASE (%)

tD 0.3473 0.3791 0.4824 0.1351 0.0318 0.1033 23.54% 76.46%

tR 0.2212 0.3432 0.4450 0.2238 0.1220 0.1018 54.51% 45.49%
rD 2.7663
rR 4.1144
Exp tax rate 0.2927 0.3696 0.4742 0.1815 0.0769 0.1046 42.36% 57.64%
D vote share 0.5166 0.7351 0.7814 0.2648 0.2185 0.0463 82.52% 17.48%
# of PUNEs 9 45 26
1980–4 Full First counterfactual Second counterfactual Total effect PBE ASE PBE (%) ASE (%)
tD 0.4025 0.4137 0.4666 0.0641 0.0112 0.0529 17.47% 82.53%
tR 0.2129 0.3638 0.4391 0.2262 0.1509 0.0753 66.71% 33.29%
rD 3.4307
rR 3.7914
Exp tax rate 0.3465 0.4011 0.4567 0.1102 0.0546 0.0557 49.48% 50.52%
D vote share 0.5609 0.7466 0.6417 0.0808 0.1857 −0.1049 229.83% −129.83%
# of PUNEs 11 42 19
1984–8 Full First counterfactual Second counterfactual Total effect PBE ASE PBE (%) ASE (%)
tD 0.3709 0.3859 0.4993 0.1284 0.0150 0.1134 11.68% 88.32%
tR 0.2392 0.3234 0.4042 0.1650 0.0842 0.0808 51.03% 48.97%
rD 2.7771
rR 3.6483
Exp tax rate 0.3109 0.3659 0.4699 0.1590 0.0550 0.1040 34.61% 65.39%
D vote share 0.4049 0.6804 0.7906 0.3857 0.2755 0.1102 71.43% 28.57%
# of PUNEs 23 52 15
1988–92 Full First counterfactual Second counterfactual Total effect PBE ASE PBE (%) ASE (%)
tD 0.3154 0.3320 0.4409 0.1255 0.0166 0.1089 13.23% 86.77%
tR 0.1504 0.3004 0.4030 0.2526 0.1500 0.1026 59.38% 40.62%
rD 2.8738
rR 4.1953
Exp tax rate 0.2870 0.3241 0.4270 0.1400 0.0371 0.1028 26.53% 73.47%
D vote share 0.5797 0.7508 0.6320 0.0523 0.1711 −0.1188 327.15% −227.15%
# of PUNEs 15 36 11
320 John E. Roemer

(a) Full model

8
r 6
4
2
0

0.08

0.02

0.01

0
0
10
20
w 30
40

(b) Model with r = r (c) Model with r = r and ρ = ρ min

8
r 6 r 6
4 4
4 2
0 0

0.03 0.08

0.02 0.02

0.01 0.01

0 0
0 0
10 10
20 20
w 30 w 30
40 40

Figure 11.2 (a) Full model

(b) Model with r = r̄
(c) Model with r = r¯ and ρ = ρmin
Party-separation “hyperplanes” for the full model and the two counter-
factual elections. The space of the domain is (w, ρ).

observation is that in passing from the full model to the first counterfac-
tual, we move from a situation in which the racial issue is very important,
to one in which politics is very largely class politics: that is, in panel (b),
the separating hyperplane appears to be almost perpendicular to the w
axis. It is important to underscore the fact that preferences of voters have
not changed in the move from panel (a) to panel (b): all that has changed
is that we have excluded the race issue as something on which parties may
take different positions. The ASE is still active in the political compe-
tition of panel (b). Thus, these figures suggest that the existence of the
 Why does the Republican Party win half the votes? 321

race issue in American politics during this period altered what would
have otherwise been predominantly class politics to politics in which race
issues are very important.
We note that this inference is quite different from the one made by
Poole and Rosenthal (1997), who argue, on the basis of congressional
vote analysis, that the race issue was relatively unimportant in this period
compared to the economic issue. One crucial difference between their
analysis and ours is that ours uses an equilibrium model, in which party
constituencies, inter alia, are endogenous. Thus we are able to simulate
how party composition would have changed, were race to have not been an
issue in American politics. This is the drama described in Figure 11.2.
I do not wish to urge overdue credence in the magnitudes of our results.
If racism has indeed lowered the tax take of the state by between 11 and
18 percentage points, I would be surprised. There are many criticisms
one can lodge against our model – and I will mention several in a moment.
But I do believe that we have made a serious effort at estimating the effect
in question; the ways that I can think of to improve our model would be
quite computationally difficult and time-consuming. Therefore, I believe
that we have shown that the race issue significantly reduces redistribution
in the US from what it otherwise would be.
Let me explain the limitations of the model just referred to. Typical
equilibrium analysis in political economy can be characterized as “one
by one” (i.e. 1 × 1); it assumes a policy space of one dimension and a
type space of one dimension. The policy is a tax rate, the type dimension
is “income.” We have employed a 2 × 2 model: both the policy space
and the type space are two-dimensional. This necessitated not only a
new concept of political equilibrium (moving from Hotelling–Downs
to PUNE), but computationally expensive procedures of calculation.
(The equilibrium computations described in this paper took about 500
computer hours.) Now it would be very nice to be able to increase the
dimension of both the policy space and the type space. For instance,
we might like to allow for non-linear taxation, or for foreign policy, or
for preferences over “values” issues – each of these moves would require
expanding the dimensions of both type and policy spaces. We might also
like to allow voters to possess different values of α – that is, of racial issue
salience – in the utility function. This, too, would require expanding the
type space to three dimensions. But adding even one extra dimension to
the type space would make our calculations of equilibrium an order of
magnitude (I am guessing) more time-consuming. My conjecture is that,
were we able to do this (or, more optimistically, when we will be able to),
our estimates of the ASE and PBE will change, and the estimated effect
of racism on equilibrium values of tax policies will probably decrease.
322 John E. Roemer

IV. Concluding remarks

I have argued that during the period 1976–92 the Republican Party
was able to maintain a significant vote share, without compromising
very much its position on redistribution and size of the public sector,
by exploiting the race issue. There is reason to believe that the main
Republican strategy has shifted in the last decade to relying on issues such
as abortion, patriotism, and anti-gay preferences. For some evidence, see
Lee and Roemer (2008). Whether this move will be as effective as the
race issue has been is, of course, an open question.
I do not wish to claim that Hirschman’s observations about the reliance
of reactionaries on the arguments of perversity, futility, and jeopardy are
out of date: I believe these strategies continue to be important ones,
whereby parties that represent the very wealthy can continue to remain
viable in a fully enfranchised democracy. Indeed, the American author
who has best exploited these strategies in recent years, and to great
effect, is Charles Murray (1994). As I said, these arguments fall under
the explanation of “the mapping from policies to outcomes,” and perhaps
I am wrong to place that explanation at a relatively low position (fourth)
on my list of explanations.
I have not discussed, very much, the issue of misrepresentation. As I
said, we used Bartels’s estimates to model the idea that politicians are
more responsive to rich voters than to poor ones. But there are many other
facets of misrepresentation that may be important, having to do with the
role of money in politics (both campaign contributions and lobbying)
and with the rules of representation under different democratic systems.
I believe I would disagree with those who would argue that the policy
bundling phenomenon I have focused upon here, and its apparent effect
of keeping extremely conservative economic policy politically viable in
a democracy, is an aspect of misrepresentation. Why? Because policy
bundling may be an unavoidable characteristic of any viable democracy.
The only obvious way of avoiding it is to run referenda on every single
issue facing the electorate: but if that is infeasible, or undesirable for
other reasons (as I think it is), then we cannot say that electoral systems
that bundle policies entail misrepresentation. After all, ought implies can.
What we may be able to conclude, however, is that democracy is not the
panacea that many appear to think it is.

BIBLIOGRAPHY
Alesina, A., E. Glaeser, and B. Sacerdote, 2001. “Why Doesn’t the US Have a
European-Style Welfare State?” Brookings Papers on Economic Activity (Fall):
187–278.
 Why does the Republican Party win half the votes? 323

Bartels, L. 2002. “Economic Inequality and Political Representation.” Princeton

University.
Graetz, M. and I. Shapiro. 2005. Death by a Thousand Cuts. Princeton, NJ:
Princeton University Press.
Hirschman, A. O. 1991. The Rhetoric of Reaction. Cambridge: Cambridge Uni-
versity Press.
Lee, W. and J. Roemer. 2006. “Racism and Redistribution: A Solution to the
Problem of American Exceptionalism.” Journal of Public Economics 90: 1027–
52.
2008. “Moral Values and Distributive Politics: An Analysis of the US 2004
Election,” in I. Shapiro, P. Swenson, and D. Donno (eds.) Divide and Deal:
The Politics of Distribution in Democracies. New York: New York University
Press.
Lindert, P. 2004. Growing Public. Cambridge: Cambridge University Press.
Luttmer, E. 2001. “Group Loyalty and the Taste for Redistribution.” Journal of
Political Economy 109: 500–28.
Murray, C. 1994. Losing Ground. New York: Basic Books.
Piketty, T. 1995. “Social Mobility and Redistributive Politics.” Quarterly Journal
of Economics 110 (3): 551–84.
Poole, K. and H. Rosenthal, 1997. Congress: A Political-Economic History of Roll
Call Voting. Oxford: Oxford University Press.
Roemer, J. E. 1994. “The Strategic Role of Party Ideology when Voters Are
Uncertain about how the Economy Works.” American Political Science Review
88: 327–35.
1998. “Why the Poor Do Not Expropriate the Rich: An Old Argument in New
Garb.” Journal of Public Economics 70 (3): 399–424.
1999. “The Democratic Political Economy of Progressive Taxation.” Econo-
metrica 67: 1–19.
2006. “Party Competition under Private and Public Financing: A Comparison
of Institutions.” Advances in Theoretical Economics 6 (1), article 2. Available
at: www.bepress.com/bejte/advances/vol6/iss1/art2.
Roemer, J. E., W. Lee, and K. Van der Straeten. 2007. Racism, Xenophobia, and
Distribution: Multi-issue Politics in Advanced Democracies. Cambridge, MA:
Harvard University Press.
Smith, M. A. 2000. American Business and Political Power. Chicago: University of
Chicago Press.
2007. The Right Talk: How Conservatives Transformed the Great Society into the
Economic Society. Princeton, NJ: Princeton University Press.
12 The impact of electoral debate on public
opinions: an experimental investigation of the
2005 New York City mayoral election

Sendhil Mullainathan, Ebonya Washington,

and Julia R. Azari

I. Introduction
Political debates have long been a part of the American polity. The early
practice was aimed primarily at promoting parliamentary and elite delib-
eration, as well as informing the broader public about issues and candi-
dates (Jamieson and Birdsell 1988: 37).1 Televised debates have altered
this dynamic, structuring debate as a practice to educate the electorate.
Yet the impact of televised debates on the quality of democracy has been
somewhat controversial, with proponents casting debates as opportuni-
ties “to provide sustained analysis of issues and close comparison of can-
didates” (ibid.: 5), and others expressing the concern that the televised
debates would increase the emphasis on image rather than substance
(Druckman 2003).
Whether debates influence opinions by showcasing candidate view-
points or by simply presenting shallow cues of candidate image, con-
ventional wisdom assumes that televised debates will somehow influence
viewers’ opinions regarding the candidates. Information in the electoral
process is presumed to help voters make decisions more in line with their
preferences (Lupia and McCubbins 1998). In light of the emphasis on
the informative potential of debates, it is natural to pose the question of
whether debates truly provide citizens with information that influences
their opinions or choices.
To this end, there is a large literature examining the impact of debates
on citizens’ political opinions and voting behavior. This research has
concluded that debates are able to alter viewers’ opinions of candidates,

We are grateful to Tiffany Davenport and Meredith Levine for research assistance,
Gregory Huber for his discussion, and ISPS for funding.
1 The Lincoln–Douglas debates of 1858 were notable in this regard. They were well
attended and addressed a number of issues related to the national question of slavery
(Jamieson and Birdsell 1988: 49).

324
 The impact of electoral debate on public opinions 325

but only in a small way. This literature states that debates have more
influence the earlier in the electoral season they occur. Further, the liter-
ature concludes that debates have greater influence over the uninformed
than over those who know quite a bit about the election.2
Debate studies typically follow a panel design, interviewing voting-
aged individuals both before and after the debate of interest. The pre-
post-debate opinion change of those who watched is compared with
the opinion change of those who did not watch in a “difference-in-
difference” style methodology. The limitation of such an analysis is the
potential endogeneity of debate watching. Perhaps only those who are
most inclined toward opinion change actually watch the debate. In such
a case the coefficient on “watch” would be an overestimate of the impact
of the debate on the average citizen. Or perhaps only political junkies,
who have the longest-held, most intransigent, political opinions, are in
the viewing audience. In such a case the coefficient on “watch” would be
an underestimate of the impact of the debate on the average citizen.
We overcome this limitation by employing an experimental panel
design. During the 2005 election season, in the days leading up to the
final debate between mayoral incumbent Republican Michael Bloomberg
and Democratic challenger Fernando Ferrer, we interviewed a random
sample of 1,000 New York City voters. We randomly assigned these
1,000 individuals to one of two groups: we asked the treatment group to
watch the November 1 debate, and we asked the control group to watch
a “placebo” program, PBS’s The NewsHour with Jim Lehrer, which aired
opposite WNBC’s debate broadcast. Our intervention was effective. Sta-
tistically indistinguishable from controls in the pre-debate interviews, in
post-debate interviews treatment group members reported watching the
debate at a rate that was more than twice as high as controls. (We verified
viewership by asking factual questions about the debate.) Thus we can
take opinion differences between treatment and controls as a measure of
the causal impact of the debate on political viewpoint.
Using this methodology we find that those in the watch group were
6 percentage points more likely to report that their opinions of one or both
candidates had changed from the first to the second interview. However,
when asked to rate the candidates on two of the most salient campaign
issues, terrorism and housing, the post-debate responses of the treatment
group were statistically indistinguishable from those of the control group.
Clearly in a short survey we could not address every possible issue on
which opinions may have changed. However, the fact that there was also
no statistical difference between treatment and control group members’

2 See Hellweg, Pfau, and Brydon (1992), Holbrook (1996), or Kraus (2000), for a review
of this literature.
326 Sendhil Mullainathan et al.

ratings of candidates generally, nor on their likelihood of voting for each

candidate, strongly suggests that although the debate led respondents
to believe that their opinions had changed, the debate actually effected no
meaningful political opinion change. The study highlights the incongruence
between respondents’ stated opinion change and respondents’ actual
opinion change, a discrepancy pointed out by Gerber and Green (1999).
Why do we see no real opinion change in this study? Several possibilities
exist: the first is that debates actually change no one’s political opinions.
Second, it is possible that debates shift everyone’s political opinions.
Because the highlights of debates are reported through the media and
over the water cooler, perhaps one does not need to watch the debate to
be affected by it.3 Third, perhaps debates on average do affect opinion
change, and this debate was an exception. In other words, perhaps the
external validity of the present work is limited. This could be due to its
timing – just days before the general election – or perhaps because of the
fact that this particular contest was never close. Or perhaps our sample –
older, more educated and more likely to vote than the average New
Yorker – is not a population whose views are swayed by debates.
In the next section we detail the history of the 2005 New York City
mayoral election. In Section III we describe the data and methodology,
before presenting results in Section IV. Section V discusses some potential
implications of the findings in the context of the broader question of
political knowledge and democracy. The final section concludes.

II. 2005 New York City mayoral election4

Former Bronx borough president, Democrat Fernando Ferrer, was never
a serious contender in the 2005 New York City general mayoral election.
The top vote-getter in the Democratic primary, he garnered only 39.949
percent of the vote. By law, 40 percent must be won by a single candi-
date to avoid a runoff. Fortunately for Ferrer, the second place finisher
decided, for the sake of the party, to refrain from participating in the
runoff. And thus Ferrer began his eight-week campaign against Michael
Bloomberg, a Republican mayor, who enjoyed a 60 percent approval rat-
ing in an overwhelmingly Democratic city. (Seventy-four percent of the
city’s votes were cast for Democrat John Kerry in the 2004 presidential
election [Andersen 2004].)

3 We turn to post-debate interviews conducted the night of the debate to control for this
diffusion effect. Unfortunately, the sample interviewed that night was too small for this
analysis to be informative.
4 The sources for this section are various articles from the 2005 New York Times: Healy
(2005a–d); Anon. (2005); Purnick (2005); Healy and Lueck (2005); Healy and Connelly
(2005); Haberman (2005); and Rutenberg (2005).
 The impact of electoral debate on public opinions 327

Even more than approval, Bloomberg had money. In what the New
York Times regarded as “drowning-by-spending,” the billionaire incum-
bent spent more than $70 million on his campaign. Ferrer’s only possible
opportunity to present himself on an equal footing with the mayor came
through the three debates endorsed by the city’s campaign finance pro-
gram. But as Bloomberg was financing his own campaign, he was under
no obligation to participate in these events. He decided to meet Fer-
rer in two of the three. Bloomberg avoided the first debate because he
did not like its format: the October 9 event included Conservative Party
candidate Thomas Ognibene who attacked Bloomberg for failing to be a
true Republican. Ferrer and Ognibene spent most of the debate throwing
criticisms at the empty lectern where the mayor would have stood.
The remaining debates occurred on October 30 and November 1,
much later in the electoral season. At this point Bloomberg enjoyed a 27
percentage point lead in the polls. Bloomberg even led Ferrer amongst
the more often Democratic subgroups of liberals, blacks and women.
Nonetheless, at 9 a.m. on Sunday, October 30, Ferrer “tore into Mayor
Michael Bloomberg.” Ferrer had what the New York Times dubbed “the
best day so far in his race for mayor.” Two days later, however, Bloomberg
evened the debate score. He was “far more assertive” than in the first
debate in which he participated. The media and our survey respondents
agreed. Bloomberg won this final debate, which is the focus of the inquiry
in this chapter. The mayor then went on to win the November 9 election
by more than 20 points, a record win for a Republican mayoral candidate
in New York City.

III. Data and methodology

We investigate the question of how this final debate between Bloomberg
and Ferrer affected public opinion of these candidates by employing an
experimental panel design. Respondents, randomly selected from the
New York City voter registration list,5 were first surveyed by telephone
between October 28 and October 31. Using this initial survey we estab-
lished respondents’ pre-debate attitudes about how life was going in New
York City and about the two mayoral candidates. Respondents were ran-
domly divided into one of two groups: we asked the treatment group to
watch the final mayoral debate on November 1, and we asked the control
group to watch a placebo program, PBS’s The NewsHour with Jim Lehrer,
which aired opposite WNBC’s debate broadcast. We then interviewed

5 Our selection procedure gave greater weight to likely voters. The survey company phoned
the home of likely voters but then asked for the adult registered in New York City whose
birthday would come the soonest.
328 Sendhil Mullainathan et al.

respondents for a second time between November 1 and November 3

to establish post-debate attitudes toward the candidates. Our final sam-
ple contains pre- and post-debate interviews with 1,000 New York City
residents: 505 in the control group and 495 in the treatment condition.
Our methodology is simple. We measured the debate’s impact on polit-
ical views by examining the difference between treatment and control
group members’ post-debate views. Given our panel design, an alter-
native strategy would have been to compare the change (from pre- to
post-debate) in political views of treatment and control group members.
We did not follow such a strategy because we did not want to ask pre-
cisely the same questions in the pre- and post-debate interviews. Given
that the pre- and post-debate interviews occurred within days of one
another, we wanted to avoid respondents’ post-debate answers being
influenced (or anchored) by their pre-debate answers. Instead of this
explicit differencing, we performed an implicit differencing strategy by
assuming that members of the treatment group held, on average, the
same political views as controls. Therefore, a comparison of post-debate
opinion differences between the groups is equivalent to a comparison of
pre-post-debate changes in opinion differences between the groups.
Thus the first assumption that underlies our identification strategy is
that treatment and control group members do not differ significantly
in political views before viewing the debate. We test this assumption in
Table 12.1 by comparing the two groups on pre-debate opinions.
The first column of the table gives the sample mean on the opinion
(demographic). For example the 0.77 in the first column indicates that
77 percent of the sample agreed with the statement, “New York City
is headed in the right direction.” The second and third columns of the
table give the means for the Watch (treatment) and Don’t Watch (control)
groups. These columns indicate that 78 and 75 percent of the Watch and
Don’t Watch groups respectively agreed that the city was headed in the
right direction. In the final column of the table we provide the results of a
test of the null hypothesis that the two groups are drawn from populations
with the same mean. The fact that the final cell in the first row of the table
is blank therefore says that the difference between the Watch and Don’t
Watch groups in opinion of whether the city is headed in the correct
direction is statistically insignificant.
The same is true for the remaining pre-debate opinions. Control and
treatment group members are statistically indistinguishable on party pref-
erence. The sample mean is 15 percent Republican and 75 percent
Democrat. (By treatment status, those figures are 16 percent Republican
and 73 percent Democrat for treatments and 13 percent Republican and
76 percent Democrat for controls.) The groups are also equally likely to
choose Bloomberg (Ferrer) as their preferred candidate. (The figures are
 The impact of electoral debate on public opinions 329

Table 12.1 Summary statistics

Don’t Test of equality of

Full Watch Watch Watch and Don’t
sample group group Watch groups

Pre-debate opinions
NYC is headed in the right direction 0.77 0.78 0.75
Republican 0.15 0.16 0.13
Democrat 0.75 0.73 0.76
Will vote for Bloomberg (asked of only 0.65 0.64 0.66
half the sample)
Will vote for Ferrer (asked of only half 0.31 0.34 0.28
the sample)
Voting history
Voted in 2001 0.96 0.96 0.97
Voted in 1997 0.80 0.78 0.81
Demographics
Age 64 63 64
White 0.71 0.68 0.74 ∗∗

Black 0.19 0.21 0.17

Female 0.61 0.61 0.61
College graduate 0.52 0.52 0.52
Post-debate checks
Watched debate 0.27 0.37 0.16 ∗∗∗

Correctly identified moderator’s race (of 0.56 0.6 0.46 ∗∗

268 who claimed to have watched)

Correctly identified moderator’s gender 0.64 0.69 0.54 ∗∗

(of 268 who claimed to have watched)

N 1,000 505 495

Note: ∗∗∗ denotes significance at the 1 percent level, ∗∗ at the 5 percent level, and ∗ at the
10 percent level. Sample size varies due to item non-response.

64 percent for Bloomberg and 34 percent for Ferrer among treatments

and 66 percent for Bloomberg and 28 percent for Ferrer among controls.)
In summary, the evidence in the first section of Table 12.1 supports the
assumption that pre-debate opinions are the same for the two groups.
We further test this assumption by looking at voting history. By match-
ing our sample to New York City voting records we examine whether
respondents voted in the mayoral elections of 2001 and 1997. The
96 percent voting rate in 2001 reflects the fact that we predominantly
sampled those who, based on their voting history, were likely to vote in the
upcoming election. In neither year is the difference in voting rate between
treatment and controls statistically meaningful. Thus voting history data
provide further support for the assumption that pre-debate treatment
and controls are the same politically.
330 Sendhil Mullainathan et al.

Finally, we test the assumption by turning to demographic charac-

teristics. On age, gender, and education there is no meaningful differ-
ence between the two groups. Only on one demographic, percentage of
whites, do we see a statistically significant difference. The Watch group
is 68 percent white; the Don’t Watch group is 74 percent white. Clearly
one statistically significant difference in 12 is no indictment of our ran-
domization procedure. Nonetheless, we will present our results with and
without controls for demographics to adjust for any pre-debate differ-
ences between the two groups.
Our assumption of equivalence of opinions among treatments and con-
trols thus seems defensible. A second assumption on which our analysis
rests is that our intervention – asking respondents to watch the debate6 –
was in fact effective. In the final section of Table 12.1 entitled “Post-
debate checks” we test this assumption. The row marked “Watched
debate” indicates that 37 percent of the treatment group but only 16
percent of the control group report having watched the debate. This dif-
ference is statistically significant at the 1 percent level. But our assump-
tion is not that those in the treatment group report watching the debate
at a greater rate than controls, which could arise out of a desire, on the
part of respondents, to please the interviewer.7 Our assumption is that
those in the treatment group actually watched the debate at a greater rate
than controls. To test the veracity of respondents’ viewing reports, we
asked those who claimed to have watched the debate to identify both the
race and the gender of the moderator. The moderator of the November
1 debate was a white man, while the panelists (questioners) included a
white man, a white woman, a black man and a Latino man.8 Fifty-six
percent of those who claimed to have watched the debate identified the
race of the moderator correctly. Sixty-four percent of those who claimed

6 Fifty-four percent of respondents asked in the pre-debate survey indicated that they
would watch the debate for us. Thirty-one percent said they would not. The remaining
respondents were not sure.
7 See, for example, Gales and Kendall (1957); Gray (1956); or Hanson and Marks (1958)
on interviewer effects.
8 The moderator’s race and gender were the same for both debates in which Bloomberg
participated. So unfortunately respondents may be given credit for correctly identifying
the moderator when they are in fact correctly identifying the moderator for the wrong
debate. Over 80 percent of the sample had their pre-debate interview before October 31 –
the date of the first Bloomberg/Ferrer debate. Thus it is possible that some treatment
members mistakenly watched the wrong debate. If half of those watching saw the debate
that Bloomberg won and half saw the debate in which Ferrer prevailed, then it is possible
that the reason we find no debate effects on average is because the debate in fact pulled
the two groups in opposite directions. However, such a split in debate watching cannot
explain the fact that on average debate watchers claimed to feel relatively more favorable
to Ferrer rather than unchanged. In the debate in which the mayor did not participate
the moderator was a black man.
 The impact of electoral debate on public opinions 331

Table 12.2 Impacts of commitment to vote on voting

(dependent variable is an indicator for having voted in
the 2005 election)

(1) (2) (3)

Commit group indicator 0.042∗ 0.043∗ 0.040∗

(0.023) (0.024) (0.024)
Demographic controls Yes Yes
Voting history controls Yes
N 795 721 721

Notes: Demographic controls include age, race, gender, and edu-

cation. Voting history controls include indicators for voting in the
1997 and 2001 New York City mayoral elections.

to have watched the debate identified the gender of the moderator

correctly. Thus it seems that respondents, on average, are truthfully
reporting their viewership.9 Amongst those who claim to have watched
the debate, treatment group members are significantly more likely to
correctly identify the moderator’s race and gender. To the extent that
incorrect answers on moderator gender and race proxy for not actually
having watched the debate, our results show that those in the treatment
group were more likely to have watched.
The idea that simply phoning an individual and asking her to under-
take some activity (in this case watching a debate) at a future date for
no remuneration would result in compliance is hard to believe. There-
fore we tested our methodology with an activity that we could actually
verify – voting. In the post-debate survey, respondents, regardless of
treatment/control debate status, were randomly assigned to a commit
condition or a control condition. Those in the commit condition were
asked an additional two questions at the end of the survey. The first was
whether they would vote in the upcoming mayoral election (96 percent
said they would) and the second was whom they would vote for (58 per-
cent said Bloomberg). Those in the control group were not asked about
future expectations for voting nor for vote choice. We then matched our
sample to voting records. As the results in Table 12.2 indicate, those in
the treatment group voted in the 2005 election at a rate that is a significant
4 percentage points higher than those in the control.

9 There is, however, the possibility that respondents learned of the moderator’s demo-
graphics by seeing excerpts of the debate on the news. Fortunately the news outlets were
more interested in giving sound bites of candidates responding than of the moderator’s
questioning.
332 Sendhil Mullainathan et al.

Table 12.3 Summary statistics for commitment to vote treatment

Test of equality of
Commit Non-committed committed and
group group non-committed groups

Demographics
Age 65 62 ∗

White 0.71 0.72

Black 0.21 0.18
Female 0.63 0.59
College graduate 0.50 0.54
Voting History
Voted in 2001 0.96 0.97
Voted in 1997 0.82 0.77
Pre-debate opinions
NYC is headed in the
right direction 0.77 0.76
Republican 0.15 0.15
Democrat 0.75 0.74
Sample size 500 500

Note: ∗∗∗ denotes significance at the 1 percent level, ∗∗ at the 5 percent level,
and ∗ at the 10 percent level. Sample size varies due to item non-response.

This is a sizable increase on a mean of 88 percent voting.10 As the final

columns of the table show, this 4 percentage point difference is robust
to the inclusion of controls for demographics and for voting history.
The means for these demographics, by voting treatment status, are pre-
sented in Table 12.3. Thus this simple voting manipulation experiment
provides evidence of the plausibility of the effectiveness of our debate
manipulation.

IV. Results
The basic results of our survey are derived from the post-debate interview.
As the results in Table 12.4 indicate, those in the Watch group were more
likely than those in the Don’t Watch group to believe that their opinion
of one or both candidates had changed since the time of their initial
interview.

10 Four percentage points is about half the expected size of the most effective GOTV (“get
out the vote”) interventions, such as personal canvassing (Gerber and Green 2000).
 The impact of electoral debate on public opinions 333

Table 12.4 Post-debate summary statistics

Don’t Test of equality of

Full Watch Watch Watch and Don’t
sample group group Watch groups

Have experienced change in opinion 0.48 0.51 0.45 ∗

toward one or both candidates

Opinion of Bloomberg changed 0.30 0.33 0.27 ∗∗

More favorable to Bloomberg 0.10 0.09 0.10

Opinion of Ferrer changed 0.35 0.37 0.32
More favorable to Ferrer 0.00 0.04 −0.03
Bloomberg seems more knowledgeable 0.83 0.82 0.85
Bloomberg thermometer 69 70 69
Ferrer thermometer 48 49 47
Bloomberg–Ferrer thermometer 21 21 21
Bloomberg better on terrorism 0.89 0.89 0.90
Bloomberg better on housing 0.53 0.50 0.55
Will vote for Bloomberg (asked of only 0.69 0.70 0.67
half the sample)
Will vote for Ferrer (asked of only half 0.28 0.27 0.29
the sample)
N 1,000 505 495

Note: ∗∗∗ denotes significance at the 1 percent level, ∗∗ at the 5 percent level, and ∗ at the
10 percent level. Sample size varies due to item non-response.

The first row of Table 12.4 shows the mean answers to precisely that
question. Forty-eight percent of the full sample said that their opinion
had changed. By treatment status, 51 percent of those in the Watch
group and 45 percent of those in the control group agreed with that
statement. This 6 percentage point difference, statistically significant at
the 10 percent level, is the intention-to-treat effect. However, we know
that not everyone in the treatment group actually watched the debate.
The difference in viewership between treatments and controls was not
100 percentage points, but merely 21 percentage points. Thus we scale
our 0.06 effect size by 0.21 and conclude that the effect of the debate
on those who actually watched was a 29 percentage point increase in the
belief that their opinion had changed over the past few days. Or if we only
consider those who correctly identified the moderator’s demographics as
having seen the debate, our treatment effect rises to a 43 percentage point
increase in the belief that their opinion had changed.11

11 Fifty-seven percent of the 37 percent of treatments who said they watched the debate
correctly identified the moderator’s race and gender. Or in other words 21 percent
(0.37 × 0.57) of treatments correctly identified the moderator’s race and gender.
334 Sendhil Mullainathan et al.

We next asked participants about opinion change regarding each of

the candidates individually. Thirty-three percent of the Watch group
felt their opinion of Bloomberg had changed; while only 27 percent
of controls agreed with that statement, a difference that is statistically
significant at the 5 percent level. This 6 percentage point intention to
treat effect corresponds with a 29 percentage point treatment effect (or
43 percentage points if we consider only the verified viewership).
While those in the Watch group were more likely to say their opinion
of Bloomberg had changed, they were no more likely than those in the
Don’t Watch group to say that their opinion of Bloomberg had changed
in a positive (or negative) direction. The third row of the table exam-
ines responses to the question of how opinions toward Bloomberg had
changed over the last few days. Respondents were offered the answer
choices of “more favorable” coded 1, “unchanged” coded 0 or “less
favorable” coded −1. On average survey respondents felt more favorable
towards Bloomberg, but there were no significant differences between the
two groups. Watch group members saw a mean of 0.09 for this variable
and Don’t Watch members had a mean of 0.10.12
Watch group members were also more likely than controls to say that
their opinion of Ferrer had changed. Watch group members expressed
this view 37 percent of the time compared to only 32 percent of the time
amongst controls. This 5 percentage point difference, however, is not
statistically significant. Treatment group members on average saw their
opinion of Ferrer rise, while controls saw their opinion fall. This 0.07
point difference (0.04 v. −0.03) is just shy of significance at conventional
levels.13 Thus when asked directly whether their opinions had changed,
those who were exogenously induced to watch the debate were more
likely than controls to indicate that their opinions had changed in the last
few days.
Yet when their opinions were measured directly, those in the treatment
group held post-debate opinions that were statistically indistinguishable
from those of treatment group members. The remainder of Table 12.4
shows these results. The first direct opinion question (shown in row 6

Viewership of only 7 percent of controls (16 percent claimed to have viewed × 43

percent of those named the moderator race and gender) was verified. Thus the dif-
ference between treatment and control verified viewership is 14 percentage points. We
divide 0.06 by 0.14 to arrive at a treatment effect of 43 percentage points.
12 This lack of significance between the two groups is caused not merely by an equal
number of positive and negative Bloomberg switchers in the treatment group. The two
groups saw comparable positive and negative change: 12 (20) percent of treatments had
a less (more) favorable opinion of the mayor and 8 (19) percent of controls had a less
(more) favorable opinion of the mayor.
13 The p value on the t test is 0.1067.
 The impact of electoral debate on public opinions 335

of the table) asked respondents whether they felt Bloomberg or Ferrer

was more knowledgeable. Eighty-two percent of the Watch group and
85 percent of the Don’t Watch group chose Bloomberg. The three-point
difference is statistically indistinguishable. Next, we asked respondents
to rate their feelings about each of the two candidates on a scale from 0
to 100.14 Once again, there were no significant differences between treat-
ment and control group members. The groups rated Bloomberg 70 and
68 respectively and Ferrer 49 and 47. The average difference in rating
between the two candidates, 21 points for both groups, was also statis-
tically insignificant. We turned to two hot button issues of the electoral
campaign next. We asked which candidate would perform better on (1)
terrorism and (2) housing. Eighty-nine percent of Watch group mem-
bers and 90 percent of Don’t Watch group members thought Bloomberg
would outperform Ferrer on the terrorism issue. On the housing issue,
Bloomberg was favored by 50 percent of treatments and 55 percent of
controls. Neither difference was statistically significant. Finally, we asked
respondents for whom they planned to vote. Bloomberg was chosen by 70
percent of Watch group members and 67 percent of Don’t Watch group
members. Ferrer was chosen by 28 and 29 percent of the two groups
respectively. Once again, neither difference was statistically significant.
Thus the evidence of the second half of Table 12.4 indicates that the
debates had no impact on political opinions.
The basic finding of an impact of debate watching on belief of opinion
change but of no impact of watching on actual opinion change is robust
to the inclusion of demographic and voting history controls, as shown in
Table 12.5.
The first column of the table reproduces the results of Table 12.4
in regression form. Thus whereas in Table 12.4 we run t-tests, in
Table 12.5 we test the equality of effect between the two groups by
running regressions of the following form:
(1) Outcome = a + B (Watch group indicator) + e.
We run a separate equation for each outcome of interest. As the tests
are equivalent the results are of course identical: we see a significant
impact of debate watching on the belief that someone’s opinion has
changed in general and that their opinion of Bloomberg has changed
in particular. In column 2 of the table we add demographic controls
(summarized in Table 12.1) to the model. Our qualitative findings remain
unchanged. We continue to see significant impacts on changes of opinion
of Bloomberg. Additionally, the finding that the Watch group has raised

14 The ordering of the candidates in the questioning varied randomly.

336 Sendhil Mullainathan et al.

Table 12.5 Impact of debate on opinions of candidates (each cell

presents the coefficient on a different regression).

(1) (2) (3)

Outcome
Have experienced change in opinion toward 0.058∗ 0.057∗ 0.049
one or both candidates (0.033) (0.034) (0.039)
Opinion of Bloomberg changed 0.06∗∗ 0.053∗ 0.055
(0.029) (0.031) (0.035)
More favorable to Bloomberg −0.011 −0.023 −0.014
(0.035) (0.037) (0.041)
Opinion of Ferrer changed 0.051 0.053 0.044
(0.031) (0.033) (0.037)
More favorable to Ferrer 0.062 0.082∗∗ 0.075∗
(0.039) (0.041) (0.046)
Bloomberg seems more knowledgeable −0.032 −0.033 −0.033
(0.027) (0.028) (0.031)
Bloomberg thermometer 1.234 0.870 −0.774
(1.731) (1.798) (1.925)
Ferrer thermometer 2.117 20.081 0.987
(1.721) (1.790) (1.974)
Bloomberg–Ferrer thermometer −0.758 −1.344 −2.341
(2.412) (2.488) (2.728)
Bloomberg better on terrorism −0.001 0.007 0.016
(0.022) (0.024) (0.025)
Bloomberg better on housing −0.045 −0.033 0.000
(0.035) (0.036) (0.040)
Will vote for Bloomberg (asked of only half 0.028 0.025 0.028
the sample) (0.045) (0.046) (0.050)
Will vote for Ferrer (asked of only half the −0.013 −0.002 −0.009
sample) (0.044) (0.045) (0.049)

Controls
Demographic controls Yes Yes
Voting history controls Yes

Note: Sample size varies among specifications due to non-response. Total sam-
ples size is 1,000. ∗∗∗ denotes significance at the 1 percent level, ∗∗ at the
5 percent level, and ∗ at the 10 percent level.

its opinion of Ferrer more than the control group has is now statistically
significant at the 5 percent level. All of the direct measures of opinion
change continue to show insignificant coefficients on the treatment status
dummy. In column 3 we add voting history controls (also summarized in
Table 12.1) to the model of column 2. Coefficients are little changed
for all outcomes as we move from column 2 to column 3. However, as
we utilize increasing degrees of freedom, the general opinion change and
 The impact of electoral debate on public opinions 337

the Bloomberg opinion change specifications no longer show significant

treatment/control differences. The only significant difference between
the groups remains in their opinions of Ferrer. The treatments believe,
more than controls, that their opinion of Ferrer has improved over the
last few days.
Thus our basic finding is robust to the inclusion of demographic and
voting history controls: we show that debate watching increased partic-
ipants’ beliefs that their opinions of the candidates had changed in the
past few days. However, when measuring directly, we find no evidence
of actual opinion change.

V. Implications
Since the experimental design used in this study overcomes the endogene-
ity issues endemic to observational studies of debate effects, we cannot
attribute the lack of change in participants’ opinions to their preexisting
strong opinions (Zaller 1992), or to some other shared characteristic of
debate watchers. Therefore, the findings suggest that, independent of the
effects of existing views or political awareness, the experience of viewing
political debates did not change opinions about the candidates. In this
section we discuss the implications of both this finding and the finding
that respondents thought their opinions had changed when in fact they
had not.
Democratic ideals depend on an informed electorate capable of making
reasoned political choices (Delli Carpini and Keeter 1996: 1). Responsi-
ble party government enforced through electoral accountability requires
mechanisms by which citizens can learn about the policy positions of the
candidates vying to be their leaders. More candidate-centered approaches
to political competition depend on such processes as well in order for
citizens to make judgments about the character or personal qualities
of candidates. As stated in the introduction, debates in the contempo-
rary era have been seen as a means of helping candidates communicate
their message to citizens and, in turn, for citizens to use the information
gained by watching debates to form opinions of candidates (Jamieson and
Birdsell 1988). In light of the intended relationship between debates and
popular rule, the results of this experiment are particularly striking: we
find that debates do not perceptibly change citizens’ opinions, and – even
more notable – despite the evidence that actual views had not changed,
participants in this study were under the impression that their views had
changed after watching the debates.
American political thought and history have long grappled with the
tension between the ideal of popular rule and the possibility of a poorly
338 Sendhil Mullainathan et al.

informed electorate. While revolutionary ideals may have rested in part

on an image of “the people” who are “wiser than their governors”
(Morone 1998: 5), a streak of skepticism about this characterization
of “the people” runs through early American political thought as well.
Madison repeatedly warns in the Federalist against the potential for the
people to rule using “passion” rather than “reason.”15 While the 1776
Revolution against British rule was based on a radical democratic ideol-
ogy, the dominant sentiment among ruling elites in the new American
republic soon took a more conservative turn, motivated in part by a heavy
skepticism about the ability of the mass electorate to make informed
political decisions (Wood 1998).
The question of whether popular rule is advisable or even possible
under conditions of limited citizen information is very much alive in con-
temporary normative and empirical studies. Contemporary inquiries into
the role of information in the functioning of American democracy call
into question the level of information possessed by the American elec-
torate (Delli Carpini and Keeter 1996), the effect of political awareness
and reception of political messages on public opinion (Zaller 1992), and
the importance of information for making reasoned political decisions
(Lupia and McCubbins 1998).
How does the literature on political knowledge bear upon the impli-
cations of a study whose findings concern the opinions, rather than the
political knowledge, of the participants? Implicit in the literature on polit-
ical knowledge is a link between information and political judgment and
behavior. Lupia and McCubbins, who investigate this link systematically,
insist that perfect information is not needed for citizens to make reasoned
political choices; what is required for reasoned choices is what the authors
call “knowledge” (ibid.: 6), defined as “the ability to predict the conse-
quences of actions.” Their point speaks directly to the implications of this
study: the purpose of practices intended to enhance political knowledge
is to give citizens the tools to change their opinions about candidates
in order to more closely match their true preferences. Our discussion
of the implications of the results of the study in this broader context
must be cautious; we do not know anything for sure about the voting
behavior of the study participants. Obviously we cannot say anything
about the ultimate impact of the perceived shift in opinions following the
debates.

15 Madison et al. (1982: 310–14). The arguments advanced in these documents center
on the concern about creating institutional arrangements that will be conducive to the
“people’s rule” using reason rather than passion, distinct from more contemporary ideas
about creating an informed populace.
 The impact of electoral debate on public opinions 339

We can, however, cautiously consider the implications of the findings

for the complex constellation of relationships at stake in the study of
mass opinion and knowledge, media events such as televised debates,
and broader notions of democratic rule. The fundamental purpose of
debates – namely, to inform citizens about candidates’ characteristics
and issue positions – has been explicitly linked with the objective of
changing opinions; contemporary models of political information do not
regard such information as an intrinsic good, but as means to an end.
That end is the ability of citizens to make political choices that better
reflect their underlying values and preferences. Here we return to Lupia
and McCubbins’s thesis that reasoned political decisions are possible with
limited political information as long as meaningful cues are available, and
that political information is costly and “provides no instrumental benefit”
if it does not change citizens’ decisions (ibid.). As stated before, we do not
know whether watching the debate had an impact on citizens’ vote choice.
However, the results suggest that if opinions of the candidates are the
bases of such decisions, we might expect that the debates did not have
this kind of impact. Further research is needed to determine whether the
stability in opinions after watching the debates is associated with stability
in electoral behavior. The absence of substantial opinion shift following
the debates may suggest that debates are costly and provide little actual
value to citizens who rate political information only insofar as it helps
them make political decisions that reflect their preferences and views. The
fact that participants thought their opinions had changed when in fact
they had not suggests that the process of incorporating new information
may be more complex than previously theorized. For instance, we do not
know whether the perceived changes in opinion of Bloomberg among
the treatment group led to changes in voting behavior. However, the
results of this study suggest that there may be a distinction between
voting based on an actual opinion, and based on a perceived opinion.
Our findings suggest that the relationship between political information,
political opinions, and candidate evaluation may be even more complex
than current scholarship indicates. Debates and other means of informing
citizens about candidates may not function in as straightforward a manner
as conventional wisdom would predict.

VI. Conclusion
Previous research on the impact of candidate debates on public opin-
ion has suffered from potential bias due to the endogeneity of debate
viewing. We overcome this limitation by employing an experimental panel
design. We randomly assign study participants to watch either the final
340 Sendhil Mullainathan et al.

2005 New York City mayoral debate or a placebo program. We find that
debate watching increased participants’ beliefs that their opinions of the
candidates had changed in the past few days. However, when measured
directly, we find no evidence of actual opinion change. Thus this study
highlights the incongruence between respondents’ stated opinion change
and respondents’ actual opinion change, a discrepancy pointed out by
Gerber and Green (1999).
We do not conclude, however, that debates never induce opinion
change. There are several possible explanations for our lack of opin-
ion change findings in this context. The first is, of course, that debates
actually change no one’s political opinions. The second is that debates
move everyone’s political opinions. Because the highlights of debates are
reported through the media and over the water cooler, perhaps one does
not have to watch the debate to be affected by it. Third, perhaps our find-
ings are not generalizable to the voting population at large. Our research
setting – New York City – is one of the most politically homogeneous
locales in the nation. Our focal debate occurred just days before the gen-
eral election. Our sample was older, more educated and more likely to
vote than the average New York City resident.
Thus we conclude with a suggestion and a word of caution. We suggest
that future work on the persuasive effects of debates be conducted using
an experimental research design. Such a design overcomes the bias from
the endogeneity of debate viewership. We also caution against the use of
stated opinion change as a proxy for actual opinion change. Our results
highlight that these two variables do not measure the same underlying
construct.

BIBLIOGRAPHY
Andersen, Kurt. 2004. “People Like Us.” New York 37 (November 22): 24–6.
Anon. 2005. “The Shunned Debate.” New York Times, October 9, p. CY11.
Delli Carpini, Michael X. and Scott Keeter. 1996. What Americans Know About
Politics and Why it Matters. New Haven, CT: Yale University Press.
Druckman, James N. 2003. “The Power of Television Images: The First
Kennedy–Nixon Debate Revisited.” Journal of Politics 65 (May): 559–71.
Gales, Kathleen and M. G. Kendall. 1957. “An Inquiry Concerning Interviewer
Variability.” Journal of the Royal Statistical Society, ser. A. (General) 120 (2):
121–47.
Gerber, Alan, and Donald Green. 1999. “Misperceptions about Perceptual Bias.”
Annual Review of Political Science 2 (June): 189–210.
2000. “The Effects of Canvassing, Telephone Calls, and Direct Mail on
Voter Turnout: A Field Experiment.” American Political Science Review 94
(September): 653–63.
 The impact of electoral debate on public opinions 341

Gray, Percy G. 1956. “Examples of Interviewer Variability Taken from Two

Sample Surveys.” Applied Statistics 5 (June): 73–85.
Haberman, Clyde. 2005. “Watch Debate? Sleep In? Boxers? Briefs?” New York
Times, October 28, p. B1.
Hamilton, Alexander, James Madison, and John Jay. 1982. The Federalist Papers.
New York: Bantam Books.
Hanson, Robert H. and Eli S. Marks. 1958. “Influence of the Interviewer on the
Accuracy of Survey Results.” Journal of the American Statistical Association
53 (September): 635–55.
Healy, Patrick. 2005a. “Weiner Concedes Race for Mayor to Avert Runoff; Says
Ferrer is Nominee.” New York Times, September 15, p. A1.
2005b. “Ferrer Makes His Candidacy Look Stronger Than the Polls Suggest.”
New York Times, October 31, p. B4.
2005c. “Attacks Escalate in Final Debate of Mayor’s Race; Rivals Go on Offen-
sive.” New York Times, November 2, p. A1.
2005d. “Bloomberg Cruises to Re-Election Victory; Corzine is Winner in
Costly New Jersey Race.” New York Times, November 9, p. A1.
Healy, Patrick, and Marjorie Connelly. 2005. “Even Among Democratic Voters,
Poll Finds Ferrer is Well Behind.” New York Times, October 28, p. A1.
Healy, Patrick, and Thomas Lueck. 2005. “Ferrer Levels Attack on Bloomberg
Over Iraq War and the Need for Low-Cost Housing.” New York Times,
October 27, p. B3.
Hellweg, Susan A., Michael Pfau, and Steven R. Brydon. 1992. Televised Presi-
dential Debates: Advocacy in Contemporary America. New York: Praeger.
Holbrook, Thomas M. 1996. Do Campaigns Matter? Thousand Oaks, CA: Sage
Publications.
Jamieson, Kathleen Hall, and David S. Birdsell. 1988. Presidential Debates: The
Challenge of Creating an Informed Electorate. New York: Oxford University
Press.
Kraus, Sidney. 2000. Televised Presidential Debates and Public Policy. Mahwah, NJ:
Lawrence Erlbaum Associates.
Lupia, Arthur and Mathew D. McCubbins. 1998. The Democratic Dilemma: Can
Citizens Learn What They Really Need to Know? New York: Cambridge Uni-
versity Press.
Madison, James, Alexander Hamilton, and John Jay. 1982. The Federalist Papers.
New York: Bantam, 1982.
Morone, James A. 1998. The Democratic Wish: Popular Participation and the Limits
of American Government. New Haven, CT: Yale University Press.
Purnick, Joyce. 2005. “So Far, a Race Without Chutzpah.” New York Times,
October 20, p. B1.
Rutenberg, Jim. 2005. “In Equal Footing of a Debate, Ferrer Gets Feisty.” New
York Times, October 31, p. A1.
Wood, Gordon S. 1998. The Creation of the American Republic, 1776–1787. Chapel
Hill: University of North Carolina Press.
Zaller, John R. 1992. The Nature and Origins of Mass Opinion. New York: Cam-
bridge University Press.
13 Swing voters, core voters, and
distributive politics

Gary W. Cox

How do political parties allocate targetable goods – such as private goods

targeted to individuals, local public goods targeted to geographic areas,
or tax breaks targeted to specific industries or firms – in order to opti-
mize their electoral prospects? A continuing debate on this question pits
those who lean toward Cox and McCubbins’s (1986) “core voter model”
against those who lean toward Lindbeck and Weibull’s (1987) “swing
voter model.” Both models envision two parties competing to win an
election by promising to distribute targetable goods to various groups,
should they be elected. Cox and McCubbins argue that vote-maximizing
parties will allocate distributive benefits primarily to their core voters. A
typical response embodying the swing voter logic is Stokes’s (2005: 317):
“voters who are predisposed in favor of [a party] on partisan or program-
matic grounds [ – that is, its core voters – ] cannot credibly threaten to
punish their favored party if it withholds [distributive] rewards. Therefore
the party should not waste rewards on them.”
In this chapter, I first review the literature and then note that extant
models focus solely on persuasion (defined as an attempt to change vot-
ers’ preferences between given alternatives). Once one brings coordination
(defined as an attempt to affect the number and character of alternatives
from which voters choose) and mobilization (defined as an attempt to
affect whether or not citizens participate in the election) into analytic
view, the argument that vote-maximizing parties should focus their dis-
tributive benefits on core voters is substantially strengthened. Lowering
the number of ideologically similar competitors on the ballot and mobi-
lizing its base are often more important to maximizing a party’s vote
than is persuasion; in such situations, parties allocate distributive bene-
fits primarily to individuals and groups providing key coordination and
mobilization services, rather than to vote-brokers expert at identifying
swing voters.
Another point I make is that parties are interested in votes not just
in the electoral arena but also in the legislative arena. Thus parties
may engage in “electoral targeting” (distributing benefits to optimize
342
 Swing voters, core voters, distributive politics 343

electoral outcomes), “legislative targeting” (distributing benefits to opti-

mize legislative outcomes), or a mixture of the two. Indeed, many studies
of the allocation of distributive benefits ignore the core voter versus swing
voter debate and focus entirely on legislative variables. Even in this part
of the literature, however, an analogous debate appears between those
who argue that legislative benefits flow primarily to senior figures in the
governing coalition (the analogs of core voters) and those who argue that
legislative benefits flow primarily to pivotal legislators (the analogs of
swing voters).
The core versus swing debate is important to understanding how and
whom parties represent. If parties focus exclusively on persuasion, and
hence target swing voters in the electorate and pivotal legislators in the
assembly, it is hard to see how they could be reliable agents of their core
voters. Core voters could not credibly threaten to punish their party,
because the party’s vote-maximization strategy focuses solely on persua-
sion and the core voters are by definition already persuaded. In contrast,
if parties focus primarily on coordination and mobilization, and hence
target core voters in the electorate and party members in the assembly,
there is much less tension between the goals of maximizing votes and
serving the interests of core voters.

I. The previous literature on electoral targeting

A typology of transfers
A preliminary question regarding the electoral payoff of distributing tar-
getable goods regards the nature of the exchange. There are three pure
types of transfer that appear to have been used in elections: (1) benefits
are delivered upon verification of an individual’s vote; (2) benefits are
given to a voter before s/he votes (and there is no subsequent effort to
verify how s/he voted); (3) benefits are promised upon victory of the
relevant candidate or party (again, no effort to verify individual votes).
The first type of transfer is what many think of when the word “bribe”
is used. It appears to have become less widespread after the introduction
of the secret ballot, simply because parties have a harder time verifying
their purchases when ballots are cast secretly. Nonetheless, practices such
as the Tasmanian dodge,1 squeaky voting machines, and so on can make

1 The Tasmanian dodge, invented soon after the introduction of a secret ballot in Australia
(first in Victoria, 1856), entailed party workers first securing a blank official ballot, filling
it in, and giving it to a voter. The voter then concealed the ballot, went to the polling
place and got a ballot, cast the pre-marked ballot, and returned the unmarked ballot to
344 Gary W. Cox

such bribes feasible even with putatively secret ballots. Moreover, bribes
are certainly possible when the parties themselves print and distribute
ballots, as in Argentina (cf. Stokes 2005).
The second type of transfer has typically been used to boost turnout
among known supporters. Upstate Republicans in New York used to
convey their supporters to the polls in carriages well stocked with rum
under the seat, for example. This sort of bribe is worth the money only
if turnout among likely supporters can be significantly boosted by it.
Turnout-enhancing bribery became less attractive after the introduction
of the secret ballot worsened parties’ ability to identify their supporters
(cf. Cox and Kousser 1981).
The third category consists of outcome-contingent transfers. Promis-
ing to deliver benefits if and only if one wins avoids the cost of verify-
ing either current or past voting behavior on an individual-by-individual
basis. Most of the electoral targeting models in the literature focus on
this type of transfer.

The Dixit–Londregan model

Dixit and Londregan (1995, 1996) provide a general model of how
outcome-contingent transfers are targeted, from which both the Cox–
McCubbins and the Lindbeck–Weibull models emerge as special cases.
I follow their exposition here.
Dixit and Londregan envision a left-wing party, L, and a right-wing
party, R, competing for votes (implicitly in a single-member district).
Each party k announces a vector of transfers, Tk = (Tl k , . . . , Tnk ), where
Tjk is the per capita transfer that party k promises to group j (voters
are partitioned into n groups). Promises are credible ex ante, and if
the relevant party wins honored
ex post. Party k’s transfer policy must
satisfy a budget constraint, j Nj Tj k = B, where Nj is the number of
voters

in group j. Party k chooses Tk in order to maximize its vote total,
j N P
j jk (TjL , TjR ), where Pjk (T jL ,TjR ) is the proportion of group j’s
members who will vote for party k, given the transfer promises TjL and
TjR . Although the model accommodates other possibilities, for exposi-
tional ease I shall consider the special case in which Tjk ≥ 0 for all j,
k.
To formalize Cox and McCubbins’s notion of “core support groups,”
Dixit and Londregan assume that the consumption benefit that mem-
bers of group j will actually receive, when party k promises an amount

the party worker, whereupon he was paid. The process was then repeated. The same
practice in the Philippines goes under the more evocative label of the “chain of love.”
 Swing voters, core voters, distributive politics 345

Tjk , is t j k = (1 − θ j k )Tj k . Here, θ jk ∈ [0, 1] denotes the proportion of the

subsidies that k intends to deliver to group j that will not actually reach
it. Group j is a core support group for party k when θ jk is relatively small.
As Dixit and Londregan (1996: 1134) point out, “A party’s core con-
stituencies need not prefer its issue position. It is the party’s advantage
over its competition at swaying voters in a group with offers of partic-
ularistic benefits that makes the group core.” In practice, core groups
tend also to provide solid support to their party, but it is important to
recognize that there are two distinct notions of what makes a group core.
For most of this chapter, I shall refer to voters with a strong preference
for a particular party as its “core” voters; in the next few pages, however,
a party’s core voters will be those it knows well and to whom it can more
effectively and credibly target benefits.
To provide some micro-foundations for the group response func-
tions, Pjk (TjL , TjR ), Dixit and Londregan proceed as follows. All vot-
ers in a given group are assumed to have the same income, denoted
yj for group j (so the groups can be thought of as income or occu-
pational strata). Voter h is assumed to have an innate preference
for party R, represented by a real number Xh . If voter h is in
group j, then h votes for L if U j [y j + (1 − θ j L )Tj L ] > U j [y j + (1 −
θ j R)Tj R] + Xh , and votes for R otherwise. Here, U j [y j + (1 − θ j k )Tj k ]
represents the utility that a member of group j derives from his or
her total income, yj + (1 – θ jk )Tjk . Letting j be the cumulative
distribution function of Xh in group j, Pjk (TjL , TjR ) = j [Uj [yj +
(1 – θ jL )TjL ] – Uj [yj + (1 – θ jR )TjR ]].
Dixit and Londregan show that, when the parties have no special
relationships with any groups (e.g., θ j L = θ j R = 0 for all j), the parties’
allocations are driven by the density of swing voters in each group – as
in the Lindbeck–Weibull model. As larger and larger asymmetries in the
parties’ abilities to deliver benefits arise, however, the parties’ allocations
are driven more and more by the core voter logic of promising benefits
to those groups to which the party can most effectively deliver benefits.2

Multiperiod models
Stokes (2005) and Diaz-Cayeros, Estévez, and Magaloni (2006) consider
multiperiod models of distributive politics in which promises by parties

2 Of less relevance for present purposes, the Dixit–Londregan model also predicts that
parties should target poor voters – because their votes should be cheaper to buy. Note
that poor voters’ labor – in mobilizing or coordinating others – should also be cheaper to
purchase.
346 Gary W. Cox

to deliver benefits or by voters to deliver votes are not necessarily credible.

Stokes views the voter–party exchange as a repeated prisoner’s dilemma.
In each stage (election), the voter would like to receive a bribe and
then vote for her most preferred party, while the party would like to
withhold any bribe and yet have the voter vote for it. In a single stage,
the equilibrium outcome (between a party and a voter who does not
rank the party first) would be mutual defection: no bribe for the voter
and no vote for the party. In repeated play, exchange can arise – but
the voters involved in the exchange will only be those who would not
otherwise support the party and are cheap to buy (swing voters). Diaz-
Cayeros, Estévez, and Magaloni also consider a multiperiod model but
allow voters’ predispositions for one party or another to shift, depending
on their past receipt of transfers. In their model, parties have a reason to
target core voters, since if they give them no transfers they will become
swing voters in the next election.

Multidistrict models
Distinct from the models reviewed thus far are the mostly empirical
studies that look at the allocation of distributive goods across multiple
electoral districts. The logic of targeting swing districts is particularly
compelling, because doing a bit better in a swing district can, by def-
inition, make the difference between losing and winning a seat. The
same cannot be said about swing groups in the models reviewed above.
Doing a bit better in a swing group just means that the party earns a
few more votes. As the parties in the standard single-district models are
vote-maximizers, they do not even consider how close the election is. In
contrast, the (implicit or explicit) maximand in cross-district models is
seat maximization, and so swing status is highly relevant.3
Note that whether swing districts are targeted is not particularly rel-
evant to the debate over whether swing groups are targeted. The Cox–
McCubbins, Lindbeck–Weibull, and Dixit–Londregan models address
the allocation of benefits within a single district. In a multidistrict con-
text, the core voter thesis would be that parties concentrate benefits on
their core voters within whatever districts they target, swing or otherwise.
For example, the Republicans certainly targeted resources toward the

3 McGillivray (2004) notes some structural conditions that affect the extent to which swing
districts are targeted, including district magnitude (with larger magnitudes, districts vary
less in their marginality) and party strength (weak parties cannot convince members to
run personal electoral risks for the sake of a more efficient overall campaign).
 Swing voters, core voters, distributive politics 347

swing state of Ohio in the presidential campaign of 2004. However, by

all accounts, they pushed those resources more into mobilizing their base
than into persuading swing voters. Thus the targeting of Ohio in no way
provides support for the swing voter thesis; it provides support for the
quite different swing district thesis.

Empirical evidence
The empirical studies conducted to date yield mixed results on how
much swing as opposed to core voters are targeted. Empirical studies
cited as supporting the thesis that material benefits are disproportionately
directed toward swing voters include Wright (1974), Stein and Bickers
(1994), Bickers and Stein (1996), Denemark (2000), Herron and Theo-
dos (2004), Stokes (2005), Dahlberg and Johansson (2002), and Case
(2001). However, all but the last three studies consider the allocation
of benefits across electoral districts rather than the allocation of benefits
within districts. These cross-district studies provide evidence that parties
target swing districts, but do not shed light on who gets benefits within
each district. Dahlberg and Johansson (2002) and Case (2001) study
the allocation of benefits across municipalities lying in several different
electoral districts, while Stokes (2005) examines the allocation of ben-
efits to individual voters. These studies thus come closer to testing the
vote-maximizing models considered thus far.4
Empirical studies supporting the thesis that material benefits are dis-
proportionately directed toward core voters include Ansolabehere and
Snyder (2003), Levitt and Snyder (1995), Balla, Lawrence, Maltzman,
and Sigelman (2002), Diaz-Cayeros, Magaloni, and Weingast (2000),
Calvo and Murillo (2004), Bickers and Stein (2000), and Chen (2008).
Most of these studies, too, examine the allocation of benefits across elec-
toral districts. Thus they too fail to provide much evidence pertinent to

4 Although the study has many strengths, there is reason to doubt that Dahlberg and
Johansson (2002) provide an appropriate test of the swing voter and core voter mod-
els. The main problem is that the program they study had a strong programmatic
content which appealed to core interests within the Social Democratic Party, whereas
the models all assume completely policy-neutral transfers of funds. As Dahlberg and
Johansson (ibid.: 27) note, the grants they investigate were “intended to support . . . local
investment programs aimed at an ecological sustainable development and at increasing
municipal employment.” It is possible that the primary beneficiaries of these grants were
Greens and public-sector union locals in each municipality, regardless of what the overall
Socialist percentage was in the municipality. Violation of the assumption that funds are
“policy-neutral” is a general problem with using data on spending that has programmatic
content.
348 Gary W. Cox

the Cox–McCubbins, Lindbeck–Weibull, and Dixit–Londregan models.

When such studies find that distributive benefits flow to the strongholds
of the governing party or parties, rather than to swing districts, they may
be discovering that governments are sometimes probability-of-majority
maximizers rather than vote-maximizers: if a government allocates bene-
fits to maximize its probability of retaining a majority, then it may some-
times decide that retaining a vice-like grip on a small majority is the
optimal strategy. Alternatively, the targeting of benefits to government
strongholds may reflect a purely legislative, rather than electoral, logic
(on which more below).

II. Extending the electoral targeting model(s)

Two key aspects of the Cox–McCubbins, Lindbeck–Weibull, Dixit–
Londregan, Stokes, and Diaz-Cayeros–Estévez–Magaloni models are: (a)
all voters vote; and (b) the number of parties is exogenously fixed at two.
The first assumption puts aside all issues of mobilization and turnout.
The second assumption puts aside all issues of coordination. Thus, in all
current electoral targeting models, parties can increase their vote shares
only by persuasion, and the means by which they can persuade voters is
limited to offering transfers (there is no endogenous spatial competition,
for example).
Let us consider the terms coordination, mobilization and persuasion
more carefully, following Cox (2005). We shall consider what actions L,
the left-wing party, can take to increase its total vote. The probability that
voter h votes for party L depends on whether h participates in the election;
whether L is the only left-wing party on the ballot; and whether h prefers
L to R. Efforts to ensure that voter h actually participates in the election –
a necessary but not sufficient condition for h to support L – fall under the
heading of mobilization. Efforts to ensure that there is just one left-wing
party on the ballot fall under the heading of coordination. Finally, efforts
to ensure that voter h prefers L to R fall under the heading of persuasion.
In general, parties can persuade by repositioning themselves along the
left–right spectrum, by convincing voters that they are more competent
at providing valence goods, by slinging mud at their opponents, and so
forth. In the models under consideration here, the method of persuasion
is offering transfers.
To generalize the Dixit–Londregan model, we could rewrite the pro-
portion of group j’s members voting for party L as PjL (tL , tR ) = Qj (tL ,
tR )SjL (tL , tR )/Mj (tL , tR ), where Qj (tL , tR ) is the turnout rate in group j,
SjL (tL , tR ) is the proportion of group j’s participating members who prefer
L to R, and Mj (tL , tR ) is the number of left-wing parties that campaign
 Swing voters, core voters, distributive politics 349

for votes among group j’s voters. If Mj (tL , tR ) = 1, then L gets votes from
all citizens who participate and prefer it to the right-wing alternative, R.
If Mj (tL , tR ) = 2, then L gets half the votes of citizens who participate and
prefer it to R; implicitly, the new entrant on the left is equally attractive
to left-wing voters, and vote coordination fails utterly on the left. This
particular functional form embodies the largest possible losses due to
miscoordination. A more flexible formulation would be that L receives a
fraction of the left-wing vote, with that fraction declining as the number
of left-wing parties increases.5
What happens if one extends the model(s) in the literature so that
the transfers party L offers to group j affect not just how many voters
in that group prefer L to R but also how many left-wing competitors L
faces within group j and how many citizens in group j participate in the
election?
Let us consider adding mobilization and turnout first. The transfers
offered in the current models affect only the vote choices made by vot-
ers, rather than their participation decisions. If turnout is invariant with
respect to transfers, then transfers should indeed go to groups where the
marginal persuasive effect will be greatest. Lindbeck and Weibull (1987)
and Cox and McCubbins (1986) disagree about which groups will yield
the biggest persuasion bang for the transfer buck, but they agree on
focusing just on persuasion.
What if voters’ preferences are immutable but their turnout decision
can be affected by transfers? Suppose, for example, the electorate is highly
polarized (almost no swing voters in any group) but not everyone has a
high probability of participating. In this case, each party should clearly
target transfers to its core support groups. The most valuable voter for
a party to offer a transfer to is one with a high probability of voting for
that party, conditional on participating; and a probability of participation
that is highly responsive to transfers. Thus, the party should target core
support groups (in the sense of groups with large proportions of members
who strongly prefer L to R) with malleable turnout rates.
Now let us consider adding coordination. The transfers offered in the
current models do not affect the decisions of political entrepreneurs or
young Turks to start a rival party which caters to the same core groups
as one of the existing duopolists. If there is some chance that one of the
duopolists might face competition on its home turf, then transfers to
the core groups have another value, as they presumably help to depress
the probability of internecine competition.

5 Another approach, in which both persuasion and mobilization – but not coordination –
are brought into analytic view, is that of Bartels (1998).
350 Gary W. Cox

Do the claims just made about the importance of coordination and

mobilization founder on the canonical swing voter objection, quoted at
the outset of the chapter, that core voters “cannot credibly threaten to
punish their favored party if it withholds [distributive] benefits” (Stokes
2005: 317)? First, the explicit or implicit threat of sitting the election out
is credible, because the individuals issuing the threat must bear private
costs to participate in the election. It is true that their abstention raises the
probability that the other (and dispreferred) party will win the election;
but the change in the probability is negligible. Thus, on an individual
calculus, it is rational to abstain and a threat to abstain is by no means
empty. Second, the explicit or implicit threat of running a rival candidate
in the election can also be credible if the group threatening to split has
some chance of forcing a re-coordination of the party system and emerg-
ing as the dominant party in its ideological niche. Thus, although only
occasionally manifest, threats by core groups to sit the election out or to
launch new parties are always possible. Indeed, the better way to view
it is that coordination is a full-time job and so is mobilization, so that
the party needs full-time “employees” or “consultants” working on these
problems, and that these agents then receive a regular and large flow of
distributive benefits as compensation for their services.
In a combined model, in which transfers can increase a party’s vote
share, either due to persuasion or to mobilization or to coordination, one
should find: (1) the less persuasion a party thinks possible, and the more
mobilization it thinks possible, the more it concentrates its transfers on its
core supporters; (2) the less persuasion a party thinks possible, and the
more its leaders are worried about splintering, the more it concentrates
its transfers on its core supporters; (3) the less able a party is to deliver
credible promises to swing groups, the more it concentrates on its core
groups.

III. Maximizing votes in the electoral and legislative arenas

In this section, I argue that the debate over whether parties target swing
voters or core voters is similar to the debate in legislative studies between
those who view pivotal legislators as the primary determinants of leg-
islative outcomes (Krehbiel 1998) and those who view senior majority-
party legislators as the primary determinants (Cox and McCubbins 1993,
2005; Aldrich and Rohde 2001; Kim 2006).
First, note that swing voters and pivotal legislators share a crucial
similarity. It is true that the Lindbeck–Weibull model is distributive,
while Krehbiel’s model is spatial. Nonetheless, both swing voters and
pivotal legislators end up being indifferent in equilibrium and thus cheap
 Swing voters, core voters, distributive politics 351

to buy. Swing voters are by definition indifferent on ideological grounds.

Pivotal legislators are those whose support may be crucial in overcoming
super-majority requirements in the legislative process, such as invoking
cloture in the US Senate. Sometimes securing sixty votes in the Senate
to invoke cloture is not a problem – when the status quo is far enough
away from the legislative median. In this case, the pivotal legislators are
not indifferent. When cloture (or other super-majority) requirements do
bind, however, then the pivotal legislators are in equilibrium indifferent
between passage and rejection of a bill – and are thus cheap to buy with
side payments. As with swing voter models, Krehbiel’s pivotal legislator
model assumes away both any need for mobilization (all legislators vote)
and any need for coordination (there are no fights over which bills and
amendments will be considered), leaving parties with only persuasion as
a strategy.
Second, note that core voters and majority-party legislators share some
similarities. The party knows the preferences of core voters and its own
legislators better than it knows the preferences of swing voters or inde-
pendent legislators. Moreover, the party may look to core voters and its
own legislators to help coordinate electoral candidacies or the legislative
agenda; and to help mobilize voters or whip legislators.
Cox and McCubbins (2005) argue that legislative parties are not pri-
marily mechanisms to purchase votes on the floor. Rather, they are pri-
marily coordinating and mobilizing devices: they distribute benefits to
members primarily to buy their agenda-setting (coordinating) and whip-
ping (mobilizing) services, and only secondarily to buy their votes (per-
suading). I would argue that electoral parties are similar. Their main
purpose is not to buy votes on the spot market on the day of the elec-
tion, or otherwise persuade during the election. Rather they are primarily
coordinating (Cox 1997) and mobilizing (Cox 1999) devices.
As evidence for this view, consider the allocation of offices – e.g. com-
mittee chairs in the legislative arena or postal sub-masterships in the
electoral arena. Offices are the largest private benefits that parties dis-
tribute. In the legislature, they go almost exclusively to members of the
party (i.e. to its “core voters”); and the recipients are being “paid” mostly
for their coordinating and mobilizing services. In the electorate, the con-
sistent finding is that political machines allocate patronage jobs almost
exclusively to core supporters, who are then expected not simply to vote
for the party but also – and more importantly – to coordinate (prevent
rivals from emerging) and mobilize (get out the vote).6

6 See e.g. Rakove (1975), Cox and McCubbins (1986), Calvo and Murillo (2004).
352 Gary W. Cox

IV. Models of legislative targeting

When investigating the allocation of distributive benefits to multiple dis-
tricts, one might argue that benefits are not targeted solely in order to
optimize electoral outcomes but also – or even instead – in order to opti-
mize legislative outcomes. Indeed, many studies seeking to explain the
distribution of targetable government expenditures across districts focus
on legislative considerations. They examine whether districts represented
by powerful senior figures, especially those in the government, get more
pork (e.g. Levitt and Poterba 1999; Denemark 2000; Golden and Picci
2006), whether districts represented by members of the relevant com-
mittee(s) get more pork (e.g. Hird 1991; Alvarez and Saving 1997), and
so on. Occasionally, such legislatively focused studies will include a vari-
able measuring how marginal a particular legislator is, on the theory that
more marginal incumbents should work harder to bring home benefits
to their districts. But the model is one in which individual legislators
pursue benefits in order to maximize their own electoral chances, not
one in which a party allocates benefits in order to maximize votes.
Other scholars have suggested that parties play a more consequential
role in the allocation of legislative side payments. Cox and McCubbins
(1993, appendix 1), Evans (2004), and Kim (2006) argue that distribu-
tive benefits are used as side payments to clinch broader legislative deals.
Kim stresses that these are payments not so much to buy pivotal votes
on the floor – the legislative analog of swing voters – but rather to pay
committee chairs for their agenda-setting services – the legislative analog
of electoral coordination. For example, Claude Pepper, then chair of a
key committee, received numerous distributive benefits for his district in
the “transition rules” attached to the Tax Reform Act of 1986. Ideologi-
cally, Pepper was nowhere near being pivotal on the floor (indeed, the bill
passed by a large margin) and thus he did not need to be paid to vote for
the bill. Rather, he was collecting his customary “toll” for allowing the
bill to pass through his committee and consume scarce (and valuable)
committee time (Kim 2006). In other words, he could slow down the bill
or speed it up and the proponents were willing to pay for the latter rather
than the former.
Another model of legislative targeting, which corresponds better to the
swing voter model in the electoral arena, focuses on the side payments
used by Speakers in the US House when they arrange “vest pocket” votes.
As explained most clearly by King and Zeckhauser (2003), Speakers
regularly negotiate “vote option contracts” with certain members before
a close vote. The contract gives the Speaker the option of calling on
the member to change his or her vote in exchange for a consideration
 Swing voters, core voters, distributive politics 353

the two have agreed upon. The important things to note about the side
payments conveyed via vest pocket arrangements is that (a) they go to
swing legislators; but (b) they are small potatoes compared to the flow of
pork controlled by committee actors.
My interpretation of results that show powerful individual legislators
getting more benefits than would seem justified on a vote-maximizing
calculation is that parties need both electoral and legislative votes; and
they pay those skilled at coordinating and mobilizing votes in both are-
nas. Thus, powerful legislators receive additional benefits to pay them for
their legislative services. I am not arguing that all parties in all contexts
perfectly control the allocation of distributive benefits and trade off elec-
toral vote-maximizing against legislative vote-maximizing at the optimal
rate. I am just pointing out that, if a party is a long-lived entity with
both electoral and legislative goals, then one cannot take the electorally
irrational bonuses that senior figures get – e.g. the notorious “bridges to
nowhere” in Alaska or Japan – as convincing evidence that the party as a
whole is weak or poorly organized.

V. The role of coordination and mobilization

in spatial models
As explained above, in electoral targeting models sans mobilization and
coordination, parties are more likely to target swing voters; but when
mobilization and coordination are brought into the analysis, parties cater
more to their core groups. A similar pattern emerges in spatial models.
The pure Downsian model – which ignores coordination (the number
of parties is exogenously fixed at two) and mobilization (all voters vote) –
predicts that both parties will adopt a platform that coincides with the
ideological position of the median voter. They both “target” the median
voter. When voters can abstain (e.g. Smithies 1941), the parties may not
converge, because when they contemplate a move toward the center, they
anticipate more vote losses due to abstention on their extreme wing than
vote gains from centrists. When the number of parties is not fixed, one
can add another reason for non-convergence: each party’s worry that it
might be outflanked by a splinter or new party if it moves toward the
center (Palfrey 1984).
Spatial and distributive models both focus on persuasion. They differ
in their conception of the tools that parties use to persuade voters. In spa-
tial models, parties announce a platform of policies that they will enact if
elected. In distributive models, parties announce a package of transfers
that they will implement if elected. Regardless of the technology of per-
suasion, the necessity of coordinating and mobilizing induces parties to
354 Gary W. Cox

cater more to their core groups – either in the sense of advocating policies
that appeal to the core groups or in the sense of advocating transfers that
appeal to the core groups.

VI. Conclusion
In this chapter, I have considered how vote-maximizing parties might
allocate benefits within a district – the purview of the Dixit–Londregan
and other models. The first main point I have urged is simply that the
extant models of electoral targeting focus too narrowly on a single vote-
getting strategy – offering transfers to voters who have already decided to
participate in an election with an exogenously fixed number of competi-
tors in order to persuade them to vote for a particular party (persuasion).
It is conceptually easy to extend these models to include other vote-
getting strategies – in particular, offering transfers to voters or groups,
in order to pay them for either (1) their efforts in mobilizing support for
the party; or (2) their efforts in coordinating the menu of choices that
appears on the ballot. I argue that empirically the bulk of distributive
benefits should flow to those who are crucial in lowering the number of
ideologically similar competitors a party faces on the ballot and to those
who are crucial in getting out the vote. Experts at buying votes on the
spot market on the day of the election can sometimes be important, too,
but need not be. Since the key agents in coordination and mobilization
are typically the leaders of core groups within the party, I expect that
distributive benefits should flow to core groups and their members.
The second point I have urged is that the parallel literature on legisla-
tive targeting – in which allocations of benefits across electoral districts
are explained in terms of the “clout” that individual legislators wield
by virtue of their committee positions, leadership positions, or majority
status – also should (and implicitly does) pay attention to coordination
and mobilization, in addition to persuasion. The legislative analog of
Lindbeck and Weibull’s swing voter hypothesis would be Krehbiel’s piv-
otal politics thesis, according to which all the action in side payments
should center on legislators whose votes are (or will be) pivotal to the
outcome on the floor. The argument that benefits should be targeted
to “core groups” within the legislature – that is, the majority party’s
senior figures – relies on points similar to those made above about max-
imizing votes in the electoral arena: senior party figures are crucial in
setting the agenda (coordination) and whipping (mobilization), and a
vote-maximizing party should certainly wish to pay for these important
services, in addition to buying pivotal votes on the floor when needed.
 Swing voters, core voters, distributive politics 355

Finally, I have noted that the necessity of coordination and mobilization

drives parties’ spatial positions away from the center to more robust left
and right positions; and drives their distributive strategies away from
swing voters to core voters. These are abstractly similar responses, in
that in both cases more benefits flow to the party’s base, as a recognition
of its crucial role in maximizing votes.

BIBLIOGRAPHY
Aldrich, John, and David Rohde. 2001. “The Logic of Conditional Party Gov-
ernment,” in Congress Reconsidered, ed. Lawrence Dodd and Bruce Oppen-
heimer, 7th edn. Washington DC: Congressional Quarterly Press, 26–92.
Alvarez, R. Michael, and Jason L. Saving. 1997. “Congressional Committees
and the Political Economy of Federal Outlays.” Public Choice 92: 55–73.
Ansolabehere, S. and J. M. Snyder. 2003. “Party Control of State Government
and the Distribution of Public Expenditures.” Unpublished paper, Depart-
ments of Political Science and Economics, Massachusetts Institute of Tech-
nology.
Balla, S. J., E. D. Lawrence, F. Maltzman, and L. Sigelman. 2002. “Partisan-
ship, Blame Avoidance, and the Distribution of Legislative Pork.” American
Journal of Political Science 46: 515–25.
Bartels, Larry. 1998. “Where the Ducks Are: Voting Power in a Party System,”
in Politicians and Party Politics, ed. John G. Geer. Baltimore, MD: Johns
Hopkins University Press, 43–79.
Bickers, K. N. and R. M. Stein. 1996. “The Electoral Dynamics of the Federal
Pork Barrel.” American Journal of Political Science 40: 1300–25.
2000. “The Congressional Pork Barrel in a Republican Era.” Journal of Politics
62: 1070–86.
Calvo, E. and M. V. Murillo. 2004. “Who Delivers? Partisan Clients in the
Argentine Electoral Market.” American Journal of Political Science 48: 742–
57.
Case, A. 2001. “Election Goals and Income Redistribution: Recent Evidence
from Albania.” European Economic Review 45: 405–23.
Chen, Jowei. 2008. “Republican Vote Buying in the 2004 US Presidential Elec-
tion.” Unpublished typescript, Stanford University.
Cox, Gary W. 1997. Making Votes Count. Cambridge: Cambridge University
Press.
1999. “Electoral Rules and the Calculus of Mobilization.” Legislative Studies
Quarterly 24: 387–420.
2005. “Electoral Institutions and Political Competition: Coordination, Per-
suasion and Mobilization,” in Handbook of New Institutional Economics, ed.
Claude Menard and Mary M. Shirley. Dordrecht: Springer, 141–61.
Cox, Gary W. and J. Morgan Kousser. 1981. “Turnout and Rural Corruption:
New York as a Test Case.” American Journal of Political Science 25: 646–63.
Cox, Gary W. and Mathew D. McCubbins. 1986. “Electoral Politics as a Redis-
tributive Game.” Journal of Politics 48: 370–89.
356 Gary W. Cox

1993. Legislative Leviathan: Party Government in the House. Berkeley: University

of California Press.
2005. Setting the Agenda: Responsible Party Government in the U.S. House of
Representatives. Cambridge: Cambridge University Press.
Cox, Gary W., and Michael Munger. 1989. “Closeness, Expenditure, Turnout:
The 1982 U.S. House Elections.” American Political Science Review 83: 217–
32.
Dahlberg, M. and E. Johansson. 2002. “On the Vote-Purchasing Behavior of
Incumbent Governments.” American Political Science Review 96: 27–40.
Denemark, David. 2000. “Partisan Pork Barrel in Parliamentary Systems: Aus-
tralian Constituency-Level Grants.” Journal of Politics 62: 896–915.
Diaz-Cayeros, Alberto, Federico Estévez, and Beatriz Magaloni. 2006. “Poverty
and Vote Buying: The Politics of Social Transfers in Mexico, 1989–2006.”
Unpublished book manuscript. Stanford University.
Diaz-Cayeros, Alberto, B. Magaloni, and B. Weingast. 2000. “Federalism and
democratization in Mexico.” Unpublished paper, Stanford University.
Dixit, A. and J. Londregan. 1995. “Redistributive Politics and Economic Effi-
ciency.” American Political Science Review 89: 856–66.
1996. “The Determinants of Success of Special Interests in Redistributive
Politics.” Journal of Politics 58: 1132–55.
Drazen, A. 2000. Political Economy in Macroeconomics. Princeton, NJ: Princeton
University Press.
Evans, Diana. 2004. Greasing the Wheels: Using Pork Barrel Projects to Build Major-
ity Coalitions in Congress. Cambridge: Cambridge University Press.
Golden, Miriam, and Lucio Picci. 2006. “Pork Barrel Politics in Postwar Italy,
1953–1994.” American Journal of Political Science 52: 268–89.
Hall, Richard. 1996. Participation in Congress. New Haven, CT: Yale University
Press.
Herron, M. C. and B. A. Theodos. 2004. “Government Redistribution in the
Shadow of Legislative Elections: A Study of the Illinois Member Initiative
Grants Program.” Legislative Studies Quarterly 29: 287–311.
Hird, John A. 1991. “The Political Economy of Pork: Project Selection at the
U.S. Army Corps of Engineers.” American Political Science Review 85: 429–
56.
Kim, Henry. 2006. “Agenda Setting, Clout, and Logrolls in the U.S. House.”
Unpublished typescript, University of California, San Diego.
King, David, and Richard Zeckhauser. 2003. “Congressional Vote Options.”
Legislative Studies Quarterly 28: 387–411.
Krehbiel, Keith. 1998. Pivotal Politics. Chicago: University of Chicago Press.
Levitt, S. D. and M. Poterba. 1999. “Congressional Distributive Politics and
State Economic Performance.” Public Choice 99: 185–216.
Levitt, S. D. and James M. Snyder. 1995. “Political Parties and the Distribution
of Federal Outlays.” American Journal of Political Science 39: 958–80.
Lindbeck, A. and J. Weibull. 1987. “Balanced Budget Redistribution and the
Outcome of Political Competition.” Public Choice 52: 273–97.
McGillivray, F. 2004. Privileging Industry: The Comparative Politics of Trade and
Industrial Policy. Princeton, NJ: Princeton University Press.
 Swing voters, core voters, distributive politics 357

Palfrey, Thomas. 1984. “Spatial Equilibrium with Entry.” Review of Economic

Studies 51: 139–56.
Rakove, Milton. 1975. Don’t Make No Waves, Don’t Back No Losers. Bloomington:
Indiana University Press.
Smithies, Arthur. 1941. “Optimum Location in Spatial Competition.” Journal of
Political Economy 49: 423–39.
Stein, R. M. and K. N. Bickers. 1994. “Congressional Elections and the Pork
Barrel.” Journal of Politics 56: 377–99.
Stokes, S. C. 2005. “Perverse Accountability: A Formal Model of Machine Pol-
itics with Evidence from Argentina.” American Political Science Review 99:
315–25.
Wright, G. 1974. “The Political Economy of New Deal Spending: As Econo-
metric Analysis.” Review of Economics and Statistics 56: 30–8.
 Index

accountability, 212, 232, 278, 279, 337 aristocratic thesis, 9, 92, 96, 271–301
legal, 279, 280 Aristotle, 260, 271, 273
political, 280 concept of democracy, 16, 288
and representation, 44, 225 elections, 290, 295–6
addresses, 3, 35 elections and lottery, 296
quantity of, 45 leadership, 289
royalist addresses, 52–3 rule by the poor, 272
statistics about, 41 Athenian government, 272, 280, 285, 288
use of, 40 accountability, 276, 279
advocacy, 114–16 citizenship, 295
Africa, 211–18 direct democracy, 276
electoral systems in, 220–5 institutional features of democracy, 291
ethnicity in, 222 lottery system, 67
formal inequalities in, 230 Atoka Agreement, 185
impact of quotas for women, 211–18, Austen-Smith, David, 293
230 authority, 25
political parties, 225, 232 and formal institutions, 52
women’s movements, 228, 231 Hobbes’s conception of, 15
women’s representation in, 217 authorization, 15, 22, 74, 278
African-Americans, 119, 122, 130, 188 by election, 278
agency, 9, 76, 125, 145, 296 Hobbes’s theory of, 17, 18–26
autonomy of agents, 275, 281, 283 and representation, 17, 25, 28
incentives for agents, 273, 283 Azari, Julia, 11, 324–40
and interests, 281
political, 282–6 Babcock, Frederick, 122
problem of, 273, 293 Baldus de Ubaldis, 76
restriction on the agency contract, 285 Banaszak, Lee Ann, 229
Alaska, 169 Banks, Jeffrey, 293
Albericus de Rosciate, 76 bargaining, 313
Alcibiades, 291 Bartels, Larry M., 90, 301, 314, 322
Alesina, A., 312 Bartolus of Sassoforrato, 76
American Indian Movement, 205 Beale, Edward F., 192
American Indian Policy Review Beckwith, Karen, 229
Commission, 206 Bee, Robert L., 202
American National Election Studies, 315 Besigye, Kiiza, 224
American War of Independence, 67 Bilgrami, Akeel, 144
anti-slavery petitions, 41 Black, Justice Hugo, 169
anti-solidarity effect, 10, 312, 316, 317 Bloomberg, Michael, 325, 326–7
Aristedes, 280 Board of Indian Commissioners, 202
aristocracy, 9 Botswana, 221
aristocratic rule, and representative Bradley, James, 44
government, 286–91 bribes, 343

358
 Index 359

Britain collective action problems, 117, 118, 238,

accountability of parliament, 44 255, 283
debate about popular representation, collective identity, 29, 31, 33, 219
48–50 of the state, 21, 26, 28, 33
electoral system, 36 Collier, John, 199–202
group-level representation of the people, commonwealth, 19, 22, 23
85 competition, 274, 277, 284, 301, 313, 337
instructions to members of parliament, Constant, Benjamin, 5, 91, 92, 98–107,
50, 51 259
petitions and addresses in, 35 and Rousseau, 99
representative authority in, 52, 53 usurpation, 100
representative system, 44, 96 constitutional assemblies, 30, 103
sedition bills, 42 constitutions, 30, 98, 102, 286
Westminster model of representation, 4, constructivism, 143, 144, 146
63–4, 84–5 contestation, 6, 80, 124–7
British Columbia’s citizens’ assembly, 65, value of, 104, 105, 120, 125, 127, 128
68, 69, 73, 77 Contract with America, 286
Bureau of Indian Affairs, 188, 195–202 cooperation, 62, 63–4, 81
Courts of Indian Offences, 195–7 coordination, 12, 342, 353–4
Indian Organization Division, 200 corporations, urban, 36
Indian police, 195–7 Cox, Gary W., 11–12, 342–55
Burke, Edmund, 1, 12, 51, 84, 92 Cox–McCubbins model of electoral
Byanyima, Winnie, 224 targeting, 348
Byrd, Colonel, 179 Creek nation, 180, 185
Creevey, Lucy, 223
Carens, Joseph, 151 Crow Dog, 170–1, 172
Case, A., 347 cultural commitments
Castile, George Pierre, 202 privatization of, 141, 142
charters, royal, 36 protection of, 141
checks and balances, 98, 100, 103, 107, cultural groups, 147, 150, 152
273 exemption from democratic politics,
Cherokee nation, 176, 184, 185, 196 143
legal status of, 165 membership in the state, 144, 145
Cherokee Nation v. Georgia, 163–5 rights of, 147
Chickasaw nation, 184, 185
Choctaw nation, 180, 184, 185 Dahl, Robert, 91, 126, 275
Christiano, Thomas, 73 Dahlberg, M., 347
citizens Dahlerup, Drude, 211, 216
as a collective force, 38 Dawes Act, 181–7
and governments, 32 Dawes Commission, 182, 183–4, 185
judgment of governments, 28 debates during election campaigns, 104,
membership rights, 26 339
citizens’ assembly, British Columbia, 65, data and methodology of investigation
68, 69, 73, 77 of, 327–32
citizens’ juries, 68 impact on public opinion, 324–32
citizenship, 139, 152, 257, 258 informative potential of, 324
culture of, 37 and opinion change, 326, 332–9
multicultural, 140, 141, 150 purpose of, 339
property requirement for, 295 Defoe, Daniel, 50
civil rights disputes, United States, 79, Delano, Columbus, 189
81 Delaware nation, 176
civil society, 18, 227, 228 deliberation, 116, 238
civilization, 165, 187 deliberative opinion polls, 68
classification of humans, 147 Deloria, Vine Jr., 187–8
Cohen, Felix, 172 demagogues, 5, 102
360 Index

democracy, 216, 232 coordination, 348, 349, 350, 353–4

aggregative, 6, 118, 121 Cox–McCubbins model, 346, 348
and interests, 118–20, 121, 126 Dixit–Londregan model, 344–5, 346,
ancient Greek view of, 2 348, 354
contestation, 128–31 generalization of, 348
and debate, 79 empirical evidence on, 347–8
definition of, 275–8 extended models, 348–50
deliberative, 6, 104, 118, 121, 125 Lindbeck–Weibull model, 346, 348,
and interests, 118–20, 121, 126 350, 354
and difference, 142 literature on, 343–8
direct, 9, 271, 291–5, 300 mobilization, 348, 349, 350, 353–4
hostility toward, 37 models of, 345–7, 353–4
impact of televised debates on quality of, turnout, 348, 349
324 typology of transfers, 343–4
indirect, 9 The Electors’ Right Asserted, 50
liberal, 98–107 Elk v. Wilkins, 204
measuring the performance of, 90 England’s Enemies Exposed, 48, 52
and multiculturalism, 147, 153 equality, 9, 80, 81, 151, 152, 256–60, 275
political culture of, 81 equilibrium, political, 312–13, 321
procedural, 142, 225–7 party unanimity Nash equilibrium, 312,
and representation, 1, 15–18, 26–31, 318
77–82 Erikson, Robert S., 90
representative, 26–31, 291–5 Estévez, Federico, 348
role of information in, 338 ethnic groups, 139, 146, 156
and self-representation, 78–82 in Africa, 222
and statutory officers and bodies, 77 creation of, 6
theory of, 124–7 and identity, 222
transition to, 214, 220 legitimacy and scope of claims of, 145,
and women’s inclusion, 214, 217 147, 148, 149
Democratic Party, constituency of, 308 Evans, Diana, 352
democratization, 139, 211–18 executive, 102
despotism, 103, 128
Diaz-Cayeros, Alberto, 348 Federal Housing Administration, 123
discursive dilemma, 75–6, 78 feminism, 213, 226
Dixit–Londregan model of electoral Ferejohn, John, 9–10, 271–301
targeting, 344–5, 346, 348, 354 Ferrer, Fernando, 325, 326
“doctrine of discovery,” 162, 165 Findlay, John M., 192–3
dominion, 23, 24 First Nations, 148; see also Native
Downs, Anthony, 10, 291–2, 353 Americans
Drake, Jemes, 50 Five Civilized Tribes, 177, 178, 179, 185,
Dred Scott case, 202 196
Dworkin, Ronald, 257 citizenship, 203
constitutional convention, 185
educational resources, racial disparity in exemption from federal laws, 181
access to, 120, 123 legal status of, 202
elections, 98, 226, 300 legislatures of, 185
determinants of results, 309 national governments of, 185
and lotteries, 295–300 and the Oklahoma Territory, 185
quality of candidates, 291, 309 Foner, Eric, 194, 195
spatial model of, 310–21 Fontaine, Peter, 179
electoral colleges, 101 Ford, Richard Thompson, 130
electoral constituencies, definition of, 241 France, parité laws, 240
electoral parties, 351 franchise, 40
electoral qualifications, 8–9, 236–61 freedom, 94, 125, 129
electoral systems, 221 and culture, 142
electoral targeting, 342 of indigenous people, 148
 Index 361

French Revolution, 42, 67, 92, ethnic identity, 143, 144

98 and ethnicity, 222
formation of, 146, 150
Garsten, Bryan, 5, 90, 109 group identity, 154, 223, 242
gender equality, 214, 232 politics of, 154
Gerber, Alan, 324, 326, 340 identity politics, 151, 154, 156
Gilens, Martin, 301 impeachment, 279, 280, 284
Glaeser, E., 312 incorporation, 63–4
Govender, Pregs, 225 India, 223
government, 32 Indian Territory, 177, 178, 181
legitimate, 126 commercial crops in, 183
limited, 99 employment in, 183
mixed, 273 natural resources of, 183
multivocal forms of, 5 poor white residents of, 184
Grand Petition, 41 railways in, 183, 194
Grant, President Ulysses S., 189, 194 Indians: see Native Americans
Green, Donald, 324, 326, 340 indigenous people, 149
group identity, 154, 223, 242 initiatives, 10, 294
groups injustice
constructed character of, 152 historical, 146, 149
cooperative, 62 inherited, 6–8
cultural differences of, 139 structural, 7
identity of, 30 institutional design, 100, 214
personality of, 29 institutions, 5, 6, 7–8, 112, 220–5
and political interests, 116 Interamerican Indigenista Institute, 199
structural origins of, 139 interests, 275
articulation of, 227–32
Hallowell v. United States, 204 constructivist conception of, 124–31
Hamilton, Alexander, 5, 102–3, 161 economic, 307
Harring, Sidney L., 195, 197, 198 formation of, 5, 6–7
Hassim, Shireen, 8, 211–33 and group membership, 116
Hayward, Clarissa Rile, 5–6, 111–32 institutional mechanisms and, 116
Helms, Jesse, 117 and policy, 276
hierarchical societies, 115, 117, 128, 132, political construction of, 118–24
223 and quotas, 227–32
Hirschman, A. O., 308, 322 role of, 111–29
The History of the Last Parliament, 50 and structural inequalities, 112
Hobbes, Thomas, 2, 16, 18–26 International IDEA, 215
dominion, 23, 24 Ireland, 190–1
influence of, 17 Italian city republics, 67, 70
Leviathan, 15, 16, 17, 18, 20, 22, 105
membership, 23–4 Jefferson, Thomas, 103, 188
personation, 22–3, 24, 26, 74, 79 Johansson, E., 347
theory of representation, 15, 20, 35, 61, Johnson, Kenneth W., 161
76, 78, 106 Johnson v. M’Intosh, 162–3
Home Owners Loan Corporation, 123 Jones, Commissioner William A., 197
Hopewell Treaty, 168, 176 Jung, Courtney, 6–7, 139–57
human rights law, 153, 214 Jura Populi Anglicani, 48
Jura Populi . . . Answer’d, 49
Ickes, Harold, 201 jury system, 67
identities, formation of, 6–7 justice, 255, 265, 305
identity, 140 compensatory, 150
collective, 31, 33, 219 distributive, 146
of the state, 21, 26, 28, 33 and legitimacy, 264
constructivist theories of, 146, 150 political conception of, 81
and culture, 143 substantive, 238
362 Index

Kagame, P., 225 policy outcomes, 116, 295

Kant, Immanuel, 272, 281, 287 public opinion, 104
Kazibwe, Speciosa, 231 representation, 91, 102, 272, 273, 281,
Kennedy, Senator Ted, 241 286
Kent petition, 44, 48, 50, 52 Magaloni, Beatriz, 348
Keyes, Captain E. D., 193 majority principle, 53, 95, 142
Kim, Henry, 352 Mamdani, Mahmood, 7–8, 159–89
King, David, 352 Manin, Bernard, 54, 111, 260
Kirshner, Alexander, 1–12 electoral representation, 35, 248
Kislansky, Mark, 274 leadership, 289–90
Klamath nation, 201 lottery system, 295
Knights, Mark, 3, 35 qualifications for office, 260
Krehbiel, Keith, 350, 351, 354 right to run for office, 259
Krook, Mona Lena, 219 Mansbridge, Jane, 117, 215, 218, 227,
Kymlicka, Will, 140–1, 148 243
Markell, Patchen, 125
Laffer, Arthur, 306 Marshall, Chief Justice John, 161–7, 174–5
Lawson (of the Carolinas), 179 Cherokee Nation v. Georgia, 163–5
leadership, 277, 289 “doctrine of discovery,” 162, 166
and agency, 285 Johnson v. M’Intosh, 162–3
and policy, 291 political power, 175
Lee, W., 310 relationship between ward and guardian,
Legion Memorial petition, 44, 48, 50, 163
52 title by conquest, 174
legislative parties, 351 Worcester v. Georgia, 165–7
legislative presence, 241, 242, 251 Mathews, Justice, 171, 172
legislative targeting, 352–3, 354 Mauritius, 221
legitimacy, 111–29, 223, 252 Mayhew, David, 108
and descriptive representation, 112–18 McCubbins, Mathew D., 338, 339, 342,
and justice, 264 346
and qualifications for office, 256 McRae, Kenneth D., 195
sources of, 37 media, 105
Leslie, Charles, 50 median voter theorem, 291, 292, 294, 297,
Levelers, 40 299, 353
liberalism, 92, 98–107, 139–53 Medill, William, 191
Lincoln, Abraham, 194 Mekeel, Scudder, 200
Lindbeck, A., 342 membership, 23–4, 139
Lindbeck–Weibull model of electoral cultural group, 140, 148
targeting, 345, 346, 348, 350, 354 rights, 7, 26, 151–6
Lindert, P., 307–8 memory, political significance of, 149, 150
Locke, John, 78 Menominee nation, 173, 201
Lone Wolf, 171 Menominee Tribe of Indians v. United States,
Longman, Timothy, 223, 224 173
lotteries, and elections, 295–300 Michels, Robert, 289
Lupia, Arthur, 338, 339 Mill, James, 114, 115
Luttmer, E., 312 Mill, John Stuart, 73
Miller, Warren E., 90
Macedo, Stephen, 142 misrepresentation, 42–51, 52, 322
Mackuen, Michael B., 90 mobilization, 12, 342, 353–4
Macleod, William Christie, 179, 204 monarchy, constitutional, 100
Madison, James, 5, 338 monitoring problem, 1–12
constitutional assemblies, 103 Montesquieu, C. de, 287
direct democracy, 286 Morgan, Commissioner Thomas J., 197–8
“doctrine of discovery,” 162, 165 Mosely-Braun, Carol, 117
leadership, 289 motivation, 73
 Index 363

Msimang, Manto Tshabalala, 229 tension between civilization and

Mullainathan, Sendhil, 11, 324–32 tradition, 179, 182
multiculturalism, 146, 147 Termination Policy, 201
and democracy, 147, 153 and trade, 160
Murray, Charles, 322 treaties with, 160, 168, 176, 177, 189
Museveni, Y., 224, 226, 230 Choctaw Treaty, 180
Myer, Dillon S., 201 Creek treaties, 180
guarantees given, 180
Namier, Lewis, 274 Hopewell Treaty, 168, 176
Napoleon Bonaparte, 100, 102 Treaties of Removal, 180
nation-state, 106 Tribal Councils, 201
nationhood, conceptions of, 151 tribal sovereignty, 159, 166, 167–75, 200
Native Americans, 7–8, 159–89 historical origin of, 172, 174
in Alaska, 169 scope of, 172
assimilation, 142, 188, 194, 199, 201, voluntary integration, 179
202 wards of the state, 161, 168, 171–5,
forced, 178–87 202–6
and the churches, 194 Navajo nation, 169
citizenship, 159, 161, 187, 202–6 New York City mayoral debate, 11, 324–32
and civil rights, 164, 204–5 candidates in, 326–7
and enfranchisement, 204 data and methodology of investigation
limitations on, 203–5 of, 327–32
types of, 203–4 experimental panel design to investigate,
civilization program for, 187 325
colonization of, 159–67, 174 implications of the investigation, 337–9
criminal jurisdiction over, 170 results of investigation of, 332–7
cultural practices, 196, 197, 199 summary statistics on the investigation,
“domestic dependent nations”, 162, 329, 333
163, 165 Newbury, Catherine, 222
education of children, 198 Newbury, David, 222
federal power over, 166, 171–5, 185 Nixon, Richard, 277
“free persons of color,” 164 North America, settlers and natives in,
group identity, 170 159–89
Indian Country, 167
Indian courts, 170 obligation, 139, 140–5
Indian states, 176–8 Oklahoma, 181, 185
land rights, 160, 167–8, 180, 182, 186, Oldmixon, John, 52
200 Olin Foundation, 310
appropriation of land, 181 Oliphant, 173
reduction in holdings, 186 Osage people, 196
right of use, 168
loss of political independence, 167 Page, Benjamin I., 90
relationship to the United States Paine, Thomas, 38
community, 159 Panel Study on Income Dynamics, 315
reservations, 188, 190–5 Parker, Ely S., 178
in California, 191–3 Parsinnen, T. M., 38
in Texas, 193 participation, 42, 154, 228, 258
rights of, 166 and electoral representation, 37
segregation of, 187–202 informal, 35
self-administration, 169 and representation, 61
self-determination, 202, 203, before representative democracy, 35
205 party policies and voter preferences, model
separate representation, 175–6 of, 10
social organization, 199 party unanimity Nash equilibrium, 312,
sovereignty claims over, 159 318
364 Index

paternalism, 176, 187, 205 electoral targeting, 342, 350–1, 353,

the people, 2, 3, 28, 64, 104 354
Pepper, Claude, 352 factions in, 313, 314
Pequot nation, 191 focus of distributive benefits, 342
Pericles, 277, 291 leadership, 354
personation, 22–3, 24, 26, 74 legislative parties, 12
persuasion, 9–12, 342, 348, 350, 353 legislative targeting, 343, 350–1, 353,
petitions, 3, 35 354
anti-slavery, 41 mobilization, 354
and counter-petitions, 39, 52 mobilization activities, 12
electronic, 54 multipartyism, 226
and the French Revolution, 42 transfers to boost turnout of supporters,
Grand Petition, 41 344
Kent petition, 44, 48, 50, 52 types of transfers of benefits, 343–4
Legion Memorial petition, 44, 48, 50, political society, 80–1
52 politics
and misrepresentation, 42–51, 52 constituency-based, 229
and the press, 40, 52, 53 distributive, 342–51
quantity of, 41, 45 fundamentals of, 306–10
right to petition, 49 identity politics, 151, 154, 156
statistics about, 41, 42 mass national, 38
Pettit, Philip, 3–5, 61, 104 oppositional, 154, 155
Pettit, Rory, 82–7 politics of presence, 114
Phillips, Anne, 73, 241 Poole, K., 321
Phillips, John, 44 popular sovereignty, 5, 33, 50, 54
Piketty, T., 308 hostility toward, 37
Pitkin, Hanna, 2, 15, 54, 74, 105, 282 and an informed electorate, 337, 338
criticism of Hobbes, 18, 19, 24 institutional locus of, 102
descriptive representation, 114 negative function of, 97, 106, 108
Plains Indians, 194 as a normative ideal, 109
Plato, 287 and representative government, 90
Plotke, David, 111 Rousseau on, 93–8
policy power, 144, 145, 146, 165, 166
and interests, 276 executive, 94
and leadership, 291 legislative, 94
mapping from policy to outcomes, 307, negotiated, 37
308, 322 relations of, 115, 127, 128
and types of democracy, 291–5 and representative institutions, 129
policy bundle, 10, 305, 310–21, 322 Pratt, Richard, 198
policy outcomes, and voter preferences, 10 press, free, 40, 52, 53, 54, 105
political autonomy, 8 Prucha, Francis Paul, 199
political debates, 11 public opinion, 91, 104, 108
political economy, model of, 298 impact of electoral debates on, 324–32
political elites, 115 and representers, 86–7, 276
political equality, 256–60, 261 public reasons, 79, 80, 81, 104
ancient, 259 public sphere, 141
modern, 259, 260 Pueblo Indians, 196
political parties, 49, 98, 351
activities of, 11 qualifications for office, 8–9, 236–61
allocation of side payments, 352 access to wealth, 244
bribes, 343 conceptual roadmap of, 240–55
constituency of, 314, 343 definition of, 240–8
coordination, 354 as the effects of rules and
coordination activities, 12 circumstances, 244–6
democratization of, 226 as probabilities of success, 243–4, 263
 Index 365

qualifications versus group and outcomes for women, 215, 216,

representation, 241–3 218
voter-endogenous and purpose of, 119, 238
voter-exogenous qualifications, determinative, 104
246–8, 262, 264 differences between the Washington and
democratic costs of, 239, 264, the Westminster models, 82–7
265 electoral representation, 35, 37, 54, 62
and democratic rights, 256 enactive, 4
justification for, 251, 265 factors in, 62
democratic distrust, 252–3 fast-tracking of representation for
and justice, 252 women, 212, 215, 220, 228
limits of voting rules, 253–4 formal and informal, 42, 44, 51
types of, 251–5 and group identity, 223
normative implications of, 255–64 group representation, 241–3
purpose of, 250 Hobbes’s theory of, 15, 20
competency of candidates, 249–50 imperfect, 305
composition of the legislature, 251 indicative, 66–71, 77, 82
correspondence of candidates, 250–1 extended conception of, 69–71
typology of, 248–51, 253 standard conception of, 66–9
quotas informal, 38, 52, 53, 54
debates about, 211 and inherited injustice, 6–8
demand for, 214, 219 interpretive, 4, 66, 78, 82, 104
and democratization, 220–5 Lockean conception of, 30
and elite representatives, 229 of marginal groups, 216, 222
gender-based, 9, 254–5 and the monitoring problem, 1–12
institutional factors and the success of parliamentary, 36, 37
quotas, 216 and participation, 43, 61
and interests, 227–32 popular, 30, 77
negative effects of, 212 professional, 250
objections to, 218, 223 proportional, 68, 69, 73, 221
and participation, 228 and quotas, 232
and procedural democracy, 225–7 before representative democracy, 2–3,
and qualifications for office, 236–61 35, 53
and representation, 232 representers and representees, 4, 35,
role of, 8–9 62–6, 68, 105
responsive, 71–7, 78, 82, 90, 212
Rawls, John, 64, 79, 81 constructive, 76
Reagan, Ronald, 306 directed responsiveness, 71–4
referendums, 10, 96 interpretive, 66, 74–7
Rehfeld, Andrew, 8–9, 112, 236–61 and socioeconomic development, 213
Rehnquist, Chief Justice, 173 substantive, 212, 241
representation, 278–81 theory of, 15
and accountability, 225 varieties of, 61
and authorization, 17, 25, 28 representative government, 247
cultural rights, 154 and aristocratic rule, 286–91
and democracy, 1, 15–18, 26–31, definition of, 275
77–82 and democratic rights, 260
democratic, 9–12, 62, 77–8 and popular sovereignty, 90
democratic value of, 111, 124, 127 principles of, 264
descriptive, 65, 112, 132, 212, 227 purpose of, 90, 91, 108
advantages of, 120 questions about, 107–9
and advocacy, 114–16 responsiveness of, 108
and deliberation, 116–18 representative institutions, 38
and legitimacy, 112–18 and contestatory democracy, 128–31
limits to, 119, 120 and interests, 121, 132
366 Index

representative institutions (cont.) right to petition, 49

non-electoral, 53 social, economic, and cultural rights,
and power relations, 129 153, 154
purpose of, 5 voting rights, 256
and structural injustice, 7–8 Robespierre, M., 2, 102, 106
types of, 38–9 Roemer, John, 10–11, 297–8, 304–18
representatives, 32, 62–4, 81 Rosenbluth, Frances, 9–10, 271–301
autonomy of, 117 Rosenthal, H., 321
delegates, 72, 73, 78, 82 Rousseau, Jean-Jacques, 16, 92, 105
deputies, 65, 72, 77 direct democracy, 93–4, 95
expressive pressures on, 86 general will, 93
independence of, 32 legislation and execution, 94, 97, 99
indicative, 65 opposition to representation, 96, 97,
indirect election of, 101 273
motivation, 73, 114, 131 popular sovereignty, 5, 78, 93–8, 99,
pressures on, 86 108
professional, 69–71 referendums, 96
proxies, 65, 69, 77 usurpation, 94, 95, 96, 100
and public opinion, 86–7 Rucht, Dieter, 229
responsive, 65, 73, 91, 108 Runciman, David, 2–3, 15
selection of, 290 Rwanda, 8, 221
as spokespersons, 74, 75 ethnic mobilization, 222
trustees, 78, 82, 84 Forum of Political Parties, 225
representees, 64, 81 women in, 222
control over representatives, 69, 72, 75, women representatives, 223, 224,
78 240
Republican Party, 10
economic policies, 304, 305–6 Sacerdote, B., 312
electoral success, 304–18 Santa Clara Pueblo v. Martinez, 205
energy policy, 305 Schmitt, Carl, 17
exploitation of the race issue, 322 Schumpeter, Joseph, 271, 289, 294
ideological power, 310 self-determination, 139, 142, 147
manipulative policy framing, 307 self-representation, 114
position on global warming, 305 Seminole nation, 177
quality of candidates, 308 Senegal, 223
Southern strategy, 312, 317 separation of powers, 103, 287
responsibility, collective, 29 Sequoyah, 186
rights, 99, 100, 151 Shapiro, Ian, 1–12
civil and political rights, 153, 154, 231 Shapiro, Robert Y., 90
collective rights, 50, 152, 155 Sieyès, Abbé, 17, 101, 105, 106
cultural rights, 153, 154 Sioux nation, 197
democratic, 9 Skinner, Quentin, 66
right of citizens to choose, 239 Smith, Mark A., 310
right of citizens to run for office, 239, Smith, Melanchton, 67
256, 257, 258–60, 261–2, 264 social groups, construction of, 143
right to an unconstrained choice, 256, socioeconomic development, and political
258, 261 representation, 213
right to vote, 256, 258 Somers, John, 49
equality of, 231 South Africa, 8, 221
group rights, 6, 141, 147 African National Congress, 225, 226,
indigenous rights, 139, 148, 149, 155 231
individual rights, 7, 152, 153, 155 Communal Land Rights Act, 231
membership rights, 7, 26, 151–6 electoral system, 221
political rights, 264 formal inequalities in, 230
and politics, 155 impact of quotas for women, 230
 Index 367

parliamentary women’s caucuses, 222 Uganda, 8, 221

space for dissent, 225 impact of quotas for women, 230
Treatment Action Campaign, 229 National Resistance Movement, 224
women representatives, 225 parliamentary women’s caucuses, 222
women’s organizations, 229 reserved seats for women, 226
Southern African Development single-party dominance, 224
Community, 216 women in, 222
sovereignty, 16, 19, 21, 22, 24, 25; see also women’s organizations, 229
popular sovereignty unions, political, 39
absolute, 98 United States
conceptions of, 7 African slaves, 161
constitutional limitations on, 30 Articles of Confederation, 160
democratic, 95 Atoka Agreement, 185
Hobbes’s theory of, 19 autonomy of suburbs, 130
parliamentary sovereignty, 51 Board of Indian Commissioners, 189,
of the people as a whole, 25 198, 202
source of, 104 Bureau of Indian Affairs, 188, 195–202
unlimited nature of, 18, 19, 24 Burke Act, 186
Spinner-Halev, Jeff, 146 Civil Rights Act of 1866, 161
the state, 2, 15, 19, 105, 106 civil rights in, 79, 81, 116
and the multitude, 23 Commissioner of Indian Affairs, 169,
obligations of, 139, 150 177, 191
personality of, 29, 31 conceptions of sovereignty in, 7
representation of, 32 Congress, 63
responsibility of, 145, 148 Congress-cum-administration, 82, 83,
State v. Forman, 159 84
Stimson, James A., 90 Constitution, 83, 103, 107, 160
Stokes, Donald E., 90 Bill of Rights, 205
Stokes, S. C., 1–12, 342, 347 and Indian policy, 161, 205
structural inequalities, 5, 112–13, 131, state and federal sovereignty, 174
246 constitutional conventions, 177
and collective decisions, 121, 123 Contract with America, 286
and descriptive representation, 120 Curtis Act, 181, 182, 186–7
and interests, 121, 127 Dawes Act, 181–7
and political interests, 112, 124 Dawes Commission, 178, 182, 185
structural injustice, 7, 145–56 Democratic Party, constituency of, 308
Summers, Anne, 227 Department of the Interior, 191
Sweden, 226 elections
Swift, Jonathan, 49, 50 anti-solidarity effect, 312, 316, 317
policy bundle effect, 312, 316, 317
Talton v. Mayes, 171 executive branch of government, 108
Tamale, Sylvia, 229 Federal Housing Administration, 123
Taney, Justice, 170, 202 formation of an Indian state, 177
taxation, 28, 304, 305–6, 310 General Allotment Act, 181–7
Taylor, Charles, 152 Harlan Bill, 177
Teller, Henry M., 195 Home Owners Loan Corporation, 123
Themistocles, 277 Indian affairs, 160, 166, 170
Thomas, Justice, 173 Indian Civil Rights Act, 205
Thucydides, 271, 287, 291 Indian Country, 167
Tilly, Charles, 38 Indian Reorganization Act, 199–202
trade unions, 310 Indian Self-Determination Act, 202
Tripp, Aili Mari, 221, 222, 224 Indian Territory, 178, 181
Truman, President Harry S., 201 Intercourse Act of 1834, 169
Tuck, Richard, 16 jurisdiction over Indian reservations,
Tully, James, 126 159
368 Index

United States (cont.) Waldron, Jeremy, 149

Major Crimes Act, 171 Washington, Ebonya, 11
National Resistance Movement, 221 Washington model of representation, 4, 63,
paternalism, 176, 187 82–4
Peace Policy, 189, 194 Webster, Daniel, profession of belief,
plenary powers over Native Americans, 236–7, 255
171–5, 185 WEDO, 215
political jurisdictions in, 6, 84, 90, 129, Weibull, J., 342
130, 131 Westminster model of representation, 4,
Public Law 280, 159 63–4, 84–5
qualifications for office, 236 whistle-blowers, 71
race in, 119, 122, 129, 321, Wilkinson, Charles, 162, 172
322 Williams, Melissa, 116, 150
railways in, 183, 194 Williams, Robert Jr., 162, 173
representation of Indians in Congress, Williams v. Lee, 169
175, 176 Wilson, Woodrow, 203
representation of individuals, 27 Wolin, Sheldon, 128
representative institutions, 7 women
Republican Party, 10, 304–18 ability of representatives to mobilize, 221
responsiveness of members of Congress, claim to inclusion, 214, 227, 232
107 deficit in representation of, 212–20
role of information in democracy in, 338 discrimination against, 214
state sovereignty, 159 fast-tracking of representation for, 212,
Supreme Court’s judicial review, 102–3, 215, 220, 228
108 influence over decision-making, 215,
system of representation, 82–4, 107 230
think-tanks, 310 participation in institutions of
trade unions, 310 decision-making, 228
Treaties of Removal, 180 political autonomy, 8, 227
two-state policy, 176, 177 quotas for, 8, 211–18, 236
United States v. Joseph, 169 representation in parliaments, 213,
United States v. Kagama, 171 218
United States v. Lara, 173 representation of women’s interests,
United States v. Nice, 203 212, 219
United States v. Rogers, 170 rights to property, 230
United States v. Sandoval, 169 support from political leaders, 224
United States v. Wheeler, 172 women parliamentarians and the
usurpation, 94, 95, 96, 100, 103 women’s movement, 227
women’s movements, 212, 225
Van der Straeten, K., 310 women’s organizations, 226
Van Devanter, Justice, 203 Wood, Elisabeth, 1–12
voter preferences, 9–12, 248, 307 Wood, Gordon, 274
anti-solidarity effect, 10 Worcester v. Georgia, 165–7, 171
policy bundle effect, 10 Wyvill, Christopher, 41
and policy outcomes, 10
voters, 305, 307, 331 Xenophon, 287
belief in opportunities for wealth, 306,
309 Yorkshire Association, 41
core voters, 12, 342–51 Young, Iris, 113
information available to, 296, 299, 324,
337, 338, 339 Zaret, David, 40
swing voters, 12, 342–51, 354 Zeckhauser, Richard, 352
voting systems, 76, 300 Zimbabwe, 226