The Politics of Uncertainty: Sustaining and Subverting Electoral Authoritarianism (Oxford Studies in Democratization) 9780199680320, 0199680329

Citation preview

T H E P OLIT I CS OF UN CERTA I N T Y

OX F O R D S T U D I E S I N D E M O C R AT I Z AT I O N Series editor: Laurence Whitehead

Oxford Studies in Democratization is a series for scholars and students of comparative politics and related disciplines. Volumes will concentrate on the comparative study of the democratization processes that accompanied the decline and termination of the cold war. The geographical focus of the series will primarily be Latin America, the Caribbean, Southern and Eastern Europe, and relevant experiences in Africa and Asia. OTHER BOOKS IN THE SERIES Democracy, Agency, and the State: Theory with Comparative Intent Guillermo O’Donnell Regime-Building: Democratization and International Administration Oisín Tansey Rethinking Arab Democratization: Elections without Democracy Larbi Sadiki Accountability Politics: Power and Voice in Rural Mexico Jonathan A. Fox Regimes and Democracy in Latin America: Theories and Methods Edited by Gerardo L. Munck Democracy and Diversity: Political Engineering in the Asia-Pacific Benjamin Reilly Democratic Accountability in Latin America Edited by Scott Mainwaring and Christopher Welna Democratization: Theory and Experience Laurence Whitehead

The Politics of Uncertainty Sustaining and Subverting Electoral Authoritarianism

ANDREAS SCHEDLER

3

3

Great Clarendon Street, Oxford, OX2 6DP, United Kingdom Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide. Oxford is a registered trade mark of Oxford University Press in the UK and in certain other countries © Andreas Schedler 2013 The moral rights of the author have been asserted First Edition published in 2013 Impression: 1 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, by licence or under terms agreed with the appropriate reprographics rights organization. Enquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above You must not circulate this work in any other form and you must impose this same condition on any acquirer Published in the United States of America by Oxford University Press 198 Madison Avenue, New York, NY 10016, United States of America British Library Cataloguing in Publication Data Data available Library of Congress Control Number: 2013937618 ISBN 978–0–19–968032–0 Printed and bound in Great Britain by CPI Group (UK) Ltd, Croydon, CR0 4YY Links to third party websites are provided by Oxford in good faith and for information only. Oxford disclaims any responsibility for the materials contained in any third party website referenced in this work.

TO LINA, AARON, AND DAMIAN

This page intentionally left blank

Acknowledgments About ten years ago, when I started thinking about the dynamics of electoral authoritarian regimes, the study of non-democratic politics was largely a matter of history and historians. Since then, we have seen emerging a rich and stimulating debate on the politics of dictatorship—theoretically, empirically, and methodologically. Within a decade or so, the comparative study of authoritarianism has turned into a vibrant, well-established field of research in political science. I myself have been contributing to these emergent debates. Readers will find some revised fragments of earlier texts in the first part of the book. Yet, to a degree that is difficult to overestimate, this book is a child of these debates. None of us who participated in the initial debates on authoritarian and (in particular) electoral authoritarian politics, stands where he or she stood at the beginning. We have all been learning from each other, revising our initial assumptions, moving forward, refining our theoretical tools, our empirical knowledge, our methodological standards. Any of our individual achievements is deeply indebted to the collective achievements of the international research community on comparative authoritarianism. Although academic research looks like a lonely enterprise, it is an intrinsically social one. We depend on communities of authors and also on communities of critics. Over the years, I have been very fortunate to receive generous criticism and encouragement from many colleagues I highly appreciate. With apologies to anyone I may be forgetting, I wish to thank Rosario Aguilar, Sonia Alonso, Alejandro Anaya, Javier Aparicio, Andreas Bågenholm, André Bank, Mitzy Baqueiro, Matthias Basedau, Rainer Bauböck, Timm Beichelt, Mark Beissinger, Paolo Benedetti, Andrew Bennett, Allyson Benton, Marcelo Bergman, Matthijs Bogaards, Carlos Bravo, Jason Brownlee, Valerie Bunce, William Case, Francesco Cavatorta, Philip Costopoulos, Larry Diamond, Alberto Díaz-Cayeros, Martin Dimitrov, Daniela Donno, Héctor Duarte, Todd Eisenstadt, Gero Erdmann, Abel Escribà-Folch, M. Steven Fish, Jennifer Gandhi, Carolina Garriga, Barbara Geddes, John Gerring, Johannes Gerschewski, Marlies Glasius, Daniela Gómez, Kenneth Greene, Jonathan Hartlyn, Bert Hoffmann, Marc Howard, Maria Inclán, Steffen Kailitz, William Keech, Judith Kelly, Charles Kenney, Sascha Kneip, Patrick Köllner, Joy Langston, Fabrice Lehoucq, Steven Levitsky, Monserratt Leyva, Staffan Lindberg, Soledad Loaeza, Claudio López-Guerra, Ellen Lust-Okar, Beatriz Magaloni, Eric Magar, Amichai Magen, Jennifer McCoy, Michael

viii

Acknowledgments

McFaul, Josef Melchior, Irma Méndez, Wolfgang Merkel, Alvaro Morcillo, Shaheen Mozaffar, Cas Mudde, Gerardo Munck, Ramón Narvaez, Gabriel Negretto, Kenneth Newton, Marc Plattner, Johannes Pollak, Jean-François Prud’homme, Diego Reynoso, Thomas Richter, Julio Ríos-Figueroa, Philip Roessler, Rodolfo Sarsfield, Frederic Schaffer, Oliver Schlumberger, Manfred Schmidt, Alexander Schmotz, Carsten Schneider, Martin Schürz, Gilles Serra, Benjamin Smith, Richard Snyder, Willibald Sonnleitner, Christoph Stefes, Kathryn Stoner-Weiss, Milan Svolik, Dag Tanneberg, Jan Teorell, Christian Timm, Guillermo Trejo, Bernhard Wessels, Perla Valdez, Nicolas van de Walle, Leonardo Villalón, Lucan Way, Sakura Yamasaki, and Michael Zürn. For research assistance at different points in time, I thank Azul Aguiar, Aranzazú Alonso, Mónica Caudillo, Alejandra Díaz de León, Jesús Leal, Edgar Moreno, Esther Schlosser, Enrique Talavera, and Alejandro Villarino. I am grateful to the Mexican National Council on Science and Technology (conacyt) for research grant 36970-D that got this book started. I am also grateful to my institution, the Center for Economic Research and Teaching (cide) in Mexico City, a precious island of academic excellence and professional integrity. It provided a most propitious environment for my research. Thanks, too, to the Social Science Research Center Berlin (wzb) whose generous hospitality allowed me to complete the manuscript during my sabbatical year in 2011–12. The usual disclaimers apply. Almost last but not least, I wish to thank Nicolas van de Walle for a casual phrase he spoke years ago over a coffee table and might not even remember. It helped me recharge my energies in moments of exhaustion: “Writing books. That’s what I like to do.” I dedicate this book to my wife, Lina, and our sons, Aaron and Damian. For most of the time, they didn’t know or notice or care I was working on that damn manuscript. Yet, without their infallible loving presence I could not have written it. Muchas gracias, and for so much more.

Contents List of Figures List of Tables

xi xiii

Introduction

1 Part I Analytical Framework

1. 2. 3. 4. 5.

The Politics of Uncertainty Shaping the Authoritarian Arena Shaping the Electoral Arena Two Levels of Struggle The Power of Elections

21 54 77 112 141

Part II Empirical Explorations 6. 7. 8. 9. 10.

Comparative Cases and Data The Struggle over Electoral Uncertainty The Calculus of Electoral Manipulation The Calculus of Electoral Protest The Struggle over Regime Change Conclusion

Appendix A: Case Selection Appendix B: Description of Variables Appendix C: Coding Manipulation and Protest Appendix D: Descriptive Statistics Appendix E: Miscellaneous Results References Author Index Subject Index

179 213 259 295 337 372 393 407 415 435 441 443 473 479

This page intentionally left blank

List of Figures A 1.1 1.2 3.1 4.1 4.2 6.1 6.2 6.3 6.4 7.1 7.2 7.3 7.4 7.5

7.6

8.1 8.2 8.3 8.4 8.5

Political regimes in the world, 1972–2004 The continuum of institutional strength The layered sources of regime uncertainty The spectrum of political regimes The nested game of authoritarian elections The electoral consequences of meta-game strategies: governmental dilemmas Frequency distributions of regime manipulation Frequency distribution of cumulative manipulation Frequency distributions of contentious events (during three years before election year) Frequency distributions of electoral protest Hypothetical effects of manipulation on margins of victory Hypothetical effects of opposition strategies on margins of victory Political institutions and competitiveness by subtype of regime Frequency of constitutional majorities and party alternation, by regime subtype Effects of meta-game strategies on predicted probabilities of constitutional majorities and party alternation in competitive regimes Effects of institutional and societal contexts on predicted probabilities of constitutional majorities and party alternation in competitive regimes Cycles of manipulation and threats Threat containment and escalation by manipulation Adjusting manipulation to rising first-order threats Estimated effects of hypothetical increases in vertical threats on manipulation Authoritarian attitudes to risk

3 24 32 78 118 133 196 200 203 205 224 230 234 247

251

254 266 268 271 281 283

xii

List of Figures

8.6 Estimated effects of hypothetical increases in societal stakes (by 10 units) on manipulation 8.7 Institutional stakes and legislative manipulation 9.1 The chain of participatory choices 9.2 Opposition protest and regime change 9.3 Repression and post-electoral protest (best fit) 9.4 Censorship and electoral protest (predicted probabilities) 9.5 Previous threats and electoral protest (predicted probabilities) 9.6 Previous margins of victory and electoral protest 10.1 Societal playing fields and regime trajectories 10.2 Electoral competitiveness and regime trajectories

286 288 297 300 323 326 329 331 346 359

List of Tables A 1.1 1.2 2.1 2.2 3.1 4.1 5.1 5.2 5.3 6.1 7.1 7.2 7.3 7.4 7.5 7.6 7.7 8.1 8.2 8.3 8.4 9.1 9.2 9.3

Theoretical perspectives on authoritarian regime dynamics Types and roles of uncertainty A typology of regime threats (with examples) Authoritarian resources and institutions of domination Designing and manipulating institutions of representation The chain of democratic choice and the repertoire of electoral manipulation Symmetric and asymmetric games Large-N research on the power of authoritarian elections Institutional determinants of leadership tenure, 1946–2002 Peaceful and violent exits from office, by regime type Elections in electoral authoritarian regimes, by world region, 1980–2002 Effects of manipulative strategies on electoral competitiveness Changes in legislative manipulation and margins of victory Effects of opposition strategies on electoral competitiveness Political institutions and electoral competitiveness Institutional determinants of electoral competitiveness in competitive regimes Effects of structural contexts on electoral competitiveness Authoritarian elections leading to party alternation in power The severity and visibility of manipulative strategies Electoral uncertainty and manipulation Vertical uncertainties and manipulation Societal stakes and manipulation Societal structures and opposition protest Political institutions and election boycotts Political institutions and electoral protest

8 22 35 56 63 84 114 151 162 166 192 223 225 228 233 240 241 248 275 278 280 285 312 319 321

xiv

List of Tables

9.4 The informational value of election protest 10.1 Trajectories of electoral authoritarian regimes: frequency distributions 10.2 Trajectories of poor and oil-dependent electoral authoritarian regimes 10.3 Societal playing fields and regime trajectories 10.4 Institutional playing fields and regime trajectories 10.5 Latent vertical uncertainties and regime trajectories 10.6 Electoral competitiveness and regime trajectories 10.7 Electoral protest and regime trajectories 10.8 Contentious action and regime trajectories 10.9 Authoritarian capacities and regime trajectories 10.10 Structural and strategic conditions of regime change A.1 Electoral authoritarianism: criteria of case selection A.2 Authoritarian election cycles, 1980–2002 A.3 Hegemonic party regimes, 1980–2002 C.1 Internal measurement reliability: agreement among coders C.2 External measurement reliability: agreement among datasets E.1 Strategic explanations of opposition protest (binary logistic regressions) E.2 Previous margins of victory and electoral protest

325 342 347 349 353 356 358 362 363 366 368 393 398 403 425 427 441 442

Introduction Electoral authoritarian regimes practice authoritarianism behind the institutional facades of representative democracy. They hold regular multiparty elections at the national level, yet violate liberal–democratic minimum standards in systematic and profound ways. Since the end of the Cold War, they have turned into the most common form of non-democratic rule in the world. Countries ranging from Azerbaijan to Zimbabwe, from Russia to Singapore, from Belarus to Cameroon have hosted electoral autocracies. In this book, I study regime struggles between government and opposition under electoral authoritarianism. I develop a theory of authoritarian elections which is grounded in a general theory of authoritarian politics. Both revolve around the politics of uncertainty. All authoritarian regimes, I posit, suffer from two forms of uncertainty. They suffer from institutional uncertainties: their hold on power is never secure. And they suffer from informational uncertainties: they can never know for sure how secure they are. Contending actors try to shape both: the actual as well as the perceived threats rulers face. In electoral autocracies they do so as well, yet within a particular institutional logic. I hold electoral authoritarian regimes to differ qualitatively from all other dictatorships. Even though they betray the democratic spirit of plural elections, they change their own inner dynamic by admitting them. Formally competitive elections make “a difference that makes a difference” (Bateson 1972). They structure the operative logic of electoral authoritarian rule. They define their distinctive nature. In electoral autocracies, the competition over uncertainty turns first of all into a competition over electoral uncertainty. In electoral authoritarian regimes, governments deploy a broad repertoire of manipulative strategies to keep winning elections. They ban parties, prosecute candidates, harass journalists, intimidate voters, forge election results, and so forth. Their proximate goal is to contain the uncertainty of electoral outcomes, their ultimate goal to prevent the uncertainties of regime change. Opposition parties sometimes refuse to participate in electoral charades, at other times they do so under protest. Whether they boycott or participate, acquiesce or protest, their proximate goal is to render electoral outcomes less certain. Their more distant goal is to make regime change more likely. When do these contending actors choose which kind of strategy? How do their strategic choices interact? How successful are they in either containing or creating electoral uncertainty? How successful in either sustaining or subverting the

2

Introduction

electoral authoritarian regime? These are the central empirical questions The Politics of Uncertainty pursues. In its first, theoretical and conceptual part, the book develops its analytical framework for the study of regime struggles under electoral authoritarianism. In its second, empirical part, it explores the internal dynamics of electoral autocracies on the basis of an original dataset on authoritarian elections worldwide during the last decades of the 20th century (1980–2002). In these introductory pages, I situate my perspective on authoritarian elections within the emergent comparative literature on elections and authoritarianism. I describe my approach as an instance of “political institutionalism” that studies the autonomous logic of political conflict as it unfolds within the institutional arena of authoritarian multiparty elections. Before taking a plunge into the comparative literature, though, let us have a brief look at the historical career of electoral authoritarianism, in particular, its global expansion since the disintegration of the Soviet empire.

THE RISE OF ELECTORAL AUTHORITARIANISM Electoral authoritarian regimes establish the institutions of liberal democracy on paper, yet subvert them in practice through severe, widespread, and systematic manipulation. They play the game of multiparty elections, as they hold regular elections for the chief executive and a national legislative assembly. These elections are broadly inclusive (they are held under universal suffrage), minimally pluralistic (opposition parties are permitted to run), minimally competitive (parties and candidates outside the ruling coalition, while denied victory, are allowed to win votes and seats), and minimally open (dissidence is not subject to massive, but often to selective and intermittent repression). The elections are not, however, minimally democratic. Governments subject them to manifold forms of authoritarian manipulation that violate the liberal– democratic principles of freedom, fairness, and integrity. Their manipulative maneuvers are neither light nor accidental, but severe and systematic enough to fracture the minima moralia of democratic elections. The use of multiparty elections as instruments of authoritarian rule, rather than “instruments of democracy” (Powell 2000), is nothing new. While the Ancient Greeks had understood elections as instruments of aristocratic rule, the American Revolution reinvented them as instruments of republican government. Subsequent struggles for parliamentary sovereignty and universal suffrage turned them into instruments of representative democracy. At the same time, however, during the 19th century, competitive elections served European monarchs and Latin American caudillos to continue exercising

Introduction

3

tyrannical power, with a little help from exclusion, fraud, and coercion. The electoral authoritarian regimes of the 20th century finally found ways of leading long and healthy lives by setting up the entire repertoire of republican institutions while subverting their democratic spirit through a broad and ever widening repertoire of authoritarian manipulation. The most long-lived dictatorship of the 20th century, Mexico under the pri, was an electoral autocracy. Continuing and renovating a deep historical tradition, the authoritarian use of multiparty elections has reached unprecedented global reach since the Cold War withered away in the late 1980s. Based on the political regimes dataset by Axel Hadenius and Jan Teorell, Figure A traces the worldwide distribution of major regime types between 1972 and 2004. As it shows, electoral authoritarianism has not spread primarily at the expense of democracy, but of non-electoral autocracies.1 The figure reflects the global expansion of democracy which we have witnessed since the early 1970s and which has been halted, yet not reverted, in more recent years. The Portuguese Revolution of the Carnations, in April 1974, was the political drama that marked (with the benefits of hindsight) the official starting point of the “third wave” of global democratization (Huntington 1991). Since then, the number of democratic regimes worldwide has roughly doubled. Although

100

Number of regimes

80

Democracy Electoral authoritarianism One-party regime Monarchy Military regime Other

60

40

20

19 72 19 74 19 76 19 78 19 80 19 82 19 84 19 86 19 88 19 90 19 92 19 94 19 96 19 98 20 00 20 02 20 04

0

Figure A. Political regimes in the world, 1972–2004 Source: Authoritarian Regimes Dataset 2.1 (Hadenius and Teorell 2007).

4

Introduction

different counts yield different pictures, the overall trend is quite clear. For instance, the annual Freedom House (fh) report on political rights and civil liberties in the world identified 42 “free” countries in the year 1974. Almost four decades later, in 2011, it judged 87 countries to be “free” (out of a total of 117 countries it classified as “electoral democracies”).2 Without doubt, the breadth as well as the resilience of the global democratic expansion has been a remarkable achievement in the history of the international system. “For the first time in history . . . democracy has become a global force” (Keane 2009: L 325). The world has not become democratic, however. On the one hand, a certain number of long-standing autocracies are surviving in different parts of the world. They may not be many. Today the grand categories of non-democratic regimes of the Cold War era—single-party systems, military regimes, and personal dictatorships—have almost disappeared. Figure A illustrates their collective descent. Nevertheless, we should not underestimate their moral or numerical significance. Repressive regimes like North Korea and Syria continue inflicting human suffering on a large scale on their populations and the Chinese single-party regime alone rules over one-fifth of humanity. On the other hand, in rough proportion to the decline of single-party and military dictatorships, we have seen the rise of new varieties of electoral authoritarianism. Many of them have been born in transitions from single-party rule (such as Gabon and Cameroon). Others have arisen from military coups (such as Algeria and Azerbaijan) and some have grown out of processes of democratic erosion (such as Venezuela and Russia). A small number are survivors from earlier periods, often rooted in struggles of national independence (such as Singapore and Zimbabwe). It has become commonplace to affirm that these new forms of “hybrid” regimes have turned into the most common type of non-democratic regimes in the contemporary world. Figure A corroborates this common diagnose which is likely to hold up in the foreseeable future (despite the dip at the end of the period covered by the figure). The remaining military and single-party regimes, China most importantly, may well join the electoral authoritarian crowd in the years to come. This may also be the case in the Middle East and Northern Africa. The Arab uprisings that ignited in the year 2011 have been targeted against a wide range of regimes: electoral autocracies (such as Tunisia, Egypt, and Yemen), quasi-traditional monarchies (such as Morocco and Bahrain), single-party regimes (Syria), and personalist regimes (Libya). In the short run, their outcomes have been as varied as their targets. Over the medium term, however, they may well lead to a further net increase in the number of electoral authoritarian regimes. As of early 2012, an incomplete list of contemporary electoral authoritarian regimes includes, in the post-soviet region, Armenia, Azerbaijan,

Introduction

5

Belarus, Kazakhstan, Kyrgyzstan, Russia, and Tajikistan; in Northern Africa and the Middle East, Algeria, Iraq, Egypt, and Yemen; in Latin America, Venezuela, Bolivia, Ecuador, Honduras, and Nicaragua; in sub-Saharan Africa, Burkina-Faso, Cameroon, Chad, Congo (Kinshasa), Ethiopia, Gabon, Gambia, Guinea, Mauritania, Togo, and Zimbabwe; and in South and East Asia, Afghanistan, Cambodia, Malaysia, and Singapore. Clearly, we live in an “era of electoral authoritarianism” (Morse 2012).

ELEC T I O N S U N D E R AU T H OR I TA R I AN I SM In response to the global spread of democracy, a huge comparative literature on political democratization has emerged since the mid-1980s. In response to the persistence of old forms and the rise of new forms of authoritarian rule, a blossoming literature on political authoritarianism has emerged since the early 2000s. Within the new field of comparative authoritarianism, scholars have given broad attention to the use of elections by authoritarian regimes. Three distinctive perspectives have emerged: 1. Elections as adornments: The so-called “new institutionalism” in the study of authoritarian regimes proceeds upon the assumption that formally democratic institutions matter, even under authoritarian governance (see Schedler 2009d and 2010a). Not everybody shares this causal assumption. Some analysts regard authoritarian elections as epiphenomena, as mere reflections of underlying power relations, that do not carry causal weight of their own (see Brownlee 2007a). 2. Elections as tools: Others, objecting to the idea that authoritarian elections are purely decorative, conceive them as instruments authoritarian rulers deploy to strengthen their grip on power. Elections can serve this purpose through a variety of ways, for instance, by confusing voters, distracting opposition actors, distributing patronage, pacifying contending elite members, or alerting the government of spikes in popular rejection. Irrespective of their concrete nature, their competitive or non-competitive character, or their local or national reach, elections work to prolong the political life expectancy of authoritarian rulers. They are utensils in the toolbox of dictators. They do not define authoritarian regimes, but authoritarian strategies across regimes (see Gandhi 2008). 3. Elections as arenas: In this book, I develop a third perspective that emphasizes the ambiguity of elections. It contends that some kinds of authoritarian elections are more than mere instruments of dictatorship. Specifically, it claims, regular multiparty elections to highest office at the national level

6

Introduction change the inner logic of authoritarian politics. They open arenas of struggle that are asymmetric, as they grant huge advantages to incumbents, and nonetheless ambiguous, as they endow opposition actors with opportunities of contestation and mobilization that do not exist in non-electoral regimes. Though unfree and unfair by design, authoritarian multiparty elections are contingent in their outcomes. They serve ruling parties to sustain authoritarian rule and opposition actors to subvert it. They do so in a complex manner in which electoral competition goes hand in hand with institutional struggles.

Conceiving authoritarian multiparty elections as arenas of asymmetric conflict, The Politics of Uncertainty studies the internal dynamics that unfold within their confines. It pursues the triple ambition of concept formation, theory development, and empirical discovery. Conceptually, the book situates electoral authoritarian regimes within the spectrum of political regimes. Unlike other authoritarian regimes, electoral autocracies establish the entire set of formally representative institutions that characterize liberal democracy. Unlike electoral democracies, they subject these institutions to severe and systematic manipulation. The book maps their formal institutional configurations, their variegated repertoire of manipulation, their contested boundaries, and their internal differentiation between competitive and hegemonic regimes. Theoretically, the book develops a novel framework for the study of electoral authoritarian regimes. It identifies the competition under and over uncertainty as the driving force of politics under authoritarian rule. Authoritarian regimes suffer from institutional uncertainties: their hold on power is never secure. And they suffer from informational uncertainties: they can never know for sure how secure they are. Contending actors try to shape both: the actual threats rulers face as well as the perceptions of threat. I describe regime struggles under authoritarianism as intertwined struggles over these twin uncertainties. They are struggles under uncertainty over uncertainty. In electoral autocracies, I argue, this “politics of uncertainty” unfolds as a two-level game in which actors simultaneously struggle over voters (at the game level of electoral competition) and rules (at the meta-game level of institutional change). Empirically, the book explores the internal dynamics of hegemonic and competitive regimes on the basis of an original dataset on around 200 authoritarian elections worldwide between 1980 and 2002. In the last instance, whether authoritarian elections serve to sustain or subvert authoritarian rule depends on the conflictive dynamics that unfold within the electoral arena. The book strives to explain actor strategies (regime manipulation and opposition protest) as well as systemic outcomes (electoral competitiveness and regime change).

Introduction

7

PO L I T I C A L I N S T I T U T I O NAL I SM The concept of electoral authoritarianism involves an institutionalist perspective. It rests upon the assumption that institutions matter. When we designate the formal institution of multiparty elections as the defining feature of a class of authoritarian regimes, we assume that formally democratic institutions carry the power to shape the nature of politics under authoritarian rule. Conceiving authoritarian elections as an arena of struggle between government and opposition, involves the further claim that politics matters within the formal framework of authoritarian institutions. When we study the dynamics of conflict within the authoritarian electoral arena, we assume that these dynamics possess intrinsic importance. We also expect them to possess causal relevance and to unfold in accordance with their own autonomous logic. As my approach pays equal attention to regime institutions as they structure conflict and to political conflict as it shapes institutions, I conceive it as an instance of “political institutionalism.” The so-called new institutionalism in political science is a wide assemblage of work united by the assumption that formal (as well as informal) institutions are central pillars of political life. Existing varieties of institutional analysis are conventionally divided into three branches: historical institutionalism focuses on critical turning points followed by long periods of institutional durability, sociological institutionalism on norms and perceptions, and economic or rational choice institutionalism on interest calculations by strategic actors (see Hall and Taylor 1996). A more recently introduced fourth branch, discursive or constructivist institutionalism, studies the dynamics of institutional change through negotiation and communication (see Schmidt 2008). What is strangely absent in the extended family of new institutionalisms in political science is a theoretical perspective that takes seriously the very subject matter of political science—politics. In political science, we have been working with historical, sociological, economic, and communicative institutionalism, yet have omitted to develop the most obvious perspective: political institutionalism. To a remarkable degree, we have left politics out of the institutional study of politics. My own analytical framework tries to correct for this omission. In its emphasis on competitive struggles over uncertainty, it lends primacy to politics as it unfolds within the institutional arena of authoritarian elections. To convey the distinctiveness of my “political institutionalist” perspective, I wish to contrast it with three alternative approaches to the dynamics of authoritarian regimes that differ in their conceptions of conflict, actors, institutions, and uncertainty. Table A provides a synthesis.

8

Introduction

Table A. Theoretical perspectives on authoritarian regime dynamics Society-centered political sociology

State-centered political economy

Elite-centered political economy

Regime-centered political institutionalism

The nature of conflict Core cleavage

Class conflict

State versus society

Factional elite competition

Regime cleavage

Spatial division

Horizontal, within society

Vertical

Horizontal, within state

Contingent

Stakes of power

Material resources

Material resources

Material resources

Political resources

Nature of conflict

Distributive

Distributive

Distributive

Institutional: authoritarianism versus democracy

Autonomy of politics

None

Absolute

High

High

Core policy

Level of taxation

Level of taxation and expenditure

Level of patronage

Civic liberties and political rights

Core logic

Sociological

Economic

Organizational

Political

Explanatory puzzle

Regime choice by dominant social class

Regime choice by Cooperation dominant political with regime class by minority factions

Subversion of power

Dictator

Regime and opposition

The nature of actors Core actors

Social classes

Elite factions

Actor coherence

High

High

High

Contingent

Social identity of government

Representative of social classes

Selfrepresentation

Corporate identity

Contingent

The nature of institutions Role of institutions

Instrumental

Instrumental

Instrumental

Ambiguous

Strength of institutions

Exogenous

Exogenous

Exogenous

Endogenous

Opacity

The role of uncertainty Level of knowledge

Reliable public information

Reliable public information

Reliable public information

Actor expectations

Reliable assessments of power, threats, and credibility

Reliable assessments of power, threats, and credibility

Reliable Uncertainty assessments of power, threats, and credibility

Predictability of outcomes

Determinate

Determinate

Determinate

Contingent

Representative texts

Boix (2003), Acemoglu & Robinson (2006)

Olson (2000), Wintrobe (1998), Gandhi & Przeworski (2007)

Geddes (1999), Brownlee (2007a), Magaloni (2008)

O’Donnell & Schmitter (1986), Bunce & Wolchik (2011)

Introduction

9

1. Society-centered political sociology: What we may call the society-centered political sociology of regimes focuses on horizontal conflicts within society. It identifies social groups as main actors and their distributive conflicts over wealth as the main conflict. Social actors select regimes for their redistributive consequences. The rich seek the protective arm of authoritarianism, the poor the redistributive benefits of democracy. For both sides, the tax rate on capital is the main bone of contention. In a distant echo of Marxian perspectives, politics is epiphenomenal, a mere reflection of social struggles, and politicians are perfect representatives of social classes. Institutions are tools of domination, constraints that victorious social groups impose on their contenders. They also serve the powerful to project images of self-restraint. The core logic of explanation is sociological, the core puzzle the selection of regimes by dominant classes. Since the societal distribution of wealth which determines actor preferences and capabilities is known, uncertainty plays (almost) no role. I identify Carles Boix (2003) and Daron Acemoglu and James Robinson (2006) as “paradigmatic” representatives of this perspective. Earlier, non-mathematical approaches to the political sociology of regimes carried certain theoretical affinities, yet tended to recognize more clearly the autonomous relevance of politics (e.g. Lipset 1981, Moore 1993, and Rueschemeyer, Huber, and Stephens 1992). 2. State-centered political economy: The state-centered political economy of dictatorship locates the center of agency at the peak of the state, rather than in society. The major cleavage runs vertically between state and society, with the former striving to enrich itself by exploiting the latter. The head of state, aka the dictator, establishes himself as principal rogue in residence who monopolizes force and crime within his territory and like a mafia boss extracts involuntary payments from society. The subjects of this extortive scheme rebel unless they get something in return. To appease them and maximize his revenue, the dictator needs to set a ceiling on his exploitative demands and provide a minimum of public goods. To make them trust that he will not renege on the deal, he can set up self-constraining institutions, like parties and legislatures. Since relations of power are transparent, actors do not act under uncertainty. The core logic of explanation is economic, the core puzzle the selection of regimes by the dictator. Mancur Olson’s analysis of roving and stationary rogues (2000) and Gandhi and Przeworski’s analysis of the role of authoritarian institutions in eliciting cooperation and compliance (2006 and 2007) are paradigmatic of this approach.3 3. Elite-centered political economy: Rather than horizontal conflicts within society, or vertical conflicts between state and society, the elite-centered political economy of authoritarian regimes stresses horizontal conflicts

10

Introduction

within the state. The key to regime stability and change, it posits, lies in the dynamics of factional conflict within the ruling elite. Under shifting names and banners, winning factions within the winning coalition compete against losing factions within the winning coalition. Authoritarian regimes remain stable as long as minorities cooperate with majorities within the ruling minority. Factional conflict is distributive, as both dominant and minority factions struggle to ensure access to privilege and patronage. Majority factions establish institutions to dissuade and appease minority factions. Since the existence of factions, their power relations, and their membership are public knowledge, uncertainty plays no constitutive role. The core logic of explanation is organizational, the core puzzles are the selection of regimes by dominant factions and the cooperation of losing with winning factions. I hold Barbara Geddes’ theory of factional conflict within military, single-party, and personalist regimes (1999, 2004, and 2005) and Jason Brownlee’s analysis of the role of ruling parties in managing elite conflict (2007a) to be paradigmatic expressions of this elite-centered approach.4 4. Regime-centered political institutionalism: Finally, my own perspective, which I describe as regime-centered political institutionalism in the analysis of authoritarian rule, focuses on the conflictive interaction between pro-regime and anti-regime actors. The former defend the political status quo, the latter strive to transform it. The locus of opposition is contingent. Regime conflicts confront the state with its citizens, but they also divide the citizenry within and often the ruling elite too. Vertical and horizontal divisions are fluid and interdependent. Their relative importance is an empirical question, not an ontological one. The primary axis of conflict is institutional, not distributive. In the first instance and perhaps even in the last, actors fight for political goods, not material ones. The key policy is the level of civil liberties and political rights. Both regime and opposition actors struggle to shape the role and strength of authoritarian institutions. Uncertainty is ineradicable. Regime actors work to contain it, opposition actors to deepen it.5 The logic of explanation is political, the core puzzle is the loss of power, its disintegration and defeat. How can the powerless succeed in removing the powerful from the summit of the state? The enigma is not domination, but its subversion. The text that comes closest to a paradigmatic expression of such a perspective is the founding text of the contemporary literature on regime change, Guillermo O’Donnell and Philippe Schmitter’s concluding essay of the four-volume Transitions from Authoritarian Rule (1986). In its emphasis on opposition politics and elections, the comparative study of the Color Revolutions in the post-soviet area by Valerie Bunce and Sharon Wolchik (2011) also shows important affinities.

Introduction

11

As these sketchy syntheses reveal, the seemingly innocuous slogan of taking politics seriously involves a series of non-obvious assumptions. It involves the idea that politics follows a logic of its own, that domination matters and opposition too, that collective action is problematic, and that uncertainty is a fact and artifact of political life. x Autonomy. Granting primacy to politics first of all demands taking its (relative) autonomy seriously. Within my “political institutionalist” perspective, the institutional struggles between regime and opposition unfold within an autonomous sphere of politics. The protagonists are political: they are individual and collective actors within state and civil society, not social groups. The identity of regime and opposition actors does not derive from their societal positions, but from their political location. The core conflict is political: it concerns the distribution of political power, not the distribution of material resources. x Domination. Conflicts over political regimes are conflicts over the nature of political regimes and thus over the rules that structure the access to state power and the exercise of state power. They are not distributive proxy conflicts whose genuine goal lies elsewhere, such as in the occupation of public office or the appropriation of economic resources. Authoritarian governments are agents of exploitation and the struggle against them mobilizes demands for justice. Yet, in the first place, authoritarian governments are agents of domination and the struggle against them is driven by demands for liberty. Taking politics seriously means taking domination seriously. It also means taking opposition to domination seriously. x Opposition. The common terminology of “autocratic” or “dictatorial” rule conveys the image of sovereign rulers imposing their solitary will on pliant societies. Yet, variations in the strength of institutions derive from competitive struggles between their supporters and their opponents. Institutions are joint ventures by all participants involved, despite their obvious differences in formal and factual power. Thus we need to widen our attention beyond the narrow circles of state elites. We need to study the obedient who stand on the sidelines, applauding the men in power, as well as the troublemakers who stand on the frontlines, opposing the men of power. x Coordination. Politics, furthermore, is a realm of collective action. Resolving problems of collective action (or aggravating them when it serves to divide and rule adversaries) consumes large quantities of political energy in any political context (see Lichbach 1998). Neither collective actors nor collective action can be taken for granted. To mount credible challenges, dissidents must resolve their problems of coordination. To disrupt them, regimes must aggravate their problems of coordination. Taking politics seriously means taking collective action seriously.

12

Introduction

x Uncertainty. Politics is not a science, but a craft. It takes place, not under laboratory conditions of transparency and control, but under real-world conditions of insecurity (“institutional uncertainty”) and opacity (“informational uncertainty”). The common theoretical vocabulary of institutional equilibrium, consolidation, or stickiness suggests that actors can sit back and relax. That at some point the present is known and the future under control. But it is not. And even less so under dictatorial conditions. Since the authoritarian future is always insecure, the struggle over its appropriation is a struggle with no end. Since the authoritarian present is always opaque, the struggle over its interpretation is a struggle with no end. The literature tends to treat the uncertainty of authoritarian institutions (insecurity) as either given or manageable and the uncertainty of authoritarian facts (opacity) as solvable. I understand both types of uncertainty to be fundamental, endogenous, and irresolvable. Their interplay drives the conflictive interaction between authoritarian rulers and their opponents. The “politics of uncertainty” is the twin competition over institutional and informational uncertainties. It is meant to describe the rational coordinates of political actors as they compete over the institutional foundations of political regimes. In the face of persistent authoritarian realities in the contemporary world, it has become commonplace to declare “the end of the transition paradigm” (Carothers 2002). Such declarations are welcome as they are meant to awake us from dreams of transition, from illusions about the whole world smoothly moving towards liberal democracy. Just as Guillermo O’Donnell alerted us early on against “illusions about consolidation” (1996), we surely need to protect ourselves against illusions about transition, too. However, in my perspective, the “transition paradigm” represents something very different. It does not denote a set of delusional expectations, but a set of analytical assumptions. It represents an approach to regime politics that at its core embraces the “political institutionalist” assumptions I outlined above. It trusts the autonomy of politics, stresses the political nature of actors and conflicts, grants opposition actors a leading role, and recognizes the fragility of collective action, and the centrality of institutional uncertainties. My notion of “political institutionalism” introduces two major elements that go beyond the transition paradigm: the endogeneity of regime uncertainty and the pervasiveness of opacity. One, I treat institutional uncertainty not as a contextual variable, but a strategic goal. I endogenize it. Institutional stability, as much as institutional change, is the work of actors. Institutional uncertainties define the initiation of regime transitions. I seek their origins in the competitive struggles of regime and opposition actors under authoritarian

Introduction

13

rule. Two, I propose to incorporate informational uncertainty into the analysis of regime struggles. Since opacity is pervasive and irresolvable, it leaves its imprint on all actions and perceptions under authoritarian rule. As the secretive and oppressive nature of authoritarian regimes distorts all public communication, all public behavior assumes theatrical qualities. It strives to make visible what we cannot see, to convey knowledge about things we cannot know (see Chapter 1). Both analytical moves distinguish my approach from others that take the strength of institutions as exogenously given and the lack of information as something that can be remedied. Both invite us to transcend the conventional boundary that separates transition studies from studies of authoritarian regimes. If we acknowledge that the opacities that characterize authoritarian regimes do not dissolve with the initiation of transition processes, we can extend the informational assumptions of regime studies to the analysis of transitions. If we acknowledge that transitional uncertainties do not fall from heaven, but arise from the competitive performances of regime and opposition actors, we can extend the political institutionalist assumptions of transition studies to the study of authoritarian regimes. In this sense, the “political institutionalism” I advocate in the study of authoritarian regimes involves generalizing and radicalizing the transition paradigm, rather than celebrating its demise.

PLAN OF THE B O OK The Politics of Uncertainty studies the internal dynamics of electoral authoritarian regimes. It pursues a triple ambition. One is conceptual: it strives to advance the conceptualization of authoritarian institutions in general and electoral authoritarian regimes more specifically. One is theoretical: the book introduces a novel modular framework for analyzing the political and institutional logic of authoritarian elections. The final ambition is empirical: on the basis of an original dataset, it strives to explain the strategic choices that contending actors make within the authoritarian electoral arena as well as the institutional outcomes they generate. The first part of the book focuses on theory and concept development, the second offers empirical explorations.

Analytical Framework In the theoretical and conceptual part of the monograph, I delineate the political rationality that drives the conflictive encounters between authoritarian

14

Introduction

rulers and opposition actors as well as the institutional structures that shape these encounters. I unfold my analytical map in five big steps. x Chapter 1 introduces the generic logic of the politics of uncertainty. Authoritarian regimes suffer from institutional uncertainties: their hold on power is never secure. And they suffer from informational uncertainties: they can never know for sure how secure they are. Contending actors try to shape both: the actual threats rulers face as well as the perceptions of threat. I describe regime struggles under authoritarianism as intertwined struggles over these twin uncertainties. They are struggles under uncertainty over uncertainty. x Chapter 2 traces the institutional configurations of modern non-democratic regimes. It introduces the distinction between institutions of domination and institutions of representation. The former serve to concentrate state power, the latter to disperse and constrain it. Since the 19th century, authoritarian regimes have established formal institutions of representation, yet kept them ineffective by manipulating them. On paper, contemporary electoral authoritarian regimes establish the whole institutional landscape of liberal democracy, above all, multiparty elections at the national level. In practice, however, they subvert these formal institutions through authoritarian manipulation. Institutional manipulation forms the practical counterpart to their institutional hybridity. x Grounded in democratic theory, Chapter 3 circumscribes electoral authoritarian regimes in a more concrete manner. Electoral autocracies can choose from a wide repertoire of strategies of manipulation that are mutually substitutive. They form a family of regimes whose precise demarcation line with electoral democracies is essentially contested. Within this large regime family, I distinguish hegemonic regimes from competitive ones. The former display low levels of institutional uncertainty. They are in equilibrium, the ruling party appears invincible. The latter show significant levels of uncertainty. They are theaters of trench warfare in which the incumbent appears vulnerable. x Chapter 4 lays out the political logic of authoritarian elections. The establishment of multiparty elections transforms the politics of uncertainty. It redefines its core actors, resources, and strategic options. Subject to unilateral manipulation, authoritarian elections are “asymmetric” games, not fair games. In such games, competition unfolds at two levels: at the game level, actors play by the rules; at the meta-game level, they fight over rules. In authoritarian elections, governments and opposition parties compete for electoral support at the same time as they compete over electoral institutions.

Introduction

15

x Finally, Chapter 5 inquires into the genetic logic of authoritarian multiparty elections. Why do autocrats convene them? Why do opposition actors often take part in them? The answer lies in the ambiguity of authoritarian elections. They serve rulers, but opposition actors too. While they expose dictators to the risk of electoral defeat, they also lower their exposure to the risk of military coups. They constrain opposition actors, but also offer them regular opportunities of mobilization. Regime actors may use them to sustain authoritarian rule, dissidents to subvert it.

Empirical Explorations If we wish to decipher the ambiguity of authoritarian elections, if we wish to understand when and why they serve to sustain or to subvert authoritarian rule, we need to understand their internal dynamics of conflict. In its empirical part, the book explores “the politics of uncertainty” contending actors pursue within the electoral arena through an original dataset on authoritarian elections worldwide. It seeks to explain governmental strategies (electoral manipulation) as well as opposition strategies (electoral protest). It also seeks to account for institutional outcomes within the electoral arena (electoral competitiveness) and beyond (regime change). x Chapter 6 presents the “Dataset on Authoritarian Elections” that contains data on electoral strategies, electoral outcomes, electoral institutions, and structural contexts for almost 200 elections between 1980 and 2002. It describes case selection and key variables—operational definitions, sources, coding procedures, and some descriptive patterns. Each subsequent chapter seeks to explain variance in one core element of authoritarian elections: electoral outcomes, regime manipulation, opposition protest, and regime outcomes. In each, I deepen and specify the idea of competitive struggles over uncertainty as they unfold in the electoral arenas of hegemonic and competitive regimes. Each explores the causal weight of actor strategies alongside the impact of institutional and societal structures. In each, mixtures of confirming evidence and negative findings redraw our maps of knowledge. Each is more comfortable with unsettling easy certainties than with producing them. x Chapter 7 explores the extent to which manipulation determines the short-term competitiveness of elections: the margins of victory of the ruling party, its control of constitutional majorities, and the occurrence of authoritarian alternation. It finds that authoritarian maneuvers are effective in containing electoral uncertainty, yet opposition protests even more effective in provoking it.

16

Introduction

x Chapter 8 seeks to decipher the logic of electoral manipulation. Through manipulation, incumbents strive to contain threats to authoritarian stability (first-order challenges). Yet, by doing so, they risk provoking contentious countermoves by their opponents (second-order challenges). According to my data, self-confident hegemonic regimes are less concerned about second-order challenges than their competitive counterparts. x Chapter 9 seeks to decipher the logic of electoral protest. When they challenge incumbents through contentious action, opposition parties need to respond to given structures of constraint and opportunity in a manner that loosens constraints and expands opportunities. They can do so by changing beliefs (the informative role of protest) and by activating threats (the provocative role of protest). My data provide evidence of both. x Chapter 10 moves from explaining inner regime dynamics to estimating the explanatory weight these dynamics carry in defining the trajectories of electoral authoritarian regimes. Highlighting the “embedded autonomy” of authoritarian elections, it finds that actor strategies matter for regime stability and change, yet societal and institutional playing fields as well. In its concluding chapter, the book draws some broad theoretical conclusions, reflects on the future of electoral authoritarianism, suggests some practical implications for international democratization assistance, and outlines some pending tasks for future research.

Notes 1. For similar figures, see Magaloni (2008: 734), Roessler and Howard (2009). Diamond (2002) and Schedler (2002b) provide convergent snapshots for the early 2000s. On the global spread of multiparty elections under dictatorship, see also Gandhi (2008). On the global spread of multiparty elections, see also Hyde and Marinov (2012). 2. Author counts on the basis of the Freedom House 1975 and 2012 Annual Report on Political Rights and Civil Liberties (www.freedomhouse.org). These figures have not changed much since the mid-1990s. For a synthesis of democratic advances and setbacks since the mid-1970s, see Diamond (2008) and Doorenspleet (2005). 3. For a synthesis, see also Geddes (2009: 323–328) and Teorell (2010: Ch. 1). 4. While the “elite-centered” approach I synthesize here focuses on “horizontal” conflicts within the authoritarian elite, more “dictator-centric” approaches to the political economy of authoritarian regimes focus on “vertical” conflicts within the authoritarian elite. Their theories revolve

Introduction

17

around the mixed dynamics of cooperation and competition between the unipersonal dictator and the individuals, groups, and organizations that surround and support him. Representative texts are Tullock (1987) and Haber (2008). 5. Political regimes are conventionally defined as sets of formal and informal rules that structure the access to state power and its exercise. At the same time, the notion is habitually used to refer to sets of actors as well. Throughout this book, I will follow ingrained habits of speaking and use the notion of “regime actors” (sometimes abbreviated to just “the regime”) to refer to the wide ensemble of actors in formal positions of state power. In this sense, the concept blurs the distinction between government and state actors (which is often difficult to draw in practice).

This page intentionally left blank

PA RT I A N A LY T I C A L F R A M E WO R K

This page intentionally left blank

1 The Politics of Uncertainty Authoritarian rulers have a twin problem of uncertainty. They have a problem of security. They can never lean back and relax. They have to continually prevent, detect, and contain threats to their hold on power. And they have a problem of opacity. They can never know for sure how good they are at preventing, detecting, and containing threats to their survival in power. We can conceive both problems as forms of uncertainty: the institutional uncertainty that originates in either actual or potential challenges to their rule and the informational uncertainty that arises from the impossibility of generating secure knowledge on these challenges. Both types of uncertainty are connected and mutually reinforcing. And both are structural. Rulers cannot wish them away but have to cope with them, and so have their opponents. For political actors, coping with uncertainty means adapting to it as well as transforming it. Uncertainty is a parameter of choice as well as an object of choice. Prevailing levels of uncertainty constrain actors and force them to react. At the same time, actors actively shape given levels of uncertainty, depressing them, regenerating them, or heightening them. Uncertainty is a powerful independent variable as well as a crucial dependent variable. Table 1.1 outlines the twin roles the twin uncertainties of dictatorship play. As constraints, the uncertainties of authoritarian rule elicit adaptive responses. Actors form beliefs, make predictions, and take decisions under conditions of relative ignorance; and they adjust to perceived levels of regime security through self-protective strategies. As objects of struggle, the uncertainties of authoritarian rule invite creative responses. Actors transform beliefs and predictions through communicative strategies and they alter levels of regime security through strategies of threat containment and threat creation. To the extent that the comparative literature has recognized the uncertainties of authoritarian regimes, it has recognized them primarily as independent variables, as drivers of political decision-making, much less as dependent variables, as targets of political conflict. In this book, I invert the conventional perspective. While recognizing the constraining force of uncertainty, I identify competitive struggles over uncertainty as the central axis of regime struggles

22

Analytical Framework

Table 1.1. Types and roles of uncertainty Uncertainty as constraint: Adaptive responses

Uncertainty as target: Creative responses

Informational uncertainty: Descriptive inference: belief Information gathering. the opacity of the present formation. Strategic choice Perception management: under imperfect information. communicative strategies. Institutional uncertainty: Predictive inference: formation Threat management: the insecurity of the of expectations. Protective containment and creation future responses to institutional of threats. threats.

under dictatorship. For rulers to sustain authoritarian governance, they need to shape dominant informational and institutional uncertainties. For opposition actors to subvert authoritarian governance, they need to do the same. The “politics of uncertainty” comprises their twin competition over both informational and institutional uncertainties. As I contend, it describes the core rationality that drives contending actors who dispute the basic rules of the authoritarian political game. Unfolding under conditions of uncertainty, their competitive struggles over uncertainty form the center of gravity of their competitive struggles over regime continuity. If authoritarian regime struggles are fights over uncertainty, I expect electoral authoritarian regime struggles to be first of all fights over electoral uncertainty. In Chapter 4, I will lay out the specific institutional logic the politics of uncertainty follows in electoral autocracies. In the present chapter, I sketch its general logic. I outline a broad framework for analysis that places the twin struggle over uncertainty at the center of institutional conflict in authoritarian regimes. I discuss institutional uncertainties and their treatment in the comparative literature on regimes. I trace their distant and proximate sources, and offer a typology of threats to authoritarian survival. I describe the informational uncertainties authoritarian governance generates through the denial of rights and liberties. Finally, I lay out the twin challenges that flow out of these twin uncertainties: the construction of political realities (the management of threats) and the construction of political appearances (the management of threat perceptions).

I N S T I T U T I O NA L U N C E RTA I N T Y Dictators are notoriously insecure. The power and wealth they control today may slip out of their hands tomorrow. The more they have, the more they

The Politics of Uncertainty

23

have to lose. For many, their occupational hazards translate into personality traits. They turn paranoid, chase real as well as imaginary enemies. Joseph Stalin, loving father of a revolutionary killing field, had millions of innocents deported, enslaved, and murdered in the presumptive pursuit of regime safety. Even when they display awesome powers, ostentatious wealth, and hedonic lifestyles, dictators are survivors. To enjoy life, they have to escape death. Often by sowing death.

The Role of Institutions Of course, the uncertainty of the future is an intrinsic part of human life. Human action is forward-oriented. We are aware of the future, and care about the future and our role in it, and the future is always unknown and uncertain to some degree. We face natural, transcendental, technological, and systemic uncertainties. Nature threatens us with earthquakes and diseases, God with eternal punishment, technology with plane crashes and nuclear meltdowns, capitalism with unemployment and financial crises. Here, I am concerned more narrowly with social uncertainties. As human action is by definition indeterminate, social interactions are by definition indeterminate. They are inexhaustible sources of uncertainty. We and others may always act differently than we are supposed to. While we can never eliminate social uncertainty, we can strive to contain it. Numerous social devices, such as personal trust and social reputation, serve to reduce the uncertainty of future interactions. The core technology for managing social uncertainty, though, are institutions. The notion of social institutions is abstract in the extreme. It covers an almost infinite range of empirical phenomena that have nothing in common except one formal property: the reduction of uncertainty. Herein lies the defining function of institutions: they stabilize social expectations. This is what disperse social phenomena we often describe as institutions, like corruption, marriage, and courts of justice have in common: they limit the uncertainty of the future. They do not turn humans into machines, but they do render their interactions predictable within reasonable bounds. Stabilizing expectations is not a matter of all or nothing, though. Strong institutions create deep certainties, weak institutions much less so.1 Now, descending the ladder of abstraction, political regimes are institutions, too. They are sets of formal and informal rules that regulate the occupation of state power: the access to power, the possession of power, the exercise of power. Their role is to stabilize expectations about the hard core of modern political life: How do actors obtain state power, how do they keep it, how do they exercise it? Like other institutions, regimes create more than factual

24

Analytical Framework Full uncertainty

High uncertainty

Intermediate uncertainty

Low uncertainty

No uncertainty

Institutional void

Weak institutions

Intermediate institutions

Strong institutions

Non-social worlds

Authoritarian regime crises

Fragile autocracies

Consolidated autocracies

Competitive regimes

Hegemonic regimes

Fragile democracies

Consolidated democracies

Democratic regime crises

Figure 1.1. The continuum of institutional strength

stability. They create expectations of stability. They provide order and constraint today by securing order and constraint tomorrow.2 Figure 1.1 illustrates the range of possibilities: At the extremes of fully uncertain and fully certain expectations, institutions do not exist. Worlds of complete uncertainty are institutional voids. Worlds of complete certainty are not of our world. They are kingdoms of an all-controlling God, technological utopia of perfect repetition, or cemeteries of irrevocable death, without flowers, worms, or superstition. In between these poles, we find situations of high, intermediate, and low institutionalization with corresponding (inverse) levels of uncertainty. At high levels of institutional strength, we find consolidated regimes, be they autocracies or democracies. At intermediate levels of institutional strength, we encounter fragile regimes, be they autocracies or democracies. In the realm of electoral authoritarianism, hegemonic regimes belong to the former category, competitive regimes to the latter (see Chapter 3). At low levels of institutionalization, we enter situations of regime crisis in which institutions lose their binding force and actor expectations turn indeterminate.

Authoritarian Insecurities Both democracies and autocracies vary in the degrees of institutional strength they possess. Competitive struggles over institutional uncertainty are central to

The Politics of Uncertainty

25

both. Battles of regime consolidation in fragile democracies are battles against institutional insecurities. Anti-democratic assaults against robust democracies are attempts to introduce institutional uncertainties. In this sense, the politics of uncertainty transcends the boundaries between democratic and authoritarian regimes. And yet. While weak democracies can be just as fragile as weak autocracies, consolidated autocracies can never be as safe as consolidated democracies. There are systemic limits to the security of dictators. In terms of leadership security, non-democratic regimes tend to suffer from four structural disadvantages: ●

●

●

●

The threat of rivals: Democracies institutionalize peaceful procedures of alternation in power. Democratic power is pro tempore, open to challenges at regular intervals. The peaks of state power do not admit permanent occupants. Few authoritarian regimes are able to regulate alternation in power. Most remain vulnerable to succession crises and to irregular power grabs by contending factions. The threat of rebels: All regimes produce victims. In democracies, losses are temporary and public decisions reversible. Losers can voice their grievances and seek redress before the courts of justice and public opinion. Non-democratic regimes, by contrast, tend to silence protest and immortalize injustice. They deny impartial remedies to the injustices they produce. Autocracies are standing invitations to rebel. The threat of retaliation: Democracies protect losers who quit positions of power after electoral defeat. They lose their jobs, but nothing else. They continue enjoying their civil liberties and political rights. Dictators are insecure in power because they are insecure in opposition. Losing power would leave them as vulnerable in the future as their opponents are in the present. Their stakes of defeat are infinite. Ignorance: Last but not least, the insecurity of dictators is grounded in their ignorance. As I will discuss below, democracies are regimes of relative transparency, dictatorships regimes of structural opacity. Under authoritarianism, all estimations of institutional uncertainty take place under conditions of informational uncertainty. Our knowledge about the uncertainty of authoritarian institutions is limited by the uncertainty of our knowledge about authoritarian institutions.

Overall, the pacification of politics, a huge civilizational achievement of modern democracy, is much harder to establish under dictatorship, subject to perennial threats of rivals, rebels, and retaliation. Under the surface of authoritarian tranquillity, the menace of violence never quite disappears. These endemic insecurities are deepened by endemic opacities. Even if dictators are strong and secure, they can never be certain they are.

26

Analytical Framework Ubiquitous Uncertainty

One might contend that social expectations are too soft a ground to base our theoretical expectations on. Too soft and volatile and subjective. Ethereal matter. When sociologists tell us that “social structures are nothing else but structures of expectation” (Luhmann 1987: 397) we can accept that as symptomatic of their disciplinary blindness.3 Sociologists know nothing about the hard structures of political power, right? And yet, if we look closer, we can see that the comparative study of political regimes is built on cognitive foundations. In manifold ways, our theories of regime change and stability are anchored in the seabed of social expectations. Actor expectations about the future form the core of our core concepts. Uncertainty is embedded within concepts such as democracy and authoritarianism, regime transition and consolidation, structural opportunities, regime threats, trust and credibility. ●

●

●

Regime types: In democratic regimes, free and fair elections are held according to stable procedures. Their rules are laid out beforehand, their results respected afterwards. Democratic elections provide procedural certainty. At the same time, electoral competition is open and its outcomes are indeterminate. Democratic elections admit substantive uncertainty (see Przeworski 1986). Authoritarian elections follow the inverse logic. Autocrats enjoy the prerogative of meddling with procedures and of determining outcomes in advance. They combine procedural uncertainties with substantive certainties. Regime transition: The founding text of the contemporary literature on regime change, Guillermo O’Donnell and Philippe Schmitter’s small concluding tractatus of the four-volume Transitions from Authoritarian Rule, entered its subject matter precisely by “introducing uncertainty” (1986: Ch. 1). As students of regime change know by heart, the two authors identified the emergence of “extraordinary uncertainty” as the defining feature of regime transitions. Transitions, they observed, lack “the relative stability and predictability” of normal times. They are times of “disorder” and “indeterminacy” (1986: 3–5). In these “moments of madness” (Zolberg 1972), the powers of agency and possibility are only matched by the promises of modern consumer culture: “impossible is nothing” (Adidas). In its emphasis on institutional uncertainty, the concept of regime transition resembles the classical notion of crises as well as the related idea of critical junctures (see Gerschewski et al. 2013). The only difference: “democratic transitions” name their purposeful goal, while crises are open-ended. Structural opportunities: Similar to notions of transition and crisis, the language of “structural opportunities” (see Tarrow 1994) rests upon

The Politics of Uncertainty

●

●

27

counterfactual claims of possible change. Opportunities involve a “sense of possibility” (Bellin 2012: 141). When new opportunities arise, “the parameters on what is politically possible and desirable are expanded” (Bunce 1999: 18). Regime consolidation: The broad notion of democratic consolidation means different things to different people in different contexts (see Schedler 1998b). According to its “classic” acceptance, it denotes firm expectations of regime survival. If the emergence of uncertainty is the defining feature of democratic transition, its recession is the defining trait of democratic consolidation thus understood. Democratic transitions begin when the rules of the authoritarian game start looking shaky; processes of consolidation reach closure when democracy starts looking secure. In transitions, regimes lose their monopolistic position as “the only game in town” and the irruption of competing political games turns feasible. In processes of consolidation, they acquire such monopolies by crowding out their systemic competitors. From this perspective, a regime appears as consolidated when all relevant actors expect it to “last well into the foreseeable future” (Valenzuela 1992: 70), when its supporters can lean back and “relax” (Di Palma 1990: 141), trusting its capacity to weather internal or external challenges. It is solidly institutionalized when even its opponents are resigned to its invincibility and accept its persistence “as an unchangeable fact” (Václav Havel, cited in Sluglett 2007: 102).4 Regime threats: While the transition literature has studied unusual junctures of heightened uncertainty and the consolidation literature its subsequent domestication, the expanding literature on the political economy of dictatorship examines structural sources of uncertainty. It analyzes permanent threats to the survival of leaders that originate in their structural dependencies and vulnerabilities.5 Authoritarian leaders are not sovereign. For their exercise of power, they depend on actors who control resources they desire: economic wealth, arms, organization, technology, human capital, legitimacy. For their survival in power, they are vulnerable to actors who control resources they fear: organized violence, above all. Thanks to their endowment with independent resources, these actors pose perennial threats to the capacity of leaders to govern and to survive in government.6 Just like the notions of transition and consolidation, the concept of regime threats captures actor expectations and institutional uncertainties. It is a carrier of prospective causal claims. If we describe x as a threat to y, we express the causal expectation that x may harm y in the future. Regime threats are not actual causes of regime breakdown, but potential causes. The damage they do lies in the future, not in the past. In contexts of regime struggles, the notion of threats commonly refers to the conditional

28

●

Analytical Framework probability that determinate actors turn into sources of institutional uncertainty: If actor x does y (with probability p), then negative institutional consequences z will follow (with probability q). Credibility and trust: If powerful political actors face existential threats from others, they can choose among two courses of action: They can either try to neutralize their adversaries by changing the prevalent correlation of power, for example, by expropriating, disarming, or eliminating them. Or they can try to accommodate their adversaries and make them accept voluntary restraints on the use of their power resources, for example, by persuading, flattering, or bribing them. The negotiation of mutual restraints is an appealing strategy, but in the absence of a Hobbesian Leviathan its soft spot is obvious: compliance does not rest on reliable enforcement by third parties, but on mutual trust between parties. As any occasional viewer of gangster movies can tell, trust-based compliance is structurally problematic. Trust is an attitude towards the future, it embraces its uncertainty with no more assurance than fallible judgments about the reliability of others. It is a bet. In the power struggles of authoritarian regimes, it is a bet with high stakes. Consequently, much of the literature on the political economy of regimes accordingly revolves around problems of trust and credible commitment. How can the rich trust that the poor will not expropriate them under democratic conditions (Alexander 2002, Boix 2003, Acemoglu and Robinson 2006)? How can they trust the military will not do so under authoritarian conditions (Olson 2000, Haber 2008)? How can rulers trust the military will not topple them (Geddes 2009)? How can elites trust their leaders will not cheat on them (Magaloni 2008, Svolik 2009)? Just like actors under authoritarian rule, the literature on authoritarian rule is haunted by the irresolvable uncertainty of the future.

Political regimes, then, like other institutions, contain the uncertainty of the future by stabilizing social expectations. Despite their air of exotic abstraction, institutional uncertainties form the core of concepts that form the core of contemporary regime studies. They are the theoretical air we breathe even when we are not aware of it. However, in the comparative study of political regimes, we have failed to understand that uncertainty is a fundamental condition of authoritarian politics, yet also a central object of struggle. We have treated it as an externally given parameter, rather than comprehending it as endogenous to political conflict. Our emphasis on reactive responses over creative approaches to uncertainty has been most obvious in transition studies. Actor-centered “transitology” grants institutional uncertainties a place of pride. As mentioned above, the very notion of regime transitions rests on the

The Politics of Uncertainty

29

presence of “extraordinary uncertainty” (O’Donnell and Schmitter 1986: 3). Yet transition scholars do not study the political dynamics that lead to the emergence of regime uncertainties, but the political dynamics that follow from the emergence of regime uncertainties. Their research interest sets in the moment institutional uncertainties set in, not before. Often the origins of regime uncertainties lie in elite splits, yet the origins of elite splits lie in the dark. They antecede the research interest of transition studies. If we wish to take authoritarian insecurities seriously, we need to endogenize them, rather than externalize them. We need to comprehend them as a purposeful object of political struggle.

LAYERS OF UNCERTAINTY If institutional expectations are the foundations of institutional stability, what are the foundations of institutional expectations? Political expectations are often unstable. They are a soft “tissue that can easily tear” (Kurzman 2004: 171). Edifices of interlocking expectations are fragile social constructions that may collapse like a house of cards when fissures appear in their self-sustaining foundations. Political regimes may look inevitable one day and on the brink of collapse a couple of days after. Only months before it fell, the demise of the Iranian monarchy presided by Shah Reza Pahlavi seemed “unthinkable” even to those who opposed it (Kurzman 2004). In the velvet revolutions in Eastern Europe, the iron cage of expectations about the solidity of political realities in the Socialist Bloc started to crumble like rotten wood when the Polish citizenry voted the Communist party out of power in June 1989 and a non-Communist government took office two months later. In the wave of mass protest and regime change that shook the Arab region in 2011, a cascade of contingent events in Tunisia, triggered by the self-immolation of Mohamed Bouazizi, a despairing street vendor, was sufficient to shake up libraries of well-informed resignation about the unassailable stability of Arab autocracies. Though potentially unstable, political expectations need not be idiosyncratic. If they were, we would have to renounce the ambition of explaining them in general ways. If we wish to build a theory of rational expectations, we need to assume that expectations rise and fall in systematic ways. We need to assume that political actors are able to form convergent expectations about the future behavior of others to the extent that their environment provides clear clues—that is, clearly visible and clearly relevant clues. If contextual information is scarce and contradictory, actor expectations will be undefined and contradictory. If it is abundant and consistent, their expectations can be strong

30

Analytical Framework

and convergent (assuming they possess similar levels of information and use them on the basis of similar causal models).7 At a high level of abstraction, it is easy to posit three broad conditions that sustain institutional equilibria. Institutions appear safe and secure as long as three generic conditions hold: (a) most actors involved behave, and are seen to behave, according to institutional prescriptions; (b) their inner motives and external contexts support, and are seen to support, their continuing compliance with institutional prescriptions; and (c) those actors who are seen to violate institutional imperatives, or are seen to possess good reasons for doing so in the future, are too few or too powerless to affect the overall institutional equilibrium. 1. Behavioral compliance. If most actors comply with the behavioral imperatives of an institution, their observable compliance provides prima facie evidence of institutional strength.8 Such behavioral benchmarks serve well to assess the strength of institutions where the imperatives of institutions are clear and consensual. When people start arguing about the practical implications of institutional rules, when they start testing and contesting the boundaries between the permissible and the forbidden, the very definition of rule-conforming behavior turns problematic. While institutional observance is a matter of institutional strength, its observation is a matter of institutional clarity. However, to assess the strength of institutions we cannot look at compliance alone. We also need to assess causal relations. 2. Conditions of compliance. On the one hand, we need to ask about the inner and external conditions that sustain open compliance. If people continue to perform the outward movements that conform with institutional requirements, but either possess strong inner motives or face strong external pressures for abandoning their conformism (as soon as others do), we cannot trust the outward appearance of institutional strength. For instance, a repressive and corrupt authoritarian regime that enriches its elites and impoverishes its citizens may well be able to maintain social peace over long stretches of time. Yet we do not trust the surface of tranquillity if we know that pressures are accumulating underneath and may erupt at any moment. We often describe such regimes with metaphors that express the perceived incongruence between observable realities and underlying conditions: time bombs, powder barrels, pressure chambers. 3. Consequences of non-compliance. On the other hand, we need to ask about the institutional effects non-compliance is likely to produce. Threats are a function of institutional vulnerabilities as much as of behavioral challenges. Although institutions can be sensitive to minute alterations in behavioral patterns, institutional equilibria generally do not require abso-

The Politics of Uncertainty

31

lute compliance, only “generalized” compliance. They process dissidence, allow for exceptions. The thresholds at which exceptions from the rule start affecting the rule, rather than confirming it, are fuzzy. Still, as long as exceptions remain exceptional, rather than symptomatic of more general trends of institutional subversion, we can observe them without doubting the general effectiveness of prevalent institutions. For instance, if we hold a political regime to be consolidated when “all relevant actors” abstain from political violence, consolidation does not demand the complete cessation of political violence. It is compatible with violent incursions by actors we deem to be “irrelevant” insofar as they are incapable of disturbing the institutional equilibrium. In sum, to assess the strength of institutions, we need to survey observable patterns of interaction. Yet attention to behavioral compliance is not enough. We also need to know something about its probable causes and likely consequences. Since present compliance may be a poor predictor of future compliance, we need to examine its underlying conditions. And since institutions can absorb certain levels of non-compliance, we need to examine their vulnerability to behavioral challenges.

C O L L E C T IV E C H A L L E N G ES While social institutions like monogamy, club bowling, or rules of courtesy, may erode through the cumulative force of individual non-compliance, the replacement of political institutions usually requires some form of collective action. And so does their maintenance. Political institutions do not stand and fall with individual initiatives, but with collective efforts. As a rule, unless challengers are able to coordinate their assault on established political institutions, they are unlikely to topple them. And unless the advocates of the institutional status quo are able to coordinate their defensive efforts, they are unlikely to hold their positions. Collective actors and collective actions thus are the first thing to look at when we strive to assess the strength of political institutions. They are the observable anchors of our assessments of institutional strength. Collective contention is the most proximate cause of institutional uncertainty, collective support the most proximate symptom of institutional strength. Individual actors and individual action remain important, but mainly as inputs for collective action. Figure 1.2 sums up my overall argument about the sources of institutional uncertainty. It depicts three causal layers of varying proximity to the outcome

32

Analytical Framework Regime strength low / high institutional uncertainty

Collective action: compliance / challenge

Individual actors: resources, interests, and norms

Societal structures: state, economy, and culture

Figure 1.2. The layered sources of regime uncertainty

variable of institutional strength: societal structures, individual actors, and collective actors. The most distant sources of institutional uncertainty are societal structures in a broad sense: economic structures, state institutions, and cultural templates. These structural factors translate into individual-level variables: the identity of individuals, their resource endowments and interests, their normative commitments and cultural dispositions. Individual actors and actions translate into collective actors and collective actions. It is the latter which most directly affect the uncertainty of political institutions. Regimes are not threatened by societal structures, nor (usually) by individual actors, but by collective challenges.

Causal Translations Of course, causal “translations” between levels are not simple and determinate, but complex and contingent. Social structures do not translate mechanically into individual choices, which do not translate mechanically into collective choices, which do not translate mechanically into institutional dynamics. Given the length and complexity of the causal chain, all theories of regime stability and change must privilege some level of explanation over others. While some focus on direct causal relations between contiguous causal layers, others

The Politics of Uncertainty

33

leapfrog intermediate layers and connect lower with more distant higher levels. For example, among those who study direct causal relations, students of public opinion often look at the relations between structural variables and individual attitudes,9 students of social movements at the relations between individual dispositions and collective contention,10 and students of political instability at the relations between collective conflict and institutional outcomes.11 Among those who study distant causal relations, students of socio-economic modernization routinely analyze the relation between broad societal structures and institutional outcomes, leapfrogging individual as well as collective actors.12 Similarly, students of institutional capacities assess the strength of state and party institutions at the bottom of my causal chain and the longevity of regime institutions at its top, bracketing actor-related variables in between.13 Students of cultural modernization analyze the relation between cultural patterns and institutional outcomes, leapfrogging the level of collective action.14 Students of political revolutions focus on the relation between state structures and organized violence, circumventing the level of individual choice.15 However. When structural or institutional theorists choose to skip the contingent world of individual and collective action, they choose to skip the micro-foundations of their theories. As one is tempted to note from the viewpoint of methodological individualism, leaving their theories without micro-foundations they leave them without foundations. My actor-based understanding of regime threats is meant to provide such foundations.

Types of Threat Threats are potential sources of institutional instability. The notion of regime threats, more concretely, designates collective actions that carry the potential of destabilizing a political regime, or, more obliquely, of setting into motion corrosive interactive dynamics that may end up destabilizing the regime. With destabilization meaning: the creation of widespread uncertainty about the sustainability of the regime. Exceptionally, actions of isolated individuals can make a difference, too, as in some cases of tyrannicide or self-immolation. Generally, though, it is only through coordinated action that dissidents can generate serious uncertainty about the institutional solidity of regimes. Not all collective challenges are equal, however. They differ in their targets, their origins, and their means. Objects of threat. In the comparative study of political regimes, it has become common to start the enterprise of theory-building, not with the

34

Analytical Framework

functional requirements of political systems, but with the functional requirements of individual rulers. According to the emergent general (and in its generality persuasive) standard account, rulers, whether presiding a pre-modern hierarchical state or the complex bureaucratic structures of a modern state, have to resolve two fundamental challenges. Whatever the substantive goals they pursue, they have to secure their ability to govern (the challenge of governance) and they have to secure their continuity in government (the challenge of political survival).16 Threats to governance derive from structural dependencies: rulers depend on the “cooperation” of their subjects whom they require to contribute labor and taxes in order to develop and maintain structures of power. They typically originate in the refusal of societal actors to contribute required resources to the state. Threats to survival derive from structural vulnerabilities: rulers are vulnerable to acts of rebellion by actors whose “compliance” they require to keep the authoritarian status quo intact. They typically culminate in the refusal of armed actors to carry out the orders of government officials. Securing political governance demands the construction of solid infrastructures of power, securing political survival the construction of solid alliances of power. Coordinated opposition actors may threaten either of these two projects. Although this book (like much of the literature) focuses on questions of regime continuity and leadership survival, we need to keep in mind that problems of governance and survival are mutually contaminating. If a government loses its capacity to govern, its hold on power turns fragile. If its continuity in power is put into question, its capacity to govern is bound to erode too. Origins of threat. Institutional uncertainty may come from anywhere, the carriers of threat may be anywhere. Abstractly speaking, we can distinguish between vertical, horizontal, and external threats. Vertical threats originate from below (the citizenry), horizontal or lateral threats from within the ruling coalition (the elite), and external threats from without national borders (the international community). Mass demonstrations and popular rebellions are typical instances of vertical threats, palace coups and military coups typical manifestations of lateral threats, and war and diplomatic pressure paradigmatic instances of external threats. Means of threat. Threats differ in their sources as well as in their form. They may be peaceful or violent. If successful, they trigger transfers of power. Bloodless, peaceful transfers of power usually take place according to established rules of succession. Examples are elections, dynastic succession, or the consensual rotation of military junta members. Bloodstained, violent transfers of power usually take place in violation of formal or informal rules of succession. Examples are military coups, leadership assassination, or foreign invasion.17

The Politics of Uncertainty

35

Table 1.2. A typology of regime threats (with examples)

Lateral threats Vertical threats External threats

Peaceful threats

Violent threats

Palace putsch Electoral competition Economic sanctions

Military coup Popular rebellion War

If we combine the latter two dimensions, the identity of actors who threaten a regime and the means they deploy, we obtain a sixfold typology of threats to political survival, as depicted in Table 1.2. At times the boundaries between types of actors and instruments are less than razor sharp. Still, the typology provides a heuristic map that allows us to situate the analytic concerns of this book. Its focus on the challenges authoritarian elections pose to authoritarian regimes involves a focus on peaceful threats from below. In principle, electoral autocracies are subject to the same uncertainties as closed regimes are. They live under the threat of palace rebellions, military coups, popular uprisings, foreign invasion, and so forth. What distinguishes them is their audacity to expose themselves to an added form of uncertainty: electoral uncertainty—the hedged uncertainty of the outcomes of national multiparty elections. By opening up this additional source of uncertainty, electoral autocracies do not necessarily expose themselves to heightened levels of threat. What they do is to change the overall configuration of threats. If authoritarian multiparty elections are successful, they have a civilizing effect on the competition for state power. Since they open up a formal venue for challenging top authorities in a peaceful manner, they lower the utility of violent challenges. Before trying to conquer power through bullets, cost-sensitive challengers should assault it through ballots.18

The Interaction of Threats Studies of democratic transitions have long been emphasizing the critical role that lateral conflicts within the regime (elite splits) play in triggering the dynamics of regime change. As Guillermo O’Donnell and Philippe Schmitter famously asserted in their seminal 1986 essay, “there is no transition whose beginning is not the consequence—direct or indirect—of important divisions within the authoritarian regime” (1986: 19). When the coherence of the ruling elite turns uncertain, the continuity of the regime turns uncertain. Subsequent studies have confirmed the empirical regularity that “most of the time the most serious challenge to dictators’ survival in office comes from high level

36

Analytical Framework

allies, not from regime opponents” (Geddes 2005: 6, see also Svolik 2012: 5). Consequently, much of the literature on the political economy of dictatorship focuses on horizontal, rather than either vertical or external threats.19 Yet, although horizontal threats follow their own, age-old logic of elite rivalries, they do not rise and fall in isolation. Threats interact. Threats from below, within, and without influence each other, and so do violent and peaceful threats. Both concrete observations and general propositions about dynamic interdependencies between threats abound in the literature. For example: Evidence of mass support “enhances cooperation within the ruling coalition” (Magaloni and Kicheli 2010: 128). Mass protests are often able to “force defections from the regime” (Beissinger 2009: 75). Under certain conditions, serious threats from below compel elites to cooperate (Slater 2010). The armed forces split in response to citizen protests, and citizen protests respond to splits in the armed forces (Lee 2009). Rulers set up infrastructures of mass mobilization to counteract the threat of military coups (Geddes 2009). Discontented subnational elites stage citizen protests to extract concessions from the center (Robertson 2010). Structural dependencies on the external world affect domestic balances of power (Levitsky and Way 2006). The presence of international observers encourages opposition parties to boycott authoritarian elections (Beaulieu and Hyde 2009). In electoral autocracies, opposition challenges are often tame and lame unless prominent politicians defect from the ruling party (Langston 2006, van de Walle 2006). And so forth. Actors at various institutional locations watch each other and respond to each other, and expect each other to watch and respond. Elections are arenas of confluence of threats. In electoral competition, vertical and horizontal uncertainties are tightly connected. To be effective, one needs the other. Elites whose electoral offers find no resonance among voters are powerless in the electoral arena. Citizens whose electoral demands find no sympathies among parties and candidates are voiceless. When authoritarian rulers introduce multiparty elections, they make themselves vulnerable, not to horizontal threats in isolation, nor to vertical threats in isolation, but to the articulation of both in the electoral arena. Dissenting elites without electoral support as well as discontented citizens without electoral agents may look troubling to electoral autocrats, but not threatening. As long as electoral supply and demand fail to meet, dictators can stay calm. Managing interdependencies among threats forms a central part of authoritarian regime struggles. Authoritarian rulers strive to interrupt the positive feedback between emergent threats, opposition actors work to reinforce it. In electoral authoritarian regimes, their efforts center on the electoral connection of horizontal and vertical threats.

The Politics of Uncertainty

37

I N F OR M AT IO NA L U N C E RTA I N T Y Autocracies pose structural limits to the systemic information actors can acquire. Authoritarian rulers, elites, and subjects, they all operate in a “poor information environment” (Magaloni 2006: 236). Some know more, others less, but all face irredeemable epistemic uncertainties: basic political facts are uncertain, their general significance is uncertain, and their causal relations are uncertain. All estimations of institutional uncertainty take place under conditions of informational uncertainty. Our knowledge about the uncertainty of authoritarian institutions is limited by the uncertainty of our knowledge about authoritarian institutions.

Authoritarian Opacities If we were to ask a philosopher, she would explain to us, patiently, as parents explain simple things to small children, that all our knowledge is uncertain. “The heavenly ideal of perfect knowledge [is] impossible on earth” (Gigerenzer 2008: L 26). A sociologist, at a slightly lower level of abstraction, would inform us that the “self-observation” of modern societies is a perennial challenge, always tentative, incomplete, and contested. True enough. And, still, profound differences exist between the relative transparency of democratic regimes and the structural opacity of authoritarian regimes. For all their internal variance, non-democratic regimes impose much tighter limits, not only on what we know, but on what we can know. Under non-democratic conditions, three structural factors conspire to produce structural opacities: the repression of subjective beliefs and desires; the hidden nature of public politics; and the untrustworthiness of information monopolies. Repressed subjectivity. Authoritarian regimes are by definition oppressive. To larger or lesser degrees, in form or in fact, they suppress the civil liberties and political rights that define modern representative democracy. In democracies, citizens can talk politically and act politically without, in principle, further restraints than the renunciation of violence. In autocracies, they can’t. Autocracies impose certain limits on what subjects can say and do and certain demands on what they must say and do.20 Some people will disregard these restraints and they are likely to pay a price for it. Others, perhaps most others, will behave the way the regime demands and display outward signs of conformity. Yet, as their behavior is coerced, not free and voluntary, we do not know whether it conforms to inner convictions or not. The denial of liberty introduces a structural divorce between the observable behavior

38

Analytical Framework

and the subjective world of personal desires, values, and beliefs. It generates systems of generalized theatricality or “preference falsification” (Kuran 1995). Everybody plays roles, wears masks, measures her words, calculates her deeds, or can be assumed to do so. And everybody means everybody, citizens as well as elites. The true beliefs of citizens as much as the true beliefs of elite members are locked up in a “black box” impenetrable to the authoritarian eye. Poor dictators. They yearn to be loved and feared, only to end up confused and ignorant about the extent to which they are either loved or feared. This is the “dictator’s paradox” (Wintrobe 1998), his irresolvable dilemma: his power and knowledge are inversely related. The more powerful he is, the more frightening to his subjects and his collaborators, the less he can know about those he rules over and those he rules with.21 Hidden action. Hidden action is a classic problem in relations between principals and agents: the former cannot see very well what the latter do day by day. It is a pervasive problem in democracies and even more so in authoritarian regimes. Consequently, while autocracies are well known for surveying their subjects, they also invest huge efforts in surveying their agents. Many of their information-gathering activities are self-directed, striving to find out what their bureaucracies do at different levels. In contemporary China, for instance, the Communist party works as a vast apparatus of meta-surveillance, watching over the watchmen of the regime. In addition, proto-representative institutions like limited media pluralism, intraparty competition in local elections, and bounded tolerance towards contentious action can be understood as information devices that alert central authorities to misbehavior by subnational officials (see e.g. Fukuyama 2012: 21–23). Everywhere principals face problems of hidden action by their agents. Inversely, though not a concern to principal–agent theory, everywhere agents face problems of hidden action by their principals. Authoritarian regimes magnify these problems to the extreme. In democracies, top decision-makers are subject to institutional demands of transparency and accountability. Even while they preserve certain times and spaces of legitimate confidentiality in which they can deliberate, negotiate, and decide behind closed doors, they have to inform about their decisions and justify them in public debate. In autocracies, the black box of politics is much wider and much darker. As authoritarian elites operate in “clandestine darkness” (Przeworski 1982: 25), much less is known about who decides what, when, how, and why in the secretive halls and corridors of palace politics (see also Barros 2011). Unreliable information. To the extent that authoritarian rulers control (or are seen to control) the form and content of public information, they discredit the form and content of public information. “The fact of monopolistic

The Politics of Uncertainty

39

control . . . causes in the general public a profound distrust of all news and other kinds of information” (Friedrich and Brzezinski 1961: 112). When official information lacks independent checks and balances, its addressees can discount it as unreliable—as no more reliable anyway than counter-sources of information like samizdat, rumors, or online blogs. Whatever an authoritarian regime declares in public is unlikely to be taken at face value by its “inmates” (Gauck 2012).

No Way Out Even though the structural opacity of authoritarian regimes is a widely accepted fact in the literature, authors often seem to suggest that rulers are able to overcome their informational dilemmas through a variety of institutional remedies. To open venues for the expression of discontent, they can, among other things, vary the degree of repression (Wintrobe 1998), establish single-party elections (Malesky and Schuler 2010), hold multiparty elections (Cox 2007), license private media (Debs 2007), or grant civil liberties (Robertson 2010). Happily, however, authoritarian rulers can only dampen, but not dissolve the inferential uncertainties that they themselves create. Democratic societies have developed all kinds of instruments to observe themselves: census bureaus, central banks, statistical offices, maps and museums, mass media, contentious politics, social sciences, representative surveys, the Internet, space stations, psychoanalysis. In efforts of institutional mimesis, authoritarian regimes often use the same demoscopic instruments. Alas, in their hands, they are almost infallibly more fallible. Distinctively authoritarian institutions of data collection, like torture, denunciation, the secret police, or bureaucratic reporting systems about the “popular mood,” are no more reliable. They do not and cannot register attitudes, only self-protective behavior observed by self-protective bureaucracies (see also Longerich 2006: L 732).22 In his uncertainty about the threats that are gathering to his sides and to his feet, it may not be much of a consolation to the dictator, but he is not alone. He is not alone in his relative ignorance. The typical king from the typical fairy tale changes dress and mingles among the people to learn what the people think about him. Yet “the people” themselves (individually) do not know what “the people” (collectively) think of their rulers. When authoritarian regimes shut down or restrict the public space, the institutional site of popular sovereignty (Habermas), they also shut down or restrict what we use to identify as “public opinion.” What is left is a disjointed collection of private opinions that are mutually ignorant of each other.

40

Analytical Framework Triple Ignorance

Under these conditions of structural opacity, where all public behavior (except for acts of oppositional heroism) appears as a product of the political regime, rather than an expression of inner motives,23 institutional expectations turn uncertain as their epistemic foundations are uncertain. Basic facts are uncertain, the general relevance of public facts is uncertain, and their causal relevance is uncertain. 1. When actors observe the political landscape they suffer from factual uncertainties: What is going on here? What is the case? What happens in the hidden worlds of authoritarian rule, on the backstage of power, in the prison cells, in the living rooms, in private minds? How does the visible relate to the invisible? 2. Actors also suffer from conceptual uncertainties: What is the meaning of what we see? How do local phenomena relate to general trends? How appearances to realities? How widespread are the signs of dissidence we observe? How reliable the manifestations of conformity? What is the rule and what the exception? What are these cases cases of? 3. While the observation and conceptualization of facts belong to the realm of descriptive inferences, causal inference is no less problematic. The formation of expectations about authoritarian institutions is also fraught by causal uncertainties: What follows from what we observe? Which are possible consequences? Which likely ones? How do societal structures affect individual calculations? How does individual behavior translate into collective action? How do collective acts affect the strength of regime institutions? Closed regimes are notoriously opaque. With secretive zeal, they shield the inner circles of decision-making from public inspection, protect the privileged knowledge of insiders, condemn the general public to generalized ignorance. By comparison, electoral autocracies are much more transparent. Their reliance on repression tends to be selective and intermittent. They allow opposition to express itself in the open. They are more vulnerable to information leakages from within the state bureaucracy. They tend to tolerate the conduct of public opinion surveys. Incapable of subjecting the public sphere to comprehensive control, they often cohabitate with independent and even hostile private media. Incapable of extracting public shows of citizen loyalty beyond electoral participation, they often show themselves indifferent to private expressions of dissent. And yet. Electoral authoritarian regimes are no glass houses. Though less opaque than closed regimes, they are less transparent than democracies. Democratic elections express the will of the electorate.

The Politics of Uncertainty

41

Authoritarian elections do not. They are uncertain amalgams of popular preferences and authoritarian machinations (as I will lay out in Chapter 4).

C O N ST RU C T I N G P OL IT I C A L R EAL I T I ES For actors under authoritarianism, the twin uncertainty of institutions and information entails the twin challenge of managing threats as well as threat perceptions. They need to shape both realities and appearances. Rulers need to exercise power and to demonstrate it. They must handle threats and manage threat perceptions, too. Dissidents need to weaken the incumbent and to show his weakness. They must exploit the vulnerabilities of the regime and shatter its image of invincibility, too. Realities and appearances are interdependent and philosophers are bound to tell us that the light-minded distinction between the two does not withstand epistemological scrutiny. Still, it does not seem extravagant to claim, paraphrasing Peter Berger and Thomas Luckmann (1966), that “the political construction of political realities” under authoritarian rule involves two simultaneous struggles: the struggle over authoritarian realities and the struggle over authoritarian appearances. Let us first have a look at the former.

Shaping Structures Societal structures are remote causes of institutional uncertainty and almost by definition they are not susceptible to short-term manipulation. They are given and fixed, exogenous to the conflictive interplay between political actors. They are the heritage of external constraints authoritarian rulers encounter upon assuming office (Haber 2008). Structural variables are beyond the reach of opposition actors, yet authoritarian governments do have some leeway in shaping them. At the level of political elites, rulers often enjoy considerable margins of maneuver in choosing the winning coalitions that sustain their project of domination. By picking winners and losers, conceding and withdrawing favors, opening and closing access to power, they select structural attributes of the elites that surround and sustain them, such as their age, profession, ideology, ethnic membership, religion, and institutional origins. At the wider societal level, structures are stickier. They are more likely to serve as parameters of authoritarian rule, rather than objects of authoritarian intervention. Few regimes have either the ambition or the power or the time to revolutionize the societies they pretend to govern. Totalitarian regimes have

42

Analytical Framework

been the historical exception, rather than the rule. Stalin, to cite a prime example, did have the revolutionary agenda, the despotic power, and the leisure to subject his society to violent transformation. Having reached power over a peasant population by means of a proletarian revolution, he resolved the contradiction by abolishing the peasants, the only societal group that posed a structural threat to his claims of exclusive dominance (see Figes 2007). Long-lasting authoritarian regimes can also induce incremental, long-term structural changes through purposive policies in fields like public education, rural development, and infrastructure development. For instance, when the Mexican revolution reached formal closure in 1917, the regime that claimed its heritage presided over a profoundly rural society. Seven decades later, when a protracted process of democratization was eating away the political pillars of the post-revolutionary regime, its societal pillars had been eroded long ago through protracted processes of industrialization and urbanization (see Schedler 2010c).

Shaping Individual Action Individuals subject to authoritarian domination can situate themselves towards their political masters in manifold ways. They can choose from a wide spectrum of possibilities, ranging from active collaboration in violent repression to active resistance against violent repression. In between these poles lies the vast landscape of active adaptation. All three categories are broad and multicolored. Collaboration, to begin with, can take many forms. Individuals may collaborate with an authoritarian regime within or outside its institutions, within its civil bureaucracy or security apparatus, at high or low levels, in formal or informal, and in direct or indirect manners. Resistance, the other pole, wears many faces as well. Individuals may confront authoritarian rule through individual or collective acts, through peaceful, transgressive, or violent means, in open or covert manners, by aiming at the high center of power or its local peripheries. Adaptation, finally, is chameleonic by nature. Individuals may adapt to the realities of authoritarian governance by taking part in public rituals and official discourse, by applauding and falling silent at the right moments, by going into inner or outer exile, by ciphering or self-censoring their political disagreements.24 In rough accordance with this threefold categorization of behavior, it is common to group individuals into three simple categories: the bad, the good, and the guilty—regime supporters, their opponents, and the silent masses in between. These distinctions are often appropriate, although realities on the ground tend to be complex, especially in the murky middle, as people struggle to find

The Politics of Uncertainty

43

their personal balances between acts of compliance and defiance. Their choices are often situational and their means of expression subtle (see Biermann 2009). Furthermore, the meaning of concrete actions depends on the meaning the regime attributes to them. Autocracies differ in where they draw the line, and how sharp a line they draw, between what they demand and what they tolerate and what they persecute.25 The same is true with respect to elite behavior. By definition, the members of the political and societal elite that sustain authoritarian governance are supportive players in the authoritarian game. Yet, the very definition of support, the political line that separates members of the governing coalition from opposition actors, differs from regime to regime. Often it is porous and shifting, open to processes of trial and error, or trial and terror. Lowering one’s hands while everybody else continues to applaud the Great Leader may go unnoticed in one regime, and destroy a man’s life in another.26 Regimes then prescribe and proscribe different things for their subjects and their elite allies. To achieve compliance, they can mobilize a broad range of means, including violence, money, and persuasion. However: compliance is never secure. Even credible threats of force may elicit resistance, rather than submission (see Ermakoff 2008: Ch. 3). Primarily through their modest means of moral appeal opposition actors will work to frustrate the ambitions of autocrats and get individuals to resist their coercive, material, and ideological enchantments. Individual actions are objects of domination as well as targets of opposition. If the regime succeeds in extracting displays of loyalty and acquiescence from individuals, it is likely to succeed in suppressing collective challenges. If the opposition succeeds in encouraging acts of individual defiance, it can kindle hope of mobilizing collective challenges.

Shaping Collective Action Dissidence at the individual level does not translate mechanically into dissidence at the collective level. It only indicates the potential of generating collective contention. The criss-crossing efforts government and opposition undertake to influence individual actors ultimately aim at either blocking or facilitating their coordination. Constraining or enabling the emergence of collective action lies at the heart of the competitive politics of collective action. Being “smart enough to know that they cannot squelch all dissent,” contemporary autocracies concentrate their authoritarian energies on inhibiting “meaningful political coordination and information-sharing” among dissidents (Walker 2012: 172). For instance, working “at the leading edge of authoritarian innovation worldwide” (Walker 2012: 172), the Chinese government has established a vast and

44

Analytical Framework

sophisticated apparatus to survey and censor Internet communication (see Diamond 2010, Keane 2012). Contrary to common assumptions, its central purpose does not seem to lie in suppressing expressions of individual discontent with the Communist party regime. Rather than to silence criticism, “the most extensive effort to selectively censor human expression ever implemented” (King, Pan, and Roberts 2012: 1) seems primarily designed to curtain collective action (see King, Pan, and Roberts 2012).

CONSTRUCTING POLITICAL APPEARANCES Authoritarian governments use their powers to shape political realities. Inevitably, they also use them to shape political appearances. Given the structural opacity of authoritarian realities, rulers need to construct realities and to communicate them too. All their acts of power are simultaneous performances of power. They are acts of domination as well as acts of communication. Whatever rulers do, they do in front of multiple audiences—elites, citizens, opposition actors, and the international community—who are at once objects of their games of power and spectators of their displays of power. Even the most physical acts of cruelty and destruction carry symbolic messages. Even acts of secrecy and hiding address the watchful public. Whatever authoritarian rulers do and decide, sentencing dissidents to death or granting pardon, repressing or appeasing protest, building walls or expelling the discontented, they always make a point. They always tell something about themselves and their adversaries. Voluntarily or involuntarily. The same holds for opposition actors. They mobilize resources to change political realities, but also to change political appearances. All their acts of opposition are simultaneous performances of opposition. Whatever it is opposition actors do, raising individual voices or gathering in large numbers, rioting in the streets or staying quiet, they always tell something about themselves and their adversaries. Due to the structural opacity of authoritarian regimes, all actors face structural problems of credible information: How can they learn basic facts about political reality within a context that distorts all factual information? In addition, they face structural problems of credible communication: How can they communicate in credible ways within a context that distorts and discredits all communication? How can they produce credible symptoms of phenomena no one can see or prove? How can they convey invisible realities to skeptical audiences who know nothing, except that they live in a system that encourages the generalized falsification of communication? How can they transmit

The Politics of Uncertainty

45

unknown beliefs and preferences of citizens and elite members in such contexts of epistemic uncertainty? The answer is: by playing theater. The defining core of theatrical interactions lies in a double binary distinction of social roles and social spheres. Theater comprises the social roles of actors who give a scripted performance and audiences who watch it; and it comprises the social spheres of front stage where the performance takes place before the eyes of the audience and back stage where the performance is prepared in seclusion from the public. The Greek θέατρον (théatron) is “a place for viewing.” In the grand theater of authoritarian politics, all actors face the same generic challenge of emitting credible messages. Different actors need to tell different stories, though. They differ in the core messages they need to draft and deliver. In essence, elite members need to communicate their loyalty to the regime, citizens their proximity or distance to the regime, rulers the coherence and popularity of the regime, and opposition actors its incoherence and lack of popular support. In other words, rulers and their allies strive to persuade their adversaries (as well as each other) that neither horizontal nor vertical threats exist. Opposition actors and their allies strive to persuade their adversaries (as well as each other) that both horizontal and vertical threats are boiling under the surface of authoritarian tranquillity. Competing struggles over reality involve competing struggles over appearances.

Elite Performances Being a member of good standing in the selected club of authoritarian regime elites is a tough assignment. It’s not enough to carry a membership card in the ruling party, or an enchanting smile whenever the dictator passes by. The problem is: nothing is enough. Nothing is enough for sure. Membership in the elite—or whatever we call it, the inner circle, the winning coalition, the ruling alliance, the power bloc—is insecure. Those who are inside, as well as those who want to get inside, surpass each other in producing observable symptoms of loyalty, be it personal loyalty to the dictator or systemic loyalty to the regime.27 Yet they never know. Regimes may punish dissidence as well as demonstrations of loyalty they consider insufficient.28 The demands for loyal behavior change, the red lines of intolerable behavior shift. The will of the dictator is as inscrutable as the will of God. What is celebrated today can turn suspicious tomorrow. In China’s Cultural Revolution, for instance, being a “revisionist,” “reactionary,” or “counterrevolutionary,” and thus a chosen victim of “red terror,” was not a matter of ideology, political activism, social class, or anything discernible

46

Analytical Framework

to the rational mind. It was an arbitrary designation according to shifting, opaque, unpredictable criteria. In the last instance, “only Mao himself could ‘detect’ revisionists, or, more accurately, decide who they were” (MacFarquhar and Schoenhals 2006: L 738). Still, authoritarian theater scripts for the role of well-behaved elite member can remain stable over long stretches of time. In institutionalized regimes it is pretty clear what elite actors must do, and must not do, to eat from the tree of power. In post-revolutionary Mexico, for instance, the behavioral requirements of political success were straightforward: Cheer your boss, keep control within your own domain of power, and don’t move if you want to remain in the picture.29

Citizen Performances Loyal elites have to play the role of loyal elites, and loyal subjects the role of loyal subjects. As a matter of fact, if it goes by the wishes of benevolent rulers, everybody has to play the role of loyal subjects. Cautious citizens, whatever their private convictions, are well advised to do so, as deviant conduct runs the risk of punishment. Successful regimes manage to make everyone perform on stage the outward movements it defines as observable signs of loyalty. Václav Havel’s famous grocery vendor is a fine example. Along with the price tags for his fruit and vegetables, he places a call for proletarian concord: “Workers of the world, unite!” As the sign makes clear to everyone, the man is not a man of collective action, not a working-class hero fighting to create a transnational labor movement, but a man of the system, someone trying to get along by doing what he expects is expected of him (in Communist Czechoslovakia). He signals his acquiescence to the demands of the regime as well as his impotence of doing otherwise. Through their public displays of deference, humble subjects collaborate in producing and reproducing the theatrical appearances demanded by the regime (see Havel 1985). Nonconformists have a harder part to play. Unless they are heroes who speak their mind without fearing the consequences, they have to perform delicate balancing acts. They have to trace their personal path between the compromises they are ready to accept and the risks they are ready to accept. Often they maneuver to avoid being perceived as regime supporters, without running excessive risks of being prosecuted as regime enemies. During their first years in power, through waves of street terror, propaganda, and state repression, the Nazis cowed the majority population into silent acquiescence to their politics of expelling citizens of Jewish descent from the national community, the state, culture, the economy, and the public sphere. Under the menace of fierce reprisals, people were unlikely to show their criticism of anti-Semitic

The Politics of Uncertainty

47

prosecution openly. They took refuge in small gestures, such as ostentatiously ignoring the yellow stars that Jewish Germans were forced to wear in public since September 1941. If they dared to voice their concerns to authorities, they often invoked arguments of national interest or economic expediency deemed legitimate by the regime (see Longerich 2006: esp. 176, L 3349, and 41, L 741).

Regime Performances Through their proclamations and decisions, as well as their silences and omissions, autocrats send all kind of messages to their expectant audiences. By prosecuting peaceful dissidents, they trace the limits of tolerated behavior; by failing to prosecute armed gangs of regime supporters, they indicate the realms of sanctioned violence and impunity. By keeping their crimes secret, they project themselves as respectful members of the civilized world; by committing their crimes openly, they project themselves as resolute executioners of the civilized world. And so forth. Yet, among the many messages autocrats craft to communicate their power and project an image of strength, two are of paramount importance: (a) to keep horizontal threats at bay, authoritarian rulers need to convey an image of elite cohesion; (b) to keep vertical threats at bay, they need to convey an image of popular support. The management of threats involves the management of threat perceptions. Staging elite cohesion. For long, the regime literature has recognized the importance of elite cohesion to the stability of authoritarian regimes. As long as the ruling elite does not show any public fissures, it is almost impossible to tumble an authoritarian regime, even an apparently weak one (although its weakness is likely to induce such fissures). Inversely, as soon as the ruling elite starts to carry out its perennial rivalries in public, changing the regime turns possible, even when its infrastructural and repressive capacities are high (although its strength tends to discourage public rivalries). Yet, if regime crises emerge when regime elites split, when do regime splits emerge? Elite divisions do not fall from heaven. They are the work of actors. Nor are they burning meteorites whose nightly descent to earth we watch in awe. Elite splits are non-objective and non-obvious. More often than not, it is not clear at all how cohesive the ruling class is, or how divided and divisible. Whether elites are divided, over what they divide, how relevant and serious their divisions are, is usually a matter of competing interpretations. And thus a matter of competing performances. The major occupation of opposition actors under authoritarian regimes is to speculate

48

Analytical Framework

about splits within the regime. They watch it closely, register its gestures, read between the lines. Since splits spell change and change means hope, their interpretations often carry a dose of wishful thinking—whose supreme expression lies in the silent desire that these speculations turn self-fulfilling: that by discerning hair splits within the elite they may actually be inducing serious fissures. Such expectations are easily self-defeating as regimes respond with defensive repression, striving to extinguish the flames of elite division at their origins. If opposition parties are the private detectives of elite fissures, authoritarian governments are the official propagandists of elite unity. To counter incessant speculations about elite conflicts, they need to craft public performances that allow them to deny their existence. A fine example is Cuba after the abdication of Fidel Castro. During the slow-motion succession to his brother Raúl, speculations flew high about possible changes arising from generational gaps, ideological conflict, and power struggles within the Communist party. The new commander-in-chief choked all anticipation of internal renewal by appointing a new politbureau of old-guard, hard-core revolutionaries. Resembling a military gerontocracy, the new government irradiated maximum cohesiveness, erasing even the faintest traces of internal heterogeneity (see Hoffmann 2011: 14–16). Political change is not on the agenda, such shows of revolutionary unity tell. It lacks potential allies within the regime.30 Staging public support. Just as they work to offer solid performances of elite unity, authoritarian regimes struggle to produce credible symptoms of popular support. Suppressing open dissent is one side of the equation, generating manifestations of open support is the other. The mass ceremonies staged by totalitarian regimes are well known: big numbers parading the streets, in military discipline and infantile enthusiasm, with flags and torches, chanting combative slogans and sentimental hymns, cheering to the words of their fatherly leaders. Carting thousands of well-functioning subjects onto public squares can be an impressive bureaucratic accomplishment. Yet, as official mass rallies tend to carry a certain artificial quality, authoritarian governments often stage more “spontaneous” expressions of “the popular will.” Sometimes they delegate violence to non-state actors who act as authentic expressions of popular indignation, directing their regime-sponsored fury against regime-designated victims. The Nazis were experts in such theatrical performances of violent popular sentiment.31 In a more peaceful mode, regimes turn civil society into an infrastructure of regime support. State corporatist regimes like post-revolutionary Mexico have done so, and contemporary regimes like Russia do so, though in a less coherent and institutionalized manner, when they organize social protest from above, quasi-governmental movements in support of regime policies (see Robertson 2010).

The Politics of Uncertainty

49

When authoritarian regimes succeed in extracting everyday manifestations of support from elites and citizens, be it through persuasion, corruption, or intimidation, they receive more than visible proofs of support. They receive proofs of power. They receive tokens of subordination whose private motives— enthusiasm or resignation, greed or fear—matter less than the public fact of compliance with the dictates of domination.

Opposition Performances If the main communicative challenge of authoritarian regimes is to demonstrate cohesion and legitimacy, the main communicative task of opposition actors is to destroy the appearance of regime cohesion and legitimacy. Scratching the image of elite unity. Opposition actors have little direct influence on the image of cohesion that a regime presents in public. Moreover, the more hermetic the regime, the more difficult it is to crack it open from the outside. To scratch the propagandistic surface of elite unity, dissidents often cannot do much more than try to redefine existing cleavage structures and bet on the self-fulfilling force of their discursive interventions. On the one hand, they can exaggerate existing or potential divisions within the regime, in the hope of actually deepening or creating them. The politics of rumor. On the other hand, they can downplay the differences that separate themselves from selected regime actors, in the hope of actually attenuating them. The politics of alliance. As it makes little sense creating divisions within the regime without building bridges towards the regime (unless one expects it to collapse and wither away miraculously), the politics of alliance is crucial. Alliance builders within the opposition treat selected regime actors as potential allies in the hope of actually turning them into their allies. They treat them as regime “softliners” in the hope of converting them into such. To do so, they need to introduce moral distinctions among authoritarian elites according to their proximity to the dictator or their involvement in human rights violations. Once they have identified acceptable interlocutors, they can offer them negotiations and future rewards in case of timely defection. To the extent that the opposition succeeds in generalizing the expectation that potential softliners will respond to its overtures and switch sides, this expectation may become self-fulfilling and induce them to actually do so. The initiation of bandwagon effects is a publicity stunt. By shattering the image of elite unity, opposition actors shatter the mechanisms of elite unity. Scratching the image of popular support. Legitimacy cannot be produced by administrative decree. A regime can offer reasons for its acceptance, but it

50

Analytical Framework

cannot make people accept them. What they can produce by decree, if they possess the requisite infrastructural and coercive powers, is behavioral acquiescence. They can make people comply with the behavioral demands they impose on them as observable signs of their political conformity. Regimes cannot make people believe, but they can make them behave, and so create the outward appearance of belief. Opposition actors need to upset the mise-enscène of popular support created by behavioral conformity. In regimes that ban opposition and claim a monopoly of popular representation, even singular acts of defiance can be important in demonstrating the existence of dissent. The mere act of making opposition visible may tear the illusion of unanimity. In more open regimes that concede spaces of political pluralism and admit opposition, the main challenge is not to document the existence of dissidence, but its strength. To the extent that democratic legitimacy dominates official discourse, government and opposition compete over numbers and their significance. Who mobilizes larger numbers in the streets, obtains larger approval ratings at the polls, obtains larger vote shares in manipulated elections? And what do these numbers mean under given limitations of political liberty?32

The Hermeneutics of Authoritarianism Sending proper messages does not guarantee their proper reception. Just as authoritarian governments and opposition actors compete in transmitting messages to domestic and external audiences, they compete in deciphering these messages. Everybody is playing theater and everybody is watching theater and trying to make sense of it. Under reigning conditions of opacity, everything is subject to competing interpretation: facts, frames, and consequences. Factual claims are controversial. Their conceptual framing is controversial. And their causal implications are controversial. In the last instance, when all is said and done and acted out, it is the competitive struggle over interpretations that determines the strength of regimes: their degree of institutional uncertainty.

C ON C LU S I O N The politics of uncertainty, the competitive struggle over institutional and informational uncertainties, is meant to describe the central coordinates of political rationality that drive regime conflicts under all forms of dictatorship.

The Politics of Uncertainty

51

In this book, I employ this general analytical framework to make sense of institutional struggles in electoral authoritarian regimes. In the subsequent analytical chapters, I lay out the institutional framework that structures the politics of uncertainty in electoral autocracies. In the empirical chapters, I use it as a heuristic device that guides the exploration and interpretation of my quantitative data. I do not pretend to fully test my theoretical arguments. My data, while close to actors, lie far from perceptions and expectations. In this sense, my theory transcends my empirical observations. I could discard the theory, or my data, or pretend the gap between the two does not exist. Instead, I uphold the gap and strive to make it productive.

Notes 1. Formal definitions of institutions tend to assume the functional role of institutions, yet commonly fail to make it explicit. Institutions are situated in a rich semantic field. Some authors define them through neighboring terms, such as rules, norms, constraints, and structures that emphasize stability, regularity, and repetition. Others define them by their observable consequences as “regularized patterns of behavior” (Hall and Taylor 1996: 939). Still others include causal elements into their definitions. They point to epistemic, normative, and behavioral requirements of institutional effectiveness when they identify those rules as institutions that are widely known, widely valued, and widely practiced (see e.g. DiMaggio and Powell 1991, O’Donnell 1994: 57). 2. Expected continuity is a defining attribute of strong institutions, but also a causal condition. Institutions are not only defined by interlocking expectations, but sustained by them. Expectations create expectations. They are self-reinforcing. Thus the vital importance for rulers of creating and recreating expectations of regime stability that generate self-fulfilling prophecies. Just as expectations of regime instability do. 3. On institutions as cognitive constructs, see also Berger and Luckmann (1966) and DiMaggio and Powell (1991). 4. Although the literature on “regime consolidation” has focused on democratic regimes, scholars have begun applying the toolkit of “consolidology” to the study of “authoritarian consolidation.” See, for instance, the workshop on “Authoritarian Consolidation” held at the University of Duisburg in May 2009 (). 5. The foundational text is Gordon Tullock’s Autocracy (1987). More recent contributions are Debs (2007), Escribà-Folch (2007), Gandhi (2008), Gandhi and Przeworski (2006 and 2007), Geddes (1999), Haber (2008), Magaloni (2006 and 2008), Svolik (2009 and 2012), and Wintrobe (1998). 6. If authoritarian consolidation requires the absence of threats, their permanence implies the impossibility of consolidation.

52

Analytical Framework

7. The macroeconomic theory of rational expectations postulates that “economic transactors make unbiased forecasts on the basis of all the information available. This means that although forecasts may be right or wrong, agents do not make systematic forecasting errors” (see “rational expectations,” A Dictionary of Finance and Banking, ed. Jonathan Law and John Smullen, Oxford: Oxford University Press, 2008). 8. Of course, the sources of compliance may lie outside the institution in question. Legal rules, for instance, often do not create social behavior, but only reinforce prevalent patterns of behavior dictated by mores or self-interest. 9. See e.g. Alesina and Wacziarg (2000) and Inglehart and Welzel (2005). 10. See e.g. Lichbach (1998) and Oberschall (2000). 11. See e.g. Ulfelder (2005). 12. See e.g. Lipset (1981) and Przeworski et al. (2000). 13. See e.g. Levitsky and Way (2010b) and Geddes (2005). 14. See e.g. Putnam (1993) and Inglehart (2000). 15. See e.g. Goodwin (2001). 16. On threats to survival, see e.g. Bueno de Mesquita et al. (2003), Haber (2008), and Tullock (1987). On threats to governance, see e.g. Bunce (1999), Gandhi (2008), and Gandhi and Przeworski (2006). 17. In their Archigos dataset, Henk E. Goemans, Kristian Gleditsch, and Giacomo Chiozza similarly distinguish between “regular” (rule-based) and “irregular” (force-based) exits from power (2009). 18. The literature review in Chapter 5 on the power of elections supports the notion that authoritarian multiparty elections have a pacifying effect on politics, insofar as they lower the incidence of violent removals from power (see also Cox 2007). 19. Notable exceptions are Wintrobe (1998) and Acemoglu and Robinson (2006). 20. Of course, non-democracies differ widely in the restraints they impose on their subjects. Totalitarian regimes extract from their subjects visible signs of conformity in all walks of life, such as the “German salute” and the wearing of swastikas in Hitler’s Germany. Non-totalitarian regimes, by contrast, often content themselves with suppressing dissent, rather than mobilizing loyalty. Non-democratic regimes also differ in the degree of violence they are prepared to unleash on their populations. While the totalitarian regimes of the 20th century acted like occupying forces against their own people and set up bureaucracies of murder against those they defined as enemies of the people, non-totalitarian regimes tend to be more selective and restrained in the political use of force. 21. The expression of human subjectivity is always problematic. The boundary between outer and inner worlds is luckily universal (except in some psychiatric disorders). Yet, under repressive conditions, the walls between the two worlds are much more imposing than under conditions of liberty and the links between attitudes and behavior much more tenuous. 22. See, however, Dimitrov (2012). On the opacity of authoritarian regimes, see also Barros (2011), Hoffmann (2009: 231–232), and Wintrobe (1998).

The Politics of Uncertainty

53

23. To some extent, inner motives are endogenous to authoritarian regimes. Autocracies often strive to mold the values and beliefs of their subjects and often they seem to be successful to certain (always uncertain) degrees. 24. On the spectrum of individual and collective responses to authoritarian rule, see e.g. Grafe (2009) and Sharp (2010: Appendix 1). 25. Autocracies also differ in their means of silencing individual voices. Not all regimes would stitch a man’s mouth to prevent him from addressing his executioners. In 1976, in the last fluttering of the Cultural Revolution, Chinese worker Shi Yunfeng was condemned as “active counterrevolutionary” and led to his execution “drugged and with his lips sewn together with surgical thread so that he would not confuse his executioners by shouting revolutionary slogans” (MacFarquhar and Schoenhals 2006: L 5076). 26. I am alluding to an episode narrated by Aleksandr Solzhenitsyn: At a district party conference, no one dared to stop a thundering standing ovation to comrade Stalin. After an agonizing eleven minutes, “the director of the paper factory assumed a businesslike expression and sat down in his seat. And, oh, a miracle took place. Where had the universal, uninhibited, indescribable enthusiasm gone? To a man, everyone else stopped dead and sat down. They had been saved!” (1998: L 891). The man was arrested the same night and sentenced to ten years’ imprisonment. His interrogator signed him off with solid advice: “Don’t ever be the first to stop applauding!” (L 895). 27. Notoriously, courts and personal dictatorships are “sites of intense competition . . . in servility and obedience whose rewards are tokens of royal notice: a nod, a precious smile, perhaps an envelope” (Kapuścińksi 2006: L 190). 28. In the civil war orchestrated by Mao Zedong under the grandiloquent title of “cultural revolution,” a schoolteacher “was sentenced ‘in accordance with the law’ to nine years in prison for having, among other crimes, written in his private diary that a certain Mao-quote gave him ‘boundless energy’, then changed that to ‘very much energy’” (MacFarquhar and Schoenhals 2006: L 3846). 29. The phrase “El que se mueve no sale en la foto” is attributed to Mexican union leader Fidel Velázquez (1900–97). 30. On the dramaturgical construction of elite cohesion by authoritarian rulers, with empirical illustrations from Cuba during the critical juncture of quasi-dynastic power transfer, see Schedler and Hoffmann (2012). 31. See Longerich (2006: 59, L 1057; 75, L 1347–1364; 23–24, L 2296–2313). 32. The imperatives of public performance are most clearly discernible when they contradict the imperatives of utilitarian rationality. For instance, according to standard decision-theoretic assumptions, dissidents should mobilize protest under favorable conditions, when costs are low, expected benefits high, and success is within reach. At times, however, they mobilize under unfavorable conditions, when costs are high, benefits nil, and failure is certain. It may still be rational for them to do so under communicative imperatives, when they need to establish a credible identity as genuine opposition actors, as in the early days of the Polish Committee for the Protection of Workers (kor) of the 1970s (see Dietz 2011).

2 Shaping the Authoritarian Arena Electoral authoritarian regimes differ from other non-democracies in the manner they shape the institutional arena of authoritarian politics. They establish the entire set of formal representative institutions we associate with liberal democracy—while deploying a broad range of manipulative strategies that prevent them from being effective. Although their global diffusion since the end of the Cold War is unprecedented, electoral authoritarian regimes are not entirely new phenomena in the history of political regimes. Rough types of predecessors have existed since the 19th century in Europe and the Americas. To understand their institutional distinctiveness we need to situate them within the broad menu of institutional choices (as well as the broad history of institutional struggles) that non-democratic rulers have been facing since the invention of modern representative institutions.

I N ST I T U T I O N S OF D O M I NAT I O N The landscape of modern authoritarian regimes comprises two sets of institutions: institutions of domination and institutions of representation. The former are instruments of authoritarian imposition, designed to produce compliance and cooperation within monopolies of political power. The latter are concessions to societal pluralism and the principle of popular representation. Designed to channel political conflict, they put formal constraints on power. To counteract these constraints, authoritarian rulers recur to institutional manipulation. Comparative scholars often describe electoral authoritarian regimes as “hybrid” or “mixed” regimes that combine authoritarian with democratic features.1 Yet what exactly do these regimes mix? The distinction between institutions of domination and institutions of representation allows us to understand their peculiar morphology. Electoral autocracies combine a contingent set of institutions of domination with a comprehensive set of institutions of representation, subject to authoritarian manipulation. Just like any other type

Shaping the Authoritarian Arena

55

of non-democratic regime, these regimes need to build solid institutions of domination if they wish to thrive and survive. Still, their defining trait lies in the comprehensive set of representative institutions they adopt—and the corresponding imperative of comprehensive manipulation they face. Most importantly, electoral authoritarian regimes convene inclusive multiparty elections at the national level—while suffocating their democratic spirit through severe and systematic manipulation. Modern Authoritarianism Electoral autocracies belong to the extensive family of modern non-democratic regimes. Pre-modern regimes are commonly distinguished from modern ones by their reliance on traditional sources of legitimacy.2 Yet, authoritarian modernity involves more than post-traditional justifications of authority: ●

●

●

●

Territorial states: Modern autocracies preside over territorial states, members of a global community of territorial states. These states claim monopolies of legitimate violence over the inhabitants of their territory. Even if unbound and arbitrary at their peak, they exercise power through the medium of law, i.e. general formal rules defined by central authorities and backed by public violence. Functional differentiation: Modern politics is an autonomous sphere of action, dependent on other autonomous spheres of action. For all their rhetoric of sovereignty, modern states are immersed in webs of interdependence. They do not constitute the “high center” (Anderson 1983: 19) of hierarchical societies, but interact with social spheres of action that follow their own internal logic, such as religion, science, and the family. Above all, states do not command the economy, but are tightly constrained by the imperatives of an autonomous market economy of global reach. Ideological contingency: Modern authoritarian regimes are post-traditional and post-transcendental. They cannot ground their right to rule on secure claims of tradition or divine will. Ideologically homeless, lacking a secure roof of legitimacy over their exposed heads, they have to tap non-religious sources of legitimacy: nationalism, socialism, Asian values, whatever. Even when, longing for the comforts of pre-modern legitimacy, they invoke timeless sources of authority, they have to invent them, like the Arab monarchies. The deeper their ideological constructions are grounded in local cultures, the higher their popular resonance. Technologies of power: Modern states command technologies of governance unknown to their predecessors. Above all: bureaucratic organization and mass communication. From the revolutionary party to the extermination

56

Analytical Framework camp, from youth organizations to the secret police, modern autocrats have been passionate builders of bureaucratic structures. Born into the era of mass communication, they have been children of the mass media. No radio, no fascism.

Defining modern authoritarian regimes by this mixture of institutional and ideological factors, all of them complex and continuous, raises immediate questions about empirical configurations of authoritarian modernity. While we concede (without further development) that all four dimensions may vary in degree, the simple baseline is: all contemporary regimes are modern.

Authoritarian Resources The state and the market provide authoritarian governments with two fundamental resources: violence and money—the core instruments of repression and cooptation. From the cultural sphere, they may extract a third resource of governance: ideology—the main ingredient of legitimation. These strategic resources serve to elicit compliance and cooperation from subjects and allies by mobilizing different motives. Violence engenders acquiescence through the generation of fear, money through the satisfaction of interests, and ideology through the creation of regime-supportive norms and beliefs (see also Table 2.1). The current literature on the political economy of authoritarian regimes tends to assume that repression is real, but not enough to hold a regime together, while ideology is fake, and thus of little use for autocratic selfTable 2.1. Authoritarian resources and institutions of domination

Authoritarian strategies Motives of acquiescence Structures of political modernity Institutions of domination

Violence

Money

Ideology

Repression

Cooptation

Legitimation

Fear

Interests

Beliefs and values

Territorial monopolies of violence

Functional differentiation state–economy Economic institutions (constitutive, extractive, distributive, and regulatory)

Post-religious justification of power

Repressive institutions (bureaucracies of surveillance and punishment)

Ideological institutions (monopolies of education and mass communication)

Shaping the Authoritarian Arena

57

preservation. According to a near-consensus in the discipline, the key to authoritarian survival lies in the creation and reproduction of political alliances through the distribution of material resources. Broad strands of comparative regime analysis share common ground in assuming the instrumental primacy of cooptation over both repression and legitimation. Numerous authors place their emphasis on the distributive logic of public revenue, as in the literature on inequality and taxation.3 Others stress the political logic of public expenditure, as in the literature on political patronage.4 Still others incorporate both sides of the equation of public finance in their accounts of regime dynamics, as in the literature on predatory states and natural resources.5 The relative inattention to repression as well as the facile dismissal of ideology are startling when one considers the totalitarian regimes of the mid20th century. Describing the ideological regimes of terror established by Hitler and Stalin as mere systems of patronage distribution would miss their very essence.6 Yet contemporary political science tends to restrict its scope of inquiry to political regimes of the Cold War and its aftermath, leaving the study of modern totalitarianism to historians. Authors who adopt wider historical horizons are more likely to take into account the constitutive role violence often plays in the reproduction of authoritarian regimes.7 Still, some emergent strands of research have been counterbalancing our disciplinary fixation on the distributive foundations of dictatorship. Some scholars have been shifting the explanation of authoritarian endurance towards repressive practices and institutions.8 Others have been more generally “bringing the state back in” to the study of authoritarianism.9 And still others have been renewing systematic attention to its normative and ideological foundations.10 Slowly but surely, the comparative study of dictatorship seems to be rediscovering the full range of authoritarian means of domination.

Authoritarian Institutions Rather than dispensing violence, material benefits, and ideological messages in occasional and improvised manners, authoritarian governments can put their strategies of repression, cooptation, and legitimation on a stable basis and expand their scope and density by institutionalizing them (see also Table 2.1). 1. Repressive institutions structure the political use of state violence. The construction of large-scale bureaucracies of political repression has been the distinctive institutional innovation of modern authoritarian regimes. If

58

Analytical Framework

we think of modern dictatorship we think of prison systems, extermination camps, the Gulag, the secret police, presidential guards, the domestic deployment of military force. 2. Economic institutions structure the relation between authoritarian states and the economy. Constitutive economic institutions lay the foundations of the economic system, like money, property rights, labor regimes, and systems of economic dispute settlement. Extractive institutions mobilize the requisite resources for financing the operation of the state, be it through taxation, rent appropriation, or productive enterprise. Distributive institutions allocate public expenditures through public employment, public investment, subsidies, transfer payments, and welfare regimes. Finally, regulatory institutions subject economic transactions to legal constraints, which include border regulations (exchange rate, international trade, capital mobility, and migration regimes). 3. Ideological institutions structure the creation and communication of regime ideologies. The mass media (the machinery of propaganda) and the education system form their vertebrae. In less salient roles, they may also include the scientific community, quasi-corporatist associations of sports, culture, and entertainment as well as regime art, architecture, and rituals (such as national holidays and single-party elections). In broad, generic terms, all modern states, dictatorships as well as democracies, rest upon a wide array of repressive, economic, and cultural institutions. Under democratic conditions, these institutions may serve as institutions of liberty. Repressive institutions are domesticated by the rule of law. Economic and cultural institutions institutionalize pluralism, rather than monopolies of power, wealth, and knowledge. Under non-democratic conditions, by contrast, these institutions work as institutions of domination. Notoriously, authoritarian regimes differ in the degree to which they mobilize, and in the forms in which they institutionalize repression, cooptation, and ideology. On the complex map of institutional configurations of authoritarian rule, totalitarianism occupies an extreme position. The totalitarian regimes of the 20th century put the administrative and technological capacities of political modernity to the service of delirious ideological projects of “total domination” (Arendt 2004: 565). In their quest for absolute power, these regimes of terror erected unbound bureaucracies of surveillance, repression, exploitation, and extermination. Striving to terrorize everyone, they also aspired to monopolize everything: politics and the state, the economy, civil society and mass communication, science and education, truth and virtue, art and entertainment, ideology and religion. Totalitarian regimes have been rare and extreme. Yet, if we fail to understand the polar experience of totalitarian domination, we will fail to comprehend

Shaping the Authoritarian Arena

59

more ordinary forms of non-democratic rule: less ideological, less repressive, less comprehensive. Totalitarian rulers sought to teach us an insane, horrifying lesson: “everything is possible” (Arendt 2004: 505). If we forget that lesson, the indelible lesson of the 20th century, we will fail to understand the nature of non-totalitarian dictatorship as well as, incidentally, the essence of liberal democracy.

I N ST I T U T I O N S OF R E P R E S E N TAT I O N To conserve and exercise power, authoritarian regimes need to develop institutions of domination. They need some minimum capacity of repression, some minimum financial sustenance, and perhaps even some minimum ideological resonance among their agents and subjects. It is as simple as this: Authoritarian regimes are (by definition) systems of political domination that are based upon institutions of domination. However. While institutions of domination are congenial to the authoritarian project, modern autocracies often incorporate formal institutions that seem puzzlingly incongruent with their nature. They introduce political institutions we commonly recognize as “republican,” “liberal,” “representative,” or “democratic,” such as independent courts and multiparty legislatures. The animating “spirit” of representative institutions does not lie in enabling the sovereign exercise of power, but in constraining the arbitrary exercise of power. Rather than establishing monopolies of power, they protect individual rights and give expression to societal and political pluralism. Over the past two centuries, countless authoritarian regimes have established such formal “institutions of representation” in a selective manner. Electoral authoritarian regimes have done so in a comprehensive manner. The far-reaching formal–institutional concessions they make to the liberal–democratic heritage represent the last line of authoritarian defense in a long history of struggle that has been unfolding since the invention of modern representative institutions.

Two Stages of Struggle The systems of representative government we have come to call “liberal democracies” rest upon a configuration of institutions whose fundamental building blocks were essentially conceived and crafted in the 18th century: constitutional government, individual rights and liberties, the rule of law, checks and balances,

60

Analytical Framework

the functional and territorial division of power, representative legislatures, popular elections, universal suffrage, civil society, mass media, and the public sphere. Only mass parties were later products of 19th-century suffrage extensions. The 20th century, while catastrophically creative in developing technologies of repression and destruction, added little to the inventory of basic democratic institutions (except for the invention of constitutional courts after the First World War and the development of international human rights law after Nazism). As we know, the (ideational and factual) emergence of representative institutions in the century of enlightenment and revolution did not lead to their immediate diffusion throughout the expanding world of independent states. It did trigger a long-term struggle for representative government, though, that has been unfolding for more than two centuries now. Since its inception, this democratizing struggle has been a two-stage fight, first over the introduction of representative institutions and then over their effectiveness. Advocates of representative government have been struggling to establish these institutions de jure and to render them effective de facto. Defenders of autocratic sovereignty have been resisting their formal introduction as well as their effective institutionalization. When making formal concessions, they generally strove to counteract them through factual manipulation. The democratic battles of the 19th century in Europe and the Americas largely revolved around the egalitarian extension of civil liberties and political rights, the formal establishment and empowerment of representative institutions, and their liberation from factual domination by powerful actors within state and society.11 The totalitarian regimes of terror that arose in the first half of the 20th century not only swept away the democratic gains of the interwar period; their projects of total domination suffocated the struggle over representative institutions. Hitler and Stalin shut down representative institutions.12 Those few they kept in name (like criminal courts in Nazi Germany and legislatures and quasi-federal structures in the Soviet Union) were colonized and integrated into their bureaucracies of repression and extermination. After totalitarianism, the struggle over representative institutions resumed under changed circumstances.13 In the new postcolonial world, it was not against European semi-constitutional monarchies or Latin American electoral oligarchies anymore that democrats strove to erect effective representative institutions, but against military regimes and single-party states. Many of these made limited concessions to the liberal heritage, such as the partial and contingent tolerance of civil liberties, civic associations, and legislative elections. Then the fall of the Berlin Wall marked a historical turning point in the authoritarian management of representative institutions. It triggered a broad shift from selective to comprehensive institutional concessions, and thus from selective to comprehensive institutional manipulation.

Shaping the Authoritarian Arena

61

At the moment of their foundation, electoral authoritarian regimes do something extraordinary. They adopt the formal institutional framework of their systemic competitor, liberal democracy. Their foundational act is an act of self-denial. Rather than suppressing representative institutions, or accepting only some, electoral authoritarian regimes embrace them all. They set up the full panoply of liberal–democratic institutions, from constitutions to constitutional courts, from legislatures to agencies of accountability, from judicial systems to federal arrangements, from independent media to civic associations. Most importantly, they hold regular multiparty elections at all levels of state power. In their institutional forms, these regimes are virtually indistinguishable from liberal democracies. However, by the iron law of autocracy, authoritarian rulers compensate formal concessions with substantive controls. As they abstain from suppressing representative institutions, electoral authoritarian regimes specialize in manipulating them.

Institutional Design If the struggle over representative institutions comprises two distinct challenges for opposition actors, the challenge of introducing them in form and of making them effective in substance, it comprises two distinct sets of choices by authoritarian governments. First, authoritarian institutional designers have to decide the rough outlines of the institutional landscape they wish to create. Within the constraints and opportunities that their societal and political environment afford them, they have to decide how to structure the political regime they wish to inhabit. They have to give shape to the institutional arena of struggle over power and policies they wish to oversee. Second, once they have opened up certain institutional spaces, they have to constrain and contain them and make sure they do not get out of hand. That is, once they have picked from the menu of institutional choices they have to pick from the menu of institutional manipulation. Of course, authoritarian rulers do not encounter an institutional tabula rasa upon taking office. They “inherit an economy, a system of property rights, a class of wealth holders, and a range of pre-existing organizations and institutions—not the least of which are constitutions, legislatures, political parties, opposition political movements, trade unions, police forces, and militaries” (Haber 2008: 696). Only modern totalitarian rulers with their revolutionary ambition of creating a new state and a new society repudiate tout court the institutional inheritance they encounter. Many autocrats are selective in accepting, modifying, or transforming given structures of rules and power.

62

Analytical Framework

Still, whether they continue, create, transform, or destroy political institutions, their first task of macro-institutional landscaping involves at least seven basic choices (see also overview Table 2.2): 1. Legislatures: Shall rulers keep the task of legislation to themselves? Or shall they create specialized collegial bodies that produce the formal general rules the central state aspires to impose on the inhabitants of its territory? 2. Courts: Shall rulers leave the adjudication of disputes in the hands of political–military authorities? Or shall they set up specialized judicial agencies? If they establish courts systems, shall they limit their jurisdiction to civil and criminal disputes among subjects? Or shall they allow them to arbitrate conflicts between subjects and authorities, and among authorities? 3. Decentralization: Shall rulers strive to monopolize decision-making in the capital city? Or shall they introduce intermediate layers of government between the center and the localities and permit subnational units to exercise bounded political autonomy? 4. Elections: Shall rulers claim sovereign authority to fill positions of state power and pick their own successors? Or shall they establish decentralized appointment procedures that make access to high positions of state power conditional upon formal ratification by their subjects? 5. Parties: Shall rulers build a regime-supporting organization that fields candidates for elections (and selects candidates for non-elective positions)? If they do, shall the official party work as an institution of domination that holds a formal monopoly on public policy and public office? Or shall rulers permit the existence of multiple parties outside formal state control? 6. Media: Shall rulers strive to monopolize mass communication? Or shall they permit the existence of private media outside formal state control? 7. Civil society: Shall rulers strive to monopolize mass organization? Or shall they permit the existence of civic associations outside formal state control? Within the tradition of representative government, the first three institutional choices concern formal divisions of power within the state apparatus. When dividing power either horizontally (between branches of government) or vertically (between center and periphery), authoritarian rulers empower agents. They delegate authority to decision-makers who act, in fact if not in name, as their representatives. The latter four choices concern the admission of societal pluralism. Rather than dividing state power, they establish societal constraints on its exercise. They empower adversaries, rather than agents, of the authoritarian government. They protect and empower citizens, as they allow them to

Shaping the Authoritarian Arena

63

form civic associations and political parties, and to have a voice in the selection of authorities and the formulation of policies. The distinction between the division of power and the admission of pluralism is less than watertight, though. Regime adversaries often use legislatures, courts, and local institutions as sites of political contention (see e.g. Moustafa 2008). And regimes often use media, civil associations, and parties as ventriloquists of governmental policies (see e.g. Wilson 2005). In the face of these institutional alternatives, electoral authoritarian regimes consistently choose the path of nominally representative designs. Conceptually they are defined by these choices.14 However, despite their Table 2.2. Designing and manipulating institutions of representation Representative institutions Formal institutions of power division Legislatures Courts

Decentralization

Formal institutions of pluralism Elections

Parties Media

Civil society

Menu of institutional choice

Menu of institutional manipulation

None / appointed / elected Disempowerment, agent control, legislature fragmentation Fusion / division of political Disempowerment, agent control, and judicial powers. fragmentation, insulation Settlement of disputes among subjects / subjects and authorities / authorities Central monopoly of decision- Disempowerment making / territorial (administrative decentralization centralization), repression (agent control), bureaucratic control (agent selection), accountability (ex post intervention)

None / irregular / regular elections. Subnational / national, legislative / executive elections None / single party / party pluralism Monopoly / pluralism of mass communication Repression / monopolization (state corporatism) / pluralism

Disempowerment, market restrictions, preference distortions, vote distortions

Restrictions on means of production, media content, or media consumption Hierarchical organization (monopoly), fragmentation, divide et impera (selective incentives)

64

Analytical Framework

institutional similarities on paper, electoral autocracies differ from electoral democracies in their determination to suffocate the spirit of representative institutions in practice.

INSTITUTIONAL MANIPULATION The formally representative creatures authoritarian regimes breed are not meant to grow and flourish in liberty. Authoritarian rulers will generally strive to constrain, contain, and control them. They will try to make sure that the nominally democratic institutions they set up remain substantively authoritarian. To keep formal institutions of representation from gaining autonomy and weight, authoritarian rulers have to move from institutional landscaping to institutional gardening. They have to shift their attention from the grand decisions of institutional macro-design to the more specific choices of institutional micro-design and micro-management. After institutional creation, they have to move on to institutional containment. Given the central role competitive elections play in the definition of political democracy, the comparative literature has paid substantial attention to electoral manipulation by contemporary as well as historical regimes. By comparison, our cartography of authoritarian interference is less developed with respect to other spheres of authoritarian institution building. Before turning to electoral manipulation in the subsequent chapter, I wish to outline the repertoire of authoritarian manipulation in other institutional arenas.

Containing Agents The following “menus of manipulation” are rather uniform with respect to those institutional fields that establish formal divisions of power. Authoritarian rulers tend to treat divided powers as delegated powers. They tend to bridge formal divisions of power by establishing principal–agent relations. Rather than conceiving formally independent state agencies like courts, legislatures, and subnational governments as countervailing powers, they tend to treat them as subordinate branches of executive power. As they are delegating power, they need to contain the agents they deploy to exercise it. The following menus of manipulation highlight four common strategies of agency control: formal constraints on the delegation of power, the control over agent selection, the management of agent incentives through repression and cooptation, and the induction of coordination problems among multiple agents.

Shaping the Authoritarian Arena

65

The menu of legislative manipulation. Most authoritarian regimes establish some kind of legislative assembly (see Gandhi 2008: 34–37). That is, they create some collective body specialized in writing the rules that the central state, backed by its reservoirs of violence, pretends to impose on the inhabitants of its territory. Given their relatively small size, legislatures are easy objects of authoritarian control. To ensure legislative subordination, rulers may pursue three broad strategies: ●

●

●

Disempowerment: They may tightly circumscribe the formal powers of legislative assemblies (see Fish 2006). Agent control: Even in the face of formally powerful legislatures, rulers may create pliant “rubber-stamp” assemblies either by controlling the selection of legislators (through direct appointment or the control of candidacies to elective legislatures) or by setting up irresistible incentive structures that push deputies towards cooperation with the executive, be it through intimidation or cooptation. Fragmentation: In case rulers cannot control the behavior of legislators, they can try to make sure nobody else can. They can disorganize the legislative assembly, for instance by encouraging the multiplication of party factions or by manipulating the legislative agenda (Riker 1986).

The menu of judicial manipulation. In principle, modern judicial systems serve to adjudicate disputes between private citizens, between citizens and public authorities, and between authorities. Modern authoritarian regimes cannot do without a specialized court system and they often tend to “operate legalistically” (Barros 2002: L 395). In order to establish authoritarian “rule by law” (Ginsburg and Moustafa 2008), rulers may clip the wings of “the least dangerous branch” through manifold devices.15 ●

●

Disempowerment: Authoritarian rulers can restrict the formal powers of judicial actors by limiting their jurisdiction to certain issue areas while withdrawing others from their purview. They can deny them investigative powers thus leaving them at the mercy of executive authorities for the establishment of relevant case facts. They can limit their margins of discretion by imposing dense networks of formal regulation (the bureaucratization of judicial decision-making). And they can neutralize the effects of judicial decision-making either by circumscribing them to individual cases (as in Mexico’s amparo system of judicial review) or by simply “under-enforcing” inconvenient court rulings. Agent control: Even in the face of formally powerful court systems, authoritarian rulers can strive to control them through a mixture of appointment procedures and incentive structures. They can select politically reliable

66

●

●

Analytical Framework magistrates or discipline them through dissuasive punishment regimes. Authoritarian regimes are huge employment agencies for loyal servants, but they are also masters of what students of public administration call “incentive compatibility.” Through mutually reinforcing sets of intra-judicial and extra-judicial incentives they can make sure that all judicial strategies except prudent “self-restraint” appear personally costly and politically self-defeating. If they wish to simplify matters, they can set up hierarchical systems of appeal that centralize and homogenize judicial rulings (see Shapiro 1981) and that allow them to constrain lower-level judges by controlling the veto player at the top. Fragmentation: Rather than establishing unified judicial systems, authoritarian rulers can “contain judicial activism by engineering fragmented judicial systems” in which executive-dominated “exceptional courts run alongside the regular court system” (Moustafa and Ginsburg 2008: 18). Special courts, often endowed with overlapping jurisdictions with regular courts, facilitate the political control of sensitive cases and arenas of conflict. As long as autocrats retain their ability to transfer politically sensitive cases or categories of cases to courts under their control, they can easily tolerate islands of judicial assertiveness (see Moustafa 2008). Insulation: Even though they pretend to work as closed systems of rule-based dispute arbitration, judicial systems, just like all other state institutions, are embedded in their societal environments (see Migdal 2001). Their capacity to protect citizens “horizontally” against resourceful private actors and “vertically” against public authorities very much depends on the surrounding network of professional and civic associations that are willing and able to challenge powerful actors (see Brinks 2008). By “incapacitating judicial support networks” (Moustafa and Ginsburg 2008: 20) authoritarian rulers can effectively preempt the emergence of judicial challenges.

The menu of local manipulation. Authoritarian governance seldom spells the end of local politics. To pre-empt the emergence of local challenges, central authorities thus face the challenge of devising “institutional mechanisms that minimize the odds that [they] will lose control over local elites” (Landry 2008: 25). Perhaps the most prominent mechanisms are repression, bureaucratic control, accountability, and arbitration.16 ●

In repressive regimes of center–periphery relations, central authorities set up parallel bureaucracies of surveillance and physical punishment, such as the Soviet secret police under Stalin, to terrorize lower-level authorities into subservience.

Shaping the Authoritarian Arena ●

●

●

67

In bureaucratic regimes, central authorities set up territorial layers of government in a hierarchical fashion and strive to control subnational authorities by controlling the “appointment game” (Landry 2008: 40) from top to bottom. In such settings, each unit of subnational government is “critically constrained by the capacity of a hierarchically superior unit to appoint, remove, or dismiss [its] leading officials” (Landry 2008: 79). In accountability regimes, authoritarian governments adopt a sort of new public management approach to center–periphery relations. Instead of micro-managing or closely regulating and monitoring subnational politics, they delegate broad authority to local actors, yet hold them accountable for severe performance failures. The criteria for such result-oriented accountability may be political, like the maintenance of social peace, or non-political, like the achievement of economic growth (see Bravo 2010, Landry 2008: 39, Langston and Díaz-Cayeros 2008). Finally, in arbitration regimes, the authoritarian ruler in the capital city acts as arbiter between rival subnational factions that compete for his favors. Similar to a regional hegemon in international relations, he appears as the overpowering external actor whose intervention tips the internal balance of power within regions and localities. Containing Adversaries

When establishing formal institutions of political pluralism, authoritarian rulers face the challenge of opening up, yet containing (often through repression as their means of last resort), spaces of contention by adversaries. As it is easier to commandeer agents than to contain adversaries, the authoritarian repertoires of manipulation are less standardized and more context-specific with respect to those politically contentious institutional fields (see also overview Table 2.2).17 The menu of media manipulation. While access to “alternative sources of information” (Dahl 1971: 3) represents an essential feature of democracy, misinformation and disinformation represent core features of authoritarianism. To minimize the exposure of citizens to competing constructions of political reality, non-democratic rulers can place restrictions on means of communication, media content, and media consumption. ●

Restrictions on private ownership in the means of production of political information typically take the form of state monopolies in print or electronic mass media. Claiming a full monopoly on legitimate political

68

●

●

Analytical Framework communication, some dictatorial states have gone even further, restricting private access to decentralized means of written communication, such as typewriters, copying machines, computers, and the Internet. Post-production restrictions on media content may take the form of official state censorship or more indirect and informal sanctions against informational transgressions, such as the withdrawal of operating licenses, the harassment of media enterprises by tax agencies, and the beating or assassination of journalists. Both legal censorship and extra-legal intimidation tend to induce self-censorship. To restrict the consumption of available information by citizens, rulers may legally prohibit or materially disable mass access to symbolic products that have been produced outside the bounds of authoritarian control (which includes information distributed by international media).18

The menu of associational manipulation. Repression and cooptation are the most obvious authoritarian strategies to keep citizens from practicing the modern “art of association.” In general terms, authoritarian rulers either work towards the subordinate organization of societal interests, the disorganization of societal actors, or the competitive division of civil society. ●

●

●

Mobilizational single-party regimes and state corporatist regimes are both grounded in the use of hierarchical organization to prevent the emergence of autonomous civil society. By contrast, demobilizing authoritarian regimes that aspire to confine atomized citizens in their private spheres bet on the disorganization of societal forces to achieve popular acquiescence. If civil society constitutes an associational realm autonomous of the state, hierarchy and disorganization represent opposite modes of controlling the birth of civil society: the former establishes organization without autonomy, the latter autonomy without organization. For the purpose of authoritarian containment, vertical control and the disruption of horizontal communication are functionally equivalent. In between these extremes lie divide et impera strategies in which rulers strive to pit existing civil society organizations against each other through the selective dispensation of punishments and favors. We find such intermediate situations, for instance, in the “divided structures of contestation” Ellen Lust-Okar has analyzed for the Middle East (2005).

Institutional manipulation is not a one-shot game. It is a permanent task. When autocrats introduce formally representative institutions, they may come to regret their decisions later on, but cannot reverse them without cost. Once created, macro-institutions tend to be sticky. It is generally cheaper to manip-

Shaping the Authoritarian Arena

69

ulate them than to abolish them. It is easier to keep revising and adapting manipulative choices than institutional selections. Which is what autocrats tend to do. Within their self-made institutional framework, they continually update and adjust their strategies of institutional containment.

N EW I N S T I T U T I O NA L I SM Historians, philosophers, and comparative political scientists have long been aware of the organizational bases of modern authoritarianism. With historical and theoretical nuance, they have been studying institutions of domination in totalitarian and other non-democratic regimes: military juntas, the secret police, prison systems, labor camps, bureaucracies of death, single parties and their ancillary organizations, macroeconomic regimes, state enterprises, welfare bureaucracies, machineries of patronage, machineries of propaganda, systems of public education.19 By contrast, nominally representative institutions have been deemed mostly irrelevant for authoritarian governance. Authoritarian regimes have been assumed to be realms in which formal constraints or “parchment institutions” (Carey 2000) are weightless in the face of factual correlations of power and informal practices of governance. Institutions of representation established in authoritarian regimes been regularly “dismissed as insignificant window dressing” (Gandhi 2008: xxi).

Institutional Relevance Certainly, formal representative institutions make for lovely decorations in the shop windows of authoritarian regimes. Aside from satisfying the aesthetic demands of the unsophisticated public, however, they are likely to carry instrumental value for authoritarian rulers. Unless institutions are somehow useful in coping with the perennial imperatives of governance and survival, why should non-democratic rulers bother to create them? The so-called new institutionalism in political science has revived the empirical study of formal political institutions (rules and organizations). It is founded upon the credo that institutions matter. In the comparative study of politics outside Western Europe, its emergence has been driven to a large extent by the emergence of democratic regimes. Liberal democracies are defined as institutional configurations. By definition, democratic formal institutions carry political weight. Under authoritarian conditions, the relevance of formal

70

Analytical Framework

institutions is less obvious. Still, over the past years, comparative scholars have been turning towards the study of formal representative institutions in nondemocratic regimes as well. Just like neo-institutional approaches to democratic politics, the new institutionalism in the study of authoritarian politics rests upon the theoretical intuition that institutions matter. Even under nondemocratic conditions, formal representative institutions are likely to matter. The comparative literature on authoritarian regimes has begun to take their institutional choices seriously and to examine domain-specific institutional configurations and their underlying strategic logic. Most scholarly attention has focused on the use of multiparty elections by authoritarian rulers. The new institutionalism in the study of non-democratic regimes, however, has gone well beyond the study of electoral autocracies. For instance, in her award-winning analysis of political institutions under dictatorship, Jennifer Gandhi (2008) studied the establishment of legislatures and parties as vehicles of policy concessions that facilitate authoritarian governance and survival (see also Gandhi and Przeworski 2006 and 2007). Barbara Geddes (2005) has analyzed the role of mobilizational regime parties in increasing authoritarian rulers’ odds of deterring or surviving military insurrections by creating counterweights to military establishments (see also Brownlee 2007a and B. Smith 2005). A number of comparative scholars of the Middle East and Northern Africa have examined the regime-supporting roles quasi-autonomous civic associations may perform by playing the role of willing victims of tyranny (see Albrecht 2005, Cavatorta 2009, Garon 2003: 180, Lust-Okar 2005). Even judicial institutions which seem least likely to escape vertical controls and enjoy spaces of autonomy under dictatorship have attracted scholarly attention. The emergent literature on authoritarian legality (“rule by law”) has been examining the multiple roles courts may play in the containment of vertical as well as horizontal threats, in particular through the “judicialization of repression” (Pereira 2008), the imposition of hierarchical controls on administrative agents and political competitors, and the simulation of rule of law as a source of political legitimacy (for an overview, see Moustafa and Ginsburg 2008, see also Barros 2002 and Moustafa 2008). Insofar as court systems work as decentralized arenas of arbitration, they disperse conflicts and deflect responsibility from the political center. Federal arrangements may work in an analogous manner. Miniature dictatorships like the city state of Singapore have no need for political decentralization. By contrast, authoritarian rulers who oversee immense countries like Argentina, Brazil, China, Mexico, and Russia have been developing federal or at least decentralized structures of governance. In large countries, the principal–agent relations that stretch from the capital to distant

Shaping the Authoritarian Arena

71

peripheries are too complex, and the informational advantages of local actors on the ground too large, to permit close oversight and control by the center. Accordingly, the new literature on “decentralized authoritarianism” (Landry 2008) and authoritarian federalism (see Bunce 1999) studies the complex balancing acts central elites perform in granting autonomy to the regions while striving to keep them under control. In post-revolutionary Mexico, for example, presidents delegated sweeping authority over policymaking and personnel selection to state governors, while holding them personally responsible for competent conflict management within their states (see Langston and Díaz-Cayeros 2008). While the “old” institutionalism in studies of non-democratic governance has focused its attention on institutions of domination and the emergent “new” institutionalism on institutions of representation, we should not overlook the forceful presence of anti-institutional or perhaps a-institutional approaches. In particular, economic analyses of authoritarian politics along the paths traced by Gordon Tullock in his seminal Autocracy (1987) have little use for the political science tradition of distinguishing non-democratic regime types on institutional grounds. Striving to reconstruct the rational calculations of individual actors within their strategic environment of contending players, political economists replace the analysis of regimes by the analysis of rulers (also known as dictators or autocrats).20

Institutional Ambivalence If formal representative institutions matter in authoritarian regimes, what do they matter for? The prevailing view in the literature on the political economy of authoritarian institution building is unambiguously instrumental: Just like institutions of domination, formal institutions of representation are instruments of authoritarian rule. They help the average dictator ease his existential problems of governance and survival. They help him elicit cooperation by societal groups and individual actors and diminish the (actual or potential) challenges they pose to his exercise of power. To cite just a few examples: Parties and legislatures allow autocrats “to co-opt potential opposition and maintain the coalitions necessary to remain in power” (Gandhi 2008: 179). They “facilitate power-sharing and thus enhance the survival of authoritarian regimes” (Boix and Svolik 2013: 301). Legislatures “help the dictator to convince the asset holders that their assets will be respected and protected” (Escribà-Folch 2003: 24). In states dependent on natural resources, their creation serves as “a mechanism for the dictator to bribe and split the opposition” (Wright 2008: 322).21

72

Analytical Framework

If institutions are useful, rulers have good motives for adopting them. And the other way round. If rulers are instrumentally rational, they cannot do otherwise than adopt useful institutions, and reject those they consider useless or even harmless. In much of the literature, the axiomatic assumption that actors are rational dictates the logical conclusion that institutions are useful. The assumption of utilitarian rationality compels us to assume that authoritarian rulers benefit (or expect to benefit) from whatever they do, whether they reject or embrace representative institutions. Of course, one may find the assumption of rationality itself to be less than compelling. For instance, the political terror unleashed by Hitler and Stalin made little sense in terms of regime security, economic gain, or success in war. They were profoundly “anti-utilitarian” projects (Arendt 2004: 457, 532). Still, the institutionalization of irrationality seems to be largely confined to the largely historical category of ideological regimes (Weltanschauungsregime). Beyond the dark realms of irrational domination, the cost–benefit calculations of ostensibly “rational” dictators are limited by ignorance or incompetence. Rulers may and often do make mistakes and miscalculate the political consequences of their institutional acts. Still, despite the emotional comfort it offers, the assumption that most dictators act stupidly most of the time cannot provide solid ground for theory-building. If we discard the notion that authoritarian institutional designers suffer from either generalized irrationality or generalized incompetence, we are back to the utilitarian baseline assumption: Authoritarian institutions serve authoritarian interests. They are tools of authoritarian rule. Otherwise, rulers should close them down, or never open them up in the first place. Does this imply that authoritarian institutions are no more than authoritarian tools? No more than pliant instruments at the service of authoritarian interests? Hardly so. The reason is simple: By establishing formally representative institutions, authoritarian rulers establish relations of dependency. To the extent that political institutions empower regime agents, rulers turn dependent on their collaboration. Rather than working as reliable servants of the regime, agents may exploit state institutions for their own purposes. To the extent that representative institutions empower regime adversaries, rulers become dependent on their cooperation. Rather than acting as loyal opposition to the regime, adversaries may either boycott institutions or appropriate them as sites of resistance. If political institutions “have the potential to undermine autocratic rule, why would any incumbent create or tolerate them?” (Gandhi 2008: xvii). The answer is straightforward: Rulers cannot have one without the other. Of course, authoritarian institutional designers dream of “purging ambivalence” (Bauman 1991: 24). No doubt, they would love to grow regime-supportive

Shaping the Authoritarian Arena

73

institutions that preclude regime-subversive possibilities. An authoritarian world without ambivalence seems to be an authoritarian illusion, however. Autocrats cannot establish effective institutional safeguards without accepting the structural risks they involve. This applies even to institutions of domination which they set up as guardians of authoritarian rule. How many dictators have fallen victims of the paramilitary security forces they organize for personal protection? How many have been deposed by factions within the single parties they created as instruments of dominance? Even the totalitarian project of a comprehensive bureaucratization of society in the name of socialism ended up self-defeating. The all-powerful institutions of the Soviet empire “that had defined [the socialist systems] and that were, presumably, to defend them as well, ended up functioning over time to subvert both the regime and the state” (Bunce 1999: 2). This is then the dilemma of authoritarian institutional design: Unless political institutions are granted minimal margins of power and autonomy, they cannot make an independent contribution to authoritarian governance and survival; and as soon as political institutions are granted minimal margins of power and autonomy, they can turn against the dictator. They open up sites of diversion and subversion, open and subterranean, with multiple actors testing in multiple manners the limits of the permissible. In autocracies, then, institutions are arenas of control and cooptation, but also of contention.22 In a standard decision-theoretic framework, actors weight the expected benefits against the expected costs of alternative choices. If actors choose a course of action that promises positive net utility, with benefits outweighing costs, the latter (expected costs) do not simply wither away. When taking their decision, actors accept them as risks. Later on, these risks may or may not materialize. They nevertheless tend to stay with decision-makers as menacing shadows, latent yet real, that infuse decisions with elements of ambivalence.23 Formal representative institutions under autocracy are no exceptions. Representative institutions established by authoritarian rulers may carry high probabilities of producing beneficial outcomes for their authoritarian creators and low probabilities of generating inimical outcomes. Still, it is the presence of latent risks, even in the presence of tangible benefits, that infuses them with ambivalence. On average, authoritarian political institutions seem indeed to fulfill regime-supporting functions. And yet, inevitably, although to variable degrees, they contain seeds of subversion. Institutions are not machines. As they are run by human beings, they cannot be subject to absolute control; and if they were, they would stop serving the purposes of their dictatorial creators. Even when authoritarian institutions work as they are supposed to, absorbing, channelling, dampen-

74

Analytical Framework

ing, deflecting, or dispersing oppositional energies, regime-critical actors may still succeed to some extent in neutralizing these institutions or even appropriating them for their purposes. Even if institutions make autocracy work, and augment the authoritarian ruler’s probability of surviving in office and governing effectively, they still contain the possibility of eroding authoritarian stability and governance. Stressing the risks and ambivalences of representative institutions under authoritarian rule involves a subtle, yet significant, change of perspective. It posits that authoritarian institutions of representation are useful for incumbents—but for their opponents, too (even if to a lesser extent). From this perspective, formal representative institutions matter to the dynamics of authoritarian regimes, not because of the instrumental value they have for rulers, but because of the relative autonomy they possess even under non-democratic conditions. They are instruments of authoritarian rule, but also instruments of resistance to authoritarian rule. They constitute spaces of contention, not just spaces of domination; means of creating uncertainty as much as mechanisms for containing uncertainty. Instead of conceiving them as simple authoritarian tools, we need to comprehend them as arenas of struggle.

C O N C LU S I O N In a world in which liberal–democratic institutions and ideas have become hegemonic, electoral authoritarian regimes represent authoritarianism’s last line of defense. While building a contingent set of repressive, economic, and ideological institutions of domination, they introduce the comprehensive set of formal representative institutions that define modern democracy: legislatures, party pluralism, regular elections at all levels, the rule of law, autonomous public spaces with independent media and civil associations, and territorial divisions of power. Institutions of representation are supposed to divide and constrain state power. To neutralize their divisive and constraining logic, electoral autocracies throw out a wide net of manipulative devices. And yet. Despite comprehensive manipulation, each formal institutional concession to liberal democracy is likely to contain a certain amount of ambiguity. Each formally representative institutional space constitutes a potential arena of struggle: an object of authoritarian manipulation as well as a site of anti-authoritarian resistance. Multiparty elections are embedded in this multifaceted landscape of contested institutions. Salient and centrally located, they form the main arena of struggle—and thus the main arena of manipulation.

Shaping the Authoritarian Arena

75

Notes 1. See, for instance, Bunce and Wolchik (2011), Diamond (2002), Levitsky and Way (2010b), Robertson (2010). 2. See, for example, Linz (2000: 143–151). 3. The foundational texts for the new literature on political regimes and social inequality are Boix (2003) and Acemoglu and Robison (2006), for the political economy of taxation, Levi (1988). On inequality and regime dynamics, see also Manzano (2005), Soifer (2009), and the thematic issue entitled “New Approaches to the Politics of Inequality in Developing Countries,” PS: Political Science & Politics 42/4 (October 2009). For an overview on taxation and authoritarian regimes, see Escribà-Folch (2007: Ch. 5). See also Gehlbach (2008) and Bräutigam, Fjeldstadt, and Moore (2008). 4. The literature on clientelism and patronage is burgeoning. Kitschelt and Wilkinson (2007) offer a fine collection of essays on clientelism and electoral competition. On patronage as the basis of hegemonic regimes, see Greene (2007) and Magaloni (2006). 5. Olson (2000) is a seminal text on predatory states. After its early formulation by Karl (1997), the literature on political regimes and natural resources keeps expanding. See, for example, Dunning (2008), Fish (2005: Ch. 5), Haber and Menaldo (2010), Ross (2001), Smith (2004) and (2007), Snyder (2006b), and Ulfelder (2007). 6. Classical treatments of the Nazi and Soviet totalitarian regimes are Arendt (2004) and Friedrich and Brzezinski (1961). Orlando Figes’ oral history of Stalin’s Russia (2007) offers a fascinating account of totalitarianism from the perspective of ordinary citizens. 7. For example, Tullock (1987) and Wintrobe (1998). 8. Examples are Bellin (2004) and Levitsky and Way (2010b). Harking back to the earlier literature on political revolutions (e.g. Skocpol 1979 and Goodwin 2001), both emphasize the coercive capacities of authoritarian states. More recently, varieties in regime outcomes during the Arab Spring have reminded us of the simple fact that state security agencies are the final arbiters of authoritarian regime persistence in the face of mass protest (see e.g. Bellin 2012: 129–130, Barany 2011). 9. See, for instance, Slater (2010), Snyder (2006a), and Way (2006). 10. Examples are Gilley (2009), Gerschewski et al. (2013) and Kailitz (2012). 11. Drake (2009) offers an excellent overview of constitutional struggles in 19th- and 20th-century Latin America. 12. As a matter of fact, in interwar Europe, representative institutions were not just murdered. Some of them committed assisted suicide. On the “collective abdications” of the parliaments of Weimar Germany and Vichy France, see Ermakoff (2008). 13. After the death of Stalin in 1953, the ussr and its satellites congealed into “posttotalitarian” facades. Few totalitarian regimes have emerged since then (with North Korea representing the last totalitarian bastion in the contemporary world),

76

14.

15.

16. 17.

18.

19. 20. 21.

22.

23.

Analytical Framework although we have been witnessing recurrent relapses into “totalitarian solutions” (Hannah Arendt), as in China’s Cultural Revolution (1966–76) and the 1994 genocide in Rwanda. Only their degree of decentralization is contingent. The necessity of delegating authority to local actors is largely a function of country size. Regardless of regime subtype, large autocracies tend to adopt decentralized or even federal schemes of territorial organization. Examples are pre- and post-revolutionary Mexico, the Soviet Union, military Brazil and Argentina, and contemporary China. On the “natural feebleness” of the judiciary as the “least dangerous” branch of government, see the Federalist Papers No. 78. My outline of authoritarian strategies of judicial containment largely follows Moustafa and Ginsburg (2008: 14–21), even if I reframe and relabel some of their analytic categories. Note that the issues of institutional creation and design I discuss for the national level reappear at subnational levels. On the manipulation of parties and elections, see Chapter 3. Throughout the book, I use the notion of manipulation as a catch-all term for all forms of authoritarian interference (carried out by purposeful agents of the state) with nominally representative institutions. Since it comprises a wide range of practices, not all of them resonate well with its semantic connotations. Above all, in ordinary language, the notion of “manipulation” suggests “the exercise of subtle, underhand, or devious influence” (Oxford English Dictionary). Several strategies I describe as authoritarian manipulation, however, are not backstage maneuvers hidden from the public eye, but open acts of democratic norm violation, like the violent treatment of dissidents or the banning of candidates from electoral competition. On efforts by authoritarian regimes to control the new information and communication technologies, see Diamond (2010), Kalathil and Boas (2003), and Zheng (2010). Among many others, see Bunce (1999), Friedrich and Brzezinski (1961), Geddes (1999), Haber (2008), O’Donnell (1973), and Snyder and Mahoney (1999). See, for instance, Escribà-Folch (2007) and Svolik (2009). Suspicions of endogeneity are never entirely dispelled, though. The creation of formal institutions may be an effect of durable regimes, rather than an independent cause of regime duration (see Pepinsky 2007b). In authoritarian Brazil under military rule, for example, “lawyers stretched the boundaries of permissible activity and speech within national security law” (Pereira 2008: 35), while journalists defied formal censorship, “trying to publish content that [was] taboo under either explicit or implicit constraints” (Stein 2007: 4). In common usage, the psychological term ambivalence does not describe objective attributes of the outside world, but mental states of individuals who harbor “contradictory emotions or attitudes (as love and hatred) towards a person or thing” (Oxford English Dictionary, 2nd edition). Here, I invert its empirical reference: the “ambivalence” of authoritarian institutions does not lie in the subjective observer, but in the objective world. Their contradictory quality does not arise from psychological, but political realities.

3 Shaping the Electoral Arena The Janus-faced concept of electoral authoritarian regimes is anchored in formal institutions (multiparty elections) as well as actual practices (electoral manipulation). On the institutional side, these regimes establish regular multiparty elections with universal suffrage at all levels of state power. Their formal institutions of democratic representation distinguish them from other types of authoritarian rule. On the practical side, they subvert the democratic spirit of competitive elections through severe and systematic manipulation. Their actual practices of manipulation distinguish them from democratic electoral regimes. These conceptual boundary lines have causal implications. They rest upon the twin assumption that authoritarian regimes with multiparty elections differ in their internal dynamics from democracies as well as from non-electoral autocracies. Still, the dividing line between electoral autocracies and electoral democracies is essentially normative. The defining criterion of electoral manipulation is the violation of essential democratic norms. A clear notion of electoral manipulation therefore requires a clear conceptualization of democratic norms and authoritarian offenses. In this chapter, I strive to provide these conceptual foundations of electoral authoritarian rule. After situating electoral autocracies within the broad spectrum of political regimes, I outline their defining formal–institutional properties. In continuation, I draw the normative borderline between democratic and authoritarian elections. I define a set of essential conditions that democratic elections have to fulfill (“the chain of democratic choice”) and describe a set of contingent authoritarian practices that undermine these conditions (“the repertoire of electoral manipulation”). The borderline between electoral democracies and electoral autocracies is often blurred and controversial. As I argue, we should comprehend it as “essentially contested” (Gallie 1956). I conclude by tracing an internal boundary within the family of electoral authoritarian regimes: the distinction between solidly hegemonic and more fluid competitive regimes.

78

Analytical Framework ELECTORAL AUTHORITARIAN REGIMES

Given their contradictory mix of democratic procedures and authoritarian practices, electoral authoritarian regimes have unsettled the conceptual routines of comparative politics. Their global spread after the end of the Cold War has triggered a wave of conceptual dispute and conceptual innovation. Within these debates, the notion of electoral authoritarianism has been somewhat polemical since it compelled us to redraw our conceptual maps. Originating from a twin process of conceptual differentiation, it introduced two new conceptual boundaries: the dividing line that separates electoral autocracies from electoral democracies and the one that separates them from closed autocracies.

The Spectrum of Regimes The idea of electoral authoritarianism takes seriously the authoritarian quality these regimes possess as well as the electoral institutions they establish. The emphasis on authoritarianism serves to distinguish them from electoral democracies, the emphasis on elections to set them apart from “closed” autocracies. Figure 3.1 illustrates the conceptual space they are located in. Alongside electoral democracies, they occupy an intermediate position in a spectrum of political regimes whose poles are formed by liberal democracies and closed autocracies, respectively. In the democratic neighborhood of electoral authoritarian regimes lie electoral democracies. These regimes lack some attributes of liberal democracy (such as checks and balances, bureaucratic integrity, and an impartial judiciary), but do conduct free and fair elections, which electoral autocracies don’t. Drawing a dividing line between democratic and authoritarian elections, and thus recognizing that electoral regimes can be authoritarian, has been the first act of border demarcation in the conceptualization of electoral authoritarianism.

Closed authoritarianism

Electoral authoritarianism

Non-democratic regimes

Figure 3.1. The spectrum of political regimes

Electoral democracy

Liberal democracy

Democratic regimes

Shaping the Electoral Arena

79

In the non-democratic neighborhood of electoral autocracies lie closed autocracies. This residual category designates all non-democratic regimes that refrain from enacting multiparty elections as the official route of access to national executive power. Drawing a dividing line between election-based and non-electoral authoritarianism, and thus recognizing that authoritarian regimes can be electoral, has been the second act of border demarcation in the conceptualization of electoral authoritarian regimes.

The Democratic Borderline Political democratization has been a disappointing process. Its outcomes have almost invariably fallen short of democratic hopes and expectations. The great majority of the new democracies that have been born since the early days of the “third wave” of global democratization (Huntington 1991) have been “democracies with adjectives” (Collier and Levitsky 1997), “defective democracies” (Merkel 2004) that fall short of our liberal–democratic ideals. Within the wide family of “diminished subtypes” of democracy (Collier and Levitsky 1997) we should distinguish two subgroups: Some fall short of standards of liberal democracy. They suffer from problems of democratic quality. Others fall short of standards of electoral democracy. They suffer from problems of democratic essence. x Electoral democracies possess all essential attributes of electoral democracy, yet lack attributes that are essential for liberal democracy. The normative yardstick of their defects lies in their distance from high-quality democracy. Examples of such “electorally competitive regimes” (Bermeo 2003: 5) with quality problems are “patrimonial democracies” that fail to establish rational bureaucracies (see Mazzuca 2007) and “clientelist democracies” that are weak on programmatic party politics (see Kitschelt 2000). x Borderline democracies, by contrast, do not fully possess all essential attributes of electoral democracy. One or more of these attributes they display in a limited or partial manner only. The normative yardstick of their defects lies in their proximity to authoritarian regimes. Examples of such democratic regimes with damaged institutional cores are “tutelary democracies” (Shils 1960) in which non-elected actors restrict the sovereignty of elected authorities and “exclusionary democracies” (Remmer 1985) in which significant “segments of all adult citizens are excluded” from suffrage rights (see Merkel 2004: 49). The debate on electoral authoritarianism transcends issues of democratic quality. It does not concern the desirable attributes of high-quality democracy,

80

Analytical Framework

but the indispensable attributes of minimal democracy. It is not merely deficient, but severely damaged democracies that provoke disputes about the demarcation between “democracies with adjectives” and “autocracies with adjectives.” Along the foggy border area that separates electoral democracies and electoral autocracies, their differences may seem subtle. Borderline democracies do not fully comply with each requisite of electoral democracy; with one or several, their compliance is only partial. In contradistinction, electoral autocracies fully violate at least one of these requisites; with one or several, their non-compliance is severe and systematic. The partial violation of at least one essential attribute of democracy qualifies an electoral regime as “diminished subtype” of democracy; its full violation qualifies it as autocracy.1 When “democratic defects” cancel the democratic essence of electoral regimes, we should call the animal by its name: autocracy. Clinging to the notion of democracy (even if qualified by disqualifying adjectives) would weaken our sense of authoritarian realities, rather than sharpening our grasp of democratic deficits. When we conceptualize non-democratic regimes as instances of democracy, however deficient, we stumble into the methodological pitfall of “conceptual stretching” (Sartori 1984).2 For instance, when we describe Russia under Putin, Egypt under Mubarak, Zimbabwe under Mugabe, or the Islamic Republic of Iran as “democratic” regimes (albeit “illiberal” ones), we stretch and strain the modern meaning of democracy beyond recognition (see Zakaria 2003). Similar concerns of conceptual validity apply to the notion of mixed regimes. Conscious of the dangers of conceptual stretching, some authors have been treating the substandard electoral regimes that inhabit the contemporary world as genuine mid-points between democracy and authoritarianism. As these regimes combine democratic and authoritarian features, scholars locate them at the very center of the conceptual spectrum and as a result consider them to be neither democratic nor authoritarian. Concepts such as “hybrid regimes” (Diamond 2002), “mixed regimes” (Mansfield and Snyder 2005: 41), “semi-democracy” (P. Smith 2005), “semi-authoritarianism” (Ottaway 2003), “semi-dictatorship” (Brooker 2000: 252), and “the gray zone” (Carothers 2002) express the idea of genuine institutional compromises situated in the messy middle ground between democracy and dictatorship. Teleological “concepts of movement” (Koselleck 1979), like “transitional regimes,” “democratizing regimes,” or “emergent democracies,” reflect similar conceptual ambiguities (for a critical review, see Carothers 2002). As electoral autocracies integrate formal institutions of representation into their systems of domination, they can fruitfully be described as hybrid regimes. Yet, they are hybrid authoritarian regimes, rather than unclassifiable mixed regimes that inhabit a no man’s land between democracy and

Shaping the Electoral Arena

81

authoritarianism. The terminology of diminished subtypes of democracy and hybrid regimes recognizes deviations from democratic minimum standards. The notion of electoral authoritarianism identifies these deviations as instances of authoritarian governance. This also holds for neighboring concepts such as “pseudodemocracy” (Diamond, Linz, and Lipset 1995: 8), “disguised dictatorship” (Brooker 2000: 228), and “competitive authoritarianism” (Levitsky and Way 2002). These and similar conceptualizations of new forms of “authoritarianism with adjectives” abandon the assumption that the regimes in question are still anchored in liberal–democratic principles. They describe them as instances of non-democratic rule that display “the trappings but not the substance of effective democratic participation” (Marshall and Jaggers 2002: 12). They involve the claim that these electoral regimes are neither democratic nor democratizing nor ambiguous, but plainly authoritarian—albeit in ways that depart from other forms of modern authoritarian rule. The Authoritarian Borderline Modern non-democratic rulers have been using the political technology of popular referenda and mass elections for a long time. The lineage of dictators who ask their subjects to formally consent to their oppression is long. Napoleon iii and Bismarck come to mind, Hitler and Stalin, Franco and Salazar, Somoza and Marcos, and many others. Even before the end of the Cold War, political elections were almost universally in use. By the late 1970s, they were held “in nearly every country in the world” (Hermet, Rose, and Rouquié 1978: viii). Elections of any kind, elections without adjectives, are likely to matter (somehow) in all kinds of authoritarian regimes. However, some categories of elections are likely to matter more than others. The notion of electoral authoritarianism involves the causal assumption that some types of elections are prone to develop a logic and weight of their own, capable of defining the internal dynamics of authoritarian regimes (and thus justifying the creation of a category of regimes distinct from dictatorships that refuse to introduce them). They involve a level of “electoral exposure” (Cox 2007) that changes the nature of the authoritarian game. Which kinds of elections are likely to do so? According to my stipulative definition, elections in “electoral authoritarian regimes” carry distinctive formal–institutional properties: x Political elections. Unlike Bonapartist regimes that orchestrate occasional plebiscites to demonstrate popular consent on constitutional matters or

82

x

x

x

x

Analytical Framework policy issues, electoral autocracies invite citizens to partake in electoral processes that serve as official selection devices for highest office. Regular elections. The concept of political regimes presupposes some minimum of continuity across time. Occasional elections or aborted elections followed by military coups or civil war do not constitute electoral authoritarian regimes, but electoral authoritarian “situations” (Linz 1973). Rather than patterns, they define exceptions. Inclusive elections. Unlike competitive oligarchies, as in 19th-century Latin America or South Africa under apartheid, contemporary electoral autocracies do not control elections by restricting the franchise to social minorities, but operate on the basis of universal franchise. They are institutional responses to mass politics, rather than its suppression. High-level elections. Some authoritarian regimes hold “limited” legislative or subnational elections “in which national power is not directly at stake” (Thompson and Kuntz 2006: 114). Under such limitations of scope, elections are unlikely to develop into serious contests for power. They will tend to serve as arenas of competition for access to state patronage (see Lust-Okar 2009, Weber 1988: 536–545). Electoral authoritarian regimes, by contrast, subject the head of government to electoral confirmation, be it though direct elections (in presidential systems) or indirect ones (in parliamentary systems). Multiparty elections. Last but not least, unlike single-party regimes that organize one-party (or national front) elections, either with or without intra-party competition, electoral authoritarian regimes allow for organized dissidence in the form of multiparty competition.

By designating qualified “elections with adjectives” as defining attributes of electoral autocracies, we place non-democratic regimes that lack these types of elections into the residual category of “closed” regimes. By drawing one boundary, we blur others. Yet, while the omnibus concept of “closed” autocracies serves the purpose of highlighting the distinctive institutional profile of electoral autocracies, it is not meant to deny the profound differences that exist among closed regimes, nor the importance that electoral processes may possess for these regimes, nor, for that matter, the relevance non-electoral factors may have in electoral authoritarian regimes. Where exactly to draw the borderline between electoral and closed autocracies is primarily an empirical matter; it depends on the causal relevance of the electoral arena. By contrast, where exactly to draw the line between electoral autocracies and democracies is primarily a normative matter; it depends on the authoritarian quality of the electoral arena. Let us now turn our attention to these normative issues.

Shaping the Electoral Arena

83

THE CHAIN OF DEMO CRATIC CHOICE A widespread, twofold consensus exists in the scholarly community regarding the role of elections for democratic governance. On the one hand, elections are essential ingredients of liberal democracy. No elections, no democracy. On the other hand, not any theatrical performance does the trick. To qualify as democratic, elections have to be competitive, free, and fair. In other words, while recognizing the centrality of elections, the current liberal–democratic mainstream is sensitive to the “fallacy of electoralism” (Karl 1995). Democratic elections are commonly considered to form the core of minimal procedural conceptions of democracy. At close examination, though, it turns out that the electoral democratic minimum involves a demanding set of normative conditions. In essence, democratic elections are mechanisms of social choice under conditions of freedom and equality. To qualify as democratic, they must offer effective choices of political authorities among a community of free and equal citizens. According to Robert Dahl’s much-cited formulation, this democratic ideal requires that all citizens enjoy “unimpaired opportunities” to “formulate” their political preferences, to “signify” them to each other, and to have them “weighed equally” in public decision-making (Dahl 1971: 2). Building upon Dahl’s abstract stipulation, I wish to delineate seven more concrete conditions that must obtain if regular elections are to fulfill the promise of effective democratic choice. The ensuing “chain of democratic choice” is summarized in Table 3.1. It covers every stage of electoral choice, from its original structure to its final consequences. Together, these conditions form a metaphorical chain which, like a real chain, holds together only so long as each of its links remains whole and unbroken. 1. Empowerment. Elections are routes to power. Their purpose lies in the binding selection of the polity’s “most powerful collective decision makers” (Huntington 1991: 7). Democratic elections are about citizens wielding power. They are the institutional expression of rule by the people. In their role as voters, citizens do not pick winners of beauty contests or responses to marketing surveys. They authorize access to state power.3 2. Freedom of supply. The idea of democratic choice presupposes the free formation of alternatives, the freedom to offer contrasting visions of the common good, conflicting policy options, and diverse sets of candidates and party organizations on the electoral marketplace. Modern democracy institutionalizes pluralism and competition. It is adversarial by design. Acclamatory “elections without choice” (Hermet, Rouquié, and Linz 1978) do not qualify as democratic, nor do elections in which

84

Analytical Framework

Table 3.1. The chain of democratic choice and the repertoire of electoral manipulation Dimensions of electoral choice

Normative premises of democratic choice

Strategies of electoral manipulation

1

Object of choice

Empowerment: Democratic • Reserved positions: limiting the elections involve scope of elective offices the delegation of • Reserved domains: limiting the decision-making authority. jurisdiction of elective offices

2

Range of choice

Freedom of supply: Citizens must be free to form, join, and support conflicting parties, candidates, and policies.

3

Formation of preferences

Freedom of demand: Citizens • Repression: restricting political and must be able to learn civil liberties about available alternatives • Unfairness: restricting access to through alternative sources media and money of information.

4

Agents of choice

Inclusion: Democracy assigns equal rights of participation to all full members of the political community.

• Formal disenfranchisement: legal suffrage restrictions • Informal disenfranchisement: practical suffrage restrictions

5

Expression of preferences

Insulation: Citizens must be free to express their electoral preferences.

• Coercion: voter intimidation • Corruption: vote buying

6

Aggregation of preferences

Integrity: One person, one vote. The democratic ideal of equality demands weighting votes equally.

• Redistributive institutions of electoral governance: biased rules of electoral competition and organization • Redistributive practices of electoral governance: electoral fraud and impunity

7

Consequences of choice

Decisiveness: Elections without consequences do not qualify as democratic.

• Tutelage: preventing elected officers from exercising their constitutional powers • Reversion: preventing victors from taking office, or elected officers from concluding their constitutional terms

• Exclusion of opposition actors: restricting access to electoral arena • Fragmentation of opposition actors: disorganizing electoral dissidence • Subversion: simulating and controlling opposition parties

Shaping the Electoral Arena

3.

4.

5.

6.

7.

85

choice is confined to a narrow menu of state-licensed options. The range of available alternatives must not be engineered by a manipulative government, but determined by active citizens themselves within a framework of fair and universal rules. Freedom of demand. Democratic elections presuppose the free formation of voter preferences. Citizens who vote on the basis of state-induced preferences are no less constrained than those who are forced to choose from a manipulated set of alternatives. Modern democracy assumes that all citizens, regardless of their formal education, wealth, social status, ethnic membership, and religious convictions, possess equal faculties of autonomous decision-making. To use their abilities of political judgment, voters must be able to learn about available choices. They may decide not to vote (thus failing to register their political demands), or to vote in complete ignorance (thus depriving their political demands of a rational core). For their collective choice to be free and democratic, though, voters need to possess the structural opportunity for getting informed. Only access to plural sources of information can avail that opportunity (see Dahl 1971: 3). To the extent that competing parties and candidates (as well as citizens themselves) are denied free and fair access to the public space, the will of the people as expressed at the ballot box turns into an expression of structurally induced ignorance. Inclusion. In the contemporary world, democracy demands universal suffrage. Restrictions of the franchise once commonly applied on the basis of property, education, gender, religion, or ethnicity are no longer deemed legitimate. The modern demos is supposed to include the entire adult population of permanent residents in a given territory. Empirical exceptions, such as convicted felons, recent immigrants, and persons afflicted by severe mental illness, are few and contested (see López-Guerra 2005). Insulation. Once citizens have freely formed their preferences, they must be able to express them just as freely. The secret ballot is designed to shield them from undue outside pressures, whether in the form of coercion, bribery, or social disapproval. Integrity. Once citizens have given free expression to their will at the polls, competent and neutral election management must count their votes honestly and weigh them equally. Without bureaucratic integrity and professionalism in the administration of elections, the democratic principle of “one person, one vote” remains an empty aspiration. Decisiveness. Just like elections without choice, elections without consequences are not democratic. Winning parties and candidates must be able

86

Analytical Framework to assume office, exercise power, and conclude their terms in accordance with constitutional rules. Here, the democratic circle closes. Elections must be “decisive” (O’Donnell 2001: 13) ex ante as well as ex post. Unless they invest winners with effective decision-making power, they turn into vacuous exercises, into fraudulent and frustrating, rather than empowering and dignifying, experiences for citizens.

The notion of a logical sequence of necessary conditions implies that elections may be considered democratic if and only if they fulfill each of them. The mathematical analogy is multiplication by zero, rather than addition. Partial compliance with democratic norms does not add up to partial democracy. Gross violation of any one condition invalidates fulfillment of all others. The links in the democratic chain are of varying strength. Some are more fragile, others more resilient. National governments may sever some of them through single-handed decisions. To cut others, they depend on the cooperation of large numbers of state agents and citizens. Nevertheless, if the chain of democratic choice is broken anywhere, be it with great ease or great effort, elections become not less democratic but undemocratic. Concept analyses in the social sciences often distinguish between two approaches to concept formation. The “classical” approach defines concepts by identifying their necessary attributes. All attributes must be present for the concept to be applicable. The “family resemblance” approach defines concepts by identifying sets of typical attributes that tend to go together, but need not be present in all instances of the concept. Attributes are mutually substitutive. The concept is applicable even if only some of them are present.4 Often, we find classical definitions at the most general level of conceptualization and family resemblances at more specific levels. Phenomena defined by abstract criteria often look dissimilar in the concrete world of appearances. Just like families do (see Schedler 2011). By conceptualizing the normative requisites of democratic elections as a coherent chain of choice I introduce a rigidly classic definition of electoral democracies: All links are necessary; each of them must remain intact for an electoral process to be judged as democratic. My conceptualization of electoral autocracies, by contrast, blends classical elements (at the definitional level) with family resemblances (at the operational level). In general terms, electoral authoritarian regimes share a set of necessary attributes. They hold non-democratic multiparty elections at the national level. In more concrete terms, however, they need not look alike. To render elections non-democratic, they can pick from a broad repertoire of functionally equivalent authoritarian strategies. They can break the chain of democratic choice at any point. In principle, if the seven links may be severed either individually or jointly, we obtain

Shaping the Electoral Arena

87

27 = 128 possible electoral authoritarian configurations (minus one, the case of democratic integrity without manipulation). If we consider choices over time, the number of hypothetical manipulative sequences reaches 7! = 5,040 (even without taking into account strategic options within each category of norm violations). Even if they share common non-democratic principles, we should not expect the members of the electoral authoritarian regime family to share a common physiognomy.

THE REPERTOIRE OF ELECTORAL MANIPULATION The normative chain of democratic choice is meant to be complete. It takes into account every step of electoral choice, from its object to its impact. There are no links to be added and none to be taken away. The toolbox of authoritarian norm violations, by contrast, is open-ended. Authoritarian rulers may always devise new tactics to sever the chain of democratic choice at one or multiple points. Even as regimes institutionalize and settle into authoritarian routines they may be innovating—fine-tuning and expanding their repertoire of electoral control.5 The limits to authoritarian innovation and imagination are not logical, but empirical. My discussion of authoritarian maneuvers (which follows step by step the seven links of the democratic chain) should therefore be understood as inherently incomplete. Here as in the world of sport games, the notion of “foul play” does not describe a finite catalogue of offenses, but an “open-ended range of behavior” (Connolly 1993: 182).

1. Disempowerment: Reserved Positions and Policies While democratic elections are routes to power, authoritarian elections may be devised as paths to powerlessness. Regimes can pre-empt potential threats emanating from popular elections by circumscribing the scope of elective office through the use of reserved positions. Some authoritarian regimes allow voters to fill subordinate positions of public authority, while keeping the “high center” of power shut off from electoral pressures. Local elections in Taiwan until the early 1990s, village elections in China since 1988, as well as legislative elections in contemporary Morocco and authoritarian Brazil (1964–85), exemplify such strategies of electoral confinement.6 For the ruling party, limited elections to local or legislative office may be useful devices. They may generate valuable information on citizens and candidates, they may keep elites

88

Analytical Framework

circulating and expectant, and they may serve as conflict-resolving mechanisms to adjudicate patronage among factions within the ruling coalition. For citizens and opposition actors, limited elections are of limited relevance, as they do not offer the formal possibility of contesting the commanding heights of state power.7 Authoritarians may also keep elected officials from acquiring real power by delineating reserved policy domains that cut these officials off from effective decision-making in certain policy fields (Valenzuela 1992: 64–66). Official posts are filled through elections, but crucial policy areas are removed from their jurisdiction. Guatemala in the late 1980s, Chile after Pinochet, and Turkey until recently furnish examples of “fencing-off ” strategies in which the military protect sensitive policy domains from democratic interference.

2. Supply-Side Restrictions: Exclusion, Division, and Subversion At times, authoritarian victories are the fruit of opposition failure, with incumbents emerging victorious from elections not thanks to their own “cleverness but the ineptitude of [their] opponents” (McFaul 2000: 23). More often than not, however, authoritarian rulers find ways to engineer the failure of opposition parties by imposing supply-side restrictions on the electoral market. They ban, split, or manufacture opposition groups. Exclusion. Authoritarian incumbents are always tempted to restrict the electoral arena by keeping out unfriendly competitors. Their means of exclusion range from institutional violence to institutional design. The actual or attempted physical elimination of salient opponents, as in Togo in 1991, Armenia in 1994, Belarus in 1999, Ukraine in 2004, and Pakistan in 2007, is only the most extreme form of candidate screening. Much “more common” is the “milder technique” (Monga 1997: 159) of banning parties and disqualifying candidates. The design of discriminatory legal provisions provides a common route to shut uncomfortable contenders out of the electoral arena. Ruling parties often hand-tailor legal instruments that permit them to exclude specific opponents, be they individuals or social groups, potential or actual adversaries. Examples abound. The electoral laws of post-revolutionary Mexico kept regional and religious parties as well as independent candidates out of the electoral arena (Molinar 1993: 27–28). In Egypt under Mubarak, the party law practically forbade all partisan activity based on non-harmonious views of society. It banned “any parties established on the basis of class, sect, community, geography, race, origin, religion, or creed” (Lust-Okar 2005: 65). In contemporary Iran, the Islamic revolutionary credentials of candidates are subject to tight

Shaping the Electoral Arena

89

layered evaluation by state agencies and religious authorities (Boroumand and Boroumand 2000: 117–119). After his 1994 coup, Gambian President Yahya Jammeh pushed through a new constitution that expelled the country’s entire political elite from the electoral arena (see Wiseman 1998: 69). In the 1990s in Côte d’Ivoire, Kenya, and Zambia, incumbent presidents engaged in blatant “manipulation of citizenship” (Daniel and Southall 1999: 43), using custom-made “nationality clauses” to prevent their most serious competitors from running (Monga 1997: 160).8 Instead of devising biased electoral rules, authoritarian governments can keep uncomfortable candidates out of the electoral arena through ostensibly non-political administrative means. They may impose burdensome registration requirements on parties and candidates, such as unreasonable amounts of notarized signatures and financial deposits. Or they may apply reasonable requirements in discriminatory manners. In the post-Soviet world as well as in the electoral authoritarian regimes of sub-Saharan Africa and Southeast Asia, “governments have arbitrarily deregistered parties and disqualified candidates, actions usually supported by pliable election commissions” (Case 2006: 101). In many instances, submissive administrative and judicial agencies show high levels of competence in banning parties and candidates on apparently technical legal grounds. Fabricated offenses are often minuscule, sometimes even “bizarre” (Wilson 2005: 82). Candidates are denied participation because of illegible signatures, misspelled names, missing photocopies, or minor factual errors.9 Guarding the gates to electoral competition through legal formalism may take rougher forms, too. The arbitrary use of the law as a weapon against political dissent frequently extends to tax bureaucracies, auditing agencies, and anti-corruption bodies. Masquerading as guardians of public integrity, these state agencies turn into willing executioners of political blackmail. Leveling legal charges like tax evasion, corruption, and embezzlement, they keep potential contenders out of the electoral arena through police arrest, judicial prosecution, and imprisonment. Ukraine under Kuchma, Belarus under Lukashenka, and Russia under Putin exemplify the selective enforcement of law and the partisan use of public agencies to pre-empt opposition figures from entering electoral competition.10 Division. In authoritarian Brazil (1964–85), military rulers banned pre-existing political parties but “their quasi-resurgence was tolerated under the all-embracing rubric of an official two-party system” (O’Donnell and Schmitter 1986: 22). In the course of several electoral contests, however, the generals came to realize that “the compulsory two-party structure ensured opposition unity” (Skidmore 1989: 35). In reaction, they abandoned their pretension of legislating the party system into a bipolar format. Instead they turned their creative energies to a project many electoral autocrats pursue

90

Analytical Framework

with enthusiasm: encouraging the fragmentation of the opposition camp (Skidmore 1989: 22).11 Since at least the time of the ancient Romans and their policy of divide et impera, authoritarian rulers have been seeking to aggravate the collective action problems of their adversaries. To outlaw opposition parties and allow only unaffiliated individuals to contest elections, as in Taiwan until 1989, or to ban parties in general, as in the Islamic Republic of Iran, represents the most radical device for disorganizing electoral dissidence. Yet even in electoral authoritarian regimes that admit multiple political parties, rulers can establish institutional environments that obstruct the creation of effective opposition parties. In particular, lopsided institutional rules that favor independent candidates and weak legislatures engender fragile, fleeting, and amorphous parties. Not just opposition parties, governing parties too may remain weak and ineffectual if overpowering presidents succeed in institutionalizing electoral politics on the basis of caudillos, clans, and cliques. Several of the Soviet successor states have taken the path of disempowering opposition parties by pulverizing the field of electoral competition. Rather than “inchoate” party systems, as identified by the literature on party-systemic institutionalization (see Mainwaring and Scully 1995), they have created anti-party systems. Rather than founding hegemonic parties to secure their political survival, they conduct multiparty elections in a manner that practically dispenses with political parties (including governing parties that might develop into countervailing forces to presidential power). In essence, rather than either proscribing or employing political parties, they have rendered them useless. The results are disorganized elections inundated and dominated by independent candidates. In Belarus under Alyaksandr Lukashenka, for instance, nominal independents conquered 48 percent of legislative seats in the 1995 elections, 73.6 percent in 2000, 89.1 percent in 2004, and 93.6 percent in 2008. In Kyrgyzstan under Askar Akeyev, independent candidates occupied 64.8 percent of legislative seats after the 1995 elections, 69.5 percent in 2000, and a full 100 percent in 2005.12 The disorganized autocracies of Soviet successor republics are extreme instances of government-induced coordination failure. Peru’s “partyless party system” under Fujimori represented a similar instance of authoritarian electoral competition without parties (see Levitsky 1999 and Roberts 2006). In more structured authoritarian party systems, electoral laws offer less dramatic tools to keep opposition parties from coalescing. In Mexico under the pri, for instance, elements of proportional representation and legal obstacles to the formation of electoral alliances served this purpose (see Díaz-Cayeros and Magaloni 2004).

Shaping the Electoral Arena

91

While formal institutions may be valuable devices for seeding division within the opposition camp, incumbents often strive to weaken and divide their adversaries through a wide array of informal practices. They destroy opposition unity through corruption and intimidation, as Kenya’s President Daniel arap Moi did when “harassing or bribing the leaders of any new parties until splits occurred or key members defected” (Barkan and Ng’ethe 1998: 33). They induce fissures in the opposition camp by awarding cabinet positions or other high public office to distinguished opposition figures. They manipulate public opinion polls to prevent opposition actors from coordinating, as security adviser Vladimiro Montesinos did in Peru under Fujimori (McClintock 2006: 256). To prevent opposition voter coordination, rather than opposition party coordination, some authoritarian rulers start creating (nominal) opposition parties themselves. In mild versions of government-sponsored “opposition party” formation, they rely on positive inducements. They encourage “the creation of ‘friendly’ opposition parties” (Sithole 1997: 136) through personal networks and monetary incentives. In a recurrent plot designed to confuse votes, governments resort to advanced biotechnology and “clone” opposition candidates. Electoral “clones” or “nuisance candidates” run under the same name (or a very similar one) as rival candidates in the same district. They have been deployed in places as distant as the Philippines under Marcos and Ukraine under Kuchma (Case 2006: 102, Diuk and Gongadze 2002: 161). Subversion. To the degree that governments actually create and control opposition parties, rather than “merely” obstructing, debilitating, and fragmenting them, they engage in a profoundly subversive enterprise: the simulation of opposition. Weak and disunited opposition forces are opposition forces nevertheless. They may be marginal and impotent, yet they keep alive voices of dissent and pluralism in the public arena. They are recognizable members of a recognizable opposition, that is, a symbolic community of political actors that are (on whatever grounds) distinguishable from the ruling party, the ruling clique, or the ruling whoever. By contrast, nominal opposition parties that are products and puppets of the government, invented or intervened by it, strip multiparty elections of their core meaning. If the distinction between government and opposition collapses, if all parties are no more than theatrical performances scripted and directed by the chief executive, the electoral arena mutates from an arena of competition (however skewed) into an arena of acclamation (however disguised). The “virtual politics” practiced in electoral authoritarian Russia exemplifies the systematic subversion of multiparty competition by the state. It includes the regular invention and simulation of

92

Analytical Framework

protest movements and opposition parties (see Aspinall 2005: Chs. 3 and 6, Robertson 2010, Wilson 2005).

3. Demand-Side Restrictions: Repression and Unfairness In markets for goods and services, consumer demand presupposes knowledge about available supply. Even when basic human needs are at stake, like food and shelter, consumers capable and willing to pay still need to know who offers what, where, and how in order to translate their needs into effective demand for available products. Ignorance kills markets. It prevents demand from meeting supply. The same applies to electoral markets. If citizens who demand alternative policies know nothing about the politicians who offer them, the former are unlikely to vote for the latter. Demand and supply of opposition programs will remain divorced. Accordingly, to prevent opposition forces from reaching out to voters and voters from demanding anything other than the authoritarian status quo, rulers often shut electoral challengers out of the public sphere. They either violate their rights to speak, peacefully assemble, or move about freely (repression) or block their access to mass media and campaign resources (unfairness).13 Repression. To qualify as democratic, elections must take place in an open and secure environment in which civil and political liberties are effectively protected. Electoral authoritarian regimes often exhibit a “strange combination of remarkably competitive elections and harsh repression” (Diamond and Plattner 2000: 107). In many sub-Saharan countries, electoral contests have been conducted under the shadow of state violence. For instance, in the early 1990s, “bogus elections” in Cameroon and Togo were accompanied by “pervasive state violence” (Joseph 1998: 6). The 1996 presidential elections in Gambia took place within an oppressive “climate of fear” (Wiseman 1998: 66) and “continual state harassment of the opposition” marked the Kenyan 1992 and 1997 national elections (Barkan and Ng’ethe 1998: 33). There are many gray areas, though. Democracy requires minimal degrees of freedom. But how much is just enough? Rather than enacting massive and continual repression, electoral authoritarian regimes tend to subject their adversaries to selective and intermittent repression. Irregular episodes of harassment and intimidation make less international headlines than systematic human rights violations. Overall the recourse to “low-intensity” repression may be no less effective than the reliance on “high-intensity” repression. Prominent examples are the 2000 presidential elections in Russia and Peru (see McFaul 2000: 30, Youngers 2000: 41–50 and 65–67).

Shaping the Electoral Arena

93

Besides, central state agencies do not need to do the dirty job of physical repression themselves. They can delegate it to societal actors. When repressive violence is not state-executed, but “only” state-sponsored or state-licensed, governments may deny responsibility and rhetorically condemn what they organize, encourage, or at least tolerate behind the scenes.14 To give an example of state-tolerated violence: By a conservative estimate, in Mexico in the 1990s, about 150 members of the emerging left-wing opposition party prd were killed “as the direct result of postelectoral conflicts” at the local level. While, to all appearances, these political assassinations were not masterminded by the central government but local “freelance” strongmen, the national executive did little to either investigate or prevent them (Eisenstadt 2004: 122). To cite an instance of state-sponsored violence: In the months leading up to the 2000 legislative elections in Zimbabwe, the Mugabe regime systematically deployed and incited armed gangs of unemployed youths (parading as veterans from the independence war) to harass and intimidate the social bases of the opposition Movement for Democratic Change (see Meredith 2002: 167–189). Repression often involves violations of physical integrity. To dodge the radar screens of human rights watch groups, it may shift its objective from physical harm to economic deprivation. Rather than murdering, jailing, or exiling opposition figures, regimes may simply bankrupt them. Moving well beyond clientelist routines of conceding and withholding favors, they may push dissidents into economic ruin through aggressive tax persecution, civil lawsuits, job loss, the imposition of ruinous fines, property confiscation, or the withdrawal of licenses, citizenship, or university degrees. Notorious examples have been Belarus under Lukashenka (Silitski 2005: 92), Russia under Putin (McFaul and Petrov 2004: 22–27), Peru under Fujimori (Conaghan 2005), and Singapore under the pap (Case 2006: 99–100). Unfairness. When authoritarian incumbents go out to face the voting public and get its electoral seal of approval, they often confront emerging opposition parties under conditions of radical unfairness. In case after case, the unfairness of electoral competition derives from inequalities in access to money and mass media. Electoral autocrats usually enjoy ample access to public funds and favorable public exposure. The whole apparatus of the state—often including government-run media—is at their beck and call, and they often can harass or intimidate privately owned media organs into ignoring opposition candidates.15 The problem of electoral unfairness extends beyond unfair conditions of competition. It includes unfair practices of competition. Regimes impose structural conditions of unfairness—and put them to practical use. In their electoral campaigns, authoritarian incumbents tend to treat their adversaries with contemptuous hostility. In addition to inundating state media with “undiluted praise” for themselves, they swamp them with “unadulterated

94

Analytical Framework

attacks” on their contenders (Wiseman 1998: 65). Going beyond gestures of dismissive arrogance, they often orchestrate aggressive campaigns destined to destroy the public reputation of opposition candidates. Official strategies of “character assassination” (McClintock 2006: 256) vary in form and substance. Generally speaking, opposition politicians can expect to see themselves portrayed as insincere, self-interested men and women of bad character, enemies of the nation, conspiratorial allies of foreign powers, “seeders of chaos” (Silitski 2005: 86), hideous instigators of the violence unleashed against them. It is not just their tone, but their institutional context, that sets such campaigns apart from negative campaigning in democratic elections. They violate minimal standards of civility and truthfulness, but more importantly they proceed without brakes, without the countervailing force of “alternative sources of information” (Dahl 1971: 3). Authoritarian attack campaigns are symptoms of a deeper disease: the colonization of the public space by the ruling party. Negative campaigns get a bad press everywhere. Yet, the ability to criticize officials and candidates forms a crucial part of our liberal, adversarial conception of democracy. It is a key resource for opposition parties and candidates trying to overcome incumbency advantages and mobilize popular discontentment against sitting governments. Besides, as John Geer has shown in his empirical study of attack ads in us presidential campaigns, negative campaign messages may be more useful and informative than positive messages, as they tend to contain more empirical evidence and demand higher standards of veracity (see Geer 2006). The democratic potential of negative campaigns, however, can unfold only in contexts of genuine competition and open media. In the face of (near) information monopolies that provide citizens “with nothing more than an endless diet of [official] propaganda” (Meredith 2002: 187), they work as simple weapons of the strong. In the absence of informational checks and balances (as well as in the absence of judicial restraints), sitting presidents can shield themselves against public criticism, while destroying their defenseless adversaries through unrestrained public attacks.16

4. Exclusive Suffrage: Formal and Informal Disenfranchisement Ever since the invention of representative government, political actors have been tempted to control electoral outcomes by controlling the composition of the electorate, whether by formal or informal means. In the contemporary world, formal disenfranchisement is a very tough “sell” both domestically and abroad. Legal apartheid is not a viable model anymore. Today, even the most hard-boiled electoral autocracies commonly grant universal suffrage to their citizens. The real growth end of the business lies in the realm of informal dis-

Shaping the Electoral Arena

95

enfranchisement, directed against categories of citizens who are known to support, or suspected of supporting, the opposition. Typical target groups are the young, the poor, residents of certain areas, and carriers of certain cultural traits or phenotypical characteristics. “Ethnic cleansing,” the persecution, physical elimination, and forced displacement of certain groups of citizens, as of non-Arab-speaking blacks in Mauritania in the early 1990s, is the most atrocious way of stripping citizens of their franchise (and much more). Less uncivil authoritarians recur to more subtle techniques. They manipulate the voter rolls, illicitly adding or deleting names, or bar voters from polling stations on trumped-up legal or technical grounds. They reduce the number of polling stations in areas of opposition support or shift their location without notice.17 They set up physical impediments on election day to keep opposition supporters from casting their votes. In the 2000 legislative elections in Zimbabwe, for example, supporters of the ruling party “set up roadblocks to prevent people from reaching polling stations and confiscated their identity cards” (Meredith 2002: 187–188). Governments may also devise registration methods, identification requirements, and voting procedures that are universal in form but systematically discriminatory in practice (see Schaffer 2008: Ch. 2).18 The other face of the coin of strategic disenfranchisement reads illegitimate enfranchisement, often in the form of multiple voting. In rural Malaysia, “phantom” or “transfer” voters work as a rough equivalent to classic gerrymandering. Rather than shifting district boundaries, official candidates shift citizens from their home districts to competitive districts where, “equipped with aliases, fake identity cards, and false addresses” (Case 2006: 103), they “arrive by the hundreds—more than enough to swing a typical rural election” (Pepinsky 2007a: 116). Either way, by disqualifying legitimate citizens from the voting process or by smuggling in illegitimate voters, rulers try to contain electoral threats by manipulating the boundaries of the demos. Instead of the people selecting its rulers, it’s the rulers selecting their people.

5. External Interference: Intimidation and Corruption Voters must be insulated from undue outside pressures if they are to choose freely. According to our modern conception of democratic citizenship, forged over long decades of democratizing struggle in the 19th and 20th centuries, the supreme political act of voting is a supremely private act. It is to be performed by autonomous individuals before the court of their personal consciousness. Voters are supposed to ponder contending political arguments while remaining impervious to extraneous means of persuasion, such as threats of force,

96

Analytical Framework

economic deprivation, and social disapproval. Securing the autonomy of individual voters requires, above all, shielding their decisions from the colonizing forces of violence and money.19 Intimidation. The threat of violence puts high burdens on the shoulders of voters. When authoritarian actors resort to violence against opposition candidates, civil society, mass media, and independent judges, they may or may not succeed in frightening them into resigned acquiescence, but unmistakably they transgress the bounds of democratic politics. The threat of violence can be suffocating in the wake of civil war settlements in which contending parties enter electoral competition before renouncing their military capacities. Held under the shadow of warring parties who are ready to resume war, “elections serve at best to ratify the balance of power of territory, people, and resources created by the war” (Snyder 2006a: 224). In the 1997 elections in Liberia, warlord Charles Taylor blackmailed voters into certifying the imbalance of power created by the civil conflict. Longing for peace and knowing that he would resume war if he lost the presidential election, they endorsed him by a landslide (see Lyons 2005: 71–75). On the eve of the 2000 legislative and 2002 presidential elections, Zimbabwe’s President Robert Mugabe orchestrated credible threats of civil war in case the ruling Zanu-pf would lose at the polls. Through a ruthless campaign of partisan violence masquerading as struggle for social justice, he strove, in the words of then-opposition leader Morgan Tsvangirai, “to intimidate the whole country into submission” (cited in Meredith 2002: 182). Open rehearsals of civil war and massive campaigns of repression are the most blatant forms of governmental threat communication. Regimes may deliver their intimidating messages to the citizenry in more coded forms, too. They may place electoral logistics in the hands of soldiers. They may beat up individual dissidents in the well-grounded expectation that others receive the message. They may prohibit street protests and threaten citizens with implacably enforcing “the rule of law” in the event of anti-government demonstrations. Et cetera. Corruption. The literature on historical as well as contemporary elections is replete with stories about parties and candidates trying to buy the sympathies of the electorate through the generous dispensation of particularistic favors (Lehoucq 2007, Schaffer 2007b). Concerns about the “clientelist control” of poor voters tend to arise whenever electoral competition unfolds in contexts of poverty and socio-economic inequality (see Rouquié 1978, Schaffer 2007c). Allegations of vote buying are not exclusive to authoritarian regimes. Yet, given the resource advantages as well as the legal impunity they enjoy, electoral autocrats are often capable of engaging in massive forms of patronage distribution (see Greene 2007, Levitsky and Way 2010a, Magaloni 2006).

Shaping the Electoral Arena

97

6. Redistributive Electoral Governance: Discriminatory Rules and Practices Regardless of the conditions of freedom that surround the electoral campaign, authoritarian incumbents may try to bend or break the will of voters by establishing self-serving institutions and practices of electoral governance. The notion of “electoral governance” (Mozaffar and Schedler 2002) refers to four fundamental aspects of the electoral process: x Rules of electoral competition: election formula, district magnitude, district boundaries, assembly size, electoral timetable, passive and active franchise. x Rules of electoral organization: voter registration, party and candidate registration, campaign financing and regulation, election observation, ballot design, location of polling stations, election material, voting, counting, and tabulating procedures, the structure of election management bodies and dispute settlement authorities. x Electoral rule application: election management.20 x Electoral rule adjudication: electoral dispute settlement. In democratic regimes, institutions and practices of electoral governance are assumed to be fundamentally neutral among contending parties.21 Thanks to their impartiality, actors can accept electoral outcomes as faithful reflections of voter choices. In electoral autocracies, by contrast, institutions and practices of electoral governance tend to be heavily “redistributive” rather than minimally “efficient.” Due to their discriminatory nature, losers perceive election results as treacherous distortions of voter choices.22 Redistributive rules of electoral competition. Authoritarian rulers who practice the evolving art of “election engineering” often establish self-serving rules of representation to keep eventual vote losses from translating into losses of power. Self-confident ruling parties that preside over a fragmented field of opposition parties tend to establish (or retain) winner-takes-all formulae that preserve their certainties of victory regardless of eventual oscillations in popular support (Pripstein 2005). In Mexico under the pri, Singapore under the pap, Zimbabwe under Robert Mugabe, and Croatia under Franjo Tudjman, majoritarian electoral rules have worked as effective depressants of opposition representation in parliament.23 To the degree that competition intensifies, however, authoritarian rulers face the rising risk of suffering loser-takes-nothing defeats themselves. When the “veil of ignorance” (Rawls 1971: 136–142) begins to descend over the electoral arena, rational autocrats should start dismantling majoritarian institutions that protect them in the present, but may harm them in an open, more pluralistic future (see Colomer 1995 and 2001).

98

Analytical Framework

Along with basic rules of representation, authoritarian incumbents often fine-tune lesser components of electoral systems. In Russia, for example, the high threshold of parliamentary representation, set at 5 percent of valid votes in 2003 and raised to 10 percent for 2007, long served as an effective device to bar the liberal opposition from the legislative arena—albeit with a little help from “a ‘little’ fraud” (Fish 2005: 79). In Mexico under declining hegemonic party rule, a generous “governability clause” ensured that the incumbent would preserve its legislative majority with little more than one-third of the vote. In other places such as Kenya, the Gambia, Jordan, Malaysia, and Venezuela, electoral autocracies have relied on gross malapportionment to keep winning despite defeat. In the Kenyan 1997 parliamentary elections, for instance, severe malapportionment was crucial to ensure a bare 51 percent governmental majority. The official party would have won less than 40 percent of seats, “had parliamentary districts been drawn to be roughly equal in numbers of registered voters” (Barkan and Ng’ethe 1998: 44).24 Redistributive rules of electoral organization. The complex set of rules that govern the administration of elections may harm opposition parties at each step of the electoral process, from voter registration to vote tabulation. While until recently rules and practices of electoral administration have been a rather “neglected variable” (Pastor 1999b: 6) in comparative politics, both practitioners and students of authoritarian elections have known for long the practical relevance of election management (see Birch 2011: 3–6, 39–41). “The election process is well under way before the formal stage of voting” (Issacharoff, Karlan, and Pildes 2001: 2). Accordingly, in most authoritarian settings, “the possibilities for democratic politics are already constrained” well “[b]efore the first vote is cast or the first ballot counted” (Issacharoff, Karlan, and Pildes 2001: 2). In non-democratic electoral systems, one bundle of organizational issues tends to overshadow all others: the structure of election management. Almost invariably, electoral autocrats and opposition parties clash over the key decisions involved: whether to leave the organization of elections in the hands of local authorities or put it under control of the national government (centralization); whether to entrust it to the ministry of the interior or establish a permanent professional agency (bureaucratization); whether to assign different functions, such as voter registration, election management, and dispute settlement, to a single body or separate agencies (specialization); whether to keep electoral authorities dependent on the executive or grant them genuine autonomy (independence); whether to place party representatives or independent experts at the top of independent electoral commissions (delegation);

Shaping the Electoral Arena

99

and whether to grant election officials wide margins of discretion or restrain them tightly through formal rules (regulation) (see Mozaffar and Schedler 2002). While most industrial democracies entrust the government bureaucracy (the ministry of the interior) with the organization of national elections, the democratizing struggles of the past decades have “put into question the legitimacy of the executive playing the role of a ‘referee’ in the competition for power” (López-Pintor 1999: 43). When authoritarian incumbents control the organization of elections, they invariably possess, or are seen to possess, the capacity and willingness to manipulate electoral procedures. Under the twin conditions of abuse of power and distrust of power that tend to prevail in authoritarian contexts, opposition parties usually fight to lead election management bodies into independence.25 Electoral fraud. Electoral fraud, an art form as old as political elections themselves, involves the introduction of bias into the administration of elections. It can take place at any stage of the electoral process, from voter registration to the final tally of the ballots. It covers such activities as forging voter id cards, burning urns, and padding the vote totals of favored parties and candidates. By definition, fraudulent practices violate the principle of democratic equality. They distort citizen preferences by silencing some citizens, while amplifying the voice of others. Thanks to its tranquilizing properties, electoral alchemy has been a popular pastime among electoral autocrats. From Haiti to Peru, from Serbia to Azerbaijan, from Burkina Faso to Zimbabwe, governing parties have engaged in vote rigging. In at least half of 81 protested elections registered worldwide during the 1990s, opposition parties claimed they had been victims of fraud (see Pastor 1999b: Appendix). Electoral impunity. “When elections go wrong” (Issacharoff, Karlan, and Pildes 2001), the system of electoral dispute settlement may offer the last wall of contention, a last chance to remedy a faulty process before political conflict overflows its institutional bounds. Under authoritarian governance, however, the prospects of judicial redress usually look faint. Judicial proceedings drag along endlessly, available legal sanctions are toothless, and electoral judges are professionally committed to blurring the criminal trails of the victorious incumbent. Accordingly, opposition parties often view electoral courts as mere public relations agencies of the regime, as “mechanisms of crowd control [set up] to diffuse mass mobilization” (Eisenstadt 2004: 82). Even if regimes strive to institutionalize electoral impunity, judicial institutions may nevertheless evolve into sites of struggle and resistance. Even under harshly repressive conditions, electoral courts designed to serve as

100

Analytical Framework

vote-laundering machines may show signs of independence against fraudulent incumbents. The example of the Egyptian constitutional court ordering the judicial supervision of polling stations in the 2000 elections made international headlines (see Brownlee 2007a: 134–137). After the opposition challenged partial results of the 2000 parliamentary elections, Zimbabwe’s president Mugabe demanded that “elections must be decided by the voters themselves and not the courts” (Meredith 2002: 200). He had evidently lost confidence in the political reliability of judicial authorities.

7. Indecisiveness: Tutelage and Reversion The notion of empowerment or authorization forms the first link in the chain of democratic choice. Elections are meaningful exercises of democratic governance only if they are consequential, if voters are able to endow elected officials with real power. Yet even if elections are decisive ex ante, with elected representatives enjoying full constitutional authority, elections may still fail to be decisive ex post. They may turn inconsequential, too, when non-elected actors put elected officials under their de facto supervision (tutelage) or straightaway prevent them from occupying their formal positions (reversion). Tutelage. Students of new democracies have long been aware of the danger of “tutelary powers” that undermine the sovereignty of democratic politics (Valenzuela 1992: 62–64). Under authoritarian tutelage, elected representatives possess their constitutional powers only on paper. In fact, they are subordinate to veto powers by unaccountable actors, be they domestic or foreign. Portugal during the first year of the Revolution of the Carnations, Algeria after the 1991 military coup, and Turkey until the constitutional referendum of 2010 are examples of military guardianship. Reversion. More visible and drastic than the assertion of tutelary powers is the classic coup, in which non-democratic actors nullify electoral outcomes by either preventing democratically elected officials from filling their posts or by ousting them before their constitutional mandate has expired. In the face of limited elections that produce irritating official results, authoritarian rulers at times resort to less dramatic corrective remedies. They adjust official election results, as in the 2000 parliamentary elections in Iran (Boroumand and Boroumand 2000: 117–120). They remove legislators and mayors from their positions, be it through judicial intervention or executive decree, as in Brazil in the 1970s and Venezuela in the 2000s. Or they persuade opposition actors through physical intimidation and material seduction to switch ranks and join the official party, as in “Operation Recruitment” carried out in Peru

Shaping the Electoral Arena

101

by the Fujimori government after it had lost its legislative majority in the year 2000 (Conaghan 2006: 111). Partial ex post corrections of either legislative or subnational election results may help an electoral authoritarian regime to smooth over minor disturbances, comfort its allies, demoralize its contenders, and adjust its strategies of control and legitimation. Overall they may allow the regime to continue doing electoral authoritarian business as usual. By contrast, if rulers decide to abort the entire electoral game by canceling elections, closing congress, chasing the sitting president from office, or jailing the winning candidate, they go much further. They transgress the defining boundaries of electoral authoritarianism. Rather than stabilizing the regime, they abolish it. By shutting down the electoral arena they abandon the logic of electoral authoritarianism and inaugurate either an interim regime or some form of closed authoritarianism. To sum up: Though often defended as modest, minimal procedural democracy is an ambitious enterprise. The requirements of democratic elections are steep, the possibilities of subverting them manifold. Still, authoritarian rulers cannot pick from the menu of electoral manipulation as they like, in absolute freedom. Their choices are subject to costs and constraints. And they are subject to risks of failure. When setting out to manipulate elections, rulers select strategies of intervention within an arena of conflictive interaction. They do not control the consequences of their choices, which may turn out ineffective or even counterproductive. If autocrats were sovereign in choosing and secure in choosing well, the politics of electoral manipulation would be no more exciting than darning socks or sweeping floors.

ES S EN T I A L LY C O N T E S T E D B O U N DAR I ES By admitting the existence of an intermediate zone between electoral democracy and closed authoritarianism, the concept of electoral authoritarianism narrows the gap between continuous and dichotomous conceptions of democracy. It nevertheless rests upon the idea that democratic regimes form “bounded wholes,” that is, coherent configurations of essential attributes whose absence marks “qualitative leaps” in the contested border regions that divide democracy and authoritarianism (see Collier and Adcock 1999: 557– 560 and 562). In principle, the qualitative logic of boundary delimitation dictated by my “family resemblance” approach is clear-cut. Within the family of multiparty electoral regimes, electoral democracies basically comply with all

102

Analytical Framework

essential conditions of democratic choice, while electoral autocracies severely and systematically violate at least one of them. Yet what does “basic compliance” with democratic minimum conditions mean in practice? And what does its opposite, their “severe and systematic” violation, mean? While clear-cut in theory, the boundary between electoral democracies and electoral autocracies is contested in practice. As I wish to argue, it is essentially contested. The notion of “essentially contested concepts” was introduced by W. B. Gallie (1956) to describe multidimensional and value-loaded concepts whose application is subject to irresolvable disputes.26 The concept of electoral authoritarianism is multidimensional and value-loaded, too. Its application is simple and clear-cut in the face of simple and clear-cut cases. Yet almost by definition, the border areas of electoral authoritarian regimes are neither simple nor clear-cut, but contain considerable ambiguities. Extreme cases of authoritarian manipulation are easy to assess. Few observers would doubt, for instance, that Belarus under Lukashenka and Egypt under Mubarak pertain to the family of authoritarian regimes. The whole point of electoral authoritarianism, however, is to escape the category of easy cases. They are normatively ambiguous by design, not by accident. While all authoritarian regimes possess certain theatrical qualities, public simulation defines the very nature of electoral authoritarian regimes. Electoral autocracies are stage performances by definition. They are purposefully contradictory creatures that stage-mange “democracy as deception” (Joseph 1998: 59). They preach the virtues of modern democracy, while practicing the ancient art of manipulation.27 The intended result is an enveloping lack of normative and empirical clarity. Less open than democratic systems and less repressive than closed regimes, electoral autocracies put our judgment to test.

Contentious Normative Thresholds In the face of ambiguous realities, the application of abstract norms to concrete regimes is bound to be contentious. Broad normative consensus over democratic fundamentals does not easily extend to specific agreement over precise minimum requirements of democracy at each step along the chain of democratic choice. The idea of a multifaceted repertoire of electoral manipulation has been broadly accepted in the scholarly community. Numerous definitions of democracy in general as well as numerous more specific treatments of the democratic quality of elections, have been conceiving the minima moralia of liberal democracy in similar, multidimensional ways.28 Scholarly disagreements concern less the normative foundations of democratic elections than the methodological possibility of clearing the “gray zone” (Carothers 2002) of

Shaping the Electoral Arena

103

“hybrid regimes” (Diamond 2002) and drawing sharp qualitative distinctions between democratic and authoritarian elections. We should not expect these disagreements to wither away in the future. They are fed by the continuous nature of both democratic norm application and authoritarian norm violation. The realization of democratic norms as well as their violation tend to take place at a distance from the poles of ideal-typical clarity. Imperfect democratic norm compliance. The constitutive norms of democracy are idealizations that admit varying degrees of realization in practice. Democratic norms are not perfectly realized anywhere, even in advanced democracies. The scope and jurisdictions of elective offices are everywhere limited; nowhere do popular elections constitute a universal route of access to all public office, and no liberal democracy grants unlimited powers to elected officials. Access to the electoral arena is always costly and never perfectly equal and the fairness of competition is an elusive ideal everywhere. The unequal distribution of societal resources like money and fame always carries the potential of distorting electoral competition. Mass media, even if plural and open in structure, never serve, nor are supposed to serve, as passive mirrors of popular opinion. They operate according to their own professional, commercial, and political criteria of relevance and tolerance. Rules of electoral competition invariably discriminate against someone, be they actual players or potential entrants to the party system. The organization of elections may conform to high standards of integrity and efficiency and still contain certain errors.29 The inescapable gap between democratic ideals and democratic realities invites two types of evaluative fallacies. We commit the fallacy of relativism if we refuse to describe any regime as authoritarian because all violate democratic ideals to some extent. Nobody is perfect, so everybody is more or less democratic. Authoritarian rulers, of course, are happy to embrace this perspective and describe their norm violations as “democratic imperfections.” We commit the opposite fallacy of perfectionism if we refuse to describe any regime as democratic because none embodies democratic ideals without flaws. Everybody is imperfect, so nobody is democratic. Not any deviation from democratic ideals qualifies as authoritarian, however. Electoral authoritarian regimes are not imperfectly democratic, but manifestly non-democratic. While relativists hold electoral regimes to excessively modest standards, perfectionists hold them to excessively ambitious standards. Imperfect authoritarian norm violations. If we recognize that all real existing democracies are situated at some distance from the democratic ideal, the normative question arises precisely how far an electoral regime has to move away from the ideal to be expelled from the democratic community. There are no clear-cut, quasi-natural boundaries we could rely on. We will all agree that slight, occasional deviations from democratic norms are not sufficient

104

Analytical Framework

to suspend a country from membership in the democratic community. To deny it democratic recognition, violations of liberal–democratic norms must be profound and systematic. Yet how profound and systematic exactly? Where shall we put the precise bar of minimum democratic accomplishments at each link in our democratic chain? How much power must elections confer? How much state repression do we tolerate? What are the bounds of legitimate exclusion from electoral competition? How much organizational independence and ideological distinctiveness do we ask from opposition parties? How much inequality in financial resources and media attention do we find acceptable? How much abuse of state resources and vote buying do we find worrisome? How severe and widespread do administrative irregularities have to look in order to convince us of the presence of electoral fraud?

Contentious Empirical Thresholds Even if we come to agree on democratic minimum standards, we may still disagree over their application to concrete cases. As laid out in Chapter 1, all authoritarian regimes suffer from informational and inferential uncertainties. Facts are unclear and contested and their causal implications too. For political actors, uncertainty represents a political problem. For academic observers, it presents a methodological problem. If basic descriptive as well as causal inferences are uncertain and contested, the regime classification they sustain will be uncertain and contested as well. Factual opacity. Closed dictatorships want everyone to know who is in charge. They want to crush and control dissidence even before it arises. Carried out in broad daylight, their repression is remedial as much as preventive. They want everybody to see how much it costs rising up against the system. Electoral autocracies are different. Rather than publicizing their repressive capacities, they aspire to practice authoritarianism by stealth. They strive to control and distort opposition actors in indirect, oblique ways. The very notion of electoral “manipulation” that is central for the idea of electoral authoritarianism hints at backstage activities hidden from the public eye. Causal uncertainty. Some authoritarian strategies require the active cooperation of opposition actors and citizens. To assess their normative relevance, we need to assess their severity as well as their causal effects on non-regime actors. For instance, governments that exercise severe repression may fail to frighten citizens into obedience. When we strive to ascertain the intensity of authoritarian manipulation, we usually have to accept considerable margins of uncertainty. These margins of uncertainty are even wider when we strive to estimate the causal impact of authoritarian strategies.

Shaping the Electoral Arena

105

Boundary Politics Given the ambiguous nature of electoral authoritarian regimes, the contentious nature of norm application, and the opacity of facts and causal effects, intense polemics over their true nature form an intrinsic element of electoral authoritarian regimes. Adversarial political interests perpetuate these controversies. Conceptual disputes about regime boundaries are political disputes about regime legitimacy. Since contending actors defend contending visions, their consensual resolution is unlikely. It is not just citizens and opposition parties who struggle with the structural ambiguities of manipulated elections. International election observers, too, have been wrestling with the empirical and normative complexities involved in evaluating flawed elections. Election-monitoring agencies have become more and more methodologically sophisticated (as well as more and more demanding) in their assessments of political elections. They have broadened their presence and perspective far beyond election day; they have developed a more precise and cautious vocabulary; they have written elaborate checklists, manuals, and codes of conduct; and they have made increasing use of statistical techniques of sampling, data processing, and inference. Nevertheless, in contexts of intense conflict where realities are opaque and contradictory, norms controversial, and actors deceptive, no amount of methodological sophistication will be able to dissipate the contested nature of regime classification.30

CO MPET I T IV E A N D H E G E M ON IC R EGI M ES Conceptual discussions on hybrid or electoral authoritarian regimes have been more concerned with establishing the external boundaries of electoral autocracies than with mapping their internal variance. It has been clear from the outset, however, that the broad category of electoral authoritarianism masks significant within-regime variation. Electoral autocracies form an extensive family of regimes whose members vary substantially in the types and intensity of authoritarian controls they impose (once they have crossed the threshold of democratic norm violation) as well as in the degree of interparty competition they display (once they have admitted organized opposition in the electoral arena) (see also Brownlee 2004 and Diamond 2002). The literature has recognized variations in levels of competitiveness by dividing electoral autocracies into competitive and hegemonic regimes (Schedler 2002b: 47). The precise labels that authors adopt vary. It is also controversial whether it makes sense placing both types of regimes in the same

106

Analytical Framework

conceptual box. Some authors draw the distinction to see how it plays out empirically (Brownlee 2002, Hadenius and Teorell 2007, Roessler and Howard 2009, Schedler 2002b), others to delimit the scope of their research and exclude hegemonic regimes from their universe of cases (Howard and Roessler 2006, Levitsky and Way 2002 and 2010b). Either way it is common practice to sort electoral regimes into two categories that display contrasting levels of party-systemic competitiveness. The notion of hegemonic regimes remits to Giovanni Sartori’s (1976) seminal conceptualization of “hegemonic party systems” as a subtype of non-competitive party systems. In his widely cited definition, Sartori emphasized the asymmetry in power between the hegemonic party and its “satellites.” The ruling party exercises tight control over the electoral arena, over its players as well as its outcomes. Only harmless opposition parties are allowed to run. Alternation of power is not an option. As Sartori formulated (1976: 230): The hegemonic party neither allows for a formal nor a de facto competition for power. Other parties are permitted to exist, but as second-class, licensed parties; for they are not permitted to compete with the hegemonic party in antagonistic terms and on an equal basis. Not only does alternation not occur in fact; in cannot occur, since the possibility of a rotation in power is not envisaged. The implication is that the hegemonic party will remain in power whether it is liked or not.

Hegemonic parties keep all relevant sites of central state power under tight control. As they also hold the power of constitutional change, they act without formal checks and balances. Due to an uncertain mix of manipulation and popularity, their vote margins are stratospheric. Opposition actors struggle at the margins, weak and disunited, perpetually torn between heroism, cooptation, and resignation. Single-party hegemony also leaves its imprint on intersubjective beliefs and expectations. Hopes for democratization are only modest, incremental, and long term. As there are no viable opposition parties vying for power, alternation in government seems close to impossible, and timid anticipations of change are fraught with fears of chaos and repression (see also Magaloni 2006). Hegemony, just like party dominance, as conceived by Maurice Duverger, is not just a matter of achieving longevity in power, but of creating beliefs of inevitability: “La domination [ . . . ] est aussi un phénomène de croyance. Un parti dominant est celui que l’opinion publique croit dominant” (1954: 342). In contradistinction, cases of “competitive” authoritarianism (Levitsky and Way 2002) allow for higher levels of electoral uncertainty. They manipulate the electoral game, but they do not control it as tightly as hegemonic regimes do. Lacking the aura of invincibility hegemonic parties possess, competitive regimes are more insecure, less institutionalized. Ruling parties keep winning those elections that allow them to occupy the chief executive office. Yet,

Shaping the Electoral Arena

107

on average, their margins of victory are smaller and more volatile; they often do not control constitutional majorities, and sometimes not even absolute majorities, in the national legislature; and they frequently have trouble controlling non-elected state officials, such as judges, security agents, and local election officials. In competitive regimes, opposition parties are still supposed to lose the big prize of the electoral game, the presidency. Yet they may win lesser prizes and can always hope to land a surprise and dislodge the incumbent in a “stunning election” (Huntington 1991: 174–180).31 In its core, Sartori’s definition of hegemonic party rule as precluding alternation in power nicely mirrors Adam Przeworski’s later conceptualization of democracy as a system of institutionalized uncertainty (see Przeworski 1988: 63). In democratic regimes, the substantive outcomes of elections are uncertain; in hegemonic regimes they are almost certain. Competitive autocracies lie in between: neither do they allow for full democratic uncertainty, nor are they able to establish practical certainty over electoral outcomes. If democracy is “a system in which parties lose elections” (Przeworski 1991: 10), electoral authoritarianism is a system in which opposition parties lose elections. Yet, while the defeat of hegemonic parties looks impossible, in competitive regimes alternation of power lies within the range of the possible.32 As I argued in Chapter 1, authoritarian regimes face the twin problem of institutional and informational uncertainty. Hegemonic and competitive regimes differ on the first dimension, not on the latter. Hegemonic regimes irradiate the certainties of institutional equilibria, competitive regimes the uncertainties of weak institutionalization. Yet the former are no less opaque than the latter. Actors do not have better information about popular support and elite coherence in hegemonic regimes. They do, however, operate on the basis of different assumptions about popular support and elite coherence. In hegemonic regimes, everybody assumes the regime to be popular and coherent—and therefore expects it to last well into the future. In competitive regimes, actors are unsure about its popularity and coherence—and thus unsure about its future prospects. The institutional certainties of hegemonic rule are grounded in different beliefs more than in different facts.

C ON C LU S I O N Electoral authoritarian regimes organize political elections with adjectives: regular, inclusive, unlimited, and competitive. These elections constitute the central arena of struggle as well as the central arena of manipulation in electoral autocracies. I strove to anchor the broad repertoire of manipulation that

108

Analytical Framework

authoritarian rulers have at their disposal in a systematic comprehension of the normative foundations of democratic elections. As I argued, a clear conception of democratic norms and authoritarian offenses does not preclude disputes over norm application. Some of these disputes are strategic, fed by contending interests. Some of them are genuine, fed by normative ambiguities, information deficits, and inferential uncertainties. For the comparative scholar, the conflictive delimitation of regime boundaries is a methodological problem of classification. For political actors, it is part of larger political struggles over the nature, the foundations, and the survival of the political regime. In this chapter, I discussed the normative logic of authoritarian elections. Now, I will turn to their political logic.

Notes 1. For a brief discussion of the general logic of “diminished subtypes” of political concepts, see Schedler (2011). 2. On conceptual stretching, see also Collier and Levitsky (1997), Schedler (2011). 3. See also Birch (2011: 14–19). The notion of political regimes that structure access to power presupposes the existence of a modern state. “[I]n the absence of a state, there is no regime, and that holds true whether or not there are . . . elections. When elections are held in the context of collapsed states and civil war . . . they do not confer power because there is no usable state apparatus that becomes available to the winner of elections.” Rather than the hallways of power, they lead to “a false door, as on a movie set, that has no real edifice behind it: a doorway to nowhere” (Snyder 2006a: 224, 223, and 225). Morlino, by contrast, introduces the twin notion of “democracies without law” and “democracies without state” (2009). 4. See Goertz (2006: Ch. 2) and Collier and Mahon (1993). The notion of “family resemblances” remits to Ludwig Wittgenstein (1968, §§ 66–71). 5. My terminology echoes analogous terms that refer to portable, replicable modes of action that varying actors may use in varying contexts for varying purposes, such as “the repertoire of contention” (Tarrow 1994) and “the repertoire of repression” (Davenport 2007). 6. Some regimes, such as contemporary Zimbabwe, post-Pinochet Chile, New Order Indonesia, and Thailand under the 1991 constitution, reserve legislative positions by granting non-elective authorities the prerogative of appointing a certain portion of seats in the national legislative assembly (see Case 2006: 100, Daniel and Southall 1999: 44). 7. As noted above, I circumscribe the notion of “electoral authoritarianism” to those regimes that subject the head of government to multiparty competition. On the limited, though still significant, role of limited elections, see Anderson (2000), Gandhi and Przeworski (2006), Lust-Okar (2009), Schmitter (1978). 8. One issue is the illegitimate exclusion of opposition candidates, another one the illegitimate inclusion of incumbents. For instance, most independent observers

Shaping the Electoral Arena

9. 10.

11.

12.

13.

14.

15.

16.

17.

109

believed that Alberto Fujimori’s 2000 bid for a third presidential term was unconstitutional (Levitsky 1999: 80, McClintock 2006: 251–252, Youngers 2000: 9 and 30). For post-Soviet illustrations, see Fish (2005: 61–67), McFaul and Petrov (2004: 54), Silitski (2005: 93) and Wilson (2005: 82–83). See Birch (2011: 35), Darden (2001), Fish (2005: 61–67), McFaul and Petrov (2004: 54), Robertson (2010: “The Insertion of Veto Points,” L 3918–4000), Silitski (2005), Wilson (2005: 82–86). Admittedly, encouraging opposition coordination by creating a two-party system, with one party acting as the government party and the other one being assigned the role of the official opposition, is a rather stupid thing to do for an electoral authoritarian regime. It is much more reasonable to proceed as Senegal’s Léopold Sédar Senghor and Egypt’s Anwar al-Sadat did: to devise a party system along the left–right continuum that places the ruling party in the political center, flanked by loyal contenders on each side (see Lust-Okar 2005: 118–125, Schaffer 1998: 26–27). In Kyrgyzstan, after the “Tulip Revolution” had toppled President Akeyev in the wake of the 2005 legislative contest, the trend reverted. In the December 2007 election, the Ak Zhol People’s Party garnered 73.9 percent of seats, initiating a broader trend of reconsolidating ruling parties in former Soviet republics. All figures from Nohlen (2001) and Inter-Parliamentary Union, parline Database on National Parliaments (). State repression, of course, is multifunctional. It may strangulate the formation of alternatives as well as their public expression; and it may affect citizens as well as parties and candidates. Aside from the repression of political opponents, some regimes outsource a broad range of dirty labors to non-state actors. To keep their hands clean, they “subcontract the necessary tasks of political manipulation” (Wilson 2005: xiv) either to professional specialists or enthusiastic amateurs. Greene (2007) offers a systematic treatment of “hyper-incumbency advantages.” See also Magaloni (2006), Levitsky and Way (2010a and 2010b). In addition to money and media, authoritarian governments often control another crucial resource: time. In Ghana in 1992 and Gambia in 1996, military rulers lifted the ban on political parties just a few months before election day, which deprived their opponents of the time necessary for putting together something resembling a party organization (see Gyimah 1999: 106–107, Wiseman 1998: 66). On the strategic use of time in politics, see Schedler and Santiso (1998); on the manipulation of electoral timetables by interim governments, see Shain and Linz (1995: 76–83). Two examples of regimes that established deeply unfair conditions and practices of public debate and political campaigning are Peru under Alberto Fujimori (Conaghan 2005: 140–162, McClintock 2006, Mauceri 2006, Youngers 2000: 65–67) and Russia under Vladimir Putin (Fish 2005: 67–77, McFaul 2000: 29, Ryabov 2004: 189–194). For syntheses, see Calingaert (2006: 139–140), Case (2006: 103), Daniel and Southall (1999: 39–41), Lehoucq (2003: 236–246).

110

Analytical Framework

18. On the legal disenfranchisement of voters through indigenous customary laws at the municipal level in Mexico, see Benton (2012). 19. Voting is a socially embedded act. Even in the absence of illegitimate pressures by parties and candidates, the notion of autonomous voters and personal preferences may be a benign fiction. Citizens may vote in either coerced or voluntary coordination with family or village members (see Schaffer 2010). 20. The overarching category of “electoral rules” includes rules of competition as well as rules of organization. 21. It is only “when elections go bad” (Issacharoff et al. 2001), as in the US presidential election in the year 2000, that this basic assumption tends to dissolve. 22. Following the terminology proposed by George Tsebelis, neutral institutions are “efficient” insofar as “they improve . . . the condition of all (or almost all) individuals or groups in a society” (1990: 104). Biased or discriminatory institutions are “redistributive” insofar they affect the distribution of costs and benefits among relevant actors (see Tsebelis 1990: 104–118). 23. In Mexico, as a matter of fact, the depressive effect of majoritarian rules on the opposition was almost lethal during the golden age of hegemonic party rule. Accordingly, in 1963, the ruling pri introduced elements of proportional representation to reanimate the agonizing opposition (see Lujambio and Marván 1997). 24. Biased rules of representation are not always illegitimate. At times, opposition parties pay little attention or even support discriminatory rules. In Brazil, Spain, and Argentina, for example, outgoing authoritarian governments devised rules of legislative seat assignment that privileged less populated, rural, and traditionally conservative regions. Nevertheless, persistently high levels of malapportionment in these countries have not stirred up major controversies. Political parties have come to accept them as a legitimate fact of political life (see Reynoso 2004). 25. On the structure, origins, and causal relevance of election management bodies, see also Birch (2011: 109–132), Eisenstadt (2004), Elklit and Reynolds (2002 and 2005), Hartlyn, McCoy, and Mustillo (2008), International Institute for Democracy and Electoral Assistance (2006 and 2012), Jaramillo (1998), Lehoucq (2002), López-Pintor (1999), Molina and Hernández (1999), Mozaffar (2002), Pastor (1999a and 1999b), and Schedler (1999, 2004b and 2005). 26. For more extensive discussions, see Collier, Hidalgo, and Maciuceanu (2006), Schedler (2011), Waldron (2002). 27. I borrow the notion of manipulation as art-form from Riker (1986), though his discussion focuses on the manipulation of voting processes in democratic contexts. 28. Illustrative syntheses of manifold strategies of authoritarian manipulation carried out in different contexts can be found, for example, in Birch (2011), Calingaert (2006), Case (2006), Conaghan (2005), Daniel and Southall (1999: 38–47), Herb (2005), Markoff (1996: Ch. 5), Ottaway (2003:137–160), and Pripstein (2005: 94–108). Schaffer (2007a) contains empirical analyses of one specific strategy, the purchase of votes, situated within the larger menu of manipulation.

Shaping the Electoral Arena

111

29. According to Issacharoff and co-authors, all elections contain “an unacknowledged ‘margin of error’—a concept that is readily identified with opinion polls and the Census calculations of the population” (Issacharoff, Karlan, and Pildes 2001: 2). This may well be true. Note, however, that the margin of error is conceptually different in the case of elections. Elections do not involve inferential statistics. Unlike public opinion polls, they do not involve generalizations from a sample to population. Elections do not aim at population estimations either. Unlike census calculations, they do not pretend to calculate the size and characteristics of an ever changing population (due to birth, migration, and death) on the basis of a periodic photographic collection of information that aspires to be accurate and comprehensive, though never quite is. The potential errors of elections are not inferential, but organizational. Most of them may be controlled internally, within the process of electoral administration. Even in large and heterogeneous countries, organizational faults may be brought down close to zero, excepting for cases of force majeure. Other sources of error, however, escape the control of bureaucratic agencies. Since people keep moving, migrating, and dying, voter registries contain unavoidable inaccuracies. Human error may be a source of bureaucratic failure, too. As we have seen in the us 2000 presidential elections in Florida, voters may err at the moment of registering their vote (the problem of the butterfly ballot) and counting authorities may have difficulties in interpreting the validity of votes (the problem of hanging, swinging, and dimpled chads). 30. Bjornlund (2004) offers a fine overview of the evolution of international election monitoring. Munck (2006) gives testimony to the efforts by election monitors to tap academic knowledge about empirical research for their practical purposes. Elklit and Reynolds (2005) translate the common checklist approach into a framework for empirical research. Kelley (2008 and 2012) traces empirical patterns, causes, and consequences of international election monitoring. 31. Schedler (2004a) explores empirical variations in competitiveness among electoral democracies and electoral authoritarian regimes. 32. In contrast to the procedural reliability of democratic regimes, institutional procedures are uncertain in both hegemonic and competitive regimes. On possible combinations of both procedural and substantive certainties and uncertainties, see Bunce (1991: 148–149 and 159).

4 Two Levels of Struggle Designating multiparty elections as the defining feature of a distinct category of non-democratic regimes makes sense only if they are more than mere adornments of authoritarian rule. The concept of electoral authoritarianism involves the claim that elections matter, even in contexts of authoritarian manipulation. In electoral authoritarian regimes, if they are to deserve their name, elections are more than rituals of acclamation. They change the game of authoritarian politics in a fundamental manner. They are constitutive of a central arena of politics. In this chapter, I will delineate the structural properties of the authoritarian electoral arena. Authoritarian multiparty elections, I will argue, are “creative” institutions that constitute a certain set of interdependent actors (ruling parties, opposition actors, citizens, and state agents), endow them with certain sets of strategies, and push them into a conflictive two-level game in which the competitive struggle for votes within given rules takes place alongside a competitive struggle over the rules of the game. I will review the structural features of this two-level competition step by step. Before, I wish to outline the general class of conflictive interaction that authoritarian elections belong to: asymmetric games.

A S Y M M E T R IC G A M E S Games are repeated, temporally and spatially bounded activities pursued for their own sake.1 Since Wittgenstein’s reflections on family resemblances (1968: §§ 66–87), we know that games come in many forms. Some are communal, others solitary; some are competitive, others cooperative; some are rule-bound, others shapeless; some are strategic, others spontaneous; some are conventional, others original; some unfold on tables, others on open fields; some are quiet and reflexive, others boisterous and corporeal; some require skill and effort, others are simple and easy. In many aspects, democratic elections resemble one specific subcategory of games: rule-bound competitive games.

Two Levels of Struggle

113

Such games, like baseball, soccer, tennis, and other sports, share a series of coherent structural features. They are bound by public, non-retroactive formal rules (rule of law). These rules are fair, as they do not discriminate players by their personal or collective identity (institutional justice). Players know and accept these rules and thus the possibility of defeat (procedural legitimacy). None of the players can change the rules by herself (symmetric powerlessness). Rules can only be changed by certain authorities through certain procedures (constitutional protection). Rule interpretation and conflict adjudication lie in the hands of third parties (procedural justice). These are the competitive games as we know and enjoy them as players and spectators. For heuristic reasons, I wish to introduce their logical opposite, a class of games I propose to call asymmetric games that violate basic principles of fair competition. Their defining feature lies in the unequal distribution of power: one of the players (or teams of players) holds the power over rule definition, rule application, or dispute adjudication. Such asymmetries of power are bound to translate into asymmetric procedures and outcomes. Either in form or in practice, they are bound to subvert the “spirit of the game,” the spirit of fair competition among equals. Asymmetric games accordingly share a coherent set of characteristics summed up in Table 4.1: x Asymmetric power: One actor either controls the definition of formal rules (constitution making) or their application and adjudication (rule implementation) or both. He is not an external Leviathan who imposes his will on everybody else, but one of the participating players (or teams of players). With a touch of lyrical hyperbole, we may call him the dictator or sovereign. x Insecure rules and practices: The powerful player holds power in order to exercise it. He will not wish to commit himself, and as a matter of fact is unable to commit himself, to keep defining and playing the game in a spirit of competitive fairness. Subject to the utilitarian calculus of the sovereign player, the formal rules of the game as well as their practical application turn unstable and uncertain (procedural uncertainty). x Unfair rules and practices: Almost invariably, asymmetries of power engender procedural injustice. If the sovereign player changes constitutional rules at convenience, the game turns unfair in form. If he manipulates the application of rules, formal rules lose their binding force and the game turns unfair in practice. Through institutional design or institutional practice the game stops being neutral and becomes structurally biased or “redistributive” (Tsebelis 1990: 104). x Unfair outcomes: Biased rules as well as the biased application of rules are designed to produce biased outcomes. Under fair rules and practices,

114

Analytical Framework

Table 4.1. Symmetric and asymmetric games Properties

Symmetric games

Asymmetric games

Power relations

Balanced: no player has unilateral Imbalanced: one player constitutional, legislative, or has power over design, judicial authority application, and adjudication of rules

Source of rules

Exogenous

Endogenous

Rule application and dispute settlement

Impartial: neutral third party

Partial: collusion between powerful player and authorities

Procedural certainty

High: stable rules

Low: contingent rules

Procedural justice

High: fairness of rules and rule application

Low: unfairness of rules and rule application

Substantive uncertainty

High: indeterminate outcomes (resulting from differences in knowledge, skills, and luck)

Low: predetermined winners and losers (resulting from differences in power)

Acceptance of rules and results

Mutual: simple games

Contingent: two-level games

Motives for accepting defeat

Normative: fairness

Instrumental: utility

Intrinsic value of game

High: the pleasures of playing

Unilateral: the pleasures of power

Motives for participation

Intrinsic and voluntary

Extrinsic or involuntary: coercion or corruption (side-payments)

Spirit of the game

Fair competition among equals

Unfair competition under conditions of inequality

the outcomes of competition are open (substantive uncertainty). Under discriminatory rules and practices, they are predetermined. In the last instance, the powerful player establishes the identity of winners and losers (substantive certainty). Procedural injustice is the mother of substantive injustice. x Unacceptable outcomes: As the outcomes of unfair games are not endogenously determined by the competitive dynamic between players, but exogenously generated by unilateral impositions of rules and practices, they turn illegitimate. Losers have no intrinsic reasons to voluntarily accept the predetermined outcomes of games whose fundamental rules and conditions they have not accepted in the first place.

Two Levels of Struggle

115

x Devalued processes: By manipulating rules and practices, powerful players establish “diminished subtypes” of the original game. Such damaged, distorted, corrupted versions of the game may still offer extrinsic benefits bestowed by the benevolent dictator, but the intrinsic rewards of playing get lost. The game becomes boring or annoying. x Withdrawal: In certain types of dictatorial games, such as the competitive struggle between slave gladiators and wild animals in the sands of the Coliseum in ancient Rome, participation is coerced. In competitive sports, it is voluntary. To the degree that competitive games lose their original meaning and enchantment, actors may keep playing because the dictator pays them well. Yet, in the absence of either coercion or external incentives, predetermined losers tend to withdraw their cooperation. They may do so by slowing down (apathy), by crying foul and assaulting the arbiter (protest), by adopting the unfair playing style of their adversaries (competitive fraud), or by stopping playing the game and retiring from sports altogether (exit). In all these scenarios, the game is not what is was anymore. In the latter, it turns into a solitary pastime in which dictators end up “bowling alone” (Putnam 1995). The unequal power structure of asymmetric games thus tends to translate into structural properties of the game itself (insecurity, unfairness, and illegitimacy). Quite logically, one thing leads to another. Asymmetries of power render rules and practices insecure and unfair; insecure and unfair rules produce unfair outcomes; unfair outcomes devalue games and render them illegitimate and unattractive (for an overview, see Table 4.1). Given their limited appeal for rational, self-respecting players, asymmetric games are scarce. They are rare pathologies of games, not normal forms. Small children or members of street gangs are sometimes bullied into participating in such corrosive games of power. The only societal realm where asymmetry appears to be widespread is politics—almost by definition a sphere of power and inequality. Authoritarian politics in particular.

T WO - L EV E L G A M E S Given their structural uncertainty and unfairness, asymmetric games tend to generate more contentious and muddled interactions among participants than “normal” symmetric games. In asymmetric games, competition is not bound by rules, but by relations of power. Rather than constrained by a set of stable rules, asymmetric competition unfolds within a set of contingent rules whose shape and

116

Analytical Framework

weight depend on decisions taken by the sovereign player. Unwilling or unable to compete within given rules, parties extend their struggle to the rules itself. Rather than focusing their competitive energies on playing the game, they get continually drawn into fights over the basic rules and conditions of the game. Asymmetric games thus resemble the “nested games” analyzed by George Tsebelis in which strategic interaction within rules goes hand in hand with strategic competition over rules (see Tsebelis 1990). Tsebelis, though, studies the politics of institutional change in highly institutionalized contexts in which players can take the effectiveness of formal rules for granted: “each player expects every other player to conform to them” (Tsebelis 1990: 94). In asymmetric games, by contrast, players struggle over the design of formal institutions as well as their effectiveness. Formal rules are structurally vulnerable to be subverted by actual practices. Rule abidance is intrinsically problematic. Authoritarian elections belong to the family of asymmetric games. As such, they are not conventional games in which players compete within a given institutional framework, known, accepted, and respected by all relevant players. They are fluid, adaptive, contested games whose basic rules actors try to redefine as they play the game itself. Due to their deep asymmetries in power and structure, they form nested or two-level games in which strategic interaction within rules goes hand in hand with strategic competition over rules. The partisan struggle over the hearts and minds of voters within the electoral arena unfolds simultaneously with partisan confrontations over the fundamental institutions and practices that shape the electoral arena. The game of electoral competition is embedded within the meta-game of institutional change. A good way of illustrating the nature of authoritarian elections is to contrast them with free and fair foundational elections that crown democratic transitions. The early third wave transitions from military rule that took place in Southern Europe and South America were based on foundational pacts and foundational elections. According to the standard script of transitions from military dictatorship, after an uncertain and conflictive period of institutional struggles, embattled authoritarian rulers ceded to pressures of democratization and withdrew from politics under the condition of obtaining certain institutional safeguards. The personal and institutional “vital interests” they commonly sought to protect against democratic impulses include property rights, military autonomy, and impunity from judicial prosecution. Once they had obtained such long-term guarantees, they called “founding elections” (O’Donnell and Schmitter 1986: 61) in which they did not participate (at least not in a direct manner). When these inaugural electoral processes were conducted within “reasonably fair rules” (O’Donnell and Schmitter 1986: 58), they provided a clear-cut threshold, a democratic watershed that certified the death of the old regime and signaled the birth of a new one.

Two Levels of Struggle

117

In classical transitions from military rule, institutional design and electoral competition thus form a temporal sequence of separate activities. Contending parties first devise a new set of fundamental rules that provide both the military government and the civilian opposition with minimal levels of “mutual security” (Dahl 1971: 16). In continuation, the “men on horseback” (Finer 2002) withdraw to enjoy the rigors of military life in their barracks, while civilian political parties move on to probe the excitements of the new democratic game. Founding elections blow the final whistle to the game of democratization and kick off a new game: the democratic competition for votes. After its inauguration, political actors may still aspire to change specific nuts and bolts of the political system. Yet, they will strive to achieve institutional change within its institutional bounds. In authoritarian elections, by contrast, actors compete for popular support without having settled ex ante their struggle over the basic rules of political competition. Given their asymmetric nature, they are much more ambiguous affairs. Neither do they spell the end of authoritarianism, nor do they inaugurate an era of democratic competition. When the velvet curtain of the electoral contest falls, the “drama of democratization” (Whitehead 1999) is far from over. Rather than inaugurating a new regime, authoritarian elections represent the focal points of a cyclical “test of forces and of legitimacy” (Lamounier 1989: 69). Rather than signaling the end of non-democratic rule, the introduction of asymmetric elections spells the continuation of democratizing struggles by other means. Figure 4.1 contains a simple graphical representation of the two-level game of authoritarian elections. In democratic elections, government and opposition parties compete at the game level over voter preferences. In authoritarian elections, their struggle over electoral outcomes unfolds at both levels. At the game level, they try to influence voter choices, at the meta-game level, institutional choices. In the remainder of this chapter, I will delineate the basic building blocks and dynamic properties of their nested interaction: its key actors and strategies, the hybrid nature of votes, the interaction of game levels, and the opacity of facts and effects.

AC TORS AND STRATEGIES Authoritarian multiparty elections are creative institutions. As they open up the two-level game of electoral competition and institutional struggle, they introduce actors, resources, strategic options, relations of conflict, and relations of dependence that are unknown to non-electoral autocracies. The main

118

Analytical Framework State agents Meta-game level: institutional struggle

Government

Opposition

Meta-game strategies of manipulation. Game strategies of voter mobilization

Meta-game strategies: Boycott & protest. Game strategies of voter mobilization

Electoral outcomes

Voters Game level: electoral competition

Figure 4.1. The nested game of authoritarian elections

protagonists in the authoritarian electoral arena are the government and the opposition. They do not battle alone, however. To achieve their conflicting goals, they crucially depend on two mediating groups of actors: citizens, the official arbiters of the game (at the level of electoral competition) and state agents, the informal arbiters of the game (at the meta-level of institutional change).

Opposition Authoritarian rule is often held to be incompatible with the admission of opposition. In autocracies, writes Piotr Sztompka, the “existence of a legal opposition is unconceivable” (1999: 149). By admitting multiparty competition for positions of state power, however, electoral autocracies legitimate the principle of political opposition. As laid out in the previous chapter, they may try to restrict the expression of organized dissidence by fragmenting, excluding, colonizing, or denigrating opposition parties. Yet they still have to live with opposition forces that enjoy at least minimal degrees of autonomy. By the simple fact of instituting multiparty politics, they abandon

Two Levels of Struggle

119

ideologies of collective harmony, accept the existence of societal cleavages, and renounce a monopolistic hold on the definition of the common good. Even if governments treat opposition actors with aggressive contempt, they admit the fundamental legitimacy of opposition the moment they admit formally competitive elections. The recognition of political pluralism entails the recognition of legitimate public dissent.2 When authoritarian rulers call multiparty elections, opposition forces face a series of critical choices. In the first instance, they have to decide whether to enter the game of unfree competition or to boo from the fences (participation versus boycott). If they enter the electoral arena, they have to think how to persuade voters and devise their strategies of electoral campaigning. Once the polls have closed and official results are published, they have to decide whether to acquiesce to the outcome or take their complaints to the media, the courts, the streets, or the international arena (acceptance versus protest). Entering the dynamics of electoral campaigning at the game level does not relieve opposition parties from meta-level concerns about institutional forms and practices. Even while they struggle for the hearts and minds of voters, they need to push for continuing reform and defend the vote. Accepting the dual challenge of simultaneous competition at both game levels was the hallmark of opposition politics in the “electoral revolutions” of the former Soviet republics. The secret of success of what Valerie Bunce and Sharon Wolchik have described as the “electoral model” of democratic transitions (Bunce and Wolchik 2011) lay in the determination of opposition actors to take both game levels seriously. They made creative and energetic efforts both to attract voters and to counteract manipulation. At the game level, they strove to unite the fragmented field of anti-government parties, recruit attractive candidates, survey public opinion, offer well-designed electoral platforms, put grassroots activists in direct contact with voters, and run intense professional campaigns. At the meta-game level, they worked hard to scrutinize the administrative preparation of elections, organize comprehensive electoral observation on election day, study the legal environment and train lawyers for judicial defense, and prepare for massive public protest in case of electoral fraud (see also Bunce and Wolchik 2007).

Government Where is the seat of power in authoritarian regimes and who occupies it? In principle, we may attribute autocratic power and agency to various individual actors, collective actors, or collections of actors. Common candidates are the chief executive (the ruler or dictator), the cabinet (the body of ministers),

120

Analytical Framework

the government (the dictator and his ministers), the top leadership (the inner circle of decision-makers), the state (a collection of civil bureaucracies, military bureaucracies, legislatures, and judicial institutions), the regime (the government and the state), a specific state agency (such as the military or the secret police), the political elite (high-level public officials), the ruling elite (the political elite and supporting social elites), and the ruling party. For the sake of simplicity, academic as well as political observers habitually assume the formal head of government (the dictator) or the official governing body (the national government) to be the central loci of authoritarian power.3 My conception of authoritarian elections as a four-player game rests upon the distinction between government and state agents. Still, I use the more encompassing language of “the regime” when agency appears shared or unclear. In addition, given my focus on elections, I often refer to “the ruling party” as the central agent of the authoritarian regime in the electoral arena. Governing parties in authoritarian regimes work as linkage organizations. Horizontally they link state agents, vertically they link citizens to the regime.4 When autocrats hold elections, they need their parties to fulfill the twin assignment of manipulating effectively (their horizontal role) and campaigning effectively (their vertical role). Electoral authoritarian regimes may display “sultanistic tendencies” (Chehabi and Linz 1998b: 9), with patrimonial rulers ratifying themselves in power through recurring multiparty elections. The organizational demands of authoritarian elections, however, limit the degree of personalism they can afford. As a rule, autocrats who wish to govern through controlled multiparty elections need a party (as well as a subsidiary state) to mobilize voters, and they need a state (as well as a subsidiary party) to control elections. Only few electoral authoritarian rulers do not invest in party building, but choose to debilitate all parties, including the governing party, in order to erect themselves as supreme arbiters over a field of weak and dispersed political organizations.5 Like their opponents, authoritarian governments have to take some key decisions regarding their strategic behavior in the electoral arena. Most importantly, they have to decide how to mix electoral manipulation and electoral persuasion in order to keep winning electoral contests. To what extent should they rely on authoritarian controls, and which strategies are they to pick from the variegated menu of electoral manipulation? And to what extent should they rely on the persuasion of voters, and which strategies are they to choose from the variegated menu of electoral mobilization? Most empirical analyses of electoral authoritarian regimes stress the pursuit of electoral manipulation over efforts at electoral mobilization. Yet, even if they retain ample capacities to distort electoral outcomes, these regimes do face a certain degree of “electoral exposure” (Cox 2007) and thus need to

Two Levels of Struggle

121

mobilize voters and obtain their active support at the ballot box. After all, a “defining feature of electoral, as opposed to military, authoritarianism is the ability of the leadership to win elections, albeit under unfair conditions” (Cameron 2006: 289). Even ruling parties that do not face tangible risks of defeat, like the Mexican pri at the height of its hegemony, often invest heavily into electoral campaigning.6 When devising their strategies of electoral mobilization, authoritarian incumbents pick from the usual menu of options. They may base their campaign appeals on parties or persons, on programs or clientelism, on economic or cultural issues, on symbols or substance, on the evaluation of the past or the construction of the future.7

Citizens Political elections recognize subjects as citizens. They endow them with “the ultimate controlling power” (Mill 1991: 97) over the occupation of the summit of the state. By establishing multiparty elections for highest office, electoral authoritarian regimes institute the principle of popular consent—even as they subvert it in practice. They establish the formal primacy of democratic legitimation, that is, legitimation through democratic procedures. Just like their non-electoral counterparts, electoral autocracies may tap various sources of legitimacy: revolutionary (the creation of a new society), transcendental (divine inspiration), traditional (hereditary succession), communitarian (nation building, anti-imperialism, ethnic supremacy), charismatic (magical leadership), or substantive (material welfare, public integrity, law and order, external security). In the last instance, however, the variegated reasons that feed popular support need to pass the institutional exam of the ballot box. Whatever their concrete claims to popular legitimacy, electoral authoritarian rulers accept having voters formally endorse them. They concede formal control over the validation of their claim to legitimate power to the citizenry. The institutional concessions that electoral authoritarian regimes make to the principle of popular sovereignty endow citizens with normative as well as institutional resources. Most importantly, elections open up avenues of collective protest. They provide “focal points” that may bundle social expectations and thus allow citizens to overcome problems of strategic coordination. Elections constitute citizens as individual carriers of political roles, but they also enable them to turn into collective actors, be it at the polls or on the streets. Generally speaking, individual citizens face a similar set of choices as opposition parties. On the game level, they have to decide whether to vote or not to vote (participation versus abstention), and if they do go to the polls, whether to support the incumbent or the opposition (pro-government versus

122

Analytical Framework

anti-government vote). On the meta-game level, they can either wave their flags in support of the regime (official loyalty), stay quiet and watch electoral protests unfold from their windows (passive acquiescence), or heed opposition calls to take to the streets in protest against the election (active mobilization). Of course, they can also choose violent courses of action on either side of the authoritarian divide. They can take up arms against the government and join rebel groups. Or they can take an active part in state repression, either formally by entering the security forces or informally by joining paramilitary organizations or state-sponsored gangs of thugs.

State Agents Electoral authoritarian governments fully control the peak of states they do not fully control. Popular “elections entail the largest peacetime mobilization of the national population in a short time span and require the coordination of millions of individuals engaged in hundreds of different activities” (Mozaffar and Schedler 2002: 5). Especially in poor, large, and diverse countries, elections pose enormous administrative and logistical challenges to governments, be they democratic or non-democratic. For rulers striving to contain the uncertainty of electoral outcomes, multiparty elections with universal suffrage involve problems of agency control on a massive scale. “Governments do one thing and one thing only: they give orders” (Ringen 2013: 59). For the implementation of their orders, they depend on others. This is true for authoritarian as much as for democratic governments. To enact their strategies of electoral manipulation, autocrats depend on active collaboration from a broad range of subordinate state actors. To communicate with citizens, they rely on reporters of state-owned media and teachers of state-controlled schools. To enact constitutional reforms as preventive policy instruments, they rely on legislators ready to applaud their bills. To introduce technical errors into voter rolls, they need a helping hand from election workers and municipal authorities. To ensure massive electoral turnout by social service recipients and the residents of public hospitals, they have to mobilize the mobilizing capacities of social workers and clinic directors. To correct misguided voter decisions at the stage of vote counting and vote tabulation, they depend on the mathematical creativity of polling station officials and software specialists. To control unruly protesters, they have to trust the “ruthless sense of right and dedication” (Heller 1994: 408) of their security forces. To convince the international community of their unwavering commitment to the rule of law, they need judges of the highest integrity willing to dismiss challenges to the electoral process as notoriously unfounded.8

Two Levels of Struggle

123

In multiple institutional arenas, central governments thus need willing executioners of their manipulative schemes whose political reliability and administrative efficiency they cannot, however, take for granted. At every location within the wide state apparatus, high up or low down, close to the center or near the periphery, they may suffer agency losses that impair the effectiveness of their electoral authoritarian blueprints. Remaining faithful to the ruling party (loyalty) is only one course of action open to subordinate public officials. They may also raise their voice in protest (dissidence), undertake covert actions of everyday resistance (subversion), leave the ruling coalition to join the opposition (defection), or withdraw from the public sphere altogether (exit). Last but not least, state agents may also openly challenge the ruling party, for instance, through military coup attempts, adverse judicial rulings, or the frustration of electoral fraud (confrontation).9 The extensive literature on principal–agent problems in economics and public administration contains a general set of practical recommendations for superiors (“principals”) struggling to make their subordinates (“agents”) heed their wishes and fulfill their commands. In order to mitigate problems of conflicting goals (“moral hazard”), principals need to set appropriate incentive structures; in order to minimize the risk of corrupt and incompetent agents (“adverse selection”), they need to tighten their selection procedures; in order to alleviate problems of asymmetric information (“hidden action”), they need to establish effective mechanisms of surveillance.10 In general terms, this is how authoritarian governments obtain the loyalty of state agents: they select friends and followers, bribe them into continual obedience, and keep an eye on them. They follow the textbook advice of principal– agent theory. Opposition actors, in turn, struggle to shake and break the loyalties that tie public officials to the government. Both sides court state agents, either to keep them on their side or to win them over. Inevitably, one of the main addressees of their competitive courtship are the security forces, the specialists in physical violence. They are the political lenders of last resort who can bail out a trembling government or push it gently over the abyss (see e.g. Barany 2011).

T H E H Y B R ID I T Y OF VOTES Success and failure in electoral competition is measured in votes. Votes are the chief prizes in electoral struggles. Even if the value of elections themselves is variable under authoritarian conditions, votes constitute the most valuable commodity on the electoral market. They represent that “obscure object of desire” (Luis Buñuel) that incumbents and opponents alike struggle to

124

Analytical Framework

possess. Votes buy access to state power. To gain power, parties must obtain votes. That’s the official name of the official game. Yet, while in democracies political parties must persuade voters to gain votes, in autocracies they must persuade voters as well as state agents. Under authoritarian conditions, votes are the joint result of voter choices and state decisions. Voters as well as state agents play crucial mediating roles in their production. Both government and opposition parties try to secure their respective market shares of votes by competing at both levels of the nested game (under asymmetric conditions). At the game level, they compete for electoral support by voters, at the meta-game level, for institutional support by state agents. Their final vote shares are determined by the joint outcome of these two levels of competition. Authoritarian election results are not children of repression and manipulation alone, as under conditions of single-party rule. Nor do they arise from the simple aggregation of citizen decisions, as under democratic conditions. Electoral outcomes reflect, to variable and unknown degrees, both citizen choices and manipulative maneuvers. In simple mathematical language: the official vote shares (v) of contending parties are a function of citizen preferences (p) and electoral manipulation (m).

v = f (p, m)

(4.1)

In democratic elections, “voter preferences are the principal source of uncertainty” (Kaminski 2002: 334). In non-democratic elections, authoritarian maneuvers constitute a competing source of substantive uncertainty. The confluence of vertical arrows in the center box of electoral outcomes in Figure 4.1 illustrates the hybrid nature of authoritarian election results. The absence of horizontal causal arrows from the government and opposition to electoral outcomes indicates the mediated nature of authoritarian vote production. In the two-level competition between government and opposition, neither of the two lead actors of authoritarian elections controls electoral outcomes in a direct manner. Both need to influence the mediating actors who do: state agents and citizens. The former co-determine electoral outcomes by implementing authoritarian measures, the latter by making electoral choices. In their interaction with state agents, governments strive to enhance the effectiveness of their manipulative schemes, while opposition parties try to erode it. In their interaction with citizens, both governments and opposition actors strive to bolster the effectiveness of their campaign strategies, while corroding the electoral appeal of their contenders. The politics of electoral uncertainty thus involves the twin struggle over the loyalty of officials and the loyalty of voters. This two-level competition unfolds in a simultaneous manner. Even worse, for actors and scholars who seek simplicity and predictability, it unfolds in an opaque and interactive manner.

Two Levels of Struggle

125

T H E O PAC IT Y OF AU T H O R I TA R I A N EL EC T I O N S Given their coercive powers, their restrictions on public debate, their penchant for secrecy, the absence of reliable mechanisms to register popular preferences, and their aversion to public accountability, non-democratic regimes are structurally opaque (see Chapter 1). Electoral authoritarian regimes make for no exception. What we can actually see in electoral authoritarian regimes are election results, the official distribution of votes and seats among parties and candidates. None of two bundles of causal and motivational factors that co-determine these results, neither authoritarian manipulation nor popular preferences, are easily observable. Two of the three variables in the basic equation of authoritarian elections are unknown.

The Opacity of Manipulation To a significant extent, electoral manipulation is an undercover activity. Some things we can see, such as the enactment of discriminatory election laws, the repression of protest marches, or the exclusion of candidates from the ballot by administrative fiat. Such manipulative efforts take place in broad daylight, mobilize agents of the central state, and invoke the language of legality and public reason for their justification. By contrast, many other authoritarian strategies of electoral control, such as the alteration of electoral lists, the intimidation of voters, or the falsification of ballots on election day, are more decentralized activities that involve myriads of public and private agents who generally do their job without leaving public traces. For all the knowledge we may be able to gather, be it episodic or systematic, narrative or statistical, the hidden realm of authoritarian electioneering constitutes an impenetrable black box we can (almost) never whiten in its entirety. Only few regimes have the panoptic aspirations of the Fujimori–Montesinos regime in Peru, whose comprehensive system of extortion, surveillance, and videotape recording allowed the public to inspect the black box of authoritarian maneuvering at least after the fact, once the regime had fallen (see Cameron 2006, Conaghan 2005). Normally, however, we will not even remotely know what non-democratic actors are up to on the invisible backstage of electoral politics. Moreover, even if we knew everything, we could not know that we know everything. The logic of distrust that prevails under authoritarian rule would make us uphold the suspicion that the worst may still be hidden from our eyes. The wysiwyg what-you-see-is-what-you-get rule never works under authoritarianism. Political actors know that what they see is most likely not

126

Analytical Framework

what they get from the authoritarian regime. They know that, in coping with authoritarian rule, they must practice the ancient art of dietrologia, the study of politics behind the scenes. 11

The Opacity of Citizen Preferences With respect to popular preferences we face a similar situation of partial knowledge built upon foundations of fundamental ignorance. Even in the “poor information environment” (Magaloni 2006: 236) that characterizes authoritarian elections, we may learn something about popular preferences, be it through access to “local knowledge” (Geertz 1983) or through representative public opinion surveys. Yet, under authoritarian conditions, we never know to what extent citizens engage in the public falsification of their private preferences (Kuran 1995). We do not know either to what extent their genuine private preferences are endogenous to authoritarian governance. In the absence of individual autonomy and freedom, popular attitudes are always suspected to be products of authoritarian manipulation. Authoritarian rule distorts the formation of popular preferences as much as their expression. Eager to know whether they are walking on solid ground or skating on thin ice, contemporary authoritarian rulers often engage in extensive public opinion polling. In climates of fear and corruption, margins of error are high, though, and even governments may fail to believe their own polls. On the eve of the 2000 presidential elections in Peru, for example, it appeared that “despite the opinion polls showing a likely Fujimori victory, the government did not feel confident and wanted to be able to rig the count if necessary” (McClintock 2006: 261). After all, as the government knew and everyone else suspected, “among the many actors bribed by [the regime] were pollsters” (McClintock 2006: 261).12

The Opacity of Causal Relations For political actors, imperfect information about basic facts involves imperfect information about causal relations. If actors know that election outcomes are the combined products of voter choices and regime manipulation, but cannot know the relative weight of each, they remain uncertain about the fundamental causal factors that explain electoral results. Their factual uncertainties about given levels of voter support and regime manipulation translate into inferential uncertainties about the causal weight of voter preferences and regime strategies. If descriptive inferences are uncertain under authoritarian conditions, causal inferences are even more so.13

Two Levels of Struggle

127

Consider, for example, the difficulties of estimating the empirical incidence of vote buying. The effective purchase of votes is a demanding business transaction whose success depends, to a large degree, on citizens’ willingness to sell their formal political rights. It requires that (a) citizens receive particularistic material offers, (b) change their voting behavior in response, and (c) understand their behavioral changes as concluding part of a commercial transaction. More often than not, we have observable evidence on the first move only, while the latter two remain hidden in the nested black boxes of individual minds and voting booths. In the agitated times leading up to multiparty elections, we can often see parties, candidates, and campaign workers delivering a broad range of material goods to individual citizens, from dollars to donuts, from hens and hoses to beans and bicycles. What we can usually not see, by contrast, are reactions and perceptions by voters. To assess the behavioral consequences that distributive campaigns trigger as well as the subjective meanings that voters attach to them we would need pieces of factual knowledge (who received what from whom to what effect) as well as counterfactual knowledge (what would voters have done in the absence of material offers) that are largely inaccessible to outside observers (see Schaffer and Schedler 2007).

The Distorting Mirror Recognizing the endemic opacity of authoritarian regimes, authors often consider political elections as mechanisms that allow dictators to solve their information problems. Multiparty elections, and even single-party and non-party elections, are said to provide ruling parties with important clues about popular dissatisfaction with the government, the party, or individual candidates. In particular, in well-established regimes with extensive electoral histories, small variations in local patterns of voting are thought to serve as early warning systems of emergent anti-regime sentiment (or waning pro-regime enthusiasm) at the local level. To gauge fluctuations in the public mood, authoritarian rulers can observe trends in the distribution of votes as well as less obvious data generated by authoritarian elections, such as registration and abstention rates, the number of invalid votes, and the use of secret, rather than public, voting.14 Conceiving authoritarian elections as sources of information seems to be incompatible with comprehending them as sources of opacity. The two viewpoints, though, are easy to reconcile. It is a simple matter of points of reference. Authoritarian elections are more informative than authoritarian rule without elections and less informative than democratic elections. Under authoritarian conditions, they produce information that would not be available in their

128

Analytical Framework

absence, or that would be more expensive to produce by alternative means, such as the bureaucratic surveillance of citizens by the political police. When compared to democratic conditions, though, the data they generate appear messy, noisy, contaminated. Thus, while authoritarian elections can serve as useful sensory systems of citizen discontent, they are a “deforming mirror” (Martin 1978: 127). They are imperfect authoritarian solutions to authoritarian problems of imperfect information. They mitigate, yet do not eliminate, the opacity of authoritarian governance. They allow regimes to overcome ignorance, yet not to achieve transparency. A distorting mirror is more reflective of reality than a blank wall. It still fails to produce faithful images.

T H E L IN KAG E OF G A M E L EV EL S The strategic calculus of antagonists who compete under conditions of relative opacity for a scarce good (votes) in a game of simultaneous moves at two levels, each of them controlled by a different set of third parties (voters and state agents), looks pretty complex. It looks even more complex if we consider the fact that the two levels of the authoritarian electoral game are not independent of each other, but interact in complicated ways. Whatever actors do on one level, is likely to affect the other level. To illustrate the nexus between game levels, I wish to discuss the logic of voter choices. When voters form their electoral decisions, we should expect them to pay attention to both levels: the game level of electoral campaigning as well as the meta-game level of institutional struggles.

Two Levels of Voter Evaluation In democracies, election results are simple aggregations of individual decisions. When we wish to explain electoral outcomes, it is clear what we need to learn about: the motives of individual citizens. If we understand the electoral calculus of voters, their perceptions and expectations, their moral and instrumental reasoning, then we can understand the overall distribution of votes. In electoral autocracies, by contrast, voter preferences do not explain but a (possibly small) part of electoral outcomes. In the presence of authoritarian manipulation, citizen choices are never the whole explanatory story. However, to the extent they do form part of that story, they may be analyzed in ways that are partially analogous to the ways in which we analyze voter choices in established democracies.

Two Levels of Struggle

129

When participating in the nested game of authoritarian elections, voters are likely to derive their electoral preferences p from two sources: their policy evaluations eg formed at the game level of electoral competition and their meta-game evaluations em concerning the strategies actors pursue at the level of institutional struggle. Assuming that the relation between these two evaluative components is additive, as in standard utility calculations where negative evaluations are subtracted from positive ones, we can write (for the sake of simplicity, without introducing subscripts for individual citizens and political parties): p = αe g + βe m (4.2) Constants α and β = (1 – α) denote the relative weights voters assign to policy and regime issues. These weights reflect the dimensionality of political conflict. To the extent that opposition parties succeed in structuring electoral competition along the axis of democratization (the authoritarian status quo versus democratic change), citizens decide their vote on the basis of their regime preferences. In the extreme, they care exclusively for the meta-game level of regime conflict (α = 0, β = 1). By contrast, to the extent that ruling parties succeed in blurring the regime cleavage by introducing cross-cutting dimensions of ideological conflict (such as left versus right, rural versus urban, or secular versus religious), citizens have to ponder their regime preferences against their ideological commitments, their collective identities, and their calculations of economic self-interest. In the extreme, they do not care at all about the meta-game level of reform struggles (α = 1, β = 0).

Game Evaluations To account for citizen evaluations of electoral campaigns under authoritarian conditions we can employ the same analytic distinctions and explanatory factors we commonly use to decipher electoral choices under democratic conditions. We can ponder the relative weight of parties versus candidates, policy platforms versus personality traits, valence issues versus instrumental considerations, positive versus negative messages, risk acceptance versus risk aversion, political versus contextual attribution of responsibility. As in democratic contexts, we can conceive the evaluation of public policies as the composite product of voters’ socio-tropic evaluations (regarding the state of the economy and the quality of governance), their pocketbook evaluations (factoring in their personal economic situation, their tax burden, the reception of subsidies, social transfers, and clientelist retributions), and their ideological affinities (including their party identification). Their policy evaluations may

130

Analytical Framework

be retrospective, prospective, or both. So far, so nice, theoretically speaking. The sad detail is that description comes before explanation and we possess very little factual knowledge about voter preferences and campaign evaluations in electoral authoritarian regimes. We do have some fine country studies, but very little in the way of descriptive cross-national data.15

Meta-Game Evaluations The notion of meta-game evaluations refers to anything relevant to voter decisions which voters observe at the meta-level of institutional struggle. As a matter of course, comparative data on meta-level evaluations are no less scarce than data on game-level evaluations. In the empirical part of this book, I will follow the example of others and bracket the level of voter attitudes. Still, as I assign a theoretically important role to voter perceptions of institutional struggles at the meta-game level, I wish to outline three dimensions of rationality that voters are likely to use in their meta-game assessments: normative, instrumental, and cognitive. Normatively, citizens assess the legitimacy of governmental strategies; instrumentally, they calculate their expected utility; cognitively, they read these strategies as sources of information on the institutional strength of the regime. Legitimacy. Political legitimacy refers to the normative judgments that citizens pass on the existing regime and its meta-game maneuvers.16 These judgments derive from their normative attitudes towards authoritarianism and their beliefs about the nature of governmental strategies. If citizens are democratic and watchful, the authoritarian strategies that governments deploy at the meta-game level are bound to diminish their electoral returns at the game level. If citizens are indifferent or ignorant, regimes may recur to electoral manipulation without affecting their levels of popular support. Like the legitimacy of government strategies, voters judge the legitimacy of opposition strategies on normative grounds. Opposition actors usually describe themselves as democratic forces who oppose authoritarian incumbents out of moral convictions. In assessing their democratic credibility, voters assess the veracity of their anti-authoritarian critiques as well as the appropriateness of their pro-authoritarian silences. If opposition actors protest, do they take to the streets because they are victims of an unfair system? Or because they are bad losers who divert attention from their own incompetence and unpopularity? If they remain passive, do they accept flawed elections because they are too weak to extract concessions? Or because they are colluding with a cheating (and corrupting) incumbent? If voters give credit to their claims, they should be more likely to reward opposition parties at the polls. If they don’t, they should punish them.17

Two Levels of Struggle

131

Utility. In instrumental terms, individual citizens anticipate the personal consequences that meta-game strategies pursued by government and opposition actors may bear on their personal welfare. With respect to government behavior, they have to calculate the expected utility of continuing authoritarian governance. Their utilitarian calculus is likely to include political as well as economic goods. Liberty and security, the enjoyment of personal freedom and physical integrity, though often excluded from narrow conceptions of personal utility, seem often more important than the enjoyment of material goods. With respect to opposition behavior, one recurrent concern of citizens is political violence. If voters anticipate that opposition protests may turn violent, provoke violent reactions by regime supporters, or ignite spirals of societal violence, they tend to shy away from supporting the opposition in order to avert the risk of political disorder. The higher the levels of post-electoral violence voters expect and the higher their personal aversion to violence and disorder, the less likely they are to support opposition parties at the polls (see Magaloni 2006: 55).18 Uncertainty. The institutional strategies by government and opposition are objects of normative and utilitarian evaluation, but also sources of information. They elicit moral judgments and define incentives, but they also shape beliefs. The manipulative strategies that governments deploy tell citizens something about the power, determination, and unity of the government. The protest strategies that opposition parties pursue tell citizens something about the power, determination, and unity of the opposition.

STRATEGIC DILEMMAS From the viewpoint of contending actors, the interaction between game levels may be mutually reinforcing or mutually frustrating. Negative feedback loops between the two levels may create serious strategic dilemmas that render strategic decisions indeterminate. One central dilemma of electoral authoritarian governments arises from the “legitimacy costs” their manipulative schemes may generate among voters.19 According to Robert Dahl’s seminal (1971) account of the history of modern democracy, socio-economic modernization involved a gradual progression from a situation in which the benefits of repression were high and its costs low to one in which the benefits of repression were low and its costs high. The decision–theoretic implication of such an evolutionary path is compelling: Actor incentives work in a consistent, convergent manner in favor of democratization. As the costs of democracy decline and the costs of authoritarianism go up, democratization obtains.20

132

Analytical Framework

Under electoral authoritarianism, governments and their allies measure the utility of repression and manipulation in the currency of votes: How large are the vote gains to be obtained through authoritarian practices and institutions at the game level (redistributive benefits)? How large are the vote losses to be assimilated due to citizens’ adverse reactions to non-democratic governance (legitimacy costs)? Electoral authoritarian regimes may find themselves in one of the equilibrium situations described by Robert Dahl in which benefits and costs align: situations of authoritarian comfort where the gains from manipulation are high and its costs low, or situations of authoritarian distress where the gains from manipulation are low and its costs high. However, they may well find themselves in more complex situations in which their meta-game strategies produce contradictory, rather than convergent, effects on the final distribution of votes. In the production of authoritarian election outcomes, popular support and regime manipulation are substitutes. If regimes have more of one, they need less of the other. Yet there are limits to substitution. While other types of dictatorships (in particular, military regimes and totalitarian regimes) may know few constraints on the intensity of repression they unleash on their subjects, electoral authoritarian regimes do not rely on naked repression. They engage in a game of illusions and contradictions. Officially, they give citizens a voice. Informally, they retain the capacity of distorting “the will of the people” as it emanates from the ballot box. And yet. While the authoritarian control of elections is designed to ensure governmental victories at the polls, it suffers the perennial possibility of over-extension. Authoritarian excesses run the risk of destroying the regime’s foundational illusion of democratic foundations. At some ill-defined point, electoral manipulation may turn counterproductive. To the extent that citizens (a) value democratic political goods and (b) perceive the existing political regime to violate democratic precepts, they create a dilemma for electoral authoritarian regimes. By manipulating the electoral arena, electoral authoritarian incumbents may compensate for their deficits of popular support. At the same time, they aggravate these deficits to the extent that manipulation makes democratic voters turn their backs on the regime. Manipulative substitutes for popular legitimacy can thus end up aggravating the very problem they are supposed to solve. At this point, the dual goals of control and legitimacy turn into “conflicting imperatives” (Gould 1999). To survive, regimes need both, but their efforts to obtain one hurt their chances of attaining the other. Figure 4.2 illustrates the space of possible outcomes that governmental meta-game strategies may bear. The vertical axis indicates the redistributive consequences of manipulation and reform (at the meta-game level), the horizontal axis their effects on voter support (at the game level). Both

Two Levels of Struggle + Neutrality line

133

Redistributive effects of meta-game strategies

Authoritarian comfort

Dilemma of manipulation –

+

Legitimacy effects of meta-game strategies

Authoritarian distress

Dilemma of reform –

Figure 4.2. The electoral consequences of meta-game strategies: governmental dilemmas

dimensions are based on the common measure of vote gains and losses. The positive upper half of the y axis depicts the vote shares rulers gain through electoral manipulation, its negative lower half the vote shares they lose through democratizing reform. The positive right hand of the x axis shows the vote shares that governments win thanks to the legitimacy gains of democratizing reform, its negative left hand the vote shares they lose due to the legitimacy costs of authoritarian manipulation (either because voters stay at home, vote for the opposition, or take part in activities that mitigate the impact of manipulation). The northeastern quadrant describes the area of “authoritarian comfort” where manipulative strategies engender positive effects at both levels of the game. The opposite southwestern quadrant indicates the area of “authoritarian distress” where governmental strategies are counterproductive at both game levels. The two remaining quadrants, by contrast, are lands of contradiction. The net effects of governmental meta-game strategies (manipulation or reform) cancel each other out along the “neutrality line” that crosses the two-dimensional space from north-west to south-east. Along this line the net effects of governmental meta-game strategies are nil. In the northwestern quadrant, governments face the “dilemma of manipulation,” since the redistributive gains they reap from manipulation are eroded by its legitimacy costs. Above the neutrality line, redistributive profits predominate and the net

134

Analytical Framework

effects of manipulation are still positive; below, delegitimizing effects prevail and the net balance of manipulation turns negative. In the southeastern quadrant, governments face a “dilemma of reform,” since the legitimacy gains they earn from democratizing concessions are counteracted by redistributive losses. Above the neutrality line, legitimacy gains prevail; below redistributive costs predominate. In the upper triangle, net effects of reform on the vote share of the ruling party are positive, in the lower triangle they turn negative. The higher and the more even costs and benefits are (and thus further away from zero and closer to the diagonal line of mutually neutralizing effects), the sharper are the dilemmas governments face in either of the two quadrants of contradictory effects. In both common language and academic debate, the notion of dilemmas often serves as a dignifying as well as somewhat dramatizing name for the various trade-offs that actors face in everyday life: the weighing of benefits and costs (including opportunity costs) of their decisions, their intertemporal choices between short-term and long-term interests, the election between the secure fruits of personal effort or the insecure benefits of social cooperation (as in the prisoner’s dilemma), and so forth. In a world of scarcity and constraint, that is, in the world we are used to living in and teach our children to live in, we may experience anything we do as dilemmatic in that simple sense: To obtain a, I need to renounce b. In a slightly richer sense, dilemmas involve only a subset of decisions: those that do not involve trade-offs, balancing acts between costs and benefits, but the election of miseries only, painful choices between painful courses of action. This corresponds to what the Oxford English Dictionary defines as the “popular use” of the word: “A choice between two (or, loosely, several) alternatives, which are or appear equally unfavourable.” To obtain a, I have to choose between the torments of b and the sufferings of c. Accepting decision costs and choosing between evils may push actors into dramatic inner conflict. The paradigmatic notion of dilemmas, however, seems to involve the more intractable problem of self-defeating action. Caught in such “genuine” dilemmas, actors face the impossibility of achieving their goals due to contradictory consequences of their decisions. They need to do b to obtain a, but b causes c, which impedes the realization of a. The decisions required to attain their objectives are incompatible with the attainment of their objectives. They cannot further their goals without undermining them. The demands of instrumental rationality push them into irrational, self-subversive courses of action. The literary model of such dilemmatic contradictions is Catch-22, the bureaucratic dilemma of air force members during the Second World War: To be relieved from flying dangerous combat missions, you need to be mentally insane. Yet, if you actually solicit relief, your request is taken as a sign of mental health, which prompts rejection (Heller 1955).

Two Levels of Struggle

135

The dilemma of manipulation seems to belong to the category of genuinely dilemmatic, self-defeating choices. To stay in power, governments manipulate elections (at the meta-game level), which may cost them power due to the legitimacy costs of manipulation (at the game level). Authoritarian elections are not inherently dilemmatic, but dilemmas are an inherent possibility in authoritarian elections.

STRATEGIC CONTROVERSIES AND SPLITS Both strategic dilemmas and informational uncertainties conspire against consensual definitions of rationality. Within each group of actors, they are likely to provoke strategic controversies, which are likely to provoke internal splits.

Strategic Indeterminacy When actors are torn between contradictory consequences of their actions, the imperatives of reason turn indeterminate. In the face of strategic ambiguities, there are no uniquely rational courses of action that actors can take. Strategic ambivalence entails rational indeterminacy. Caught between frustrating courses of action, actors do not know what to do, and we academic observers do not know what to expect them to do. Similarly, their rational calculations lack orientation if they lack information. If actors grope their way in the dark, the light of reason is dim and flickering. If parties and candidates ignore the sources of election results, they ignore the electoral consequences of their strategies. Informational uncertainties (and controversies) are bound to produce strategic uncertainties (and controversies). Opacity breeds controversy. To the extent that actors diverge in their causal beliefs about the origins of electoral outcomes, they will diverge in their beliefs about the effectiveness of alternative strategies. If they believe that election outcomes are children of manipulation, they will focus their strategic energies on the meta-game of institutional struggle. If they believe that official results mostly do reflect voter preferences, they will direct their creative efforts at the game level of electoral campaigning. Finally, if they are convinced that elections are determined by authoritarian strategies as well as voter choices in a roughly even way, they will take both game levels seriously. Instead of privileging one level over the other, they will engage in institutional battles with the same energy and determination with which they lead the competitive struggle over the hearts and minds of voters.

136

Analytical Framework

Of course, even if the mumbling voice of rationality sounds faint and incoherent, actors have to act nevertheless. They usually cannot afford to freeze under the paralyzing stare of strategic contradictions and contextual opacities. In the face of strategic dilemmas, they may not possess a blinking sword to dissolve the Gordian knot of strategic dilemmas with a clean cut. Still, they can devise simplifying decision rules that allow them to take some decision and move on. The most simple decision rule is repetition. As long as authoritarian incumbents are successful, they can go on as ever. They can rest on habit, old tricks, business as usual. Opposition actors, by contrast, are perpetually forced to adapt and learn, to innovate and experiment (see also Bunce and Wolchik 2011). Rather than forcing them to renounce judgment, the structural opacity of authoritarian elections thus obliges actors to rely on informational shortcuts, such as tradition, the fate of neighbors, or the heuristics of “representativeness” (Kahneman, Slovic, and Tversky 1982). Regardless of how they decide, though, their decisions are bound to be controversial and contingent, rather than consensually derived from given structures of choice. Simple decision rules or informational heuristics may lead individual actors to embrace firm convictions. They are unlikely, however, to create common sense—shared intersubjective perceptions about “the right thing to do” in the face of strategic dilemmas and opaque realities.21 When actors face genuine strategic dilemmas, we social scientists face genuine explanatory dilemmas. In the second part of this book, I will strive to explain the strategic choices that actors take within the two-level game of authoritarian elections (electoral manipulation and electoral protest) as well as the outcomes of their choices (electoral competitiveness and regime change). Through statistical analysis I will try to uncover average trends across cases. Here as in other fields of social inquiry, convergent patterns are likely to emerge only if the strategic environments that actors encounter are sufficiently clear and compelling to push them towards convergent readings and responses. If they are not, however, if strategic contexts are opaque and ambiguous, the terms of strategic rationality turn opaque and ambiguous as well. Instead of pursuing convergent courses of rational action, actors are likely to embark on irresolvable controversies about rational courses of action. The search for average patterns will be in vain.

Internal Splits Internally heterogeneous and potentially fissiparous, all four groups of actors that take part in the nested game of authoritarian elections are beset by

Two Levels of Struggle

137

coordination problems. If they succeed in presenting themselves as unitary actors to the outside world, they commonly do so under strenuous efforts to achieve and maintain internal coherence. Moreover, in addition to facing perennial threats of internal scission, all actors strive to erode the cohesion of their adversaries. As ancient Romans emperors knew, if you wish to rule securely, divide the people and your adversaries, too. For the subjects of divisive governance, the same wisdom applies symmetrically. If you wish to disturb the self-confident security of the emperor, unite the people and divide the rulers. The strategic controversies induced by dilemmatic and opaque structures of choice tend to exacerbate the coordination problems that are endemic to all participants in authoritarian elections. As a matter of fact, given their intractability, these controversies may easily induce actors to split. As they affect ruling elites as much as opposition actors, they may trigger “the classic four-player game of transition” (Linz and Stepan 1996: 265) that has characterized transitions from military rule. Strategic disagreements may lead incumbents to split into hardliners and softliners and the opposition into radicals and moderates. Among other things, these subgroups are likely to differ in their causal assumptions about the origins of electoral results. Regime hardliners tend to attribute their electoral victories to manipulation, softliners to popular support. When searching for explanations of electoral defeat, opposition moderates tend to give causal primacy to the relative popularity of the incumbent, radicals to its authoritarian maneuvers. Accordingly, softliners and moderates tend to grant strategic priority to the game level, hardliners and radicals to the meta-game level.22

C ON C LU S I O N Conceiving authoritarian elections as arenas of struggle involves the claim that these elections are not pure instruments of domination. Even if we assume that authoritarian governments are sovereign in their decision to establish multiparty elections, their sovereignty erodes as soon as multiparty elections are in place. The “asymmetric game” of authoritarian elections redefines the internal dynamics of authoritarian governance. It introduces a new set of actors, resources, and conflicts, hand in hand with a new set of structural dependencies, informational demands, and strategic complexities. The entire empirical part of this book is predicated on the assumption that authoritarian elections constitute a site of political conflict that possesses its own logic and relative autonomy (even while it is embedded in its societal and international environment).

138

Analytical Framework

If elections were simple tools of authoritarian governance, they should make the life of rulers easier and let them live longer. The literature is divided on this point. Schizophrenic almost. Some authors study elections as instruments of authoritarian survival, others as instruments of democratization. If they constitute a conflictive game, both outcomes are plausible. Before exploring the empirical relations between the moving parts inside the “nested game” of authoritarian elections, I wish to review the comparative literature on the effects of multiparty elections on authoritarian stability. If they are straightforward instruments of authoritarian survival, the entire two-level struggle between government and opposition is irrelevant, a chimera, a neat stage performance by the government and a delusional exercise for the opposition. However, if their regime consequences are more ambivalent, it makes more sense a priori to look into the black box of authoritarian multiparty competition and take its autonomous dynamics seriously.

Notes 1. Games are not work. They are “unproductive” activities that lack instrumental value. They are ends in themselves (even if, paraphrasing Max Weber, professional sportsmen live, not only for the game, but from the game). Generally games are pursued for the sake of entertainment. Players may enter them for a variety of motives, though, and as Wittgenstein noted (1968: § 66) they vary in their entertainment value. Some are exciting, others boring. However, boring games are commonly regarded “diminished subtypes” of games, vacuous and pointless. 2. Of course, the formal recognition of pluralism does not necessarily imply its discursive recognition. In the 2006 elections in Belarus, for instance, the “entire opposition movement was demonized as a treasonous plot” (Diamond 2008: 85). Such strategies of rhetorical delegitimation are common among electoral autocrats who have troubles adapting to the discomforts of public debate they can neither repress nor embrace (see also Ch. 3: “Demand-Side Restrictions,” pp. 92–94). 3. Common typologies of political regimes rest upon attributions of power to individuals or organizations. They respond to the question of who controls the levers of state power: a party, the military, a dynasty, or an individual. Given the opacity of authoritarian regimes, however, locating the center of power is often a difficult business. In the comparative study of politics, we tend to bridge these difficulties of empirical observation through audacious assumptions (see Barros 2011). 4. On party linkages in democratic contexts, see Lawson (1988). 5. Mexico’s Porfirio Díaz would be a historical example of (something like) a “party-less electoral autocrat” (see Bravo 2010), Kyrgyztan’s Askar Akeyev a more contemporary one. On “sultanistic” electoral authoritarian regimes, see Thompson and Kuntz (2006).

Two Levels of Struggle

139

6. See Greene (2002 and 2007), Langston and Morgenstern (2009), and Magaloni (2006) 7. On the menu of electoral mobilization, along with the guiding distinction between “clientelistic” and “programmatic” mobilization, see Kitschelt (2000). 8. On the general organizational demands of electoral manipulation, see also Way (2006) and Levitsky and Way (2010b). 9. Quite transparently, my conception of strategic options for state officials is indebted to Hirschman (1970) and Scott (1985). 10. For a fine synthesis, see Kiewiet and McCubbins (1991: Ch. 2). 11. I owe the notion of dietrologia to Philippe Schmitter (see his corresponding entry in Les Intraduisibles: The Dictionary of Untranslatable Terms in Politics, ). On the general research problems generated by secrecy under dictatorship, see Barros (2011). 12. In his analysis of electioneering under General Jaruzelski in late communist Poland, Marek Kaminski (1999) offers a telling account of authoritarian rulers investing heavily in surveying the public mood—and erring heavily in overestimating their own popularity (see also Kaminski 2002). 13. See also Ch. 1, “Informational Uncertainty,” pp. 37–41, and Ch. 3, “Contentious Empirical Thresholds,” p. 104. 14. See also Cox (2009), Hermet (1978: 12), Lohmann (1993), Magaloni (2006), Miller (2009), Pravda (1978: 177), Rozenas (2009), and Ch. 5, “Authoritarianism by Elections.” 15. Magaloni (2006) offers a nuanced model of electoral choice under authoritarian conditions. See also Domínguez and McCann (1996), Greene (2007), Hale (2008), and Rose and Mishler (2008). 16. The contemporary literature on the political economy of authoritarianism tends to employ the aggregate notion of “loyalty” as a shorthand for citizens’ behavioral acquiescence, irrespective of motives (see Wintrobe 1998). My narrow conception of political legitimacy covers only normative sources of political loyalty so broadly understood. It excludes instrumental considerations, such as personal gain and personal security. 17. Like the legitimacy of government strategies, voters also judge the legitimacy of opposition strategies on normative grounds. On the problems of credibility opposition parties face when challenging the credibility of an election, see Eisenstadt (1999: 93–95), Hartlyn and McCoy (2006: 51–52), Magaloni (2006: 233–239), Schedler (2007a), and Whitehead (2007). 18. Notoriously, authoritarian governments are adept at exploiting the “fear factor” and portraying themselves as guarantors of order and stability. More often than not, their warnings against chaos and violence are little more than veiled acts of blackmail which threaten voters with violent retaliation should they dare to grant victory to the opposition. 19. See also Birch (2011: 56). Of course, governments do not hold a monopoly on strategic dilemmas. The meta-game strategies opposition actors pursue (election boycotts, above all) also may generate contradictory effects at the two levels of the electoral game (see Schedler 2009a: 186–188).

140

Analytical Framework

20. See, in particular, Dahl (1971: 16, Figure 1.4). For an application of Dahl’s reasoning to the decision calculus of electoral authoritarian regime change, see Lindberg (2009d). 21. On the general indeterminacy of strategic choice, see Hardin (2003). On political contingency, see Schedler (2007b). 22. For the original formulation of this 2–2 scheme of transition actors in the context of “classical” transitions, see O’Donnell and Schmitter (1986: 15–17, 24, 63, and 71).

5 The Power of Elections The existence of authoritarian multiparty elections poses a double puzzle of rationality. Why do authoritarian rulers expose themselves to the complexities of electoral competition, rather than enjoying the tranquillity of authoritarian politics without elections? And why do opposition actors participate in charades of electoral competition, rather than enjoying the innocence of political contention outside the electoral arena? The adoption of elections by authoritarian rulers is pointless unless elections serve as tools of authoritarian governance. Yet, if they do, why should opposition actors take part in them? Their participation in authoritarian elections is pointless unless these elections serve as tools of anti-authoritarian resistance as well. The notion of institutional ambivalence captures the contradictory nature of authoritarian elections that serve their authoritarian masters and their anti-authoritarian servants, too. As a matter of course, they do not serve them equally well. The asymmetric game of authoritarian elections cannot but give rise to asymmetric ambivalence. Governing parties are likely to prevail, even if they face a residual risk of failure, while opposition parties are likely to fail, although they possess a small chance of success. At each election, authoritarian success is the rule (the probable outcome), opposition success the exception (the possible outcome).1 While this book focuses on the internal dynamics of conflict that unfold within electoral authoritarian regimes, it cannot ignore the consequences multiparty elections bear on the very stability of electoral authoritarian regimes. Their macro consequences impinge on the genetic logic of regime origins. To the extent that non-democratic elections prove useful for rulers in prolonging the status quo, we can comprehend the instrumental rationality of authoritarian institutional design. To the extent they prove useful for opposition parties in subverting the status quo, we can understand the rationality of their institutional engagement. In addition to providing rational motives of institutional choice and participation, the systemic consequences of elections also impinge on the political logic of regime struggles. Unless the two-level competition between government and opposition bears systematic effects on the fate of regimes and

142

Analytical Framework

governments, neither of the two parties should take it (that) seriously, nor should comparative scholars.2 If authoritarian multiparty elections work as institutional causes whose mere presence determines regime outcomes, rather than institutional arenas whose contingent dynamics of conflict determine regime outcomes, the nested game of authoritarian elections amounts to no more than shadowboxing, theatrical and inconsequential. So far the theory. So far the puzzles. But what do we know empirically about the consequences that authoritarian multiparty elections bring to bear on regime and leadership dynamics? Does the comparative literature provide evidence for the asymmetric ambivalence of authoritarian multiparty elections? For the contingent nature of their effects on the duration and trajectories of regimes and rulers? Although the present state of comparative research does not offer definitive answers, a review of the emergent literature provides valuable insights.

T H E DI V ER SI T Y OF T H E OR E T I C A L EX PEC TAT I O N S Authoritarian elections carry great, yet contrasting, expectations on their shoulders. Some view them as tools of authoritarian governance, others as sources of instability, and still others as levers of democratization. Individual country trajectories suggest that the systemic consequences of authoritarian elections are not predetermined. Everything seems possible: stability, regression, and democratization. Authoritarian multiparty elections can form the foundation of long-term hegemonic party rule, as in post-revolutionary Mexico and Paraguay under Stroessner. They can facilitate gradual processes of “democratization by elections,” as in Senegal or Taiwan in the 1990s. They can trigger the sudden collapse of authoritarianism through “electoral revolutions,” as in the Philippines in 1986 and Serbia in 2000. They can constitute an arena of authoritarian regression, as in Russia and Zimbabwe after 2000. Just like their democratic counterparts, authoritarian elections may also provoke the closure of the electoral arena through military intervention, as in Azerbaijan in 1993 and Côte d’Ivoire in 1999. All these contrasting trajectories fall within the realm of the possible. Yet, are some more likely than others? Which are the average effects of authoritarian elections across cases? Over the past years, we have seen growing scholarly interest in deciphering “the power of elections” (Di Palma 1990: 85) under authoritarian governance. In conceptual, theoretical, and empirical terms we have learned a lot from the new, blossoming literature on the causal role of authoritarian elections. However, defying the scientific goal of accumulation of knowledge, we have

The Power of Elections

143

seen a disquieting bifurcation of debate. On the one hand, the literature on the political economy of dictatorship has been emphasizing the regime-sustaining value of authoritarian elections. On the other hand, comparative studies of democratization by elections have been stressing their regime-subverting potential, because elections tend either to destabilize authoritarian regimes or to democratize them. Until recently, these strands of theoretical inquiry and empirical analysis have been developing in a state of peaceful coexistence and mutual ignorance.

Authoritarianism by Elections Over three decades ago, in their edited volume Elections without Choice (1978), Guy Hermet, Richard Rose, and Alain Rouquié called upon scholars to pay more attention to the empirical relevance of authoritarian elections. As one of the editors wrote, “state-controlled elections” had often been “denied any significance.” Observers tended to dismiss them as hollow, fraudulent, predetermined rituals, heeding them with “objective contempt” and ignoring the “numerous functions” they perform (Hermet 1978: 1), such as regime legitimation, voter education, the communication between rulers and subjects, and the resolution of conflict within the ruling class (see Hermet 1978: 13–17).3 For quite some time, Elections without Choice represented the state of the art in the comparative study of authoritarian elections. It was well received and often cited—yet inconsequential in its impact on the research agenda of comparative political studies. Hermet and his co-authors had identified and surveyed a research lacuna that the discipline was not capable of filling at the time. It is only in recent years that we have seen a new generation of largely quantitative empirical research that studies the use of elections (along with other formal institutions) as instruments of authoritarian rule. This new literature has replaced the language of functions by the language of rationality. Instead of searching for systemic roles elections accomplish for authoritarian regimes, it reconstructs the instrumental utility they have for authoritarian rulers. The basic intuitions are very similar, though: Unless elections would serve a purpose, dictators would hardly bother to convene them. Following the pathbreaking Autocracy by Gordon Tullock (1987), most of the political economy literature on political regimes assumes that elections under dictatorship serve one overwhelming purpose: the political survival of the dictator. Scholars “now generally view the establishment of elections as a means by which dictators hold onto power” (Gandhi and Lust-Okar 2009: 404). The basic assumption is simple: Authoritarian rulers want to survive in office and elections (just like other formally representative institutions like

144

Analytical Framework

parties and legislatures) help them to accomplish this. They do so by containing structural threats that linger over their hold on power. Much of the literature on the political economy of authoritarian elections focuses on horizontal threats that arise from within the ruling elite. Rather than conceiving elections as (intrinsically deceptive) devices to create popular legitimacy, authors understand them as tools of elite persuasion, designed to nudge potential challengers from within the ruling circles into continuing acquiescence. How do they do so? At times, the underlying “mechanisms through which [elections] promote regime survival remain opaque” (Pepinsky 2007b: 1), but authors have been suggesting a broad range of possibilities. In some way or other, they tend to treat authoritarian elections either as solution to informational problems or as institutions that change the relations of interdependence among actors. Informative elections: According to one set of perspectives, elections mitigate opacity. They serve as two-way streets of political information flows. (a) Elections produce information. They allow rulers to gather information on allies, voters, and adversaries (surveillance, fact finding). (b) Elections disseminate information. They permit rulers to spread information on the power and determination of the regime (reputation building, perception management). Constraining elections: According to another set of hypotheses, by introducing new structures of interdependence, elections serve as double-edged constraints on political actors. (a) Authoritarian elections constrain rulers. To the extent that it is costly for rulers to close the electoral arena at their convenience, authoritarian elections offer “a credible exit option” (Magaloni 2008: 728) for discontented elite members. Both the dictator and his allies know that internal challengers may defect to the opposition in case the dictator fails to pay and protect them properly. By rendering rulers vulnerable, elections create the structural basis for institutionalizing credible distributive and policy coalitions. Instead of creating legitimate routes of access to power, they create credible routes of access to patronage. The possibility of exit strengthens the force of voice (see Hirschman 1970, Magaloni 2008). (b) Authoritarian elections constrain adversaries. They make rulers dependent (to a certain degree) on electoral support—but everybody else too. Institutions are often said to play constraining as well as enabling functions. They define limits and create opportunities.4 Authoritarian elections constrain by enablement. By creating opportunities of challenging the regime at the polls they create constraints on non-electoral challenges. Opposition actors who are able to challenge the incumbent at the polls are encouraged to grasp the institutional opportunities that elections afford them. Those who are unable to do so, by contrast, are deterred from choosing extra-institutional routes of dissidence. Unless they have a good chance of mobilizing voters,

The Power of Elections

145

they have a poor chance of mobilizing anybody else. As they are enabled to challenge regimes electorally opposition actors find themselves constrained to challenge them electorally.5 The constraining roles of authoritarian elections are parasitic on their informational roles. Without information, neither constraints nor opportunities become operative. For instance, only if authoritarian elections provide shows of popular regime support that are credible to the inner circles of the elite (who know all about the dirty tricks of the regime), can they “serve to deter rivals” (Geddes 2005: 10) from mounting either electoral or non-electoral challenges to the incumbent. Instability by Elections While scholars working in the political economy tradition tend to view elections as straightforward instruments of authoritarian rule (since any other assumption would violate basic notions of instrumental rationality), scholars who work in the tradition of transition studies tend to view elections, in particular multiparty elections, as a destabilizing element in authoritarian regimes. It is perfectly plausible that rulers establish multiparty elections under the expectation that they fulfill regime-sustaining roles. Yet authoritarian rulers are not solitary players on the electoral planet. They are not bowling alone in the electoral arena. When opening up multiparty elections, they do not open up a tool chest, but a competitive field. Their efforts to appropriate the benefits of elections in an exclusive fashion are unlikely to go uncontested. Thus the regime-subversive potential of multiparty elections. When analyzing the opening moves of military and single-party dictatorships, the early transition literature had observed that political liberalization often constituted a self-defeating strategy of regime survival. Unlocking a Pandora’s box of societal conflicts, liberalization was too conflictive to be contained within the bounds of autocratic control. Due to their institutional inconsistency, liberalized authoritarian regimes or “dictablandas” (O’Donnell and Schmitter 1986: 9) seemed condemned to failure. Once softliners within an authoritarian regime had taken the first steps to introduce liberalizing measures, the subsequent outflow of political protest forced them to either revert to repression or take bolder steps towards democratization. Right after the democratic revolutions of Eastern Europe, Adam Przeworski stated what was to become the conventional wisdom in the field: Projects of liberalization launched by forces from within the authoritarian power establishment are invariably intended as controlled openings of political space . . . Yet liberalization is inherently unstable . . . Once repression lessens, for whatever reason, the first

146

Analytical Framework

reaction is an outburst of autonomous organization in the civil society . . . Once that happens, liberalization can no longer continue . . . Liberalizations are either reversed, leading to grim periods euphemistically termed normalization, or continue to democratization. (Przeworski 1991: 57–60)

Scholars on electoral authoritarian (ea) regimes often choose these earlier insights on the inherent instability of authoritarian liberalization as their theoretical reference point. To indicate the destabilizing potential of authoritarian multiparty elections, they cite Samuel Huntington’s famous dictum: “liberalized authoritarianism is not a stable equilibrium; the halfway house does not stand” (1991: 137).6 The question is: If the contradictory mix of authoritarian governance and liberalization is unlikely to produce a sustainable equilibrium, does the same hold for the contradictory mix of authoritarian governance and political elections? Just like the liberalizing measures taken by military and single-party dictatorships, authoritarian elections are designed to stabilize authoritarian regimes. Are rulers able to fulfill their dreams of reaffirming their hold on power through controlled elections? Or are they destined to be “almost always disappointed” (Huntington 1991: 175) by authoritarian elections spinning out of control just the way processes of authoritarian liberalization tend to spin out of control?

Democratization by Elections If multiparty elections introduce systematic tensions into authoritarian regimes that make them prone to instability, do they push these regimes towards one side of the spectrum of regimes? If the “inconsistent regimes” located between the poles of democracy and closed dictatorships are inherently unstable (see Gates et al. 2006), do they collapse with equal probabilities to either side of the regime spectrum? Are they equally likely to enjoy democratic transitions as to suffer authoritarian regressions? Are processes of “democratization by elections” rare random events or do authoritarian multiparty elections exert a systematic pull towards political democratization? In my own earlier work, I was ambiguous on this point. On the one hand, I admitted that authoritarian elections could fulfill regime-sustaining as well as regime-subverting roles and declared their indeterminate effects to be contingent on the unfolding dynamics of the nested conflict between government and opposition. On the other hand, I claimed that “even if democratic progress is not inevitable, the inner logic of the game pulls it away from authoritarianism”: To the extent that opposition parties succeed in accumulating strength in the electoral arena, they improve their chances to extract institutional reforms from the

The Power of Elections

147

ruling party. And vice versa, to the extent that they succeed in improving the conditions of electoral competition and electoral governance at the meta-game level, they improve their chances of capturing votes and seats at the game level. Electoral success furthers electoral reform which again furthers electoral success. Rather than establishing a “self-enforcing” equilibrium, ambivalent elections thus tend to trigger a “self-subversive” spiral that over time undermines both the institutional and the electoral bases of the authoritarian incumbent. (Schedler 2002a: 111, emphases added)

The key qualification here is “to the extent” which turns the statement almost tautological. My claim hinges on the interaction between the two game levels: if opposition parties accrue gains at the game levels, those are likely to spill over into gains at the meta-game level; if they accrue gains at the meta-game level, those are likely to spill over into gains at the game level. Even if we assume the causal link between the two game levels to hold, my formulation pushes the question back to the origins of democratizing changes at either level. If nothing happens at either level, nothing happens at the other. Yet why should we expect anything to happen? If multiparty elections are designed to prolong the lifespan of authoritarian regimes, why should we expect them to produce counter-intentional consequences? The answer is rather simple: To the extent that multiparty elections make authoritarian rulers dependent on others, they render them vulnerable to others. For rulers, the structural dependencies that elections introduce create structural vulnerabilities. For their allies, agents, and adversaries, they create structural opportunities for challenging the prevailing balance of power. When authoritarian rulers admit opposition parties, they admit the possibility that these parties are less loyal and timid than they expect them to be (in gratefulness for being admitted to the official arena of politics). Yet the structural threats emanating from authoritarian elections go beyond the obvious possibility that opposition parties take their role seriously. Electoral opportunity structures may invite regime allies, state agents, as well as citizens to withdraw their cooperation with the regime. 1. The desertion of state agents. Popular “elections entail the largest peacetime mobilization of the national population in a short time span and require the coordination of millions of individuals engaged in hundreds of different activities” (Mozaffar and Schedler 2002: 5). Especially in poor, large, and diverse countries, they pose enormous administrative and logistical challenges to governments, be they democratic or non-democratic. For authoritarian governments striving to contain the uncertainty of electoral outcomes, multiparty elections with universal suffrage involve problems of agency control at a massive scale. Their strategies of electoral manipulation depend on the active cooperation of state agents capable and willing to carry them out in an effective fashion. Their authoritarian schemes

148

Analytical Framework

may fail due to the technical incapacity or the political unreliability of state actors. It is purely and simply a problem of large numbers. Just imagine. If electoral autocrats had to control a small face-to-face electorate only, such as an assembly of notables, the central committee of the ruling party, or a town hall meeting, they would need no more than a small staff of loyal political operators. Even their personal presence might suffice to carry the vote. They would have (almost) none of the principal–agent problems they face when trying to domesticate competitive mass elections—that dizzying modern invention, egalitarian, bureaucratic, and decentralized. 2. The desertion of elites. As they provide institutional means for reward and punishment, training and exposure, competition and compliance, and selection and rotation of state personnel, authoritarian elections often serve as effective devices of elite coordination. They may also, however, serve as effective triggers of elite desertion. Under conditions of hegemony, incumbents are practically immune to internal splits as long as they can convince (temporarily) discontented members of the ruling alliance that it is only within the hegemonic party that they can maintain “a fair chance they will win posts and privilege in the future” (Langston 2006: 60). In competitive autocracies, by contrast, the perennial threat of ambitious regime insiders crossing over to the ranks of the opposition is more immediate and pressing. Existing opposition parties and candidates are often too disorganized and unpopular to unseat the incumbent at the polls. If anyone is capable of defeating the incumbent, it is someone from the inner ranks of the regime. The Ukrainian electoral rebellion of the year 2004 illustrates these dynamics, as virtually “the entire leadership of the Orange Revolution had . . . been closely allied with the president just a few years prior” to the collapse of the regime (Way 2008: 63). In both hegemonic and competitive regimes, opposition actors are often resigned to the apparent fact that serious challenges may only arise “from within,” forcing them to “wait on a split within the regime” (Pepinsky 2007a: 126). 3. The desertion of citizens. In addition to “usable states” (Linz and Stepan 1996), electoral manipulation requires pliable citizens who let themselves be persuaded, intimidated, or corrupted. Authoritarian elections, however, afford more than opportunities for individual passive resistance. They provide structural opportunities for contentious collective action. Few areas of political research possess anything resembling a unifying paradigmatic core. Exceptionally, the study of social movements does so. Students of social movements basically know what it needs to get protesters onto the streets. They know what to look for if they want to explain the emergence of contentious collective action: grievances (appropriately framed to resonate with relevant audiences), repertoires of collective action (such as public

The Power of Elections

149

demonstrations, strikes, and street barricades), mobilizing structures (such as social networks or formal organizations), and political opportunities (positive expected utilities of collective action under given circumstances). Authoritarian elections provide all this. They create profound democratic grievances as well as a powerful master frame of protest: the defense of democracy. They provide the opposition with a simple low-cost medium of communication (the vote) and a simple common message (a protest vote against the regime). They circumscribe the time and place for collective action. They energize political parties, the organized pillars of competitive elections, mobilize civic associations, and activate social networks in alliance with either parties or civil society organizations. They offer a range of roles citizens may assume, as voters, grassroots activists, party leaders, or candidates, as well as a range of collective activities they may carry out, such as campaigning, voter education, electoral observation, and protest demonstrations.7 Authoritarian multiparty elections, in sum, place severe constraints on opposition actors, yet also provide structural opportunities for individual as well as collective challenges. They create focal points of defection and coordination that facilitate subversive action by regime dissidents at both elite and mass levels.

T H E D I V E R S IT Y OF E M PIR I C A L F I N D I N GS What do we know empirically about the systemic consequences of authoritarian elections? Do they matter, how much, and what for? Over the past years, a growing number of comparative studies have put the contrasting hypotheses about “the power of elections” to statistical testing. The preliminary balance sheet of this emerging stream of large-N studies of the effect of elections on the fate of authoritarian governments is not evident at first sight. As in other substantive areas of cross-national statistical inquiry, results are mixed, sometimes contradictory, and not readily reconcilable, since studies differ in data selection, model specification, and statistical procedures as well as in geographical and temporal coverage. In quantitative studies of elections and authoritarian survival, the diversity of findings in the main seems to originate from broad differences in research purpose: the types of elections whose causal impact authors wish to test (the definition of independent variables) and the kinds of phenomena they want to explain (the definition of dependent variables). Regarding their independent variables (explanans), some studies look at broad categories of national elections, including de jure or de facto single-party

150

Analytical Framework

elections. Others focus more narrowly on multiparty elections and some further distinguish between hegemonic and competitive contests. Only a few authors study the internal dynamics of authoritarian elections. Regarding their dependent variables (explanandum), some studies try to explain the occurrence and direction of regime change, others the survival and fate of authoritarian rulers, still others continuous changes in levels of political rights and civil liberties. These incongruent conceptualizations of independent and dependent variables have given rise to incongruent explanatory enterprises. If we wish to make sense of the diversity of empirical findings, we need to be aware of the diversity in purpose and design of extant empirical studies. In the following, I will review cross-national statistical inquiries into the “power of authoritarian elections” according to their explanatory purpose. Table 5.1 provides an overview over about 20 large N-studies, their temporal and spatial coverage, their definition of explanatory and dependent variables, their variables of control, and their essential empirical findings. I will order and discuss these studies, in the first place, according to their conception of causal variables (types of elections); in the second place, according to their conception of dependent variables (consequences on regimes and rulers).

Multiparty Elections and Authoritarian Survival If we ask whether multiparty elections (mpe) prolong the duration of political regimes or tenure of political rulers, their natural points of comparison are regimes without elections and regimes with single-party elections (spe). The null hypothesis is, of course, that all authoritarian parties and elections are irrelevant, much institutional ado about nothing. The opposite assumption that authoritarian institutions do matter allows for four theoretically coherent possibilities: 1. Causal homogeneity: Single-party and multiparty elections show similar effects: they both increase the life expectancy of regimes (spe = mpe). 2. Superiority of single parties: Both types of elections prolong the lifespan of regimes, but single-party elections work better (spe > mpe). 3. Superiority of multiple parties: Both types of elections make regimes more durable, but multiparty elections work better (spe < mpe). 4. Contrary effects: Single-party elections prolong the life of regimes, multiparty regimes shorten it (spe ≠ mpe). The first hypothesis is consistent with the idea that institutions are authoritarian tools. Any institution that dictators care to establish should serve them well. The second hypothesis corresponds to the notion that single-party

Table 5.1. Large-N research on the power of authoritarian elections Authors and studies

Independent variables / units of analysis

Dependent variables

Statistical controls

Sample coverage

Empirical findings

Regime duration

None

1946–2005, Military and personalist regimes that authoritarian hold (at least occasional) direct regimes national elections tend to survive worldwide longer on average than those without elections. Single-party regimes survive longest (the category includes hegemonic regimes).

All elections → Regime change Geddes 2005

Authoritarian regimes with / without elections, regular / irregular elections

Multiparty elections → Regime duration and leadership survival Brownlee 2007a

Formal multiparty (“limited”) elections

Authoritarian regime breakdown

gdp per capita, growth, region, regime age

1975–2000, Military regimes are less, single-party authoritarian regimes more stable than personalist regimes regimes. The presence of elections worldwide does not affect the probability of authoritarian breakdown (limited elections include de jure multiparty elections in which only one party wins seats).

Magaloni 2008

Electoral authoritarian (ea) regimes (de facto multiparty “hegemonic” regimes)

Regime duration

None

1950–2000, Annual survival probabilities of ea world, except regimes are almost identical to advanced single-party regimes. Democracies oecd have higher, military regimes lower democracies probabilities of survival. (Continued)

Table 5.1. (Continued) Authors and studies

Independent variables / units of analysis

Gandhi 2008

Ordinal measure of Authoritarian regime institutionleadership alization (no / tenure single / multiple legislative parties)

Dependent variables

Statistical controls

Sample coverage

Empirical findings

Military regimes, 1946–2002, Institutions neutralize societal threats. external and authoritarian Controlling for levels of threat, internal war, past regimes institutionalization bears no effect on coups, territory, worldwide the tenure of authoritarian rulers. gdp per capita

Reanalysis of Gandhi Dummies: Authoritarian 2008 single-party / leadership multiparty regimes tenure

Military regimes, 1946–2002, The relation of “institutionalization” external and authoritarian and leadership survival is curvilinear. internal war, past regimes Single parties prolong the tenure of coups, territory, worldwide chief executives by more than three gdp per capita years, multiple parties by about one. Institutions affect leadership tenure regardless of pre-existing levels of threat.

Gandhi & Przeworski Deficient / optimal Authoritarian 2007 / excessive leadership institutionalization tenure (actual relative to predicted numbers of legislative parties, as a function of survival threats)

Military and 1946–1996, Rulers without institutions have shorter civilian regimes, authoritarian tenures (ø = 4.3 years) than those mineral exports, rulers (with presiding multiparty (ø = 8.2 years) or previous political completed single-party regimes (ø = 9.8 years). parties, previous spells) Rulers who under-institutionalize leadership worldwide in response to threats survive less changes, long (ø = 3.3 years) than those who democracies in institutionalize optimally (ø = 6.9 the world years) or over-institutionalize (ø = 9.4 years).

Multiparty elections → Direction of regime change and exit routes of leaders Cox 2007

Political regimes: democracy / nonelectoral / singleparty / multiparty autocracy

Authoritarian None leadership survival, modes of exit

1975–2000, all regimes worldwide

Cox 2009

Political regimes: as in Cox 2007

Authoritarian gdp, civil war, leadership country and survival, modes regional fixed of exit effects

1975–2000, Authoritarian leaders face lower risks authoritarian of violent exit in electoral regimes. regimes Risks of peaceful exit are higher worldwide in multiparty regimes than in single-party regimes.

Schuler & Gueorguiev 2010

Elections with / Probability and without opposition direction of (with / without transition one-year lag)

Wright and Escribà-Folch 2009

Presence of political parties (ordinal scale: no / single / multiple parties)

gdp level and Unclear (N = growth, 3,514 regime ethno-linguistic years) fractionalization, regional and decade dummies, military regimes

Continuity in office is higher in ea regimes than in either democracies or non-party autocracies. Multiparty autocracies have higher rates of peaceful exit than military regimes, yet lower rates of violent exit. Both peaceful and violent exits are least frequent in single-party regimes.

Multiparty elections dampen the risk of irregular exits from power as well as the risk of democratization (although they heighten both risks in post-election years). Single-party elections dampen the risk of irregular transfers of power, yet have no impact on the odds of democratization, and involve no post-election risks.

Probability and Geddes regime 1946–2002, In non-personalist regimes, parties direction types (military, authoritarian increase the likelihood of democratic of regime single-party, regimes transition and decrease the likelihood transition; personalist), gdp worldwide that leaders suffer imprisonment or post-exit fate of per capita, growth, death after quitting power. leaders democratic neighbors (Continued)

Table 5.1. (Continued) Authors and studies

Independent variables / units of analysis

Dependent variables

Statistical controls

Sample coverage

Empirical findings

Magaloni & Kricheli 2010

ea regimes (“dominant-party regimes”)

Probability and direction of transition

None

1950–2000, all regimes worldwide [unclear]

Monarchies and democracies are the most stable, military regimes the least stable type of regimes. Like single-party regimes, ea regimes show intermediate annual survival probabilities. They are more likely to transition to democracy than single-party regimes, yet less likely to descend into “anarchy” (civil war or caretaker governments). Mutual transitions between single- and dominant-party regimes are frequent.

Period effects: 1987–1994: ea regimes are less stable than democracies, no less stable than closed regimes. 1995–2006: Post-Cold War hegemonic regimes are more stable than closed regimes, less than democratic regimes. Competitive regimes are the least stable regime category.

Hegemonic vs. competitive elections → Probability and direction of regime change Roessler & Howard 2009

Hegemonic / competitive ea

Regime duration

None

1987–2006, all regimes worldwide

Brownlee 2009

Hegemonic / competitive authoritarian elections

Democratization (direction of regime breakdown)

gdp per capita, growth, Middle East, regime age, Geddes regime types (military, single-party, personalist)

1975–2004, Competitive elections augment the cases of probability of democratization, authoritarian hegemonic elections do not. Except regime for military-personalist regimes, breakdown none of the conventional regime worldwide types affect the direction of regime breakdown.

Hadenius & Teorell 2007

Hegemonic / competitive ea (“dominant / non-dominant multiparty regimes”)

Regime change (probability and direction)

None

1972–2003, all regimes worldwide

Closed regimes (in particular monarchies and single-party regimes) and democracies are most durable, ea regimes most fragile. Most transitions from single-party and military regimes lead to ea regimes. Transitions from hegemonic regimes tend to lead to competitive regimes, transitions from competitive regimes to democracy.

Multiparty elections → Political liberalization Lindberg 2006b

Multiparty elections

Continuous gdp per capita, 1990–2003, annual changes growth, human sub-Saharan in Freedom development, Africa House (fh) oda, oil exports, scores of civil religion, ethnic liberties fractionalization, popular protests

Unbroken strings of multiparty elections improve civil liberties.

Lindberg 2009c

Multiparty elections

Continuous Similar controls annual changes as in Lindberg in civil liberties 2006b (aggregate and disaggregate fh scores)

Unbroken strings of multiparty elections improve civil liberties (above all, individual rights and rule of law).

1990–2006, sub-Saharan Africa

(Continued)

Table 5.1. (Continued) Authors and studies

Independent variables / units of analysis

Weiss 2007

National elections Continuous gdp per capita, size (either presidential annual changes of government, or parliamentary) in civil liberties trade openness, (fh scores) prior levels of political rights

Teorell & Hadenius 2009

All elections / Continuous Colonial 1974–2002, multiparty annual changes background, all regimes elections / legacy of in political religion, size, worldwide elections rights and civil socio-economic liberties (fh development, scores) growth and inflation, democratic diffusion, political conflicts (among others)

No effects of non-competitive elections. No robust short-term effects of multiparty elections. Substantively small effects of historical legacies of elections.

Teorell 2010

Regime types: Continuous Comprehensive democracy, annual changes controls similar monarchy, military, in political to Teorell and one-party, and rights and civil Hadenius (2009) multiparty liberties (fh autocracy. scores) Intervening variables: external shocks

“Multiparty autocracies democratize to a larger extent both in and of themselves and in response to both popular protest and short-term economic performance” (L 3045).

Dependent variables

Statistical controls

Sample coverage

Empirical findings

1972–2005, all regimes worldwide

At intermediate levels of civil liberties (fh scores 4–6), elections bear a positive impact on civil liberties. At extreme levels (in closed autocracies or democracies), they bear none.

1974–2005, all regimes worldwide

Actor strategies → Political liberalization Howard & Roessler 2006

Actor strategies: opposition coordination and protest, candidacy by incumbent

“Liberalizing Prior levels and 1990–2002, electoral changes in civil competitive outcomes” liberties, system ea regimes (leos): of government, worldwide election-year gdp growth, oda, improvements fdi in fh and Polity scores

Relative frequency of leos: 15/50. Internal factors that determine the democratizing effects of competitive authoritarian elections: opposition behavior and electoral participation by incumbent. Contextual factors are irrelevant.

Pop-Eleches & Robertson 2008

Intensity of manipulation: non-competitive / seriously flawed / somewhat flawed / free and fair elections

Improvements in fh and Polity democracy scores (“liberalizing moments”)

Most elections are not liberalizing, but most liberalizations are associated with elections. Neither sham elections nor free and fair elections are liberalizing. Flawed elections “are critical political events that have an independent effect on the likelihood of political liberalization” (p. 19). In election years, protests mobilization, ethnic fractionalization, inequality, corruption, external democracy assistance, and democratic neighbors reinforce the liberalizing potential of elections.

gdp per capita, 1992–2006, all growth, inflation, regimes with natural resources, fh Political inequality, Rights scores democratic >1 diffusion, population, urbanization, prior levels of political rights

158

Analytical Framework

elections are institutions of domination and multiparty elections institutions of representation. The former are effective tools in the hands of dictators, the latter more ambivalent arenas (which is the analytical perspective I embrace in this book). The third hypothesis derives from the idea that multiple parties represent higher levels of authoritarian institutionalization. The more parties, the better for the ruler (see below). The fourth hypothesis is consistent with approaches that hold multiparty elections to be intrinsically destabilizing or liberalizing or democratizing. To what extent does the extant literature allow to arbitrate between these competing hypotheses? If we wish to learn something about the differential effects multiparty and single-party elections bear on the endurance of authoritarian rule, we need at least three things: (a) appropriate levels of conceptual differentiation; (b) appropriate units of analysis; and (c) systematic comparison. (a) Conceptual differentiation. A couple of studies have explored the most general theoretical expectation according to which authoritarian elections of any kind serve the authoritarian purpose of prolonging the status quo (beyond its counterfactual life expectancy without elections). They tend to support this general expectation. According to Barbara Geddes (2005), authoritarian elections work as they are expected to, extending the lifespan of non-democratic regimes. As her figures for the years 1948 through 2005 suggest, both regular and irregular elections serve this purpose and they do so in any kind of dictatorship, whether military, personalist, or single party.8 Geddes’ data lend support to the notion that dictators are better off with elections than without. For her empirical explorations, she could rely on a simple registry of national elections, without further conceptual distinctions. Yet, if we wish to assess whether single-party elections serve dictators better than multiparty elections, we need to distinguish between these types of elections in valid and reliable ways. Since de jure multiparty systems often mask de facto single-party dominance, drawing this distinction is not as quick and easy as it may seem at first glance. Conceptual stretching is a perpetual danger. Over-inclusive notions of single-party systems are one form of conceptual overstretching, over-inclusive notions of multiparty regimes another. From my perspective, we conceptualize single-party systems too broadly when we classify hegemonic multiparty autocracies as one-party regimes (e.g. Geddes 1999 and Magaloni and Kricheli 2010).9 And we conceptualize multiparty systems too broadly when we take them also to include formal multiparty regimes that run de facto single-party elections (e.g. Brownlee 2007a, Lindberg 2006b).10 In

The Power of Elections

159

the former case, we grant too little importance to existing opposition parties, in the latter too much to non-existing opposition parties. Either way, we blur the distinction between single-party and multiparty elections. (b) Units of analysis. Single-party regimes are closely associated with authoritarian durability. They have been the champions of longevity among modern autocracies (see Geddes 1999). Existing cross-national studies probably overstate their durability. The high average duration of single-party regimes masks “wide variation in regime tenure” (Smith 2005: 423) within the family of single-party regimes. Case samples are historically truncated. Since few cross-national data are available for earlier years, most studies focus on the period after 1945 and thus exclude the comparatively short-lived Nazi regime and its fascist allies inside and outside Europe. Also, following dubitable conventions of cross-national research, authors commonly treat the East European satellites of the Soviet Union as independent cases, even though their position within the regional power bloc was analogous to the dependent status of subnational units in federal states. It seems clear nevertheless that single-party regimes have indeed enjoyed “exceptional durability” (Gandhi and Przeworski 2007: 1292) during the second half of the 20th century (even if few of them survived the end of the Cold War). It is unclear, however, whether single-party elections carry any causal weight of their own on the longevity of single-party regimes. Studies of both single-party and multiparty autocracies often do not focus on elections, but parties (e.g. Wright and Escribà-Folch 2009). Yet, while multiparty systems and elections are closely related, single-party regimes often manage without elections. According to the Arthur Banks Cross-National Time–Series Data Archive (cnts), among those regimes that do not admit multiple parties, 59.9 percent have a non-elective head of government and 38 percent either have no legislature or a non-elective one. By contrast, among those (authoritarian as well as democratic) regimes that do allow for multiple parties (albeit at varying levels of inclusiveness), only 5.3 percent have non-elective chief executives and 1.7 percent do not have elective legislatures.11 Thus while elections are a regular feature of multiparty regimes, their presence is contingent in single- and non-party regimes. (c) Comparison. Following the lead of Jennifer Gandhi, some scholars conceive single-party and multiparty systems as graded instances of one underlying phenomenon: the degree of authoritarian institutionalization. In her Political Institutions under Dictatorship, Gandhi (2008) studies the role that “nominally democratic institutions” (p. xviii) play in staving off vertical threats to the political survival of authoritarian leaders (rather than political regimes). She conceives both single-party and multiparty elections as

160

Analytical Framework

“nominally democratic” institutions and aligns them along an ordinal scale of authoritarian regime “institutionalization”: non-institutionalized dictatorships contain no legislatures or no parties (coded 0), semi-institutionalized dictatorships endow a single party with control over the national assembly (coded 1), and institutionalized dictatorships accommodate multiple legislative parties (coded 2). Within my own analytical framework, the distinction between single-party and multiparty systems is a qualitative one, a difference in kind, not degree. Single-party systems (as well as associated elections) are institutions of domination, multiparty systems (as well as associated elections) formal institutions of representation. The former are simple tools of authoritarian governance, the latter complex carriers of contradictions. By admitting multiple parties instead of one, authoritarian regimes do not admit more of the same thing. They take a qualitative leap into a different political world. If we understand single-party and multiparty regimes as continuous phenomena, we unsettle our coordinates of comparison. Whatever the effects of “authoritarian institutionalization,” we never know which types of “world jump” they derive from. They may originate in systemic transformations from no parties to single parties, from single parties to multiple parties, or from no parties to multiple parties. The potential differences between single-party and multiparty elections turn invisible.12 In her work, Gandhi finds confirming evidence for the third hypothesis listed above (on p. 150), the superiority of multiparty elections for authoritarian survival. However, as I wish to argue, measurement matters. If we treat her institutional variables as categorical, rather than ordinal, her data lend credence to the second hypothesis, the superiority of single-party elections. Gandhi’s analytical framework centers on the management of societal threats. In her theory, authoritarian leaders establish small decision-making bodies, like military juntas or royal councils, to handle lateral threats from within the ruling clique. If these prove insufficient, they establish formally representative institutions as a “second institutional trench” (Gandhi 2008: 166) to stave off “threats from larger groups within society” (Gandhi 2008: 164). In tranquil situations of low threats, rulers will survive without institutions. In dangerous situations of high threats, rulers will do so with the help of institutions. As Gandhi suggests, we should not expect the introduction of representative institutions to extend the political life of dictators in absolute, only in relative terms—relative to the intensity of threats to their political survival. Just as the intake of certain drugs will not prolong our lives in the absence of the diseases they are supposed to cure, the establishment of institutions will not extend the political lives of dictators in the absence of the societal threats they are supposed to contain.

The Power of Elections

161

To test the conditional effects of representative institutions, Gandhi uses her ordinal scale of authoritarian institutionalization as independent variable. Since she expects the “degree of institutionalization” to extend the tenure of authoritarian rulers only in the presence of strong societal threats, she controls for the presence of violent societal threats (previous military coups, international and domestic war) as well as for other factors likely to influence the intensity of societal conflict (country size and wealth). Her parametric event history model yields an intriguing set of negative results: nothing is significant, except war and military regimes, both shortening the life expectancy of rulers (Gandhi 2008: 177). Gandhi interprets the apparent irrelevance of political institutions as theory-confirming. From the fact that “[d]ictators with institutions on average do not survive longer than those without” (Gandhi 2008: 176), she concludes that rulers “respond with the appropriate degree of institutionalization” to the “threats germinating within society” (Gandhi 2008: 177). These negative findings as well as their positive interpretation are suggestive. What happens, though, if we abandon the assumption that single-party rule and party pluralism are no more than graduations along a one-dimensional continuum? What happens if we abandon the corresponding expectation that the relationship between no elections, single-party elections, and multiparty elections with authoritarian leadership duration should be lineal? Why not expect a curvilinear relation? As straightforward institutions of domination, single parties should work better in securing the longevity of dictatorship than multiple parties, two-edged formal institutions of representation. To explore the impact of authoritarian elections on the survival of dictators in a conceptually less restricted manner, I did a simple reanalysis of Gandhi’s data: (i) I transformed her ordinal variable of institutionalization into two dummy variables, one for single-party regimes and one for multiparty regimes. (ii) I explored the possibility that institutions may bear unconditional, rather than mediated effects on leadership survival. (iii) Sacrificing methodological sophistication for transparent simplicity I ran lineal ols regressions instead of log-likelihood survival models. The results presented in Table 5.2 lend credence to the notion that ordinal measures of institutionalization mask non-lineal relationships between institutional configurations and the term length of authoritarian leaders. No-party, single-party, and multiparty regimes show an inverted-U relationship to leadership survival. If I employ Gandhi’s ordinal data, each step along a three-point scale of institutionalization prolongs the political life expectancy of dictators by little more than half a year (b = .66). If I use my categorical measures instead, the introduction of single parties appears to prolong the average tenure of authoritarian heads of government by almost 3.5 years (b = 3.44), the admission of multiple parties by roughly one year (b = 1.16). The

162

Analytical Framework

Table 5.2. Institutional determinants of leadership tenure, 1946–2002 Ordinal scale of institutionalization B Without controls Constant Institutionalization Single-party regimes Multiparty regimes Standard error R2 N With controls Constant Institutionalization Single-party regimes Multiparty regimes Military regime War Turnover by coup Land Per capita income Standard error R2 N

9.867 .664

Institutional dummies p

.000 .000

9.480 .003 4.298 11.865 .591

–.324 –3.061 –15.813 –.046 .088 8.999 .103 3.245

.000 .004

.358 .000 .000 .000 .042

B

p

9.028

.000

3.441 1.165 9.387 .023 4.298

.000 .001

9.993

.000

3.843 1.347 .150 –2.863 –14.347 –.032 .181 8.886 .125 3.245

.000 .001 .671 .000 .000 .010 .000

Source: Gandhi (2008). Dependent variable: age = years in office of effective head of government. Institutionalization: 0 = no national legislature or no parties, 1 = legislature + single party, 2 = legislature + multiple parties. Sample: all non-democratic regimes worldwide, 1946–2002. ols linear regression. B = unstandardized regression coefficients. Shaded coefficients = statistically significant (p d .15).

inclusion of controls does not alter the overall magnitude of these non-lineal institutional effects, yet puts them into perspective. The burdensome construction of a single party adds three meager years to the life expectancy of an authoritarian ruler, while a previous history of military coups takes off almost a decade and a half (b = –14.34). Thus, if single-party elections are more effective than multiparty elections in prolonging the grip of authoritarian rulers on power, their comparative advantage seems to be small. Other comparative inquiries confirm this pattern. In a statistical study covering the last quarter of the 20th century,

The Power of Elections

163

Jason Brownlee (2007a) rejects the hypothesis of electoral destabilization. He finds that the mere presence of de jure multiparty elections fails to affect the probability of authoritarian regime breakdown. Similarly, within the universe of authoritarian regimes in the second half of the 20th century, Beatriz Magaloni (2008) does not encounter significant differences in regime duration between single-party and multiparty autocracies. She posits that multiparty elections stabilize distributive coalitions. They empower the members of the ruling alliance vis-à-vis the chief executive and thereby keep the latter from reneging on his clientelist commitments. Her data support the idea that multiparty elections fulfill stabilizing roles. Both single-party regimes (p = .049) and multiparty regimes (p = .066) have lower annual transition probabilities than military dictatorships (p = .104), though higher ones than democracies (p = .019) (p. 736). Based on the same theoretical intuitions and institutional data as Gandhi (2008), Gandhi and Przeworski (2007) hypothesize that rulers defensively choose levels of regime institutionalization (the number of legislative parties) in order to neutralize observed societal threats through policy concessions.13 To estimate “optimal” levels of institutionalization, the authors estimate predicted levels of institutionalization by regressing levels of societal threat on given numbers of legislative parties (none, one, or multiple). They measure threat levels by access to resource rents, previous leadership instability, previous institutions, the presence of military regimes, and the percentage of democracies in the world. Based on these predictions, they compare predicted with actual numbers of legislative parties. As they find, regardless of threat levels, rulers without institutions have shorter mean tenures (ø = 4.3 years) than those presiding multiparty (ø = 8.2 years) or single-party regimes (ø = 9.8 years). Rulers who under-institutionalize in response to threats survive less long (ø = 3.3 years) than those who institutionalize optimally (ø = 6.9 years) or over-institutionalize (ø = 9.4 years). Although doubts about the ordinal nature of the distinction between no-party, single-party, and multiparty regimes remain, these findings support the claim that political institutions serve to counteract structural vulnerabilities of authoritarian regimes. They also support the notion that single-party regimes work marginally better than multiparty regimes in diffusing societal threats. They do raise the question, though, why rulers would strive to achieve no more than average rates of political survival (by balancing threats through “optimal” levels of institutionalization) rather than maximum rates (by pre-empting threats through “excessive” levels of institutionalization). Why do dictators want to rule as long as everybody else, instead of aspiring to rule as long as possible?

164

Analytical Framework Multiparty Elections and Regime Trajectories

If authoritarian rulers wished to maximize the duration of the regimes that they preside over, available data suggest that they should set up one of the three most durable regime types: a monarchy, a single-party regime, or a democracy. Between 1972 and 2003, monarchies endured on average for 25.4 years, single-party regimes for 17.8 years, and democracies for 17.5 years (Hadenius and Teorell 2007: 150). Unfortunately, all three regime types carry certain disadvantages. The disadvantage of democracy is the loss of power: “dictators are dictators because they cannot win competitive elections” (Gandhi and Przeworski 2007: 1281). Although authoritarian rulers are likely to care about regime continuity, their primary concern should be their personal survival in power. The disadvantage of monarchy is its lack of feasibility. Few dictators in the modern world command the cultural and communicative means to invent the tradition that sustains the legitimacy of hereditary succession.14 The drawback of single-party rule is similar to monarchies. With few exceptions, single-party rule is not a feasible option anymore. It no longer forms part of the choice set that contemporary autocrats face. At the end of the Cold War, most sub-Saharan countries and all post-Soviet countries had already lived through long nights of single-party rule in the second half of the 20th century. Finding themselves under multiple domestic and international pressures after the disintegration of the Soviet Union, they could not simply choose to perpetuate the old political monopoly that had steered their countries into political oppression and economic ruin. The remaining systemic alternative to electoral authoritarianism is non-electoral, no party dictatorship, either in the form of direct military rule or in the form of personal rule with military backing. Military regimes are relatively short-lived. Hadenius and Teorell report an average lifespan of 11.1 years (Hadenius and Teorell 2007: 150). They tend to end soon, but even worse, they tend to end badly. If we broaden our concerns from issues of regime and leadership duration to issues of regime and leadership trajectories (forms of political death and afterlife), we can appreciate more easily the comparative value of multiparty regimes. A couple of years ago, Gary Cox (2007) introduced an important distinction into the debate on authoritarian institutional choice: the contrast between violent and non-violent exits from power. He hypothesizes that rulers care about their political welfare and survival—but even more so about their physical welfare and survival. The average dictator resents losing power, but even more going to jail or losing his life. His first preference will be to remain in power. Yet when confronting the alternative of being evicted from office through either peaceful or violent means, he will opt for the former as the

The Power of Elections

165

lesser evil. Political survival matters, but forms of death do so, too. It is the potential trade-off between political and physical survival that may motivate the organization of multiparty elections under authoritarian rule. Elections come at a cost. They “always involve some risk” (Geddes 2005: 4) and may end up shortening the career of the authoritarian ruler. Rational dictators can trade this marginal risk against the benefit of lowering the probability of suffering a violent overthrow. To test his theory of exit preferences, Cox uses annual data from the Archigos Database on Political Leaders for the years 1975 through 2000 on three career trajectories of national leaders: their survival in office, their peaceful exit, and their violent exit from office. Table 5.3 shows the frequency distribution he finds across four types of political regimes: non-electoral regimes, single-party regimes, electoral autocracies, and democratic regimes.15 The data suggest some compelling regularities: Democratic leaders are most likely to exit office peacefully, and least likely to do so violently. Dictatorships without parties or elections suffer few peaceful exits, yet are most vulnerable to violent ones. Single-party dictatorships are the best of all authoritarian worlds: they suffer the fewest exits from power, be they peaceful or violent. On both accounts, multiparty autocracies fare worse than single-party regimes. Why then should autocrats ever conduct multiparty contests instead of single-party elections? If single-party elections were still a feasible option, dictators might go for them more often. Yet since the end of the Cold War they usually aren’t. Contemporary autocrats thus adopt multiparty elections not because they are preferable to single-party elections, but because they are preferable to the remaining feasible alternatives: military rule without elections or a full transition to genuine democracy. Gary Cox’s data show why. On the one hand, although multiparty autocracies do contain some degree of electoral risk, their annual vulnerability to electoral turnovers is about two-thirds lower than under democratic conditions. On average, democratic governments suffer electoral defeat every five years, electoral authoritarian governments every fourteen years. On the other hand, their exposure to irregular removals from office is about three-quarters lower than in non-electoral regimes. On average, closed autocracies experience violent removals from power once in ten years, electoral authoritarian governments once in twenty-five years (see Table 5.3). In sum, when single-party rule is not a feasible option, authoritarian rulers choose multiparty arrangements over democracy because they want to survive in office; and they choose multiparty arrangements over non-party regimes because they do not want to die in office. Multiparty elections improve the chances of achieving both goals, even if they contain the irrepressible risk of failure in either of them. The benefits of macro-institutional choices are

166

Analytical Framework Table 5.3. Peaceful and violent exits from office, by regime type

Continuity Peaceful exit Violent exit Total

N % N % N % N %

Non-electoral autocracies

Single-party autocracies

Electoral autocracies

Electoral and liberal democracies

793 84.5 51 5.4 94 10.0 938 100.0

785 93.9 44 5.3 7 0.8 836 100.0

722 90.3 58 7.3 20 2.5 800 100.0

1,633 78.3 432 20.7 20 1.0 2,085 100.0

Source: Cox (2007: Figure 1). Primary source: Archigos Database on Political Leaders, 1875–2004. Note: Data are annual event counts of the fate of national leaders (chief executives), worldwide, 1975–2000. Ch2 = 418.2, p = .000.

probabilistic, not deterministic. Multiparty elections prolong the expected lifespan of non-democratic regimes. They do not buy them immortality. They depress the risk of electoral defeat, yet do not eliminate it. They depress the risk of military coups, but do not eliminate it either (see Clark 2006). Other large-N studies that have examined the comparative trajectories of authoritarian regimes, rather than their mere duration, tend to support the notion that electoral autocracies play a two-sided protective role: they shield rulers from the rough winds of democratic competition as well as from the tempests of military interference. Paul Schuler and Dimitar Gueorguiev (2010) find that authoritarian multiparty elections tend to dampen the risk of irregular exits from power as well as the risk of democratization (although they heighten both risks in post-election years). Joseph Wright and Abel Escribà-Folch (2009) establish that parties and legislatures do not affect the fate of leaders in personalist regimes. That is, they are inconsequential in those regimes in which political institutions are inconsequential by definition. By contrast, in more institutionalized regimes, the number of political parties (again aligned along the ordinal scale of no, single, and multiple parties) increases the likelihood of democratic transition and decreases the likelihood that leaders suffer imprisonment or death after quitting power. Finally, Beatriz Magaloni and Ruth Kricheli (2010) calculate that multiparty autocracies are more likely to transition to democracy than single-party regimes and less likely to descend into “anarchy” (civil war or caretaker governments).16

The Power of Elections

167

Hegemonic versus Competitive Elections Whatever it is that empirical studies find about the systemic effects of multiparty elections, it is likely to be confounded by the fact that multiparty regimes are internally heterogeneous. As I posited in Chapter 3, the family of electoral authoritarian regimes divides into two branches, one hegemonic branch in equilibrium and one competitive branch in disequilibrium. We should not expect these branches to be causally homogeneous. Rather, we should expect them to differ in their average duration as well as in their trajectories. By definition, hegemonic regimes are long-lived. If they change, we should expect them to turn into competitive regimes. Given the resourcefulness of hegemonic parties, they are unlikely to collapse or to give way to full democratization. Competitive regimes, by contrast, are more fl uid and contested and thus more open to experience electoral transitions to democracy. If we throw all authoritarian multiparty elections into a single conceptual basket, we mask the divergences between hegemonic and competitive regimes. We also overlook the transitions that may take place between the two. Unfortunately, only few large-N studies on authoritarian multiparty elections distinguish between hegemonic and competitive regimes. They tend to confirm their contrasting dynamics and thus the empirical relevance of the conceptual distinction. Mapping disaggregate regime trajectories since the late 1980s, Philip Roessler and Marc Howard (2009) encounter pronounced differences in hegemonic and competitive regime dynamics, in particular for the period between 1995 and 2006. In earlier years, between 1987 and 1994, the rate of regime change was similar for all types of authoritarian regimes. All were more vulnerable to regime change than democracies. Since the mid-1990s, however, hegemonic regimes have become more stable than closed regimes, while competitive regimes have turned into the least stable regime category (Roessler and Howard 2009: 121). In their comprehensive examination of regime transitions from 1972 through 2003, Axel Hadenius and Jan Teorell (2007) uncover similar patterns. They locate hegemonic regimes within a group of intermediate longevity. Less stable than monarchies, democracies, and single-party regimes, they last about as long as military regimes and no-party dictatorships. Again, the authors find competitive autocracies to be the least stable category of regimes. They also observe that a majority of transitions from hegemonic regimes lead to competitive authoritarianism—and from there a majority of transitions lead to democracy. Competitive autocracies thus appear to form “a typical stepping-stone to democratization” for hegemonic as well as for closed regimes

168

Analytical Framework

(Hadenius and Teorell 2007: 150–153).17 Consistent with these findings, Jason Brownlee (2009), in his study of authoritarian regime breakdown worldwide (1975–2004), encountered that transitions from competitive authoritarianism are more likely to lead to democracy than transitions from any other kind of authoritarian regime (except personalist military dictatorships). The overall conclusion seems straightforward. Theorists of authoritarian multiparty elections have formulated contradictory expectations. Some have expected such elections to sustain stable authoritarian rule, others to undermine it. The distinction between hegemonic and competitive elections dissolves the apparent contradiction. Hegemonic regimes appear to account for the longevity of electoral authoritarian regimes, competitive regimes for their vulnerability. Conceptual differentiation unveils causal heterogeneity.

Liberalization by Elections The notion that authoritarian elections affect the dynamics of political regime change need not be understood in categorical, all-or-nothing terms. Regime change may be conceived in qualitative terms, as transition from one category of regime to another, or in gradual terms, as incremental change within a regime category. Rather than provoking either qualitative institutional discontinuities (regime transitions) or personal alternation in power (leadership change), multiparty elections may have more modest, incremental effects on the nature of political regimes. In particular, the literature on “democratization by elections” (Schedler 2002a) stipulates that authoritarian multiparty elections may allow opposition actors to extract and accumulate small democratizing gains from the authoritarian incumbent. Through a series of battles, often at many levels and points, extended through space and time, elections may allow the opposition to establish a virtuous circle of progressive liberalization and democratization (see also Lindberg 2009a). Several large-N studies have put this hypothesis of the silent subversive power of elections to statistical testing. In his research on electoral cycles in sub-Saharan Africa since 1989, Staffan Lindberg (2006b) found that the regular conduct of multiparty elections tends to go hand in hand with steady improvements in their integrity, legitimacy, and competitiveness. In addition to increasing the “democratic qualities” of elections, uninterrupted sequences of electoral contests tend to irradiate into non-electoral spheres, too. In particular, they appear to carry “a significant and positive causal effect” (Lindberg 2006b: 144) on the improvement of civil liberties (as rated by Freedom House). These findings hold when controlling for a series of factors commonly associated with prospects of democratization (Lindberg 2006b: 139). As Lindberg succinctly concluded: “the more successive

The Power of Elections

169

elections, the more democratic the regime” (Lindberg 2006b: 139). In later updates, he replicated and refined these findings on the basis of disaggregate civil liberty scores (see Lindberg 2009c). Lindberg’s interpretation of regime trajectories in sub-Saharan Africa represents a strong version of “democratization by elections,” or more precisely, liberalization by elections. In and by themselves, Lindberg contends, elections deepen civil liberties. Regardless of how they look and how they go. They need not be perfect, they are self-perfecting. Repeating them does the trick. To examine the more general validity of Lindberg’s hypothesis of the liberalizing power of repeated elections, other authors have extended statistical testing to other regions and time periods. Their work has helped to specify the scope conditions of the underlying theoretical argument. The democratizing effect of elections is not supposed to be universal, but contingent on the type of election under analysis. In essence, it is circumscribed to national multiparty elections in authoritarian regimes.18 Neither limited elections nor elections organized by closed regimes nor those conducted under democratic conditions should be expected to provoke, through their mere existence, improvements in civil liberties or political rights. Limited elections that are circumscribed to subnational government or legislative assemblies are bound to be limited in their subversive potential. Single-party elections aid authoritarian rulers to perpetuate their denial of civil liberties and political rights. Expecting them to fulfill liberalizing functions would mean to misunderstand their very nature as institutions of domination. We should not expect either that democratic elections that take place under fundamentally liberal conditions trigger further gains in civil liberties or political rights. One reason is empirical: democratic elections are not democratizing in and by themselves. One is methodological: even if democratic elections were democratizing, our current democracy indicators would not show it. A third reason is theoretical: One subtype of elections, founding elections, are outcomes of democratization, not causes. Due to the structure of our annual datasets, though, they tend to create illusions of electoral transitions. 1. Powerless elections. Liberal–democratic theory holds electoral accountability to constitute the core mechanism that ensures the continual responsiveness of political elites to citizen demands (see Sartori 1987). High normative expectations rest on the shoulders of democratic elections. Yet the extent to which they actually work as “instruments of democracy” and deliver on the democratic promises of accountability, representation, and responsiveness, is contingent on institutional arrangements, party-systemic configurations, and political dynamics (see Powell 2000 and 2004). Empirically, in electoral

170

Analytical Framework

democracies, elections in and by themselves have been patently insufficient to move regimes towards democratic completion. As Philip Roessler and Marc Howard find (2009), most contemporary electoral democracies are stuck where they are, neither progressing towards liberal democracy nor reverting to authoritarian governance. Although meaningful elections are vital to the quality of democracy, they cannot, it appears, sustain the open-ended project of “democratizing democracy” on their own. If unsupported by “horizontal accountability” (O’Donnell 1994) as well as variegated forms of “societal accountability” (Smulovitz and Peruzzotti 2000), elections may be powerless to protect even basic rights and liberties, not to speak of the deepening or extension of democracy (see McCoy and Hartlyn 2009). The mere presence of elections may help to achieve democracy, but not to democratize democracy. 2. Truncated scales. In liberal democracies that have already reached the ceiling of cross-national democracy scores, truncated scales like Freedom House or Polity would be unable to register further democratization, even if elections were actually able to produce democratic deepening. 3. Optical illusions. In single-party regimes or military regimes that undergo transitions to democracy, the “founding elections” (O’Donnell and Schmitter 1986: 57) that they call are symptoms of democratization, not causes. Rather than preceding democratic breakthroughs, they follow them. Common measures of civil liberty and political rights tend to reflect these changes only afterwards, in the post-election year. If we credit these changes to the force of elections we get the direction of causality wrong. The flow of data that reports elections one year and improved country rating the next year creates optical illusions. Founding elections are democratic, not democratizing. In most cross-national electoral datasets, it is easy to identify single-party elections and distinguish them from multiparty elections. It is more difficult to delimitate authoritarian from democratic multiparty elections. Some statistical tests therefore aggregate both into the heterogeneous group of multiparty elections. For instance, in a comprehensive, careful study of electoral regimes in the 20th century, Jan Teorell and Axel Hadenius (2009) examine the short-term effects of elections on annual variations in civil liberties as well as the cumulative consequences of historical experiences with elections (controlling for about everything one can reasonably control for). They find feeble support for both types of democratizing effects: “weak but still notable” short-term effects (Teorell and Hadenius 2009: 99) as well as “cumulative although not very substantial” longer-term consequences (Teorell and Hadenius 2009: 100). To the extent that countries succeed in building continuous electoral histories,

The Power of Elections

171

the “marginal utility of elections” (Teorell and Hadenius 2009: 99) decreases over time. The general expectation according to which “elections have a democratizing effect” (Teorell and Hadenius 2009: 98) thus seems to be only weakly supported by longitudinal cross-national evidence. Yet, in its very generality, in its broad reference to “elections without adjectives,” this expectation seems to be overdrawn from the outset. If we wish to test its empirical validity, we need to subject the hypothesis of liberalizing or democratizing elections to appropriate scope conditions (excluding single-party elections, limited elections, and democratic elections). In his Determinants of Democratization, Jan Teorell (2010) accomplished this by examining the liberalizing propensities of various types of regimes (rather than elections): democracies, monarchies, military regimes, single-party, and multiparty autocracies. Tracing the trajectories of political regimes in the wake of “exogenous shocks,” he found that “[m]ultiparty autocracies democratize to a larger extent both in and of themselves and in response to both popular protest and short-term economic performance” (Teorell 2010: L 3045). Statistical explorations that maintain elections as their units of analysis, yet exclude both single-party elections and democratic elections, tend to confirm that the liberalizing power of elections is limited to intermediate categories of political regimes. Controlling for initial levels of civil liberty, David Weiss examined the short-term effects that elections carry on subsequent annual ratings of civil liberties within a worldwide sample of political regimes from 1972 through 2005. As he found, elections carry liberalizing effects only at intermediate levels of civil liberty. In terms of political liberalization, they are inconsequential at both extremes of the spectrum of political regimes (2007) (see also Lindberg 2009b: 13). If authoritarian multiparty elections (competitive elections in particular) carry the possibility of advancing democratization, probabilistic statistical techniques that estimate their average effects may still underestimate their systematic contribution to democratic change. If elections do indeed create opportunities for change, they do not impose anything like an iron law of electoral transformation. Consider, for instance, Marc Howard and Philip Roessler (2006) who examine the occurrence of “liberalizing electoral outcomes” in the wake of national elections in competitive authoritarian regimes. According to their data, most elections do not trigger short-term improvements in political rights and civil liberties (as measured through a combination of Freedom House and Polity data), but almost one-third does (30 percent, 15 out of 50). In a similar fashion, although relying on a different notion of political liberalization centered on the expansion of political rights, Grigore Pop-Eleches and Graeme Robertson (2008) analyze the capacity of authoritarian elections

172

Analytical Framework

to generate “liberalizing moments” in the post-Cold War era. As they find, most elections do not carry liberalizing effects (and “sham elections” do not carry any at all), yet most instances of liberalization are “closely associated with elections” (Pop-Eleches and Robertson 2008: 5). Both Roessler and Howard (2009) and Pop-Eleches and Robertson (2008) confirm the liberalizing potential of authoritarian multiparty elections. Significantly, to uncover the conditions under which elections activate this potential, both studies go beyond the analysis of structural conditions external to the electoral arena. Both do what this book sets out to do: examine actor dynamics within the electoral arena. Both take a glimpse into the black box of authoritarian elections.

Inside the Electoral Black Box When conceiving authoritarian elections as nested games in which ruling and opposition parties compete simultaneously for electoral support and electoral institutions, I conceive elections as arenas of struggle between actors, rather than causes of actor behavior. Arenas dictate neither actor strategies nor game outcomes. They open up spaces of conflictive interaction whose outcomes are contingent . . . upon the very dynamics of conflictive interaction. It is not the existence of the arena as such that determines the outcomes of conflicts, but the correlations of force and the strategic choices of actors within the institutional arena. Thus the theoretical intuition: To explain their consequences, we have to look inside the black box of authoritarian elections. The statistical studies on the liberalizing power of authoritarian elections by Howard and Roessler (2006) and Pop-Eleches and Robertson (2008) are distinctive insofar as they take seriously the causal weight of actor strategies, in addition to contextual factors. Howard and Roessler underline the causal relevance of opposition coordination, opposition protest, and the previous withdrawal of the incumbent chief executive from electoral competition. Pop-Eleches and Robertson highlight the causal relevance of regime manipulation and opposition protest. Both studies are consistent with the idea that authoritarian elections are arenas of unequal struggle between government and opposition which grant governments the tools to win most times (the probable outcome) and opposition actors the opportunity to advance sometimes (the possible outcome).19 In contrast to most large-N research, a growing number of qualitative comparative studies have been treating authoritarian elections not as causal events, but as institutional spaces whose perpetuation or transformation is contingent on internal causal dynamics. While statistical studies have been tracing

The Power of Elections

173

the context-invariant effects of authoritarian elections, qualitative studies have been focusing on their context-dependent effects; and while the former have been examining the external contexts that condition the consequences of authoritarian elections, the latter have been reconstructing the internal configurations of actors and dynamics of conflict that condition their outcomes. To cite just a few examples: Conceiving manipulation as an art form, William Case (2006) explains the fate of electoral authoritarian regimes by the degree of skill with which governments execute their manipulative maneuvers. Nicolas van de Walle (2006) and Joy Langston (2006) study the interdependent coordination problems of government and opposition. The more divided the government and the more united the opposition (with one depending on the other), the more likely it is that authoritarian elections change the correlation of force between the two antagonists. Steven Levitsky and Lucan Way (2010b) identify the organizational infrastructure of the regime as the major domestic determinant of electoral authoritarian regime trajectories after the end of the Cold War. In his explanation of democratic breakthroughs in post-communist countries, Michael McFaul (2005) analyzes the broader correlation of force between civil society and the state. As he infers from examining commonalities among successful electoral transitions, if civil society is strong and able to mobilize and the regime weak and paralyzed, electoral rebellions are likely to happen and likely to be successful. Valerie Bunce and Sharon Wolchik (2011) tell a similar story: If authoritarian rulers turn vulnerable and complacent, while their opponents take seriously the twin challenge of attracting voters and counteracting manipulation, electoral revolutions have a fair chance of succeeding.

C ON C LU S I O N Even if diverse in research purposes and research designs, the emergent comparative literature permits us to draw some tentative conclusions on the general impact that multiparty elections bear on the fate of authoritarian regimes and leaders. If rulers were free to select the institutional configuration that grants them the highest expectations of political life, they would choose to be monarchical heads of dynasties or chairpersons of single parties. For most contemporary rulers, neither of these two are feasible options. Electoral authoritarian rule thus emerges as their third-best alternative. It affords them two-sided protection against democratic competition and military intervention. In a manipulated multiparty environment, autocrats can expect to conserve power longer than under democratic conditions, and to lose it in

174

Analytical Framework

less traumatic ways than under military rule. While improving the odds of both survival in power and peaceful exit from power, electoral authoritarianism does not offer absolute protection. It is not a source of tranquillity, but a strategy of risk management. Multiparty elections do not spell the end of authoritarian struggles for survival. They change their nature. They change the magnitude and nature of the risks autocrats accept, yet do not eliminate these risks, nor the need to contain them through continual vigilance and conflictive political maneuvering. The risks of rulers are the opportunities of their opponents. Electoral authoritarian regimes are structurally dependent on the cooperation of elite members, state agents, and citizens—and thus structurally vulnerable to defection and contestation by these groups of actors. Since multiparty elections afford multiple opportunities to exploit these vulnerabilities, they may give rise to either incremental or sudden processes of democratic change. Political opportunities are not self-realizing, though. They must be seized by political actors. Electoral transitions are not the predictable, law-like outcome of authoritarian multiparty elections. They are the marginal risk that authoritarian rulers assume when calling multiparty elections for the purpose of political (as well as physical) survival. Whether they happen or not is contingent on the dynamics of conflictive interaction between government and opposition in the electoral arena. The relevance of authoritarian elections does not reside in their causal power, but in the causal power they grant to the two-level competition that unfolds in the electoral arena (embedded in the larger arena of societal conflicts). The power of elections derives from the power of electoral politics. Just like other political institutions, elections shape the arena of politics. They do not substitute for politics.

Notes 1. On the more general ambivalence of authoritarian institutions of representation, see Chapter 2, pp. 71–74. On asymmetric games, see Chapter 4, pp. 112–115. 2. Of course, the internal dynamics of authoritarian elections may carry relevant effects below the level of regime change. They may influence, for instance, the design of public policies or the level of civil liberties (see e.g. Lindberg 2006b). 3. In an insightful piece published in the same year as Elections without Choice, Victor Zaslavsky and Robert Brym (1978) analyzed the processes and functions of single-party elections in the ussr from a participants’ perspective. Rather than legitimating single-party dictatorship, they posited, Soviet elections served multiple other purposes for multiple agents. They allowed citizens to express dissent by abstaining (through the request of absentee certificates). They allowed party leaders to identify dissents (who refused to register). They allowed party activists

The Power of Elections

4. 5.

6.

7.

8.

9.

10.

11.

12.

175

to earn credit for “social service.” They allowed state authorities to control population movements (by auditing voter registries) and to improve their capacities of large-scale mobilization. And they allowed the regime to demonstrate, not its legitimacy, but its capability of extracting popular acquiescence to a system of ideological hypocrisy (Zaslavsky and Brym 1978: 367–371). See e.g. Helmke and Levitsky (2004: 726). The Arab rebellions of 2011 point to the limits of such constraints on opposition behavior. The lower the enabling capacity of elections, the lower their constraining capacity. When electoral autocrats restrict electoral competition in excess, they push contention outside the electoral arena. The architectonical metaphor as such is hardly persuasive, in particular for someone residing in a country full of halfway houses, started in times of prosperity and abandoned to their fate and the weeds and the dust of decades after the family coffers turned empty. If one visualizes the silent resilience of semi-finished ruins under the Mexican sun, one cannot but wonder about the possible capacity of authoritarian halfway houses to withstand the teeth of time. See also McFaul (2005) and Bunce and Wolchik (2009). For readings of authoritarian elections through the lenses of social movement studies, see Pop-Eleches and Robertson (2008), Thompson and Kuntz (2004), and Tucker (2007). Since Geddes distinguishes seven pure and mixed types of regimes, her case numbers are small. Some observed differences may not be statistically significant (see Geddes 2005: Table 2). When authors describe authoritarian multiparty regimes as “one-party regimes” they may find themselves asking somewhat paradoxical questions about the number of parties that inhabit these nominal single-party regimes (see e.g. Wright and Escribà-Folch 2009). Note, though, that not all instances of de facto single-party elections within de jure multiparty frameworks arise through authoritarian engineering (as in the fake multipartyism of the German Democratic Republic). Some simply result from the concerted withdrawal of all opposition parties (as in the 2004 elections in Venezuela). Despite their full boycott by opposition parties, such cases should count as multiparty elections. cnts 2008 edition, with data from 1815 through 2006 (). Units of analysis are regime years (N = 15,333). To establish the presence of elections, I used variables S22F5 legislative selection (0 = no legislature, 1 = non-elective legislature, and 2 = elective legislature) (with score 2 indicating legislative elections) and S21F6 executive selection (1 = direct, 2 = indirect election, 3 = non-elective) (with scores 2 and 3 indicating executive elections). To establish the presence of multiple parties, I used variable S19F6 party legitimacy (0 = no parties, or all but dominant party and satellites excluded, 1 = significant exclusion of parties, 2 = exclusion of minor or extremist parties, 3 = no parties excluded) (with scores 1–3 indicating, generously, the admission of multiple parties). A conceptual side note: if the strength of institutions lies in their ability to stabilize expectations and interactions (see Chapter 1), it is not clear why we should, by

176

13. 14. 15.

16.

17.

18.

19.

Analytical Framework definition, attribute higher degrees of “institutionalization” to multiparty systems. In terms of institutional strength and stability, the average single-party regime scores higher on “institutionalization” than the average multiparty autocracy. The authors are emphatic in ascribing the causal effects of multiparty regimes not to elections, but to legislatures (Gandhi and Przeworski 2007: 1291). On the institutional logic of hereditary succession outside the world of presumptively traditional monarchies, see Brownlee (2007b). Cox chooses slightly different regime labels: no-party autocracies, one-party autocracies, multiparty autocracies, and democracies. He does not distinguish between hegemonic and competitive electoral autocracies, nor between horizontal and vertical threats to authoritarian survival. More than one-quarter of all transitions from multiparty autocracy lead to democracy (28.6 percent, N = 63), while less than one-fifth of all transitions from single-party regimes do so (19.3 percent, N = 57). Almost one-third of all transitions from military government lead to “anarchy” (30.4 percent, N = 92), while only a bit over one-sixth of all transitions from multiparty autocracies do so (17.5 percent) (Magaloni and Kricheli 2010: Table 1). The authors’ terminology differs somewhat from mine. They refer to hegemonic electoral authoritarianism as “dominant-party multiparty regimes,” and to competitive electoral authoritarianism as “limited multiparty regimes.” See also Pop-Eleches and Robertson (2008: 10). Kelley offers a similar argument regarding the potential impact of international election monitoring on the quality of elections (2012: Ch. 7). Schneider (2007) offers a critical reassessment of Howard and Roessler’s statistical findings through Qualitative Comparative Analysis (qca).

PA RT I I E M P I R I C A L E X P L O R AT I O N S

6 Comparative Cases and Data In the preceding theoretical chapters, I described authoritarian elections as arenas of asymmetrical conflict that obey their own complex logic. In the subsequent empirical chapters, I will trace their internal dynamics through quantitative comparative analysis. My statistical explorations will center on the meta-game strategies that the two leading actors in the nested game of authoritarian elections pursue: strategies of manipulation on behalf of the government and strategies of protest on behalf of the opposition. Guided by the overarching notion of a competitive struggle over uncertainty, I will study the mutual interaction between these strategies: the degree to which manipulation responds to protest (Chapter 8) and protest to manipulation (Chapter 9). In a previous step, I will examine to what extent these bundles of strategies explain degrees of electoral uncertainty (Chapter 7), and in a subsequent step, to which extent they account for probabilities of regime change (Chapter 10). In each chapter, I combine theory-building and theory-testing. I develop more specific hypotheses on the politics of uncertainty and I examine their empirical plausibility on the basis of original data on authoritarian elections worldwide before and after the end of the Cold War (1980–2002). In the contemporary quantitative study of comparative politics, most of our theoretical explanations focus on actors, yet most of our empirical tests rely on institutional and structural variables. Our theory-building tends to revolve around individual or collective actors, their competition for resources, their preferences, beliefs, expectations, and mutual perceptions, their strategic calculations and choices. Our theory-testing, by contrast, tends to rely on contextual variables like economic wealth, social inequality, natural resources, religion, forms of government, and regime type. Constrained by the limited availability of cross-national data on actors and actor strategies, we tend to stow them away in the black box of unobservable causal mechanisms. To narrow the gap between actor-centered theory and structure-centered empirics, in the empirical part of this book, I center my attention on political actors. I adopt an actor-centric perspective within an election-centric theory. I still have to accept a yawning gap between my expectation-centered theory and my action-centered empirics. Constrained by the unavailability

180

Empirical Explorations

of cross-national data (and the practical impossibility of generating them), I have no choice but to bracket actor perceptions of uncertainty, despite the central role they play in my theoretical framework. Lacking information about their beliefs and expectations, I take recourse to actors’ observable behavior: manipulative strategies by regime actors, contentious strategies by their opponents. The privilege I grant to actor strategies does not oblige me to ignore their institutional and structural contexts. The concept of electoral authoritarianism concedes causal weight and autonomy to the electoral arena. It vindicates the twin notion that political institutions matter and politics, too. At the same time, it does not prevent me from recognizing the embedded nature of electoral authoritarian politics. The nested game of authoritarian elections is nested within its larger societal and political environment. Given the relative complexity of my theoretical framework, its empirical explorations through statistical data analysis carries an inevitable touch of heroism. Aside from black-boxing uncertainty, my theoretical master variable, my analysis suffers from limitations I wish to state from the outset. I have been able to collect data on meta-game strategies by government and opposition, yet not on voter perceptions and preferences at the game level. The coverage of my dataset is limited to little more than two decades and my case numbers are rather low for the purpose of statistical data analysis, above all when I segment my sample by competitive and hegemonic regimes. I employ individual elections as units of analysis, which allows me to estimate short-term effects, but not long-term developments over various electoral cycles. Although I recognize the theoretical possibility of decision-making dilemmas, I examine average patterns of co-variation across cases. Dilemmatic decisions are indeterminate. They push actors into opposite courses of action. Their contradictions wash out in the estimation of average effects. Throughout, I will strive to keep my analysis as simple, intuitive, and accessible as possible, thus sacrificing statistical sophistication in favor of statistical comprehensibility. All in all, my empirical analyses are what I say they are: explorations. Tentative, rather than definitive, they blend deductive and inductive reasoning and feed theory-building as much as theory-testing. Before proceeding to the statistical analyses of Chapters 7 through 10, in this chapter I present the contours of the original data source that feeds the empirical part of the book: the Dataset on Authoritarian Elections (dae). After some introductory reflections on the measurement of complex concepts, I describe my definition and classification rules for electoral authoritarian regimes as well as for the internal distinction between competitive and hegemonic regimes. In continuation, I sketch the sources, definitions, and frequency distributions of my three core groups of variables: regime manipulation, opposition protest, and electoral competitiveness. Appendices

Comparative Cases and Data

181

a–d offer more detailed information on cases, operational definitions, rules of exception, the treatment of borderline cases, sources and coding procedures, internal and external reliability checks, and descriptive statistics.

CLASSIFYING ELECTORAL AUTHORITARIANISM How do we recognize an electoral authoritarian regime when we see one? It is easier to define electoral authoritarian regimes in general terms than to distinguish them case by case for the purpose of cross-national comparison. By definition, clear cases are easy to classify. In hard, borderline cases, however, the normative ambiguities and informational uncertainties of electoral authoritarian regimes conspire against their consensual empirical identification (see Chapter 3). In delimitating the electoral authoritarian universe, I use two criteria: (a) the formal–institutional criterion of multiparty elections at the national level and (b) the practical subversion of electoral institutions through authoritarian manipulation. For the latter purpose of establishing the authoritarian quality of elections I will rely on judgmental data (annual Freedom House political rights scores). This requires some justification.

Measurement and Judgment Standard methodological advice tells us to base our measurement decisions on “observations, rather than judgments” (Przeworski et al. 2000: 55). It presupposes that we partition the process of measurement into two phases. In the first stage, make all the judgments necessary to select and define the empirical phenomena we admit as observational evidence, as well as to devise the coding rules that permit us to assign categories or numbers to cases. In the second stage, by contrast, we ban judgmental elements and limit ourselves to apply our self-made rules of codification in a mechanical fashion. The first phase is deliberative, demanding the intersubjective justification of conceptual and operational decisions; the second one is observational, demanding the transparent collection of information and the quasi-bureaucratic application of rules. The design of general rules and their implementation by impersonal coders allows political measurement to be replicable and reliable. However, the methodological ideal of establishing a functional separation between judgment and observation rests upon a demanding set of assumptions. To be able to ban judgmental elements from the final stages of the measurement process, we need: (a) transparent empirical phenomena whose observation does not

182

Empirical Explorations

depend on our judgmental faculties; (b) full public access to reliable information on these phenomena; and (c) complete, consistent, and determinative rules of measurement that minimize discretion in the assignment of numbers to observations. These conditions are intrinsically problematic in the comparative study of politics (see Schedler 2012b). For all practical purposes, they are impossible to meet in the measurement of complex concepts, that is, in the operationalization of abstract, multidimensional, and aggregate concepts— such as electoral authoritarianism. Electoral autocracies resemble electoral democracies in their formal–institutional framework. They differ from them in the manipulative practices they unleash against these formal representative institutions. Identifying these regimes thus requires detecting and assessing their practices of electoral manipulation. It requires monitoring, aggregating, and weighting manifold violations of abstract democratic norms at multiple sites and multiple stages of the electoral process. Elections are complex processes. Electoral integrity as well as its opposite, electoral manipulation, are accordingly complex concepts. As election experts Jørgen Elklit and Andrew Reynolds write, the “systematic study of election quality” demands the systematic study of eleven “steps” or dimensions of electoral processes: legal framework, electoral management, constituency and polling district demarcation, voter education, voter registration, ballot design, party and candidate nomination and registration, campaign regulation, polling, vote counting and vote tabulation, electoral dispute resolution, and post-election procedures. As the authors explain, to ascertain the democratic quality of these eleven dimensions we have to address around 50 questions or “performance indicators.” Many of them contain further subdimensions and demand the collection of information on the entire voting process, with thousands of locations and millions of actors (see Elklit and Reynolds 2005: Table 1). The complexity of the measurement task at hand is staggering. If we would try to devise coding rules that would allow election observers to assign numbers to all 50 performance indicators in a mechanical fashion, without exercising discretionary decision-making power, we would need at least a dozen precise rules for each indicator. Most likely, we would need many more. If it were just ten rules per indicator, we would need 500 coding rules (!) to assess the democratic quality of an election—and many more to handle informational lacunae and inconsistencies, to incorporate unexpected and unclear cases, and to determine the weight and aggregation of each dimension. A regulatory and administrative nightmare, a recipe for mental as well as methodological insanity. No wonder Elklit and Reynolds conclude that most of their indicators have to be ascertained through “expert panel assessments” (Elklit and Reynolds 2005: Table 1).

Comparative Cases and Data

183

If we wish to measure complex phenomena like the integrity of elections whose empirical observation is not amenable to full bureaucratic regulation we can choose between two alternative paths: We can either strive to identify easily observable symptoms of the phenomena we are interested in. Or we can embrace the complexities of intersubjective judgment to assess them (see also Schedler 2012b). The alternation rule proposed by Adam Przeworski and co-authors represents the first path, the annual data on political rights and civil liberties in the world provided by Freedom House the second one.

The Alternation Rule In their widely (and justly) acclaimed study Democracy and Development, Adam Przeworski and his collaborators identify democratic regimes on the basis of three institutional attributes: (a) executive selection: the head of government is elected in popular elections; (b) legislative selection: the legislature is elected; and (c) party pluralism: there is more than one party (see Przeworski et al. 2000: 28–29). All three attributes of political democracy are present in electoral authoritarian regimes as well. What distinguishes electoral autocracies from electoral democracies are not the formal properties of political elections, but their authoritarian qualities. Electoral authoritarian regimes, just like their democratic counterparts, hold multiparty elections for presidents and legislative assemblies. However, they subject these formally democratic procedures to authoritarian controls so systematic and severe that they deprive them of their democratic substance (see Chapter 3). While formal institutional facts are fairly easy to ascertain, practices of electoral manipulation are much less accessible to public inspection. As outlined in Chapter 4, what we can see in electoral authoritarian regimes are election results, the official distribution of votes and seats among parties and candidates. Rather than expressing the free will of the people, these outcomes represent the hybrid product of authoritarian manipulation and popular preferences. Neither of these is accessible to direct observation. Przeworski and his co-authors therefore hold that passing judgment on the authoritarian quality of elections is an elusive enterprise. Attempts “to assess the degree of repression, intimidation, or fraud . . . cannot be made in a reliable way” (Przeworski et al. 2000: 24). One way of dealing with the endemic problems of imperfect information that characterize authoritarian elections is simply to work with what we have: the factual realm of official election results. Numerous scholars have done so. Knowing that we cannot take official figures as simple expressions of voter preferences, they have treated them as simple proxies for electoral

184

Empirical Explorations

manipulation. The underlying justification for inferring procedural qualities from substantive outcomes is simple. Given the open and conflictive nature of modern democracy, in most places and at most times, low levels of interparty competition speak of low levels of democracy. The less competitive an election is, the less democratic we may assume it to be.1 If incumbents earn qualified majorities of the popular vote in two subsequent elections, Adam Przeworski once asserted, we can safely assume the system is not democratic. “No country in which a party wins 60 percent of the vote twice in a row is a democracy” (1991: 95).With his co-authors in Democracy and Development, he adopted an even tighter standard: the alternation rule. As long as regimes that hold multiparty elections have “not lost an election” at least once, they count as dictatorships (Przeworski et al. 2000: 29). In democracy, alternation in government testifies both to the vigor of interparty competition and to the vitality of the democratic system. The conquest of power by opposition parties is the crowning act of interparty competition, the acceptance of defeat by governing parties the crowning act of democratic consolidation.2 In addition, according to liberal–democratic theory, the threat of electoral alternation in power constitutes the very foundation of democratic accountability and responsiveness (see Sartori 1987 and Powell 2000). Incorporating alternation in power into our operational definition of democracy thus resonates well with democratic norms and theories. Grounded in the attractive conception of democracy as “a system in which parties lose elections” (Przeworski 1991: 10), the “alternation rule” proposed by Adam Przeworski and his co-authors (2000: 23–29) rests upon a twin assumption: (a) democracy requires alternation; and (b) alternation requires democracy. Under this rule, if we see a regime that holds elections it never loses, we should not classify it as democratic. Inversely, if we observe that a regime holds elections and actually loses one, we may deduce that it had been democratic all along (since its last fundamental change of rules).3 By definition, democracy does not demand the actual, but only the possible occurrence of alternation in power. However, until alternation actually happens we can never know for sure whether a ruling party would be willing to hand over power in the case of electoral defeat. We can only make informed guesses. Few autocrats are as honest and explicit as Zimbabwe’s president Robert Mugabe on the eve of the 2000 legislative elections when he thundered to his supporters: “The [leading opposition party] MDC will never form the government of this country, never ever, not in my lifetime or even after I die” (cited in Meredith 2002: 177). The alternation rule makes sense in normative–democratic terms and offers a clear-cut, easily discernible criterion to classify hard cases (except for bogus alternations in which successor governments are dissimulated reincarnations of their predecessors). The rule also avoids the uncertainties that

Comparative Cases and Data

185

come along with counterfactual reasoning and allows the analyst to stick with comparatively simple observables, rather struggling to make sense of myriads of disperse facts. Still, taking election results and, in particular, alternation in office as primary evidence of procedural integrity runs the risk of misclassifying numerous regimes (a risk Przeworski and his co-authors readily acknowledge). Although the causal association of alternation with democracy must be regarded a solid empirical generalization, it does allow for significant exceptions. In authoritarian contexts, alternation in power may happen and its significance is less clear and predetermined than under democratic conditions. Depending on the integrity of the election and the identity of the winner, turnovers under authoritarian (or post-authoritarian) rule may belong to one of three categories: democratic revelations, democratic revolutions, or authoritarian reshuffles. 1. Democratic revelations: post-transitional alternation. In cases in which electoral authoritarian governments oversee a process of democratizing reform, their final defeat in national elections serves to confirm the seriousness of the democratic advances they allowed to happen before. In such instances of post-transitional alternation, the defeat of the ruling parties at the polls fulfills an epistemic function: it reveals the presently democratic nature of a previously authoritarian system. The 2000 presidential victories of opposition candidates Vicente Fox in Mexico, Abdoulaye Wade in Senegal, and Chen Shui-bian in Taiwan were neat examples of such “democratic revelations” (Schedler 2000). Under the alternation rule, we would classify these regimes (correctly) in a retroactive fashion “as democratic for the entire period [the outgoing] party was in power under the same rules” (Przeworski et al. 2000: 24), that is, since the last substantive reform of electoral institutions.4 2. Democratic revolutions: transitional alternation. Even if electoral autocracies are systems in which opposition parties are supposed to lose elections, and actually do lose elections most of the time, alternation in power may occur even under authoritarian conditions. Electoral autocrats expose themselves to some degree of electoral uncertainty, however small. Through institutional design, authoritarian manipulation, and electoral persuasion they may be able to contain electoral uncertainty, but rarely to eliminate it altogether. Even if minuscule, the possibility of suffering defeat in “stunning elections” (Huntington 1991: 175, Thompson and Kuntz 2006) remains alive. In such cases, opposition parties must commonly take to the streets and stage “electoral revolutions” to make their surprise victories stick, as in the Philippines in 1986, Côte d’Ivoire and Serbia in 2000, Georgia in 2003, and Ukraine in 2004. On only rare occasions, as in Olusegun Obasanjo’s

186

Empirical Explorations

victory in the 1999 presidential elections in Nigeria, opposition candidates emerge victorious from fraudulent elections and are able to take office without large-scale conflict. When opposition parties manage to win the uphill race of an authoritarian election against all odds, we should not infer from the failure of authoritarian manipulation that the process was free and fair from the very beginning.5 In “democratic revolutions” (Thompson 2004), alternation in power does not reveal the democratic nature of the regime. It only shows its unpopularity and weakness—its lack of popular support as well as its lack of resources and resolve to ensure its own survival. Instances of democratizing or “transitional” alternation do not serve as indicators, but causal forces, of democratic advance.6 3. Authoritarian reshuffles: alternation without transition. In an exceptional manner, non-democratic elections may lead to changes in government without changing the nature of the authoritarian regime. Such rare instances of authoritarian alternation in power seem to happen most often in situations where the central government exercises only weak control over the electoral arena: (a) in cases where ongoing civil wars undermine the democratic foundations of electoral competition, as in Sri Lanka in 1994, Algeria in 1999, or Colombia in 2002; (b) in cases where military tutelage drains the authority of elected officials, as in Guatemala in 1985, Pakistan in 1990, 1993, and 1997, or Niger in 1999; and (c) in cases of inaugural, yet still authoritarian, elections in the wake of civil war or military rule, as in Turkey in 1983, Georgia in 1992, or Cambodia in 1993. Needless to say, in such governmental reshuffles under continuing authoritarian conditions, elections do not perform democratizing roles.7

Expert Judgments An alternative way of dealing with the opacity of electoral manipulation lies in the use of multiple sources of information in order to reach comprehensive judgment on the authoritarian quality of an election. Rather than staying with the doubtful objectivity of official election results, observers may expand their camp of vision and gather information about the two game levels whose confluence determines election results. By combining their knowledge of official election results with evidence on regime manipulation (at the meta-game level) and popular preferences (at the game level), they may reach broad judgment about the authoritarian quality of the electoral process under scrutiny.

Comparative Cases and Data

187

Even if expert judgments on the democratic quality of elections are often controversial, at least in complex and ambiguous cases, disinterested observers generally do reach convergent, or at least overlapping, assessments that have a fair chance of surviving public interpellations, provided they are well documented and well reasoned.8 As Judith Kelley concludes in her comparative analysis of international election observation: “Although the truth of an election’s quality is by nature elusive . . . most of the time the assessments of international election monitors are fairly uncontroversial, and align well with a version of reality that most other commentators are willing to accept” (2012: 156–157). However, even if well-justified expert judgments on the authoritarian quality of multiparty elections seem feasible in principle, in practice established data providers in comparative politics do not offer such judgmental assessments. Major cross-national classifications of non-democratic regimes are not designed to identify electoral autocracies. They either work with binary categories, dividing the world into democracies and dictatorships (e.g. Przeworski et al. 2000). Or they employ conventional categories of non-democratic systems, such as the distinction between military, personal, and single-party regimes, that do not account for the existence of electoral autocracies (e.g. Geddes 1999 and 2004). To delineate its universe of cases, the literature on electoral authoritarianism has therefore by default been relying on graded measures of democracy that either contain less than full assessments of the democratic quality of national elections, or more. The former is true for the Polity dataset on political regimes and transitions (), the latter for the annual Freedom House reports on civil liberties and political rights (). The widely used Polity dataset “Political Regime Characteristics and Transitions” is of limited value for the task of identifying electoral autocracies. It is centrally concerned with the institutionalization of politics (the “regulation” of power struggles and popular participation) as well as with horizontal limitations on the exercise of power (“executive constraints”). Where it covers aspects relevant to the democratic quality of national elections, it relies on coarse categorical measures (often mistaken as ordinal ones) that blend multiple substantive concerns in methodologically opaque manners. For instance, the Polity measures of “the competitiveness of participation” mix concerns about political repression (the exclusion and harassment of opposition), cleavage structures (local versus national, parochial versus general, overlapping versus crosscutting, ethnic versus secular conflicts), the renunciation of violence (peaceful leadership succession), and the stability of competition (enduring parties) (see Marshall and Jaggers 2002: 25–26). The resulting scores align well at the poles inhabited by liberal democracies and closed autocracies

188

Empirical Explorations

(which most regime datasets have few troubles in identifying). But they are essentially incomprehensible in the intermediate range where authors are free to read them as indicators of hybrid regimes, semi-democracy, institutional inconsistency, democratic transitions, or electoral authoritarianism.9 By comparison, despite inbuilt methodological opacities (see Munck and Verkuilen 2002), Freedom House seems to do much better in evaluating the democratic quality of electoral regimes. In its assessments of political rights, the New Yorker think tank asks more questions than we need, yet still asks the right questions we need, in order to judge the democratic quality of electoral processes. In its survey of political rights, some items (regarding the sovereignty, integrity, and accountability of elected decision-makers) relate to the exercise of power, rather than the access to power. Yet, most concern the procedural integrity of elections: Are the chief executive and the national legislative assembly elected “through free and fair elections”? Do citizens enjoy freedom of association? Are there “fair electoral laws, equal campaign opportunities, fair polling, and honest tabulation of votes”? In addition to electoral procedures, Freedom House considers electoral outcomes as well. Concerned about the intensity of electoral competition, Freedom House experts do more than evaluating electoral results after the fact. They strive to assess the ex ante uncertainty of outcomes that characterizes democratic elections (see Przeworski 1986: 58–61). Is the political system, they ask, “open to the rise and fall of . . . competing parties”? Do we observe “a significant opposition vote, de facto opposition power, and a realistic possibility for the opposition to increase its support or gain power through elections”?10 Freedom House formulates its normative and empirical questions at a fairly high level of abstraction and it bundles concerns about elections (the access to power) with concerns about governance (the exercise of power). Still, its judgments on the quality of electoral processes seem reasonable. Above all, its assessments of whether “citizens are able to change their government through regular elections” (which always introduce its qualitative evaluations of political rights in individual country reports) look fundamentally sound and defensible in the light of available evidence and democratic norms.11 Of course, as all graded measures of democracy, Freedom House data fit uneasily with the notion that political regimes form “bounded wholes” (Sartori 1987: 184) that are set apart by qualitative differences. Their measurement scales do not offer obvious cutting points for drawing regime boundaries. Since it is unclear what specific scores and differences between scores are meant to mean, any effort to translate these ordinal numbers into qualitative regime categories is liable to charges of arbitrariness. Scholars have nevertheless been establishing conventional cut-off points to categorize

Comparative Cases and Data

189

regimes. For instance, an average Freedom House score of 2.5 is usually considered to be indicative of a liberal–democratic regime (just as a score of 6 or higher on the Polity iv Democracy index is commonly accepted as a measure of stable democracy).12 Such scholarly conventions do not necessarily, or even recognizably, rest on substantive criteria. Yet, to the extent that the underlying continuous measures make sense (and thus fulfill the criterion of validity), mapping them onto qualitative categories is a normative choice as much as a methodological problem. By setting the threshold that separates democratic from non-democratic regimes we choose the level of normative demands we make on those regimes we accept to describe as democratic. Since we treat differences in degree as if they were differences in kind, we are bound to misclassify at least some cases, regardless of where we draw our precise qualitative boundary. The question is what kind of error we accept to accept when trying to identify non-democratic regimes: to label regimes as authoritarian that are not (false positives or type i errors) or to classify regimes as democratic that are not (false negatives or type ii errors). “Err we must; the question is in which way” (Przeworski et al. 2000: 23). By adopting the alternation rule, Przeworski and his co-authors resolved to tolerate false positives. By adopting a rather high cut-off point along the Freedom House scale of political rights, I strove to avoid them.

THE UNIVERSE OF CASES My Dataset on Authoritarian Elections (dae) pretends to cover (almost) the full set of authoritarian multiparty elections worldwide from 1980 through 2002.13 Within that time-span, it actually takes a census, rather than a sample. I delimitated the universe of cases in two steps. At first, I used Freedom House data to define a broad pool of intermediate authoritarian regimes as candidates for inclusion (among countries with population > 1 million). In continuation, I introduced additional information to weed out all countries that lacked the defining formal–institutional features of electoral authoritarianism and did not fulfill minimal conditions of internal and external sovereignty. x Authoritarian boundaries. Sandwiched between electoral democracy and closed regimes, electoral authoritarian regimes populate an intermediate territory on the continuum of regimes. Though not democratic, they do not occupy the pole of unrestrained dictatorial regimes either. By opening up the electoral arena to multiparty competition, they have to open at least minimal breathing spaces to dissenting citizens and politicians. In order

190

Empirical Explorations

to demarcate the intermediate terrain of semi-liberties in which electoral autocracies flourish, I chose all countries that received Freedom House political rights scores between 4 and 6 during at least four consecutive years between 1980 and 2002 (a criterion of minimum stability intended to filter out short-term fluctuations of countries into the relevant range of Freedom House scores). Regimes with political rights scores of 3 or lower are considered to be too democratic to deserve inclusion, regimes with political rights scores of 7 too dictatorial. Equipped with the razor blade of Freedom House scores, I applied this simple selection rule with bureaucratic zeal, granting only minor exceptions (see Appendix a). x Multiparty elections. While my initial delimitation of candidate countries relied on the criterion of political freedom, in a second step I employed institutional criteria, with a narrow focus on the electoral arena, to demarcate my universe of cases. Freedom House scores alone do not tell us anything about the institutional infrastructure of political regimes. Among the countries located at the intermediate levels of political freedom according to Freedom House, only those qualify as electoral authoritarian that held at least one full set of multiparty elections with universal suffrage for the chief executive as well as the national legislature.14 This bundle of institutional requirements excludes non-party regimes, de jure as well as de facto one-party regimes, military regimes, competitive oligarchies, traditional monarchies, as well as regimes that limit electoral competition either to the legislative or the subnational arena. The criterion of “full sets” of elections also excludes incomplete election cycles in which a single presidential or legislative election is followed by regime breakdown (through military coup, executive coup, armed insurrection, or external occupation). x State sovereignty. As the idea of authoritarian regimes presupposes the presence of governments endowed with a minimum of external and internal sovereignty, my dataset also excludes non-sovereign protectorates, like Lebanon and Bosnia-Herzegovina, as well as cases of state collapse that fail to maintain political structures we could recognize as political regimes, like Liberia between 1990 and 1995 and Somalia in the 1990s. x Entry and exit. My units of analysis are authoritarian elections. The dataset covers national elections cycles conducted without interruption under conditions of electoral authoritarianism in the time period from 1980 to 2002. The delimitation of the research period derives from a mix of theoretical and pragmatic considerations. Authoritarian elections have mushroomed after the fall of the Berlin Wall. Since they are not exclusive to the post-Cold War period, however, I strove to include Cold War elections as well, moving as far back in time as available data permitted. At its end point, the dataset goes as far as my research resources allowed me to go. Although the last

Comparative Cases and Data

191

elections covered date from the year 2002, for the purpose of analyzing regime trajectories the dataset includes information on the posterior fate of the electoral authoritarian regime (within the subsequent electoral cycle). Regimes enter the dataset with their first set of national authoritarian elections held after 1979. Six electoral authoritarian regimes included in the dataset were founded before 1980: Mexico (1929), Paraguay (1954), Indonesia (1968), Malaysia (1957), Philippines (1972), and Singapore (1965). At its endpoint, the dataset includes the last national election held in an electoral authoritarian regime before 2003. Regimes exit the dataset earlier if they go through a democratic transition, defined by a sustained improvement of Freedom House political rights scores (over at least four consecutive years), or if they suffer an extra-constitutional interruption of their electoral cycle by military coup, foreign intervention, insurgency, or presidential self-perpetuation in power. Table A.1 in Appendix a contains a more detailed account of general selection rules, specific examples and exceptions, and the treatment of borderline cases. The resulting sample contains 194 authoritarian elections that were held in 51 countries. Most of these elections were held after the end of the Cold War. Only 38 took place before 1990 (19.6 percent), the remaining 156 afterwards (80.4 percent). Table A.2 in the appendix provides the full list of election cycles included in the dataset. It gives the number of legislative and presidential elections as well the years of the first and last elections included in each cycle. In addition, it indicates the nature of the predecessor and successor regimes and gives some hints at the mode of transition to and from each electoral authoritarian regime. Table 6.1 shows the distribution of the corresponding 121 legislative and 73 presidential elections across world regions. Reflecting the subcontinent’s turn from one-party rule to electoral authoritarianism in the last decade of the 20th century, sub-Saharan Africa concentrates the largest share of authoritarian election cycles (37.1 percent). Due to the peace transitions in Central America and the demise of hegemonic party rule in Mexico and Paraguay, Latin America and the Caribbean come only second, contributing close to one-fifth of cases (20.1 percent). All other world regions hosted around one-tenth of authoritarian elections each between 1980 and 2002.

CLASSIFYING HEGEMONIC REGIMES Within the broad family of electoral authoritarian regimes, hegemonic regimes succeed in establishing a stable set of interlocking perceptions and

192

Empirical Explorations

Table 6.1. Elections in electoral authoritarian regimes, by world region, 1980–2002 World region

Legislative elections

Presidential elections

Total

%

All electoral autocracies 1 Latin America & Caribbean 2 Eastern Europe 3 Central Asia & Caucasus 4 Middle East & Northern Africa 5 Sub-Saharan Africa 6 South & East Asia Total

21 13 12 12 41 22 121

18 7 10 4 31 3 73

39 20 22 16 72 25 194

20.1 10.3 11.3 8.2 37.1 12.9 100.0

Competitive regimes 1 Latin America & Caribbean 2 Eastern Europe 3 Central Asia & Caucasus 4 Northern Africa & Middle East 5 Sub-Saharan Africa 6 South & East Asia Total

16 12 12 6 20 7 73

14 7 10 3 14 3 51

30 19 22 9 34 10 124

24.2 15.3 17.7 7.2 27.4 8.1 100.0

Hegemonic regimes 1 Latin America & Caribbean 2 Eastern Europe 3 Central Asia & Caucasus 4 Northern Africa & Middle East 5 Sub-Saharan Africa 6 South & East Asia Total

5 1 0 6 21 15 48

4 0 0 1 17 0 22

9 1 0 7 38 15 70

12.8 0.1 0 10.0 54.3 21.4 100.0

Source: Dataset on Authoritarian Elections.

expectations: the ruling party always wins, and is expected to, for it is popular and powerful, and perceived as such. In competitive regimes, the nested game of authoritarian elections is more open. Both the power and the popularity of the ruling party are more uncertain and contested, and the outcomes of elections less predictable. If the conceptual distinction between hegemonic and competitive regimes is grounded in expectations (degrees of institutional uncertainty), their empirical classification should be as well. Yet, since cross-national data on regime uncertainty are unavailable, I rely on two observational proxies to distinguish hegemonic from competitive regimes: a minimum duration of ten years (since the assumption of power by the ruling

Comparative Cases and Data

193

coalition) and the continuous control of legislative supermajorities (with the ruling party holding at least two-thirds of seats in the Lower House). 1. Duration: The criterion of duration relates to the institutionalized nature of hegemonic party regimes. Hegemonic parties are not shooting stars that illuminate the party system during fleeting electoral episodes. Founded at the end of civil war, the achievement of national independence, or the imposition of military rule, they rule their countries over long stretches of time. Their original strength feeds their longevity and their longevity feeds their continual strength. Institutional duration solidifies institutional expectations. It breeds self-confidence by incumbents and resignation by their opponents. Their capacity to hold on to power for long periods, either avoiding or weathering serious challenges, makes hegemonic parties look invincible, even inevitable. 2. Constitutional majorities: The criterion of continuous legislative supermajorities derives from the notion that hegemonic regimes strive to assemble heterogeneous “oversized coalitions.” Their rather inclusive and overpowering alliance structure permits them to be invincible, and appear invincible, in the electoral arena. In the arena of constitution making, it allows them to control the basic rules of the political game and to manipulate them at their convenience (see Magaloni 2006: 15). To capture the logic of hegemonic party rule, both criteria seem essential. Some authors have been relying on electoral results alone to distinguish competitive from hegemonic regimes. The precise threshold varies. Howard and Roessler set it at 70 percent of the popular vote in either legislative or presidential elections (2006), Diamond at 75 percent of presidential votes (2002: 32). Symbolically, in terms of popular perceptions, presidential results may be more important. Institutionally, in terms of policy-making, legislative seat shares seem more relevant. Yet, legislative or presidential landslides won by the incumbent (and occasionally even by opposition figures like Georgia’s Mikhail Saakashvili who won the snap presidential contest called in the wake of the Rose Revolution in January 1994 with 96 percent of the vote) are not always indicative of hegemonic equilibria, but may arise from contingent political junctures. At times they are the simple result of opposition boycotts. The criterion of regime duration prevents us from misclassifying such transient supermajorities as hegemonic rule. Among the electoral autocracies I classify as competitive, I register fourteen legislative elections in which incumbent parties earn constitutional majorities. In ten of them, opposition parties had either fully or partially boycotted the contest (71.4 percent).15 Table A.1 in the appendix explains these classification rules as well as minor rules of exceptions in greater detail. Table A.3 contains the resulting list of

194

Empirical Explorations

countries that were ruled by hegemonic parties at some point during the period under study (1980–2002). It also indicates the name of the ruling party, its personal leadership (if applicable), the origins of hegemony, and the year and mode of termination (as by the time of writing). Of the 194 elections in the dataset, a good third were conducted in hegemonic regimes (70 elections or 36.1 percent). Of these, more than two-thirds (68.5 percent) were conducted after the end of the Cold War. Even if several longstanding hegemonic regimes predate the dataset, hegemonic party regimes should not be considered a Cold War phenomenon. Some ruling parties, in particular in sub-Saharan Africa, have managed to survive the transition from single-party rule to electoral authoritarianism by converting themselves (seamlessly) into hegemonic parties. Table 6.1 displays the distribution of competitive and hegemonic elections by world regions. Latin America and sub-Saharan Africa concentrate the largest numbers of competitive authoritarian elections (about one-quarter each), while the Middle East and Northern Africa and South and East Asia host the lowest numbers (less than 10 percent each). In the world of hegemonic elections, sub-Saharan Africa stands out among all other regions. It hosted more than half of all hegemonic contests in the dataset (54.3 percent). A notable one-fifth of hegemonic-party elections (21.4 percent) took place in South and East Asia, still an intriguing “storehouse of historical and contemporary electoral authoritarianism” (Case 2006: 95). Due to the recent independence of post-Soviet countries, until the early 2000s hegemonic elections were practically unknown in Eastern Europe, Central Asia, and the Caucasus.

MEASURING MANIPULATION Existing cross-national datasets do not contain reasonably proximate measures for most strategies within the broad repertoire of authoritarian manipulation. Some authoritarian strategies pose methodological problems to cross-national measurement that have to be regarded insurmountable for all practical purposes. This seems true in particular for the intimidation of voters by state and party agents, the unfair distribution of campaign resources and media attention, acts of vote buying, and the informal disenfranchisement of voters. Other measures of electoral control are more amenable to quantification. In the present study, I include rough indicators of four authoritarian core strategies. I tap existing data sources to measure human rights violations and media restrictions. In addition, I gathered my own data from various news sources and election reports to capture electoral fraud and the exclusion of opposition actors. In the following, I will introduce each of these four measures and present their frequency distributions.

Comparative Cases and Data

195

Appendix b contains summary descriptions of all variables (including a host of control variables). Appendix c details sources, coding procedures, coding rules, as well as internal and external measures of reliability for my original measures in the dataset (on authoritarian manipulation as well as on opposition protest). Appendix d offers descriptive statistics for all variables by regime subtype.16

Human Rights Violations To capture the intensity of political repression in electoral authoritarian regimes, I use the annual Physical Integrity Rights Index constructed by the Cingranelli–Richards (ciri) Human Rights Data Project (). The ciri index measures the intensity of governmental violations of human rights on a 9-point scale (0–8) for 195 countries since 1981. Its primary sources are annual reports by Amnesty International and the us State Department. The index is additive. It sums up ordinal measures on four types of repressive state action: extrajudicial killings, disappearances, torture, and political imprisonment. Each of the disaggregate measures registers the absolute frequency of human rights violations (in a given country, in a given year) on a three-point scale. When zero violations occur, countries are credited with full respect for human rights (score 2). When 50 or more instances of violation are reported, they obtain the worst score (score 0). In the broad band in between, in the range from 1 to 49 recorded violations, they receive intermediate grades (score 1).17 Figure 6.1 displays the frequency distribution of our cases (election years) by regime subtype along the inverted ciri index (where higher scores indicate more frequent violations of physical integrity rights). Unsurprisingly, most cluster around the middle of the scale. Less than one-tenth (8.3 percent) earn the most benign scores (0–1), less than one-fifth (17.3 percent) the most repressive ones (7–8). Hegemonic regimes are significantly less repressive than competitive regimes, though. Four-fifths of hegemonic elections (80.4 percent), yet only half as many competitive contests (43.6 percent), are located on the less repressive left-hand side of the scale (scores 0–4).18

Media Restrictions To measure limitations on media freedom, a core dimension of governmental violations of civic liberties, I calculate the arithmetic mean of annual Freedom House Press Freedom scores () and the inverted Freedom of Speech and Press scores from the Cingranelli–Richards

196

Empirical Explorations

human rights dataset (both rescaled to zero). The result is an ordinal 0–2 scale, in which the lowest score indicates the absence of media restrictions, the highest their systematic presence.19 As Figure 6.1 shows, limitations of press freedom distribute similarly as human rights violations. While almost no authoritarian elections take place in 40.0%

50.0%

Competitive regimes Hegemonic regimes

40.0%

30.0%

30.0% 20.0% 42.1%

20.0%

40.4%

34.0% 28.1%

10.0%

10.0%

18.1%

19.3%

10.5% 6.4%

0.0% 0

1

2

3

4

5

6

7

8

0.0%

0.0

0.5

1.0

1.5

2.0

Violations of media freedom

Violations of physical integrity

60.0%

50.0%

50.0%

40.0%

40.0% 30.0% 50.0%

30.0%

54.2%

43.1%

20.0%

45.2%

37.3%

20.0% 28.8% 29.2%

26.0%

10.0%

16.7%

0.0%

10.0%

19.6%

27.3%

22.7%

0.0% No fraud

Irregularities

Fraud

No fraud

Irregularities

Legislative fraud

60.0%

60.0%

40.0%

40.0% 68.2%

61.6%

20.0%

Fraud

Presidential fraud

60.8%

58.3% 41.7%

38.4%

0.0%

39.2%

20.0%

31.8%

0.0% Inclusion

Exclusion

Legislative exclusion

Inclusion

Exclusion

Presidential exclusion

Figure 6.1. Frequency distributions of regime manipulation

Comparative Cases and Data

197

a context of unambiguously free media, typically they do not involve the complete suppression of media freedom either. A free press (scores 0 or 0.5) exists in less than one-tenth of our cases (8.6 percent), while full media restrictions (score 2) are in place in less than one-fifth (18.5 percent). By far most elections in our dataset (72.9 percent) took place at intermediate levels of media restrictions (scores 1 or 1.5). According to these data, hegemonic and competitive regimes do not differ systematically in their treatment of mass media.

Electoral Fraud Given its opaque and contested nature, vote rigging is “difficult for social scientists to study.” As it involves “clandestine efforts to shape election results . . . [n]o one who stuffs the ballot box wants to leave a trail of incriminating evidence” (Lehoucq 2003: 233). Whatever evidence exists is likely to be biased. Given the partisan nature of electoral disputes, “the most abundant sources for studying fraud—from first-hand accounts to partisan denunciations of electoral shenanigans—are not ‘objective’” (Lehoucq 2003: 233– 234).20 Notwithstanding such methodological difficulties, over the past years a growing number of comparative scholars have taken up the challenge of constructing cross-national datasets that try to capture the incidence of fraud in national elections. Their data differ from mine in coverage, conceptions, and operational definitions of election fraud. Eventually, the scholarly community would benefit from coordinating its overlapping and competing ventures of cross-national data development (see Schedler 2012a). In the meanwhile, I have made use of emergent data on fraud (as well as on exclusion and opposition protest) to conduct rough external reliability checks of my own measures (which I report at the end of Appendix c). For the purpose of cross-electoral measurement, I defined election fraud in a narrow manner as the manipulation of electoral administration for partisan advantage at any stage of the electoral process (before, during, and after election day). The ordinal measure I constructed is trichotomous. The category of fraud indicates partisan interference with the organization of election that is either widespread and systematic or, if less intense, serious enough to affect outcomes (score 2). The intermediate category of irregularities denotes deviations from formal rules that are neither systematic and widespread nor serious enough to alter results (score 1). The final category of electoral integrity (“no fraud”) pertains to situations in which no more than minor irregularities are reported (score 0). Like most similar measurement efforts, I built my dataset through content analysis of international news sources and election observer reports (for more detailed coding rules, sources, and coding procedures, see Appendix c).

198

Empirical Explorations

Again, the case frequencies displayed in Figure 6.1 show a convex distribution. Roughly speaking, about one-half of authoritarian elections fall into the (somewhat fuzzy) intermediate category of irregularities and about one-quarter into each of its adjacent categories of integrity and fraud. Presidential elections register a somewhat higher incidence of fraud than legislative contests (a rough 10 percent gap) and competitive elections a higher incidence than hegemonic elections (again, by around 10 percentage points).

Exclusion To measure the exclusion of opposition actors from competition at national elections I constructed a dichotomous variable from the same information sources as in the measurement of fraud (see Appendix c). I classified elections as exclusionary if state agencies (belonging to any branch of government) prevented active opposition parties or candidates from participating, be it through constitutional provisions, legal rules, executive decrees, judicial rulings, or administrative decisions. Examples are the denial of registry to opposition parties on political or administrative grounds, or the enactment of ad hominem rules of candidacy that bar prominent opposition actors from electoral competition. Note that my threshold for coding elections as exclusionary is rather low. I require excluded parties or candidates to be active at the time of the election, but do not apply any criterion of size or relevance. I thus ignore the exclusion of latent actors, but incorporate the exclusion of minor ones. A presidential election counts as exclusionary if state authorities disqualify any presidential candidate, regardless of his or her importance. In legislative elections, my criterion is slightly more demanding. The widespread exclusion of “uncomfortable” individual candidates as well as the wholesale exclusion of opposition parties qualify as exclusion, while the disqualification of individual candidates in a few districts does not. Appendix c provides more specific coding rules and procedures as well as indicators of internal and external reliability. Within the universe of legislative authoritarian elections, cases of inclusion (53.7 percent) and exclusion (46.3 percent) hold a rough balance. Among presidential elections, the distribution leans towards the inclusionary side. In less than two-fifths of presidential contests (37 percent), rulers strive to contain electoral uncertainty by excluding opposition candidates. As Figure 6.1 shows, hegemonic regimes hold more exclusionary legislative elections than competitive regimes.21

Comparative Cases and Data

199

Patterns of Manipulation Given the authoritarian toolbox they have at their disposal, what are the configurations of tools that regimes chose to keep elections under control? Two questions arise. One is quantitative: How many strategies do rulers deploy at a time? The other one is qualitative: What strategies do they deploy at any given time? Manipulative intensities. In general terms, rulers face the choice between two broad strategic alternatives: the mutually reinforcing combination of authoritarian strategies (cumulative control) and the mutually substitutive selection of strategies (selective control). The universe of contemporary electoral authoritarian regimes contains numerous exemplifications of both possibilities. Egypt under Mubarak exemplified the repressive rigor of comprehensive strategies of control, including arbitrary arrests, party bans, voter intimidation, and ballot rigging, all of them embedded in decades of coercive emergency rule. By contrast, Singapore under the pas illustrates the silent efficacy of selective strategies of control, in which tight restrictions upon speech and association, a lame parliament, and discriminatory electoral rules coexist with respect for due process and electoral integrity. Mapping electoral authoritarian regimes would be easy if they would cluster at the poles of cumulative and selective control. If some would choose to overrun the opposition with all means at their disposition, while others would strive to neutralize it through surgical violations of democratic norms, we could sort them neatly into two discrete boxes: harsh regimes that behave as cumulative maximizers of electoral control versus light regimes that act as selective optimizers of electoral control. In theory, all things being equal, we would expect utilitarian rulers to lean towards selective strategies of control. Rather than doing everything to keep opposition parties from winning elections, they should only do what is necessary to prevent defeat. They should strive to control the electoral game in subtle and skillful ways, by deploying selective rather than comprehensive strategies of control. To capture the empirical degree of strategic selectivity electoral authoritarian regimes pursue, I constructed an additive index of “cumulative manipulation.” The index normalizes my four measures of manipulation to a scale from 0 to 1 and sums them up for each election (counting exclusion and fraud in the legislative and presidential arenas only half in the case of concurrent elections). It yields a theoretical minimum value of zero (no recourse to any of the four strategies) and a maximum value of four (full recourse to all four). The frequency distribution in Figure 6.2 shows that our cases practically cover the entire spectrum, reaching from very low levels of manipulation (minimum = 0.38) to almost full manipulation

200

Empirical Explorations 14

Number of cases

12 10 8 6 4 2 0 3.88 3.75 3.63 3.50 3.38 3.25 3.13 3.00 2.88 2.75 2.63 2.50 2.38 2.25 2.13 2.00 1.88 1.75 1.63 1.50 1.38 1.25 1.13 1.00 0.88 0.75 0.38 Cumulative index of manipulation

Figure 6.2. Frequency distribution of cumulative manipulation

(maximum = 3.88). Their distribution is close to a bell-shaped normal distribution around the mean of 2.17 (SD = 0.73, N = 148). While the distinction between “light” (selective) and “heavy” (cumulative) electoral authoritarian regimes appears to make empirical sense, most regimes occupy intermediate positions. Manipulative configurations. Should we expect to find any systematic pattern in manipulative choices? Do rulers face compelling reasons to either combine certain strategies or to renounce some when implementing others? Perhaps not. Physical repression may either reinforce or replace any other strategy. Exclusion and fraud should work as close substitutes: to the extent that regimes bar opposition actors from electoral competition, they need not tamper with electoral results. Only media controls seem to be the ideal company for any other strategy. To the extent that violations of media freedoms go unnoticed, they may ensure that other violations of democratic norms go unnoticed as well. Within our sample, most bivariate correlation coefficients between the four strategies of authoritarian control are close to zero and statistically non-significant. Predictably, manipulative maneuvers in concurrent presidential and legislative elections correlate highly, yet otherwise inter-item coherence is low. The only exception are exclusionary strategies in the legislative arena that show systematic negative associations with repression (r = –.21, p d .05) and positive ones with media controls (r = .35, p d .001). Exclusion

Comparative Cases and Data

201

and repression, it seems, work as mutual substitutes, while exclusion and censorship reinforce each other. The general weakness of lineal associations between authoritarian strategies implies that regimes enjoy considerable margins of discretion at the moment of selecting and combining their strategies. There is no “inner logic” of authoritarian strategies that pushes them towards certain configurations of control. Rather than a coherent set of strategies that hang together in law-like fashion, authoritarian strategies seem to form an ensemble of options that rulers are picking from with few immanent constraints. The metaphor of a “menu” of manipulation seems quite appropriate after all to describe the structure of choice electoral authoritarian rulers face.22

M E A S U R I N G P ROT E S T To reconstruct cycles of protest, students of contentious politics tend to employ content analyses of domestic newspapers and periodicals, national as well as local ones. In most contexts, local media sources can be safely assumed to be “superior to regional and global ones in terms of extent of coverage and richness of detail” (Aditjondro, Kowalewski, and Peterson 2000: 121). They will often be politically imbalanced, but selecting several outlets with contrasting editorial tendencies is likely to cancel out particular biases (see Ball 2005). Under authoritarian conditions, the coverage of domestic media will be more limited and their reporting more systematically biased than under democratic conditions. Nevertheless, even under military rule and open censorship, national media may provide reasonable coverage of contentious action (see Stein 2007: 25).

Vertical and Horizontal Threats Despite its weaknesses, content analysis of domestic news sources is still the methodological gold standard in the quantitative study of contentious action. Unfortunately, it is a feasible method only for comparative studies that cover few countries. Large-N studies need to rely on global news sources or (so-called) international newspapers (at times complemented by non-periodical sources of specialized information, like election reports). Accordingly, the only global longitudinal dataset on contentious politics available today, the Arthur Banks Cross-National Time–Series (cnts) “conflict event data,” constructs its event counts on the basis of daily international news reported in the New York Times.

202

Empirical Explorations

It covers various forms of political contention that bear no necessary relation to electoral conflicts: political assassinations, general strikes, guerrilla warfare, riots, revolutions, and anti-government demonstrations. Although cnts conflict data register only a tiny fraction of contentious actions reported in local news sources (as Ronald Francisco documented for selected countries and periods),23 I will be using them as a supplementary source on contentious action in electoral authoritarian regimes. For most purposes, I will aggregate cnts event counts into three additive indicators. My indicator of “vertical threats” sums up three instances of mass protest: anti-government demonstrations, general strikes, and riots. My category of “armed rebellion” includes two expressions of organized political violence: guerrilla movements and revolutions. My measure of “horizontal threats” adds cnts counts of government crises, purges, and military coups. Figure 6.3 displays the frequency distributions of these three additive indicators for the three years preceding each election. It confirms the commonsensical expectation that hegemonic regimes are more tranquil places than competitive regimes. Only about one-third of competitive elections (34.8 percent), but more than half of hegemonic elections (50.9 percent) take place in the complete absence of previous vertical threats. In somewhat less than half of competitive elections (48.3 percent), yet more than three-quarters of hegemonic elections (77.2 percent), cnts registers no instances of armed political violence in antecedent years. More than half of competitive contests (53.9 percent), and almost all hegemonic elections (91.2 percent), are free of preceding horizontal threats.24

Electoral Protest To obtain information on election-related contention, I collected more specific data on electoral protest. In accordance with the temporal structure of electoral processes, I distinguished pre-electoral from post-electoral protest. In addition, I created a separate measure for election boycott, the refusal of opposition parties or candidates to participate in an electoral process.25 1. Opposition boycott. My ordinal measure of opposition boycott focuses on the behavior of “major” opposition parties, ignoring minor players as well as regime-loyal creatures. It captures three possibilities. The category of participation indicates that all major opposition parties participate (score 0). When some, but not all, major opposition parties refrain from participating in the election, I register a partial boycott (score 1). A full boycott requires that all major opposition parties stay away from the polls (score 2).

Comparative Cases and Data 60.0%

203

Competitive regimes Hegemonic regimes

Percentage of cases

50.0% 40.0% 30.0% 20.0% 10.0% 0.0%

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 22 26 27 36 CNTS Vertical threats (protest demonstrations + strikes + riots)

80.0%

Percentage of cases

60.0%

40.0%

20.0%

0.0%

0 1 2 3 4 5 6 7 13 14 CNTS Armed rebellion (guerrilla movements + revolutions)

100.0%

Percentage of cases

80.0%

60.0%

40.0%

20.0%

0.0% 0 1 2 3 4 5 CNTS Horizontal threats (government crises + purges + coups)

Figure 6.3. Frequency distributions of contentious events (during three years before election year)

204

Empirical Explorations

2. Pre-electoral protest. As my sources do not allow for gathering more precise information across all cases, I measure pre-electoral opposition protest as a coarse dichotomous variable. Opposition parties and candidates may criticize any of the steps leading up to the election (before election day). They may voice their discontentment with the legal framework, election authorities, or administrative decisions. Yet I code their behavior as acquiescence if they refrain from mobilizing their followers in protest against an upcoming election (score 0). I register active protest if they take recourse to the “repertoire of contention” (Charles Tilly) typical of modern social movements, such as marches, street blockades, strikes, and sit-ins (score 1). The category of active protest mobilization includes opposition calls on voters to boycott elections. It also captures legal appeals that opposition parties submit against biased electoral institutions, electoral practices, or electoral authorities. 3. Post-electoral protest. Conceptually, my measure of post-electoral contention is similar to my measure of pre-electoral protest, only a bit finer, as it includes an intermediate category. Acquiescence implies either explicit or tacit acceptance of defeat by losing parties or candidates (score 0). It includes cases in which losers complain (only) about non-decisive irregularities. Rejection involves the claim that results were falsified and thus failed to reflect the will of the electorate (score 0.5). It may take the form of rhetorical rejection (public complaints that elections were undemocratic, that manipulation was decisive, that irregularities invalidated results); judicial recourse (demands for annulment, appeals to domestic or international courts); or symbolic protest (such as boycott of the presidential inauguration or inaugural session of the parliament). If opposition parties boycotted an election, I assume them to reject its results, even if I have no further reports on post-electoral manifestations of discontent. Active protest captures the collective mobilization of followers in protest against election results (score 1). It may involve public demonstrations, civic resistance, occupation of public buildings, street blockades, a boycott of the legislative assembly, or even the spontaneous outbreak of violence. Figure 6.4 shows the frequency distributions of electoral protest by regime subtype across legislative and presidential elections. Before election day, in both electoral arenas the modal choice of opposition parties is acquiescence. It is only in 29.8 percent of legislative elections and 43.8 percent of presidential elections that opposition actors step beyond verbal criticism and actively protest the preparation of an election, be it on the streets, in the courts, or in the legislature. In the face of hegemonic parties, opposition actors resort even less frequently to contentious collective action. Their quiescent anticipation is

Comparative Cases and Data 80.0%

Competitive regimes Hegemonic regimes

60.0%

205

80.0%

60.0%

40.0%

40.0%

77.1%

72.7%

65.8%

51.0%

49.0%

20.0%

20.0%

34.2%

27.3%

22.9%

0.0% Acquiescence

Active protest

0.0%

Acquiescence

Active protest

Presidential pre-electoral protest

Legislative pre-electoral protest

80.0% 60.0% 60.0% 40.0% 68.5% 68.8%

40.0%

20.0%

70.6%

40.9%

20.0%

31.8%

27.3%

20.5% 20.8% 11.0% 10.4%

0.0%

17.6%

11.8%

0.0% Participation

Partial boycott

Full boycott

Participation

Legislative boycott

Partial boycott

Full boycott

Presidential boycott

40.0%

50.0% 40.0%

30.0%

30.0% 20.0%

36.99% 37.5%

50.0% 35.62%

33.33% 29.17%

27.4%

41.2% 40.9%

20.0% 31.4%

27.5%

10.0%

10.0% 9.1%

0.0% Acquiescence

Rejection

Active protest

0.0%

Acquiescence

Legislative post-electoral protest

Rejection

Active protest

Presidential post-electoral protest

Figure 6.4. Frequency distributions of electoral protest

most pronounced in the presidential arena. In hegemonic regimes, opposition parties actively mobilize against upcoming presidential elections only half as often (27.3 percent) as in competitive regimes (51 percent).26 The figures also indicate that opposition parties do not embark lightly on what constitutes their pre-electoral protest strategy of last resort: electoral boycott. All major opposition parties participate in more than two-thirds of

206

Empirical Explorations

legislative elections (68.6 percent) and more than three-fifths of presidential contests (61.6 percent). Still, in almost one-third of legislative contests (31.4 percent) they do abstain from participating, though mostly in a fractured way. Full boycotts take place in only 10.7 percent of legislative elections. In the presidential arena, boycotts under conditions of opposition unity are twice as frequent (21.9 percent). Participation rates in competitive and hegemonic regimes are practically identical in legislative elections, yet diverge again in the presidential arena. The rate of presidential boycotts in hegemonic contexts (59.1 percent) doubles the boycott rate in competitive contexts (29.4 percent).27 On election day and during the often tense and conflictive days and weeks to follow, it is only in a minority of authoritarian elections that opposition parties accept official results without strong public criticism. In the legislative arena, opposition responses distribute in a roughly even way between acquiescence (37.2 percent), rejection (33.1 percent), and active protest (29.8 percent). In the presidential arena, responses by opposition parties are skewed towards protest: in only about one-fifth of cases they acquiesce to the result (21.9 percent), while in about two-fifths they either reject it rhetorically (37.0 percent) or mobilize active protest (41.1 percent). Although the rate of post-electoral opposition acquiescence to presidential election results is lower in hegemonic regimes than in competitive ones (9.1 against 27.5 percent), levels of post-electoral protest do not differ significantly between the two regime subtypes (see Figure 6.4).

M EA S U R I N G C OM PE T IT I V EN ESS The concept of electoral competitiveness is grounded in intersubjective perceptions. Its defining feature is not the closeness of results, but their uncertainty (see Przeworski 1986). In non-competitive elections, substantive outcomes are certain: the ruling party always wins, and everybody knows it will. In competitive elections, results are uncertain: the ruling party may lose, but no one knows for sure. Uncertainty, however, is an elusive phenomenon, volatile and controversial (see Schedler 1998a). To estimate the substantive uncertainty citizens and elites perceive ahead of a given election, we would have to reconstruct their public declarations and private perceptions through field work and survey research, which is as yet unfeasible for the purpose of large-N cross-national comparison. In the absence of systematic information on actors’ intersubjective perceptions, we have to rely on “objective” electoral data to gauge levels of com-

Comparative Cases and Data

207

petitiveness. At each round in the game of authoritarian elections, election authorities pay off parties and candidates in the currency of votes and seats. The official results they pronounce freeze formal correlations of power for constitutionally defined time-spans. They provide objectively observable and politically relevant indicators of interparty competitiveness. Nevertheless, official election outcomes are only pale reflections of pre-electoral perceptions of electoral uncertainty. Election results may fail to reflect pre-electoral perceptions of competitiveness for five simple reasons: (a) Given the opacity of authoritarian rule, actors may miscalculate patterns of popular support. (b) To mold public perceptions, they may publicly misrepresent their estimates of voter support. (c) Even if prevailing perceptions of voter preferences are accurate, voters may change their minds. (d) Even if popular preferences remain stable, citizens may alter their behavior, in particular their turnout rates. (e) Even if everybody is correct and honest and nothing changes, electoral authorities may falsify election results. Although authoritarian election figures distort “real” levels of electoral competitiveness, it is more instructive, one may contend, to look into a stained mirror than to stare at a blank wall. Comparative election studies work with a broad variety of quantitative measures that translate election results into indicators of interparty competitiveness. From among the broad variety of available options, I will employ two measures of electoral competitiveness: x Margins of victory: Legislative margins of victory are the difference in seat shares between the largest and the second-largest party, presidential margins of victory the difference in valid vote shares between the winning candidate and the up-runner (in the first round of presidential elections in ballotage systems). The closeness of electoral results (margins of victory) will serve as the principal measure of electoral competitiveness. Margins of victory are simple and intuitive and they seem relevant to political actors themselves (who may, or may not, spend their time computing more arcane indicators of competitiveness, such as the effective numbers of parties and candidates). Margins of victory also offer a straightforward lineal measure of competitiveness: the larger they are, the less competitive is the party system. I calculate margins of victory in a manner that takes into account the occurrence of alternation of power. When incumbent parties or candidates lose an election, governmental margins of victory turn into margins of defeat. Since I use the closeness of elections as a measure of incumbent strength, “margins of victory” show negative signs in such instances of incumbent defeat (for a listing of cases, see Appendix b).

208

Empirical Explorations

x Opposition fragmentation: Legislative opposition fragmentation is the degree of dispersion of legislative seats among opposition parties, presidential opposition fragmentation the degree of dispersion of opposition votes in presidential elections (Rae index). Rae indices of legislative opposition fractionalization indicate the probability that any two random opposition deputies belong to different parties. In situations of opposition unity, the probability is zero, in situations of complete individualization of legislative dissidence, it is one. In an analogous fashion, presidential opposition fractionalization indicates the probability that two random opposition voters support the same opposition candidate. If the opposition fields a single candidate, the probability is zero; if every opposition voter runs for president (and votes for herself), it is one.28 Margins of victory correlate highly with alternative measures of party-system concentration, such as party-systemic fragmentation (Rae index) and the effective number of parties and candidates (Laakso–Taagepera index). By contrast, against common assumptions about the crucial role of opposition coordination in debilitating authoritarian incumbents, the closeness of outcomes is only weakly related to the fragmentation of the opposition camp.29 The occasional occurrence of incumbent defeat depresses average winning margins in competitive regimes, which lie at 27.5 percent of legislative seats and 31.1 percent of presidential votes. It also creates broad variance (ranging from –47.5 to 90.5 percent of legislative seats and –63 to 98 percent of presidential votes). In hegemonic regimes, in which incumbents by definition have never lost an election, mean margins are substantially higher, reaching 70.9 percent in legislative and 61.3 percent in presidential elections (see Appendix d). In both subtypes of regimes, the legislative fragmentation of opposition parties ranges from complete unity (Rae index = 0) to almost perfect disorganization (Rae index = 0.99). In presidential elections, the maximum dispersion of opposition parties is somewhat lower (Rae index = 0.88 in competitive and 0.77 in hegemonic regimes). Quite notably (considering the reductive effect of electoral rules), opposition parties are on average more fragmented in the legislative arena (Rae index = 0.69 in competitive and 0.52 in hegemonic regimes) than they are in presidential elections (Rae index = 0.46 in competitive and 0.39 in hegemonic regimes).30

C O N C LU S I O N Taking authoritarian elections seriously is a leitmotiv of my theoretical framework. It is the guiding idea of my empirical explorations as well. It

Comparative Cases and Data

209

involves taking seriously the intrinsic relevance of actor choices and conflictive interactions that unfold within the electoral arena. It also means taking seriously the causal weight these choices and interactions carry inside the nested game of authoritarian elections and in the determination of regime outcomes. The explanatory thrust of the following empirical chapters reflects the intrinsic relevance I attribute to the inner dynamics of authoritarian elections. Each chapter takes one central element of the electoral authoritarian two-level game as its dependent variable: election outcomes (Chapter 7), regime manipulation (Chapter 8), and opposition protest (Chapter 9). It is only in a last step that I move from the explanation of within-regime dynamics to the explanation of regime change (Chapter 10). My empirical explorations also reflect the causal assumption that these game dynamics matter. Although I study authoritarian elections as embedded within their structural and institutional environments, the main causal weight of my explanatory enterprise rests upon the shoulders of actor strategies within the authoritarian electoral arena. In essence, my explanations are endogenous and interactive. I strive to explain the internal dynamics of electoral authoritarian regimes through the internal dynamics of electoral authoritarian regimes. And their trajectories too.

Notes 1. For a critical review, see Bogaards (2007). 2. See, for example, Anderson et al. (2005), Huntington (1991: 266–267), Przeworski (2003), Rapoport and Weinberg (2001), Whitehead (2007). 3. See note 4 below. 4. “Whenever a ruling party eventually suffered an electoral defeat and allowed the opposition to assume office, the regime is classified as democratic for the entire period this party was in power under the same rules”(Przeworski et al. 2000: 24). The sameness of rules is not always obvious. Which rules count and which count as identical? When do reforms count as profound, when as superficial? For instance, in the decade before the 2000 alternation in power in Mexico, the country had undergone four waves of electoral reform at the national level (see Schedler 2005). Which serves best to reclassify the regime as democratic in retrospect? Formal rules often change in minor or inconsistent ways that do not bring fundamental change. How do we know when things stay “essentially” the same even if they changed “at the surface”? 5. Naturally, if we apply the “alternation rule” to cases of democratizing or transitional alternation, we misclassify the electoral autocracies these instances of alternation bring to an end as having been democratic all along. For critiques of the alternation rule, see also Brownlee (2008) and Tanneberg (2012).

210

Empirical Explorations

6. Citing Leonardo Morlino, Valerie Bunce asks: “Are elections [causally] important in themselves, or are they merely visible and efficient summaries of more long-term and complex developments on the ground?” (2008: 2). With respect to alternation in power, the response is split: electoral revelations are mirrors of preceding democratic progress, electoral revolutions motors of democratic advance. 7. Chapter 7 presents empirical patterns in authoritarian alternations in power. 8. Schedler (2012b) offers some guidelines for the “rationalization” of judgmental data development. 9. For example, Mansfield and Snyder (2005) picture regimes with intermediate Polity scores as “emerging democracies.” Regan and Henderson (2002) treat them as “semi-democracies.” Gates et al. (2006) talk of “inconsistent regimes.” Roessler and Howard (2009) use Polity in conjunction with Freedom House data to identify electoral authoritarian regimes. For critiques of Polity data, see Bowman, Lehoucq, and Mahoney (2005), Gleditsch and Ward (1997), and Munck and Verkuilen (2002). 10. The quotes are from the “Political Rights and Civil Liberties Checklist” in the methodological appendix to the 2003 Survey (Karatnycky, Piano, and Puddington 2003: 697). 11. These judgments are not without problems, of course. In particular in the past, they were not. In its early years, Freedom House ratings were conspicuously lenient towards right-wing military dictatorships in Latin America. Later on, the think tank has been continually adjusting its methodology without ever updating and recoding its historic data. Less visibly and less explicitly, Freedom House has been shifting its normative criteria, too. Apparently, throughout the 1990s, it has been tightening its standards and elevating its threshold of acceptability of political elections. 12. Matthijs Bogaards (2010) offers a critical overview over the manifold manners in which scholars have measured not political regimes, but processes of regime change, through the use of ordinal-scaled Polity and Freedom House data. 13. The Dataset on Authoritarian Elections (dae) is available in pasw/spss at the cide Data Archive for Applied Research in the Social Sciences (). All data presented in the tables and figures of this and the subsequent chapters are based on the dae. 14. I demand direct popular elections for the chief executive in presidential regimes. In case of parliamentary regimes, I demand direct elections for the legislature that selects and maintains the government. Borderline cases included in the dataset are Egypt, Ethiopia, Indonesia, Pakistan, and Turkey. 15. For further details and rules of exception, see Appendix a. 16. Appendix b also lists all control variables I introduce in subsequent chapters and Appendix d shows their descriptive statistics. Statistical controls include country characteristics (world region, size, population density), social structure (urbanization, literacy, income inequality, ethnic and religious polarization), socio-economic modernity (gdp per capita, communication infrastructure), economic structure (oil exports), economic performance (growth and inflation),

Comparative Cases and Data

17.

18. 19.

20.

21. 22.

23. 24.

25.

26.

27. 28.

211

political institutions (system of government, electoral rules, subnational elections, legislative effectiveness, personalism), state capacity (societal war, non-tax income), and external linkages (trade, foreign direct investment, multilateral debt, and official development assistance). See also Cingranelli and Richards (1999). The unequal breadth of the three categories is likely to involve some measurement bias: high sensibility towards occasional violations and low sensibility towards extreme violations of human rights. In addition, the (defensible) practice of recording absolute frequencies of human rights violations seems to contain an in-built bias against large countries. In my dataset, the ciri index correlates positively and significantly with population size (r = .38, p < .000, N = 149). For a critical review of existing datasets on political rights and civil liberties, see Skaaning (2006). Differences in mean levels of repression between competitive and hegemonic regimes are statistically significant (bilateral t-tests, p = .000). In their original 3-point scales, Freedom House distinguishes between a free, partly free, and not free press, while ciri observes whether there are no, some, or complete government restrictions on the freedom of press, speech, or expression. On the methodological challenges of studying electoral fraud, see also Alvarez, Hall, and Hyde (2008a), Hartlyn and McCoy (2006), Kelley (2012), and Lehoucq (2003), and Lehoucq and Molina (2002). Differences in mean levels of legislative exclusion between competitive and hegemonic regimes are statistically significant (bilateral t-test, p = .031). Similarly, Judith Kelley (2012: Ch. 5) concludes that the choice of authoritarian strategies is less structured than common assumptions of manipulative sophistication would make us believe. She finds that various types of manipulation are positively correlated, though, rather than unrelated. Under the scrutiny of international monitors, governments do not switch from open to veiled manipulation. Instead, “bad things tend to go together . . . safer and riskier forms of irregularities are complements, not substitutes” (Kelley 2012: 86, emphasis removed). See Francisco (2006: Table 4). See also Schedler (2012a: 247–248). Differences in mean levels of vertical threats (p = .028), armed rebellion (p =.000), and horizontal threats (p = .000) between competitive and hegemonic regimes are statistically significant (bilateral t-tests). I proceeded as before, in gathering my data on regime manipulation. I used the same sources, implemented the same coding procedures, and employed the same validity and reliability checks (for details, see Appendix c, for descriptive statistics, Appendix d). Differences in mean levels of pre-electoral protest between competitive and hegemonic regimes are statistically significant in the presidential arena (bilateral t-test, p = .062) and “weakly” significant in the legislative arena (p = .185). Differences in mean levels of presidential boycott between competitive and hegemonic regimes are statistically significant (bilateral t-test, p = .037). For sources and formal definitions of both measures, see Appendix b. Appendix d provides descriptive statistics by regime subtype.

212

Empirical Explorations

29. Margins of victory and indices of fractionalization covary both in the legislative arena (r = –.786, p = .000, N = 119) and in presidential elections (r = –.822, p = .000, N = 73). Victory margins also correlate highly with effective numbers of presidential candidates (r = –.718, p = .000, N = 73), yet less so with effective numbers of legislative parties (r = –.268, p = .003, N = 119). Correlations with opposition fractionalization are low and non-significant, in the legislative arena (r = –.045, p = .636, N = 113) as well as in the presidential arena (r = –.177, p = .143, N = 70). 30. See Appendix d. Differences between legislative and presidential opposition fragmentation are statistically significant (bilateral t-tests, p = .003).

7 The Struggle over Electoral Uncertainty All authoritarian regimes have to cope with the twin problem of uncertainty: with the opacity of their support among subjects and allies and with the insecurity of their support. In electoral authoritarian regimes, the core uncertainty, the mother of all uncertainties, is the uncertainty of electoral outcomes. Hegemonic regimes are in equilibrium. They are built upon certainties of electoral victory by the ruling party. Incumbents strive to preserve these certainties, dissidents to shake and break them. Competitive regimes are in disequilibrium. They contain disquieting uncertainties about the electoral strength of the ruling party. Incumbents work to dissolve these uncertainties, dissidents to deepen them. In this chapter, I will explore the effectiveness of their strategies. More concretely, I will trace the short-term impact that authoritarian manipulation and opposition protest bear on three objective measures of interparty competitiveness: margins of electoral victory, the control of legislative supermajorities by the incumbent, and party alternation in power. My initial hypotheses are somewhat paradoxical: Hegemonic regimes extract high levels of compliance from citizens as well as elites through moderate levels of manipulation. Their low-enforcement equilibria rest upon the generalized perception that they are popular, united, and strong. Open manifestations of dissidence by either citizens or elites disturb these mutually supportive beliefs. Manipulation, too, introduces doubts about the popularity of the regime, yet also reinforces its reputation of ruthlessness. Thus, I expect dissidence to be more disturbing to hegemonic regimes and manipulation more comforting.

T H E R E D U C T I O N O F C OM P LEX I T Y This chapter strives to explain variations in the uncertainty of authoritarian electoral outcomes (as indicated by officially acknowledged levels of electoral competitiveness). Its central question concerns the electoral effectiveness of

214

Empirical Explorations

authoritarian manipulation. To what extent, it asks, do manipulative strategies succeed in keeping opposition parties weak and thus levels of electoral competitiveness low? For this is what electoral manipulation is supposed to be good for. “[D]ictators are dictators because they cannot win competitive elections, because their preferences diverge from those of the majority of the population” (Gandhi and Przeworski 2007: 1281). They are supposed to recur to authoritarian strategies in order to win elections they could not win otherwise. Or at least to manufacture the margins of victory they could not obtain otherwise. The very point of electoral manipulation is to contain the uncertainty of electoral outcomes by decoupling electoral results from popular support. The whole enterprise (as well as the whole concept) of electoral authoritarianism rests upon the premise that incumbents subvert the conditions of freedom and fairness of elections in a manner that affects the outcomes of these elections. In this sense, the main research question of this chapter is x-centered. It focuses on the effects of a determinate explanatory variable x (electoral manipulation), rather than the causes of a determinate dependent variable y (electoral competitiveness). However, any serious x-centered analysis needs to take into account alternative explanations of y. In the present explanation of electoral competitiveness, governmental manipulation will compete against its closest theoretical (as well as political) contender, namely, opposition protest.1 In addition, I will explore the impact that institutional and structural environments leave on levels of electoral competition. Statistical data analysis is a technique of complexity reduction. Its search for common patterns among large numbers of cases always rests upon simplifying assumptions. Before embarking on the explanatory exploration of electoral uncertainty, I wish to highlight three initial simplifications I embrace in the following analysis: my focus on meta-game, rather than game strategies; my focus on short-term, rather than long-term effects; and my vulnerability to problems of endogeneity. The black box. When rulers establish authoritarian multiparty elections, they open up an asymmetric struggle over votes at two levels. At the game level of electoral competition, government and opposition actors struggle over citizen preferences, at the meta-game level of institutional change, they struggle over the basic conditions of electoral competition. Election results are the joint outcome of their competitive interactions at both levels. They are more than simple aggregations of voter choices. They are the joint result of decisions taken by the non-sovereign people and the semi-sovereign state (see Chapter 4). Unfortunately, for the purpose of cross-national research, citizen preferences are locked away in a black box impenetrable to systematic observation. The same is true for the communicative efforts that political actors expend to

Struggle over Electoral Uncertainty

215

influence voter decisions. Cross-national data on either voter preferences or election campaigns in electoral autocracies simply do not exist. Even worse, their collection appears unviable at the present stage of cross-national data development (see Schedler 2012a). Thus forced to bracket the base level of electoral competition, I focus my attention on the upper level of the nested game of authoritarian elections: the meta-game strategies pursued by regime and opposition actors. In essence, in this as well as in the subsequent chapters, my empirical inquiries probe the extent to which we can understand the internal dynamics of electoral authoritarian regimes without observing the game-level dynamics of interparty competition and voter choice. Shallow impact. To trace the impact of manipulation and protest on electoral outcomes, I will look at lineal short-term associations between these meta-game strategies and levels of electoral competitiveness. That is, I will examine their “shallow impact” in the short run, rather than their “deep impact” over the long run. In a simple world of authoritarian sovereignty over the electoral arena, in which dictators manufacture vote tallies at will, the short-term relation between manipulation and competitiveness should be straightforward. Holding everything else constant (the usual crucial counterfactual caveat), high levels of electoral manipulation should yield low levels of electoral competition, lower levels of manipulation should permit more intense competition. Opposition challenges to the grand manipulator should erode the effectiveness of manipulation, their absence should enhance it. We should not, however, place excessive expectations on encountering lineal short-term associations between meta-game strategies and electoral competitiveness. Even though such associations are plausible, they are not compelling. Authoritarian strategies as well as opposition choices may influence electoral uncertainty in manifold ways. Above all, the consequences of their decisions may be long term, rather than immediate. As a matter of fact, theories of “democratization by elections” do not expect one-shot effects of isolated authoritarian elections, but cumulative effects of iterated elections (see Lindberg 2006b, 2009b, and 2009d, Schedler 2002a and 2006a). The elephant. Quietly and melancholically trampling the porcelain room of statistical analysis is the ubiquitous elephant of endogeneity—the problem of reverse causality. In principle, as my explanatory variables are temporarily prior to final election results, I mitigate problems of endogeneity. In the world of social interaction, though, temporal priority provides limited insurance against endogeneity. Human intelligence is forward-looking and their dense “local knowledge” (Geertz 1983) of societal structures and political junctures may enable actors on the ground to anticipate voting trends well ahead of election day. Thus, even if we uncover positive short-term associations between meta-game strategies and electoral certainty, we cannot be certain about the

216

Empirical Explorations

direction of causality. Levels of competitiveness may be the consequence of actor choices as well as the other way round: rational, forward-looking actors may devise their strategies in anticipation of electoral outcomes. Backward-looking causal interpretations compete against forward-looking instrumental readings. The problem with instrumental interpretations of authoritarianism is their indeterminacy. They always make sense. Whatever authoritarian rulers do seems rational. When they feel safe, they can pre-emptively resort to either manipulation or democratizing reform. When they feel insecure, they can defensively do the same, either manipulate or reform. Causal interpretations, by contrast, are puzzling when high manipulation is followed by high competitiveness. Authoritarian strategies that debilitate the ruling party are counterproductive and the notion that rulers adopt measures that on average frustrate their own objectives runs against standard presumptions of actor rationality. Yet, positive short-term associations between manipulation and competitiveness lend credit to the idea that two-level games of authoritarian elections are systemic sources of dilemmas and mistakes (see Chapter 4).

THE VULNERABILITY OF HEGEMONY The ostentatious purpose of manipulation lies in the control of electoral outcomes. It is meant to contain the substantive uncertainty of multiparty elections. In this chapter, I examine the short-term effectiveness of manipulation in keeping the lid on the simmering pot of electoral competitiveness. But I also examine the short-term effectiveness of opposition protest in heating it. I expect these strategies to vary in their effectiveness across regime subtypes. Somewhat paradoxically, I expect both governmental and opposition strategies to be more effective in hegemonic contexts. Even though opposition actors do not enjoy direct access to the levers of state power, they need not, and commonly do not, stand on the sidelines and watch passively how governments subject electoral institutions and processes to their control. Through acts of vigilance, protest, and resistance, by detecting, denouncing, and obstructing manipulative maneuvers, they may well subvert, or even revert, the electoral impact of manipulation. They may monitor the confection of voter rolls, document and denounce violations of the electoral law, instruct voters to accept bribes from the ruling party and still vote against it, observe the polling process, conduct quick counts, and so forth. Although their strategic goals are often less immediate and transcend individual electoral junctures, opposition actors may well be able to influence the short-term

Struggle over Electoral Uncertainty

217

effectiveness of authoritarian manipulation. In addition to their capacities of institutional subversion, contentious actions contain the possibility of informative subversion. As I wish to argue, vertical threats from below (as well as horizontal conflicts within the political elite) tend to be particularly irritating to hegemonic belief systems.

Interlocking Expectations Hegemonic regimes are in equilibrium. On average, they display solid and consistent combinations of ruling party dominance, popular acquiescence, and elite cohesion at moderate levels of repression. Four basic facts of hegemonic party rule, which we have already run across in the previous chapter, reflect the basic comforts of hegemonic equilibria: x Electoral dominance: On average, hegemonic regimes enjoy wide and stable margins of victory, while competitive regimes show wider oscillations around more narrow margins. The average margin of victory lies at 66.9 percent in hegemonic elections (sd = 23.1) and at 29.6 percent in competitive elections (sd = 36.6).2 x Vertical tranquillity: In addition to low levels of electoral uncertainty, hegemonic equilibria involve low levels of vertical uncertainty. During the three years preceding the election year, hegemonic regimes on average witness less than half as many anti-government demonstrations, general strikes, political assassinations, guerrilla wars, and revolutionary movements than competitive regimes. x Horizontal tranquillity: Just like manifest threats emanating from society, public challenges that arise from within the state are much rarer in hegemonic regimes than in competitive regimes. Hegemonic parties experience almost no horizontal threats: no coups, no purges, and almost no government crises. Overall, competitive regimes are not only more competitive but also more contentious. x Manipulative restraint: While highly unequal in their overall levels of competitiveness and contention, the two regime subtypes do not differ much in their average intensity of manipulation. Competitive regimes are somewhat more repressive and slightly more fraudulent, while hegemonic regimes are a little more exclusionary. More powerful, more stable, more tranquil, yet no more repressive than competitive regimes, hegemonic party systems appear as more efficient manipulators. At intermediate levels of manipulation, they elicit high levels of societal cooperation and elite cohesion. In comparison to their competitive

218

Empirical Explorations

counterparts, they get a much higher “vote value” out of each “unit of manipulation” they expend. What explains this pattern of high compliance at low levels of coercive enforcement? Do hegemonic parties maintain elite unity and dominate electoral competition because they are more popular than their contenders? Is their institutional equilibrium rooted in popular consent? If this were the case, we could not know it. Given the intrinsic opacities of authoritarian rule, we cannot know how popular hegemonic regimes are. We can only know how popular they appear to be. When we observe actors complying with coercive institutions, we tend to attribute their compliance to coercion. When we observe them complying with liberal institutions, we tend to describe their compliance as voluntary. Hegemonic regimes mix coercive and liberal institutions. Their proven ability to extract long-term compliance from their citizens as well as their elites tends to generate the twin belief that they rest on both voluntary and involuntary compliance, that they are both popular (in the first instance) and coercive (in the last). The tranquil seas of hegemonic party rule thus appear as what economists describe as a “self-enforcing equilibrium” which no participant has an interest to upset. Self-reinforcing beliefs engender self-reinforcing expectations. As I proposed in Chapter 1, any institutional equilibrium rests upon the cognitive foundations of interlocking expectations. And so does the equilibrium of hegemonic party rule: Everybody expects everybody else to expect that the regime is strong and cohesive, capable of weathering political challenges, because it is popular. Potential challengers from within are resigned to the apparent fact that they cannot obtain significant popular support outside the ruling party. Potential challengers from below are resigned to the apparent fact they cannot obtain significant elite support within the party. Together they are resigned to the apparent fact that, thanks to its popularity, the regime shall answer serious challenges, should they arise against expectations, with decisive repression, without fissures, without hesitation.

Paralyzing Resignation Resignation has a simple mathematical expression: low subjective probabilities of change. This is the great pedagogical success of hegemonic regimes: They persuade all actors that the chances of change are infinitesimal. They create images of invincibility. The regime has been here for long and it is here to stay, full point. The behavioral implications for utility-sensitive individuals are straightforward. It is not discontentment, but hope, that pushes dissidents

Struggle over Electoral Uncertainty

219

from loyalty to defection. Even if people value regime change, they should not do anything about it as long as it looks close to impossible.3 The expected utility of an outcome approximates zero as long as its probability approximates zero. Classical decision theory can serve to illustrate the primacy of expectations over evaluations in explaining challenges to hegemonic regimes. Decision theory postulates actors who maximize expected utility. The expected utility μ of a decision outcome is the product of its utility (benefits minus costs) times the probability of its occurrence. In electoral autocracies, the expected utility of opposition victory μ(opp) is the product of its utility u(opp) times its probability of occurrence p(opp):

μ (opp) = u (opp) * p (opp)

(7.1)

Utilities as well as probabilities assume values between 0 and 1. As the values of incumbent and opposition victory mirror each other, the expected utility of authoritarian regime continuity μ(reg) is the product of its utility u(reg) = 1–u(opp) times its probability p(reg) = 1–p(opp): μ (reg ) ⎡⎣1 − u ( opp)⎤⎦ * ⎡⎣1 p ( opp)⎤⎦

(7.2)

Utility maximizing actors, be they citizens or state agents, should defect from the regime when they expected the utility of opposition victory μ(opp) to be larger than the utility of the hegemonic status quo μ(reg). This condition holds when the utility of opposition victory is larger than the probability of regime continuity: μ ( pp) > μ ( g ) if

(

pp) > p (reg )

(7.3)

For instance, if individuals who are fed up with a hegemonic regime, yet continue to think it’s too strong to be defeated, calculate the utility of opposition victory to lie at 0.9 and the chances of opposition victory at 0.1, they end up indifferent between regime change and continuity, as the expected utilities of both outcomes are identical: μ ( pp)

09 * 01

μ (regg ) = (1

9) * (1

1) = 0.1 * 0.9

(7.4)

Under these simple assumptions, hegemonic regimes can hypothetically confront an entire population of disaffected individuals with u(opp) ≤ 0.9 who

220

Empirical Explorations

will nevertheless keep supporting the ruling party as long as their subjective probabilities of regime change are lower than the critical threshold of p(opp) ≤ 0.1. They do not need the approval of their subjects, only their sense of impotence. Thus the importance of expectation management. Hegemonic parties can afford to alienate the people, but not to inspire hope. As Adam Przeworski wrote decades ago, “[w]hat matters for the stability of any regime is not [its] legitimacy … but the presence or absence of preferable alternatives” (1986: 51–52). Hegemonic regimes are masters in foreclosing the emergence of feasible alternatives.

Irritating Signals Mutually reinforcing expectations of popular support, elite cohesion, and manipulative muscle lend hegemonic parties great security. But they also render them vulnerable. If fissures appear at just a single point in the virtuous circle of congruent expectations, the whole edifice of beliefs in the invulnerability of the regime may start crumbling. Hegemony is like love: eternal while it lasts. But when do doubts about its eternity emerge? Which are the irritating events that may produce fissures on the surface of hegemonic equilibria? More specifically, when does their firm grip over the electoral arena start to loosen? I wish to propose three hypotheses: 1. The informative value of vertical challenges: Silence grants consent. Those who fail to raise their voice against the status quo invite the assumption that they are happy with it. This applies to any political context and it should apply to hegemonic regimes as well. If hegemonic equilibria are anchored in generalized assumptions of regime popularity, these assumptions are likely to be anchored in generalized popular acquiescence. A visible surface of electoral support and public quiescence invites observers to attribute deep support to hegemonic regimes. By contrast, open manifestations of popular discontentment against hegemonic regimes invite observers to doubt the solidity of their popular support. 2. The informative value of horizontal challenges: Open conflicts within the hegemonic elite upset rulers’ lateral certainties. They introduce doubts about the internal cohesion of the regime. But they also irritate their vertical certainties. They introduce doubts about popular loyalties, since political elites are unlikely to defect unless they anticipate receiving substantial support by citizens.

Struggle over Electoral Uncertainty

221

3. The informative ambiguity of manipulation: High or increasing levels of manipulation convey contrasting pieces of information. They, too, introduce doubts about the popularity of the regime, as they signal self-doubts that the government harbors about its own popularity. But they also convey messages of authoritarian capacity and resolve. By stepping up their manipulative efforts, hegemonic regimes signal their willingness to contain emergent institutional uncertainties by authoritarian means. In hegemonic regimes, I expect the latter message to overshadow the former one. Given their history of authoritarian effectiveness, hegemonic regimes should be capable of making manipulation work. In contrast to their hegemonic siblings, competitive regimes are not built upon solid reputational foundations. They are more contested and thus more familiar with contentious challenges. They are less coherent and thus more familiar with the periodic reshuffling of elite alliances. They are less stable and their control of state institutions tends to be less secure. Yet their relative weaknesses, I propose, involve paradoxical strengths. Since their reputation is weak, they are relatively immune to reputational damages. Unable to project credible images of popularity, coherence, and strength, they are less vulnerable to events that put their popularity, coherence, and strength into doubt. When hegemonic expectations erode, behavioral patterns are likely to follow suit. Emergent doubts about vertical and horizontal loyalties towards a hegemonic party may encourage voters as well as state agents to abandon the leaking ship. Continuing convictions about the manipulative strength of the regime may encourage them to stay on board. Accordingly, I expect authoritarian electoral competitiveness to be sensitive to contentious actions, horizontal troubles, and manipulative maneuvers—in hegemonic regimes more so than in competitive ones: H1 Vertical challenges are more informative in hegemonic regimes than in competitive regimes. I therefore expect them to be more successful in strengthening the opposition in the electoral arena. H2 Horizontal challenges are more informative in hegemonic regimes than in competitive regimes. I therefore expect them to be more consequential in weakening the incumbent in the electoral arena. H3 Manipulation conveys mixed signals. It indicates weakness (the erosion of voluntary compliance) as well as strength (the mobilization of coercive compliance). In hegemonic regimes, I expect the message of strength to dominate the message of weakness. I expect manipulation to contain electoral competitiveness more successfully than in competitive regimes.

222

Empirical Explorations EXPLAINING MARGINS OF VICTORY

Margins of victory represent plausible measures of interparty competitiveness and thus electoral uncertainty. Authoritarian incumbents may not always be able to maximize margins, only to optimize them. Sometimes, they manufacture smaller margins in order to render electoral outcomes more credible. Instead of “winning big” (Simpser 2004), they content themselves with less than overwhelming victories. Still, it is plausible to assume that more is better. The wider their margins, the safer their grip on power. The inverse holds for opposition parties. Suffering narrow defeat may be annoying to them, if the closeness of the result lends undeserved credibility to authoritarian incumbents and their artful micromanagement of fraud. In general, however, they should prefer suffering defeat by small margins to “losing big” against overwhelming incumbents. We should expect actors to deploy their instruments of power in order to shape electoral outcomes in accordance with these broad preferences. We should also expect the impact of their strategic choices to vary by regime subtype.

The Weapons of the Strong Now, what do our data tell us about the short-term association between manipulation and competitiveness? As the multivariate ols regression results at the bottom of Table 7.1 indicate, the calibration of only four authoritarian strategies (repression, censorship, exclusion, and fraud) explains almost one-third of variation in hegemonic margins of victory (R2 = .31).4 Apparently, the explanatory force of manipulation derives primarily from the weight of exclusion. Exclusion alone explains more than one-fifth of variance in hegemonic margins of victory (R2 = .22). By banning adversaries from the electoral arena, hegemonic parties upholster their margins of victory by more than 20 percentage points. Inclusionary hegemonic regimes enjoy average winning margins of 55.1 percent, exclusionary ones of 76.9 percent. Repression and fraud, too, covary with hegemonic margins of victory. Their coefficient signs are negative, though, which invites two alternative interpretations. We may read them as indications of defensive manipulation: Hegemonic parties tighten their authoritarian controls when they expect competitive pressures to intensify. Or we may read them as symptoms of counterproductive manipulation: Voters desert hegemonic parties when these turn to most severe violations of democratic norms. If they move from full tolerance to full

Struggle over Electoral Uncertainty

223

Table 7.1. Effects of manipulative strategies on electoral competitiveness Margins of victory hr

Margins of victory cr

Constitutional majority cr

Party alternation in power cr

B

p

B

p

eb

p

eb

p

Repression Constant se / pac R2 / c&s R2 N

–3.00 76.45 22.42 .060 55

.068 .000

.06 28.74 36.87 .000 90

.972 .004

.91 .36 80.3% .005 71

.547 .156

1.13 .12 81.3% .009 91

.358 .004

Censorship Constant se / pac R2 / c&s R2 N

8.41 55.01 22.84 .024 55

.250 .000

10.79 13.75 36.52 .014 89

.259 .314

1.85 .09 81.4% .009 70

.427 .041

.69 .38 81.1% .003 90

.583 .302

Exclusion Constant se / pac R2 / c&s R2 N

21.85 55.06 20.56 .225 56

.000 .000

6.72 26.47 36.69 .008 93

.384 .000

.85 .26 80.8% .001 73

.790 .003

1.53 .18 80.9% .007 94

.437 .000

Fraud Constant se / pac R2 / c&s R2 N

–9.54 75.49 22.29 .089 56

.024 .000

8.36 20.28 36.32 .028 93

.105 .003

1.64 .13 80.8% .020 73

.233 .001

.36 .58 80.9% .072 94

.012 .195

Manipulation Repression Censorship Exclusion Fraud Constant se / pac R2 / c&s R2 N

–2.32 –3.29 22.01 –6.31 72.57 19.69 .315 55

.134 .657 .001 .114 .000

1.27 11.03 7.01 7.58 –3.67 35.51 .049 89

.518 .251 .386 .158 .843

1.01 2.10 1.24 1.50 .04 81.4% .024 70

.910 .362 .737 .357 .064

1.06 .64 .60 .41 .87 81.1% .075 90

.677 .529 .406 .030 .919

Note: hr = hegemonic, cr = competitive regimes. se = Standard error, c&s R2 = Cox & Snell R2, Lineal ols regressions for margins of victory (reporting unstandardized B coefficients, standard errors, and R2 values), binary logistic regressions for constitutional majority and alternation in power (reporting odds ratios eb, Percentages of Accuracy of Classification pac, and Cox & Snell R2 values). Repression and censorship for year previous to election year. Shaded coefficients = statistically significant (p ≤ .15).

repression (along the 0–8 scale of violations of physical integrity), they suffer a loss of 24 percentage points in their lead vis-à-vis their closest competitor. If they switch from electoral integrity to electoral fraud (from score 0 to 2),

224

Empirical Explorations

Hypothetical changes in margins of victory

they lose 19.1 points. Under this perspective, the strategic dilemmas outlined in Chapter 4 appear as more than a mere possibility. For hegemonic regimes, they would seem to be a tangible reality.5 In competitive regimes, the relations between manipulation and margins of victory are more consistent, yet also more tenuous. While all coefficient signs are positive, which speaks of authoritarian success in widening margins of victory, only fraud and censorship approach conventional levels of statistical significance. Taken together, the four manipulative strategies explain no more than a fraction in variations of competitive margins of victory (R2 = 0.05). Figure 7.1 illustrates the bivariate regression results presented in Table 7.1. The vertical bars indicate the magnitude of changes in margins of victory that would result from comprehensive shifts in manipulative strategies along their entire scales, from full absence to full presence. White bars indicate results below statistical significance (at the permissive level of p ≤ .15). They hold within our sample of elections, but not beyond. The figure highlights the outstanding roles of exclusion in hegemonic and fraud in competitive regimes. If

30.0 20.0 10.0

16.8 0.5

0.0 –10.0

–24.0

21.6

21.9

16.7 6.7

–19.1

–20.0

R H

io n es s

n es sio

Re pr

Re pr

Ce CR ns or sh ip Ce H ns R or sh ip Ex CR clu sio n H Ex R clu sio n CR Fr au d H R Fr au d CR

–30.0

Figure 7.1. Hypothetical effects of manipulation on margins of victory Note: The bars show predicted changes in margins of victory, in response to hypothetical changes in individual strategies of manipulation from complete absence to full presence (calculated on the basis of bivariate ols regression results presented in Table 7.1). Shaded bars denote statistically significant effects (p ≤ .15).

Struggle over Electoral Uncertainty

225

hegemonic parties switch from inclusionary (score 0) to exclusionary policies (score 1), they can expect a plus of 21.9 percent in their margins of victory. If competitive parties move from integrity (score 0) to fraud (score 2), they can expect a bonus of 16.7 percent. Media restrictions seem to be a “productive” strategy in both subtypes of regimes, although we cannot generalize beyond my case sample. What happens to the associations between manipulation and competitiveness when we switch from a static to a dynamic perspective? If levels of manipulation show systematic relations with levels of competitiveness (without taking into account short-term changes on either side), to what extent do changes in the degree of manipulation relate systematically to changes in the degree of competitiveness (without taking into account prevailing levels on either side)? Table 7.2 provides a tentative answer. It compares the average magnitude of changes in legislative margins of victory across three scenarios: when manipulation decreased in intensity since the previous election (relaxation), when it remained constant (no change), and when it increased (tightening). Hegemonic regimes show no consistent pattern. If anything, changes in their manipulative strategies produce puzzling effects. When they tighten the screws, their competitive advantage erodes. When they loosen them, it widens. Manipulation, I posited above, sends mixed signals. It conveys authoritarian troubles as well as authoritarian resolve. In hegemonic manipulation, I

Table 7.2. Changes in legislative manipulation and margins of victory Hegemonic regimes

∆ Repression

∆ Censorship

∆ Exclusion

∆ Fraud

Tightening No change Relaxation Tightening No change Relaxation Tightening No change Relaxation Tightening No change Relaxation

Competitive regimes

∆ MV

N

p

∆ MV

N

p

–3.23 11.50 9.42 –1.81 6.54 6.27 –12.34 4.77 26.35 4.31 6.12 –3.49

10 6 12 5 17 7 1 28 1 4 23 3

.375

15.83 –3.57 –11.39 4.70 –5.26 3.16 18.95 –.10 –31.86 10.65 –8.42 –6.37

9 11 13 7 19 7 2 25 4 8 15 8

.109

.786

.508

.805

.699

.056

.294

Note: ∆ mv = changes in legislative margins of victory. ∆ Exclusion and fraud = level changes with respect to previous election. ∆ Repression and censorship: level changes from previous election to pre-election year. P-values from Analyses of Variance (anova). Shaded cells denote statistically significant differences (p ≤ .15).

226

Empirical Explorations

expected signals of strength to trump signals of weakness. My data suggest otherwise. Unlike their hegemonic kin, competitive regimes seem to benefit clearly from short-term adjustments in their authoritarian strategies. If ruling parties relax their authoritarian controls, they lose ground; if they tighten controls, they gain ground. The short-term effects of changes in manipulation on changes in competitiveness are most pronounced with respect to repression and exclusion. Tightening repression pays fine electoral dividends (+15.8 percent) and so does the banning of opposition actors (+18.9 percent). Inversely, easing the repressive treatment of dissidents makes electoral margins dwindle (−11.4 percent), readmitting banned opponents into the electoral arena makes them collapse (−31.9 percent). We need to insert a twin caveat here. When we wish to estimate the causal effects of manipulation on competitiveness we should take into account two probable selection effects. First, the selection of manipulative strategies is likely to be contingent on the expected usefulness of manipulation. If strategies are only needed under certain conditions, rulers should only choose them under these conditions. To know whether solutions are effective or not, we need to know the magnitude of the problems they address. We cannot look at outcomes in isolation. The true measure of pharmaceutical effectiveness is not that my medicine keeps me healthy while I am not even sick. Its effects must be assessed relative to the gravity of my illness. If manipulation serves to contain institutional uncertainties, its success must be measured relative to the level of uncertainty that rulers face at the moment of manipulative choice. Second, the selection of manipulative strategies is likely to be contingent on the expected effectiveness of manipulation. If strategies are only effective under certain conditions, rulers should only choose them under these conditions. Effective manipulation requires the cooperation of state agents and the acquiescence of citizens (see Chapters 4 and 5). Broadly speaking, the stronger the state and the weaker civil society, the more likely it is that manipulative maneuvers turn out successful. If instruments of authoritarian rule depend for their effectiveness on the presence of (knowable and discernible) enabling conditions, we should expect rulers to adopt them only if these conditions are present. They should not take authoritarian measures unless these measures can be expected to be successful. I tried to trace both selection effects through various statistical exercises, but to no avail.6 In any case, both effects should weaken the direct relationship between manipulation and competitiveness as they render it contingent on antecedent conditions. The fact that I observed such direct associations speaks for their strength.

Struggle over Electoral Uncertainty

227

Summing up: Hegemonic regimes are highly effective manipulators. At similar levels of manipulation, they enjoy much higher levels of institutional security than their competitive counterparts. As my data suggest, the magic wand of their authoritarian effectiveness is the exclusion of competitors. Hegemonic regimes are more exclusionary on average than competitive regimes, and exclusion stands out as their most effective authoritarian strategy in the everyday production of their overwhelming margins of victory. More than we commonly recognize, their electoral hegemony seems to be grounded in their exclusionary nature. Which makes sense. In their origins, most hegemonic parties are the victors of dramatic violent conflict: independence, revolution, civil war. Staging themselves as saviors and protectors of the nation, they claim a monopoly over its legitimate representation and thus the right to declare minorities illegitimate and expel them from the political arena. Not all hegemonic regimes are exclusionary, but those that are clearly benefit from subjecting the electoral arena to political cleansing. Resting uneasily upon a more fragile and volatile support base, competitive regimes deploy a more diversified set of authoritarian means of production to augment their margins of victory. Fraud and censorship appear as their most productive instruments, yet short-term adjustments in repression and exclusion, too, generate impressive immediate gains in electoral dominance. The dynamic comparison of changes in manipulation and competitiveness shed light on further differences between the two subtypes of regimes. In competitive regimes, contrary to my initial expectations, levels of electoral competitiveness seem rather sensitive to short-term policy changes.

The Weapons of the Weak Let us now shift our attention from the manipulative weapons that rulers wield to contain electoral uncertainty towards the contentious weapons opposition actors possess to ignite electoral uncertainty. To assess the short-term impact that opposition protest bears on average on levels of authoritarian electoral competitiveness, I use five indicators of political dissidence. From the Dataset on Authoritarian Elections (dae), I use two measures of opposition protest that are proximate and explicitly related to the conduct of elections: pre-electoral protest (acquiescence versus active mobilization) and opposition boycott (full participation, partial boycott, and full boycott).7 From the Arthur Banks Cross-National Time–Series Data (cnts), I construct three aggregate measures of conflictive events between elections, during the three years preceding each election year: (a) a measure of various forms of “vertical threats” (the sum of anti-government demonstrations, general strikes, and

228

Empirical Explorations

riots during these years); (b) a more narrow indicator of organized political violence or “armed rebellion” (the sum of guerrilla warfare and revolutions); and (c) an indirect measure of internal elite opposition or “horizontal threats” (the sum of government crises, purges, and military coups). Although horizontal threats emanate from a different universe of actors than vertical ones, I will examine the short-term effects that the two categories of threats have on electoral competitiveness side by side. Table 7.3 shows bivariate relationships between contentious events and electoral outcomes. While some of the links between manipulation and competitiveness that we examined above looked blurred and inconsistent, these lineal regression results reveal rather straightforward patterns of covariation. All relations show the expected signs and most are statistically as well as Table 7.3. Effects of opposition strategies on electoral competitiveness Margins of victory hr B

p

Margins of victory cr B

Constitutional majority cr

Party alternation in power cr

p

eb

p

eb

p

Vertical threats Constant Standard error / pac R2 / Cox & Snell R2 N

–1.18 .122 –1.48 69.54 .000 35.77 22.84 34.95 .034 .077 56 88

.009 .000

.93 .31 80% .018 70

.331 .001

1.05 .16 80.9% .026 89

.118 .000

Armed rebellion Constant Standard error / pac R2 / Cox & Snell R2 N

–3.33 .230 –2.74 68.76 .000 34.69 23.05 35.57 .026 .044 56 88

.048 .000

.64 .40 80% .077 70

.078 .010

1.18 .14 82.0% .042 89

.051 .000

Horizontal threats Constant Standard error / pac R2 / Cox & Snell R2 N

–8.68 .235 –5.60 68.01 .000 34.77 23.05 35.62 .025 .041 56 88

.056 .000

.46 .41 80% .076 70

.056 .014

1.42 .14 82% .036 89

.064 .000

Pre-electoral protest Constant Standard error / pac R2 / Cox & Snell R2 N

–12.40 .081 –3.81 70.00 .000 30.88 22.71 36.80 .054 .003 56 93

.621 .000

1.71 .16 80.8 .010 73

.406 .001

1.61 .17 80.9 .008 94

.382 .000

Struggle over Electoral Uncertainty

229

Table 7.3. (Contined) Margins of victory hr

Constitutional majority cr

Party alternation in power cr

p

eb

p

eb

p

6.68 .092 16.59 63.31 .000 21.32 22.75 34.78 .051 .109 56 93

.001 .000

4.93 .07 86.3% .202 73

.000 .000

.45 .31 80.9% .034 94

.111 .000

–1.21 –1.87 –2.48 –7.29 15.62 36.11 33.219 .205 88

.029 .158 .385 .347 .004 .000

.99 .71 .68 6.99 7.64 .03 88.6% .311 70

.961 .310 .445 .064 .001 .003

1.05 1.14 1.34 1.83 .62 .07 80.9% .108 89

.160 .154 .150 .392 .402 .001

B Boycott Constant Standard error / pac R2 / Cox & Snell R2 N Contentious action Vertical threats Armed rebellion Horizontal threats Pre-electoral protest Boycott Constant Standard error / pac R2 / Cox & Snell R2 N

Margins of victory cr

–.27 –.01 –12.63 –13.90 8.09 68.11 22.10 .169 56

p

.771 .998 .148 .077 .041 .000

B

Note: hr = hegemonic, cr = competitive regimes. Vertical threats = anti-government demonstrations, general strikes, and riots; armed rebellion = guerrilla warfare and revolutions; horizontal threats = government crises, purges, and coups (sum of events during three years previous to election year). Source: cnts. Lineal regressions for margins, binary logistic regressions for supermajorities and alternation. pac = percentage of accurate classification. Shaded coefficients = statistically significant (p ≤ .15).

substantively significant. All sources of threat, vertical as well as horizontal, peaceful as well as violent, undermine margins of victory, in hegemonic as well as in competitive regimes. Each single “vertical threat” that arises during the three years preceding the election year, that is, each single anti-government demonstration, general strike, or riot, erodes official margins of electoral victory by more than one percentage point. The effects of armed rebellions, horizontal threats, and pre-electoral protests are even more pronounced. Competitive regimes are more competitive in contexts of organized violence and armed rebellions also chip away at the electoral dominance of hegemonic parties (although below conventional levels of statistical significance). Horizontal threats are serious and rare events. They are almost inexistent in hegemonic regimes (only five hegemonic elections in the dataset are preceded by horizontal threats) yet if they occur they depress margins of victory

230

Empirical Explorations

by more than 12 percentage points. The same holds for opposition mobilization against hegemonic elections. Pre-electoral protests are a prelude of competitiveness. Their occurrence depresses hegemonic margins of victory by almost 14 percent (see multivariate regression at bottom of Table 7.3). While competitive regimes seem almost immune, hegemonic regimes appear highly allergic to protest mobilizations against upcoming elections.8 Only boycotts, as could hardly be otherwise, inflate the winning margins of authoritarian incumbents, in particular in competitive regimes. If opposition parties proceed from full participation to full withdrawal, competitive regimes augment their winning margins by 31.2 percent, hegemonic regimes by 13.4 percent. This asymmetry is unsurprising, since opposition parties are by definition weaker in hegemonic regimes. If they withdraw, the overall results do not change that much. Finally, the overall explanatory capacity of contentious action is appreciable. In hegemonic regimes, these conflict events explain about one-sixth of variations in interparty competitiveness (R2 = .17), in competitive regimes one-fifth (R2 = .20).

20.0 31.2 13.4

0.0

–3.6

–4.5

–10.0

–3.8

–8.2 –16.8

–12.4

–26.1

–20.0

Ve r

tic a Ve l thr ea rti ts Ar cal H R m thr ed ea t r s Ar e CR m bel H ed lion or r izo ebe HR nt llio H or al th n C Pr izon rea R t eele tal t s H R h c r to Pr ra eats el ele C pr ot R ct or es tH al pr ot R es Bo t CR yc ot Bo t H yc R ot tC R

Predicted changes in margins of victory

40.0

Figure 7.2. Hypothetical effects of opposition strategies on margins of victory Note: The bars show predicted changes in margins of victory, in response to hypothetical changes in protest events: from no vertical threat, armed rebellion, or horizontal threat to three events per year (in the three years previous to the election), from pre-electoral acquiescence to active mobilization, and from full participation to full boycott (calculated on the basis of bivariate ols regression results presented in Table 7.3). Shaded bars denote statistically significant effects (p ≤ .15).

Struggle over Electoral Uncertainty

231

Figure 7.2 illustrates the bivariate regression results in a more tangible fashion. The vertical bars indicate predicted changes in margins of electoral victory in response to changing levels of extra-electoral uncertainty. Levels of inter-election conflict (vertical threats, armed rebellion, and horizontal threats) change from zero to three events during the three years preceding the election year (one per year). Levels of election-related conflict change across their full scale, pre-electoral protest from acquiescence to active mobilization (a one-unit change) and opposition boycott from full participation to full boycott (a two-unit change). White bars indicate results that do not hold outside my sample, as they lie below my (permissive) level of statistical significance (p ≤ .15). Summing up: It is a rather rare blessing, if the political world, in all its infinite complexity, grants us the gift of lineal short-term relations between presumptive independent and dependent variables. As we have just seen, such simple associations do exist, with some exceptions and qualifications, between contentious action and electoral competitiveness. The active mobilization of opposition actors between and before national elections goes hand in hand with substantive downward swings in ruling parties’ electoral margins. If there is a causal story behind these findings, it is decidedly encouraging for opposition parties. Those anti-regime actors who are strong and bold enough to actively mobilize against upcoming authoritarian elections have a fair chance of increasing their competitiveness. If they succeed in inducing splits within the regime, their chances are even better. Hegemonic regimes seem somewhat less sensitive to societal threats in between elections. Yet, confirming my theoretical expectations, they seem more sensitive to fissures in the political elite as well as to contentious actions that target the electoral process itself. Opposition parties that choose to protest by withdrawing from elections have to know that they grant substantial immediate advantages to their adversaries in power. I’m agnostic about the long-term consequences of election boycotts. They may retard the prospects of democratization, as Staffan Lindberg finds (2006a). Or they may trigger political reform, as Emily Beaulieu suggests (2006). Yet, in particular in competitive regimes, the short-term damage opposition boycotts impose on electoral competitiveness is substantial.

The Weight of Institutional Design Structural and institutional factors are even more distant determinants of electoral outcomes than the stirrings of opposition, either outside or within the ruling coalition. Authoritarian elections do not unfold in a societal vacuum, though. They are embedded within institutional and societal contexts that define constraints and opportunities and allocate politically relevant

232

Empirical Explorations

resources to contending parties. Accordingly I will explore a small bundle of institutional and structural factors as contending (or rather, complementary) explanations of electoral uncertainty. Men, as Karl Marx famously wrote, make history, yet not under self-selected circumstances.9 Electoral authoritarian regimes make history, too, yet under self-selected circumstances. By definition, hegemonic party regimes control the levers of constitutional choice. Competitive regimes often do so as well, be it through their formal control of legislative supermajorities or their factual control of legislatures. Insofar as they do, political institutions are endogenous to electoral authoritarian regimes. Just like short-term strategies of manipulation, they constitute policy instruments at the service of regime survival, even when governments do not adjust them from election to election. To estimate the effects institutional variables carry on margins of victory, I shall explore two dimensions of the institutional context of authoritarian elections: institutional design and institutional strength. On the former, I will include data on forms of government, electoral systems, subnational elections, and electoral timetables; on the latter, measures of the electoral cycles, legislative strength, and personal rule. Since most institutional arrangements carry different implications for assembly elections than for executive elections, I segment my sample by electoral arena (presidential versus parliamentary contests). Again I also try to sort out differential effects by subtype of regime. Table 7.4 presents comparisons of mean margins of victory between the major institutional categories. It also indicates the statistical significance of group differences in Analyses of Variance (anova). Figure 7.3 visualizes selected patterns of association between political institutions and average margins of victory in both subtypes of electoral authoritarian regimes. Since hegemonic regimes are able to tame and tailor institutions at their convenience during long time-spans of learning and adjustment, I expect variations in institutional design to matter less for hegemonic outcomes. Competitive regimes are more likely to err and choose designs that end up hurting their vote- and seat-maximizing ambitions. Hegemonic parties either choose appropriately the institutions that serve their enterprise of electoral domination or they adapt appropriately to the institutions they choose. If institutional arrangements exist that diminish their electoral returns in general, across structural and strategic contexts, they should not pick them. Forms of government: Students of democracy have long been discussing whether and under which conditions presidential forms of government constitute sources of democratic conflict and instability. Non-democratic regimes raise the inverse concern: whether and under which conditions presidential forms of government constitute foundations of authoritarian conflict management and stability. Authoritarianism and presidentialism are often thought

Table 7.4. Political institutions and electoral competitiveness Legislative mv ø N institutional design Form of government Presidentialism or Semi-P. Parliamentarism Electoral system Proportional representation Majoritarian or mixed Electoral timing Non-concurrent elections Concurrent elections State elections No Yes Municipal elections No Yes institutional strength Electoral cycle 1st elections 2nd and 3rd elections 4th + elections Legislative strength Effective legislature Ineffective legislature Personalism Non-personalist regime Personalist regime

[p = .137] 41.41 53.04 [p = .003] 13.80 44.31 [p = .339] 46.58 39.80 [p = .768] 46.87 43.81 [p = .109] 42.46 15.36 [p = .016] 36.02 44.24 61.37 [p = .106] 39.21 50.11 [p = .904] 47.79 49.00

90 29 17 63

30 31 15 15

52 42 25 64 55 78 18

Presidential mv ø N

40.18 [p = .054] 13.22 36.98 [p = .871] 41.02 39.59 [p = .487] 42.60 33.28 [p = .806] 29.52 23.90 [p = .693] 39.70 38.39 53.74 [p = .009] 27.54 50.04 [p = .559] 38.07 44.81

73

10 41 30 43 18 16 6 10

41 27 5 32 41 40 15

Hegemonic mv ø N

Competitive mv ø N

[p = .471] 65.43 70.24 [p = .995] 55.70 63.11 [p = .610] 67.81 64.03 [p = .562] 63.26 67.55 [p = .145] 72.57 50.45

[p = .678] 29.84 24.91 [p = .109] 12.38 29.10 [p = .962] 29.39 28.99 [p = .099] 29.73 9.26 [p = .090] 24.34 –4.51

[p = .083] 58.72 65.66 75.10 [p = .061] 73.43 61.88 [p = .109] 69.49 57.41

39 18 1 28 44 13 18 16 7 6

18 18 21 25 32 45 12

[p = .651] 29.41 31.28 16.17 [p = .000] 15.26 44.51 [p = .027] 18.20 43.09

83 11 17 54 64 30 19 21 10 14

53 35 6 49 45 44 16

Note: mv ø = mean margins of victory. P-values from Analyses of Variance (anova). Shaded cells are statistically significant (p ≤ .15). Sources: for subnational elections: World Bank Database on Political Institutions; for personalism: Barbara Geddes Authoritarian Regime Data. Shaded cells denote statistically significant differences (p ≤ .15).

234

Empirical Explorations Competitive regimes

Competitive regimes

Hegemonic regimes

Hegemonic regimes

60.00

40.00 63.12 55.71

20.00

Mean margin of victory

Mean margin of victory

60.00

40.00 65.14

20.00 30.15

29.11 12.39

8.98

0.00

0.00 Proportional representation

80.00

Majo a ritarian or mixed-member system

No

Yes

Subnational elections

Competitive regimes

80.00

Competitive regimes

Hegemonic regimes

60.00

40.00

75.11 65.67 58.72

20.00

Mean margin of victory

Hegemonic regimes

Mean margin of victory

62.66

60.00

40.00 73.44 61.89 44.51

20.00

31.29

29.41

16.17

15.26

0.00

0.00 1st elections

2nd or 4th + elections 3rd elections Sequence of elections

Effective legislature a s Ineffective legislature a s

Competitive regimes Hegemonic regimes

Mean margin of victory

60.00

40.00 69.50 57.41 43.09

20.00 18.20

0.00 Non-personalist regimes

Personalist regimes

Figure 7.3. Political institutions and competitiveness by subtype of regime Note: Shaded bars denote statistically significant differences (p ≤ .15).

Struggle over Electoral Uncertainty

235

to carry elective affinities, as both tend to concentrate power at the peak of the executive. My basket of cases does not quite support this notion. Little more than one-tenth of competitive authoritarian elections take place in parliamentary regimes (11.7 percent), yet almost one-third of hegemonic elections do so (31.6 percent). The construction of lasting hegemonies on the basis of parliamentary rule is clearly a feasible option. While the electoral autocracies of Latin America and the former Soviet Union have been exclusively presidential, the Asian route to hegemonic rule passes through parliamentary institutions. In terms of its capacity to contain interparty competitiveness, parliamentarism should not be expected to be inferior to presidentialism. As my data suggest, it even shows a slight comparative advantage. Legislative margins of victory are higher by about ten percentage points in parliamentary systems (where they matter more) than in presidential regimes (see Table 7.4).10 Electoral systems: According to conventional wisdom, majoritarian systems favor the ruling party (for as long as it continues to be the majoritarian party), while proportional representation (pr) grants larger spaces of representation to the opposition (even though it encourages its fragmentation). Due to its propensity to further the concentration of power, majoritarianism, just like presidentialism, is thought to possess elective affinities with authoritarianism. The distribution of electoral rules within my basket of authoritarian elections lends qualified support to this notion. Hegemonic regimes almost never resort to proportional representation (the only one case is Nicaragua in 1984), while a minority of competitive elections is held under pr (23.9 percent). More than half of hegemonic elections (51.7 percent) and almost two-fifths of competitive elections (39.4 percent) are held under pure majoritarian rules. pr rules seem to belong to the category of generally harmful rules electoral autocrats generally stay away from. For the purpose of comparing levels of electoral competitiveness, I fuse majoritarian and mixed-member systems as they show similar mean margins of victory. Due to low case numbers, I cannot infer anything about hegemonic regimes. In competitive regimes, however, the impact of electoral systems looks quite dramatic. Ruling parties more than duplicate their margins of victory if elections are held in mixed-member or majoritarian systems (29.1 percent) rather than pr (12.4 percent). Within the full sample of electoral autocracies, legislative elections under pr deliver margins of victory that are less than one-third (13.8 percent) of those under more majoritarian systems (44.3 percent). Even presidential margins of victory shrink to a similar extent when legislative elections are governed by proportional representation. While the combination of presidentialism and proportional representation has been regarded as “inimical to stable democracy” (Mainwaring 1990: 168), it may

236

Empirical Explorations

also be inimical to comfortable authoritarianism. In the largely presidential world of competitive regimes, majoritarian formulae seem to work as effective tools of authoritarian governance, proportional rules as effective instruments of opposition (see also Figure 7.3).11 Electoral timing: Should we expect incumbents to enhance their competitive advantages by manipulating electoral cycles? It is not clear that they are able to do so. The mutually “contaminating” concurrence of legislative elections with presidential contests may cut either way in terms of electoral competitiveness. Non-concurrence may be competition-depressing in legislative elections if opposition parties find it harder to mobilize in elections of low salience. Yet concurrence may have the same effect if incumbents throw the whole weight of the state apparatus behind dual campaigns of high salience.12 These ambiguous expectations translate into non-ambiguous non-findings: mean margins of victory do not vary between concurrent and non-concurrent elections (see Table 7.4). Subnational elections: The conventional focus of comparative politics lies on the central politics of national states. It involves a twin blindness towards international influences as well as towards subnational dynamics. Over the past years, though, comparative scholars have been discovering the potential of subnational comparison (see Snyder 2001). Students of electoral authoritarian regimes, in particular, have been conducting research on subnational elections both at the municipal level and at the state level in federal systems.13 Subnational elections are commonly thought to decompress and decentralize conflict. Their institutional design and substantive outcomes are often subject to vetoes from the center. They nevertheless carry the potential of developing into sites of “centripetal democratization” (Mizrahi 2002). To the extent that elections below the national level indeed generate democratizing impulses that irradiate from the periphery to the center, we should expect electoral autocracies that hold subnational elections to be more competitive at the national level than those that do not. To trace the “subversive” force of subnational elections, I compare mean levels of competitiveness of polities that conduct either municipal or state elections with those that do not (according to the World Bank Dataset on Political Institutions). Due to missing data, case numbers are low, yet still indicative. While common sense suggests that higher-level elections to subnational units in federal systems should be more relevant than lower-level elections, my data indicate the primacy of local politics. The organization of state-level elections marks a significant difference in competitive regimes. In their absence, incumbents win national elections by comfortable margins (29.7 percent). In their presence, these margins shrink by two-thirds (9.3 percent). Yet, the effects of municipal elections on national competitiveness look even stronger and more

Struggle over Electoral Uncertainty

237

consistent. Their presence involves a shrinkage in margins of victory by about 20 percentage points, in competitive elections as well as in hegemonic contests. The same applies for legislative elections within the entire sample. Only presidential elections, more dependent on persons and mass media than party organizational networks, seem untouched by the organization of municipal politics (see Table 7.4 and Figure 7.3).

The Weight of Institutional Strength The defining difference between hegemonic and competitive regimes lies in their different degrees of institutionalization: hegemonic regimes are firmly institutionalized, competitive regimes much less so. Within these contrasting worlds of macro-institutionalization, however, micro-institutions may vary considerably along the same dimension of institutional strength. I will examine three specific aspects of institutionalization within electoral autocracies: the length of electoral cycles, the strength of legislatures, and the degree of personalism at the peak of state power. Electoral cycles: To the extent that the two-level dynamics of authoritarian elections unfold over time, the length of uninterrupted election cycles may affect levels of electoral competitiveness. In principle, electoral institutionalization could cut both ways. If the repetition of elections induces a process of learning, their overall effect depends on the learning capacity of everybody involved. If the ruling party gets better and better either at mobilizing or manipulating voters, the institutionalization of authoritarian elections should widen its margins of victory. The result would be authoritarian consolidation. If opposition parties get better and better either at persuading citizens or at countering electoral manipulation, the extension of electoral cycles should lead to steady increases in interparty competition. The result would be democratization by elections. If both sides learn and get more and more effective, or if neither does, little should happen to aggregate levels of competitiveness. In practice, the length of authoritarian election cycles seems to go hand in hand with decreasing levels of competitiveness. In particular, legislative margins of victory almost double from first elections (36 percent) to fourth and subsequent elections (61.4 percent). This trend roughly matches the patterns Staffan Lindberg reported for elections in sub-Saharan Africa: even if the length of uninterrupted electoral cycles carries benign effects on civil liberties as well as electoral legitimacy, it does not generate systematic increases in electoral competitiveness (Lindberg 2009c: 33, see also Rakner and van de Walle 2009).

238

Empirical Explorations

Competitive regimes produce much fewer uninterrupted sequences of four elections or more. They account for only six of the twenty-seven fourth or subsequent legislative elections in my sample (22.2 percent) and for three out of seven fourth or subsequent presidential elections (42.9 percent). The depressive effect electoral institutionalization exerts on legislative competitiveness seems to derive primarily from hegemonic parties that succeed in consolidating their electoral dominance over time. If hegemonic parties enjoy ample average margins of victory in first elections (58.7 percent), they succeed in widening these margins even further in fourth and subsequent elections (75.1 percent) (see Table 7.4 and Figure 7.3).14 Legislative strength: Authoritarian governance is often associated with executive dominance over weak legislatures. By keeping legislatures weak, electoral authoritarian regimes keep opposition parties weak. Legislative disempowerment castrates political parties. If national assemblies lack genuine decision-making powers and oversight capacities, legislative elections are unlikely to develop into serious contests for power. They tend to serve as arenas of competition over access to state patronage (see Fish 2006, Lust-Okar 2009, Weber 1988: 536–45). Imbalances in executive–legislative relations should find almost direct expressions in electoral imbalances. In the absence of legislative checks and balances, overpowering executives should find it comparatively easy to manufacture wide margins of victory. My data confirm the close association between the ineffectiveness of national legislatures and the electoral ineffectiveness of opposition parties.15 In electoral autocracies with weak legislatures, winners enjoy much wider margins of victory than in the presence of at least partially effective legislatures. The contrast is most pronounced in competitive regimes in which legislative ineffectiveness translates into a drop in average electoral competitiveness by almost 30 percentage points. Effective legislatures thus seem to be capable of inducing higher levels of electoral competitiveness—although it may well be the other way round: higher competitiveness strengthens national legislatures. The faithful elephant of endogeneity keeps grazing on our premises. One additional irritation derives from hegemonic legislatures. Under hegemonic rule, legislative effectiveness goes hand in hand with wider, rather than narrower, margins of victory. One may wonder about the meaning of “legislative effectiveness” under conditions in which a hegemonic player exerts full control over the legislative assembly. Yet, as it appears, granting a modicum of power to the legislature that it controls seems to benefit the hegemonic incumbent (see Table 7.4 and Figure 7.3). Personalism: A third indicator of institutional strength is more generic: the personalistic nature of the regime. Conventional typologies distinguish

Struggle over Electoral Uncertainty

239

authoritarian regimes by the organization that controls the levers of state power: Military regimes are ruled by the military, single-party regimes by the ruling party, and personalist regimes by a single person, rather than an organization. Personalism is thus synonymous with institutional weakness. It renders regimes more dependent on the “brilliant genius of humanity” who happens to occupy the helm of the state.16 It also renders them more vulnerable to succession crises and unmediated struggles for power. Still, when it comes to running electoral campaigns, the personal appeal of the chief executive may compensate for the weakness of his organizational infrastructure. To assess the impact of personalist features on authoritarian electoral outcomes, I employ the authoritarian regime data developed by Barbara Geddes. Dichotomizing her regime categories, I distinguish pure and mixed personalist regimes on one hand from non-personalist regimes on the other. As simple comparisons of means indicate, personalist rule leaves traces on electoral outcomes in both hegemonic and competitive regimes, yet in contrasting manners. While hegemonic regimes fare better if they can avoid the personalization of power, competitive regimes benefit substantively from the presence of personal rulers. Margins of victory in competitive regimes with personalist traits are almost twice as high as in non-personalist regimes. Personal domination over the electoral arena thus appears to be a liability in the highly institutionalized contexts of hegemonic rule and an asset in the more fluid contexts of competitive authoritarianism (see Table 7.4 and Figure 7.3). These simple (uncontrolled) comparisons of mean margins of victory under different institutional arrangements lend credit to the idea that the design as well as the strength of political institutions matter under electoral authoritarianism. When we regress margins of victory on several of these institutional variables in competitive regimes (case numbers are too low to include hegemonic regimes), legislative ineffectiveness emerges as the most powerful institutional variable. The overall explanatory weight of institutions is quite high. Only four institutional variables account for about one-eighth of variance of competitive margins (R2 = .127) (see Table 7.5).

The Weight of Structural Contexts My account of electoral authoritarian regimes rests upon two strong assumptions: first, elections are the central arena of struggle between government and opposition; second, their outcomes are the joint product of voter choices and regime manipulation. This election- and actor-centered conception of regime dynamics does not deny the potential relevance of structural contexts. Authoritarian elections, just like democratic ones, are not suspended in thin air,

240

Empirical Explorations

Table 7.5. Institutional determinants of electoral competitiveness in competitive regimes Margin of victory

Parliamentarism Majoritarian electoral system Legislative ineffectiveness Electoral sequence Constant Standard error / pac R2 / Cox & Snell R2 N

B

p

3.118 3.961 25.297 2.436 4.725 35.998 .127 70

.595 .489 .008 .727 .740

Constitutional majority eb 3.967 5.928 .790 1.117 .004 87.7% .222 57

Party alternation in power

p

eb

p

.025 .023 .798 .883 .010

1.598 .827 2.315 .974 .171 76.1 .065 71

.203 .610 .211 .953 .073

Note: Parliamentarism = trichotomous (presidentialism, semi-presidentialism, and parliamentarism). Majoritarian electoral system = trichotomous (majoritarian system, mixed-member system, and proportional representation). Electoral sequence = trichotomous (1st, 2nd + 3rd, and 4th and subsequent elections), legislative ineffectiveness = dummy. Lineal ols regressions for margins of victory, binary logistic regressions for constitutional majority and alternation in power. pac = percentage of accurate classification. Shaded coefficients = statistically significant (p ≤ .15).

but deeply embedded in their societal environment. Elections are no islands. The conflictive encounters between government, state agents, opposition parties, and citizens in the electoral arena take place within a larger societal playing field. The nested game of authoritarian elections is itself nested within a context of societal and state structures that sets constraints and opportunities for actors within the electoral arena. Recognizing the embeddedness of authoritarian elections does not deny the relative autonomy of the electoral realm. We should not assume elections to be mirrors of their structural contexts. They follow their own logic with certain degrees of insulation. We should not expect structures to determine political outcomes. Games are fluid, parameters constant. Numerous filters and branching points lie along the extensive causal chain that leads from structural factors to electoral outcomes. Now, what is it that contextual structures do to electoral competition? In general terms: what do structures do to actors? Essentially, they provide (and withhold) resources. Some do so in a symmetrical manner (affecting all sides equally), others in an asymmetrical way (benefiting some and hurting others). When societal structures exert symmetrical effects on government

Struggle over Electoral Uncertainty

241

Table 7.6. Effects of structural contexts on electoral competitiveness Margins of victory hr

Margins of victory cr

Constitutional majority cr

eb

p

.025 .005 139.32 .012 83.6% .128 73

1.543 .056 80.6% .003 93

.626 .338

.178 .000

1.015 .805 .237 .000 81.2% .001 69

1.095 .218 83% .024 88

.178 .000

–.083 31.934 37.377 .000 89

.841 .070

1.005 .861 .179 .196 81.7% .000 71

.975 .701 80% .009 90

.376 .762

.269 .000

.173 17.730 36.476 .020 93

.171 .056

1.005 .648 .176 .018 80% .003 51

.986 .569 80.9% .021 94

.195 .416

.477 .415 .055 .775 .940

–25.278 –.690 .362 .187 85.592 35.751 .083 82

.070 .356 .439 .235 .111

1.180 1.054 .931 .968 16.386 83.1% .103 83

.874 .446 .054 .043

B

p

B

p

Wealth Constant Standard error / pac R2 / Cox & Snell R2 N

22.157 –8.692 21.860 .124 56

.007 .751

–27.021 119.81 36.127 .048 92

.036 .006

Growth Constant Standard error / pac R2 / Cox & Snell R2 N

3.232 61.284 21.678 .152 55

.003 .000

–.974 28.212 35.513 .021 87

Inequality Constant Standard error / pac R2 / Cox & Snell R2 N

.709 38.731 19.621 .073 46

.066 .024

International trade Constant Standard error / pac R2 / Cox & Snell R2 N

.092 57.068 21.938 .025 50

Societal structures Wealth Growth Social inequality International trade Constant Standard error / pac R2 / Cox & Snell R2 N

9.386 1.123 .789 .024 –2.857 17.598 .206 39

eb

Party alternation in power cr

.007 1.028 1.034 1.005 679.72 83.6% .186 67

p

.000 .896 .955 .978

Note: hr = hegemonic, cr = competitive regimes. Wealth = log of gdp per capita in current Purchasing Power Parities (ppp). Growth = average gdp per capita growth in the five years before the election. Social inequality = income inequality (Gini coefficient). Trade = percentage of international trade of gdp. Source: World Bank Development Indicators. Lineal ols regressions for margins of victory, logistic regressions for constitutional majority and alternation in power. pac = percentage of accurate classification. Shaded coefficients = statistically significant (p ≤ .15).

242

Empirical Explorations

and opposition actors, their overall consequences on the political balance of power will be contradictory and thus indeterminate. When they carry asymmetric consequences, their implications will be lineal. For example, well-developed communication networks facilitate coordination within the government camp as well as within the opposition. By contrast, the availability of mineral oil grants unilateral resource advantages to governments. Just as electoral playing fields may be skewed, societal playing fields may be skewed too. Such societal asymmetries are likely to translate into electoral asymmetries. In this section, I explore the lineal causal effects of only four thin contextual variables: economic development (monetary wealth per capita), the depth of societal cleavages (social inequality), medium-term economic performance (average growth rates in the five years before the election year), and external economic dependence (international trade as percentage of gdp). Through lineal regression analysis, I estimate both the individual and the joint effects that these variables carry on margins of victory in hegemonic and competitive regimes (see Table 7.6). Given their presumptive ability to draw on reservoirs of popularity and repression, one would expect hegemonic regimes to be less dependent on their structural environment. At the same time, given their power and their longevity in power, I expect their structural environment to be more dependent on them. If a regime is the only game in town for as long as the pri in Mexico or Golkar in Indonesia, one may well consider the town to be endogenous to the regime. The structural environment these regimes face is the wealth they create, the growth rates they achieve, the inequality they generate, the international linkages they permit or encourage. Some of the most successful developmental dictatorships of the 20th century have been electoral autocracies: post-revolutionary Mexico, New Order Indonesia, Taiwan under the kmt, Malaysia under the National Front, Singapore under the pap. To the extent that (a) determinate favorable structural contexts help hegemonic regimes reproduce their electoral dominance and (b) these regimes succeed in creating their own favorable conditions, we should observe a close association between contextual variables and electoral outcomes under conditions of hegemonic rule. Economic development: Since the 1950s, with some ups and downs and shifts in depth and terminology, debates about the structural sources of democratization have been in essence debates about the impact of socio-economic “modernization” on the establishment and consolidation of democratic regimes. For pragmatic reasons of data availability, much of the literature has studied national levels of economic development, measured in

Struggle over Electoral Uncertainty

243

monetary income per capita, as proxy for societal modernization. The underlying sociological intuition is simple: To the extent that monetary wealth is reflective of a modern economy, it entails a bundle of factors that should make it harder to keep a society under the lid of authoritarian control: a more egalitarian and mobile society, more educated and resourceful citizens, higher levels of societal differentiation. The empirical findings offered by Przeworski and co-authors (2000) still represent the basic state of the art: Economic development contributes to consolidate democratic regimes, yet not to create them. While regime stability is the fruit of economic wealth, regime change is the work of actors; from a structural point of view, its occurrence is random. I will discuss transitions from electoral authoritarian regimes in the final chapter of this book. Here, I am asking more narrowly whether socio-economic development tends to affect, not the dynamics of regime change, but the internal dynamics of electoral autocracies, in particular the balance of force between government and opposition in the electoral arena. In line with sociological modernization theory, I expect economic wealth to change the balance of power in favor of society, that is, in favor of opposition parties striving to mobilize voters against authoritarian rulers. My statistical results (reported in Table 7.6) strongly support this expectation—yet only for competitive regimes, in which average wealth (the logarithm of current per capita income in Purchasing Power Parities) shows a strong negative relation with margins of victory. To paraphrase Seymour Martin Lipset’s famous dictum: The more well-to-do a competitive regime, the lesser the chances that it will sustain high margins of victory. In hegemonic regimes, by contrast, wealth works the way it works under democracy: stabilizing the status quo. The more well-to-do a hegemonic regime, the greater its chances of sustaining dizzying margins of victory. If poverty is a source of electoral security in competitive regimes, in hegemonic regimes it is wealth. The former are safe under conditions of scarcity, the latter under conditions of affluence.17 Economic performance: Students of voting behavior in established democracies have known for long that economic considerations may constitute powerful motives for electoral choices. This is true for general macroeconomic evaluations (“socio-tropic voting”) as well as for particularistic considerations of personal well-being (“pocketbook voting”). Under non-democratic conditions, substantive policy issues like economic performance compete against polity issues like authoritarian governance. Still despite possible countervailing considerations, voters are likely to take macroeconomic conditions into account when forming their electoral preferences.

244

Empirical Explorations

In the absence of individual-level data on economic perceptions, I use a standard indicator of macroeconomic conditions to assess the impact of economic junctures on authoritarian electoral outcomes: annual changes in per capita income (average values for the five years preceding the election year). Again, regression results reported in Table 7.6 are nothing less than startling. Hegemonic regimes behave as expected. Just as they benefited from average wealth, they benefit from medium-term growth. Competitive regimes, by contrast, move counter the clock of economic conditions. Just as they were harmed by affluence, they seem hurt by growth. Stranger than paradise. Rather than bolster the incumbent, past economic growth rates seem to embolden the opposition. Competitive regimes, it seems, suffer from macroeconomic success. Societal cleavages: The presence of “objective” societal cleavages does not indicate the actual presence of political conflicts, only their potential emergence. Societal cleavages do not translate mechanically into political contention. Only if successfully politicized by societal and political actors do they enter the political agenda and structure political conflict. For authoritarian governance, societal cleavages bear ambiguous implications. They provide grievances that allow opposition actors to mobilize against the regime. But they also provide lines of conflict governments may use to erode the appeal of democratizing mobilization. Cross-cutting cleavages may serve to undermine governmental as well as opposition claims of representing citizens’ best interests. To trace the impact of social cleavages on authoritarian electoral politics, I focus on one prominent structural variable: income inequality (Gini index). According to the burgeoning literature on social inequality and political regimes, authoritarian regimes are systems of minority rule that favor the rich. If levels of inequality are low, dictators can yield to democratic demands. If they are high, they will defend the status quo against the disempowered and dispossessed majority that fights for majoritarian rule.18 Existing theories do not offer predictions about how variations in levels of inequality affect the internal dynamics of authoritarian regimes. In principle, we should expect inequality to erode the electoral support of incumbents and aid the cause of opposition parties. Yet, again, my data frustrate this simple hypothesis. While social inequality seems too distant a factor to touch election outcomes in competitive regimes, in hegemonic regimes it seems to aid the electoral dominance of incumbents rather than debilitating it (see Table 7.6). Apparently, the production of social inequality is a rational strategy for hegemonic regimes. When these regimes produce high levels of social inequality, they also benefit from them. Post-revolutionary Mexico is a fitting example:

Struggle over Electoral Uncertainty

245

during its decades of slow decline, the ruling pri received disproportionate electoral support from the stratus of the poor it had produced itself (see also Magaloni 2006). External dependence: Although the primary focus in the comparative study of democratization has been domestic, a growing number of studies have been emphasizing the role of international actors and factors. Scholars have turned their attention, for instance, to the political consequences of international integration and dependence (Levitsky and Way 2010b), the diffusion of democracy (Brinks and Coppedge 2006) and democratizing protest (Weyland 2009), the learning of both opposition actors and governments from international experience (Bunce and Wolchik 2011), the rise of international election observation and democracy promotion (Hyde 2011; Kelley 2012), and the liberating potential of new information and communication technologies (Diamond 2010). Just as the translation of societal cleavages into political conflict is a complex process of uncertain results, so is the translation of international pressures into domestic balances of power. Causal effects from international sources have a long distance to travel to reach their final destination. No wonder they get sidetracked at times. No wonder if such a thin and one-dimensional indicator of international dependence as international trade (as percentage of gdp) fails to leave discernible traces in the electoral arena. International trade, one might expect, should render electoral authoritarian rulers more concerned about their international reputation and thus more embarrassed if they win elections by obscene margins of victory. The motivating force of such embarrassments seems to be low, however. If anything, international trade relations upholster authoritarian margins of victory. The (non-significant) coefficient signs for both hegemonic and competitive regimes are positive (see Table 7.6). If we regress margins of victory on all four structural variables (as reported in the multivariate regression in Table 7.6), we obtain some notion of their overall relevance, even while some of the uncontrolled individual effects get blurred. As expected, structural contexts covary more closely with electoral outcomes in hegemonic regimes, where they account for about one-fifth of variation in competitiveness (R2 = .206), than in competitive regimes, where they explain less than one-tenth (R2 = .83). In sum: Contextual variables lie at a considerable distance from the political arena. Their political implications are often ambiguous, as they strengthen or weaken all parties in conflict in a symmetrical fashion. The fact that we found strong patterns of covariation between isolated features of the structural environment and electoral outcomes is almost surprising. It strongly speaks of the embedded nature of authoritarian elections.

246

Empirical Explorations EXPLAINING QUALITATIVE DEFEAT

Up to now, I have assumed that authoritarian incumbents simply prefer higher margins of victory over lower ones. However, the continuous measure of margins of victory does not reflect well the non-continuous utility of electoral results. Rulers may delimit zones of indifference, where they care little about variations in their margins of victory, and zones of hypersensitivity, within which they are allergic to even small reductions in their margins. Caring about the exercise of power, ruling parties should be allergic to the loss of legislative supermajorities that allow them to control the game of constitution-making without depending on the cooperation of others. Caring about the conservation of power, they should be even more allergic to the loss of executive power (regardless of its implications for the nature of the regime). Constitutional majorities and alternation in government constitute critical thresholds that incumbents will defend as fiercely as their opponents will attack them.

The Distribution of Supermajorities Hegemonic regimes control constitutional supermajorities by definition. My dataset includes only three cases in which hegemonic regimes suffered transitory “electoral accidents” and lost temporary control of legislative supermajorities (≥ two-thirds of lower chamber seats): Gabon in 1990, Guinea in 1995, and Togo in 1994. In all cases, incumbents were quick to recover their qualified legislative majorities in subsequent elections. While almost all hegemonic parties thus hold qualified legislative majorities (93.8 percent), only a minority of competitive regimes formally controls the levers of constitutional change (19.1 percent) (see Figure 7.4).

The Occurrence of Alternation For all their structural reliance on ruling parties, electoral authoritarian regimes often show personalistic traits and become deeply identified with the person controlling the chief executive office. Examples are Indonesia under Suharto, Egypt under Mubarak, Gabon under Bongo, and Paraguay under Stroessner. The more personalistic a regime, the more important it is for its beneficiaries to guarantee the continuity in power of the head of state. Some electoral autocracies with personalistic traits have carried out successful leadership

Struggle over Electoral Uncertainty 100.0%

Competitive regimes Hegemonic regimes

100.0%

80.0%

80.0%

60.0%

60.0% 100.0%

93.8% 40.0%

80.8%

40.0%

80.9%

20.0%

20.0% 19.2% 0.0%

247

6.3% No supermajority

Legislative supermajority

19.1% 0.0%

Continuity of ruling party

Alternation in power

Figure 7.4. Frequency of constitutional majorities and party alternation, by regime subtype

successions without putting the regime in jeopardy. Examples are the handovers of power from Lee Kuan Yew to Goh Chok Tong in Singapore, from Benjamin Mkapa to Jakaya Mrisho Kikwete in Tanzania, and from Heydar Aliyev to his son Ilham Aliyev in Azerbaijan. Yet only a few regimes have managed to institutionalize the regular rotation of government personnel while keeping the ruling party in power. Post-revolutionary Mexico’s clockwise succession of presidents after six-year terms in office has been unique in the history of non-democratic rule.19 While changes in the person occupying the chief executive position are often compatible with regime continuity, changes in the party controlling the chief executive are usually a reliable sign of regime change. Such party alternation in power may either crown previous democratizing progress (“democratic revelations”) or result from contentious elections in which incumbents are evicted by mass mobilization (“democratic revolutions”). In a third often overlooked scenario, party alternation takes place without disrupting electoral authoritarian continuity (“authoritarian reshuffling”). Some “color revolutions” in former Soviet republics ended in reaffirmations of authoritarian rule, as in Georgia and Kyrgyzstan (see Chapter 6, pp. 185–186). Within my sample of elections, I register several instances of electoral alternation in government. Some of them led to democracy, others to the breakdown of the electoral regime, and still others to electoral authoritarian continuity. In all cases, I limit my attention to “electoral” alternations in which official election outcomes designate a winning party distinct from the incumbent party. I thus exclude irregular removals from power that precede or succeed an election, such as military coups or electoral rebellions.20

248

Empirical Explorations

Hegemonic regimes rarely suffer the loss of legislative supermajorities and they never lose executive power through elections. At least not within my case sample. First they lose their hegemony, then they lose power. I count three cases of personal alternation in power in hegemonic regimes (Mexico 1982, Tanzania 1995, and Singapore 1991), yet none of partisan alternation. In competitive regimes, by contrast, I register 27 instances of personal alternation in power and remarkable 18 cases of partisan alternation of power. The former affect 28.7 percent, the latter 19.1 percent of elections (N = 94). There are no cases of party alternation without personal alternation, but eleven instances of personal succession without party alternation. Figure 7.4 above shows the frequency distribution of party alternation across both subtypes of regimes. Table 7.7 lists all cases as well as the corresponding regime outcomes: nine cases of regime continuity (50 percent), seven of subsequent democratization (38.9 percent), and two of breakdown (11.1 percent).21 Table 7.7. Authoritarian elections leading to party alternation in power Countries

Year

Type of election

Cambodia Belarus Pakistan Turkey El Salvador Guatemala Colombia Sri Lanka Georgia Albania Turkey Algeria Pakistan Pakistan Niger Albania Armenia Guatemala N %

1993 1994 1990 1983 1984 1985 2002 1994 1992 1997 1999 1999 1997 1993 1999 1992 1998 1995

L P L L P C C C L L L P L L C L P C 18 100

1st Civil war authoritarian election • • • • • • • • •

9 50.0

• • • • • • • •

8 44.4

Note: P = presidential, L = legislative, and C = concurrent election.

Military tutelage

• • • •

• • • • •

9 50.0

Subsequent fate of regime Breakdown ea continuity ea continuity Democratization ea continuity Democratization Democratization Democratization ea continuity Democratization Democratization ea continuity Breakdown ea continuity ea continuity ea continuity ea continuity Democratization

Struggle over Electoral Uncertainty

249

Alternation is an enigmatic event in the life of an authoritarian regime. It is not supposed to happen. And yet it does. It loses much of its mystery, though, if we consider that electoral authoritarian governments do not always control electoral outcomes. They may fail to control them because they are too weak or inexperienced. This can happen in first elections in which incumbents either do not take part (as in Turkey in 1983) or overestimate their electoral support and lose out to authoritarian competitors (as in Belarus in 1994). Governments may also fail to control election outcomes because others do. This tends to be the case in regimes in which the subversion of free and fair elections originates at least in part from actors outside the electoral arena: armed rebels waging civil war or the military high command subjecting the government to tutelary controls.22 In situations of either civil war or military tutelage, authoritarian actors may permit alternation in government because they care little about who controls the government. These three possibilities are not mutually exclusive. As Table 7.7 indicates, in all except three instances of electoral alternation, at least one of these three conditions was present. Among the exceptions, the Albanian 1992 election could be counted as a first election (although technically it wasn’t), while Armenia in 1998 and Guatemala in 1995 approximated conditions of civil war. The question is: Are these broad contextual configurations the whole story? Or are actors capable of changing the probabilities of regime defeat through their meta-game strategies?

The Secrets of Defeat How effective is authoritarian manipulation in preserving constitutional majorities and preventing the loss of executive power? How effective is opposition protest in eroding qualitative majorities and promoting the alternation in power? To establish the isolated impact that actors’ meta-game strategies bear on this pair of qualitative election outcomes, I regress both on the same strategic and contextual variables as before. Since both dependent variables are categorical, I use binary logistic regression. Since hegemonic regimes by definition control constitutional majorities and in fact never experience alternation, I circumscribe the analysis to competitive regimes. When reporting logistic regression results, I always indicate odds ratios eb rather than regression coefficients. Odds ratios indicate the extent to which the odds of occurrence of the outcome variable change if the independent variable undergoes a one-unit change. Close to one they tell us that nothing changes, below one they indicate decreases, above one increases in the dependent variable. The baseline odds of constitutional majorities in competitive

250

Empirical Explorations

regimes are 0.23 (14 cases of supermajorities divided by 59 without constitutional majority) and the odds of alternation 0.23 (18 cases of alternation divided by 76 of continuity). Manipulation. If fraud appeared as the golden route for competitive regimes to augment their margins of victory, it also seems to be their silver bullet for obtaining constitutional majorities and preventing the alternation in power. According to bivariate regression results included in Table 7.1, each step along the 3-point scale from integrity to fraud increases the odds of constitutional majorities by almost two-thirds (eb = 1.64) and decreases the odds of electoral alternation by almost two-thirds (eb = .37). Multiple regression results with all four strategies of manipulation confirm the privileged role for fraud in preserving power (although its contribution to the control of constitutional majorities turns statistically insignificant). Albeit not statistically significant and thus not generalizable either, the impact of censorship looks remarkable. Within the sample, it doubles the odds of constitutional majorities (eb = 2.1) (see bottom of Table 7.1). For the purpose of illustration, Figure 7.5 charts predicted probabilities calculated from these multivariate regression results.23 The upper two graphs show how outcome probabilities change if governments move from electoral integrity to fraud when one other strategy of manipulation is kept at its maximum, all others at their minimum. They primarily depict the effects of vote rigging and its interplay with censorship. With all authoritarian strategies set at their minimum, the probability that competitive regimes obtain constitutional majorities is close to zero (p = .076). If they introduce electoral fraud, this probability augments to .16 and if they recur to simultaneous full media restrictions, it more than doubles to .36. The mutual reinforcement of these two strategies is again evident with respect to alternation. With all strategies set at their minimum, the probability of alternation in competitive regimes is .31. With censorship alone, this probability descends to .19, with fraud alone to .07. With both at their maximum, it approximates zero (p = .037). In sum, without fraud, competitive regimes face low chances of conquering constitutional majorities and substantial chances of losing power. With fraud, they face a reasonable chance of conquering constitutional majorities and almost no risk of losing power through elections. Contention. Just as protests from below and dissidence from within seemed to act as effective depressants of margins of victory, they also seem to shape the qualitative outcomes of competitive authoritarian elections. As bivariate logistic regression results suggest (see Table 7.3), any instance of armed rebellion reduces the odds of constitutional majorities by one-third (eb = .64), any instance of horizontal threat by more than one-half (eb = .46). Horizontal threats, in addition, appear as powerful triggers of alternation. Each lateral

1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 0.00

Max repression

1.00 Electoral fraud

Exclusion

2.00

1.00

1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 0.00

5 Armed rebellions Pre-electoral protest

2.00 3.00 4.00 Horizontal threats

Max repression

5.00

1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 0.00

Max censorship

1.00 Electoral fraud 10 Armed rebellions 20 Vertical threats

Predicted probability off alternation

Predicted probability of constitutional majo a rities

Full boycott 10 Vertical threats 1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 0.00

Max censorship

Predicted probability off alternation

Predicted probability of constitutional majo a rities

Exclusion

1.00

2.00

Full boycott Pre-electoral protest

2.00 3.00 Horizontal threats

4.00

Figure 7.5. Effects of meta-game strategies on predicted probabilities of constitutional majorities and party alternation in competitive regimes Note: Predicted probabilities calculated with X-Post (Cheng and Long 2000) from multivariate logistic regressions presented in Tables 7.1 (for fraud) and 7.3 (for horizontal threats).

5.00

252

Empirical Explorations

conflict raises the odds of alternation by 42 percent (eb = 1.42). The effects of boycott are predictable. Full opposition boycotts are no necessary condition, but almost a sufficient condition, for the incumbent to achieve a constitutional majority and prevent electoral alternation. If the opposition moves by two units, from full participation to full boycott, the odds of constitutional majorities multiply by ten (eb = 4.93) and the odds of alternation fall to one-quarter (eb = .46). The multiple regressions at the bottom of Table 7.3 blur the effects of some variables and highlight the effects of others. All contentious events covary with qualitative election outcomes as expected (even if not all relations are significant). Only the figures for pre-electoral protest are counterintuitive. It is unclear why opposition mobilization against an upcoming election would multiply the odds of constitutional majorities by seven (eb = 6.99). Overall, the joint explanatory power of contentious action looks high, in particular for the control of constitutional majorities (R2 = 31). Figure 7.5 illustrates the powerful facilitating role that opposition boycotts play for the control of constitutional majorities by competitive regimes, yet also the powerful counteracting force of armed rebellions and horizontal threats. If all major opposition parties boycott in the absence of threats, the probability that incumbents obtain a qualified legislative majority lies at .68. If they boycott in the presence of five instances of armed rebellion in the three years preceding the election year, this probability descends to .28 and in the presence of five horizontal threats to .24. When the opposition enters the electoral arena, armed rebellions and horizontal threats practically eliminate the probability of qualified majorities (p = .007 and .005 respectively). Figure 7.5 also shows the boost that horizontal threats give to the probability of alternation, in particular in association with other types of threat (although potentially neutralized by boycott). When all expressions of dissidence are held at their maximum, the probability of alternation is very high (p = .9). At their minimum, it is very low (p = .07). The mere occurrence of five horizontal threats elevates the probability of alternation to .24. Boycotts halve it (p = .11), the presence of twenty vertical threats doubles it (p = .47). Institutions. If the design and strength of political institutions serves to compress margins of victory, it should also serve to improve the chances of conquering constitutional majorities and preserving executive power. Multiple regression results in Table 7.5 indicate that institutional choices matter more for the former goal than for the latter. A hypothetical move from presidentialism to semi-presidentialism, or from there to parliamentarism, multiplies the odds of qualitative majorities by four (eb = 3.96), a move from majoritarian to mixed-member systems, or from there to proportional representation, by six (eb = 5.92). Intriguingly, however, both parliamentarism and legislative

Struggle over Electoral Uncertainty

253

ineffectiveness seem to erode the authoritarian grip on power. Though slightly below conventional thresholds of statistical significance, both increase the odds of alternation. Again we illustrate these patterns of association by plotting predicted probabilities (see Figure 7.6). With all other variables set at zero, the combination of presidentialism and proportional election renders the obtainment of constitutional majorities almost impossible (p = .005). The opposite combination of parliamentarism and majoritarianism makes it very likely (p = .715). Somewhat ironically, parliamentarism increases the chances of alternation in power as well. Competitive regimes that establish the unusual combination of presidentialism with reasonably strong legislatures face a modest probability of alternation (p = .14), those that establish the self-contradictory combination of parliamentarism with rubber-stamp parliaments have an even chance of losing power through elections (p = .49). Societal structures. If average wealth seemed to inflict severe damage to the competitive standing of competitive regimes, it seems equally damaging to their prospects of achieving constitutional majorities. As the multivariate logistic regression results reported in Table 7.6 indicate, wealth flattens the odds of constitutional majorities almost to zero (eb = .007)—while it leaves the odds of alternation untouched. By contrast, the unequal distribution of wealth as well as the international integration of the economy seem to strengthen the position of authoritarian rulers. A hypothetical ten-point increase of the Gini index reduces the odds of alternation by 69 percent (eb = .93). A hypothetical expansion of international trade by 10 percent of gdp reduces it by 32 percent (eb = .97). Apparently, rather than sources of instability, both structural variables constitute sources of authoritarian stability. The predicted probabilities in Figure 7.6 illustrate the depressive impact wealth carries on the probability of constitutional majorities, and inequality on the probability of alternation. The impact of wealth is overpowering. It takes place regardless of the values that the other structural variables assume. The impact of inequality is more conditional. If preceded by high negative (!) growth rates, high inequality (Gini = 58) brings the probability of alternation close to zero (p = .09). If preceded by high positive growth rates, the chances of alternation remain non-trivial (p = .23). The impact of economic crises on regime stability has been the object of intense comparative debate. Adam Przeworski and his co-authors found that “most deaths of democracy are accompanied by some economic crises,” while dictatorships “die under all kinds of economic conditions” (2000: 109 and 111). They inferred that “economic circumstances have little to do with the deaths of dictatorships” (2000: 117). Perhaps, though, competitive authoritarian regimes tick differently. Perhaps, economic fair weather conditions (in which change looks

1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 0.00

1.00 Majo a ritarian electoral rules

Majo a ritarianism Ineffective legislature a

2.00

1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00

0.00

Positive growth Max trade

3.00 3.20 3.40 3.60 Wealth (log current PPP per capit a a)

3.80

1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 28.00

Proportional representation Repeated elections

1.00 Parliamentarism Negative growth Affluence

Predicted probability off alternation

Predicted probability of constitutional majo a rities

Negative growth Max inequality 1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 2.80

Presidentialism Repeated elections

Predicted probability off alternation

Predicted probability of constitutional majo a rities

Parliamentarism Ineffective legislature a

33.00

2.00

Positive growth Min trade

38.00 43.00 48.00 Social inequality

53.00

58.00

Figure 7.6. Effects of institutional and societal contexts on predicted probabilities of constitutional majorities and party alternation in competitive regimes Note: Predicted probabilities calculated with X-Post (Cheng and Long 2000) from multivariate logistic regressions presented in Tables 7.5 (for institutions) and 7.6 (for social structures). Extreme values: Poverty / affluence in competitive regimes = 570 ppp p.c. (log = 2.76) / 8,080 ppp p.c. (log = 3.91). Negative / positive growth = average decline / increase of 10 percent in gdp per capita over the 5 years preceding the election year. Minimum / maximum inequality = Gini 28.2 / 59.1. Minimum / maximum trade = 24.9 percent / 166 percent of gdp.

Struggle over Electoral Uncertainty

255

less menacing for citizens as well as governments) are more propitious than times of disasters for subverting competitive authoritarian regimes through elections.24

C ON C LU S I O N Despite the complexity of the real world, despite the variety of sources that determine the uncertainty of authoritarian elections, we found some simple lineal relationships between meta-game strategies of manipulation and protest on the one hand and electoral competitiveness on the other. Through their strategic choices, actors shape the balance of power within the authoritarian electoral arena in direct and immediate ways. While hegemonic regimes live on their reputation of invincibility, they possess clear sources of fragility. Their vertiginous margins of victory depend to a surprising degree on political exclusion and are vulnerable to a surprising degree to active protest against upcoming elections. Sensitive to manifestations of popular discontent in between elections, hegemonic parties are even more sensitive to manifestations of elite disunity. Emergent fissures in popular acquiescence and elite unity seem to send shock waves to the cognitive foundations of hegemonic equilibria: the generalized assumption of popular support. While highly institutionalized by definition, hegemonic regimes benefit from further strengthening their institutional framework. Their power and duration allow them to shape their societal environment. In part, their electoral dominance derives from their ability to create the structural conditions that support electoral dominance: wealth, growth, and social inequality. Competitive regimes are neither invincible nor believed to be so. In their perpetual struggle to contain electoral uncertainties, fraud and censorship seem to constitute their most effective weapons. Both serve to inflate margins of victory, ensure the control of legislative supermajorities, and avoid alternation in power. With similar consistency, manifest dissidence from either above or below serves to undercut these goals. Vertical threats, armed rebellions, and horizontal threats all tend to narrow margins of victory, render the conquest of constitutional majorities more difficult, and increase the probability of alternation. When opposition parties mobilize against upcoming elections, the short-term consequences are contradictory. When they withdraw, they are clear: authoritarian incumbents reap large immediate benefits from opposition boycotts. While sensitive to the design of institutions, competitive election outcomes seem even more sensitive to their strength. Above all, weak

256

Empirical Explorations

legislatures strengthen the incumbent. While actors matter for electoral competitiveness, societal structures do so as well, although in unexpected ways: the poorer a competitive regime is and the poorer it gets, the more tranquil its hold on power.

Notes 1. On the distinction between x-centered and y-centered analyses, see Gerring (2001: 137). 2. sd = Standard Deviation. Here as well as in most of my subsequent empirical analysis, I segment my sample of elections by subtypes of regime, yet merge data for legislative and presidential elections in order to prevent case numbers from dropping too much. Unless I distinguish explicitly between electoral arenas, my data on manipulation, protest, and competitiveness therefore refer to either legislative or presidential elections. In concurrent elections, these data seldom diverge substantially. Nevertheless I need rules of arbitration in case they do. To construct my figures on electoral competitiveness, I choose presidential election data in all cases of concurrent elections. Whenever my measures of fraud, exclusion, boycott, and protest differ for concurrent legislative and presidential elections, I choose the higher value. 3. In equilibria of resignation, problems of collective action deepen the paralysis of individuals. I bracket issues of opposition coordination here to discuss them in Chapter 9. 4. For definitions of authoritarian strategies, coding rules, and frequency distributions, see Chapter 6. 5. On the severity of manipulative strategies, see Chapter 8, where I also return to discuss the dilemmas of electoral manipulation. 6. Concretely, I checked for effects baseline threats carry on the selection of manipulative strategies by controlling for preceding electoral, vertical, and horizontal threats. I checked for effects prevailing conditions of effectiveness carry on the selection of manipulative strategies by regressing manipulative choices on levels of state capacity (measured by the share of tax revenue in gdp) and societal strength (measured by the rate of illiteracy among the population over age 15). To test whether the consequences of manipulative measures are contingent on their structural conditions of success, I also introduced interactive terms between manipulation and these facilitating conditions into ols regressions (with margins of victory as their dependent variable). Surprisingly, none of these analyses yielded any discernible (comprehensible) traces of selection effects. The corresponding results, together with a discussion of the differential dependence of manipulative strategies on state and society, are available at request from the author. 7. For definitions, coding rules, and frequency distributions, see Chapter 6.

Struggle over Electoral Uncertainty

257

8. In competitive regimes, though, changes in pre-electoral protest covary consistently with changes in margins of victory. When protest abates, margins rise (+7.9 percent), when nothing changes, they slightly descend (–3.9 percent), when protest ignites, they fall more steeply (–10.8 percent). Due to low case numbers (N = 31), these differences in mean changes are not statistically significant (p = .52). 9. “Die Menschen machen ihre eigene Geschichte, aber sie machen sie nicht aus freien Stücken, nicht unter selbstgewählten, sondern unter unmittelbar vorgefundenen, gegebenen und überlieferten Umständen” (115). “Men make their own history, but they do not make it as they please; they do not make it under self-selected circumstances, but under circumstances existing already, given and transmitted from the past” () (accessed May 7, 2010). 10. Data on forms of government are from Gerring, Thacker, and Moreno (2005). They are trichotomous, distinguishing between parliamentary, presidential, and semi-presidential systems. Since the latter two show similar levels of competitiveness, I merge them for the present purpose. 11. Again, data on electoral systems are from Gerring, Thacker, and Moreno (2005). The original measure is trichotomous, distinguishing between proportional representation, mixed-member majority or bloc vote, and majoritarian or preferential vote. Gerring and his co-authors sort electoral systems into mutually exclusive categories. The World Bank Database on Political Institutions, by contrast, only indicates whether electoral systems contain some element of proportional representation (variable pr) or whether pr rule governs the majority of Lower House seats (variable housesys). These coding rules provide “no consistent way to discern mixed systems” (Teorell and Lindstedt 2010: 439). Other comparative data sources on electoral rules either do not cover non-democratic systems, or capture electoral rules that seem of little utility for purposes of manipulation. Notably, no usable dataset on electoral rules in presidential elections exists to date. For a theoretically guided review of available cross-national data on electoral system properties, see Teorell and Lindstedt (2010). 12. The locus classicus in the empirical study of election cycle effects on party systems under democratic conditions is Shugart and Carey (1992). 13. The literature on subnational elections in authoritarian regimes is particularly rich with respect to large federal countries like Mexico (see, among many others, Beer 2003, Eisenstadt 2004, and Mizrahi 2002). On the rediscovery of international influences on democratization, see also p. 245 below (“External Dependence”). 14. I classify former single parties as “hegemonic” when they keep dominating the electoral arena after transiting to multiparty competition. Most first hegemonic elections in my dataset derive from such instant mutations from long-lived single-party to hegemonic-party rule (see Appendix a, p. 397). 15. To capture the strength of national legislatures, I use the data on legislative effectiveness from the Arthur Banks cnts Data Archive (variable s22F4). For

258

16.

17.

18. 19.

20. 21.

22. 23.

24.

Empirical Explorations the present purpose, I invert and dichotomize the measure. Scores of 0 indicate effective or partially effective legislatures, scores of 1 ineffective “rubber-stamp” legislatures or inexistent legislatures. Within my sample, cnts registers seven “inexistent” legislatures. These are cases in which first legislative elections are held late in the calendar year or in which presidential contests inaugurate an electoral regime, with legislative elections following later. The quote refers to one humble title Joseph Stalin claimed for himself (see , accessed March 6, 2011). For a conceptual discussion of personal rule, see Guliyev (2001). Lipset’s original proposition on democracy and development: “The more well-to-do a nation, the greater the chances that it will sustain democracy” (Lipset 1981: 31). See the seminal texts by Boix (2003) and Acemoglu and Robinson (2006). Some personalist regimes, as the Dominican Republic under Balaguer, simulate alternation in power. The strong man puts a straw man in the presidential office and keeps pulling the strings of power behind a facade of formal rotation in power. The puppeteer moving his puppets. Some electoral autocracies have innovated the techniques of officially denying yet factually reproducing personal power. The dictator moves from one formal position of power to another, and real power follows him. State power, rather than being attached to formal offices, gets attached to persons. Examples are Serbia under Milošević and Russia under Putin. The only two cases of electoral rebellion in the dataset are the Philippines in 1986 and Côte d’Ivoire in 2000. For brevity, I will subsequently refer to party alternation by elections as “alternation.” As explained above, in these cases margins of victory turn into margins of defeat and show negative signs (see Chapter 6). Cases of “civil war” show levels of societal warfare of 2 or higher (State Failure Dataset, for variable description see Appendix b, p. 411). I calculate all predicted probabilities with “X-Post: Excel Workbooks for the Post-estimation Interpretation of Regression Models for Categorical Dependent Variables” developed by Simon Cheng and J. Scott Long (Bloomington, Indiana University, 2000) (). On the impact of wealth and growth on electoral authoritarian regime change, see Chapter 10.

8 The Calculus of Electoral Manipulation Common sense tells us that electoral manipulation is a purposeful enterprise. Rulers employ it to keep electoral threats at bay. In hegemonic regimes, they strive to prevent the emergence of uncertainty, in competitive regimes to contain its spread. Common sense also tells us that manipulation may fail. It may fail in the short term when state agents or citizens refuse to play their part. It may also fail in the longer term, when it brings citizens or state agents to oppose a manipulative government they perceive as illegitimate. Perhaps the primary purpose of authoritarian elections is not to create regime legitimacy. Yet, destroying it may well be their ultimate consequence. In this chapter, I intend to capture the dilemmatic potential of electoral manipulation by conceiving it as a strategy of circular threat management. Its twin challenge lies in containing present threats to the regime (“first-order uncertainties”) without provoking future troubles (“second-order uncertainties”). In their search for “optimal” levels of manipulation, rulers have to consider both: their capacity to control present threats as well as their propensity to provoke future threats. The former depend on their capacities of institutional control, the latter on their capacities of information control. My empirical explorations differ from extant studies in their empirical scope and their conceptual precision. In terms of scope, I strive to explain manipulation across subtypes of electoral authoritarian regimes, rather than across all types of political regimes. Conceptually, I choose a rather high level of disaggregation of my dependent as well as independent variables. On the dependent side, I strive to explain a bundle of four distinct authoritarian strategies, rather than just one. On the independent side, I distinguish between electoral and non-electoral and between violent and peaceful threats. In my explorative incursions into the relation between threats and manipulation, I assume that survival in power has the same utility to all authoritarian rulers. It is only towards the end of the chapter that I shed this assumption and introduce variations in the utility of power as possible sources of variations in the intensity of manipulation.

260

Empirical Explorations THE UTILITY OF MANIPULATION

According to standard decision-theoretic assumptions, actors compute the expected utility of available courses of action (across all possible outcomes) and choose the one that promises the highest net benefits.1 In their forward-looking calculations, they form probabilistic expectations about possible futures, yet never know for sure what the future brings once they have taken their choice. The utility μ of alternative futures is not a quantity known with certainty, but an estimation of risks and opportunities. It is the product of their net utility u times the probability p of their occurrence. As decisions may produce a wide range of possible outcomes, their expected utility represents the sum of expected utilities across all these possible outcomes. Thus, if electoral manipulation M strives to conserve the status quo, its expected utility μ is the sum of the expected utility of its alternative outcomes (at some future point in time t+1): the continuity of the status quo (sq) and regime change (sq’). μ (M ) p (SQ ) u (SQ )

(1-p) (SQ ) u (SQ )

(8.1)

Institutional certainties have a simple mathematical meaning: the (perceived) probability of continuity of the authoritarian status quo equals one [p(sq) = 1]. Threats to regime survival imply that this probability is lower than one [p(sq) ≤ 1]. For most of this chapter, I will pay exclusive attention to this probabilistic side of the manipulative calculus. In doing so, I will assume that holding and losing power has the same value for all rulers [u(sq) = 1]. Only in the last section of the chapter will I turn to explore the utilitarian side of the manipulative calculus, the stakes of the political game. In doing so, I will assume that all rulers face similar levels of threat to their political survival.

Authoritarian Rationality and Its Limits Since the world is home to rationality as well as madness, we should be cautious to claim that political repression is always rational in an instrumental sense. The bureaucratic mass destruction of human life carried out by the totalitarian regimes of the 20th century did not serve either the purpose of political survival or the purpose of political governance. It was only after the regimes of Hitler and Stalin had extinguished any genuine challenges to their rule that they built up their bureaucracies of mass murder to full scale, and both the Holocaust and the Gulag were counterproductive for the war efforts of Germany and the Soviet Union, respectively.2 And yet. As social scientists

The Calculus of Electoral Manipulation

261

we can hardly proceed otherwise than assuming that authoritarian strategies are rational. That in general they are useful instruments of authoritarian self-aggrandizement and self-preservation. The exceptions of totalitarian ideology prove the rule of authoritarian rationality. As philosopher Ludwig Wittgenstein famously formulated, what we cannot speak about we must pass over in silence.3 In the social sciences, what we cannot speak about in rational terms we must pass over in silence—and refer to the psychology department. What we cannot understand, in terms of either wider or narrower conceptions of rationality, we cannot explain.4 Accordingly, comparative scholars tend to conceive authoritarian strategies as rational instruments of governmental self-preservation. They understand violations of human rights, civil liberties, and political rights as the means that authoritarian governments choose to deflect perceived threats either to their ambitions in power or their continuity in power. Just as the old political sociology of regimes tended to be hyper-functionalist and discount the possibility of harmful structures, the new political economy of regimes tends to be hyper-rationalist and discount the possibility of harmful strategies.5 Yet, I would contend, even while we embrace the assumption of authoritarian instrumental rationality, we should recognize its limits. Authoritarian strategies of self-defense may be self-defeating. They are costly, they are not always effective, and they may even generate counterproductive consequences.

Authoritarian Benefits: Institutional Containment Electoral manipulation is the exercise of power over access routes to power for the sake of conserving power. Almost by definition, we should expect its core purpose to lie in the containment of institutional threats to the political survival of authoritarian rulers—in the first instance, in the containment of electoral uncertainties. The simple notion that authoritarian rulers manipulate elections to win them, be it by thin or overwhelming margins, seems to be a commonsensical proposition. Some authors have described manipulation as an art form (see Case 2006, Riker 1986), yet no one seems to think that rulers perform it for its own sake, as authoritarian l’art pour l’art. In devising their strategies of electoral manipulation, competitive and hegemonic regimes face contrasting challenges. The former need to control disquieting electoral uncertainties, the latter to conserve comforting electoral certainties. Competitive regimes live under the continual threat of losing power through elections. They have to suppress electoral challenges to preserve power in the short run. Hegemonic regimes, by contrast, live in a secure world of electoral invulnerability. They have to prevent the emergence of electoral

262

Empirical Explorations

challenges to keep enjoying power over the long run. Electoral manipulation should serve both purposes: the containment of short-term electoral risks by competitive regimes as well as the prevention of long-term electoral risks by hegemonic regimes. In both subtypes of electoral authoritarianism, the preservation of power appears as the raison d’être of electoral manipulation. Again, my empirical explorations shall focus on four specific strategies of manipulation: repression, media restrictions, exclusion, and electoral fraud. Scholarly discussions of authoritarian strategies commonly proceed on either more specific or more aggregate levels. Authors either single out one specific authoritarian strategy for attention or bundle various strategies under broad conceptions of authoritarian intervention.6 In general, though, the literature rests upon the idea that governments implement these strategies to prolong their political lifespans. x The survival benefits of repression. The broad, largely quantitatively oriented literature on state repression has traced covariates of violations of physical integrity rights by state authorities.7 To explain variations in levels of political repression across regimes, most authors have adopted a rough decision-theoretic framework according to which central state authorities choose levels of repression on the basis of a simple calculus of costs and benefits. According to “the law of coercive responsiveness” (Davenport 2007: 7), governments resort to repression when they face open threats to the regime, its policies, or personnel. Although little is known about its actual effectiveness,8 scholars assume repression to work well in neutralizing these threats. Given its presumed effectiveness, its benefits are proportional to the intensity of the challenges that rulers confront. In the language proposed by Robert Dahl decades ago, the literature takes the benefits of repressing opposition forces to be perfectly symmetric to the costs of tolerating them (see Dahl 1971: 15–16).9 x The survival benefits of censorship. If one conducts a bibliographic search for keywords like “censorship” and “media restrictions,” much of the resulting literature deals with limitations on the liberty of expression in established democracies. Most demands for restrictions on free speech under democratic conditions originate in issues of violence, sexuality, racism, religiosity, or public health. Studies on media restrictions by authoritarian governments are less frequent. Not too dissimilar from their liberal–democratic counterparts, authoritarian governments tend to implement media restrictions in the name of public goods, like moral decency, public security, and political stability. Yet, in contrast with democratic governments, autocrats control the media for clear partisan purposes. Authoritarian censorship is designed to hurt the opposition, be it through silence or slander, and to favor the incumbent, be it through positive reporting or non-reporting.

The Calculus of Electoral Manipulation

263

Again, we possess little systematic knowledge about the effectiveness of media restrictions in shaping the political perceptions and preferences of media consumers under authoritarian rule. Citizens are not naïve, and blatant authoritarian censorship often generates political cynicism, rather than political conformity (see Longerich 2006). Besides, modern decentralized communication technologies like the Internet and cell phones make it easier for critical citizens to circumvent centralized controls of mass communication (see Diamond 2010). Nevertheless, even if authors recognize the limits of censorship, they tend to posit that it serves authoritarian rulers. It serves them to shape the beliefs and values of their subjects and their political action, too, either by mobilizing supporters or by demobilizing opponents (see Debs 2007, Edmond 2011, Gehlbach and Sonin 2009: 29, King, Pan, and Roberts 2012, Snyder 2000). x The survival benefits of fraud. The notion of electoral fraud describes interferences with the organization of elections for partisan gain. The literature on its causes and consequences is bifurcated according to its objects of research. Those authors who study elections and plebiscites in singleand hegemonic-party regimes tend to emphasize the role of elections in reaffirming structures of political domination. In these contexts, electoral fraud appears as just one element, and usually a minor one, in a larger electoral choreography of political domination.10 By contrast, those authors who study non-democratic elections under more open conditions tend to emphasize the role of fraud in tipping the balance in close elections. In these more competitive contexts, electoral fraud appears as a strategic recourse, and sometimes even a decisive one, for controlling the outcome of elections.11 Reconstructing the symbolic role of fraud in reproducing relations of power has been largely the province of historians. Reconstructing the material role of fraud in regulating the access to power has been largely the province of political scientists. x The survival benefits of exclusion. Just like the grounds and boundaries of legitimate censorship, the grounds and boundaries of legitimate bans on political parties has been a matter of intense political debate in liberal democracies. Since Hitler ascended to power in 1933 through democratic elections (and the strategic threat of civil war), excluding the enemies of democracy from electoral competition has often been considered a legitimate instrument of democratic self-protection (see Fox and Nolte 1995). In the wake of global democratization, the dilemma-ridden debates about the limits of democratic tolerance have spread to new democracies. New candidates for democratic exclusion have been, among others, ethnic parties in plural societies, communist parties in post-communist countries, and religious parties in Islamic countries.12

264

Empirical Explorations

Although authoritarian self-defense by exclusion is a widespread and incisive phenomenon, the corresponding literature is much thinner and more recent than the literature on democratic self-defense by exclusion.13 In authoritarian regimes, the terms of public debate sometimes sound superficially similar to democratic debates. Yet, even when authoritarian governments invoke impartial arguments, the partisan goals they pursue when pushing adversaries out of the electoral market are more transparent: the self-serving limitation of electoral competition. Exclusion is not a ballot-proof instrument of authoritarian manipulation. Banned opposition elites may found successor parties or ally with approved parties, and opposition voters may vote for successor parties or shift their support to like-minded alternatives. Still, as the emergent literature on authoritarian electoral exclusion has shown, exclusionary policies often do inflict grave damage on anti-authoritarian opposition forces.14 Recognizing the benefits that rulers derive from manipulative measures demands little theoretical imagination. They are transparently self-serving instruments of authoritarian rule that respond to electoral risks and contain electoral risks. Even if they are often less effective than rulers would like them to be, we find it unsurprising when they succeed in their ostentatious purposes: when repression represses dissent, when vote rigging rigs the vote, when media controls limit the public space, and when exclusion truncates electoral competition. The puzzle arises when these strategies fail. Most of the time they do not. Otherwise rational rulers would hardly adopt them. Yet, alongside the authoritarian gratifications they promise, they all carry risks, however small, of triggering counter-intentional consequences.15

Authoritarian Risks: Corrosive Information The manipulative meta-game maneuvers by authoritarian governments are likely to reverberate throughout the entire two-level game of authoritarian elections. Electoral manipulation responds to perceived uncertainties in the electoral arena. Yet, rather than containing these uncertainties it may contribute to escalating them. The dynamics of escalation may start at the top and spiral downwards, or start at the bottom and spiral upwards. Manipulation may either awaken horizontal threats, by shaking the loyalty of state agents, and in consequence aggravate vertical threats. Or it may provoke vertical threats, by breaking the loyalty of citizens, and in consequence open up horizontal fissures. When rulers manipulate the electoral process, they alter their institutional environment. They also alter their information environment. Manipulative

The Calculus of Electoral Manipulation

265

acts convey mixed messages of regime weakness and strength. They signal that rulers are nervous about their bases of support but also that they are willing to secure loyalty through coercive means.16 Both signals may prove corrosive for the loyalty of state agents as well as citizens. Here, I will focus on the latter: the counterproductive consequences manipulation may have on citizen attitudes, perceptions, and behavior: x Normative costs: the erosion of legitimacy beliefs. To the extent that citizens hold democratic norms and observe their systematic violation by the government, electoral manipulation may actively dry up the regime’s normative sources of support in civil society.17 x Cognitive costs: the erosion of legitimacy perceptions. Since manipulation and popular legitimacy are substitutes in the production of authoritarian election results, the exercise of manipulation points to deficits in legitimacy. The extent of authoritarian maneuvers indicates the extent to which governments think they need them to keep winning elections, be it by close or high margins. From observing the intensity of regime manipulation, observers can infer the depth of regime support. Even when their own legitimacy beliefs remain stable, their second-order beliefs, their beliefs about the legitimacy beliefs of their co-citizens, may change in response to manipulation. x Behavioral costs: the erosion of loyalty. Even when manipulation saps perceived or actual levels of popular support, it need not trigger contentious action. In the face of multiple obstacles to collective action, the translation of attitudes into behavior is inherently problematic. Still, authoritarian manipulation constitutes the fundamental grievance opposition protests are supposed to respond to. If citizens hold the regime to be illegitimate, they may well start behaving in a disloyal manner. Scholars often refer to these communicative risks of manipulation as its “legitimacy costs” (see e.g. Birch 2011, Schedler 2002a). In and by themselves, both normative and cognitive costs may be of little concern to manipulative governments. It is their potential of generating “loyalty costs” among voters as well as elites that makes them worrisome. When citizens withdraw their support from the regime, or think others have been doing so, their electoral loyalty to the ruling party turns uncertain. Consequently, the loyalty of state agents turns uncertain as well. When the authoritarian containment of threats induces either voters or state agents to defect from the regime, the chain of manipulative causation turns into a self-defeating circle. Figure 8.1 illustrates the dilemma: Manipulation ends up aggravating the very threats it strives to contain. Rulers need to respond to uncertainties that they helped to create themselves. The dog of manipulation bites its tail and reacts to the pain.18

266

Empirical Explorations

Electoral manipulation

Second-order threat escalation

First-order threat containment

Corrosive information: Loyalty costs

Figure 8.1. Cycles of manipulation and threats

The idea of manipulative dilemmas (see also Chapter 4) differs from the commonsensical notion of the costs of repression that the literature routinely evokes since Dahl (1971: 15–16). Students of state repression tend to conceive the costs of repression in a manner analogous to fixed production costs that business firms operating in stable markets factor into their investment plans. Such “costs of repression” are a matter of authoritarian bookkeeping. They represent calculable costs that the accountants of manipulation simply deduct from their balance sheet of expected returns. They cover everything that may worry the masterminds of manipulation: financial as well as political costs, societal as well as personal disadvantages, certain as well as uncertain consequences, moral as well as strategic considerations, problems of inefficiency as well as problems of ineffectiveness (and at times even the anticipation of counterproductive effects).19 The counterproductive risks of manipulation, by contrast, are a matter of authoritarian anxiety. They represent uncertain, self-defeating consequences that put the very rationality of manipulation into question. Though rather wide, the concept of self-defeating risk refers to a bounded set of possibilities: the management of uncertainties may provoke the deepening of uncertainties, strategies of survival may create problems of survival.

OPTIMAL MANIPULATION If authoritarian rulers confront two types of threats, the first-order threats they try to contain through manipulation and the second-order threats

The Calculus of Electoral Manipulation

267

they risk provoking through manipulation, what are rational ways of coping with these conflicting imperatives? How can they balance the need of extinguishing present threats against the danger of inflaming future threats? It is easy to name ideal outcomes of manipulation. In the ideal dictatorial world, manipulation eliminates present as well as future threats. It is fully corrective with respect to present challenges and fully dissuasive with respect to future challenges. Yet, since authoritarian rulers are neither divine nor sovereign nor solitary actors, they “can only choose a strategy, not an outcome” (Hardin 2003: 1).

Configurations of Threat Authors often suggest that authoritarian rulers can choose optimal levels of manipulation that allow them to earn all its benefits and avoid all its costs. For instance, Gartner and Regan (1996) stipulate that “ideal” (1996: 276) levels of repression are “consistent” (1996: 278) with the intensity of the challenges a government faces. Proportional to the threat it meets, “appropriate” (1996: 277) repression is neither “insufficient” nor “excessive” (1996: 278). It neither fails to neutralize the challenge, nor does it provoke retaliation by domestic or international actors. Concepts such as the “optimal level of opening” (Eisenstadt 2000: 14) or the claim that rational autocrats do not distort the electoral process as much as possible, but only “as much as necessary to assure a win” (Monga 1997: 159), reflect similar notions of authoritarian optimization. The notion of optimal manipulation assumes that authoritarian rulers can have their cake and eat it. That they can avoid trade-offs between costs and benefits by choosing the correct level of authoritarian intervention. Like medicine, manipulation is conceived as a matter of dosage. If you prescribe too little, you don’t cure the disease. If you prescribe too much, you risk poisoning the patient. Underlying is the idea that the costs of manipulation bear a non-lineal, U-shaped relation with the intensity of manipulation. Excessive leniency in tolerating threats is costly, but so is excessive zeal in suppressing them.20 While often plausible, the U-curve does not represent an invariant, law-like relationship. Rather, we should understand it as the contingent result of variable relations between first-order and second-order threats. Figure 8.2 plots several hypothetical configurations. Figure 8.2a illustrates the authoritarian ideal: when dictators tighten electoral manipulation, first-order and second-order descend in a convergent manner. Maximum manipulation keeps both to a minimum. Figure 8.2b, by contrast, traces the U-shaped dilemma of authoritarian rulers. It derives it from the interplay between first-order

268

Empirical Explorations (b)

Effective v containmen m t, escalation at high g levels of manipulation Threats to regime surviva r l

Effec ff tive containmen m t off present and future threats Threats to regime surviva r l

(a)

Present threats

Future threats

Present threats

O

O

Electoral manipulation

Electoral manipulation

Present threats Future threats

O

Effec ff tive containment, escalation at low w level v s off manipulation Present threats

Future threats

O

Electoral manipulation

Electoral manipulation

Effec ff tive containmen m t, escalation at high g level v s off manipulation Threats to regime surviva r l

(e)

(d)

Threats to regime surviva r l

Effec ff tive containmen m t, low w propensity off escalation Threats to regime surviva r l

(c)

Future threats

Present threats

Future threats

Present threats = first-order threats Future threats = second-order threats O = Equilibrium off optimal m manipulation (present threats ≤ future threats) O Electoral manipulation

Figure 8.2. Threat containment and escalation by manipulation

The Calculus of Electoral Manipulation

269

and second-order uncertainties: the former decline with the increase of manipulation, the latter increase with the increase of manipulation. At low or “insufficient” levels of manipulation, rulers pay the cost of leaving first-order threats unchecked. At high or “excessive” levels, they pay the cost of triggering second-order threats. At medium or “appropriate” levels of manipulation, they achieve the best of all worlds: they contain first-order threats without letting second-order threats escalate.21 The symmetrical U-shaped relationship between threats and manipulation is only one possibility among others. It is a special case. Nothing compels these threat functions of either first or second order to behave this way. Either of them may start at low or high levels, proceed in lineal or curvilinear fashion, slope downwards or upwards, or just lie flat like the American Midwest. Figure 8.2 illustrates three further possibilities. Figures 8.2c and 8.2d alter the slope of second-order uncertainties. Both maintain the assumption that manipulation is effective in controlling present threats. Yet the former lowers its propensity to trigger future threats, while the latter allows it to escalate at low manipulative levels. Figure 8.2d alters the slope of first-order uncertainties. It reproduces the assumption of exponential increases in second-order threats, yet decreases the effectiveness of manipulative efforts at containing first-order threats.

Manipulative Equilibria How should rational rulers respond to these configurations of threats? Given the different time horizons of first-order and second-order threats, their optimal responses depend on their time preferences. To what degree do they discount future costs and benefits, valuing them less than immediate ones? If they hold neutral time preferences (discount factor δ = 1), they should choose levels of manipulation that minimize the overall level of threat (i.e. the sum of present and future threats).22 If they see themselves under grave and imminent danger, they should dismiss future risks (δ = 0) and minimize present threats. If they believe they can weather present threats, yet have to worry about emergent challenges (δ ≥ 1), they may ignore momentary uncertainties and minimize future threats (up to the point where the present devours the future). While actors may follow any of these rules in certain situations, none of them provides a persuasive criterion of optimality across contexts. As a more general decision rule, I wish to suggest a fourth possibility: dictators

270

Empirical Explorations

balance present and future utility, yet they do so under the primacy of the present (δ < 1). Under this rule of asymmetric balance, the masterminds of manipulation minimize first-order threats up to the point that second-order threats start exceeding them. If manipulative maneuvers promise to contain present threats only at the cost of provoking future threats that are higher than the present ones, rulers will refrain from engaging in these maneuvers. They choose manipulative strategies only if the conditional probability of threat containment (c) under manipulation (m) is larger than or equal to the conditional probability of threat escalation (e) under manipulation.

(

p C M

) p (E M)

(8.2)

All points of intersection between the threat curves in Figure 8.2 satisfy this condition. They determine the levels of “optimal” manipulation O where no other manipulative choices yield lower levels of first-order threat without provoking levels of second-order threats that surpass them. These optimal points of manipulation grant regimes the comforting knowledge that they cannot do more to suppress present challenges without provoking excessive future challenges. Having reached these optimal decision points may grant regimes certain comforts. But getting there may involve difficult dilemmas. Dilemmas of dynamic adaptation may also arise at optimal points themselves when the curves of first-order threat shift upwards, for instance in response to external shocks such as economic crises, foreign intervention, or the international diffusion of protest waves.

Manipulative Adjustments The decision rule that orders electoral autocrats to minimize present threats up to the point where future threats start exceeding them yields equilibria points that look plausible except for one class of threat configurations. When present threats are inelastic, yet future threats very sensitive to manipulation (as in Figure 8.2e), optimal manipulative strategies purchase small decreases in present threats with large increases in future threats. When we abandon the static world of equilibria and consider the logic of dynamic adjustments, the corresponding dilemma of authoritarian decision-making becomes more transparent. In the face of exogenous changes to levels of threat, authoritarian governments need to adjust their levels of manipulation. In such dynamic scenarios, they need to estimate the marginal benefits of changes in manipulation. If they value the present as much as the future (or more), they

The Calculus of Electoral Manipulation

271

should require changes in the conditional probability of threat containment to be equal or higher than changes in the conditional probability of threat escalation:

(

Δp C M

)

(

Δp E M

)

(8.3)

To the extent that they discount the future, they will accept trading small decreases in first-order threats against larger increases in second-order threats. We can express their intertemporal calculations by multiplying changes in second-order threats by discount factor δ:

(

Δp C

) ≥ δ Δp (E M)

(8.4)

Threats to regime survival

Figure 8.3 illustrates the decision-making dilemma that emerges when authoritarian hardening provokes second-order threats that exceed the first-order threats it suppresses. It shows a world in which first-order threat functions are rather flat (so that additional manipulation yields little present gains) and second-order threat functions rather steep (so that additional manipulation generates large future risks). When exogenous shocks cause first-order threats to climb from low levels at t1 to high levels at t2 (without changing their slope), rulers face a strategic catch. If they want to be marginally safer today, they risk

Present threats t2

Future threats

A B

Present threats t1

C

O1

O2

Electoral manipulation

Figure 8.3. Adjusting manipulation to rising first-order threats O = Equilibrium of optimal manipulation (present threats ≤ future threats)

272

Empirical Explorations

being dramatically unsafe tomorrow. If they follow the static decision rule of minimizing absolute levels of present threats (until they exceed absolute levels of future threats), they should move from the initial equilibrium O1 to the subsequent point of optimal manipulation O2. Yet, the corresponding reduction in first-order threats at t2 (line segment a–b) is much smaller than the corresponding increase in second-order threats (line segment b–c). Moving from one static equilibrium to another would be myopic. Rational rulers who value the future almost as much as the present should refrain from making such a move. Instead they should follow our dynamic decision criterion—and not do anything at all. Since they cannot contain present threats without provoking disproportionate future threats, they should keep levels of manipulation where they are.

Contextual Variance If optimal levels of manipulation depend on empirical configurations of first-order and second-order threats, we should expect them to vary across contexts (regime subtypes) and tools (authoritarian instruments). If hegemonic and competitive regimes differ in their typical constellations of firstand second-order threats, the authoritarian management of threats should differ among regime subtypes. If different manipulative strategies differ in their intrinsic propensities to provoke second-order threats, authoritarian choices should reflect these differences. How should we expect hegemonic and competitive regimes to differ in their management of threats? Hegemonic regimes defend an institutional equilibrium that competitive regimes struggle to build. We know that both regime subtypes are similar in their levels of manipulation, yet differ in their levels of threat. Hegemonic regimes are realms of political quietness, competitive regimes arenas of political agitation. As the former obtain lower levels of threat per unit of manipulation, their manipulative investments appear more effective on average. Their first-order threat curves lie lower, and are likely to descend steeper, than those of their competitive counterparts. Hegemonic regimes have less to fear from first-order threats and probably from second-order threats as well. Their institutional equilibrium is grounded in beliefs about their invincibility, which are grounded in beliefs about their popular legitimacy and manipulative capacity. People believe that the regime is still legitimate and strong. Their cushion of interlocking perceptions is likely to grant hegemonic regimes more leeway in containing authoritarian threats without appearing to be either weak or illegitimate. Competitive regimes, by

The Calculus of Electoral Manipulation

273

contrast, are more likely to face steeply rising legitimacy costs when they step up their manipulative defenses. They are more likely to find themselves in the dilemmatic situation depicted in Figure 8.3. If they strive to compress emergent threats by increasing manipulation, the reduction in first-order threats they achieve is much smaller than the additional second-order risks that they provoke. If these assumptions hold, and if the first-order threat function of competitive regimes lies higher and runs flatter, and the second-order function runs steeper, than in hegemonic regimes, we should expect these two subtypes of electoral authoritarian regimes to differ in their mean strategies of threat management. Hegemonic regimes should be less inhibited in pulling their weapons of mass manipulation in the face of increasing first-order uncertainties. They should be more likely to resort to defensive manipulation. Competitive regimes, by contrast, should respond more cautiously to rising tides of threat. The deterrent force of second-order uncertainties should block their authoritarian instincts of defending themselves through manipulative escalation. They should be more likely to resort to defensive concessions.

Instrumental Variance While the choice of authoritarian manipulation is likely to reflect contextual constraints, it may also reflect the intrinsic properties of various manipulative strategies. If authoritarian tools differ in their inherent propensity of threat provocation, the instrumental choice among them is likely to reflect forward-looking concerns about second-order threats. As I wish to propose, all authoritarian strategies are equal (in their capacity of subverting democratic norms), yet some are more equal than others (in their capacity of provoking democratic opposition). For authoritarian manipulation to provoke second-order uncertainties, it needs to affect the legitimacy of the regime. For manipulation to affect regime legitimacy, citizens need to be good democrats as well as good observers. They need to care about democratic norm violations and they need to be aware of them. If all citizens are equal in their normative and observational propensities, the impact of manipulative strategies should vary to the extent that they vary either in their normative implications (severity) or their accessibility to public inspection (visibility). Egregious norm violations that are non-observable should leave citizen evaluations untouched, and so should visible strategies that do not violate their normative sensibilities. The legiti-

274

Empirical Explorations

macy costs of manipulation appear as the interactive product of severity and visibility: Legitimacy costs = severity * visibility of manipulation The visibility of manipulation. Despite their deliberate theatricality, electoral authoritarian regimes can never acquire full monopolies on “the social construction of reality” (Berger and Luckmann 1966). No political regime ever can. Some things are difficult to hide, like the expulsion of popular opposition candidates from electoral competition or their forced change of residence to penal institutions. Other things may be drawn into the open by vigilant opposition groups. Well-organized election monitors, for instance, may not be able to prevent electoral fraud from being committed, but from going undetected. Also, whether media controls work as effective instruments to establish “authoritarianism by stealth” depends on the resistance the media offer to their political subordination, the credibility citizens lend to biased media, and the access that they enjoy to subterranean sources of alternative information.23 Even though the visibility of the four authoritarian strategies under investigation thus depends on numerous contextual factors, I would roughly rank them in the following way (in descending order of visibility): Exclusion > repression > fraud > censorship Since my category of exclusion refers to the corrective banning of active parties and candidates, rather than the preventive prohibition of potential electoral contenders, there is no way to conceal it from the affected parties and their followers. Repression, too, cannot be hidden from its direct victims, although the general public may be unaware, or choose to be unaware, of its occurrence. Fraud involves a wide array of activities of varying degrees of publicity. If opposition parties are weak or if governments wish to show them they are, it can be blatant and practiced in the open. Yet, more often than not, electoral fraud, like any other form of fraud, is “extremely difficult, if not impossible, to observe directly” (Birch 2009: 399). Finally, effective media restrictions are effectively invisible. They are the magic hood of authoritarian governments that erases their misdeeds from the public sphere and in the act erases the eraser itself. As media restrictions shape the window through which citizens see the political world, the window itself is not readily visible to them. The severity of manipulation. Strategies of electoral manipulation are not a matter of all or nothing. They are graded phenomena. Rulers may deploy them lightly, with a touch of innocence, or impose them ruthlessly, with an iron fist. Whether they imply more or less severe violations of democratic norms depends to a large extent on whether rulers impose them more or less severely. Still, individual strategies do contain certain intrinsic elements of

The Calculus of Electoral Manipulation

275

Table 8.1. The severity and visibility of manipulative strategies

More severe norm violations Less severe norm violations

Less visible norm violations

More visible norm violations

Fraud Censorship

Repression Exclusion

severity. I would roughly rank them in the following way (in descending order of severity): Repression > fraud > exclusion > censorship My major criteria are harmfulness and irreversibility. In terms of harm, the violation of physical integrity is clearly the most damaging strategy. Beating up political opponents, throwing them into jail, torturing and injuring them, or even physically eliminating them inflicts deep harm on victims.24 Though less damaging than repression, electoral fraud is a mode of political expropriation. It silently confiscates the political capital that opposition parties have acquired in the form of voter support. It deprives them of entitlements (in terms of votes and seats), while exclusion denies them no more than opportunities. Censorship, too, primarily shapes the world of political possibilities. It distorts political beliefs and preferences, the electoral consequences of which are difficult to estimate. In terms of irreversibility, some violations of physical integrity may heal, eventually, but that does not make them reversible. Fraud is final, too. Once all instances of appeal are exhausted, fraudulent outcomes cannot be overturned except through extra-institutional means. Acts of exclusion, by contrast, do not seal the electoral fate of opposition actors. They may find ways around electoral bans by supporting proxy candidates. Finally, the denial or distortion of information through censorship often knows partial redress: access to unofficial sources of alternative information. In courageous simplification, Table 8.1 dichotomizes the locations of these four strategies along the two dimensions of visibility and severity. If rulers care little about counterproductive second-order effects, they should prefer the more severe and more visible measures. If they worry about the legitimacy costs of manipulation, they should pick the lighter and less visible strategies.

C ON TA IN IN G T H R E ATS If the “optimal” manipulative balance depends on the effectiveness of manipulation in containing first-order threats and its propensity to provoke

276

Empirical Explorations

second-order threats, that is, if it depends on the manipulation elasticity of present as well as future threats, then my empirical explorations into the logic of electoral manipulation should address both questions: To what extent are manipulative strategies direct responses to first-order uncertainties? And to what extent are these responses mediated by the anticipation of second-order uncertainties? Regrettably, due to data limitations, I have to confine my empirical examinations to the first question. After having developed a rather complex set of theoretical arguments, I will proceed to examine some of their empirical implications in a most simple and straightforward manner. I will examine the direct, unconditional responses of rulers to various types of vertical threat: electoral competitiveness (as measured in previous and predicted margins of victory), peaceful and violent vertical threats in between elections (anti-government demonstrations, general strikes, riots, and armed rebellion), and election-related protest on the eve of national elections (protest mobilization and electoral boycott). By exploring the lineal short-term associations between first-order threats and manipulation I wish to see to what extent rulers respond to these sources of uncertainty without taking into account the countervailing risks of provoking second-order uncertainties. The basic hypothesis is simple: H1 Rulers strive to contain emergent uncertainties through authoritarian measures. The higher the level of first-order threats, the higher the level of manipulation. Yet, even simple questions about direct responses to present threats are likely to be “contaminated” by considerations of future threats. According to the hypotheses developed above, the lineal, positive relation between first-order threats and manipulation should be mediated, and thus attenuated, by two intervening variables: subtypes of regimes and types of strategies. First, hegemonic and competitive regimes, as I argued above, differ in their optimal responses to first-order threats. Hegemonic regimes are more likely to address rising threats through rising manipulation. Competitive regimes are more likely to temper their defensive responses: H2 In hegemonic regimes, we should see strong positive associations between levels of threat and levels of manipulation. In competitive regimes, the lineal positive relation between threat and manipulation should be more tenuous and may even turn negative. Second, different strategies of manipulation, as I further argued, differ in their intrinsic propensity to provoke second-order threats, and thus in their optimal deployment in the face of first-order threats. More severe and more visible authoritarian strategies are likely to generate higher legitimacy costs, and thus

The Calculus of Electoral Manipulation

277

higher risks of escalating second-order threats. Rulers should be cautious in imposing such high-risk strategies: H3 Low-cost strategies (such as censorship and exclusion) should show lineal positive relations with levels of threat, high cost strategies (such as repression and fraud) weaker or even negative relations.

Electoral Uncertainty To assess the extent to which authoritarian manipulation responds in a direct manner to electoral uncertainty, the queen of regime uncertainties in electoral autocracies, I use two measures of electoral threat: lagged and predicted margins of victory. The former indicate margins of victory in previous elections. The latter derive from three contextual variables that emerged as strong predictors of electoral competitiveness in the previous chapter: wealth, growth, and legislative ineffectiveness. In ols regression analyses (not reported),25 these three variables jointly explain about one-fifth of variation in margins of victory in hegemonic as well as in competitive regimes. I calculated “predicted margins of victory” from the resulting regression coefficients for these three predictor variables.26 Previous margins of victory. In a first step, I regress individual strategies of manipulation on previous margins of victory while controlling for previous levels of manipulation. Table 8.2 displays the results. All estimations derive from lineal ols regressions (with the table showing unstandardized B coefficients and p values), except for the dichotomous dependent variable exclusion, for which I use binary logistic regression (with the table showing odds ratios eb and p values).27 As I lose all first elections when incorporating lagged values, case numbers are depressingly low. The results are still illustrative, at least for one point: the stickiness of manipulation. Past levels of manipulation explain present levels of manipulation. Regression coefficients for lagged manipulation are large and significant, and the overall explanatory power of the regression models is high. With one exception: fraud in competitive regimes seems less a matter of authoritarian inertia than a matter of authoritarian discretion. Previous fraud is barely significant (at p ≤ .20) and explains almost nothing (R2 = .05).28 While current levels of manipulation seem strongly related to previous levels of manipulation, they seem largely unrelated to previous margins of victory. Previous levels of electoral certainty appear to account only for variations in exclusion. More secure hegemonic regimes are more exclusionary, more secure competitive regimes more inclusionary. Overall, however, the control

278

Empirical Explorations

Table 8.2. Electoral uncertainty and manipulation Repression B Previous margins of victory hr Previous mv –.011 .501 Manipulation t−1 Constant 2.389 Standard error / pac 1.522 .307 R2 / Cox & Snell R2 N 32 cr Previous mv .001 .429 Manipulation t−1 Constant 2.495 Standard error / pac 1.502 .192 R2 / Cox & Snell R2 N 42 Predicted margins of victory –.054 hr Predicted mv Constant 7.160 Standard error / pac 1.817 .097 R2 / Cox & Snell R2 N 54 cr Predicted mv –.019 Constant 5.396 Standard error / pac 1.932 .024 R2 / Cox & Snell R2 N 86

Censorship

Exclusion

Fraud

p

B

p

eb

p

B

p

.364 .002 .024

.001 .567 .479 .321 .454 33 –.001 .385 .847 .371 .154 42

.635 .000 .034

1.047 42.773 .008 88.2% .524 34 .973 8.102 .465 79.1% .241 43

.097 .001 .024

.000 .692 .192 .550 .481 32 –.002 .226 1.154 .678 .053 42

.542 .157 .000

.006 .990 .441 .020 55 .012 .974 .392 .179 86

.304 .009

.921 .004 .005

.021 .000

.155 .000

.609 .011 .000

.000 .000

1.114 .001 64.3% .203 5 1.019 .467 51.7% .019 87

.055 .009 .253

.003 .004

.197 .116

–.027 2.679 .675 .162 55 .000 1.074 .732 .000 86

.542 .157 .000

.002 .000

.972 .000

Note: hr = hegemonic, cr = competitive regimes. Previous mv = electoral margins of victory in previous election. Manipulation t−1 = levels of manipulation in previous election. Predicted margins based on ols regression (wealth, growth, and legislative ineffectiveness) (available at request from author). ols lineal regressions for repression, censorship, and fraud (reporting unstandardized B coefficients, standard errors, and R2 values), binary logistic regressions for exclusion (reporting odds ratios eb, Percentages of Accuracy of Classification pac, and Cox & Snell R2 values). Shaded coefficients = statistically significant (p ≤ .15).

variable (lagged manipulation) ends up dominating the explanatory variable (lagged competitiveness). This is what lagged values of dependent variables often do. To ascertain the relative irrelevance of electoral uncertainty, the presumptive mother of all institutional uncertainties, is frustrating nevertheless. Theory frustrating. What may account for this negative finding (except the usual culprit, my low case numbers)? Perhaps, rulers do not react to official election results, their self-made fictional figures, but to underlying trends, as more genuine indicators of interparty competitiveness.

The Calculus of Electoral Manipulation

279

Predicted margins of victory. By regressing individual strategies of manipulation on predicted margins of victory (estimated from wealth, growth, and legislative weakness), I try to gauge the extent to which rulers choose levels of manipulation, not in response to official competitiveness in the past, but in response to contextual factors that co-determine present levels of competitiveness. How do rulers respond to “bad times” (contextual factors that presage lower margins of victory), how to “good times” (contextual factors that presage higher margins of victory)? Table 8.2 shows the results. Under conditions of tranquillity (at high margins of victory), hegemonic regimes relax repression and fraud (coefficient signs are negative), while competitive regimes resort to censorship and exclusion (coefficient signs are positive). In turn, under conditions of stress (at lower margins), the former resort to repression and fraud, while the latter relax censorship and exclusion. In sum: In accordance with theoretical expectations, hegemonic regimes are inclined to resort to defensive manipulation, competitive regimes to defensive concessions. Yet, against theoretical expectations, electoral authoritarian rulers tend to deploy the most severe strategies (repression and fraud) in a defensive manner (in the presence of higher competitiveness) and the least severe strategies (censorship and exclusion) in a pre-emptive manner (in times of electoral tranquillity). This negative finding resonates with Judith Kelley’s empirical finding that manipulative governments are less selective and less sophisticated than we are inclined to expect (2012: Ch. 5). They do not respond to the presence of international election observers by shifting from visible to hidden strategies of electoral manipulation. Governments “who cheat are by and large neither subtle nor safe” (Kelley 2012: 93).

Contentious Action If official indicators of electoral uncertainty do not seem to trigger consistent manipulative responses, what about other types of observable vertical threats? Do various forms of contentious action (anti-government demonstrations, general strikes, and riots) prompt consistent authoritarian responses? Does organized societal violence directed against state authorities do so? Or do forms of opposition protest that target the electoral process in an explicit manner meet with similar authoritarian responses across our diverse sets of cases? The regression results reported in Table 8.3 provide some tentative answers. In hegemonic regimes, contentious action explains both repression and fraud rather well. In the face of vertical uncertainties, hegemonic regimes take decisive corrective action: they turn more repressive and more fraudulent.

280

Empirical Explorations

Table 8.3. Vertical uncertainties and manipulation Repression

hr Vertical threats Armed rebellion Pre-electoral protest Electoral boycott Constant se / pac R2 / Cox & Snell R2 N cr Vertical threats Armed rebellion Pre-electoral protest Electoral boycott Constant se / pac R2 / Cox & Snell R2 N

Censorship

Exclusion

Fraud

B

p

B

p

eb

p

B

p

.102 .688 .724

.098 .002 .150

–.002 .059 –.101

.935 .356 .510

.985 1.577 .731

.866 .170 .620

.053 .012 .246

.054 .897 .270

.324 2.585 1.511 .399 55

.218 .000

.005 1.344 .463 .026 56

.951 .000

.623 1.382 68.4% .076 57

.190 .400

.196 .605 .673 .198 56

.095 .000

–.012 .320 .412

.670 .000 .310

–.023 –.007 .111

.001 .662 .253

.939 1.081 1.788

.123 .337 .225

.018 –.058 .086

.135 .044 .606

–.460 4.301 1.749 .234 88

.098 .000

–.026 1.399 .419 .133 88

.689 .000

1.197 .552 64% .056 89

.577 .107

.051 1.062 .721 .077 88

.655 .000

Note: hr = hegemonic, cr = competitive regimes. Vertical threats and armed rebellion: sum of events in three years previous to election year. Repression and censorship in election year. Binary logistic regressions for exclusion. Otherwise ols lineal regressions. se = Standard error. pac = Percentage of accurate classification. Shaded coefficients = statistically significant (p ≤ .15).

In competitive regimes, contentious action accounts best for repression and censorship. In the face of opposition boycotts, these regimes ease repression, in the face of armed rebellion they step it up. Rather than triggering tighter restrictions, contentious actions against competitive regimes seem to have a liberating effect on the media. Figure 8.4 illustrates the estimated effects that increases from zero to three in my indicator of vertical threats (the sum of demonstrations, general strikes, and riots in the three years previous to the election year) carry on levels of manipulation.29 The picture looks strikingly similar to the effects of predicted levels of electoral competition: Under the pressure of vertical threats, hegemonic regimes aggravate repression and fraud, competitive regimes alleviate censorship and exclusion. All effects are standardized to a scale from zero to one. Although significant in statistical terms, they are minuscule in substantive terms. Still, they conform to our theoretical expectations. As hegemonic parties are more effective in containing first-order uncertainties and have to worry less about provoking

Hypothetical changes in manipulation

The Calculus of Electoral Manipulation

281

0.06 0.08

0.03 0.04 0.03

0.00

–0.01 –0.03

–0.04

–0.03

CR

H R

au d Fr

n

au d Fr

sio

n

clu

Ex

sio clu

Ex

CR

H R

CR

R

hi p

H Ce ns o

rs

hi p

n

Ce ns o

rs

sio

n

es

sio es

pr Re

pr Re

CR

H R

–0.06

Figure 8.4. Estimated effects of hypothetical increases in vertical threats on manipulation Note: The bars show changes in levels of manipulation, in response to hypothetical changes from zero to three vertical threats (anti-government demonstrations, general strikes, and riots) during the three years preceding the election year (calculated on the basis of regression results presented in Table 8.2). All strategies of manipulation are standardized to a scale of 0 to 1. Values for exclusion are changes in predicted probabilities of exclusion (all other threats held at zero, calculated with X-Post). Shaded bars denote statistically significant effects (p ≤ .15).

second-order risks, they are free to respond to emergent threats by tightening, rather than relaxing, authoritarian manipulation.

P ROT E C T I N G S TA K E S If rulers use manipulative strategies to tighten their grip on power, they should find them most useful when their hold on power is most insecure and when their control of power is most valuable. In the preceding sections, I have examined manipulative responses to variations in insecurity. Now I wish to explore manipulative responses to variations in the value of state power, i.e. in the stakes of the electoral game. Since Robert Dahl’s seminal remarks (1971: 15–16) on “the costs of tolerance” (that is, the costs of democratizing change to authoritarian rulers), students of political regimes have been factoring “the stakes of politics” into

282

Empirical Explorations

their equations of regime politics. If the stakes of the political game are high, if rulers have much to lose when losing power, they will resist democratizing pressures. In turn, if the stakes of the political game are more moderate, if rulers have little to fear from the loss of executive power, they will more readily concede to democratizing demands. The stakes of politics may influence authoritarian strategies either in a conditional manner, contingent on prevailing levels of threat, or in a direct manner, regardless of prevailing levels of uncertainty. As outlined earlier in this chapter, within a decision-theoretic framework, rulers should not worry about their stakes as long as they do not run any palpable risk of losing them. The intensity of regime conflicts c should be the product of the subjective value of power u(sQ) and perceived threats to power p(sQ’):

c = p(sq’)*u(sq)

(8.5)

Both variables interact. If power is worth nothing, rulers should care little about losing it. If power is valuable, yet secure, rulers should worry little about losing it. Only if the control of power is both valuable and insecure, should rulers fight hard for preserving it. However. Even if the political game looks secure and rulers are confident in swiping all the bets, the sheer magnitude of their potential losses may induce them to be risk averse. They may play safe and maneuver to protect their bets even in the absence of credible challengers. In such cases, the intensity of power struggles simply mirrors the magnitude of potential losses:

c = u(sq)

(8.6)

In either scenario, political stakes play a simple role: they shape authoritarian attitudes towards risk. High stakes induce risk aversion, low stakes risk tolerance. If the ultima ratio of electoral manipulation lies in the defense of the authoritarian status quo, the same logic of risk management should apply. If the stakes of power are high, rulers should be risk averse and intensify their defensive efforts against emergent threats. They may even step up their manipulative defenses in the absence of manifest challenges. If the stakes of power are low, they can be risk tolerant and more permissive in the face of rising institutional uncertainties. If they conceive of their interests in qualitative rather than continuous terms, their manipulative defenses will follow the logic of thresholds: when vertical uncertainties start threatening their “vital interests” they will escalate their defensive efforts.30 Figure 8.5 illustrates the dependence of manipulative responses on authoritarian risk attitudes.

The Calculus of Electoral Manipulation

First-order threats

Risk tolerance

283

N

S

Risk aversion

Electoral manipulation

Figure 8.5. Authoritarian attitudes to risk Note: N = Risk neutrality. S = Threshold effects.

Ascribing contrasting risk attitudes to actors often works as a convenient labeling device that does not resolve explanatory puzzles, but shifts them into a psychological “black box” (Boudon 1998: 180). Given the concentration of power in authoritarian regimes, it is always tempting to explain dictatorial decisions by personal characteristics of the dictator. Yet, if certain features of the political environment either compel actors to be risk averse or allow them to be risk tolerant, we need not dig up the psychological foundations of risk attitudes. In the previous section, I looked at second-order risks modulating authoritarian attitudes towards first-order risks. Here, I will look at political stakes playing the same role: conditioning the authoritarian management of visible challenges. I will examine first a bundle of societal factors and then a cluster of institutional variables that tend to influence the stakes of politics.

Societal Stakes The exercise of power is said to provide multifarious satisfactions to the wielders of power: material comfort, social prestige, a purpose in life, the aggrandizement of the ego, aphrodisiac powers (Heinrich dixit).31 Whether the control of state power yields large or small benefits in narrow financial terms, depends on the ratio of economic opportunities (reo) inside and outside public office.

284

Empirical Explorations

With Em standing for income opportunities in market employment and Es for financial opportunities in state employment, we can write:

reo = Em / Es

(8.7)

If reo is smaller than one, it is comparatively attractive to seek public office. If it is larger, less so. If authoritarian nations are poor, unequal, and oil rich, the payoffs of public employment often surpass the penuries of private employment many times over. • Poverty: To the extent that economic development creates sources of wealth outside the state, it lowers the comparative financial advantages of holding public office. Under conditions of widespread poverty, the routes to affluence are few and narrow. Private access to public resources is usually the most obvious one (see e.g. Hadenius 1992: 78). I use the relative size of the rural population (as percentage of total population) as a proxy for poverty. • Inequality: To the extent that authoritarian politics is an expression of social warfare by other means, and to the extent that authoritarian regimes act as armed agents of the rich and protect the prevailing distribution of income and property, the stakes of authoritarian politics rise with rising social inequalities (see e.g. Boix 2003). I use Gini coefficients as indicators of the social distribution of monetary wealth. • Oil: To the extent that states control the appropriation of easy income (rents), be it in the form of natural resources, foreign aid, or external loans, occupying public office equals to occupying sources of effortless wealth. The more abundant these sources, the higher the stakes of political struggle (see e.g. Karl 1997). I use the relative size of mineral oil exports (as percentage of total exports) to measure the availability of rents from natural resources. If high political stakes induce authoritarian rulers to enact cautionary measures, regardless of prevailing levels of threat, we should see direct relationships between the magnitude of stakes and the intensity of manipulation. If high political stakes induce them to intensify their defensive responses to prevailing threats, we should observe interactive effects of stakes and threats on the intensity of manipulation: the presence of high threats should magnify the impact of high stakes on manipulation, and vice versa. To trace these effects, I regress the four strategies of manipulation (repression, censorship, exclusion, and fraud) on the three indicators of societal stakes (poverty, inequality, and oil wealth). Table 8.4 presents the results.

The Calculus of Electoral Manipulation

285

Table 8.4. Societal stakes and manipulation Repression

Censorship

Exclusion

Fraud

B

p

B

p

eb

p

B

p

hr Poverty Inequality Oil Constant Standard error / pac R2 / Cox & Snell R2 N

.038 .001 .024 1.308 1.755 .295 33

.004 .980 .099 .537

–.002 –.021 –.002 2.441 .379 .150 34

.489 .030 .538 .000

.892 .769 1.001 6.310 80% .454 35

.053 .024 .956 .024

.014 –.021 .014 .691 .574 .479 34

.001 .144 .005 .316

cr Poverty Inequality Oil Constant Standard error / pac R2 / Cox & Snell R2 N

.011 .069 .022 1.084 1.898 .151 69

.464 .007 .011 .454

.006 –.002 .004 .997 .436 .107 69

.065 .681 .034 .004

1.003 .975 1.017 1.329 64.3% .074 70

.842 .343 .089 .857

–.013 .004 .001 1.557 .724 .085 69

.026 .687 .792 .006

Note: hr = hegemonic, cr = competitive regimes. Poverty = rural population (% total population). Inequality = Gini. Oil = mineral fuel exports (% total merchandise exports). Repression and censorship in election year. Binary logistic regressions for exclusion. Otherwise ols lineal regressions. Shaded coefficients = statistically significant (p ≤ .15).

Although low case numbers oblige us to read the results with caution, societal stakes seem to do a fair job in explaining variations in hegemonic manipulation. Together, poverty, inequality, and oil wealth explain almost one-third of variance in repression (R2 = .29), about one-seventh of censorship (R2 = .15), and almost one-half of exclusion (R2 = .45) and electoral fraud (R2 = .48). In accordance with expectations, poor yet oil-rich hegemonic regimes are more repressive and more fraudulent. Against expectations, unequal regimes trample their subjects with a lighter foot. They put lesser restrictions on the media, are less exclusionary, and less prone to vote rigging. In competitive regimes, by contrast, societal stakes seem to explain less, even if a fair number of coefficients turns out significant. The overall picture is similarly incoherent, though. Only the effects of oil are consistent with theoretical expectations: Oil wealth renders competitive regimes more repressive, more media restrictive, and more exclusionary. In turn, poverty augments censorship, yet dampens fraud. Inequality invites higher repression, yet nothing else. Figure 8.6 graphically illustrates this mixture of confirming findings, disconfirming evidence, and non-findings. It plots estimated changes in manipulation in response to hypothetical 10-unit changes in societal stakes.

286

Empirical Explorations

Estimate a d effects of increases in rural population (by 10%)

0.10 0.07

0.05

0.03

0.00

–0.07

–0.10 –0.25

–0.20

Estimate a d effects of increases in Gini index (by 10 points)

Re

Re

pr es sio n H pr R es Ce sion ns or CR s Ce hip H ns R or sh i p Ex C clu R sio n Ex H clu R sio n CR Fr au d H R Fr au d CR

–0.30

0.09

0.00

–0.06

–0.11

–0.11

–0.20 –0.57

–0.40

Re

Re

pr es

sio n

H pr R es Ce sion ns or CR s Ce hip H ns R or sh ip Ex clu CR sio n Ex H clu R sio n CR Fr au d H R Fr au d CR

–0.60

Estimate a d effects off increases in oil exports (by 10%)

0.08

0.06 0.04 0.07 0.04

0.02 0.03

0.03

0.02

0.00 –0.01

Re

Re

pr es

sio n

H pr R es Ce sion ns or CR s Ce hip H ns R or sh ip Ex clu CR sio n Ex H clu R sio n CR Fr au d H R Fr au d CR

–0.02

Figure 8.6. Estimated effects of hypothetical increases in societal stakes (by 10 units) on manipulation Note: The bars show changes in levels of manipulation, in response to hypothetical increases of poverty, inequality, and oil exports by 10 units (calculated on the basis of regression results presented in Table 8.4). All strategies of manipulation are standardized to a scale of 0 to 1. Values for exclusion are changes in predicted probabilities of exclusion (everything else held at its mean, calculated with X-Post). Shaded bars denote statistically significant effects (p ≤ .15).

The Calculus of Electoral Manipulation

287

The resulting dance of bars nicely shows the inconsistent effects of poverty, the largely decompressing effects of inequality, and the consistently oppressive effects of oil wealth. They also visualize the higher sensitivity of hegemonic regimes (dark shaded bars) to these structural factors in comparison to competitive regimes (light shaded bars).32

Institutional Stakes If features of the societal environment leave traces on the intensity of electoral manipulation, features of the self-made institutional environment are even more likely to do so. They are more proximate to the electoral arena. Rather than defining generic stakes of politics they define the stakes of electoral politics. I will examine patterns of association between levels of manipulation and four institutional variables which I expect to augment the stakes of legislative elections, and thus the intensity of legislative manipulation: • Parliamentarism: In parliamentary systems, legislative elections are supposed to determine the composition of the government. As long as the ruling party holds comfortable majorities, it does not need to worry about the power of parliament. Still, given the formal role of legislatures in parliamentary systems, I expect parliamentary governments to subject legislative elections to tighter manipulation than in presidential systems. • Majoritarian electoral systems: Majoritarian electoral systems are winner-takes-all systems. In electoral authoritarian regimes, they tend to work, and are intended to work, as ruler-takes-all systems. The incumbent party wins everything, the opposition nothing. Authoritarian ruling parties can be expected to protect their fortunes under these protective rules and make manipulative investments to ward off a possible inversion of roles between winners and losers (see also Birch 2011: 71–72). • Legislative effectiveness: Effective legislatures are arenas of contestation, ineffective ones theaters of acclamation. As the former matter more than the latter, controlling them should matter more to authoritarian rulers. • Concurrent elections: In authoritarian as well as in democratic presidential systems, voters often pay little attention to legislative elections that are held outside the presidential election cycle. Authoritarian rulers, too, may treat non-concurrent legislative elections with benign neglect. They are more likely to pay full manipulative attention to concurrent elections. Higher institutional stakes should make authoritarian governments resort to higher levels of manipulation. To what extent do our data bear out this expectation? When we compare mean levels of legislative manipulation across these four institutional features, a mixed picture emerges.33 Most differences

288

Empirical Explorations Hegemonic regimes

Competitive regimes

2.0

1.0 1.56 0.5

1.22

1.20

Mean violations of media freedom (p = 0.018)

Mean violations of media freedom (p = 0.038)

1.25 1.5

Semi-presidential (N = 5)

Parliamentary (N = 18)

1.32 1.06

1 0.8 0.6 0.94 0.4 0.60 0.2

0.32

Presidential Semi-presidential Parliamentary (N = 46) (N = 16) (N = 11) Mean legislative exclusion (p = 0.090)

Mean legislative exclusion (p = 0.000)

1.39

0.5

0.0 Presidential (N = 25)

0

0.4 0.3 0.48 0.2 0.25

0.1

0.18

1.5

1.0 1.59 1.05

0.5

Presidential (N = 46)

Parliamentary (N = 18)

0.50

Mean legislative fraud (p = 0.022)

Semi-presidential (N = 5)

2.0

Semi-presidential (N = 16)

Parliamentary (N = 11)

1.2 1 0.8 1.22

0.6

1.04 0.4 0.2

0.56

0

0.0 Mixed electoral system (N = 11)

Proportional representation (N = 16)

Majo a ritarian system (N = 11)

1.2 1 0.8

1.12

0.4 0.61 0.2

Mean violations of media freedom (p = 0.000)

Proportional representation (N = 1)

0.6

0.5

0 Presidential (N = 25)

Mean violations of media freedom (p = 0.007)

0.75

0.25 0.0

Mean legislative fraud (p = 0.007)

1.0

Mixed electoral system (N = 18)

Majo a ritarian system (N = 23)

1.5

1.0 1.50 1.16 0.5

0.0

0 Ineffective legislature a (N = 25)

Semi-effective legislature a (N = 23)

Ineffective legislature (N = 32)

Semi-effective legislature (N = 41)

Figure 8.7. Institutional stakes and legislative manipulation Note: Legislative elections only. P-values from comparisons of means through Analyses of Variance (anova).

The Calculus of Electoral Manipulation

289

in means are not statistically significant. Figure 8.7 displays those that are (in anova Analyses of Variance, p ≤ .10).34 In hegemonic regimes, variations in mean levels of legislative manipulation generally conform to the expectation that higher institutional stakes translate into more intense manipulation. Hegemonic parties that run parliamentary systems of government tend to subject legislative elections to tighter media restrictions. Almost invariably, they also ban parties or candidates from the electoral arena. If they conduct legislative contests under majoritarian electoral rules, they impose higher levels of censorship as well. Counterintuitively, they interfere less with the administration of elections to semi-effective than to ineffective parliaments (see Figure 8.7). In competitive regimes, the picture is much less clear. Only electoral formulae hold what they promise. Under proportional representation, legislative elections are less tainted by fraud than under either mixed or majoritarian election rules. Otherwise, competitive authoritarian regimes do not seem to set their levels of manipulation in proportion to the magnitude of institutional stakes. Quite to the contrary. In the context of parliamentary systems, competitive regimes practice censorship to a similar degree, and exclusion to a lesser degree, than in presidential contexts. Furthermore, they rig elections to semi-effective legislatures less than those to ineffective assembliess (see Figure 8.7). These counterintuitive findings may derive from a scissor effect of misaligned incentives and opportunities. In the presence of presidential systems and ineffective legislatures, competitive regimes may face lower incentives, yet higher opportunities, to manipulate elections. Thus they manipulate. Because it is easy, even if not quite important. Opportunity makes a thief, not necessity.

C ON C LU S I O N In this chapter, I analyzed strategies of manipulation as tools that authoritarian rulers employ to contain threats to power and protect the stakes of power. The more insecure their grip on power, and the more they have to lose when losing power, the more they manipulate. So far so simple. So far, in essence, the conventional account of authoritarian manipulation. I introduce theoretical complications by extending the causal chain into the future and by admitting the possibility that threats turn endogenous to manipulation. Authoritarian efforts to contain emergent uncertainties (“first-order threats”) can provoke the emergence of future uncertainties (“second-order threats”). Less so, I hypothesized, in hegemonic regimes than in competitive ones, and more so when rulers resort to more severe and more visible strategies.

290

Empirical Explorations

I conducted my empirical analyses faithful to the spirit of exploration and conscious of the fact that my analytical rucksack carries a theory that is more complex than my data. The resulting mix of confirming evidence, negative findings, and disconfirming results lends the chapter an air of theoretical excess. A surplus of theory appears to encounter a deficit in empirical confirmation. Still, even if the chapter has not spoken (nor would ever dream of speaking) the last word on the logic of electoral manipulation, it does contain at least three broad and broadly important findings. First, electoral manipulation indeed responds to short-term variations in observable threats, yet it does so in a highly differentiated manner. Authoritarian reactions to threats differ by strategies, types of threat, and subtypes of electoral authoritarian regimes. If nothing else, these variegated findings confirm that the strategy of empirical and conceptual disaggregation makes sense. Authoritarian manipulation does not appear to respond to a simple uniform logic. Generalizing inferences across strategies, types of threat, and subtypes of regimes are likely to be misleading. Second, grosso modo, hegemonic regimes respond to rising stakes just as they respond to rising threats: by deepening repression and fraud. Competitive regimes, in turn, are more likely to respond with concessions, with retreating to lower levels of manipulation. These divergent patterns lend credence to the notion that these regimes choose their levels of manipulation within different configurations of first-order and second-order threats. Competitive regimes can be less confident than hegemonic regimes about their capacity to suppress present threats and they need to worry more about the risks of provoking future threats. Thus their greater proclivity to make concessions in times of uncertainty, rather than tightening authoritarian controls. Third, standard decision theory assumes that actors take their decisions on the joint basis of two parameters: the costs and benefits of available courses of action and the probability that these costs and benefits materialize. If the costs of losing power are high, yet any change to the status quo is highly improbable, rulers can let their guards down. If political life is uncertain, yet the costs of losing office are modest, they may relax their defense of the status quo as well. In short, rulers are supposed to calculate expected utilities. Neither mere expectations nor mere utilities should drive their actions. Yet they do, apparently. Both threats and stakes bear direct relations with levels of manipulation. We can infer that the average electoral autocrat values power strongly enough to respond to mere variations in threat and feels insecure enough to respond to mere variations in stakes. Their fundamental love of power renders them sensitive to its uncertainties, and their fundamental insecurity in power makes them protective of its privileges.

The Calculus of Electoral Manipulation

291

Notes 1. Classic utility theory actually does not serve explanatory purposes. It is a descriptive device for representing actor preferences over decision outcomes: It is not the utility of outcomes that determines actor choices, but actor choices that reveal the utility of outcomes (see Morrow 1994: 33). 2. The slave economy of the Gulag did generate substantial contributions to the Soviet economy, though (see Figes 2007: 111–114, 423–427, 467–468). On the “anti-utilitarian senselessness” of totalitarian violence, see Arendt (2004: 620–623). 3. “Wovon man nicht sprechen kann, darüber muß man schweigen” (Tractatus logico-philosophicus § 7). 4. My underlying notion of explanation is Weberian: The social sciences strive to comprehend social action and thus to explain it causally by deciphering its meaning (see Weber 1985: 1). 5. Notable exceptions are, among others, Lamounier (1989) on the destabilizing consequences of the two-party system decreed by the Brazilian military, Przeworski (1991) on the counter-intentional effects of authoritarian liberalization, Bunce (1999) on the regime-subversive long-term consequences of Soviet institutions, and Moustafa (2008) on the unforeseen strength of civilian courts in Egypt under Mubarak. 6. For instance, in its narrow sense (which I embrace throughout this book), the notion of political “repression” refers to state violations of physical integrity rights. Yet scholars often use it in a wider sense that goes beyond political coercion to include non-violent restrictions on civil liberties like media freedom (e.g. Davenport 1996, Escribà-Folch 2010). Similarly, in its narrow sense (which I embrace throughout), the notion of electoral “fraud” refers to partisan interference with the administration of elections. Yet scholars often use it in a wider sense that goes beyond the administrative arena and covers the entire repertoire of electoral manipulation (e.g. Magaloni 2010). 7. For insightful analytical syntheses, see Davenport (2007) and Rivera (2010). 8. In a recent study, though, Abel Escribà-Folch found that “[r]epression, in all its various forms, is effective” (2010: 20) in decreasing annual probabilities of losing power. 9. Given its assumed effectiveness, the costs of repression are commonly treated as negligible (if not in theory, in actual research practice). Beneath vague conceptions of the “costs” and “benefits” of repression lie vague conceptions of the “threats” that motivate it (see also Davenport 2007: 8). 10. For an insightful overview, see Jessen and Richter (2011). In his more general explanation of blatant fraud, Alberto Simpser too stresses the indirect and long-term consequences of electoral manipulation (2013). 11. Alvarez, Hall, and Hyde (2008b), Birch (2009 and 2011: Ch. 2), and Lehoucq (2003) offer critical syntheses of the historical as well as contemporary literature.

292

12.

13.

14.

15.

16. 17.

18.

19.

20. 21.

Empirical Explorations Donno and Roussias (2012) study the effect of electoral manipulation on the effective number of legislative parties. For general reflections on (the limits of) “decommunization” in Eastern Europe, see Holmes (1994). On party bans in Africa, see Basedau et al. (2007) and the special issue of Democratization 17/4 titled “Ethnic Party Bans in Africa” (2010). On the participation of Islamic parties in electoral politics, see the thematic cluster in the Journal of Democracy 19/3 titled “Islamist Parties and Democracy” (2008). The new literature on social inequality and the political economy of regime dynamics (the seminal texts are Boix 2003 and Acemoglu and Robinson 2006) assumes that authoritarian regimes are exclusionary by nature. As institutions of minority rule they keep the majority out of the political arena. This literature does not expect purely politically motivated (rather than socially grounded) acts of exclusion nor does it consider variations in the scope and targets of exclusionary policies within authoritarian regimes. On the politics of selective exclusion in Arab countries, see Lust-Okar (2005). On Zambia, see Bratton and Posner (1999: 392–403), on Côte d’Ivoire, Crook (1997), on Russia, Fish (2005: 61–67). Although electoral manipulation primarily responds to challenges of authoritarian survival, it often carries secondary effects on problems of authoritarian governance. These effects can be constructive, as when the repression of vertical threats lays the foundations of state building (Slater 2010). They can also be self-subversive, as when censorship aggravates problems of agency control in the relations between central state authorities and subordinate public officials (Egorov, Guriev, and Sonin 2009, Lorentzen 2009). See Chapter 7, “Irritating Signals,” pp. 220–222. To the extent that public officials (a) are democratically minded and (b) realize that their government is acting undemocratically, manipulation may also dry up the regime’s normative support within the state apparatus. Of course, manipulation does more than create legitimacy costs. For opposition actors, it also decreases the expected utility of electoral participation. Barking and biting, the dog of manipulation may scare away its competitors and have their cake too. On the costs of repression, see, for example, Bellin (2004: 146), Davenport (2007: 17), Gartner and Regan (1996: 276–277), Goldstone and Tilly (2001: 187), and Lindberg (2009d: 320). For an explicit justification and graphical representation, see Gartner and Regan (1996: 278). Note that the notion of U-shaped costs of repression (as well as my conception of escalating second-order threats at high levels of manipulation) is incongruent with the inverted-U curve the literature on contentious politics holds to exist between repression and social mobilization. Social protest is highest at intermediate levels of repression, lowest at low and high levels of repression (see Inclán 2008, Tarrow 1994). The apparent contradiction may arise from different perspectives on causal

The Calculus of Electoral Manipulation

22.

23.

24.

25. 26.

27.

28.

29.

30. 31. 32.

293

probability. The literature on social mobilization centers on probable outcomes of repression, while the literature on state repression highlights possible outcomes (even if in an implicit manner only). The rule is that repression suffocates mobilization (by raising its costs), the exception that it ignites mobilization (by raising its grievances). Time discount factors δ conventionally vary between 0 and 1. As future payoffs are multiplied by these parameters, they represent the value actors ascribe to future benefits relative to present ones. For myopic actors, future benefits are worthless (δ = 0). For patient actors, they are worth as much as present gains (δ = 1). The notion of “authoritarianism by stealth” paraphrases Laurence Whitehead’s description of the Mexican electoral transition as “democratization by stealth” (Whitehead 1995). Sure enough, the magnitude of discredit that state repression casts upon a political regime may vary according to the social status of its victims and the moral infrastructure of the citizenry (see Brinks 2006 and 2008). Results available from the author. Some comparative studies of state repression have explored the explanatory weight of elections in and by themselves (Davenport 1997, Escribà-Folch 2010). In contradistinction, I do not treat elections per se as threats, “clear, visible, and predictable” (Escribà-Folch 2010: 2), but expect two of their attributes (competitiveness and contention) to involve varying degrees of threat. To indicate the overall fit of regression models, I report, here as elsewhere, standard errors and R2 values for lineal regressions, and percentages of accurate classification (pac) and Cox & Snell R2 values for logistic regressions. Note that I only look at election-time variations in manipulation (occurring either in the election year or in the previous one). Relative to average levels of manipulation between elections, rulers may temporarily either tighten or relax authoritarian restrictions in the proximity of elections (see Booth 1995: 9, Davenport 1997: 521). Three events seem few, but only 33.7 percent of competitive and 17.5 percent of hegemonic elections in my sample are preceded by higher numbers of contentious events. On continuous versus categorical conceptions of regime utility, see Schedler (2010b). “Power is the ultimate aphrodisiac” (Henry Kissinger). If societal stakes matter for the choice of manipulation regardless of societal threats, they should matter even more if we take into account their interaction with societal threats. The emergence of palpable threats should intensify the authoritarian defense of high stakes, just as the presence of high stakes should intensify the defensive manipulation against emergent threats. They should, yes. But they don’t. At least not according to my thin data. Rather than mutually reinforcing, stakes and threats seem to be mutually debilitating. Interactive regression models (not reported here) explain less than the simple models. The signs and directions of change are rather erratic. Some coefficients turn non-significant,

294

Empirical Explorations

their magnitudes remain minuscule anyway. There is no coherent story to be told (results available at request). 33. I exclude presidential elections, since the institutional features under discussion can be expected to affect the stakes of legislative competition only. 34. With two exceptions: (a) Among hegemonic regimes, mean levels of legislative repression also vary significantly between systems of government (p = .026). The pattern is U-shaped, with semi-presidential systems being less repressive (ø = 1.4, N = 5) than either presidential (ø = 3.6, N = 25) or parliamentary systems (ø = 4.0, N = 17). (b) In concurrent elections, hegemonic regimes admit (or commit) more administrative regularities, if not outright fraud, in the organization of elections (ø = 1.08 in concurrent, N = 17, and 0.71 in non-concurrent elections, N = 13, p = .077). Competitive regimes, by contrast, show lower levels of media control in concurrent elections (ø = 4.30, N = 30) than in non-concurrent contests (ø = 5.06, N = 32, p = .097).

9 The Calculus of Electoral Protest Politics, as the old saying has it, is the art of the possible. This truism rings particularly true for opposition politics under authoritarian rule. In electoral authoritarian regimes, opposition parties take part in a game that is designed to defeat them. They enter electoral contests under biased rules that they have neither devised nor consented to and that are binding only on them, but not their adversaries. They fight an uphill battle against a state that commands superior organizational, communicative, economic, and coercive resources. And yet. They do struggle to grasp and expand the elusive opportunities that competitive regimes offer and to create the tenuous possibilities that hegemonic regimes deny. Within the nested game of authoritarian elections, the conflictive interaction between government and opposition unfolds within the parameters as well as over the parameters of the game. Electoral authoritarian rulers respond to changes in their environment by changing their environment. Opposition parties struggle to accomplish the same. The former strive to reduce the uncertainties of electoral competition, the latter to augment them. At the meta-game of authoritarian elections, both are caught in the inherent circularity of institutional struggles. To achieve their goals, they have to adapt to their political environment. Yet their goals are transformative. They strive to alter the very environment they must adapt to. If we wish to explain the choices that actors make at the meta-level of electoral competition, we need to consider the interactive relationship between contextual constraints and contextual interventions. We need to understand the reactive nature of contentious action as well as its constructive role. The empirical explorations of the previous chapter revolved around the “second-order uncertainties” that authoritarian rulers need to contain when trying to contain emergent “first-order” uncertainties through manipulation. The empirical explorations of this chapter will center on “second-order opportunities” opposition actors wish to generate when responding to given “first-order” opportunities through protest. When deciding to boycott or protest an authoritarian election, opposition actors need to observe existing constraints and incentives. At the same time, they strive to redraw these constraints and incentives by calling attention to the authoritarian nature of the election

296

Empirical Explorations

(the informative role of protest) and by deepening emergent challenges to the regime (the provocative role of protest).

THE NATURE OF ELECTORAL PROTEST Given their characteristic mix of repression and tolerance as well as their contradictory impulses towards the mobilization of citizens (pushing them into the private realm one day and dragging them into public participation the next), electoral authoritarian regimes tend to give rise to contentious politics in multiple realms, across a wide range of actors and issues. Collective contention in these regimes often bears no ostensible relation to electoral politics. At times, it does not even bear relation to opposition politics but is sponsored by actors within the state or the ruling party as a means of competition within the authoritarian elite.1 In this chapter, I do not look at contentious politics in general, but more specifically at opposition mobilization against authoritarian elections. Electoral protest is a narrow subcategory of political contention. Unlike other forms of social protest, it does not pursue policy change, but polity change, a transformation of the basic rules governing the access to power. It does not unfold outside the electoral arena, but forms part of the two-level game of electoral competition. Commonly it does not take place within an autonomous sphere of non-partisan civil society, but is embedded within the party system. It is not a spontaneous expression of discontent from below, but one led by opposition parties and candidates. It is not an instrument of authoritarian mobilization either, but an instance of opposition protest against the regime. In synthesis, we are talking about election-centered protest orchestrated by the counter-elite of opposition actors. Much of the literature on political protest revolves around problems of collective action. How can protest movements prevent discontented individuals from surfing the waves of mobilization for free? How can they bring them to participate in collective protest if it does not make any perceptible difference to the outcome whether they individually do so or not and if they can avoid the costs of failure and still enjoy the fruits of success by simply staying at home? Electoral protests, just like any other form of collective contention, face deep problems of coordination. Most opposition leaders who mobilize protest against authoritarian elections are party leaders. They have some minimal organizational infrastructure at their disposal and thus enjoy a competitive advantage over movement entrepreneurs who lack organizational resources. Their organizational advantages, however, do not redeem opposition leaders from all problems of collective mobilization.

The Calculus of Electoral Protest

297

Before opposition actors reach the gates of authoritarian electoral arenas and ponder their strategic options of boycott and protest, citizens as well as elites have to pass through a long chain of individual decisions: whether to remain in the private realm or participate in politics; whether to support the ruling party or work against the regime; whether to take up arms or embrace peaceful means of resistance; whether to partake in spontaneous protest or joint political organizations; and whether to practice the art of association in the civil realm or support a political party. It is only at this point, when political parties already exist, that the decision calculus of electoral protest swings into motion: whether to boycott or partake in electoral contests called by the authoritarian regime; and whether to play within the confines of formal institutional politics or take to the streets in protest against electoral processes, be it before or after election day. Figure 9.1 sketches the lengthy decision-path that leads from the comforts of private live to the agonies of electoral participation. While the emergence of opposition parties constitutes an important puzzle of authoritarian electoral competition (see Greene 2007), explanations of

Citizen choices

Political participation

Private life

Anti-regime

Violent

Pro-regime

Peaceful

Organized

Unorganized

Political parties

Civil society

Electoral participation

Acquiescence

Electoral boycott

Protest mobilization

Figure 9.1. The chain of participatory choices

298

Empirical Explorations

electoral protest already presuppose their presence. Both the decision to exit electoral competition (boycott) and the decision to challenge the electoral process (active protest mobilization) are taken by opposition elites who control at least embryonic forms of party organization. The main task therefore is not to explain why party members join electoral protests, but when party leaders convoke electoral protests. When do they mobilize their followers to protest an authoritarian election? When do they demobilize them and protest by withdrawing from the electoral arena? However, even when opposition leaders control capable party organizations, they still need to mobilize citizens beyond the small nucleus of party activists. Their calls for collective action have to make sense to their followers. When they call voters to take to the streets, they need to offer compelling reasons that make them heed their calls. It is easy to see why opposition parties reject authoritarian election processes and outcomes. It is less clear why they should do anything about it. In principle, genuine opposition parties have strong normative as well as utilitarian reasons for refusing to embrace the results of manipulated elections.2 On the normative side, democratic opposition parties are supposed to accept the uncertain outcomes of democratic elections, yet not the prearranged outcomes of authoritarian elections. In electoral autocracies, losing parties have no normative reasons to accept the verdict of the ballot box. On the utilitarian side, infringements of democratic minimum standards do more than offend the moral sensibilities of opposition actors. They hurt their vital interests. Everything else being equal, each authoritarian norm violation decreases their chances of political success. For opposition parties, the struggle for political opening is often a matter of moral convictions—and always a matter of political survival. Democratic principles and organizational self-interest are neatly aligned. Authoritarian elections thus are sources of political grievance. They provide normative as well as utilitarian reasons for rejecting the political status quo. Notoriously, though, grievances alone are never sufficient to propel actors into contentious action. In and by themselves, neither common interests nor common sentiments of injustice push aggrieved parties from quiescence to protest. Why should opposition parties go beyond rhetorical declarations of discontent? Why should they resort to active electoral protest? Why should they believe that either their withdrawal from electoral competition or the mobilization of street protest helps them change the authoritarian status quo? In Alfred Hirschman’s terms (1970), why should they hold either exit (election boycott) or voice (protest mobilization) to be effective instruments of change? To cut my long answer short: because they serve to publicize grievances and to broaden opportunities.

The Calculus of Electoral Protest

299

T H E T WI N ROL E O F E L E C TO R A L PROT EST By protesting against authoritarian elections, opposition actors may strive to transform their non-democratic quality through either electoral reform or electoral reversion. (a) Electoral reform changes the institutional foundations of elections. It involves procedural concessions that affect institutions of electoral governance (the rules and practices governing the organization of elections). It may take the form of either de jure changes in formal institutions or de facto changes in actual practices of state agents. (b) Electoral reversion overturns the official outcomes of elections. It involves substantive concessions that affect either the overall outcome or partial results of national elections. In electoral revolutions, mass mobilization in the aftermath of a “stolen election” (Thompson and Kuntz 2004) succeeds in toppling the central government.3 Given their temporal logic, boycotts as well as pre-electoral protests should aim primarily at electoral reforms, post-electoral protests at electoral reversions. Yet why should we expect either types of protest to achieve these goals? Why should we expect public protests against flawed elections to contribute to either the enactment of “wider political reforms” or “the removal of the authoritarian player” (Cavatorta 2009: 139)? What would be the causal mechanisms that make authoritarian rulers respond to protest mobilization either by democratizing concessions or by democratizing exits from power? Despite the vocabulary of threat and challenge that is common in the study of political protest, contentious actions in and by themselves seldom constitute a “clear and present danger” to ruling elites. Whatever the final goals electoral protests wish to achieve, they cannot achieve them directly. They can only achieve them obliquely, through the pursuit of intermediate goals. In the two-level game of authoritarian elections, opposition parties do not have direct access to electoral conditions or to electoral outcomes. If they wish to influence the former (through electoral reform) or the latter (through electoral reversions) they need to influence those actors that do have direct access: governments, state agents, and citizens. To do so, I claim, they have to engage in the twin politics of uncertainty: in the competition over opacity as well as in the competition over insecurity. They need to shape public beliefs about the regime (the informational role of protest) and they need to induce further threats against the regime (the provocative role of protest). The causal chain I have in mind looks rather simple: The generation of information and the provocation of threats alter the prevailing balance of power and thus induce the government to grant (or deny) procedural or substantive concessions. Figure 9.2 illustrates the causal circle that starts anew when governmental responses (concession or regression) feed a fresh cycle of protest (or acquiescence).

300

Empirical Explorations

Opposition protest

Regime relaxation or tightening

Changing balance of power

Information

Provocation

Figure 9.2 Opposition protest and regime change

Changing Beliefs: The Informative Role of Protest Within the structural opacity that characterizes authoritarian elections, electoral opposition protests can fulfill a double informational role. They can convey information about both the regime (its nature and strength) and the opposition (its nature and strength). Information about the regime. Electoral protests can shape prevailing beliefs about the regime. (a) Regime nature: By staging public protest against flawed elections, opposition actors try to inform domestic as well as international audiences about the non-democratic nature of the regime and the elections it organizes. By denouncing authoritarian maneuvers they also strive to establish a causal story: it is not the will of the people, but regime manipulation that accounts for their defeat at the polls. Protests are public declarations of electoral illegitimacy that counter governmental discourses about democracy at work. (b) Regime strength: Often protests also provide hints about the strength of the incumbent. For instance, if a government prohibits protest or threatens dissidents, but is unable or unwilling to make its threats and prohibitions effective, observers will read the occurrence of protest as a symptom of governmental weakness. Information about the opposition. Electoral protests also generate information about the opposition. (a) Opposition nature: Electoral protests contain clues about both the ideological and tactical nature of opposition parties: their

The Calculus of Electoral Protest

301

commitment to political democratization, their accommodative or confrontational profile, their disposition to use peaceful, disruptive, or violent means of protest. (b) Opposition strength: Protests also convey information about the strength of opposition parties: their mobilizing capacity, their organizational and communicative skills and resources, their degree of unity and coherence, the degree of control that leaders command over their followers, their ability to form alliances and their wider resonance in civil society.4 Of course, the elements of “information” that opposition protests convey do not eliminate the opacities that surround authoritarian elections. Protesters are not impartial senders that broadcast objective pieces of factual information to neutral receivers. Rather, they are engaged in strategic attempts to “frame” electoral processes. Given the adversarial nature and epistemic uncertainties of authoritarian elections, all framing efforts are inevitably controversial. Whatever opposition parties try to communicate through contentious action has to assert itself against competing constructions of political reality. The meaning of boycotts, above all, is subject to inescapable controversy. Opposition parties who participate in a flawed election under protest are vulnerable to charges of inconsistency. Why do they take part in a competitive game they denounce as unfair? If they protest before the election, they face pressures to draw the logical conclusions and withdraw. If they protest afterwards, they look like bad losers who create a fuss because they can’t stand losing. Boycotting parties appear more coherent, but their logical consistency does not save them from problems of political credibility. Critics are bound to advert that boycotting parties do not declare the electoral system bankrupt, but reveal themselves to be electorally broke. We have to remain alert to the possible misrepresentations of reality that opposition parties offer. In general, though, despite some caution, comparative scholars tend to embrace the idea that opposition parties protest and boycott elections because they have good normative reasons to do so. Protests against fraud are not fraudulent themselves. It is neither greed (the pursuit of patronage) nor narcissism (the offense at the lack of affection by voters), but genuine grievance (the observation of authoritarian manipulation) that drives opposition parties. If they boycott or protest, they do so “to publicize fraud” (Beaulieu 2006: 41).5

Activating Threats: The Provocative Role of Protest In isolation, street demonstrations are seldom massive and menacing enough to constitute a serious threat to regime survival. Authoritarian governments

302

Empirical Explorations

are often able to sit out cycles of electoral protests without major adjustments. As I hypothesize, non-violent opposition protest becomes genuinely threatening only to the extent that: it induces significant shifts in voter preferences, generates a massive cascade of contention, spills over into violent protest, reinforces the activities of pre-existing rebel groups, triggers elite defection at higher or lower levels, or sparks military intervention. In other words, electoral protest turns threatening only to the extent that it deepens either vertical or horizontal threats. Deepening vertical threats. Contentious action can be self-reinforcing. Peaceful protests may generate more peaceful protests, but they may also spill over into violent forms of contention: (a) Reinforcing peaceful protest: Explanations of contentious action are dynamic. They take into account that public protest possesses self-reinforcing qualities. Common metaphors such as cycles, spirals, thresholds, tipping games, bandwagons, and cascades of protest reflect the much-observed fact that “collective dissent begets more collective dissent” (Lichbach 1998: 63). Peaceful election-centered protest today may breed peaceful protest outside the electoral arena tomorrow (and inside as well). (b) Triggering spontaneous violence: Even when leaders proclaim their commitment to peaceful means of expression, non-violent mobilization often contains “the possibility of violence.” As a matter of fact, its threatening quality derives from the “absence of predictable limits” (Tarrow 1994: 101–102). When passions run high, electoral protest may give rise to (more or less) spontaneous and disorganized violence. (c) Intensifying organized violence: Non-violent challengers must be cautious in not getting themselves associated with violent dissent. Still, the depth of their grievances and the breadth of their popular support may serve armed movements as a public justification to intensify their aggressions against the state. If ballots fail as a pathway to power, bullets may seem like a legitimate alternative (see Lehoucq and Molina 2002, Rapoport and Weinberg 2001: 28–30, Slater 2010: 104–105). Deepening horizontal threats. Vertical threats feed vertical threats, but they also provide nourishment to horizontal threats, either by waking dormant or by reinforcing manifest uncertainties: (a) Provoking high-level elite defection: Ruling parties face the perennial risk of ambitious regime insiders defecting to the ranks of the opposition. As long as dissidents face dim chances of success outside the ruling party, they will tend to remain loyal to the regime (see Langston 2006). Yet, when the independent opposition starts attracting visible popular support, the calculus of exit from the electoral authoritarian coalition changes. Opposition parties that are able to mount massive public challenges against an authoritarian election may convince regime members that the time has come to leave the rocking ship. “Political protest threatens to undermine elite cohesion” (Robertson 2010: L 304). (b) Provoking lower-level elite defec-

The Calculus of Electoral Protest

303

tion: Just as opposition protest may generate tensions at the high echelons of state power, it often introduces tensions into lower levels of state bureaucracies. Mass protest may erode the willingness of soldiers, police officers, and civil servants to secure the administrative foundations of a regime that mistreats their co-citizens. It may push them into “disaffection and noncooperation” (Sharp 2010: 63). Again, these levels of institutional uncertainty interact. Signs of disobedience at the bottom of the state may induce dissidence at its top, and the other way round. (c) Inviting military defection: In the last instance, the survival of electoral authoritarian regimes in the face of either vertical or horizontal challenges depends on the military and the security forces. If the agents and agencies that administer the state monopoly of physical force withdraw their support from a government under siege, the government falls from the tree of power like Newton’s apple. Electoral autocracies open up venues for processing and diffusing conflicts that reduce their exposure to military coups (see Cox 2007). Yet, the organization of contentious authoritarian elections does involve the risk that civilians invite the military to intervene into the electoral arena, and that the military feels entitled to do so, in particular when it observes a downward spiral of tightening authoritarianism and deepening illegitimacy (see Clark 2006).6 Comprehending electoral protest as a strategy opposition parties deploy in the twin struggle over uncertainty resonates well with many bits and pieces of theoretical reasoning and empirical evidence in the democratization literature. Yet, the idea of dissidents who struggle to publicize electoral grievances and to deepen vertical uncertainties does not sit easily with the literature on political contention and collective action. The sociological literature on contentious action conceives opportunities as causal forces, not goals, of contentious action. The economic literature holds both grievances and opportunities to be irrelevant.

R ET H I N K I N G C OL L E C T IV E C O N T EN T I O N The study of political protest has been dominated by two contrasting approaches: the economic tradition of Mancur Olson and the sociological tradition of Charles Tilly and others. Scholars working in the former tradition have analyzed individual disincentives for collective action, placing emphasis on the improbability of mobilization. Scholars working in the latter tradition have studied the social logic of political protest, placing their emphasis on the possibility of mobilization. While sympathetic to core insights of both approaches, my conception of the protest politics of uncertainty differs from both in non-trivial ways.

304

Empirical Explorations Rethinking the Logic of Contentious Politics

Few areas of political research possess anything resembling a unifying paradigmatic core. Exceptionally, the study of social movements does so. Students of social movements basically know what is needed to get protesters onto the streets. They know what to look for if they want to explain the emergence of contentious collective action: (a) grievances (appropriately framed to resonate with relevant audiences); (b) repertoires of collective action (such as public demonstrations, strikes, and street barricades); (c) mobilizing structures (such as social networks or formal organizations); and (d) political opportunities (positive expected utilities of collective action under given circumstances, a favorable balance between expected success and expected repression).7 My approach to electoral protest highlights the first and the last elements: grievances and opportunities. Yet, it grants them a twin status: as independent as well as dependent variables. Grievances provide moral justifications for collective protest: we need to do something. Opportunities provide strategic justifications: we can achieve something. Neither of the two are objective. Both are a matter of discursive construction (“framing”). Protest scholars know this well, yet still tend to treat them primarily as explanatory factors, rather than strategic goals. In a shift of emphasis, I contend that we should understand both the perception of grievances and opportunities and their creation as purposeful goals of political protests, not just as causal parameters. Since both grievances and opportunities are prominent causal factors in the emergence of social protest, opposition leaders have a practical interest in shaping them actively, rather than just responding passively to their fortuitous presence. Electoral protest arises when the conditions are conducive to the rise of protest. Yet these conditions are partially endogenous to protest itself. While existing grievances and opportunities can be expected to exert a powerful influence on collective protest, collective protest can also be expected to exert a powerful influence on perceived grievances and opportunities. To the extent that actors exploit this circular relationship, political protest appears as a conscious effort to produce the conditions that are conducive to political protest. It is driven by the strategic goal of raising its own chances of success.8 As Mark Lichbach states in his encyclopedic treatment of the dilemmas of collective action that political rebels confront, “the fundamental competition between the regime and the dissidents is over solutions to the Rebel’s Dilemma” (1998: 26). That is, the main concern that drives the politics of collective action is to manipulate the parameters of collective action. Governments and rebels alike turn key parameters of individual decision-making, like the depth of political grievances and the probability of success of opposition protests, into fiercely disputed variables.

The Calculus of Electoral Protest

305

In conclusion: Contentious action responds to grievances and opportunities, yet grievances and opportunities also respond to contentious action. We should therefore expect protest entrepreneurs to be adaptive to “first-order opportunities”—yet also alert to “second-order opportunities,” that is, to opportunities of communicating grievances and creating opportunities.

Rethinking the Logic of Collective Action According to microeconomic analyses of collective action in the tradition of Mancur Olson (1965), powerful countervailing incentives tend to deter individuals from participating in the construction of public goods. Four broad factors generally conspire to deter coordination among actors who share a common interest in joint action: (a) The private costs of participation: Political participation carries personal costs in terms of time and resources. In authoritarian regimes, it often involves high personal risks in terms of physical integrity and economic well-being. (b) The private value of public goods: Public goods, such as roads, parks, or clean air, are conventionally defined as indivisible. Mancur Olson assumed them to be divisible. Each participant enjoys only a fraction of the public good. The private utility of the public good depends on the size of this fraction, which depends on the size of the overall group. The larger the group, the smaller the personal benefit each group member derives from the public good.9 (c) The individual contribution to success: The successful production of public goods depends on participation rates, yet participation rates do not noticeably depend on single actors. Each individual decision whether to participate or not carries only negligible marginal effects on the overall probability of collective success. (d) The private consumption of public goods: Public goods are non-excludable. Citizens can enjoy them whether they participate or not in their production. When participation is costly, when private benefits from common enterprises look modest, when individual participation seems to matter little, and when the fruits of collective success are available to all, rational actors should abstain from participating and ride the train of history for free (unless someone purchases their participation through the provision of selective goods). In this account of the logic of collective inaction, neither grievances nor opportunities matter. Why do grievances matter? The relative irrelevance of political grievances stands at the very core of Olson’s analysis: common interests do not lead to common action. However, even if it is easy to see (after Olson) that grievances are not sufficient to ignite political protest, it is difficult to overlook that they do tend to form a necessary condition. Most people who take to the streets to

306

Empirical Explorations

challenge political authorities do not act out of a sense of bored satisfaction, but in response to experiences of hardship and injustice (see e.g. Robertson 2010: L 1887). Although grievances seem alien to an Olsonian approach, they can be easily integrated. It suffices to adjust one theoretical premise: the idea that individuals enjoy only fractions of public goods. The sociological terminology of “grievance” sounds alien to the ears of economists. Its frame of reference is justice, not utility. Its underlying meaning, though, lies close to the concerns of public economics. Collective action is about the creation of public goods. Political grievances point to their absence. They are, simply, demands for the provision of public goods. Their irrelevance within Olson’s microeconomic framework derives from the small private utility it assigns to public goods. If we shed the assumption that the enjoyment of public goods is divisible, the motivating potential of political goals comes to the fore. Let us start by clarifying the counterfactual meaning of grievances. The demands for public goods we describe as political grievances (g) are not simple expressions of present suffering. They measure the distance between the utility of the present (up) and the utility of possible futures (uf ).10

g = uf–up

(9.1)

To gauge the intensity of political grievances it is not sufficient to see how bad people are (or think they are), but how good they can get (or think they can). If the status quo is bad, but a better future impossible, grievances are low. In the absence of viable remedies, there is nothing to complain about, nothing to demand. The notion of grievances rests upon the secular idea that the future lies in the hands of men, not God or fate or chance. If we hold our suffering to be inevitable (or self-inflicted), we will be in pain, but not aggrieved. The utility of the better future that actors strive to build through collective protest depends on the utility of the public goods they strive to produce. As mentioned before, Olson took the enjoyment of public goods to be divisible. As he assumed, citizens do not enjoy public goods in their entirety. Within a community of N members, each individual only suffers a fraction –1/N from the absence of public goods, and enjoys a fraction 1/N from their presence. In large societies, both expressions are bound to be very small, and so is the intensity of personal political grievances—the difference in personal utility between the public bads of today and the public goods of tomorrow. As Olson’s analysis focused on economic organizations like labor unions and employer associations and economic goods like levels of remuneration and taxation, assuming the divisible consumption of public goods makes sense. Each worker enjoys only a fraction of a general pay raise, each firm only a fraction of a general tax cut. In the realm of political goods like democratic

The Calculus of Electoral Protest

307

liberty, however, this assumption rings implausible. A political prisoner in the torture chamber does not suffer a minuscule fraction of authoritarian repression, but its whole weight. A critical citizen expressing herself in a free public space does not enjoy a minuscule fraction of civil liberties, but freedom tout court. Citizens’ personal grievances against the status quo therefore are much more intense, and their individual gains from collective action much larger, than conventionally assumed. Hence, their calculus of participation may very well not be dominated by cost considerations. Rather than being negligible, the personal benefits of public goods may outweigh the personal risks of participation.11 Why do opportunities matter? Within an Olsonian framework of analysis, the “probability of winning is not supposed to matter” either and “yet the evidence that it does is overwhelming” (Lichbach 1998: 84). Time and again, students of revolution, social movements, and regime change have emphasized the crucial role that perceived possibilities of change play in the calculations of opposition actors. Revolutionaries take up arms against weak and vulnerable states, not strong and unassailable ones. Social movements take advantage of windows of opportunity, rather than banging their heads against fortified walls. Protests against authoritarian rule gain momentum when regimes start looking divided and hesitant, rather than monolithic and resolute.12 Again, although probabilities of success seem alien to an Olsonian approach, they can be easily integrated. It suffices to adjust one theoretical premise: the idea that individuals (always) dismiss their own marginal contributions to the production of public goods. In standard utility theory, probabilities of success are central parameters of decision-making: actors do not calculate the costs and benefits of alternative courses of action, but their expected costs and benefits. Olson could afford to ignore expectations of success thanks to four (mostly implicit) empirical assumptions: 1. The success of collective action is unproblematic. Collective enterprises vary widely in their chances of success. Simple tasks of work coordination are certain to succeed, if only cooperation comes forward. We want a clean city? We have to stop throwing garbage on the street. By contrast, in situations of social conflict, the success of collective action is often much less predictable. Whether the cooperation of friends is successful depends on the cooperation of their foes. Olson studied the organization of economic interests and thus collective enterprises whose expectations of success are solidly grounded in underlying structures of power. The power of labor organizations derives from the structural power of workers, the power of business associations from the structural power of capital. If we know these structural foundations, we know the probable payoffs of collective action.

308

Empirical Explorations

2. The expected utility of public goods is negligible. As Olson assumed the private utility of public goods to be small, their expected utility is bound to be small too, regardless of whether collective action is certain to succeed (p = 1) or doomed to fail (p = 0). 3. The costs of participation are modest and stable. Olson pondered the modest costs of participation in democratic contexts (in terms of time and money) rather than the imponderable risks of participation under authoritarian conditions (in terms of death and imprisonment). The former are entirely predictable, the latter intrinsically uncertain. 4. Individuals discount their personal contribution to the success of collective action. Within Olson’s framework, the chances of success of collective ventures do not matter much since individual actors can do little to affect these chances. The marginal impact of individual participation on the overall probability of success of collective action is close to zero. From the viewpoint of each individual, it makes no perceptible difference for the fate of the joint enterprise whether she abstains or participates. Therefore, even if collective success is close to certain, she should abstain. By abstaining she avoids the personal costs of participation without renouncing the fruits of collective success. She can take the ride for free without visibly hampering the collective journey towards the public good. In utilitarian terms, the self-interested calculus of the free rider is (almost) impeccable and almost invincible.13 Without doubt, it has strangled and drowned countless enterprises of collective action. However, if we admit the possibility of moral calculations, guided by the imperatives of responsibility, rather than utility, the marginal irrelevance of individual participation looks less paralyzing for our personal decisions. Under normative considerations, if we abstain alleging that it does not matter whether we participate or not, we act against the demands of fairness and equality. We are avowedly unfair to those who carry the burden of participation for all others, including ourselves. And we are tacitly anti-egalitarian towards our co-citizens if we find depressing what is no more than a simple democratic fact: that ours is just one voice among many. Our contribution to the overall welfare is minuscule—yet it is the only contribution that we can make. Morally, in a liberal society of co-equals, we should not aspire to more. In a republican spirit of common responsibility, we should not aspire to less. The moral calculus of individual responsibility should lead us to contribute our grain of salt, to do what we can do, within the scale of human possibilities. Besides God and the dictator we wish to tumble, only megalomaniac minds should find it frustrating that they cannot determine the course of complex societies in which we are supposed to take part in equal shares.

The Calculus of Electoral Protest

309

Once we accept that our individual contribution to the success of collective enterprises is minuscule—and still consider participating, then we need to start calculating their prospects of success. Even if the modesty of our personal role does not prevent us from joining common projects, no one should expect us to participate in collective efforts that are doomed to failure. In particular when the risks of participation are high, no one should expect us to sacrifice ourselves for futile causes (although some may well choose to do so). To sum up our theoretical excursus into sociological and economic analyses of political protest: With some, rather minor adjustments of analytical lenses, both sociological and microeconomic perspectives on collective contention license my interrelated claims that political grievances matter and probabilities of success, too, which is why protest leaders should struggle to influence both—by shaping intangible perceptions (the informational role of protest) and by provoking tangible threats (the provocative role of protest). In the subsequent empirical parts of this chapter I will trace the informational use of electoral protest by exploring its association with electoral manipulation. As I hypothesize, the informational value of protest varies with the severity and visibility of manipulation. Opposition actors should be more likely to protest when they confront more severe and the more opaque strategies of manipulation. I will trace the provocative role of electoral protest by exploring its association with preceding threats. As I hypothesize, the prospects of activating institutional threats through protest depend on the intensity of antecedent threats. Opposition actors should be more likely to protest when they can build upon a history of challenges from below. My insistence on the “constructive” role of protest is meant to complement, not to supersede, its “reactive” logic. Even when protest movements aim at shaping their environment, they also respond to their environment. They respond to given opportunity structures as well as to given incentive structures. To estimate the weight of given opportunities, I will explore the association of electoral protest with elements of its societal context. To estimate the weight of given incentives, I will explore its association with elements of its political and institutional context. I expect these societal and political contexts to play different roles with respect to the two basic strategies of opposition protest that I will be examining: voice (active mobilization) versus exit (election boycott).14 Through boycott as much as through active mobilization, opposition actors try to convey a message of protest against the election and convince others to follow suit. Boycott and mobilization may thus fulfill similar “constructive” roles. Yet, they should obey opposite “reactive” logics. When parties withdraw from electoral competition, they demobilize their followers, rather than mobilizing them (even though they tend to redirect their mobilizing energies from

310

Empirical Explorations

election campaigns to boycott campaigns).15 If external conditions for mobilization are good, opposition parties should mobilize. If they are bad, they should demobilize, i.e. boycott. Furthermore, as boycotting parties renounce the potential rewards of electoral participation, election boycotts involve opportunity costs that opposition parties can avoid if they protest, yet still participate.

S O CI ETA L OP P O RT U N I T Y S T RU C T U R ES If common explanations of political contention highlight the political opportunity structures that protest movements face, I will start my empirical explorations of opposition protest against authoritarian elections with a more distant set of contextual factors: the societal opportunity structures that opposition actors face. I use the notion of societal opportunity structures to denote environmental resources outside the realm of politics that opposition actors can mobilize for the purpose of political mobilization. Some societal structures are symmetrical in their distributive implications: they affect contending groups of actors in similar ways. Others are asymmetrical: they favor some groups over others.16 Here I hypothesize economic wealth to benefit the opposition, economic growth to harm it. Societal cleavages as well as international linkages are likely to have ambiguous consequences for the societal resource balance between government and opposition.

Economic Development To the extent that levels of economic development reflect levels of societal modernization, they are bound to carry complex and often contradictory consequences for the mobilizing capacities of opposition parties. Overall, though, I expect economic wealth to favor opposition mobilization. I use three thin indicators: average wealth (the log of current ppp per capita income), basic education (literacy rates among the population over age 15), and communication infrastructure (domestic and cell phone lines per 1,000 inhabitants). Each of these three structural factors provides potential resources for opposition mobilization. Societal wealth is supposed to grant citizens the material independence they need to confront repressive states. Education is supposed to provide them with the skills and knowledge that enable them to participate effectively in public affairs. The infrastructure of communication facilitates horizontal communication among citizens that transcends face-to-face

The Calculus of Electoral Protest

311

meetings. While authoritarian governments can often easily control centralized one-way media like newspapers and television, they cannot as easily tap decentralized two-way systems like telephone communication. Thus I expect these measures of economic development to be positively associated with election protest and negatively with boycott. If the societal environment favors mobilization, opposition parties should mobilize; if it does not, they may consider boycotting. To establish whether these distant, contextual factors carry consistent effects on levels of opposition protest against either hegemonic or competitive elections, I regress my three measures of election protest on this small bundle of economic indicators: opposition boycott (participation, partial boycott, and full boycott), pre-electoral protest (acquiescence and active mobilization), and post-electoral protest (acquiescence, rejection, and active mobilization). Table 9.1 shows the results.17 As they suggest, economic development explains a decent share of variations in electoral protest, though not in an entirely consistent fashion. If anything, economic development seems to depress electoral protest. Literacy levels bear inconsistent relations with protest levels. Average wealth is unrelated to electoral protest in competitive regimes, yet considerably discourages it in hegemonic regimes. Access to phone communication is unrelated to electoral protest in hegemonic regimes, yet slightly discourages it in competitive regimes. Hegemonic parties, it appears, maintain the loyalty of their subjects by enriching them, competitive regimes by connecting them (before, one suspects, the appearance of web-guided political flash mobs).

Societal Cleavages In its narrow sense, the notion of political cleavages describes structural conflicts that have objective bases in society, subjective relevance for citizens, and organized expressions in politics.18 Here I will use it in a looser sense to indicate potential fault lines of societal conflict that do have a structural basis in society, yet may be neither experienced as conflictive by citizens nor treated as such by politicians. As a matter of fact, since the seminal essay by Seymour Martin Lipset and Stein Rokkan on West European cleavage structures and party systems (1967), contemporary political sociology has studied the politicization of societal conflict—the translation of structural cleavages into political discourse and political organization. Since cleavage structures are not self-expressive, we should not expect them to bear lineal, mechanical relations to political conflict. In electoral authoritarian regimes, cleavage structures serve as potential bases of political mobilization for both government and opposition actors and

Table 9.1. Societal structures and opposition protest Boycott Hegemonic regimes

Competitive regimes

Pre-electoral protest

Post-electoral protest

Hegemonic regimes

Hegemonic regimes

Competitive regimes

Competitive regimes

B

p

B

p

eb

p

eb

p

B

p

B

p

Economic development Wealth (log current ppp per capita) Education (% adult literacy) Communication (phones per 1,000) Constant Standard error / pac R2 / Cox & Snell R2 N

–1.213 .004 .000 4.430 .769 .184 42

.062 .600 .836 .022

.124 –.007 –.001 .651 .694 .100 76

.728 .132 .297 .545

.066 1.078 1.000 10.390 81.4% .153 43

.167 .037 .936 .682

1.471 1.014 .996 .114 51.9% .026 77

.708 .314 .222 .488

–.403 –.005 .000 2.189 .334 .397 42

.150 .125 .565 .010

.023 .001 –.002 .511 .392 .119 76

.908 .844 .018 .400

Societal cleavages Inequality (Gini) Linguistic polarization Religious polarization Population size (millions) Constant Standard error / pac R2 / Cox & Snell R2 N

.034 –.659 –.022 –.002 –.559 .739 .108 46

.087 .225 .966 .344 .473

–.004 –.146 –.431 –.002 1.013 .719 .041 84

.658 .584 .124 .326 .032

.998 .412 140.111 1.017 .018 83.0% .178 47

.977 .653 .058 .055 .267

.982 1.271 .125 .994 5.680 61.2% .079 85

.477 .757 .015 .436 .211

.000 –.664 –.157 –.002 .908 .372 ,204 46

.971 .018 .547 .154 .024

.001 –.175 –.206 –.001 .677 .420 ,039 84

.845 .262 .208 .588 .015

–.015 –.133 –.016

.565 .002 .384

–.020 –.035 –.003

.371 .176 .356

1.133 .840 .941

.232 .220 .317

.904 1.037 1.010

.112 .619 .257

–.012 –.079 –.005

.352 .000 .524

–.002 .833 –.027 .054 –.001 .411

Economic performance gdp p.c. growth (election year) gdp p.c. growth (5 years before ø) Inflation (election year)

Inflation (5 years before election ø) Constant Standard error / pac R2 / Cox & Snell R2 N

.027 .746 .722 .271 49

.138 .000

.000 .637 .748 .071 70

.466 .000

1.016 .432 78.0% .055 50

.786 .158

1.001 .722 59.2% .065 71

.553 .307

.011 .623 .348 .356 49

.227 .000

.000 .221 .586 .000 .395 .097 70

External dependence International trade (% gdp) Foreign direct investment (% gdp) Total debt service (% gni) Official dev. assistance (% gni) Constant Standard error / pac R2 / Cox & Snell R2 N

–.003 .067 –.005 .031 .533 .805 .099 48

.436 .230 .892 .072 .111

–.003 .010 .018 .005 .533 .738 .024 86

.302 .466 .489 .626 .035

1.007 .785 .989 .939 .405 79.6% .069 49

.450 .174 .916 .257 .362

1.003 .930 .963 .953 1.102 64.4% .051 87

.703 .265 .607 .169 .891

–.002 –.021 –.009 .011 .688 .379 .134 48

.295 .430 .589 .178 .000

–.002 .004 .028 .004 .482 .406 .067 86

.200 .566 .057 .511 .001

Note: ols lineal regressions for boycott and post-electoral protest (reporting unstandardized B coefficients, standard errors, and R2 values), binary logistic regressions for pre-electoral protest (reporting odds ratios eb, Percentages of Accuracy of Classification pac, and Cox & Snell R2 values). Shaded coefficients = statistically significant (p ≤ .15).

314

Empirical Explorations

therefore look ambiguous to both sides. To be successful in their democratizing endeavor, opposition actors need to establish the political anti-regime cleavage as the central axis of political conflict. Paraphrasing William Riker (1986), the struggle against electoral manipulation requires opposition parties to carry out the quintessential maneuver of political manipulation: to redefine the ideological space of conflict by introducing a new axis of competition. Discontent arising from class, religious, regional, or other societal cleavages may help the opposition to mobilize against the incumbent. At the same time, criss-crossing lines of division can also frustrate its efforts to build catch-all alliances against the regime.19 In sum, societal cleavages may encourage opposition mobilization, but they also facilitate divide et impera strategies by the regime. They encourage contention, but block coordination. To assess the net effect, I regress my three types of electoral protest on four proxies of societal cleavage structures: social inequality (Gini coefficient of household income), linguistic and religious polarization, and population size (millions). Mapping socio-economic disparities and cultural pluralism, the first three indicators are rather self-explanatory. The last reflects the old Madisonian idea according to which size correlates with social heterogeneity: “Extend the sphere [of society] and you take in a greater variety of parties and interests.”20 The regression results shown in Table 9.1 suggest that societal cleavage structures are too far away from the political arena to exert a direct influence on the dynamics of electoral protest, although they seem more relevant in hegemonic than in competitive regimes. Except for boycotts in hegemonic regimes (showing a positive relation), levels of electoral contention seem largely unrelated to levels of social inequality, the master variable in the recent literature on the social bases of political regimes. Size and religious polarization spell pre-electoral troubles in hegemonic regimes. Otherwise, coefficient signs for both cultural pluralism and size are mostly negative, small, and non-significant. If anything, rather than inflaming opposition mobilization against authoritarian elections, pre-existing societal cleavages seem to inhibit it. Hegemonic parties, above all, seem to reap the dividends of division induced by social cleavages that cut across the democratizing divide.

Economic Performance While levels of economic development mold structures of opportunity, levels of economic performance open windows of opportunity for political contention. The former influence whether dissidents inhabit good or bad places to protest, the latter whether they inhabit good or bad times for protest. Economic crises

The Calculus of Electoral Protest

315

make governments nervous and opposition actors hopeful everywhere, in all types of regimes, democratic or dictatorial. Thanks to the discontent they are bound to generate, bad economic times are good moments for general political protest (even though rising unemployment notoriously discourages labor protest). Table 9.1 presents regression results for two standard indicators of macroeconomic performance: annual rates of gdp per capital growth and annual consumer price inflation rates. All models include data for election years as well as average values for the previous five years.21 Inflation rates, it appears, bear scant relations to electoral protest rates. By contrast, medium-term rates of economic growth (rather than election-year variations), induce opposition parties to participate in authoritarian elections and acquiesce to their results. Sound macroeconomic management seems to be a sound strategy of conflict prevention. Overall, electoral protest in hegemonic regimes seems to follow the trails of economic junctures, while it appears more detached from economic ups and downs in competitive regimes. Apparently, when everything is stable, as under hegemonic party rule, opposition actors take their clues from the moving parts of their environment to decide whether it makes sense mobilizing their followers or not.

External Dependence The international embeddedness of national societies has infinite implications for domestic societies, and thus for domestic politics. International linkages are likely to affect balances of power in the domestic political arena either by altering the distribution of resources among internal actors or by altering their international alliance structures. To the extent that international societal relations render national economies dependent on external actors who value liberal democracy, they make authoritarian rulers more vulnerable to democratizing external pressures (see Levitsky and Way 2010b). The presence of potential external allies should encourage opposition actors to make themselves heard. Open protest against authoritarian elections may help translate international dependencies into political pressures. By informing external audiences about the authoritarian quality of elections, the opposition may be able to activate external threats to the regime. In this sense, external linkages contain meta-opportunities: the possibility of creating the possibility of mobilizing external threats (by mobilizing vertical threats). For the purpose of empirical exploration, I use four variables that denote the structural dependence of national economies on external actors: international trade (as percentage of gdp), foreign direct investment (as percentage of gdp), total debt service (as percentage of gni), and official development assistance

316

Empirical Explorations

(as percentage of gni). We know that not all kinds of external dependence are equal. Some regional powers, like Russia and China, do not exert systematic democratizing pressures (to say the least). Still, my theoretical expectation is simple: I expect opposition parties to show higher protest rates at higher levels of foreign dependence. The results of the regression analyses reported in Table 9.1 are disappointing. The overall explanatory capacity of external linkages is low, most individual coefficients are statistically insignificant and substantively small, and their signs drift between positive and negative. The cord of causation, it seems, frays out on the long march from aggregate relations with the world economy to protest patterns in the electoral arena.

PO LI T IC A L I N C E N T IV E S T RU C T U R ES If societal opportunity structures shape the chances of success that motivate opposition protests, political incentive structures shape the structures of costs that inhibit them. Under the wide conceptual roof of “political incentive structures” I will discuss two aspects of the political environment: formal political institutions and actual levels of state repression. The former primarily determine the opportunity costs of boycott, the latter the risks to personal integrity that anti-regime dissidents assume. Both are more proximate to the electoral arena than societal factors.

The Opportunity Costs of Boycott What is the explanatory puzzle that opposition actors pose who stand at the opening gates of authoritarian electoral arenas? Is it puzzling if they enter the arena or if they stay outside? When opposition parties boycott an authoritarian election, they either look like consistent democrats who are true to their principles or like weak competitors making up excuses (see Pastor 1999b: 1). When they participate under protest, they appear inconsistent. When they participate in silent resignation, they look lame. As a matter of fact, under the imperatives of instrumental rationality, “opposition parties that are doomed to lose should not form” (Greene 2007: 3) in the first place. But once they are in place, why should we expect them to enter the authoritarian electoral circus, play the clown, and entertain the king? For democratic opposition actors, joining the two-level game of authoritarian elections carries an ambiguous promise: They may be able to exploit the

The Calculus of Electoral Protest

317

structural ambiguities of authoritarian elections, erode the authoritarian grip on the electoral arena, and eventually transform the asymmetric game into a free and fair contest. Such prospects of institutional transformation are often vague and distant. Yet, in addition, taking part in authoritarian elections offers more concrete short-term payoffs to opposition parties, such as public exposure, public office, public finance, and access to patronage. If they boycott elections, they forgo these potential material benefits. While the long-term consequences of opposition boycotts are abstract and uncertain, their short-term opportunity costs are quite concrete and predictable. Boycotts form part of the modern repertoire of political contention (see Beaulieu 2006: 22–28, Friedman 1999). The most common variety of boycotts, consumer boycotts of products or firms, generally impose costs on producers and sellers, yet not on the boycotting consumers themselves who can shop for alternative products. Election boycotts are different. By unilaterally vacating the electoral arena from competition, boycotting parties declare elections to be an authoritarian charade. To the extent that they succeed in conveying this message to the citizenry (as well as to international actors), they impose “legitimacy costs” on the incumbent.22 You need two to tango, and to run a pseudo-democratic election too. The legitimacy costs that boycotting parties impose on incumbents are often elusive. The opportunity costs that they impose on themselves tend to be more tangible. These incommensurable factors are hard to weight. Given their uncertainties and ambiguities, boycott decisions are often described as “painful” (Silitski 2009: 87), a tough dilemma (see also Monga 1997: 158). By nature, dilemmatic decisions are hard decisions. Due to their ambiguities and uncertainties, they cannot be resolved through the simple computation of costs and benefits, but require political judgment. Accordingly, they always contain room for legitimate disagreement. Boycott debates are fissiparous. They generate splits within opposition parties and among them. Opposition parties often disagree both internally and among themselves about the wisdom of participating in or withdrawing from authoritarian elections. Even the simple cost calculations involve a catch. While high opportunity costs should deter opposition parties from boycotting an election, they also involve reputational benefits. They make boycotts look more heroic and thus more attractive as means of communication. High exit costs improve the credibility of boycotting parties and thus the informational value of boycotts. If opposition parties have a lot to lose from boycotting an election, they can more easily dissipate the suspicion that they withdraw for opportunistic reasons, because they are too unpopular to make a decent standing at the polls. Still, for reasons of simplicity, I expect that their wish to avoid the opportunity costs of boycott tends to prevail over their desire to reap the reputational

318

Empirical Explorations

benefits of costly boycotts. I expect high opportunity costs to induce lower boycott rates.

Political Institutions and Boycotts When opposition parties or candidates consider quitting electoral competition, we can expect the institutional environment to form an important part of their cost calculations. The literature on political contention has not paid much attention to “the role of both formal and informal political institutions in shaping elite incentives and so protest patterns” (Robertson 2010: L 8824). It has recognized the encouraging effects that the general openness of the political systems exerts on political protest, and the depressing effects that closure has. But it has largely ignored the impact of more concrete political institutions. In this section, I will explore the association of opposition boycotts with formal political institutions that are likely to shape the opportunity costs of boycotts. As reported in Chapter 6, opposition parties boycott around one-third of legislative elections in our sample. Yet for each full boycott there are two partial boycotts in which parties fail to coordinate. Presidential boycott rates lie at a similar level in competitive regimes, yet almost twice as high in hegemonic regimes (with full and partial boycotts holding a rough balance).23 Table 9.2 shows simple, uncontrolled comparisons of mean rates of boycott across various institutional arrangements: forms of government, the strength of legislatures, electoral systems, electoral timetables, and electoral sequences. Given the distinctive institutional logic of legislative and presidential elections, I segment my sample by electoral arenas, rather than subtypes of regimes. Subtypes of regimes: Since opposition parties are structurally weaker in hegemonic regimes, they have less to lose by boycotting an election. To the extent that their boycott decisions are driven by cost considerations, opposition parties should therefore be less reluctant to boycott hegemonic elections. This is not the case in the legislative arena, but very much so in presidential contests. Presidential boycott rates in hegemonic regimes (59.1 percent) are almost double those in competitive regimes (29.4 percent). Electoral arenas: Given the unipersonal and thus indivisible nature of the chief executive office, presidential elections are winner-takes-all (wta) contests. Opposition candidates, the predestined losers of authoritarian elections, have little to lose if they withdraw their candidacies. They may hurt their vote-getting potential in concurrent legislative elections as well as in future presidential contests. Yet they do not lose anything tangible. They renounce their predicted vote share, but presidential votes don’t buy them anything.

The Calculus of Electoral Protest

319

Table 9.2. Political institutions and election boycotts

All regimes Subtypes of regime Competitive regime Hegemonic regime Forms of government Parliamentarism Semi-presidentialism Presidentialism Legislative strength Semi-effective legislature Ineffective legislature Electoral systems Proportional representation Mixed-member systems Majoritarian systems Electoral timetables Concurrent elections Non-concurrent elections Electoral sequences 1st elections 2nd and 3rd elections 4th and subsequent elections

Legislative boycotts (% of all L elections)

N

Presidential boycotts (% of all P elections)

N

31.4 [p = .976] 31.5 31.3 [p = .011] 10.3 28.6 40.8 [p = .017] 22.0 41.8 [p = .003] 11.8 6.9 41.2 [p = .307] 28.2 37.2 [p = .010] 38.5 38.1 7.4

121 121

38.4 [p = .017] 29.4 59.1 [p = .412] … 27.3 40.3

73 73

121

73

121 … … 80

121

121

… … … [p = .465] 43.3 34.9 [p = .939] 36.6 40.7 40.0

73

73

Note: Boycotts = sum of partial and full boycotts. Shaded areas denote statistically significant differences (p ≤ .15). Chi square tests (bilateral significance). P-values shown in brackets. (...) = not applicable. Shaded cells denote statistically significant differences (p ≤ .15).

Ceteris paribus, presidential elections should therefore display higher boycott rates. They do this in hegemonic regimes, yet only to a modest extent across regime types. Forms of government: In presidential systems, legislative elections matter (or are seen to matter) less than in parliamentary systems where they are (on paper) the decisive contests. Opposition parties should therefore be more hesitant to boycott legislative elections in parliamentary systems. And they are. They boycott two-fifths of all elections in presidential systems (in either electoral arena), yet only one-tenth of national elections in parliamentary systems. Legislative powers: Conquering even a fistful of seats in a reasonably strong and assertive legislature may be a valuable price of political competition. In a weak window-dressing assembly, it may not be worth the trouble. Why bother

320

Empirical Explorations

to win fake assembly seats in fake elections? Why not better denounce the non-authenticity of both by withdrawing from the election? In accordance with such political utility considerations, boycott rates to ineffective legislatures are twice as high (41.8 percent) as those to effective or semi-effective legislatures (22 percent).24 Electoral formulae: The rules that govern the translation of votes into legislative seats have direct implications for the costs of exit by minority parties. In proportional and mixed-member systems, even small parties (depending on effective thresholds of representation) have something to lose by withdrawing from electoral competition. In majoritarian wta systems, by contrast, they often cannot aspire to win seats even when they enjoy considerable popular support. In the face of certain defeat, they should be rather content to boycott. My data support this proposition. Opposition parties almost never boycott legislative contests held under pr or mixed electoral systems, yet withdraw from about two-fifths of majoritarian elections (41.2 percent). Electoral schedules: When legislative and presidential elections take place concurrently (in the same calendar year), opposition parties may adopt convergent strategies across electoral arenas and either participate in both or boycott both. Alternatively, they can split their strategies to reap the informational benefits of boycott while avoiding its opportunity costs. They can withdraw from the presidential arena, where boycott is cheap, and keep up fighting in the legislative arena, where boycott is more costly. As Table 9.2 indicates, the latter seems to be the case. Opposition parties tend to split their strategies when concurrent elections grant them the opportunity of doing so. When elections are non-concurrent, boycott rates are very similar across electoral arenas. When they are concurrent, legislative boycott rates go down and presidential rates up (in both cases, by somewhat fewer than ten percentage points). Opposition parties are rational. They try to have their cake and eat it. Electoral sequence: Electoral boycotts are weapons that weak parties use to tell others they are victims of authoritarian manipulation. They are exceptional measures. Parties cannot adopt boycotts as their dominant strategy. If they always refuse to participate in elections, they lose their status as party organizations. Political parties, after all, are defined as organizations that field candidates for political elections (Sartori 1976). As regimes start accumulating histories of authoritarian elections, the opportunity costs of boycott may not change much, but they may become more apparent. Even if authoritarian legislatures do not form central sites of state power, they still tend to be relevant sites of regime struggle (see Fish 2006). When opposition parties renounce the possibility of entering national assemblies, they renounce access to limited, yet valuable, sources of institutional power and public exposure. If boycotts

The Calculus of Electoral Protest

321

fail to generate beneficial reforms, opposition parties may come to reevaluate their opportunity costs. Over the course of several authoritarian elections, they may come to discover either the futility of institutional exit or the value of institutional voice. As Table 9.2 indicates, such a process of learning (and resignation) indeed seems to take place. After three elections in a row, opposition parties practically stop boycotting legislative elections (yet not presidential contests). Up to now, I have reviewed direct bivariate associations between political institutions and electoral boycotts. To expand my explanatory horizon a little, I regress binary versions of all three forms for electoral protest on a few features of the institutional environment: forms of government, electoral rules, and electoral sequences. To keep case numbers from shrinking excessively, I use my full, undivided sample of authoritarian elections for these multivariate regressions. The results displayed in Table 9.3 suggest some straightforward conclusions: Majoritarian electoral rules, and to a lesser degree presidential forms of government, are consistent sources of electoral conflict. A hypothetical move from proportional representation to pure majoritarian rules (leapfrogging mixed systems) multiplies the odds of boycott by six (eb = 2.9), the odds of pre-electoral protest by three (eb = 1.6), and the odds of post-electoral mobilization by four (eb = 1.9). By contrast, the longer the strings of uninterrupted authoritarian elections, the calmer the waters of opposition protest. Each step down the road of electoral institutionalization halves the odds of both boycott and pre-electoral troubles (eb = 0.5).

Table 9.3. Political institutions and electoral protest Boycott

Presidentialism Majoritarianism Electoral sequence Constant pac Cox & Snell R2 N

Pre-electoral Protest

Post-electoral protest

eb

p

eb

p

eb

p

1.830 2.896 .504 .105 73% .151 100

.106 .005 .075 .046

1.312 1.645 .501 .550 68% .085 100

.377 .103 .037 .496

1.341 1.934 .733 .234 62% .064 100

.335 .037 .319 .113

Note: Non-segmented sample of authoritarian elections. Binary logistic regressions. Presidentialism = parliamentary. All dependent variables are dichotomized. Boycott = partial and full boycott. Post-electoral protest = active protest mobilization. Presidentialism = trichotomous (parliamentarism, semi-presidentialism, and presidentialism). Majoritarianism = trichotomous (proportional representation, mixedmember system, and majoritarian electoral system). Electoral sequence = trichotomous (1st, 2nd + 3rd, and 4th and subsequent elections). Shaded coefficients = statistically significant (p ≤ .15).

322

Empirical Explorations Political Repression and Protest

Political repression, the deployment of state violence against dissidents, forms a measure of last resort of authoritarian governments. The complex, contingent, and interactive relations between repression and mobilization have been subject to considerable debate in the literature on contentious politics. Numerous authors have found evidence of an inverted-U relationship between repression and mobilization. The underlying causal logic is simple. When repression is low, as in liberal democracies, dissidents do not need to take to the streets to make themselves heard. When repression is high, as in oppressive military regimes, dissidents dare not do so. In the broad middle, public protests are more likely: political grievances are high, effective repression is less than certain, and the chances of protest success are higher than nil.25 The family of electoral autocracies is censored at both tails, as it excludes minimally democratic as well as maximally repressive regimes. As with all intermediate types of political regimes, it nevertheless accommodates broad variation in degrees of repression. In my sample of authoritarian elections, violations of physical integrity span the entire range of measurement, from minimum to maximum violations. Hence we may expect the inverted-U relationship between repression and protest to hold in electoral authoritarian regimes too, like a kind of theoretical Russian doll, an inverted U within the inverted U.26 Probing for the existence of non-lineal relationships, I calculated “best fits” among both lineal and curvilinear (squared and cubic) estimations of bivariate associations between violations of physical integrity (in the year preceding the election year) on the one hand and the incidence of boycott and post-electoral protest in legislative and presidential elections on the other.27 As these uncontrolled bivariate estimates suggest, post-electoral protests are more sensitive to levels of repression than electoral boycotts. When governments orchestrate campaigns of violence against their adversaries, electoral boycott may emerge as a sensible act of self-defense by opposition actors. For instance, in the second round of the 2008 presidential elections in Zimbabwe, opposition parties responded to escalating repression by withdrawing from electoral competition. On average, the association between violence and boycott does not seem to hold under hegemonic party rule. In competitive regimes, high levels of repression do appear to induce higher frequencies of boycott, albeit at weak levels of statistical significance. The relation is linear in presidential elections (R2 = .034, p = .207, N = 45), and concave in legislative elections (R2 = .048, p = 191, N = 69). With respect to post-electoral mobilization, repression works as intended, as a straightforward depressant of protest, in hegemonic presidential races as well as competitive legislative contests. In these two arenas, repression

The Calculus of Electoral Protest Hegemonic legislative elections

Competitive legislative elections

1.0

1.0

0.8

0.8

0.6

0.6

0.4

0.4 R2 = .181, p = .034, N = 46

0.2 0.0

0

2

4 Repression

6

R2 = .036, p = .115, N = 70 0.2 0.0

1

Hegemonic presidential elections 1.0

0.8

0.8

0.6

0.6 R2 = .130, p = .100, N = 21

0

2

4 Repression

6

3

4 5 6 Repression

7

8

0.4 R2 = .218, p = .011, N = 47

0.2

0.2 0.0

2

Competitive presidential elections

1.0

0.4

323

0.0

1

2

3

4 5 6 Repression

7

8

Figure 9.3. Repression and post-electoral protest (best fit) Note: ols regressions. Figures show best fit among lineal, squared, and cubic specifications. Repression = violations of physical integrity in the year previous to election year.

deters contention. In the remaining two, somewhat mysterious cubic functions emerge as best fits. In hegemonic legislative elections, repression seems to deter protest, albeit not in the middle. In competitive presidential elections, repression seems to encourage protest, albeit not in the middle. Inverted-U curves did not happen to emerge (see Figure 9.3). Summing up: In electoral autocracies, repression and mobilization seem to interact as they do in other non-democratic regimes: in complex and contingent ways. There are no iron laws of mobilization and no iron links between repression and its outcomes. All in all, though, hegemonic regimes appear somewhat more effective in deterring opposition protest through repression.

324

Empirical Explorations T H E I N FO R M AT IV E RO L E O F PROT EST

Even as contentious actors adjust their protest strategies to their environment, as I hypothesized above, the main purpose of protest is to overcome environmental structures of constraint. Opposition parties who protest against authoritarian elections may loosen constraints and engender opportunities through two causal mechanisms: by communicating grievances and by activating threats. Both pathways towards the transformation of existing opportunities raise the question of meta-opportunities: When do regime challengers face opportunities for widening opportunities by publicizing grievances and provoking threats? What are the conditions that render electoral protest informative? And what are those that render it provocative? Let us begin with the first question. If protests against authoritarian elections help opposition parties to inform various audiences (such as the citizenry, the military, and the international community) about the non-democratic nature of the regime, then we should see their incidence vary with the severity of electoral manipulation as well as with its opacity. When authoritarian manipulation is severe, opposition grievances run deep. When it is opaque, ignorance about its occurrence runs deep. Both factors enhance the need to tell “third actors” what they cannot learn through official channels of communication. We may thus understand the informative value of protest (ivp) as a multiplicative function of these two variables. ivp = severity * opacity of manipulation

(9.2)

Protests lose their informative function when manipulation is obvious, regardless of its severity. Protests also lack informative value when manipulation is trivial, regardless of its visibility. In the previous chapter, I proposed to rank strategies of manipulation along these two dimensions of severity and visibility.28 If we multiply their ranks, as shown in Table 9.4, we obtain a rough idea of how informative opposition protests are when they are targeted against specific manipulative strategies: x Repression: At the high end of information value lies repression. Since violations of physical integrity are grave and not entirely transparent, protest against repression should be highly informative. Nevertheless, I expect the dissuasive role of repression to trump the informative value of contention. x Exclusion: At the other pole lies exclusion. Parties or candidates banned from electoral competition may not need to inform voters about the fact of exclusion, but still may need to convey its relevance to them.

The Calculus of Electoral Protest

325

Table 9.4. The informational value of election protest

Repression Fraud Censorship Exclusion

(1) Severity

(2) Opacity

ivp = (1) x (2)

4 3 1 2

2 3 4 1

8 6 4 2

Note: Figures in columns 1 and 2 denote ranks (with 1 = least and 4 = most severe / opaque). ivp = informational value of protest.

x Fraud: The information value of protest against fraud and censorship occupies intermediate positions. Unless regimes perform electoral fraud as a pr stunt to demonstrate their power and impunity, vote rigging is a clandestine activity. “No one who stuffs the ballot box wants to leave a trail of incriminating evidence” (Lehoucq 2003: 233). Like other fraudulent enterprises, it is “only its victims” who want to have it “publicized” (Lehoucq 2003: 235). The more incisive the rigging of votes and the better hidden, the more important it is that its victims raise their voice. x Censorship: According to the literature on social movements, media restrictions limit the ability of dissidents to reach their relevant audiences and thus discourage political protest, while media freedom facilitates political contention: “as media coverage of collective dissent increases, collective dissent increases” (Lichbach 1998: 90). Here, I stipulate the inverse relationship: censorship breeds contention. To the extent that regime-subservient mass media restrict or distort political information, anti-regime actors need to open up alternative channels of communication. Public contention may be one of them. Do our empirical data confirm these expectations according to which the severity and (above all) the opacity of manipulative strategies determine the informational value and thus the probability of protest? Figure 9.4 presents predicted probabilities of electoral protest (boycott, pre-electoral, and post-electoral protest) at varying levels of censorship in hegemonic regimes (left-hand column) and competitive regimes (right-hand column). The horizontal lines show predicted probabilities of protest when one strategy of manipulation (repression, exclusion, or fraud) is set at its maximum and all others at their minimum. The “minimum manipulation” line shows predicted probabilities of protest when all strategies are held at their minima. All simulations derive from binary logistic regressions with dichotomous versions of the three types of protest as dependent and all four manipulation strategies as independent variables (see Table E.9.1 in Appendix e).

326

Empirical Explorations HEGEMONIC REGIMES

1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 0.00

0.50

1.00 Censorship

0.50

1.00 Censorship

1.50

2.00

0.50

1.00 Censorship

1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 0.00

Min. Manipulation

1.50

2.00

1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 0.00

Max. repression Exclusion

1.50

2.00

0.50

0.50

Min. Manipulation

1.00 Censorship

Max. repression Exclusion

1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 0.00

0.50

1.50

2.00

Min. Manipulation

1.00 Censorship

Fraud Min. Manipulation P Post-electoral protest

P Post-electoral protest

Fraud Min. Manipulation 1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 0.00

Max. repression Exclusion

P Pre-electoral protest

P Pre-electoral protest

Max. repression Exclusion 1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 0.00

Min. Manipulation

P Boycott

P Boycott

Max. repression Exclusion

COMPETITIVE REGIMES

1.00 Censorship

1.50

2.00

Max. repression Exclusion

1.50

2.00

Figure 9.4. Censorship and electoral protest (predicted probabilities) Note: All dependent variables dichotomized: Boycott: 0 = participation, 1 = partial or full boycott. Post-electoral protest: 0 = acquiescence or rejection, 1 = active mobilization. Repression and censorship in year previous to election year. All elections. Calculations of predicted probabilities with X-Post (Cheng and Long 2000) on the basis of binary logistic regressions (see Table E.9.1 in Appendix e).

As Figure 9.4 nicely illustrates, censorship is almost unrelated to political protest in hegemonic regimes. Yet, in competitive regimes, it multiplies the probability of boycott (more than sevenfold) as well as the probability of pre-electoral protest (more than fourfold). In accordance with our expectations, too, vote rigging exerts a powerful pull on post-electoral protest, in

The Calculus of Electoral Protest

327

particular in hegemonic regimes (see the graphs at the bottom of Figure 9.4). Once the polls have closed, protest is unlikely to remedy the consequences of fraud. It can only publicize them. In the face of electoral theft, its victims are unlikely to recover their electoral belongings. Yet, it appears, they at least want to denounce the thief.

T H E PROVO C AT I V E RO L E O F PROT EST When dissidents summon public protests, they want these protests to be more than informative. They also want them to be impressive. If they only wished to inform the ignorant public, then they could simply go round distributing leaflets or sms messages. Aside from conveying information, they need to convey a sensation of threat. To do so, it helps if they can mobilize large numbers of citizens. But even “[s]mall acts of protest by isolated groups can grow quickly into major challenges” (Robertson 2010: L 4284). They may trigger spirals of contention and defection that “can put the survival of the incumbents at risk” (Robertson 2010: L 4284). For contentious actors then the challenge is either to appear big or to look provocative. Either they can assemble numbers large enough to make an impression on tremulous elites. Or they succeed in presenting themselves as the famous butterfly capable of triggering an earthquake. Elites may fear even a modest performance if they perceive it as the tip of a contentious iceberg, a small symptom of larger discontent boiling below the surface of political tranquillity. If contentious action indeed serves to activate further vertical or horizontal threats to the authoritarian government, we should expect strategic dissidents to summon protests at times and places that are likely to generate such additional threats. The question is: When is electoral protest likely to be provocative? When does it trigger peaceful contention beyond the narrow juncture of elections? When does it provoke violent protests? When does it motivate elite actors to leave the ruling coalition? When does it push the military to intervene in politics? As a matter of course, the consequences that electoral protests have on subsequent threats are a matter of complex, contingent, and fluid calculations. Rather than broad general contexts, it is thick local knowledge that provides the basis of actor calculations. Nevertheless, we can assume both opposition parties and ruling parties to rely (tentatively) on a simple rule of thumb: the past shows the path to the future. Antecedents can be treated as precedents. If threats have surfaced in the past, present protests are more likely to activate them in the future.

328

Empirical Explorations Protesting in the Wake of Vertical and Horizontal Challenges

Since hegemonic regimes constitute tranquil equilibria in which both vertical and horizontal threats are rare events, opposition parties are more likely to build upon histories of dissidence under hegemonic party rule. In a landscape where nothing moves and everybody lies still, they will clutch at any straw of contention that floats on the surface of hegemonic calm. However, while opposition parties should be willing to follow the trail of vertical and horizontal threats in both subtypes of regimes, they should be hesitant to reinforce violent forms of dissidence. Armed rebellions tend to generate spirals of conflict in which societal and state violence reinforce each other. Military coups shut down the electoral arena in either transient or enduring ways. Even when military leaders undertake them in the name of democracy, they afford no guarantees of democratizing progress. I therefore expect protest to breed protest—as long as it remains peaceful. Organized violence dissuades civil protest. Do empirical protest dynamics on average lend plausibility to my expectations of contentious path-dependence? Do past protests provoke present protests into provoking future protests? Figure 9.5 presents predicted probabilities of three forms of electoral protest (boycott, pre-electoral, and post-electoral protest) when the number of preceding vertical threats (the sum of anti-government demonstrations, general strikes, and riots during the three years preceding the election year) moves from zero close to the empirical maximum in our sample. Their empirical maxima are about 20 in hegemonic regimes (left-hand column) and 30 in competitive ones (right-hand column). The curves show predicted probabilities of protest when armed rebellions (the sum of preceding revolutions and guerrilla wars) are set close to their empirical maxima and horizontal threats (the sum of preceding government crises, purges, and successful military coups) at their empirical minima of zero events (and the other way round). The curve of “no other threats” shows predicted probabilities of protest when both are held at zero. All simulations derive from binary logistic regressions with dichotomous versions of the three types of protest as dependent and all three types of threat as independent variables (see Table E.9.1 in Appendix e). In hegemonic contexts, as Figure 9.5 shows quite impressively, electoral protest is indeed highly sensitive to antecedent threats. As few as two horizontal threats suffice to induce opposition parties to boycott a national election. As vertical threats reach their empirical maximum, the active mobilization of opposition parties against hegemonic elections becomes almost certain. Each additional vertical threat in the three preceding years augments the odds of

The Calculus of Electoral Protest HEGEMONIC REGIMES

1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 0.00

5.00

P Pre-electoral protest

15.00

20.00

5.00

1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 0.00

3 armed rebellions

10.00 Vertical threats

15.00

20.00

1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 0.00

3 armed rebellions

10.00 Vertical threats

15.00

20.00

5.00

5.00

5.00

25.00

30.00

10 armed rebellions

10.00 15.00 20.00 Vertical threats

No other threats 5 horizontal threats 1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 0.00

10 armed rebellions

10.00 15.00 20.00 Vertical threats

No other threats 5 horizontal threats

P Post-electoral protest

P Post-electoral protest

No other threats 5 horizontal threats

P Pre-electoral protest 5.00

No other threats 2 horizontal threats 1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 0.00

3 armed rebellions

10.00 Vertical threats

No other threats 2 horizontal threats 1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 0.00

COMPETITIVE REGIMES

P Boycott

P Boycott

No other threats 2 horizontal threats

329

25.00

30.00

10 armed rebellions

10.00 15.00 20.00 Vertical threats

25.00

30.00

Figure 9.5. Previous threats and electoral protest (predicted probabilities) Note: Binary logistic regressions. All dependent variables dichotomized (see Figure 9.4). All elections. Calculations of predicted probabilities with X-Post (Cheng and Long 2000) on the basis of binary logistic regressions (see Table E.1 in Appendix e).

pre-electoral protest by 54 percent (eb = 1.536) and the odds of post-electoral protest by 20 percent (eb = 1.196) (see Table E.1). In competitive regimes, boycotts are most likely in the absence of either vertical or horizontal threats, which supports the idea that they are “weapons of the weak” (Scott 1985). Opposition parties do not deploy them to reinforce

330

Empirical Explorations

pre-existing protest movements, but to compensate for their inability to mobilize. Furthermore, the presence of horizontal threats stimulates pre-electoral mobilization, while the presence of armed rebellion discourages it. By contrast, electoral protests look rather disconnected from preceding levels of mass contention.

Protesting in the Wake of Electoral Challenges When they can build upon antecedents of political dissent outside the electoral arena, the electoral protests that opposition parties call are likely to resonate more strongly among its target publics (both in terms of the turnout that they achieve and the impression that they make). When opposition parties can build upon emergent uncertainties within the electoral arena, their protests should carry even stronger resonance. The electoral arena, after all, is supposed to be the central arena of political struggle in electoral authoritarian regimes. If extra-electoral challenges get electoral autocrats to worry, electoral challenges should alarm them. Thus, if present challenges serve to magnify past ones, opposition parties who do well in electoral competition should do even better in subsequent political contention. If electoral protest is a continuation of electoral threats with other means, we should see protest levels rise as margins of victory shrink. In hegemonic regimes, in particular, we should see a close association between the stirrings of competitiveness and protest. During the long winters of hegemonic electoral dominance, opposition parties may hibernate in impotence. There is nothing they can do. Except perhaps do less than nothing, demobilize and leave the electoral arena. They should only start mobilizing their followers when “something moves.” Only when the “iron cage” of hegemonic party rule begins to crack, does it make sense to shake its bars in protest. It is the emergence of incipient electoral opportunities that creates meta-opportunities for opposition actors: the opportunity to create opportunities for change by challenging the hegemonic regime on the streets. I expect the association between competitiveness and protest to be less clear in competitive regimes. In hegemonic regimes, high margins of victory reinforce the belief that the ruling party is popular as well as strong. In competitive contexts, actors are more likely to read high margins of victory, not as strong indicators of a strong regime, but as fraudulent indicators of a fraudulent regime.

The Calculus of Electoral Protest LEGISLATIVE ELECTIONS Mean margin of victory (previous election)

60.00

40.00 72.09

65.27 55.24 58.33

20.00

39.20 19.50

PRESIDENTIAL ELECTIONS 80.00 Mean margin of victory (previous election)

Hegemonic regimes Competitive regimes

80.00

0.00

40.00

78.23

20.00

34.62

22.83

72.90 59.29 43.93

20.00 31.75

Hegemonic regimes Competitive regimes

60.00

0.00

50.00 40.00 30.00

57.80 48.71 42.05

20.00 31.11

10.00 0.00

Acquiescence

Active protest

Acquiescence

Pre-electoral protest

73.36 62.94 40.04

20.00

32.89

60.00

40.00

77.45

72.07 49.35

20.00

44.58 30.42

12.96

0.00

Hegemonic regimes Competitive regimes

80.00

Mean margin of victory

60.00

76.27

Active protest

Pre-electoral protest

Hegemonic regimes Competitive regimes

80.00

Mean margin of victory

35.96

Participation Partial boycott Full boycott

Mean margin of victory (previous election)

Hegemonic regimes Competitive regimes

60.00

40.00

64.73

60.71

0.00

80.00 Mean margin of victory (previous election)

Hegemonic regimes Competitive regimes

60.00

Participation Partial boycott Full boycott

40.00

331

11.13

0.00 Acquiescence Rejection Active protest Post-electoral protest

Acquiescence Rejection Active protest Post-electoral protest

Figure 9.6. Previous margins of victory and electoral protest Note: Shaded columns denote statistically significant differences (p ≤ .15).

332

Empirical Explorations

Figure 9.6 plots the mean margins of victory that rulers earned in the elections that preceded the strategic choices of opposition parties.29 For hegemonic elections, the (positive) associations between previous competitiveness and electoral protest conform to theoretical expectations. Electoral protest against hegemonic elections takes place when margins of victory have started to dwindle. In competitive regimes, however, patterns of postelectoral protest are inverted. Opposition parties tend to mobilize their followers in the wake of elections in which incumbents secured high levels of victory. Pre-electoral protest, by contrast, seems unrelated to the competitiveness of preceding elections. The same applies to boycotts in hegemonic regimes. In competitive contexts, the relation between boycotts and competitiveness is non-linear and puzzling. Opposition parties partially withdraw from presidential elections when they are least competitive and from legislative elections when they are most competitive.

C ON C LU S I O N The empirical evidence presented in this chapter, even if coarse and indirect, lends credit to the idea that opposition parties adapt to given opportunities and incentives, while they also strive to alter existing structures of costs and opportunities. In their efforts to reshape the conditions of their own success through electoral protest, opposition parties partake in the twin politics of uncertainty. They strive to shape factual uncertainties by dissipating the opacity of manipulation; and they strive to generate institutional uncertainty by reinforcing emergent threats to the regime.

Notes 1. For an insightful analysis of (elite-sponsored) labor protest and other forms of political contention in electoral authoritarian Russia, see Robertson (2010). 2. Of course, the distinction between “genuine” and “fake” opposition actors is often less than clear-cut. When authoritarian regimes use multiparty elections as mechanisms of patronage distribution, opposition parties and candidates of all ideological shades tend to seek and cherish access to economic benefits (see Albrecht 2005, Eisenstadt 2000, Lust-Okar 2009, Robertson 2010). According to Emily Beaulieu, small opposition parties are patronage seeking and by definition stage “minor” boycotts in the pursuit of particularistic goals, while large opposition parties are

The Calculus of Electoral Protest

3.

4.

5. 6.

7.

8.

9. 10.

11.

333

reform oriented and enact “major” boycotts in the quest for democratizing reform (2006: 28–35). In more limited cases, elite negotiations or judicial rulings overturn partial outcomes of national elections. On the extra-institutional negotiation of local election results during Mexico’s transition from hegemonic party rule, for example, see Eisenstadt (2004). At the game level, a third possibility exists: electoral conversion. Meta-game strategies can work as campaign strategies: electoral protests persuade voters to align with the opposition. The literature on social movements has been sensitive to the communicative, self-presentational role that political protest has on political authorities as well as towards broader audiences. The late Charles Tilly postulated four communicative goals of contentious performances: worth (a just cause), unity (a cohesive movement), numbers (wide support), and commitment (moral resolve) (wunc) (Tilly 2004: 4). See also Bratton (1998: 24) and Lindberg (2006a). The Arab rebellions of 2011 provided ample evidence of all three chains of interaction between social mobilization and authoritarian elite defection (see e.g. Barany 2011, Bellin 2012, Masoud 2011, Schraeder and Redissi 2011). More recently, some of the founding fathers of social movement research have been engaged in an analytical scissor movement. On the one hand, they have been climbing up the ladder of conceptual abstraction to study the broader category of “contentious politics” (encompassing all kinds of collective challenges to power, including movements, riots, and revolutions). On the other hand, they have been climbing down the ladder of theoretical abstraction to supersede the original set of theoretical tools by more concrete “causal mechanisms” that explain the genesis, reproduction, and decay of contentious politics (see McAdam, Tarrow, and Tilly 2001). In my view, the creation of a disconnected collection of causal mechanisms moves into the limelight what has always been the major deficit of social movement research: its lack of anchorage in social theory, in either a sociological or political theory of action. Although the literature tends to treat “political opportunity structures” as externally given explanatory parameters, protest scholars do often recognize their dependence on perceptions as well as the possibility of reverse causation (see e.g. Goldstone and Tilly 2001). For discussions of the notion of political opportunity structures, see also Gamson and Meyer (1996), McAdam (1996), and Tarrow (1994). “The gain to an individual member of the group [depends] upon the ‘fraction’ Fi of the group gain he got” (Olson 1965: 23). Since political grievances derive from notions of justice, it would be more appropriate to state that they measure the distance between the injustice of the present and the injustice of possible futures. Note that I am not playing different kinds of motives against each other. Individual preferences for civil liberties are not a matter of moral or altruistic

334

12.

13.

14. 15. 16. 17.

18. 19.

20.

Empirical Explorations or non-material motivations. The demand for liberty is no more normative than the demand for security, which is commonly conceived as its competitor. Under conditions of security, citizens are able to lead their private lives without suffering assault from fellow citizens. Under conditions of liberty (in its classic liberal, negative conception), citizens are able to lead their private lives without suffering assault from the state. The quest for liberty seeks protection from the state (the control of vertical violence or repression), the quest for security protection by the state (the containment of horizontal violence or crime). The desire to live a free life without the threat of being killed, tortured, or abducted by police officers is no more “moral” or “altruistic” than the “material” and “self-regarding” desire to live a secure life without the threat of being killed, tortured, or abducted by neighbors. See also Chapter 1. Lichbach offers a synthesis of the literature on the role of expectations of success in collective protest (1998: 62–82). See also Chong (1991) and Kuran (1995). It is, of course, self-defeating if everybody does as the free rider does and stays at home. Furthermore, the free rider’s individual contribution to collective defeat is magnified by the extent to which others observe and follow her example. On the distinction between “voice” and “exit,” see Hirschman (1970). Boycotting parties often “engage in a campaign for their boycott that is remarkably similar to an election campaign” (Beaulieu 2006: 44). See also Chapter 7, pp. 240–241. All estimations derive from lineal ols regressions (with the table showing unstandardized B coefficients and p values), except for the dichotomous dependent variable pre-election protest, for which I run binary logistic regressions (with the table showing odds ratios eb and p values). To indicate the overall fit of regression models, I report, here as elsewhere, standard errors and R2 values for lineal regressions, and percentages of accurate classification (pac) and Cox & Snell R2 values for logistic regressions. For variable descriptions, see Appendix c, for descriptive statistics, Appendix d. See Lipset and Rokkan (1967), Gallagher, Laver, and Mair (2001: Ch. 9). See also Chapter 7, “The Weight of Structural Contexts,” pp. 239–245. On ideological divisions among opposition parties, see also Albrecht (2007), Greene (2007), Howard and Roessler (2006), Lust-Okar (2005), Magaloni (2006), Posner (2007). On the strategic logic of opposition coordination, see also Gandhi (2010), Gandhi and Reuter (2013), Langston (2006), and van de Walle (2006). Madison, Hamilton, and Jay (1788: Federalist Paper No. 10, p. 127). Within my overall sample of authoritarian elections, linguistic and religious fractionalization are correlated, yet not identical (r = .415, p = .000, N = 146). Indices of linguistic and religious polarization are transformations of Rae fractionalization indices from Alesina et al. (2003): Ranging from 0 to 1, fragmentation indices

The Calculus of Electoral Protest

21.

22.

23. 24.

25.

26.

27.

28.

335

express the probability of belonging to different subgroups of society. In case of cultural differences, 0 indicates societal homogeneity, 1 complete fragmentation. At the intermediate value of 0.5, the society is polarized between two equal groups. In terms of potential political conflict, polarization seems more problematic than either full homogeneity or full fractionalization. I thus transform the fragmentation index into a polarization index: (a) I subtract 0.5 from the fragmentation index, which produces an index ranging from –0.5 to 0.5. (b) I take its absolute value, multiply it by 2, and invert the result, which yields a scale from 0 to 1, with 0 indicating either homogeneity or fragmentation and 1 bipolar polarization. For variable descriptions and sources, see Appendix b, for descriptive statistics, Appendix d. Naturally, these measures are related, yet not to an extent that would involve prohibitive problems of collinearity. Election-year per capita growth correlates moderately with the five-year average before the election (r = .238, p = .004, N = 144), while election-year inflation is surprisingly unrelated with the five-year average before the election (r = .036, P = .697, N = 121). While I focus on the communicative role of election boycotts towards domestic audiences, Beaulieu (2006: Ch. 5) and Beaulieu and Hyde (2009) analyze their interplay with international monitors. See Chapter 6, Figure 6.2; see also Table 9.2. Note, though, that legislative ineffectiveness may be endogenous to the weakness (or even absence) of legislative opposition in the wake of election boycotts. See, for example, Davenport (2007), Francisco (2005), Gartner and Regan (1996), Goldstone and Tilly (2001), Inclán (2008), Kurzman (2004: Chs. 2 and 6), Lee, Maline, and Moore (2000), McPhail and McCarthy (2005), Regan and Henderson (2002), Robertson (2010: Ch. 1, “Protest in Hybrid Regimes”), and Schock (2005). For an adaption of the inverted U to election boycotts, see Beaulieu (2006: 49–50). In more general terms, applying the inverted-U hypothesis for the purpose of comparison (be it across countries or across time) has been difficult, as we lack empirical anchors that would allow us to identify “high,” “medium,” and “low” levels of repression that are equivalent across contexts—according to actors’ shared perceptions, which is what matters for their strategic decisions. Needless to say, the Cingranelli–Richards scale of physical rights violations I rely on measures “objective” intensities of repression without pretending to gauge actors’ beliefs about repressive intensities. Base line estimations are lineal. If they turned out significant (p ≤ .15), the figure shows them unless either squared or cubic specifications improved their explanatory capacity (by an R2 increase of at least 0.05). See Chapter 8, pp. 273–275.

336

Empirical Explorations

29. For post-electoral protest, the bars show average margins of victory at those (current) elections the protest is directed against. For boycotts and pre-electoral protests, they show margins at preceding contests and thus exclude first elections. As I segment my sample by regime subtype as well as electoral arena, some of the resulting case numbers are very low (as can be appreciated in Table E.2 in the appendix). Table E.2 also contains p-values from Analyses of Variance (anova).

10 The Struggle over Regime Change In the preceding three chapters, I strove to explain the moving parts within the nested game of authoritarian elections: electoral outcomes, regime manipulation, and opposition protest. Here, I will trace the impact that political dynamics within the electoral arena have on short-term regime outcomes. How do competitiveness, manipulation, and protest contribute to produce the big outcome variable of regime change? In accordance with the idea of embedded elections, I will also consider a series of contextual factors: societal and institutional structures that may affect the balance of power between government and opposition; latent vertical threats in the form of societal cleavages and economic crises that may translate into manifest vertical threats; and state capacities that may translate into authoritarian capacities of threat containment. Before plunging into the empirical analysis, I explicate the nature of my theoretical enterprise: I explain variations within one category of authoritarian regimes (rather than across various categories) and I do so by primarily considering regime-specific factors (rather than more generic ones).

EXPLAINING REGIME CHANGE What explains stability and change of authoritarian regimes? When do they endure? When do they fall? In the comparative study of political regimes, we have been addressing this question at varying levels of abstraction, conceptually as well as theoretically.

Levels of Conceptual Abstraction Conceiving our objects of explanation at varying levels of conceptual aggregation, our theories strive to explain authoritarian regime dynamics by drawing comparisons: among all political regimes, between subtypes of regimes (democracies and autocracies), among authoritarian regimes, between

338

Empirical Explorations

subtypes of authoritarian regimes, or within subtypes of authoritarian regimes. Depending on how we slice the universe of regimes, our comparative frames of references change: 1. Comparing political regimes without adjectives: General theories of political regime dynamics strive to explain the emergence, persistence, and transformation of political regimes in general, regardless of their internal structure, their democratic or authoritarian nature. The goal is to explain variation in political stability. Samuel Huntington’s classical study on political mobilization and institutionalization in modernizing societies exemplifies this perspective (1968). 2. Comparing authoritarian and democratic regimes: General theories of regime types strive to explain the persistence and transformation of authoritarian regimes in comparison to democratic regimes. The goal is to explain variation in the durability of democratic and non-democratic regimes. The landmark study on economic development and political regime dynamics by Adam Przeworski and his collaborators (2000) is a prominent example. 3. Comparing authoritarian regimes: General theories of authoritarian regime dynamics strive to explain the persistence and transformation of authoritarian regimes in general, regardless of their internal structure. The goal is to explain variation in the durability of non-democratic regimes. Gordon Tullock’s influential essay on the political economy of autocratic regimes (1987) exemplifies this analytical perspective. 4. Comparing subtypes of authoritarian regimes: More specific theories of authoritarian governance look at patterns of stability and change across subtypes of non-democratic regimes. The goal is to explain variation in the durability of various categories of authoritarian regime. Barbara Geddes’ work on the duration and internal logics of single-party regimes, military regimes, and personalist regimes (1999) laid the foundation for this line of comparative inquiry. 5. Comparing regimes within authoritarian subtypes: At the most concrete level, comparative work on distinct subtypes of non-democratic rule seek to uncover the sources of stability and change within single categories of authoritarian regimes. The goal is to explain variation within one specific regime category. The edited volumes by Samuel Huntington and Clement Moore on single-party regimes (1970) and H. E. Chehabi and Juan Linz on personalist regimes (1998a) are examples of such inquiries into the inner logics of specific subtypes of authoritarianism. Most of the literature reviewed in Chapter 5 on the impact of multiparty elections on authoritarian survival assumes the penultimate perspective: It compares the durability of electoral autocracies across regime subtypes.

The Struggle over Regime Change

339

In contradistinction, in this chapter, I adopt the last perspective: I compare regime trajectories within the category of electoral authoritarian regimes. In comparative inquiries across regime subtypes, it is the general characteristics of subtypes that explain differences in their average performance. In comparative inquiries within regime subtypes, it is variations within subtypes that explain differences in the performance of individual regimes. In cross-regime comparisons, multiparty elections appear as the causal variable that explains the distinctive trajectories of electoral autocracies. In my within-regime comparisons, elections appear as an arena of struggle. Differences in electoral authoritarian regime trajectories cannot derive from the shared fact that these regimes hold multiparty elections. They must derive from systematic differences among them, that is, from differences in the structural context of elections, in the architecture of the electoral arena, or in the conflictive interactions that unfold within. Within the blossoming comparative literature on authoritarian politics, two important monographs have pursued essentially the same theoretical ambition I am pursuing in this chapter, which is, explaining stability and change of electoral authoritarian regimes: Competitive Authoritarianism: Hybrid Regimes After the Cold War by Steven Levitsky and Lucan Way (2010b) and Defeating Authoritarian Leaders in Postcommunist Countries by Valerie Bunce and Sharon Wolchik (2011). Both focus on more recent developments, Levitsky and Way on post-Cold War dynamics, Bunce and Wolchik basically on the 2000s (and post-communist countries). Their language is different, as the former talk about competitive authoritarian regimes and the latter about mixed regimes. Their concepts also differ somewhat from mine. Both pairs of co-authors are slightly less restrictive in drawing the line towards electoral democracies and more restrictive in drawing the boundary vis-à-vis hegemonic regimes. Levitsky and Way exclude hegemonic regimes by definition and Bunce and Wolchik by coincidence, as the post-Soviet space in the 2000s did not host any hegemonic regimes. In addition, Bunce and Wolchik strive to explain alternation in power, rather than regime change. Still, the overall explanatory project that both of these qualitative comparative inquiries pursue is very similar to mine here.

Levels of Theoretical Abstraction Now, when we go about explaining patterns of stability and change within one subtype of political regime, we can either choose to mobilize general explanations that are valid for any type of political regime. Or we can craft more specific explanations that are grounded in the institutional and strategic

340

Empirical Explorations

dynamics that are particular to the regime type in question. In this regard, the two monographs, Competitive Authoritarianism and Defeating Authoritarian Leaders, take contrasting routes. Levitsky and Way seek general explanations. They focus on two broad explanatory factors that are applicable to any type of authoritarian regime: the international embeddedness of regimes (linkage and leverage) and their organizational infrastructure (coercive capacity and party strength). As the authors posit, in situations of high linkage, when societies are densely interwoven with the external environment, competitive autocracies are likely to democratize. When countries are more isolated, such regimes are likely to remain stable unless they are structurally vulnerable to democratizing pressures from Western powers. Their argument about the primacy of international factors is meant to be time-specific rather than regime-specific. It is applicable to the post-Cold War period in general, not limited to competitive regimes in particular (although competitive authoritarian regimes happened to flourish in the post-Cold War period). Bunce and Wolchik, by contrast, focus their attention on domestic dynamics that are exclusive to electoral authoritarian regimes: the strategies opposition actors adopt towards multiparty elections. Asking about the electoral behavior of opposition actors only makes sense in contexts where multiparty elections actually take place. No elections, no electoral strategies, no electoral transitions. As the authors posit, if and only if opposition actors adopt the “electoral model” of transition, can they win against powerful incumbents. If they refrain from doing so, they will keep losing. Within the analytical framework of two-level games, the “electoral model” in essence implies that opposition actors take both levels of authoritarian elections seriously. On the one hand, they take the authoritarian nature of the electoral game seriously and struggle for institutional change at the meta-game level. Thus the importance of voter registration, electoral law reform, impartial electoral management, parallel vote tabulation, and so forth. On the other hand, they take the electoral nature of the authoritarian game seriously and struggle to win the hearts and minds of voters at the game level. Thus the importance of running creative and energetic campaigns, mobilizing the youth, increasing turnout, crafting positive messages, and so forth. Without institutional reform, winning votes is of limited use in authoritarian elections. Without effective campaigns, achieving institutional change is of limited use in competitive elections. While the two books differ in the regime specificity of their explanations, they also differ in the type of explanatory theories they offer. Levitsky and Way privilege structural factors and discard the weight of actor dynamics, Bunce and Wolchik privilege the choices of actors and hold structural factors to be secondary. The former seek explanatory variables that are distant from

The Struggle over Regime Change

341

regime outcomes, the latter variables that are more proximate. Levitsky and Way grant primacy to external factors, Bunce and Wolchik to domestic factors (even when they analyze the dynamics of international diffusion). My own conception of authoritarian elections as nested games in which regime and opposition actors compete over electoral uncertainty bears substantive affinities to Valerie Bunce and Sharon Wolchik’s approach. It involves a similar emphasis on regime-specific dynamics—on elections, actors, proximate causes, and domestic factors. Similarly, while my theoretical perspective privileges conflictive interactions within the electoral arena, I recognize their structural and institutional embeddedness.

REGIME TRAJECTORIES In electoral authoritarian regimes, multiple parties compete for positions of national state power in regular elections that transgress democratic minimum standards in severe and systematic ways. These regimes offer power-holders better prospects of political survival than democratic elections as well as better prospects of physical survival than military regimes. Yet, while authoritarian elections help governments to tighten their grip on power, they also contain the potential of loosening the chains of authoritarian control (see Chapter 5). Much of the contemporary literature on transitions from electoral authoritarianism has been focusing on the dynamics of “democratization by elections” (Schedler 2002a). Yet, when electoral autocracies fall they do not always fall on the democratic side. They may also suffer from regressions and terminate in civil war, military dictatorship, or some other kind of “closed” authoritarian regime. In this final empirical chapter I wish to explore the sources of both types of regime change: progressive democratizing change as well as the regressive closure of the electoral arena. To capture the direction of regime change, rather than just the simple fact of change, I construct a categorical measure that registers the fate of electoral authoritarian regimes from one election to the next regularly scheduled national election. If the subsequent election takes place under continuing authoritarian conditions (with a Freedom House political rights score of four or higher) I register regime continuity. If it takes place under minimally democratic conditions (with a Freedom House political rights score of three or lower), I register democratization. If the authoritarian electoral cycle comes to a halt through civil war, military coup, the suppression of opposition, or the extension of presidential mandates by more than half of the original term (by decree or referendum), I register a case of interruption. If hegemonic

342

Empirical Explorations

Table 10.1. Trajectories of electoral authoritarian regimes: frequency distributions N

%

Competitive elections Continuity Democratization Interruption Total

68 20 5 93

73.1 21.5 5.4 100.0

Hegemonic elections Hegemonic continuity Competitive authoritarianism Democratization Interruption Total

45 6 3 3 57

78.9 10.5 5.3 5.3 100.0

Source: Author’s Dataset on Authoritarian Elections (1980–2002).

governments lose their legislative dominance (a qualified majority of twothirds of parliamentary seats), yet remain authoritarian, I register a transition from hegemonic to competitive authoritarianism. Table 10.1 displays the frequency distribution of different regime trajectories within my sample of elections. The overall conservative penchant of non-democratic elections is clear. In an overwhelming majority of cases, authoritarian elections breed authoritarian elections. Almost three out of four competitive elections (73.1 percent) and almost nine out of ten hegemonic elections (89.5 percent) are followed by another round of authoritarian elections. Interruptions of authoritarian election cycles are rare events. In both subtypes of electoral autocracies, they do not exceed 5 percent of cases. Remarkably, though, more than one-fifth of competitive authoritarian elections pave the way for democratic change (21.5 percent), while a significant portion of hegemonic elections lead either to democratization (5.3 percent) or a transition to competitive authoritarianism (10.5 percent). Given my low case numbers, in the subsequent analysis I merge the last two outcomes into the aggregate category of “hegemonic opening” (15.8 percent).

S O C IE TA L S T RU C T U R ES One leitmotiv of the preceding chapters has been the “embedded autonomy” of authoritarian multiparty elections: they have a logic of their own, yet are inserted into constraining and enabling contexts. Structural contexts are multi-consequential. They affect the formation of actors, their resources, their

The Struggle over Regime Change

343

motives. In theories of regime change, societal structures play two fundamental roles. They carry implications for the distribution of resources and thus the balances of power both within society and between state and society. And they determine the stakes of politics and thus the intensity of regime conflicts. The debate on democracy and development, for example, tends to emphasize the former aspect, the debate on democracy and inequality the latter. Economic development changes balances of power. It empowers private individuals, as it permits them to secure their personal welfare outside the realm of politics. Social inequality affects calculations of interest. It makes the rich dependent on political protection and thus motivates them to support authoritarian regimes. While enlightening, the distinction between political resources and political stakes collapses in authoritarian contexts where actors compete for resources one of them controls.

Resources and Stakes: Two Sides of the Same Coin Authoritarian elections are “asymmetric games” in which one powerful actor controls the rules of the game while all others compete against him under unfair and insecure conditions (see Chapter 4). Struggles over authoritarian regime change in general pertain to this class of asymmetric conflicts. In these unequal power games, players compete over the resources one of them commands. The stakes of the game are not external to the distribution of resources among players. They are defined by it. In symmetric games, like professional soccer matches, the resources that players control are not at stake. Players possess similar resources and they compete over a variable prize that is exogenous to the game itself. Competing teams may differ in their physique, playing skills, or the color of their boots. But they do not differ in their basic resource endowments. None has more players than the other, or smaller goals, or the power to change rules at convenience. Both teams play under equal conditions for the same objective price, whatever it is and entails, be it the joy of winning or the fifa World Cup. If they win, they do not earn the right to enslave their contenders. If they lose, they do not have to fear expropriation by the winning team. In power struggles, the prize of competition is power. Under democratic conditions, these struggles are minimally symmetric: incumbents (the temporary occupants of positions of power) cannot use the resources of power to crush their opponents. The competition between roughly equal contenders is fundamentally exogenous to the possession of state power. Under authoritarian conditions, by contrast, power struggles are asymmetric: incumbents deploy the resources of power to keep their opponents from winning power.

344

Empirical Explorations

The competitive balance between deeply unequal contenders is endogenous to the distribution of state power. In asymmetric power games, players of radically diverging resource endowments compete, not over some external prize, but over the very position of the dominant player, over the very resources he controls and uses to reproduce his dominance. His resources are the stakes of the game. The resources that authoritarian incumbents control therefore determine both power relations and interest calculations. First, the higher the resource advantages of rulers, the greater their power, and thus their likelihood of prevailing in regime struggles. Second, the higher the advantages of incumbency, the higher the stakes of politics. Rulers have more to lose and challengers more to win when their resource endowments are highly asymmetric. The implication is simple: well-endowed rulers have both the power to defend the status quo and an interest to do so with high dedication. When all is said and done, though, what is likely to carry the game is not the intensity of their desire to defend their privileges, but their power to do so. Stakes define motives, resources capacities. Here I will emphasize the latter. I will understand structural contexts primarily as societal playing fields that grant or deny material resources to political actors.1

Hypotheses To trace the impact that societal structures have on the fate of electoral authoritarian regimes I will look at three broad properties of the structural environment: societal wealth, the availability of rents, and international economic integration. As indicator of wealth, I use annual gdp per capita income in current Purchasing Power Parities (ppp) and as indicator of rent endowment, the percentage share of mineral fuel exports in all exports. As measures of international linkage, I use the share of international trade in Gross Domestic Product (gdp), the share of foreign direct investment in gdp, and the share of official development assistance in Gross National Income (gni).2 Except for international linkages, my hypothetical expectations are straightforward: Average wealth: Economic development widens opportunities of legitimate self-enrichment outside the state. It also broadens societal access to education and long-distance communication. Overall, it tends to shift the balance of power between state and society in favor of the latter. As we know from Adam Przeworski and his co-authors (2000), national wealth stabilizes democracy, yet fails to destabilize dictatorship (see also Boix and Stokes 2003). Still, wealth might be able to destabilize one specific category of dictatorship: electoral

The Struggle over Regime Change

345

autocracies, where private property owners can invest with relative ease in peaceful regime-subversive activities, such as party building and electoral campaigning. Rent endowment: Rents are earnings without effort. Rentier states enjoy access to quick and easy wealth that is not available in the private sector. The more rents they have at their disposal, the easier it is to sustain authoritarian rule. Rentier autocracies possess the means to seduce and oppress people, to purchase their hearts and minds and all the necessary weapons of mass repression, too. International integration: International economic linkages make everybody more dependent on the outside environment, state agents as well as societal actors. Dependence on liberal–democratic countries should strengthen agents of democratization, dependence on non-democratic powers like Russia and China should strengthen authoritarian rulers. On average, trade dependence should therefore leave the prospects of electoral authoritarian regime change unaffected. By comparison, I expect both foreign direct investment (fdi) and official development assistance (oda) to show more consistent effects on regime stability, albeit with opposite signs. Assuming international firms to be primarily interested in political stability, not democracy, I expect fdi flows to favor the authoritarian status quo. Assuming international donors to be genuinely interested in democracy, despite countervailing foreign policy concerns, I expect official development assistance to erode authoritarian governance (slightly, at its margins).

Findings Let us first consider the association between domestic structures and regime trajectories. If high rents as well as high poverty tilt the playing field between state and society in favor of the former, do the structural incumbency advantages they create affect the probability of regime change in the wake of authoritarian elections? Figure 10.1 displays the average levels of per capita income and oil exports that electoral authoritarian regimes experienced in election years before they took different trajectories.3 With respect to wealth (gdp per capita in current ppp), it reveals a pronounced asymmetry between the two regime subtypes. Competitive regimes that democratize are richer on average (ø = 3.446 ppp) than those that break down (ø = 1.378 ppp). In hegemonic regimes, it is the other way round: those that open up are comparatively poor (ø = 2.310 ppp), those that remain stable relatively rich (ø = 4.177 ppp).4 Average wealth appears to play opposite roles in competitive and hegemonic regimes. In competitive regimes, wealth favors democratic change, while

GDP per capita (thousands PPP) (mean)

346

Empirical Explorations

5.00

4.00

4.00

3.00

3.00 2.00 2.00

3.45

4.231 2.67 2.31

1.00

1.00 1.38 0.00

0.00 Continuity (N = 45)

Opening (N = 9)

Continuity (N = 67)

Oil exports (% merchandise exports) (mean)

Hegemonic regime change

Democratization (N = 20)

Interruption (N = 5)

Competitive regime change

30.00

30.00

20.00

20.00 28.12

25.02

10.00

10.00 14.64 7.31

6.54 0.00

0.00 Continuity (N = 36)

Opening (N = 7)

Hegemonic regime change

Continuity (N =46)

Democratization (N = 20)

Interruption (N = 3)

Competitive regime change

Figure 10.1. Societal playing fields and regime trajectories Note: Blank columns = non-significant differences in means (p ≤ .15). Significance tests for hegemonic regime change: bilateral t-tests for independent samples; for competitive regime change: analyses of variance; significance of intergroup differences: Tukey hsd post hoc Test.

poverty is destabilizing. Inversely, in hegemonic regimes, wealth strengthens the status quo, while poverty favors processes of opening. These contrasting effects may derive from the potential scissor effect that economic development has on political opposition. It empowers society, but also anesthetizes it. It makes society richer in resources to challenge the state, yet poorer in motives for doing so. If the inmates of authoritarian regimes are well-off economically and attribute their well-being to the political regime, they have reasonable reasons for supporting it. Hegemonic regimes are by definition long-lasting regimes. Voters are likely to give them credit for the level of material wealth they help to create over their lifespan. As my data suggest, if hegemonic parties set their countries on paths of authoritarian development, they are able to reap the fruits of their own success, rather than being undermined by it.

The Struggle over Regime Change

347

The endowment of regimes with easy income (oil rents) shows more consistent associations across regime subtypes. In both competitive and hegemonic regimes, the availability of oil rents appears to favor authoritarian continuity. Both in competitive and hegemonic regimes, conservative elections that reproduce the status quo take place in countries that on average earn more than a quarter of their export income through mineral fuels. Transformative elections take place at much lower levels of oil dependency in both subtypes of regimes (although in hegemonic regimes differences in means are not statistically significant due to low case numbers). Dropping for a moment the distinction between competitive and hegemonic regimes, Table 10.2 highlights the implications of poverty and oil dependence for all electoral authoritarian regime trajectories. Just as the literature on natural resources and political regimes suggests (and as our disaggregate data indicated), oil is a powerful stabilizing factor. Authoritarian elections in oil-exporting countries (with a share of mineral fuel exports ≥ 25 percent of total exports) almost never lead to anything other than another round of authoritarian elections (96.9 percent). By contrast, just as the literature on economic development and political regimes suggests, poverty is a powerful destabilizing factor. While only half of elections in my sample take Table 10.2. Trajectories of poor and oil-dependent electoral authoritarian regimes Non-poor country

Poor country

Non-oilexporting country

Oilexporting country

Regime continuity

N Row % Column % Total %

59 50.0% 76.6% 39.6%

59 50.0% 81.9% 39.6%

57 64.8% 69.5% 50.0%

31 35.2% 96.9% 27.2%

Democratization

N Row % Column % Total %

17 73.9% 22.1% 11.4%

6 26.1% 8.3% 4.0%

21 95.5% 25.6% 18.4%

1 4.5% 3.1% .9%

Interruption

N Row % Column % Total %

1 12.5% 1.3% .7%

7 87.5% 9.7% 4.7%

4 100.0% 4.9% 3.5%

N Row % Column %

77 51.7% 100.0%

72 48.3% 100.0%

82 71.9% 100.0%

Total

0 .0% .0% .0% 32 28.1% 100.0%

Note: Poor country = annual gdp per capita < 2,000 ppp (current). Oil-exporting country = mineral fuel exports ≥ 25 percent of total exports.

348

Empirical Explorations

place in “non-poor” countries (with annual gdp per capita ≥ 2,000 ppp), these comparatively well-off countries host almost three-quarters of all democratizing elections (73.9 percent). In addition, practically all experiences of electoral authoritarian breakdown in my sample (seven of eight) have taken place in poor countries (87.5 percent). In the overall sample of electoral autocracies, the level of economic development seems to matter less for the probability of change than for the direction of change. Wealth favors democratization, poverty breakdown. The availability of oil rents, by contrast, favors regime stability. Oil breeds electoral authoritarian continuity (in 96.9 percent of all cases). To estimate the magnitude of effects that wealth and oil have on electoral authoritarian regime trajectories, I calculate a series of logistic regressions.5 Table 10.3 displays the results. Just like previous tables that reported logistic regression results, it does not show regression coefficients, but odds ratios (eb) that estimate the change in the odds of different regime outcomes for one-unit increases in independent variables. Odds ratios of one indicate nil effects, below one negative effects, and above one positive effects. As in subsequent logistic regressions, the reference category for both competitive and hegemonic elections is the status quo: regime continuity. When interpreting the results, we should keep in mind that both subtypes of electoral authoritarian regimes are rather stable. Transitions are the exception rather than the rule. The baseline odds of regime change are calculated by dividing the frequency of regime change by the frequency of regime continuity. With my sample, the odds that competitive regimes democratize are 0.29, that they break down 0.07. The odds that hegemonic elections are followed by a process of regime opening are 0.19 (see Table 10.1 above). The regression results basically bear out the asymmetrical relationships that we saw in our descriptive data. In competitive regimes, wealth (the decimal logarithm of per capita income in current ppp) multiplies the odds of democratization by four (eb = 4.38), while it compresses the odds of interruption almost to zero (eb = .06). Note that a one-unit change in the logarithm of wealth implies a “big leap” from one world into another, as from an annual per capita income of one thousand (log = 3) to a per capita income of ten thousand (log =4). In my dataset, Kenya in 1997 comes closest to the former, Singapore in 1988 closest to the latter. In hegemonic regimes, such a hypothetical transformation of worlds depresses the odds of opening by four-fifths (eb = .23) (albeit below conventional levels of statistical significance). Logistic regression results also confirm that access to discretionary sources of patronage inhibits democratization. Every additional percentage share of mineral resources in total merchandise exports lowers the odds of competitive democratization by almost 5 percent (eb = .955) (while its negative impact on hegemonic opening is non-significant).

The Struggle over Regime Change

349

Table 10.3. Societal playing fields and regime trajectories cr Democratization

Societal wealth Wealth Constant Chi2 / pac Cox & Snell R2 N Rent endowment Mineral fuel exports Constant Chi2 / pac Cox & Snell R2 N External dependence International trade Foreign direct investment Official dev. assistance Constant Chi2 / pac Cox & Snell R2 N

cr Interruption

hr Opening

eb

p

eb

p

eb

p

4.383

.102 .045

.062

.140 .283

.230 27.724 83.3% .035 57

.187 .368

.054 .430

.961

.429 .002

.978 .315 83.3% .036 42

.271 .029

.220 .271 .059 .493

.962 .954 1.001

.199 .798 .993 .877

.967 1.581 .964 1.065 84.8 .114 46

.078 .124 .630 .957

6.018 .063 92 .955 8.978 .122 69 .987 .869 .897 12.760 .136 87

Note: Multinomial logistic regressions for competitive regimes (cr), binary logistic regressions for hegemonic regimes (hr). Reference category: regime continuity. Wealth: log of current ppp per capita. Mineral fuel exports: % of all exports. International trade: % of gdp. Foreign direct investment: % of gdp. Official development assistance: % of gni. pac = Percentage of accurate classification. eb = odds ratios. Shaded coefficients = statistically significant (p ≤ .15).

Let us finally turn our attention to external factors. The regression results for international linkages support the intuition that different types of links differ in their implications for electoral authoritarian trajectories. Surprisingly, foreign investment goes hand in hand with higher odds of opening in hegemonic regimes. Each additional percentage point of fdi (as share in gdp) elevates the odds of hegemonic opening by more than 50 percent (eb = 1.581). Depressingly, external development assistant decreases the odds of democratization in competitive regimes. Each additional percentage point of oda (as share of gni) lowers them by 10 percent (eb = .897). International trade integration seems to work in favor of authoritarian stability as well. It consistently depresses the odds of regime change (although below conventional levels of significance in

350

Empirical Explorations

competitive regimes). Together these three thin indicators account for more than 10 percent of variance in regime trajectories (with Cox and Snell R2 of .136 for competitive and .114 for hegemonic trajectories). Overall, integration into the global economy appears to favor the political status quo, rather than encouraging authoritarian openings. Only foreign direct investments in hegemonic regimes seem to favor change, rather than continuity. Summing up: In competitive regimes, wealth works as it is supposed to, strengthening societal pressures for democratizing change, while poverty makes these regimes more vulnerable to authoritarian regression. Hegemonic regimes, by contrast, benefit from economic development. Paraphrasing Seymour Martin Lipset (once again), the more well-to-do a hegemonic regime, the greater the chances that it will sustain authoritarian rule (see Lipset 1981: 31). The economic success of hegemonic regimes translates into political security. The authoritarian blessings of oil rents, by contrast, stabilize both subtypes of regimes. The resource asymmetry between state and society which oil wealth creates works as a powerful brake on the probabilities of regime change. Electoral petro-autocracies seem much more difficult to assail than rent-poor regimes. Finally, international linkages, too, appear to favor electoral authoritarian continuity, rather than subverting it.

I N S T I T U T I O NA L S T RU C T UR ES While their structural environment is largely exogenous to electoral authoritarian regimes, although much less so to hegemonic regimes, their institutional environment is largely endogenous to them, although much more so in hegemonic regimes. According to Jennifer Gandhi, small governing bodies and consultative councils are “the first institutional trench” that authoritarian regimes excavate in their perpetual war of political survival. Legislatures and elections constitute the “second institutional trench” (Gandhi 2008: 166). From this perspective, we can understand the institutional architecture of governance and elections as the “third institutional trench” of electoral authoritarian self-defense.

Hypotheses Not all micro-institutional arrangements grant clear advantages to the ruling party. Some are more ambiguous and concede competitive opportunities to the opposition. I hypothesize that majoritarian electoral formulae and

The Struggle over Regime Change

351

parliamentary forms of government establish clear institutional incumbency advantages, while concurrent elections and subnational elections involve significant institutional opposition opportunities. Electoral systems: Majoritarian winner-takes-all systems tend to obstruct the access of opposition parties to legislative representation. They favor the ruling party, as long as a majority of votes favors the ruling party (be it through popular support or authoritarian manipulation). Parliamentarism: In presidential systems, electoral autocrats conserve executive power by manipulating elections. In parliamentary systems, they enjoy an additional layer of manipulation. To prevent alternation in government, they can manipulate legislative parties and assemblies. Concurrent elections: When legislative and presidential elections take place in temporal vicinity, the stakes of electoral politics go up. Ruling as well as opposition parties are therefore driven to step up their campaign efforts in concurrent elections. Even though their additional efforts are likely to neutralize each other, I hypothesize that electoral concurrence renders incumbents more vulnerable. Subnational elections: Subnational elections in authoritarian regimes are ambiguous affairs. They open up safety valves for governments who need to release pressure from below. Yet, perhaps more importantly, they also constitute opportunities for opposition parties to learn, grow, and consolidate, and eventually to provoke and propel processes of “centripetal democratization” (Mizrahi 2002). As I posited in previous chapters, issues of institutional design need to be distinguished from issues of institutional strength (even if institutional weakness is often a matter of authoritarian design as well). What are the implications of micro-institutional strength for electoral authoritarian survival? My expectations are contingent on the nature of the institutions in question. Regimes should be more resilient if institutions of domination are stronger and institutions of representation weaker.6 I take the degree of personalism to indicate the strength of institutions of domination, and the power of legislatures and the length of electoral cycles to indicate the strength of formal institutions of representation. Personalism: Where institutions are strong, the role of actors recedes to the background. Where they are weak, actors move to center stage. By definition, personalist regimes are weakly institutionalized regimes in which one individual leader dominates the political scenery. He does not rule with institutions, but with fluctuating coalitions of clients. Such regimes may last for long stretches of time. Still, due to their dependence on singular rulers who refuse to constrain themselves through formal rules, they live in perpetual anticipation of success crises. Multiparty elections, furthermore, grant structural

352

Empirical Explorations

opportunities for elite division, the defection of state agents, and opposition mobilization (see Chapter 5). I therefore expect personalist electoral autocracies to be more vulnerable to regime change in the wake of national elections than their impersonal counterparts. Legislative weakness: Authoritarian legislatures are potential sites of opposition building and party building. By keeping them weak, oppositions and parties are both kept weak (see Fish 2006). Electoral sequences: In and by themselves, elections play an ambiguous role for regime longevity (see Chapter 5). While they prolong the life expectancy of authoritarian regimes they also involve the risk of leading them down the road of either slow or sudden death by elections.7 Whether the longevity of elections themselves matters for the longevity of regimes has been a matter of considerable controversy. Staffan Lindberg has argued that the institutionalization of multiparty elections, their mere repetition over uninterrupted cycles, encourages the expansion of democratic liberties (see Lindberg 2006b and 2009c). McCoy and Hartlyn (2009), among others, have contested this claim.

Findings To what extent do these multifarious features of the micro-institutional environment shape the trajectories electoral authoritarian regimes take in the aftermath of national elections (until the arrival of the next regularly scheduled election)? Given my low case numbers (in particular for hegemonic regimes), I refrain from running logistic regressions and only compare the distribution of institutional attributes across regime trajectories. Table 10.4 displays average scores. Only two variables show significant and consistent associations with regime trajectories: majoritarian electoral systems and legislative weakness. These two appear as the institutional foes of democratizing change in electoral authoritarian regimes. In hegemonic regimes, not a single case of opening took place under pure majoritarian electoral rules and only two under ineffective legislatures. In competitive regimes, nine or 47.4 percent of all transitions took place under pr systems, while only eight or 16.7 percent of all continuity-reproducing elections did so. Only three instances of competitive regime change took place in the presence of rubber-stamp legislatures. While majoritarian electoral systems and weak legislatures tend to stabilize all electoral autocracies, concurrent elections tend to destabilize competitive regimes, but not hegemonic regimes. While less than one-third of all elections in competitive regimes are concurrent (31.9 percent), more than one-half of

The Struggle over Regime Change

353

Table 10.4. Institutional playing fields and regime trajectories Competitive regimes

Hegemonic regimes

Competitive Democratic Interruption continuity change

Hegemonic continuity

Regime opening

Institutional design Majoritarian ø electoral system n Parliamentarism ø n Concurrent ø elections n Subnational ø elections n

.83 48 .10 68 .24 68 .54 28

.53 19 .20 20 .55 20 .63 8

1.00 3 .00 5 .60 5 .75 4

1.00 25 .36 45 .18 45 .53 30

.75 4 .22 9 .33 9 .40 5

Institutional strength Personalism ø n Legislative ø ineffectiveness n Sequence of ø elections n

.33 39 .54 68 1.46 68

.12 17 .15 20 1.65 20

.33 3 .80 5 1.40 5

.22 45 .60 45 2.07 45

.22 9 .22 9 1.89 9

Note: Comparison of means, differences highlighted in gray are statistically significant (p ≤ .05) (anova Analyses of variance). Parliamentarism: 0 = presidential or semi-presidential system, 1 = parliamentarism. Majoritarian electoral system: 0 = Proportional representation, 1 = majoritarian or mixed-member system (Source: Gerring, Thacker, and Moreno 2005). Legislative ineffectiveness: 0 = effective or partially effective legislature, 1 = ineffective legislature (Source: cnts). Subnational elections: 0 = neither municipal nor state elections, 1 = either municipal or state elections or both (Source: dpi). Personalism: 0 = non-personalist regime, 1 = personalist regime or mixed regime with personalist traits (Source: Geddes 1999). Sequence of elections: 1 = 1st election, 2 = 2nd or 3rd election, 3 = 4th or subsequent election.

all democratizing elections are (55 percent).8 None of the other institutional variables shows systematic covariation with the post-electoral fate of regimes. In their short-term trajectories, parliamentary electoral autocracies resemble their presidential counterparts, just as personalist regimes resemble their more impersonal counterparts. Also, the countervailing effects of subnational elections on electoral authoritarian survival seem to cancel each other out. The most remarkable negative finding, though, concerns the institutionalization of elections. On average, electoral authoritarian regimes are no more likely to endure, or to fall, or to fall in either direction, as they grow older. In competitive as well as in hegemonic regimes, different regime trajectories are preceded by similar mean numbers of elections. The length of electoral cycles does not allow us

354

Empirical Explorations

to predict the post-electoral fate of electoral authoritarian regimes. To achieve democratic gains it is not enough to kick off the wheel of elections and keep it spinning. To repeat my mantra: Elections are not causes, but arenas. The cause of democratizing change cannot be won unless opposition gladiators step into the arena and face the authoritarian lion. No fight, no victory.

V E RT IC A L U N C E RTA IN TI ES Neither societal nor institutional structures threaten authoritarian regimes. They constitute sources of threat (or tranquillity) only to the extent that they facilitate (or obstruct) the emergence of collective challenges. Much of the literature on authoritarian regime dynamics focuses on horizontal, rather than vertical or external challenges. As Guillermo O’Donnell and Philippe Schmitter famously asserted in their foundational essay on transitions from authoritarian rule, “there is no transition whose beginning is not the consequence—direct or indirect—of important divisions within the authoritarian regime” (1986: 19). When the coherence of the ruling elite turns uncertain, the continuity of the regime turns uncertain. The centrality of street protests in “fourth wave” transitions in Eastern Europe and sub-Saharan Africa, in subsequent “Color Revolutions” in post-communist countries, and in the “Arab rebellions” of 2011 has led scholars to reevaluate the potential of civic protest in provoking democratizing change.9 Yet, while single-party regimes and military dictatorships are occasionally vulnerable to challenges from below, electoral authoritarian regimes make themselves systemically vulnerable to such challenges. Multiparty elections open up formal venues to challenge the authoritarian regime in a peaceful manner. If they constitute “the first and most important arena of contestation” (Levitsky and Way 2002: 54) in electoral autocracies, we should expect them to constitute the first and most important source of threat to the survival of these regimes. Thus, paraphrasing O’Donnell and Schmitter, I hypothesize that “there is no transition from electoral authoritarian rule whose beginning is not the consequence—direct or indirect—of uncertainties about the electoral support of the regime.” Conceptually, electoral uncertainty is a form of vertical uncertainty. Causally, it originates in vertical as well as in horizontal uncertainties. If authoritarian election results are the joint product of voter preferences and regime manipulation (see Chapter 4), their uncertainty is the joint product of vertical and horizontal uncertainties. It arises from vertical uncertainties about the electoral support of the regime as well as from horizontal

The Struggle over Regime Change

355

uncertainties about the support of the regime by state actors. In addition, these two sources of uncertainty feed each other. Once voters turn unreliable it is hard to prevent state agents from turning unreliable too. Vertical uncertainties tend to erode horizontal certainties. These interdependencies justify my deterministic proposition about vertical uncertainty as sine qua non of electoral authoritarian transitions (a necessary, though not sufficient cause). Note, though, that electoral authoritarian interruptions through military coup or civil war may unfold in determined indifference to voter preferences. My hypothesis concerns processes of democratization: progress from electoral authoritarianism to electoral democracy depends on the emergence of electoral uncertainties. My thin quantitative data do not allow me to trace the role of vertical and horizontal uncertainties, both evolving and interacting, in the reproduction and transformation of electoral authoritarian regimes. They do allow me, though, to examine short-term associations between various “latent” as well as “manifest” sources of vertical uncertainty on the one side and subsequent regime outcomes on the other.

Latent Vertical Threats So far I have looked at contextual factors as sources of power. Now I wish to consider them as potential sources of grievance. I will examine one set of structures—societal cleavages that make it more difficult for regimes to generate societal support—and one set of junctures—economic crises that make it more difficult to maintain societal support. Since these contextual factors lie at a causal distance from institutional struggles, I treat them as “latent” causes of vertical uncertainty. Societal cleavages: In principle, although dictators are often adroit in exploiting societal divisions, it should be easier to run an authoritarian regime (any kind of regime, as a matter of fact) in a homogeneous society than in a heterogeneous one. Through logistic regression, I explore the effects that four indicators of societal cleavages have on electoral authoritarian regime trajectories: social inequality (Gini coefficient of monetary income distribution), linguistic and religious polarization (with both indicators measuring the “cultural bipartisanship” of national societies), and population size (under the assumption that larger countries are bound to be more diverse).10 As the regression results in Table 10.5 indicate, the overall explanatory capacity of societal cleavages for short-term regime trajectories is low. These “latent” sources of threat appear to leave the dynamics of competitive regimes untouched. Yet, their individual effects on hegemonic opening look

356

Empirical Explorations

Table 10.5. Latent vertical uncertainties and regime trajectories cr Democratization

Societal cleavages Inequality (Gini) Linguistic polarization Religious polarization Population size (millions) Constant Chi2 / pac Cox & Snell R2 N Economic performance gdp p.c. growth (election year) gdp p.c. growth (5 years before ø) Inflation (election year) Inflation (5 years before election ø) Constant Chi2 / pac Cox & Snell R2 N

cr Interruption

hr Opening

eb

p

eb

p

eb

p

1.029 .500 .692 .998

.349 .446 .695 .817 .277

1.025 1.199 .210 1.012

.678 .927 .368 .338 .266

1.124 .052 32.731 1.011 .001 82.2% .089 45

.116 .157 .195 .241 .047

.268 .992 .390 .629 .051

.752 1.127 .950 .955

.095 .552 .388 .466 .173

1.077 .764 .936 1.143 .112 85.4% .150 48

.541 .166 .349 .075 .003

3.660 .043 84 .920 .999 .989 .999 9.508 .164 70

Note: Multinomial logistic regressions for competitive regimes (cr), binary logistic regressions for hegemonic regimes (hr). Reference category: regime continuity. pac = percentage of accurate classification. eb = odds ratios. Shaded coefficients = statistically significant (p ≤ .15).

intriguing (although case numbers are low and most coefficients lie below conventional thresholds of statistical significance). Social inequality, often conceived as the social glue of authoritarian governance, raises the chances of hegemonic opening instead of lowering them (eb = 1.12). Linguistic and religious “bipartisanship” work in opposite directions. Linguistic polarization goes hand in hand with impressive stability, religious polarization with impressive instability. Never trust a stranger (a stray correlation in search of a theory). Yet if we can believe these figures, the former pushes the odds of hegemonic opening almost to zero (eb = .05), the latter multiplies them by more than thirty (eb = 32.73). Economic performance: Times of economic crisis are prone to lead regimes into times of political unease. When the economy goes well, observers tend to suspect that authoritarian regimes enjoy a cushion of “performance legitimacy” or “specific support” (David Easton). When the economy runs into trouble, observers tend to suspect that authoritarian

The Struggle over Regime Change

357

regimes may do the same. When the reservoirs of patronage resources dry up, the reservoirs of “instrumental” regime support are bound to do so as well. Poverty, unemployment, and inflation are unlikely to make voters happy with those who have brought these calamities upon them through persistent political mismanagement. Through logistic regression, I explore the effects that two standard measures of economic performance have on electoral authoritarian regime trajectories: gdp per capita growth and changes in consumer prices. For both indicators I use annual figures for the election year as well as average values for the five years before. As the regression results in Table 10.5 indicate, these four indicators of economic performance carry considerable explanatory capacity. They account for more than 15 percent of variance in competitive as well as in hegemonic regime trajectories. However, the two subtypes of regimes respond in contrasting ways to economic junctures. Competitive regimes are sensitive to short-term oscillations, hegemonic regimes to medium-term changes. Competitive autocracies primarily respond to short-term variations in per capita growth. Each additional percentage of growth in the election year lowers the odds of democratization by almost 10 percent (eb = .92) and the odds of breakdown by a dramatic 25 percent (eb = .75). Hegemonic autocracies respond to medium-term changes in both growth and inflation. Each additional percentage of average growth during the five years preceding a national election lowers the odds of subsequent opening by one-quarter (eb = .76). Each percentage of average consumer price inflation increases the odds of opening by more than one-eighth (eb = 1.14). Less dependent on their medium-term performance, competitive regimes earn the fruits of current bonanzas and fall victims of present crises. Less dependent on short-term success, hegemonic regimes earn the fruits of past achievements and fall victims of medium-term failure (see also Magaloni 2006).

Electoral Uncertainty Latent threats can lie latent forever. Governments may worry about them and their worries may be instrumental in keeping them latent. Yet once latent uncertainties break through the surface of public visibility and manifest themselves in collective challenges, their nature changes. They lose their hypothetical quality, become tangible, and demand tangible responses. In the remainder of this section, I will look at three public manifestations of vertical uncertainty: electoral uncertainty, electoral protest, and political protest outside the electoral arena.

358

Empirical Explorations

As they derive from uncertain mixes of voter choices and state decisions, authoritarian election outcomes are hybrids. The uncertainties they generate are hybrids, too. They are uncertain mixes of vertical uncertainties that originate in citizen behavior and horizontal uncertainties that originate in the behavior of state officials. To simplify matters, I here discuss electoral results as manifestations of vertical uncertainties. I use two rough indicators: the fragmentation of opposition parties (Rae index) and the winning party’s margin of victory (vote percentages in presidential and seat percentages in legislative elections).11 The unity of opposition parties is often understood as an indispensable ingredient of electoral turnover. In numerous authoritarian elections, electoral coordination by opposition actors appeared as a crucial facilitating condition of authoritarian regime change (see Bunce and Wolchik 2011, Howard and Roessler 2006, van de Walle 2006). According to my data, by contrast, different regime trajectories are not preceded by different mean levels of opposition fragmentation. In comparisons of means (not reported) as well as in logistic regressions (see Table 10.6), opposition fragmentation appears to be unrelated to the subsequent fate of electoral authoritarian regimes. In and by themselves, neither opposition fragmentation, nor unity seem to set regimes on diverging trajectories. Differences in mean margins of victory, by contrast, are striking across competitive regime trajectories. Elections that predate democratization are more competitive than those that presage continuity. Their legislative margins of victory are almost 20 percent lower, and their presidential margins little less than 30 percent. However, the interruption of electoral cycles in Table 10.6. Electoral competitiveness and regime trajectories cr Democratization

Margins of victory Opp. fragmentation Constant Chi2 / pac Cox & Snell R2 N

cr Interruption

hr Opening

eb

p

eb

p

eb

p

.982 1.123

.020 .911 .181

.986 .475

.305 .667 .070

.990 .423 .558 83.3% .014 54

.528 .528 .630

6.611 .072 88

Note: Multinomial logistic regressions for competitive regimes (cr), binary logistic regressions for hegemonic regimes (hr). Reference category: regime continuity. Opp. fragmentation = opposition fragmentation (Rae index). pac = percentage of accurate classification. eb = odds ratios. Shaded coefficients = statistically significant (p ≤ .15).

Presidential margin of victory (mean)

The Struggle over Regime Change 50.00 60.00

40.00 30.00

40.00

49.58 57.81

61.43

20.00

37.76

20.00 10.00 8.37 0.00

0.00 Continuity (N = 15)

Opening (N = 5)

Continuity (N = 34)

80.00

40.00

60.00

30.00

40.00

Democratization Interruption (N = 13) (N = 3)

Competitive regime change

Hegemonic regime change

Legislative margin of victory (mean)

359

20.00

73.04

33.18 57.99

20.00

10.00 14.65

0.00

16.67

0.00 Continuity (N = 38)

Opening (N = 7)

Hegemonic regime change

Continuity (N = 50)

Democratization Interruption (N = 18) (N = 5)

Competitive regime change

Figure 10.2 Electoral competitiveness and regime trajectories Note: Blank columns = non-significant differences in means (p ≤ .15). Significance tests for hegemonic regime change: bilateral t-tests for independent samples; for competitive regime trajectories: analyses of variance; significance of intergroup differences: Tukey hsd post hoc Test.

competitive regimes seems to be the work of domineering executives (ø margins = 49.6 percent) who lack domineering parties (ø margins = 16.7 percent) (see Figure 10.2). Hegemonic regimes by definition enjoy legislative supermajorities and high degrees of electoral certainty. On the eve of hegemonic regime change, presidential margins of victory remain at dizzying heights. Yet processes of hegemonic regime opening are preceded by significant fissures in legislative dominance. In elections that lead to regime opening, hegemonic margins of victory are about 15 percent lower than in legislative elections that spell regime continuity (see Figure 10.2). Multiple regression analyses for all elections do not quite bear out these patterns of association. Non-lineal associations as well as differences between legislative and presidential contests may account for these negative findings. The only significant relationship: higher margins of victory lower the odds of

360

Empirical Explorations

democratic change in competitive regimes by a ratio of 1:2. Each percentage increase in margins of victory depresses the odds of democratizing change by 1.8 percent (eb = .982) (see Table 10.6).

Electoral protest Even when opposition parties fail to create electoral uncertainty by mobilizing large numbers of supporters at the polls, they may succeed to introduce vertical uncertainties by mobilizing large numbers of protesters on the streets. More often than not, opposition protest does not tangibly affect election results. But it may well introduce informational uncertainty about their origins. As electoral authoritarian regimes retain power through messy mixtures of popularity and manipulation, no one knows for sure how popular they are nor how manipulative. Massive street protests may introduce and reinforce doubts about the popular support of the regime. They may persuade observers that electoral outcomes are the product of authoritarian maneuvers, rather than an expression of genuine voter preferences. As discussed in the previous chapter, opposition protest may work as an alternative source of information. Of course, it would be silly to expect any minor protest event to shake the foundations of the regime. So-called electoral revolutions, in which massive street protests against fraudulent elections end up overturning the official outcome of the election, are rare. My dataset contains only two cases, the Philippines in 1986 and Côte d’Ivoire in 2000. Still, it seems reasonable to ask whether the incidence of certain protest events augments the probability of subsequent regime change, and be it at the margins only. Once again, I use rough indicators of two types of opposition protest: election-related protest and contentious collective action between elections. The former refers to forms of contention (boycotts and street protests) that take place in the temporal vicinity of elections and are explicitly directed against the authoritarian quality of electoral procedures. The latter refers to a variety of protest events (anti-government demonstrations, general strikes, and mass riots) that take place after the election (in the three years subsequent to the election year) and are often unrelated to issues of regime change. Election boycott. Opposition boycotts of elections have a bad press among students of democratization because they seem to be self-defeating. While their long-term benefits are unclear, their short-term costs tend to be high. They deprive boycotting parties of experience, visibility, votes, and legislative positions, and thus mute their voice in the national political arena.12 Boycotting parties, however, know very well that they pay a price when they withdraw

The Struggle over Regime Change

361

from the electoral arena. They pay it in the hope of extracting democratizing concessions in the future. Yet to what extent do these hopes come true? Previous comparative research on the institutional effects of election boycotts has been divided. In his study of elections and liberalization in sub-Saharan Africa, Staffan Lindberg found that opposition boycotts favored the authoritarian status quo. Opposition parties had boycotted only 8 percent of elections that led to democracy, but 54 percent of those that did not. As Lindberg concluded, “the participation of opposition parties plays an important role in transforming electoral autocracies into democracies” (2006a: 161). By contrast, in her global study of election boycotts, Emily Beaulieu found that “major boycotts appear to encourage political reform” (2006: 129). When all opposition parties dutifully participate in an authoritarian election, nothing happens afterwards (in 96 percent of cases). When at least some major opposition parties boycott, the likelihood of subsequent reform triples. More than one-quarter of major boycotts (26 percent) are followed by subsequent political reforms (Beaulieu 2006: 129). Yet, as Beaulieu does not distinguish between progressive (democratizing) and regressive (authoritarian) reforms, the institutional net effects of opposition boycotts remain open. Intriguingly, within my sample of authoritarian elections, patterns of association between opposition boycotts and subsequent regime trajectories lend support for both propositions, for Lindberg’s conclusion that participation favors democratization as well as for Beaulieu’s finding that boycott favors reform. Table 10.7 displays the distribution of frequencies. In competitive regimes, democratizing elections show the highest rates of opposition participation as well as, at the same time, a high incidence of full opposition boycotts. As it seems, both full participation and full boycott augment the chances of democratic change, while partial boycotts reinforce the authoritarian status quo. Notably, four of the five cases of interruption were also preceded by either partial or full boycotts. While the overall probability of electoral authoritarian breakdown is low, it tends to occur after elections in which at least some opposition actors had withdrawn from electoral competition. The key to opposition success, it seems, does not lie in the nature of their choices, but in the coordination of their choices. Whatever opposition parties do, whether they participate or boycott, they should do together. In the realm of hegemonic party rule, electoral boycotts appear to assume a less ambiguous role. Even if the differences between boycott rates and subsequent regime trajectories lie below conventional thresholds of significance, transformative hegemonic elections were preceded by a higher incidence of full opposition boycotts than conservative ones. Three out of five full boycotts trigger processes of hegemonic opening (if triggering is what they do). Just like their competitive counterparts, hegemonic regimes seem to have less

362

Empirical Explorations

Table 10.7. Electoral protest and regime trajectories Competitive regimes

Hegemonic regimes

Competitive Democratic Hegemonic Regime continuity change Interruption continuity opening Election boycott Participation Partial boycott Full boycott

N Column % N Column % N Column %

45 66.2% 16 23.5% 7 10.3%

15 1 75.0% 20.0% 1 2 5.0% 40.0% 4 2 20.0% 40.0% Chi2 = 9.05, p = .06

31 5 67.4% 55.6% 10 1 21.7% 11.1% 5 3 10.9% 33.3% Chi2 = 3.20, p = .20

40 58.8% 28 41.2%

12 60.0% 8 40.0%

2 40.0% 3 60.0% Chi2 = .72, p = .69

38 5 82.6% 55.6% 8 4 17.4% 44.4% Chi2 = 3.23, p = .07

22 32.4% 26 38.2% 20 29.4%

9 0 45.0% .0% 5 1 25.0% 20.0% 6 4 30.0% 80.0% Chi2 = 7.34, p = .12

18 0 39.1% .0% 13 5 28.3% 55.6% 15 4 32.6% 44.4% Chi2 = 5.54, p = .06

Pre-electoral protest Acquiescence Active protest

N Column % N Column %

Post-electoral protest Acquiescence Rejection Active protest

N Column % N Column % N Column %

trouble in weathering partial boycotts than in managing united opposition boycotts. The logistic regression results reported in Table 10.8 reaffirm the destabilizing potential of boycotts. The non-lineal relationship between boycott and democratizing change in competitive regimes (where both full participation and full boycott facilitate regime change) washes out in regression analysis. Yet, each step along the trichotomous scale of boycott roughly doubles the odds of competitive interruption (eb = 2.26) as well as the odds of hegemonic opening (eb = 1.9) (although somewhat below conventional levels of significance). Pre-electoral protest. Opposition protests against an upcoming election are quite unlikely to affect the trajectory of the regime in the years following the election. In competitive regimes, as a matter of fact, they don’t. In hegemonic

The Struggle over Regime Change

363

Table 10.8. Contentious action and regime trajectories cr Democratization

Electoral protest Boycott Pre-electoral protest Post-electoral protest Constant Chi2 / pac Cox & Snell R2 N

cr Interruption

hr Opening

eb

p

eb

p

eb

p

1.147 1.052 .599

.742 .930 .502 .007

2.260 .689 30.937

.175 .727 .118 .004

1.900 4.275 3.844 .037 87.3 .115 55

.205 .089 .223 .001

.469 .019 .000

.975 1.299

.717 .015 .000

1.637 .069 .104 83.3% .157 55

.054 .098 .000

8.431 .087 93

Post-electoral contention Vertical threats .968 1.232 Armed rebellion Constant 9.968 Chi2 / pac .102 Cox & Snell R2 N 93

Note: Multinomial logistic regressions for competitive regimes (cr), binary logistic regressions for hegemonic regimes (hr). Reference category: regime continuity. pac = Percentage of accurate classification. eb = odds ratios. Shaded coefficients = statistically significant (p ≤ .15). Postelectoral contention: sum of events during the three years after election year. Vertical threats = anti-government demonstrations, general strikes, and riots. Armed rebellion = guerrilla warfare and revolutions (Source: cnts).

regimes, however, they do. Under hegemonic rule, the incidence of active mobilization in opening elections is more than twice as high (44.4 percent) as in stabilizing elections (17.4 percent). One-third of pre-electoral protests in hegemonic regimes lead to subsequent openings (see Table 10.7). Logistic regression results confirm the surprising sensitivity of hegemonic regimes to pre-electoral protests. When opposition parties decide to challenge upcoming elections through contentious action, the odds of hegemonic opening rise fourfold (eb = 4.27) (see Table 10.8). Post-electoral protest. With respect to post-electoral protest, we find similar dissimilarities between competitive and hegemonic regimes. In hegemonic regimes, acquiescence breeds continuity. Most protests do not lead anywhere, yet not a single instance of opening was preceded by opposition acquiescence to the previous elections. Protest, even in its mild, rhetorical form, seems almost a necessary condition for hegemonic change. Under competitive conditions, post-electoral mobilization also works as a catalyst of change, in both directions. Ten out of thirty instances of active protest are followed by regime change, six by democratization, four by interruption. By contrast, the

364

Empirical Explorations

ambiguous middle path of rhetorical rejection (condemnation without active mobilization) does not appear to move anything. It is basically a recipe for prolonging the status quo (see Table 10.7). Logistic regression results confirm the relevance of post-electoral protest for the fate of electoral autocracies. Again, the U-shaped relation between protest and competitive regime trajectories washes out the impact of active protest on democratizing change. Yet each step along the trichotomous scale of post-electoral protest multiplies the odds of hegemonic opening by four (eb = 3.84) (although below conventional levels of significance) and the odds of competitive breakdown by thirty (eb = 30.9) (see Table 10.8).

Post-Electoral Contention In addition to articulating electoral grievances in the temporal vicinity of elections, citizens may participate in contentious actions whose demands bear no direct relation to electoral issues. Even if not explicitly concerned with regime struggles, such acts of extra-electoral and inter-electoral contention may contribute to the vertical uncertainties that I hold to be drivers of regime change. To examine their relationship with electoral authoritarian trajectories, I use, as in previous chapters, two additive indicators of conflict event counts from the Arthur Banks cnts dataset. Under the heading of “vertical threats” I add all anti-government demonstrations, general strikes, and riots that took place in the three years following each election year, and under the heading of “armed rebellion” all instances of guerrilla warfare and revolution. As the logistic regression results displayed in Table 10.8 indicate, the number of conflict events in the three years posterior to an election shows a remarkable explanatory capacity, for competitive regime trajectories (R2 = .102) as well as for hegemonic trajectories (R2 = .157). Competitive regimes are particularly vulnerable to organized violence. Each instance of armed rebellion augments the odds of democratization by almost one-quarter (eb = 1.232) and the odds of breakdown by almost one-third (eb = 1.299). Hegemonic regimes respond differently to societal conflict. Vertical conflicts threaten the status quo, armed rebellions reinforce it. While each instance of “vertical threat” augments the odds of hegemonic opening by almost two-thirds (eb = 1.637), even a single armed rebellion pushes these odds almost to zero (eb = .069). The weight of vertical threats derives primarily from peaceful demonstrations. If we feed logistic regressions (not reported here) with disaggregate event data, each additional anti-government demonstration almost triples the odds of hegemonic opening (eb = 2.98).

The Struggle over Regime Change

365

AU T H O R I TA R I A N C A PAC IT I ES Whether favorable societal or institutional structures translate into manifest challenges against a regime depends not least on the ability and determination of opposition actors. Whether opposition protests translate into genuine threats to regime continuity depends not least on the resources and responses of the regime itself. The magnitude of our problems never depends on the magnitude of our problems alone, but also on our capacity to cope with them. Here I will explore the isolated average impact that four distinct aspects of authoritarian capacities carry on electoral authoritarian trajectories: state infrastructural power, repressive capacities, elite fragmentation, and actual efforts of electoral manipulation. For the first three aspects, I use hair thin indicators: for state infrastructural power, the percentage share of tax revenue in Gross Domestic Product (gdp); for repressive capacities, the percentage share of military personnel in total labor force; and for elite division (“horizontal threats”), the sum of government crises and purges during the three years following the election year.13 To prevent case numbers from falling flat, I run uncontrolled bivariate logistic regressions for each of the three variables. I measure actual manipulation with my four variables from the Dataset on Authoritarian Elections: repression (violations of physical integrity), censorship (violations of media freedom), the exclusion of parties or candidates from electoral competition, and the commission of electoral fraud (administrative interferences with the organization of elections). I run multivariate logistic regressions to estimate their joint impact. All results are summed up in Table 10.9. Repressive capacity. In the last instance, the outcome of regime struggles depends on the governmental control of state violence. As long as authoritarian governments are willing and able to mobilize state agencies of repression against existential challenges, they are likely to prevail (except for the intervention of external powers). Their capacity and willingness to repress are impossible to estimate in cross-national datasets. Practically impossible, but essentially impossible, as well, for the willingness to repress is not a fixed parameter. Rather it is the contingent result of collective choice within security forces under evolving political conditions. All we can try is to capture its structural conditions through crude proxies like the one I employ here: the size of the military (in terms of labor force employment). As the regression results in Table 10.9 suggest, the weight of the military as employer is irrelevant for post-electoral trajectories

366

Empirical Explorations

Table 10.9. Authoritarian capacities and regime trajectories cr Democratization eb Repressive capacity Constant Chi2 / pac Cox & Snell R2 N State capacity Constant Chi2 / pac Cox & Snell R2 N Horizontal threats Constant Chi2 / pac Cox & Snell R2 N Regime manipulation Repression Censorship Exclusion Fraud Constant Chi2 / pac Cox & Snell R2 N

.774

cr Interruption

p

eb

p

eb

p

.426 .051

.770

.663 .009

.209 .589 84.8% .079 33

.196 .549

.342 .016

.963

.681 .113

.982 .185 87.9% .001 33

.828 .241

.370 .000

1.342

.423 .000

4.308 .159 85.2% .073 54

.109 .000

.452 .003 .087 .085 .016

.950 .216 .824 .468

.831 .211 .841 .267 .833

1.410 .286 .612 1.404

.155 .224 .573 .608 .243

.808 .014 59 1.032 1.178 .021 55 1.219 1.185 .013 93 .896 .099 .329 .481 20.087 .194 93

hr Opening

85.2% .104 54

Note: Multinomial logistic regressions for competitive regimes (cr), binary logistic regressions for hegemonic regimes (hr). Reference category: regime continuity. Repressive capacity = military personnel (% of total labor force). State capacity = tax revenue (% gdp). Horizontal threats = sum of government crises and purges during three years after election year (Source: cnts). pac = percentage of accurate classification. eb = odds ratios. Shaded coefficients = statistically significant (p ≤ .15).

of competitive regimes, yet depresses the odds of opening in hegemonic regimes (although somewhat below conventional levels of statistical significance). Conservative elections in hegemonic regimes are overseen by military forces twice as numerous as those that accompany transformative elections. The former employ on average 1.1 percent of the labor force (N = 28), the latter 0.56 percent (N = 5).14 Infrastructural capacity. Along with its coercive capacities, the infrastructural power of a state determines the institutional resources it controls to

The Struggle over Regime Change

367

respond to emerging threats. The relative magnitude of tax revenues captures one core dimension: its extractive capacity, the degree to which it enforces its monopoly of taxation. The richer the state, the more resilient its regime, one might think. Yet, for the few cases for which I have data, the explanatory capacity of tax revenue is nil (see Table 10.9). If the capacity of taxation is important for electoral authoritarian regime trajectories, it does not show in the short-term relationships that I am examining. Horizontal threats. Modern authoritarian regimes that command solid infrastructures of power are almost unassailable. The only way to topple them is to upset their internal capacity of coordination. Thus the centrality of elite cohesion for the survival of authoritarian regimes. The events that my additive indicator of “horizontal threats” registers (government crises and purges) are very rare in hegemonic regimes, but if they occur, they are highly destabilizing. Each instance of a purge or government crisis multiplies the odds of hegemonic opening by four (eb = 4.308). By contrast, these thin manifestations of horizontal tensions do not seem to affect the short-term fate of competitive regimes (see Table 10.9). Regime manipulation. For the wielders of power, it is best to possess impressive infrastructures of power without ever having to use them. When they fail to obtain what they desire through the mere display of might, when dissuasion is not enough and they have to make good on their threats of sanction, they look weak. Punitive enforcement is a veiled concession of defeat and it always carries the risk of creating more troubles than it resolves. When electoral authoritarian rulers respond to emergent threats by tightening their manipulative grip on elections, their success in containing these threats is by no means assured. Nevertheless, we should expect electoral authoritarian rulers to benefit from electoral manipulation, at least on average and in the short term (see Chapter 8). However, if we expect intensified manipulation to increase the chances of regime political survival in relative terms (considering the magnitude of challenges that they confront), should we also expect it to increase them in absolute terms (irrespective of the magnitude of challenges)? Overall, the regression results shown in Table 10.9 indicate that they do. In competitive regimes, manipulative maneuvers consistently work as expected: depressing the odds of regime change. Repression seems to be least effective, but all other strategies appear as powerful weapons to defend the competitive authoritarian status quo. Exclusion lowers the odds of democratization by two-thirds (eb = .33). Each step along the trichotomous scale of fraud depresses them by 50 percent (eb = .481) and each step along the trichotomous scale of censorship even by 90 percent (eb = .099). Fraud and cen-

368

Empirical Explorations

Table 10.10. Structural and strategic conditions of regime change Competitive regimes

eb

Democratization Margin of victory Electoral boycott Post-electoral protest Regime manipulation Wealth Growth Oil Constant

.974 1.996 2.145 .245 6.276 .934 .962

.027 .238 .411 .021 .156 .198 .140 .393

.895 22.603 158.26 .812 .001 .737 .924

.068 .127 .295 .889 .233 .183 .219 .338

Interruption Margin of victory Electoral boycott Post-electoral protest Regime manipulation Wealth Growth Oil Constant Chi2 Cox & Snell R2 N

39.421 .440 68

p

Hegemonic regimes Opening Electoral protest Electoral boycott Repression Censorship Vertical threats Armed rebellion Horizontal threats Constant

pac Cox & Snell R2 N

eb

8.187 5.357 2.002 .018 2.046 .008 5496.2 .015

p

.048 .070 .135 .087 .068 .097 .045 .068

92.6% .365 54

Note: Multinomial logistic regressions for competitive regimes (cr), binary logistic regressions for hegemonic regimes (hr). Reference category: regime continuity. Electoral data refer to national elections (presidential, legislative, or concurrent elections). Regime manipulation = cumulative index of manipulation (0–4). Wealth = log current ppp per capita. Growth = gdp per capita change in election year. Oil = mineral fuel exports (% of total exports). Vertical threats, armed rebellion, and horizontal threats during the three years following the election year. Shaded coefficients = statistically significant (p ≤ .15).

sorship also appear as rather solid safeguards against the breakdown of competitive authoritarianism (although below conventional levels of statistical significance). Overall, among the bundles of explanatory factors reviewed here, authoritarian manipulation appears as the most powerful, as it accounts for almost one-fifth of observed variance in competitive regime trajectories (R2 = .194). In hegemonic regimes, the role of electoral manipulation looks different. Its explanatory capacity is lower (R2 = .104) and its association with short-term regime trajectories less consistent. While censorship appears as an effective safeguard of hegemonic continuity, repression goes hand in hand with hegemonic opening (although both coefficients lie somewhat below conventional levels of statistic significance). The elephant of endogeneity, it appears, has

The Struggle over Regime Change

369

made another quiet entry: Hegemonic regimes turn repressive when they are ripe for change . . . and unable to prevent it. Do these associations between manipulation and regime trajectories hold even when we control for contextual factors and underlying threats? To check their robustness, I run multivariate logistic regressions that include measures of manipulation along with a series of controls. From the battery of variables that I have explored in this chapter, I select indicators that displayed significant associations with regime trajectories and at the same time have low numbers of missing cases. To prevent the number of variables from escalating, I use a composite measure of manipulation for competitive regimes (the sum of all four strategies, each normalized to a scale of 0–1) and an aggregate measure of electoral protest for hegemonic regimes (the sum of pre-electoral and post-electoral protest). Given the classical disproportion between “many variables” and “small N” (Lijphart 1971: 686), the results shown in Table 10.10 must be taken with caution. Still, they fundamentally confirm, not only the weight of manipulation in determining electoral authoritarian regime trajectories, but also the relevance of the behavioral and strategic variables that we had examined before in relative isolation.

C O N C LU S I O N What do we know about regime change by elections after skating, with more speed than artistry, the thin ice of my quantitative data? Which is the preliminary balance between structures and agency, contexts and elections, general and specific factors in the explanation of electoral authoritarian regime change? How much do the politics of electoral uncertainty explain? How much structural imbalances of power? Unsurprisingly, my answer is conciliatory: the conflictive interactions between government and opposition in the electoral arena carry considerable explanatory weight, but so do some contextual factors that define the societal playing field in which these electoral encounters take place. To begin with: the politics of electoral uncertainty matter. Institutional struggles over the quality of elections are not acts of shadowboxing by delusional actors who think they make a difference, but in the end don’t. The meta-game strategies governments and opposition parties pursue do make a difference for the fate of electoral authoritarian regimes. Even in the short run. Their impact is discernible even if we know nothing about electoral campaigns and voter preferences at the game level. And it is discernible in both subtypes of regimes,

370

Empirical Explorations

even if they differ in their causal sensitivities. Competitive regimes are more sensitive to variations in electoral competitiveness. Democratic change, when it takes place, tends to take place when margins of victory are relatively low. In hegemonic regimes, this applies only to the legislative arena. Presidential margins of victory are unrelated to subsequent regime trajectories. At the same time, defending the status quo through electoral manipulation seems to be less effective in hegemonic regimes. Manipulating the public perception of political realities through media censorship seems to be the best authoritarian card that hegemonic parties hold to cope with emergent troubles. In competitive regimes, censorship works even better in minimizing the odds of regime change, while exclusion and fraud appear as effective instruments of authoritarian stabilization as well. Confronting the solidity of hegemonic regimes, opposition actors face the challenge of introducing vertical uncertainties. Confronting the relative fluidity of competitive regimes, they face the challenge of deepening vertical uncertainties. Given their reputation of invulnerability, mobilizing protest is more difficult in hegemonic regimes. Yet, given their vulnerability to reputational losses, mobilizing protest is potentially more effective under hegemonic conditions. Echoing our findings from Chapter 7 on the sources of electoral competitiveness, hegemonic regime trajectories are remarkably responsive to oppositional challenges, inside as well as outside the electoral arena. The exception is organized violence. While non-violent challenges corrode hegemonic regimes, violent challenges reinforce them. As their equilibrium rests upon the untested assumption that they are popular, sowing, through popular protest, doubt about their popular support seems to be a more effective strategy of institutional subversion than armed rebellion. The hegemonic emperor is less vulnerable to material challenges that force him to draw his sword than to symbolic challenges that show he is fighting naked. The vulnerability of hegemonic regimes to the erosion of equilibrium-sustaining appearances may also lie beneath the strong nexus we saw between horizontal challenges and subsequent processes of regime opening. As soon as cracks open in the facade of elite unity, the whole edifice of hegemonic invincibility starts to crumble. However. Even if we assign primacy to the interplay of actors in the electoral arena in our explanations of electoral authoritarian regime change, my empirical findings also lend firm support to the idea that structural and institutional contexts matter as well to electoral authoritarian regime trajectories. Both subtypes of electoral autocracies are hurt by economic downturns, competitive regimes in the short term, hegemonic regimes in the medium term, and both benefit from establishing majoritarian electoral systems and from keeping legislatures weak. Wealth is a source of instability in competitive regimes, yet reinforces the tranquillity of their hegemonic counterparts. Oil eases

The Struggle over Regime Change

371

and greases the reproduction of both. The simple finding that oil-exporting regimes in my sample have experienced almost no instances of regime change, while poor regimes have experienced almost all instances of electoral authoritarian regime breakdown, is a powerful reminder that actors in the electoral arena are not sovereign, but embedded in structural environments that create opportunities and constraints, generate resources and scarcities, empower and impoverish.

Notes 1. For a discussion of continuous versus categorical conceptions of political stakes, see Schedler (2010b: 6). 2. All data from World Bank World Development Indicators. For descriptive statistics, see Appendix d. 3. In competitive regimes, oil exports and per capita income are moderately correlated (r = .211, p = .08, N = 69), in hegemonic regimes they are unrelated (r = .027, p = .86, N = 43). 4. These differences of means are statistically significant (p ≤ .10) (anova for competitive trajectories, bilateral t-tests for hegemonic trajectories). 5. Since my case numbers are rather low and missing data rather high, I run separate bivariate logistic regressions for both independent variables. 6. On formal authoritarian institutions of domination and representation, see Chapter 2. 7. On the distinction between “slow” and “sudden” deaths of regimes, see O’Donnell (1992). 8. By contrast, within my sample of authoritarian elections, presidential contests do not set regimes on different trajectories than parliamentary elections. 9. See, for instance, Bratton and van de Walle (1997), Bunce and Wolchik (2011). 10. See also Chapter 7, p. 244 and Chapter 9, pp. 311–314. 11. Intriguingly, the two measures are unrelated, except in legislative elections in competitive regimes (r = .26, p = .03, N = 69). 12. See Chapter 7 as well as Hartlyn and McCoy (2006), Lindberg (2006a), and Schedler (2009a). 13. Sources are World Bank World Governance Indicators for tax revenues and military personnel, cnts event counts for horizontal threats. For the present purpose, I exclude military coups from my index of “horizontal threats” since they overlap with the outcome variable “interruption.” 14. Despite low case numbers, this difference is statistically significant (bilateral t-test).

Conclusion What happens when authoritarian rulers convoke multiparty elections? In the theoretical part of this book, I sought to comprehend the dynamics of authoritarian elections at a high level of abstraction. Under authoritarian conditions, I claimed, regime struggles unfold as twin competition between regime and opposition over the uncertainty of knowledge (opacity) and the uncertainty of the future (insecurity). Under electoral authoritarian conditions, these competitive struggles unfold as two-level conflicts over the hearts and minds of voters (at the game level of electoral competition) and over the nuts and bolts of the political system (at the meta-game level of institutional reform). In the empirical part of the book, I explored the politics of uncertainty that unfolds in authoritarian electoral arenas through the analysis of short-term associations between two sets of meta-game strategies (regime manipulation and opposition protest) and two sets of game outcomes (electoral competitiveness and regime change). In these concluding pages I will avoid repeating what I have said before. Instead, I will draw some general theoretical conclusions, delineate possible future trends, suggest some practical implications, and outline some pending tasks for future research.

THE THEORY OF ELECTORAL AUTHORITARIANISM Social sciences are torn between the embrace of complexity and the quest for simplicity. In case of doubt, I have chosen to err on the side of doubt, that is, on the side of complexity. In each of the preceding empirical chapters, my theory is more complex than my data and my findings are less determinate than my hypotheses. In the end, given the mixture of positive and negative findings that I present, it is difficult to see the empirical wood from the statistical trees. Besides, the book has been more an enterprise of theory development than theory testing. Its empirical findings have been explorative, at times inductive, always tentative. Nevertheless, leaving the details and caveats to the preceding chapters, I wish to highlight two fundamental theoretical claims that have sustained my empirical explorations and have been sustained by them: the creativity and the relative autonomy of electoral politics.

Conclusion

373

Political Creativeness Classically political actors conceive their trade as “the art of the possible.”1 We students of politics tend to analyze it as the science of the probable. We seek to comprehend actors who pursue their goals within certain sets of material constraints. In our explanations of political behavior, the identity of actors does not matter. We assume them to possess similar skills and motives. What matters is their structural environment. Contexts determine outcomes. Fascinated by the reactive intelligence of actors, we tend to neglect their creative capacities. Like David Copperfield who made the Statue of Liberty disappear, we make agency disappear in our empirical theories of political action. By definition, actors are sources of indeterminacy (also known as freedom). They can always act otherwise, or else they are not actors, but puppets or billiard balls. We need to understand how political actors adapt to given constraints, but also how they transcend and transform given constraints. Otherwise we ignore their defining essence: their capacity for action.2 In what I have described as the politics of uncertainty, contending actors invest their creative energies in constructing, maintaining, and transforming authoritarian institutions. Political institutions are cognitive realities (structures of expectation) that are grounded in material realities (structures of power), and vice versa. To build, reproduce, or destroy them, actors have to build, reproduce, or destroy both: patterns of expectations and relations of power. By attributing critical roles to political actors in the construction of the institutional constraints that they confront, we redefine common ideas of institutional equilibria as well as common conceptions of political agency under authoritarian rule. First, states of equilibrium do not run on autopilot. Actors do not drop out of equilibrium games, either in happy relaxation at the top of the regime or in unhappy resignation at its bottom. Dissidents keep challenging the status quo, governments keep calibrating their arsenal of manipulation. As any tightrope walker can confirm, preserving equilibria requires constant active attention. The construction of institutional certainties is the work of actors, and so is their maintenance and eventual destruction. Second, actors do not run on autopilot either. Our reactive conceptions of dictators and dissidents tend to understate the proactive roles of both. In the comparative study of authoritarianism, we tend to focus our analytic attention on calculating rulers. We conceive them as rational investors who manage their political fortunes in competent and effective manners. They possess the requisite information and foresight to identify and anticipate threats to their power and to contain them through appropriate remedial or preventive measures. We overlook the fragility of the equilibria over which they preside,

374

Conclusion

the constant efforts they have to invest in their maintenance, for instance, by adaptively deploying multiple strategies of electoral manipulation. We also tend to overlook the proactive role that opposition actors take in disrupting established equilibria. Opposition to authoritarian rule is irrational, microeconomically speaking. It is risky and does not pay, not in the short term and perhaps not even in the long run when we are all dead. The odds against opposition success should be paralyzing. And yet. Opposition actors do exist and protest and participate in authoritarian elections. Even under depressing conditions of self-evident hopelessness as in hegemonic regimes. And they are remarkably powerful. They are rarely granted the satisfaction of toppling a long-reigning dictator after less than a month of street protests, as in Tunisia in January 2011. The empirical analyses of this book, however, offered encouraging evidence for “the power of the powerless” (Havel 1985) in electoral authoritarian regimes. Electoral protests breed regime uncertainties in all types of electoral autocracies. They erode the electoral dominance of hegemonic parties and provoke them into defensive responses. In competitive regimes, contentious actions between elections raise electoral uncertainties and inhibit authoritarian manipulation.

Embedded Autonomy Studying formally representative institutions under authoritarian rule, this book forms part of the “new institutionalism” in the comparative study of authoritarianism. The idea that such institutions matter is built into the very conception of its object of research. The notion of electoral authoritarianism rests upon the assumption that elections make a difference. They structure the operative logic of authoritarian rule. They define the distinctive nature of electoral autocracies. Yet what does it mean that institutions matter? Does it mean that they carry invariant effects? That their mere existence produces determinate outcomes? Not at all. In the introduction, I described my approach as political–institutionalist. Political institutionalism takes institutions seriously and politics, too. It comprehends institutions as arenas of politics that structure political conflict, yet do not replace it or predetermine its outcomes. It therefore comprehends multiparty elections, not as causal factors, but arenas of (asymmetric) competition whose outcomes are contingent. They depend on the conflictive interactions that unfold within their bounds. From this perspective, the idea that authoritarian elections matter involves three things. It means that the conflictive interactions that they generate possess a logic of their own, an importance of their own, and effects of their own. They are autonomous,

Conclusion

375

rather than epiphenomenal. They are not mere reflections of external circumstances. They are intrinsically important, for elites as well as citizens, rather than essentially trivial. We should care about their dramatic unfolding. And they are consequential, rather than weightless. The dynamics of the electoral game affect regime outcomes. In its empirical explorations, this book examined the internal dynamics of electoral authoritarian regimes. Holding authoritarian elections to be intrinsically relevant, it provided manifold supportive evidence for both their causal weight and their autonomous nature. On the one hand, the internal dynamics of authoritarian elections are consequential for correlations of power as well as for regime trajectories. On the other, these internal dynamics possess substantive autonomy from contextual factors. Put simply, what regime actors do in the electoral arena matters for what opposition actors do and the other way round; and what either of the two does matters for the outcomes of the game, for its competitiveness as well as for its very continuity. As a matter of course, though, the autonomy of institutions is always relative, and so is the autonomy of authoritarian elections. Elections do not happen in outer space, among floating astronauts, but are embedded in societal contexts. Thus the continual emphasis this book lay on the embeddedness of elections. The conflictive interactions that unfold in the electoral arena follow a logic of their own and possess a weight of their own. Yet, the protagonists of electoral competition are neither fully autonomous, nor fully sovereign. They are embedded in institutional and structural surroundings that define asymmetric constraints and opportunities. Features of the institutional environment (such as the design of electoral rules and the strength of legislative assemblies) matter for strategic choices as well as for systemic outcomes, and so do societal structures (such as wealth, social inequality, and the availability of rents).

THE FUTURE OF ELECTORAL AUTHORITARIANISM Even while recognizing the limits of prediction in the social sciences, it is tempting to ponder the future of electoral authoritarianism. After the end of the Cold War, electoral autocracies experienced impressive growth. Within a few years, they turned into the most common form of dictatorship around the globe. Does their global expansion represent a lasting trend or no more than a passing fad? With the exception of Singapore, all hegemonic party regimes that predated the fall of the Berlin Wall have disappeared. They have given way either to democracy or to competitive authoritarianism. Of the thirty-five

376

Conclusion

competitive autocracies that Steven Levitsky and Lucan Way identified in the early 1990s, fifteen had democratized by 2008. Among the remaining twenty, no more than ten appeared as stable autocracies, the rest as unstable (2010b: 341). One-fifth of all competitive authoritarian elections in my own sample led to subsequent democratizing change. Overall, over the past decades, competitive authoritarian regimes have emerged as “the typical stepping stone to democratization” (Hadenius and Teorell 2007: 152). Should we conclude that “the era of electoral authoritarianism” (Morse 2012) is bound to end soon, giving way to renewed advances of electoral democracy? Well, not quite, not yet. The reasons are simple: A fair number of electoral autocracies defy the laws of political mortality and continue to hang on. More importantly, a continuous stream of new cases continues to repopulate both subtypes of electoral authoritarianism. Both hegemonic and competitive autocracies continue to gain new members from their neighboring regime categories. The Regeneration of Hegemonic Regimes Hegemonic demise. Hegemonic regimes are long-lived by definition and inherently stable. They are not immortal, though. Except for Singapore, Gabon, and Tanzania, none of the hegemonic regimes in my dataset has survived beyond the year 2012. That does not mean they have all democratized. Actually only a handful did. The two stellar cases are Mexico and Indonesia. Mexico completed its transition to democracy after a stopover in the category of competitive authoritarianism. After President Suharto resigned in 1988 in the wake of economic crisis and popular unrest, Indonesia took a direct leap from hegemonic party rule to electoral democracy. Other hegemonic regimes have evolved into competitive regimes. Malaysia did so in 2008 when the governing umno lost its legislative supermajority. Although the Mexican path of gradual opening may seem like a natural tendency for hegemonic parties, hegemonic regimes have shown themselves vulnerable to leadership changes and extra-institutional threats. The death of its long-serving president plunged Guinea into political instability. In Togo, it made the military intervene in order to ensure a dynastic leadership succession from father to son with electoral authoritarian continuity. In Côte d’Ivoire and Mauritania, hegemonic party rule was terminated by military coups. In Indonesia, Tunisia, and Egypt, street pressure forced aging presidents to resign, thus paving the way for the fall of the regime. These disorderly patterns of change suggest that hegemonic regimes possess two areas of structural vulnerability: x Over-dependence on personal leaders: Just like other types of dictatorship, hegemonic parties often rest upon contradictory combinations of strong

Conclusion

377

institutionalization and high personalization. To the extent that they make themselves dependent on personal rulers, they make themselves vulnerable to success crises. x Hegemonic overstretching: The trick of electoral hegemony is to dominate the opposition completely without demoralizing it completely. When hegemonic parties fail to maintain the minimum balance between electoral restraints and opportunities that keeps the opposition in the electoral game, they displace dissidence from the electoral arena towards extra-institutional spaces of contention. In cases of authoritarian overstretching, hegemonic success destroys its own conditions of sustainability. The fact that almost all hegemonic regimes that had existed in the late 1980s and early 1990s have withered away over the subsequent two decades, does not mean that electoral hegemonies are a matter of the past. Both competitive regimes and closed autocracies may transform themselves into hegemonic regimes. They have done so in the past and are likely to do so in the future. Transitions from competitive authoritarianism: Competitive authoritarian rulers often strive to transform their precarious incumbency advantages into solid hegemonic domination. Some have failed, at least for now. Russia’s Vladimir Putin is a prominent example. Others, however, have succeeded. In the post-soviet space, Nursultan Nazarbayev of Kazakhstan, Alyaksandr Lukashenka of Belarus, and Ilham Aliyev of Azerbaijan have conducted successful transitions from competitive authoritarianism to personalist breeds of hegemonic rule. Transitions from closed authoritarianism: When closed autocracies introduce multiparty elections they can hope to establish instantaneous hegemonic domination. They possess huge initial advantages over their competitors. They have the organizational infrastructure, the administrative capacity, the appearance of popular support, and the military power that they need to face the emergent winds of electoral competition without so much as disheveling their hairdo. No doubt, the most important case of a possible future transition from closed to hegemonic authoritarianism is China. As with everybody else, the Chinese single-party empire enjoys an open future. It may continue to thrive for quite a while, stagnate and muddle through, transform itself into a repressive military regime, or disintegrate in civil war. Yet, if it undergoes a process of political opening, which is a real possibility over the next two decades, it is unlikely to democratize. It is much more likely to transform single-party into hegemonic-party rule. When social pressures keep mounting and the system of crony capitalism starts creaking, when political reforms become, not a speculative luxury, but an imperative of survival, Chinese leaders are unlikely to take a “great leap forward” to democracy.

378

Conclusion

More likely, they will continue to experiment with limited elections, with the gradual introduction of electoral competition under the control of the ccp. Hegemonic electoral authoritarianism would not grant them eternal life. But it could very well afford them some additional decades of bureaucratic omnipotence, unsustainable economic growth, and personal self-enrichment. Hegemonic persistence. The exciting panorama of transitions from and to hegemonic party rule should not make us forget existing instances of continuing hegemonic persistence. The most puzzling case is Singapore under the pap. Singapore is a scary creature. On its glittering surface, it projects the frightening image of a consumerist dictatorship where everybody is happy and quiescent, content with inhabiting a clean and orderly desert of concrete, postmodern design, and the latest generation of smart phones. The public image of consensual authoritarianism may be deceptive, though. Southeast Asia’s “Disneyland with the death penalty” (Gibson 1993) is often hailed for its administrative and judicial integrity, its impeccable rule by law. The regime is implacable in imposing its full weight on its frail and feeble opposition. It is one of those hyper-allergic dictatorships that respond with regulatory hysteria to the smallest manifestations of dissidence. And yet. The regime, a Westminster-style parliamentary autocracy with majoritarian election rules, has enjoyed an unbroken near-monopoly of legislative seats since its foundation. However, in all general elections since 1984 (except for 2001), it has received less than two-thirds of the legislative votes. Despite the self-evident futility of the act, more than one-third of Singaporeans regularly waste their votes on opposition candidates. At some point, which we cannot predict but anticipate, the brave new world of the clinically clean dictatorship without corruption, protesters, or chewing gum on the streets, may exhale its artificial life without prior warning.

The Regeneration of Competitive Authoritarianism Unlike hegemonic regimes, competitive authoritarian regimes are not in equilibrium. They are battlegrounds, congealing systems of domination, not yet solidified. Their battles are asymmetric, between contenders of unequal standing, yet not predetermined in their outcomes. They represent “the most volatile regime type” (Roessler and Howard 2009: 103) of all. Many have democratized, such as Peru, Serbia, and Ghana. Others have been cut short by military coups and political disorder, like Côte d’Ivoire and Togo. Nevertheless, the species of competitive authoritarianism is not in danger of extinction. A handful of regimes, like Russia, Algeria, Ethiopia, and Zimbabwe, are still surviving, hanging on, muddling through. Some, like Kenya and Ukraine,

Conclusion

379

have installed themselves along the contentious borderline of “ambiguous regimes” (Diamond 2002) which experts disagree as to how to classify. Others, like Chad and Cambodia, have been drifting towards the opposite end, moving closer and closer to the category of closed regimes. Still others, like Côte d’Ivoire, Madagascar, and Kyrgyzstan, have been oscillating in and out of states of disorder, military intervention, and competitive authoritarianism. While struggling to keep its quarrelsome pack together, the family of competitive autocracies has been admitting new members from closed autocracies, like Afghanistan and (as it appears at the time of writing) Myanmar. But more importantly, it has been admitting new members from electoral democracies. The transformation of electoral democracies into competitive autocracies has been most notable in Latin America. At the beginning of the 21st century, Latin America was a region of democracy. Except for Cuba’s bureaucratic authoritarianism and Haiti’s descent into political disorder, all countries of the region looked minimally democratic. Military rule seemed a matter of the past. Central America had made successful civil war transitions. The long-standing hegemonic party systems of Mexico and Paraguay had given way to electoral democracy. The Fujimori regime had collapsed under the weight of its own corruption. One decade later, the regional picture looks very different. New challenges to democracy have emerged from organized societal violence. And old challenges to democracy from overpowering executives have reappeared with renewed vigor. In Venezuela, Bolivia, and Ecuador, anti-political-establishment actors (Schedler 1996) have taken power through democratic means, concentrated power through dubious means, and then subverted the competition for power through authoritarian means (see Levitsky and Loxton 2013). Today, all three arguably belong to the category of competitive electoral autocracies. The same applies to Nicaragua after the return of Daniel Ortega and Honduras after the 2009 military coup (Sonnleitner 2008). Argentina’s “delegative democracy” (O’Donnell 1994) is creeping towards the borderline. In the early 1990s, in the midst of democracy optimism, Guillermo O’Donnell had already warned us against the ever-present dangers of democratic erosion. Authoritarian regressions, he alerted, may occur through the “sudden death” of democracy or its “slow death” (1992: 19). After the age of military coups, the spirit of the times favors the latter. Contemporary electoral democracies have proved remarkably resilient (Diamond 2008). Yet when they fall, they tend to stumble. Rather than aborted by military coups, they are undermined by executive encroachment and electoral manipulation. Even within their propitious external environment of high Western linkage (Levitsky and Way 2010b), Latin American democracies remain vulnerable to subversion by electoral authoritarian wolves in the clothing of radical–democratic sheep.

380

Conclusion THE FUTURE OF DEMO CRATIZATION ASSISTANCE

Laypersons tend to think of science as a fountain of secure practical knowledge. The comparative study of politics, just like the social sciences in general, cannot meet that expectation. If our causal findings come to be relevant for policy-makers, it is less through the provision of instrumental certainties than through the introduction of doubt. Comparative politics, rather than providing secure foundations for political decision-making, reveals its uncertain grounds. It works better in unsettling common assumptions than in establishing common sense. Its appropriate role is not to close political debates, but to get them rolling, not to make public decisions easy, but to make them hard. And yet. Even if we are skeptical about the possibilities of translating cross-national empirical findings into policy advice, do my empirical explorations license practical recommendations for democratization assistance in electoral authoritarian regimes? What should international actors do in the face of authoritarian multiparty elections?3

Embracing Elections In democracies, elections often have a bad press. They are only thin and occasional exercises of citizen participation. When political parties and candidates crowd the ideological center and the world economy narrows their policy options, critical observers frequently denounce them as meaningless rituals. Minimal procedural democracy, we often hear, is not enough. It is a fraud, an institutionalized deception. Electoral authoritarian regimes allow us to value anew the historical achievement that democratic elections represent. They allow us to appreciate how demanding even minimal democracy is. The dynamic of authoritarian elections also allows us to apprehend the enduring normative and political force that the ideal of democratic elections continues to command at the beginning of the 21st century. A first piece of practical advice to international democratic actors thus is simple: embrace multiparty elections. Bad elections are better than no elections. Electoral autocracies are preferable to closed autocracies. If nothing else, the regular holding of authoritarian multiparty elections carries one big advantage: it opens up the possibility of democratization by elections. It is an abstract possibility, not a concrete one. Rather than practical, the nexus is logical. No elections, no electoral transitions. Whether electoral transitions actually do take place is not a matter of logic, but of power. And conflicts over power. As this book has shown, elections are arenas of conflict that possess

Conclusion

381

significant degrees of autonomy and causal weight. Establishing them does not secure outcomes. It merely allows that opposition actors can fight for democratizing outcomes through electoral reform and electoral mobilization. This may not seem like much, but probably enough to welcome the introduction of multiparty elections by calculating dictators. They tend to believe they can run the show without risking their jobs as cheerful ringmasters. Let them have their belief.

Strengthening Dissidence The wide-ranging activities that run under the title of international “democracy promotion” essentially aim at strengthening actors and institutions that are deemed supportive of liberal democracy: civil society, political parties, elections, courts, and legislatures. With respect to authoritarian elections, this approach makes sense. It makes sense to strengthen the capacity of opposition parties to run effective electoral campaigns (at the game level of authoritarian elections). And it makes sense to strengthen their simultaneous capacity to stage effective protest campaigns (at the meta-game level of authoritarian elections). While my empirical explorations provided confirming evidence for the power of the latter, they remained silent on the former. Opposition parties sometimes neglect the necessity of persuading voters. Some do so because they take popular support for granted. They imagine themselves as carriers of “natural majorities” who will be swept to power as soon as voters are allowed to express themselves freely, without the constraints of electoral manipulation. The Mexican left has been losing one presidential election after another since 1994 on the basis of such self-deceptive beliefs. Other opposition actors ignore the imperatives of electoral campaigning for opposite reasons, as they take the popularity of the regime for granted. In Venezuela under Chávez, for instance, it took the opposition about a decade to take electoral competition seriously. Apparently convinced that it was facing a “popular autocrat” (Dimitrov 2009) it could not beat at the polls, it resorted to a broad range of extra-electoral forms of contentious action, including mass protest, a general strike, a recall referendum, a national election boycott, and even a coup attempt. Its radicalism radicalized the regime and pushed it further down the electoral authoritarian path. Only in recent years, the opposition has started to conduct serious election campaigns. Only since then it has begun to erode the electoral dominance of the country’s thundering petro-populism (see Corrales 2011). Taking electoral competition seriously is a sine qua non for any opposition who aspires to win an election, be it democratic or authoritarian. International

382

Conclusion

actors can help opposition parties to understand that neither their victory nor their defeat are inevitable. They can also help them to attract the creative talent, acquire the knowhow, and build the organizational infrastructure they need for successful electioneering. Under authoritarian conditions, though, attracting voters is not enough for winning an election. Transforming authoritarian institutions is needed as well. Actors have to play the game of electoral competition with passion and skill, but the meta-game of institutional struggle, too. At this level, international actors can help by putting reluctant rulers under pressure.

Reforming Institutions Monitoring elections is one major task of international election monitoring. The other one is institutional reform. Election observers wish to observe. Through their surveillance, they wish to establish the democratic quality of elections. But they also wish to transform. Through their expertise, they wish to improve the democratic quality of elections. Their diagnostic task is complex and their reformative task no less so. Electoral governance is a wide and multifaceted field, and so are authoritarian manipulation and democratizing reform. The lists of complaints and reform recommendations that professional election observers submit to electoral authoritarian authorities tend to be broad in scope and deep in legal and technical detail. My empirical explorations touched upon no more than a few dimensions of electoral governance and authoritarian manipulation. Nevertheless, they allow us to identify at least two critical areas of electoral reform that international election observers should place high on their agenda: the reform of exclusionary practices and institutions and the strengthening of formally representative institutions. Weakening exclusionary institutions. To keep electoral competitiveness low, many electoral authoritarian regimes exclude uncomfortable competitors from the electoral arena. Hegemonic regimes in particular seem to benefit from the institutionalized exclusion of their historic adversaries. Electoral exclusion is easy to spot, yet often goes unnoticed. It is also easy to remedy, yet is often met with resignation and remains in place over the long haul. While democrats need to counter authoritarian manipulation in all its forms, they also need to set priorities. Electoral market restrictions seem like a good place to start. Under authoritarian conditions, majoritarian electoral rules are another form of exclusion. While restrictions on party and candidate registration exclude inconvenient contenders from electoral competition, majoritarian electoral rules exclude them from parliamentary representation. In competitive

Conclusion

383

and hegemonic regimes alike, majoritarian systems dampen electoral competitiveness and inhibit democratizing change. Understandably, they invite opposition protest and electoral boycotts. International actors should hook up with such protests to step up pressures for counter-majoritarian reforms that mitigate the winner-takes-all features of authoritarian electoral systems. They may sweeten their pressures by appealing to the long-term rationality of rulers: They benefit from majoritarian systems only as long as they are able to manufacture secure majorities. They may lose this ability at some point. Strengthening subversive institutions. Liberal democracies are “bounded wholes” (Sartori 1970) formed by a complex set of representative institutions which support each other in a consistent manner (often by creating tensions and inconsistencies among each other). Electoral authoritarian regimes are “inconsistent regimes” (Gates et al. 2006) in which democratic forms and authoritarian practices interact in ways that rulers intend to be mutually supportive. Their intentions may get frustrated, though. As formally representative institutions under authoritarian rule are arenas of conflict, rather than simple tools of authoritarian control, opposition actors may appropriate them and turn them into “subversive institutions” (Bunce 1999). My empirical analyses repeatedly pointed to two representative institutions that international actors should concentrate their subversive energies on: legislatures and media. If left to sovereign authoritarian control, both interact with the electoral arena in ways that corrode interparty competitiveness and solidify the authoritarian status quo. Weak legislatures facilitate electoral fraud and hinder regime change. Controlled media make competitive regimes less competitive and more durable. In hegemonic regimes, they undercut the ability of opposition parties to mobilize electoral protest. By disturbing the monochrome tedium of rubber-stamp legislatures and acclamatory media, international actors can deepen the subversive potential of multiparty elections.

Inducing Structural Changes In policy circles and public debate, participants often voice the old sociological conviction according to which political democratization follows economic development. This venerable axiom of modernization theory carries practical consequences. It justifies putting economic growth first, in contemporary China for instance, under the expectation that political development will follow suit. In the comparative study of regimes, however, the long-standing debate on the impact of economic development on democratization has produced one rather firm conclusion that runs counter to modernizing optimism:

384

Conclusion

economic growth does not destabilize authoritarian regimes. The locus classicus is Przeworski et al. (2000). The practical implication is evident: promoting economic growth promotes economic growth (at best), but not political democratization. My own data have provided a split picture: Wealth stabilizes hegemonic regimes, but destabilizes competitive regimes. We can hardly conclude that we should induce poverty in hegemonic contexts and wealth in competitive ones. The certain moral implications of poverty transcend its probabilistic impact on regime change. Still, the association of wealth with hegemonic stability may heal us from self-delusions when we promote economic growth in hegemonic regimes. And the association of wealth with competitive fragility may alleviate our conscience when we promote economic growth in competitive regimes. In the former case, we back up authoritarianism, in the latter, we help to undermine it. In both instances, we can do something else: promote green energies, so that one day we can stop greasing electoral authoritarian governments by stuffing oil money into their pockets. My empirical explorations did not pretend to be the last word on the subject. Nor do my practical recommendations. It is always political actors who have the last word. They can change their mind or betray their word. They can act differently today than they did in the past. And they can respond to almost any recommendation by anticipating and subverting it. Whenever we try to get practical, we run the risk of “contaminating our dataset.”

THE FUTURE OF COMPARATIVE AUTHORITARIANISM Social research, by its very nature and by the nature of society, is a story without ending, an argument without closure. It is always open to be continued and rewritten by new evidence, new concepts, new theories. Any individual piece of investigation cannot be but a small contribution to a larger collective enterprise. The most natural way of concluding a research journey is therefore to start thinking about the routes that lie ahead. What are its implications for future research? Although this book has been primarily a study of electoral authoritarianism, it intended to contribute to the more general literature on authoritarianism as well. Its analytical framework, though tailored to the specific institutional configuration of electoral autocracies, has been grounded in a general theory of politics under authoritarian rule. My concluding reflections on pathways of future research will therefore touch upon both, the comparative study of authoritarianism in general and electoral authoritarianism in particular.

Conclusion

385

Deepening the Study of Authoritarian Uncertainty Under authoritarian rule, the future is insecure and the present opaque. Much more so than under democratic conditions. The “politics of uncertainty” is the competitive struggle over these intrinsic uncertainties of authoritarian governance: its existential insecurities (institutional uncertainty) and its structural opacities (informational uncertainty). As I maintained in this book, this twin struggle forms the core of regime conflicts under dictatorship. In principle, both forms of uncertainty, the uncertainty of the future and the uncertainty of the present, are central to our theories of authoritarian politics. At least as independent variables, if not as dependent variables. In practice, however, both are marginal to contemporary empirical research. We study neither actor beliefs nor expectations empirically. We bridge them through proxies, place them into black boxes of axiomatic assumptions. I have done so myself in this book. While programmatically actor-centered, my thin quantitative data strove to capture actor behavior, not actor expectations. My theory of the politics of uncertainty did not guide the collection of data, but their interpretation. Yet, if we wish to take uncertainty seriously in the comparative study of authoritarianism, we need to rethink our methodologies, our concepts, and our theories. We also need to reconsider the implications of opacity for our own research. Methodology. In terms of basic description, we have neglected to study empirical variations in actor beliefs and expectations. Above all, we know very little, and have cared very little to know, how political actors perceive the strength of institutions. For instance, we possess abundant public opinion data about popular support for democracy, yet very few on popular perceptions of democratic resilience.4 In part, our reluctance to gather comparative data on subjective perceptions seems to derive from methodological cautions. Taking uncertainty seriously means taking actor expectations seriously. We are often prepared to do so at a conceptual and theoretical level, but not at an empirical level. In the study of political regimes, as in contemporary political science in general, we tend to suffer from methodological schizophrenia. We build our theories on the pillars of strictly unobservable phenomena, such as rational calculation, yet demand our empirical research to rest on strictly observable foundations, such as public behavior. Yet beliefs and expectations are not observable. We cannot grasp them through direct visual inspection. To learn what actors believe and expect, we would need to ask them and listen (critically) to what they say in public speech, text, and deed. Linguistic acts, however, do not count as legitimate evidence in positive political science. Only behavior does.

386

Conclusion

The underlying justification is self-deceptive. We tend to imagine behavioral acts as observable (and nothing but observable) and speech acts as symbolic (and nothing but treacherous). The distinction does not withstand scrutiny. Behavior is no less symbolic than language. Observing human conduct requires similar social knowledge and symbolic competence as understanding linguistic expressions. In the social sciences, we “do not study inanimate, objectively given realities, but symbolic, socially constructed and interpreted realities. To grasp them, we need to understand them. Simple and pure observation won’t do” (Schedler 2012b: 26). Concepts. In the comparative literature on political regimes, we work with sophisticated notions of strategic action. Our conceptions of strategic communication, by contrast, are exceedingly simple. In consequence, we are unable to grasp the communicative imperatives of authoritarianism that arise from its informational opacities. Since the foundations of authoritarian power are opaque, actors need to dramatize them. They need to render visible what cannot be reliably observed by relevant audiences. In the comparative literature on authoritarianism, dictators employ one central communication technique to let others know what they want them to know: signaling.5 The underlying model of communication is rather mechanical. A signal is “a gesture, action, or sound that is used to convey information or instructions, typically by prearrangement between the parties concerned.”6 Signals are actions that trigger actions. They do so on the basis of shared meaning. For actions to work as signals, participants need to understand them as such. They must be conventional or commonsensical. If we wish to study the communicative foundations of dictatorship, we need to free ourselves from the conceptual straitjacket of signals. Signals are commonly exchanged by actors who cannot speak directly to each other, either because they are at a certain distance from each other or because they cannot speak openly in the presence of third parties. In our theories of communication under authoritarianism, political actors do not speak either. They just act and watch, understand, and respond to each other. They inhabit simple worlds of simple acts and simple meanings. No one talks, no one acts outside pre-established scripts. Observations are evident, their significance is known. Neither linguistic nor symbolic acts are relevant, neither doubts nor confusion exist. All messages are behavioral and they are all clear and transparent. The opacity of authoritarian rule means that nothing is obvious under authoritarian rule. In our theories of signaling, by contrast, about everything seems obvious. Needless to say, simple acts of signaling do not exhaust the repertoire of communicative strategies under conditions of opacity, neither in ordinary life (Goffman 1959) nor in ordinary dictatorship (Schedler and Hoffmann 2012).

Conclusion

387

Theories. We have neglected to sharpen the conceptual tools that would allow us to grasp the management of uncertainty under dictatorship. We have also neglected to build theories that would allow us to understand the formation of beliefs and expectations under dictatorship. We have given little systematic thought to how given structural and strategic contexts translate into actor perceptions. Instead, we have tended to treat political expectations as mere epiphenomena, as simple and direct reflections of contextual factors. Scholars of the political economy of regimes, for instance, derive actor perceptions from the objective distribution of material resources. Social groups, they assume, feel threatened by others who are believed to have an objective interest in exploiting them and the means to do so (Boix 2003). In the absence of explicit theories of expectation formation, threat perceptions themselves tend to drop out of sight. Their presence is assumed, rather than established. In his book on contentious politics and state building in Southeast Asia, Dan Slater (2010) provides an outstanding example of systematic thinking about the structural bases of political expectations. Social and economic elites, he reasons, are unlikely to band together and invest in the construction of state infrastructural power unless they are threatened by political conflicts that demand the protective hand of the state, its “iron fist” (Slater 2003). The question is under which conditions such common threat perceptions emerge that possess the power to coalesce political elites into defensive action. Elites tend to accept the sacrifices necessary to finance a strong state only when they perceive social conflict to be “endemic and unmanageable” (Slater 2010: 14). As Slater argues, such threat perceptions are likely to arise only when violent “class conflict afflicts urban areas and exacerbates communal tensions” (Slater 2010: 14, emphases removed). Unless conflict unfolded along class cleavages, unless it turned violent, unless it moved from the countryside to the city, and unless it reinforced explosive ethnic tensions, social elites in 20th-century Southeast Asia were unlikely to form convergent perceptions about clear and imminent dangers from below (see Slater 2010). Political actors are able to form convergent expectations about the future behavior of others to the extent that their environment provides clear clues, that is, clearly visible and clearly relevant clues. If contextual information is abundant and consistent, they can easily form strong and convergent expectations. If it is scarce and contradictory, their expectations tend to be undefined and contradictory. Coordinating expectations under conditions of high informational uncertainty is hard. In the absence of common understandings of reality, it becomes a matter of highly contingent social interactions. In his magnificent study of legislative self-abdication, historian and political theorist Ivan Ermakoff shows that contextual factors are not determinative when actors struggle to coordinate their expectations under uncertainty. In

388

Conclusion

March 1933, the German Reichstag voted to cede sovereign power to chancellor Hitler. In July 1940, the National Assembly of the French Third Republic voted to transfer full constitutional authority to Marshal Pétain. These were momentous decisions, taken under momentous uncertainty. Though none of them were taken under democratic conditions, both ratified the irrevocable death of democracy. Ermakoff (2008) reconstructs both processes of collective decision-making by reconstructing the perspectives of actors: what they saw and knew, what they said and wrote, what they feared and desired, with whom they interacted and how. In his historically meticulous and theoretically insightful account, he shows in an exemplary fashion how we can come to understand the convergence of political expectations under extreme uncertainty by reconstructing, not the external environment, but the internal viewpoints of participating actors. Self-reflection. If it is true that authoritarianism breeds opacity and that dictators have to cope with it as a fundamental fact of political life, then we students of authoritarianism have to cope with it as well. We cannot go on pretending that we are privileged observers who know about everything from the casual observation of salient structural and institutional factors. We are commonly referring to things like the strength of the ruling party, the loyalty of the security apparatus, the popularity of the regime, the cohesion of the elite, the credibility of the dictator, or the threat of rebellion, as if we knew about them. But we don’t. Not for sure. Nobody can. Just like everybody else, we can only make educated guesses. We can form beliefs that are as shaky as the evidence they are based on and as persuasive as the rules of inference they derive from. Deepening the Study of Authoritarian Elections I conceived authoritarian elections as two-level games in which the struggle for voters at the game level goes hand in hand with the struggle over rules at the meta-game level. My own research, just like most of the empirical literature on authoritarian elections, focused on the meta-level of institutional conflict. The void of comparative research is evident: there is very little that we know about the game level of authoritarian electoral competition. There is little that we know about party organizations, candidate selection, electoral alliances, election campaigns, public discourse, media content, media consumption, and voter behavior. Once we know more about these game-level structures and processes, we will be in a position to trace their interactions with meta-game structures and processes. The nested game of authoritarian elections is nested in other games. Elections are nested in national societies. Regime actors are nested in the state,

Conclusion

389

opposition actors in civil society. Local elections are nested within national elections. National politics is nested in international politics. Studying the internal interaction between the two levels of authoritarian elections may seem complex enough. Yet, if we wish to better comprehend the dynamics of authoritarian elections, we need to study their external linkages as well. Comparative research about them is barely commencing. Over the past decades, we have seen an impressive growth of cross-national quantitative data in comparative politics (see Schedler and Mudde 2010 and Schedler 2012a). In the new and exciting research field on electoral authoritarianism, too, numerous scholars have engaged in the development of cross-national datasets, above all, on electoral manipulation and electoral competitiveness. To push comparative research on authoritarian elections onto the next stage, we need to revise, integrate, complement, and consolidate the data we have collected so far. And we need to construct new and better data on almost all aspects of authoritarian elections: x Election results: Incredible, but true. Despite countless private initiatives of data collection, access to historical and contemporary national election data, not to speak of subnational data, is still precarious. We still need to institutionalize the systematic collection of data on elections and parties across the world (see Schedler 2012a: 256–257 and 259). x Electoral governance: We possess certain cross-national data on rules of electoral competition (see Teorell and Lindstedt 2010), but very few on institutions of electoral governance, such as suffrage rights, rules of voter and party registration, and the structure of election management and electoral dispute settlement. x Voter preferences: If we wish to apply the analytical tools of electoral studies to authoritarian contexts, we need to collect individual-level data on voter attitudes. Electoral autocracies tend to inhabit the middle ground between full liberty, where pollsters can work freely and publics respond freely, and full repression, where public opinion polls make little sense. As a matter of fact, electoral autocracies like Russia under Putin often permit the development of a professional community of public opinion polling. Still, though we possess rich data on individual countries, we possess few cross-national data on voters under authoritarian conditions. We still need to incorporate electoral authoritarian regimes into cross-national political surveys like the Comparative Study of Electoral Systems (). x Election campaigns: Collecting comparative data on authoritarian regimes is difficult. Collecting them on processes of electoral competition is difficult even in democratic contexts. Still, data projects like the Comparative Manifesto Project (mapor) that codes the content of election platforms

390

Conclusion

should in principle be adaptable to authoritarian settings (). x Electoral protest: The only available worldwide source of cross-national longitudinal data on contentious action, the political conflict data in the Arthur Banks Cross-National Time–Series (cnts), captures no more than a minuscule fraction of the contentious events that actually take place in any country in any year (see Schedler 2012a: 247–248). In a joint venture with scholars of contentious action, students of electoral politics might set out to develop more accurate data on electoral protest, and on extra-electoral protest as well. As a matter of course, even within a quantitative framework, we need not study everything in cross-national perspective. We also need systematic observations on subnational elections and electoral histories in single countries. If almost half of all quantitative comparative research (published in top journals between 1989 and 2007) covers single countries (Schedler and Mudde 2010: 421), there is no reason why the quantitative study of electoral autocracies should be different.

Bringing Domination Back In By paying attention to formally representative (and formerly neglected) institutions under dictatorship, the “new institutionalism” in the comparative study of authoritarianism has been producing an insightful stream of research. It is barely in its infancy, though. We have learned most about elections and need to know much more about the concrete structures and empirical roles of other institutions, like legislatures, mass media, courts, and constitutions. At the same time, we must not forget that authoritarian regimes are founded, not upon formal institutions of representation, but upon institutions of domination. It is surprising how little we know about them in comparative terms. We know more about practices and outcomes than about institutional structures. For instance, we have more and better cross-national data on state repression than on state agencies of repression. We lack comparative data on basic facts like the division of power between agencies of state violence, the size of the secret police, or the appointment procedures and tenure of the military command. The “new institutionalism” in the comparative study of dictatorship risks turning into a lopsided enterprise if it fails to take institutions of domination seriously, too.

Conclusion

391

Bringing History Back In In view of the remarkable rise of electoral autocracies since the end of the Cold War, we tend to forget that authoritarian multiparty elections have a long history, in particular in Europe and the Americas of the 19th century. Through research alliances between political scientists and historians we might be able to extend the comparative study of authoritarian elections deep into the past. Bringing history back in seems like a good idea as well for the study of authoritarianism in general. When studying electoral authoritarian regimes, we are often oblivious to the modern history of authoritarian elections. Similarly, when studying authoritarian regimes, we political scientists tend to be oblivious to the modern history of dictatorship. Due to pragmatic reasons of data availability, we commonly start our inquiries in the 1970s or 1980s, but almost never before 1945. The totalitarian regimes of Hitler and Stalin fall outside of our purview. We delegate their study to historians and pretend these big civilizational catastrophes of the 20th century have never happened. Clearly, in addition to the “historical turn” in democratization studies (Capoccia and Ziblatt 2010), we need a historical turn in the comparative study of authoritarianism. In conclusion: If the hill is steep and the rock feels heavy, if social research is a never-ending story and the agenda of comparative authoritarianism looks dizzyingly broad and grimly demanding, we should remember Albert Camus: “One must imagine Sisyphus happy” (1991: 123).

Notes 1. The original phrase is attributed to Otto von Bismarck (). 2. On contrasting assumptions prevalent in politics and political science about the contingency of actors, action, and events, see Schedler (2007b: 58–61 and 69–70). 3. When the democratic international community intervenes in the domestic affairs of authoritarian regimes, it often uses the language of democracy assistance. This makes for a strange choice of terminology. Who would describe chemotherapy as “health assistance” to cancer? My term “democratization assistance” is more precise, even if somewhat over-optimistic. 4. See Schedler (2001) and Schedler and Sarsfield (2007). 5. See e.g. Debs (2007), Edmond (2011), Magaloni (2008), Malesky and Schuler (2011), and Myerson (2008). 6. “signal,” New Oxford American Dictionary, ed. Angus Stevenson and Christine A. Lindberg (Oxford: Oxford University Press, 2010, 3rd edn.).

This page intentionally left blank

APPENDIX A

Case Selection Table A.1. Electoral authoritarianism: criteria of case selection Criteria of case selection and exclusion

Rules of selection, exclusion, and exception (with cases and examples)

Authoritarian governance Exclusion of democratic Operational definition of authoritarianism on the basis of Freedom regimes House (fh) annual reports on Freedom in the World: Regimes that obtain Political Rights (pr) scores of 4 or higher (worse). All regimes with pr scores of 3 or lower are classified as democracies and thus excluded from dataset. Rule of exception: gaps t 2 between pr and Freedom House Civil Liberty (cl) scores (with cl > pr). In general, fh evaluations of political rights and civil liberties match closely. Only in a handful of instances, the two diverge by more than one point. Regimes that received political rights scores of 3, but civil liberty ratings of 5, seemed illiberal enough to merit inclusion (rather than be treated as democratic). Cases: El Salvador in 1984 and Pakistan in 1993. Adjustment of annual scores: As Freedom House qualifies entire calendar years, its annual scores usually do not reflect the quality of “deviant” elections held early or late in the year. The dataset corrects for false authoritarian positives as well as for false negatives. False authoritarian positives: In some cases, countries hold foundational or transitional elections that involve democratic improvements that Freedom House does not register until the subsequent year. Despite their “bad” fh ratings, we exclude such democratizing elections from the basket of “authoritarian” elections. Cases: Kenya in 2002, Indonesia in 1999, Korea in 1987, Madagascar in 1992, Mozambique in 1994. False authoritarian negatives: In some cases, authoritarian elections are followed by democratic breakthroughs later in the year. We treat the pre-transition election as authoritarian, even if the fh annual rating reflects the posterior democratic opening. Case: the 2000 presidential elections in Peru. (Continued)

394

Appendix A

Table A.1. (Continued) Criteria of case selection and exclusion

Exclusion of closed authoritarianism

Exclusion of short-term fluctuations

Multiparty competition Institutional requirement Exclusion of regimes that do not admit legal multiparty competition (even when they fall within the required range of fh scores)

Rules of selection, exclusion, and exception (with cases and examples) Case exception: Mexico 1982. Between 1979 and 1984, Mexico’s fh political rights scores improved temporarily from four to three. As it makes little sense to conceive these years as an interlude of democratization (even if some liberalization did take place), I included the 1982 concurrent elections into the dataset. This decision was not based on rules, but judgment (my subjective aversion against excluding an election from my “concept-generating case” on quasi-bureaucratic grounds). Operational definition on the basis of Freedom House: Electoral autocracies (included in dataset) earn Political Rights scores of 6 or lower, closed autocracies (excluded from dataset) obtain pr scores of 7. Rule of exception: gaps t 2 between pr and cl scores (with pr > cl). Regimes that received political rights scores of 7, but civil liberty ratings of 5, seemed liberal enough to merit inclusion (rather than be discarded as closed autocracies). Cases: Cameroon in 1997, Niger in 1996, Indonesia in 1997, and Togo in 1997. The notion of a “political regime” implies some minimum of institutional stability. I included elections only if the country remained within the specified range of pr scores (4–6) during at least four consecutive years, including the election year. Implication: Exclusion of electoral democracies with only three (or less) consecutive years of “bad” pr scores of 4, such as India (from 1993 to 1995) and Colombia from 1995 to 2005 (when its pr scores tended to improve every four years when national elections were held and to worsen in between). At least one full set of multiparty elections for the national legislature (Lower House) and the chief executive (in presidential regimes). Exclusion of military dictatorships: Argentina (until 1983), Bolivia (1980–81), Chile (until 1989), Uruguay (until 1985). Exclusion of traditional monarchies: Bahrain, Bhutan, Oman, United Arab Emirates, Saudi Arabia, Brunei, Swaziland, Nepal (before 1991). Exclusion of de jure non-party or single-party regimes (even if they allow for intra-party competition): Cuba, China, Libya, Poland (until 1989), Hungary (until 1989), the Islamic Republic of Iran, Taiwan (until 1989), Tanzania under Nyerere, Uganda under Museveni (until 2005).

Appendix A

395

Criteria of case selection and exclusion

Rules of selection, exclusion, and exception (with cases and examples)

Exclusion of regimes that do not admit factual multiparty competition

Exclusion of multiparty regimes in which opposition parties fail to win seats in legislative elections (without explicit opposition boycott): Cameroon in the 1980s, Côte d’Ivoire in 1980 and 1985, Gabon in 1980 and 1985, Kenya in the 1980s, Mali in the 1980s, Niger in 1989, Singapore before 1984, Togo in 1985 and 1990. Exclusion of multiparty regimes in which winning candidates run unopposed in presidential elections (without explicit opposition boycott): Burkina Faso in 1991, Cameroon in the 1980s, Côte d’Ivoire in 1980 and 1985, Kenya in the 1980s, Niger in 1989, Togo in 1986, Tunisia before 1999. Almost none of the “closed” authoritarian regimes as defined above (on the basis of Freedom House pr scores of 7) hold minimally competitive elections. Those that do are still excluded from the dataset. Cases: the 1996 presidential elections in Gambia, in which coup leader Yahya A. J. J. Jammeh won “only” 55.8 percent of valid votes; and the 1992 legislative elections in Mauritania, in which the ruling prds won 67.7 percent of valid votes.

Exclusion of multiparty elections under repressive conditions

Chief executive elections Exclusion of electoral Exclusion of traditional monarchies, even if they allow legislative regimes that hold assemblies to be elected in multiparty elections: Jordan, “limited” elections Kuwait, Morocco. only Exclusion of non-democratic regimes that hold national legislative elections, but not elections for the chief executive: authoritarian Brazil until 1989, South Korea 1980–89. Exclusion of non-democratic regimes that hold subnational elections, but not elections for the chief executive: Taiwan before 1996. Borderline categories: Inclusion of two borderline categories: the parliamentary systems under military tutelage of Pakistan (until 1999) and Turkey (until 2002), and the quasi-presidential regimes of Egypt, Ethiopia, and Indonesia (before 2004) in which elected parliaments chose presidents with fixed terms. Universal suffrage Exclusion of exclusionary regimes External sovereignty Exclusion of countries with formal or de facto suppression of national sovereignty

Exclusion of electoral regimes with de jure restricted suffrage (“competitive oligarchies”): South Africa under apartheid. Exclusion of states under open foreign tutelage. Polity iv category “interruption” (“a country is occupied by foreign powers during wartime, terminating the old polity”). Examples: Afghanistan 1979–93, Cambodia 1979–90, Lebanon (under Syrian occupation after 1975), Bosnia and Herzegovina (under international protection since 1995), Poland and Hungary before 1989 (as members of the Warsaw Pact under Soviet tutelage), former Soviet Republics before formal independence in 1991. (Continued)

396

Appendix A

Table A.1. (Continued) Criteria of case selection and exclusion Internal sovereignty Exclusion of countries without minimal statehood

Database entry and exit Time period

Entry rules

Exit rules

Rules of selection, exclusion, and exception (with cases and examples) Exclusion of cases of state collapse, identified through Polity iv categories “transition” and “interregnum” (“a complete collapse of central political authority”). Examples: Angola 1992, Burundi 1993–95, Chad 1979–83, Congo-Kinshasa 1992–2002, Ethiopia 1991, Lebanon 1975–89, Lesotho 1998, Liberia 1990–95, Nicaragua 1980, Sierra Leone 1997–2000, Somalia 1991–2002, Uganda 1985. The dataset covers national election cycles conducted without interruption under conditions of electoral authoritarianism (as defined above) in the time period from 1980 to 2002. Regimes enter the dataset with their first set of national authoritarian elections held after 1979. Six electoral authoritarian regimes included in the dataset were founded before 1980: Mexico (1929), Paraguay (1954), Indonesia (1968), Malaysia (1957), Philippines (1972), and Singapore (1965). The dataset includes the last national election cycle held in an electoral authoritarian regime before the year 2003. Regimes exit the database before 2003 in one of two cases: (1) They go through a democratic transition, defined by a sustained improvement of Freedom House pr scores (over at least four consecutive years). (2) They suffer an extra-constitutional “interruption” of their electoral cycle (e.g. by military coup, foreign intervention, insurgency, or plebiscitary self-perpetuation in power). Term extensions of sitting chief executives either through plebiscitary means (referenda) or through factual delays (postponement of elections) count as interruptions if they prolong the presidential term by more than one half of the original constitutional term. Forms of interruptions of electoral cycles: Military intervention. Examples: 1990 military coup in Haiti, ousting President Jean-Bertrand Aristide, elected earlier that year; 1992 military rebellion in Azerbaijan, ousting President Abulfaz Elchibey, elected earlier that year. Extra-constitutional removal from office. Overthrow of President Zviad Gamsakhurdia in Georgia in 1992; overthrow of President Rakhman Nabiyev of Tajikistan, who was elected in 1991 and forced to resign in 1992 by Emomali Rakhmonov (who in continuation provoked civil war and authoritarian closure). Ex post extension of terms of office. 1995 referendum in Kazakhstan, extending the term of President Nursultan Nazarbayev, who had been elected in 1991 for a five-year term, for another full term. Note: the 1996 referendum in Belarus, extending the term of President Alyaksandr Lukashenka, who had been elected in 1994 for a five-year period, by “only” two years from 1999 to 2001 (less than half a presidential term), does not count as “interruption.”

Appendix A Criteria of case selection and exclusion Size Exclusion of small polities

397

Rules of selection, exclusion, and exception (with cases and examples)

Exclusion of small states with population lower than 1 million in the year 2002. Source: United Nations, Department of Economic and Social Affairs, Population Division ().

Data availability Exclusion of cases for lack of data

One country that fulfilled all selection criteria, but was excluded due to pervasive data problems: Yugoslavia (Federation of Serbia and Montenegro).

Hegemonic regimes To distinguish hegemonic from competitive regimes, I relied on two criteria: (1) A minimum duration of ten years (since the assumption of power by the ruling coalition). (2) The continuous control of legislative supermajorities (with the ruling party holding at least two-thirds of seats in the Lower House). Continuous rule: As it refers to the occupation of state power by the “ruling coalition,” my duration requirement does not invariably demand a decade of overwhelming victories in multiparty elections. A ruling party may be classified as hegemonic at the onset of multiparty elections if its overwhelming victory in first multiparty elections were preceded by at least a decade of single-party rule. An extreme case is Albania. I classify the Socialist Party (the former Albanian Labor Party) as hegemonic only for one election, the first multiparty election it convoked in 1991 after 55 years of single-party rule. The reconverted Communist party won by a landslide—before losing power by a landslide in the legislative elections held in the year after. Rules of exception: Military might: In a few cases in which rulers had come to power through military coups (and thus were in a good position to send credible signals of hegemonic ambitions at the moment of convoking elections), I handle the 10-year rule with certain flexibility. Example: when Burkina Faso held its first multiparty election in 1991, only eight years had elapsed since the military coup that brought President Blaise Compaoré to power. I nevertheless count the regime as hegemonic from 1991 to 2002 (when it lost its legislative supermajority). Transitory defeat: I also grant minor exceptions to the rule of continuous supermajorities. The governing parties of Gabon (in 1990), Guinea (in 1995), and Togo (in 1994) suffered transitory losses of their comfortable supermajorities. All three were quick to repair their electoral “accidents” and recovered their qualified legislative majorities in the subsequent elections.

Table A.2. Authoritarian election cycles, 1980–2002

Region and countries

First election in dataset

1 Latin America & Caribbean Colombia 2002

Last election in Successor regime (mode of transition dataset and subsequent election)

#L

#P

Electoral democracy (erosion, civil war)

2002

1

1

1

1

1

1

2

1

3

2

5

3

1 1 2

1 2 2

Predecessor regime (mode of transition)

El Salvador

1984

Military regime (opening)

1985

Guatemala 1

1985

Military regime (opening)

1985

Guatemala 2

1994

Electoral democracy (erosion)

1995

Haiti

1995

2000

Mexico

1982

Nicaragua Panama Paraguay 1

1984 1984 1983

Paraguay 2

1989

Paraguay 3

1998

Military regime (external intervention) Electoral authoritarianism (since 1929) Single-party regime (opening) Military regime Electoral authoritarianism (since 1954) Electoral authoritarianism (interruption, military coup) Electoral democracy (erosion)

Peru

1995

Electoral democracy (interruption by executive coup)

1994 1984 1989 1988

Electoral democracy (borderline) (re-equilibration) (concurrent elections 2006) Electoral democracy (peace accord) (legislative elections 1988) Electoral democracy (peace accord) (concurrent elections 1990) Electoral democracy (re-equilibration) (concurrent elections 1999) Interruption (armed revolt 2004) Liberal democracy (opposition victories 1997 and 2000) Electoral democracy (opposition victory 1990) Interruption (rejection, external intervention) Interruption (military coup 1989)

1989

Electoral democracy (concurrent elections 1993)

1

1

1998

Regime continuity (concurrent elections 2003)

1

1

2000

Liberal democracy (executive abdication, concurrent elections 2001)

2

2

2 Eastern Europe Albania

1991

Single-party regime (opening)

1997

Electoral democracy (legislative elections 2001)

4

0

Belarus

1994

2001

2

1992

1997

Regime continuity after 2002 (legislative elections 2004) Liberal democracy (opposition victory 2000)

2

Croatia

2

2

Macedonia

1994

1994

Electoral democracy (borderline fluctuations)

1

0

Moldova

1994

1994

0

1990

1992

Electoral democracy (presidential elections 1996) Liberal democracy (legislative elections 1996)

1

Romania

2

2

Russia

1999

Single-party regime (independence 1991) Single-party regime (independence 1991) Electoral democracy (interlude, erosion) Single-party regime (independence 1991) Single-party regime (palace coup and opening) Electoral democracy (erosion)

2000

Regime continuity (legislative elections 2003)

1

1

1999

Regime continuity (presidential re-election 2 2003) Regime continuity (hereditary re-election 2003) 2

2

3 Central Asia & Caucasus Armenia 1995 Azerbaijan

1993

Georgia

1992

Kazakhstan

1995

Kyrgyzstan

1991

Single-party regime (independence 1991) Military coup (interrupting democratic interlude after independence in 1991) Violent coup (interrupting democratic interlude after independence in 1991) Single-party regime (independence 1991) Single-party regime (independence 1991)

2000

2

2000

Regime continuity (legislative elections 2003, electoral revolution)

3

2

1999

Regime continuity (presidential re-election 2005) Regime continuity (legislative elections 2005, electoral revolution)

2

1

2

2

2000

(Continued)

Table A.2. (Continued)

Region and countries

First election in Predecessor regime (mode of dataset transition)

Last election in dataset

Successor regime (mode of transition and subsequent election)

Tajikistan

1999

Violent coup (interrupting democratic interlude after independence in 1991), followed by civil war and authoritarian closure.

2000

Regime continuity (legislative elections 2005)

1

1

2002

Regime continuity (presidential re-election 2004)

2

2

2000

Regime continuity (presidential re-election 2005) Regime continuity (presidential re-election 2004) Electoral democracy Electoral democracy (re-equilibration) Continuity (presidential re-election 2006)

5

0

1

1

1 2 1

0 0 1

3

1

3

2

2

2

4 Northern Africa & Middle East Algeria 1995

Egypt

1984

Single-party regime (transition aborted by military coup, followed by civil war). Single-party regime

Tunisia

1999

Single-party regime

1999

Turkey 1 Turkey 2 Yemen

1983 1995 1997

Military regime Electoral democracy (erosion) Civil war in 1994 (after state foundation in 1990 and legislative election in 1993)

1983 1999 1999

5 Sub-Saharan Africa Burkina-Faso

1992

Single-party regime

2002

Cameroon

1992

Single-party regime

2002

Chad

1996

Single-party regime

2002

#L

Regime continuity (presidential re-election 2005) Regime continuity (presidential re-election 2004) Regime continuity (presidential re-election 2006)

#P

Côte d’Ivoire 1

1990

Single-party regime

1995

Interruption (military coup)

2

2

Côte d’Ivoire 2 Ethiopia

2000 1995

Interruption (military coup) Single-party regime

2000 2000

1 2

1 0

Gabon

1990

Single-party regime

2001

3

2

Gambia

2001

2002

1

1

Ghana

1992

Military regime (after abolition of electoral democracy by military coup) Military regime

Interruption (reversion through mass protest) Regime continuity (contentious legislative election 2005) Regime continuity (presidential re-election 2005) Regime continuity (presidential re-election 2003)

1

1

Guinea

1993

2002

2

2

Kenya Mauritania Niger 1

1992 1996 1996

2 2 1

2 1 1

Niger 2

1999

1

1

Senegal Tanzania

1983 1995

Military coup (ending single-party regime) Single-party regime Single-party regime Military coup (ending electoral democracy) Military coup (interrupting electoral autocracy) Single-party regime (opening) Single-party regime (opening)

1998 2000

4 2

3 2

Togo

1993

Single-party regime (opening)

2002

3

2

Zambia

1996

Electoral democracy (erosion)

2001

2

2

1992

1997 2001 1996 1999

Electoral democracy (concurrent elections 2006) Regime continuity (presidential re-election 2003) Electoral democracy (opposition victory 2002) Interruption (military coup 2005) Interruption (military coup 1999) Regime continuity (presidential re-election 2004) Electoral democracy (opposition victory 2002) Electoral democracy (election of official candidate 2005) Regime continuity (presidential re-election 2003) Electoral democracy (presidential re-election 2006)

(Continued)

Table A.2. (Continued) Last election in dataset

#L

#P

Regime continuity (legislative elections 2005, presidential re-election 2008)

4

3

1993

Interruption (military coup 1997)

1

0

1998 1997

Regime continuity (legislative elections 2003) Electoral democracy (opposition victory 1999)

1 4

0 0

1999

Regime continuity (legislative elections 2004)

5

0

1997

Interruption (military coup 1999)

3

0

1986

Liberal democracy (electoral revolution 1986)

1

2

2001

Regime continuity (legislative elections 2006)

6

0

1994

Electoral democracy (re-equilibration) (presidential elections 1999)

1

1

Region and countries

First election in dataset

Predecessor regime (mode of transition)

Zimbabwe

1985

Democratic interlude after independence 1980. Single-party regime

2002

6 South & East Asia Cambodia 1

1993

Cambodia 2 Indonesia

1998 1980

Malaysia

1982

Pakistan

1990

Philippines

1980

Singapore

1984

Sri Lanka

1994

Single-party regime (war transition) Military coup Electoral authoritarianism (since 1968) Electoral authoritarianism (since 1957) Democratic interlude after military regime Personal dictatorship (since 1972) Electoral authoritarianism (since 1965) Electoral democracy (erosion, civil war)

Successor regime (mode of transition and subsequent election)

Note: # L = Number of legislative elections, # P = Number of presidential elections. Successor regimes after 2002 (end of dataset): If regime holds another full set of authoritarian elections, counted as regime continuity.

Table A.3. Hegemonic party regimes, 1980–2002 Country

Ruling party

Albania1

Albanian Labor Party PT (Socialist Party in 1991) Institutional PRI Revolutionary Party National Republican ANR Association— Partido Colorado

Mexico

Paraguay

Acronym

Initiation Regime origin of rule

Personal leadership

Years in office

Hegemonic Mode of regime termination termination

1946

Single-party rule, electoral opening

...

1992

1929

Civil war

...

1988

1954

Military coup (party Alfredo founded in 1887) Stroessner

1954–1989

1993

Egypt

National Democratic Party

NDP

1952

Military coup 1952, party foundation 1978

Hosni Mubarak

1981–2011

2011

Tunisia

Constitutional Democratic Rally

CDR

1956

Independence, soft coup 1987

1987–2011

2011

Burkina Faso

Congress for Democracy and Progress Democratic Party of Côte d’Ivoire

CDP

1983

Military coup, electoral opening

Zine el-Abidine Ben Ali Blais Compaoré

1983–

2002

PDCI

1960

Independence, electoral opening

Henir Konan Bedie

1993–1999

1999

Côte d’Ivoire

Electoral alternation in power Loss of legislative supermajority Loss of legislative supermajority (after coup 1989) Presidential resignation after popular uprising Presidential exile after popular uprising Loss of legislative supermajority Military coup

(Continued)

Table A.3. (Continued) Country

Ruling party

Acronym

Gabon

Gabonese Democratic PDG Party

Initiation Regime origin of rule

Personal leadership

Years in office

Hegemonic Mode of regime termination termination

1960

Omar Bongo

1967–2009

...

Ali Ben Bongo Ondimba Lasana Conté

2009–

...

1984–2008

2008

1984–2005

2005

1981–2000

1998

Benjamín Mkapa Gnassingbé Eyadéma

1995–2005

...

1967–2005

2005

Independence

Guinea

Progress and Unity Party

PUP

1984

Military coup, electoral opening

Mauritania

Social Democratic Republican Party

PRDS

1978

Military coup, electoral opening

Senegal

Socialist Party

SP

1960

Tanzania

Chama Cha Mapinduzi Rally of the Togolese People

CCM

1961

RPT

1967

Independence, electoral opening Independence, electoral opening Military coup, electoral opening

Zimbabwe African National Union— Patriotic Front Golongan Karya (Functional Groups)

ZANU-PF

1980

Independence

Robert Mugabe

1987–

2000

Golkar

1965

Military coup

Suharto

1965–1998

1999

Togo

Zimbabwe

Indonesia

Maaouya Ould Sid Ahmed Taya Abdou Diouf

Death of president, military coup Military coup

Loss of supermajority ... Death of president, military coup Loss of legislative supermajority Resignation of president after mass protest (1998)

Malaysia

Singapore

United Malays National Organization People’s Action Party

UMNO National Front PAP

1963

Independence

Mahathir Mohamad

1981–2003

1965

Independence

Lee Kuan Yew 1965–1990 Goh Chok 1990–2004 Tong Lee Hsien 2004– Loong

2008

... ...

Loss of legislative supermajority ... ...

...

...

Note: Initiation of rule =Initiation of regime governed by same party, person, or ruling coalition. Personal leadership = dominant chief executive under hegemonic party rule (not single-party period). 1

After 55 years of Communist single-party rule under Enver Hoxha, Albania counts as hegemonic system only for its first multiparty elections in 1991.

This page intentionally left blank

APPENDIX B

Description of Variables Regime

Election

Region

Post-Cold War Regime manipulation Exclusion

Fraud

Repression

Censorship

Cumulative index of manipulation

Subtype of electoral authoritarian regime. 0 = competitive regime. 1 = hegemonic regime (constitutional majority by ruling party ≥ 2/3 legislative seats, regime duration ≥ 10 years) (see Appendix a for case listings). Type of national election. L = legislative election (Lower House), P = direct presidential election, and C = concurrent election (within same calendar year). World region. 1 = Latin America & Caribbean, 2 = Eastern Europe, 3 = Central Asia & Caucasus, 4 = Northern Africa & Middle East, 5 = Sub-Saharan Africa, 6 = South & East Asia. Timing of election. 0 = Cold War election (–1989), 1 = Post-Cold War Election (1990–). Exclusion of parties and candidates from legislative elections (legislative exclusion), presidential elections (presidential exclusion) or either (electoral exclusion = highest value of both). Dummy: 0 = openness, and 1 = exclusion. Coding by author (see Appendix c). Interference in electoral administration for partisan advantage at any stage of legislative elections (legislative fraud), presidential elections (presidential fraud) or either (electoral fraud = highest value of both). Range 0–2, with 0 = no fraud, 1 = irregularities, and 2 = fraud. Coding by author (see Appendix c). Violations of physical integrity (extrajudicial killings, disappearance, torture, and political imprisonment). Range 0–8, with 0 = full respect for basic human rights, and 8 = gross violation of human rights. Inversion of Physical Integrity Rights Index from Cingranelli–Richards Human Rights Data Project (ciri) (). Limitations on freedom of speech and mass media. Range 0–2, with 0 = no restrictions, 1 = partial restrictions, and 2 = systematic restrictions. Arithmetic mean of Freedom of the Press (Freedom House) and inverted Freedom of Speech and Press (ciri). The index normalizes each of the preceding four measures of manipulation to a scale from 0 to 1 and ads them (counting exclusion and fraud in the legislative and presidential arenas only half in the case of concurrent elections). Hypothetical range: 0–4, with 0 = no recourse to any of the four strategies, and 4 = full recourse to all four.

(Continued)

408

Appendix B

(Continued) Opposition protest Boycott

Pre-electoral protest

Post-electoral protest

Withdrawal of main opposition parties from legislative elections (legislative boycott), presidential elections (presidential boycott) or either (boycott = highest value of both). Range 0–3, with 0 = participation, 1 = partial, and 2 = full boycott. Coding by author (see Appendix c). Active mobilization of followers by opposition (e.g. through public demonstrations, street blockades, strikes) in protest against upcoming legislative elections (legislative pre-electoral protest), presidential elections (presidential pre-electoral protest) or either (pre-electoral protest = highest value of both). Dummy: 0 = acquiescence, and 1 = active protest through contentious collective action. Coding by author (see Appendix c). Opposition protest on election day or afterwards against legislative elections (legislative post-electoral protest), presidential elections (presidential post-electoral protest) or either (post-electoral protest = highest value of both). Range 0–1, with 0 = acceptance, 0.5 = rejection of electoral outcomes (verbal, judicial, or symbolic protest), and 1 = active protest mobilization. Coding by author (see Appendix c).

Electoral competitiveness Legislative margin of s1–s2 where s1 is the seat share of the largest party, and s2 the seat share victory of the second party (Lower Chamber). Note: In cases of alternation in power, positive governmental margins of victory turn into negative margins of defeat. Cases: Albania 1992 and 1997, Cambodia 1993, Colombia 2002, Georgia 1992, Guatemala 1985 and 1995, Niger 1999, Pakistan 1990, 1993 and 1997, Sri Lanka 1994, Turkey 1983 and 1999. Presidential margin of v1–v2 where v1 is the vote share of the winning presidential candidate, victory and v2 the vote share of the second-placed candidate. Note: In cases of alternation in power, positive governmental margins of victory turn into negative margins of defeat. Cases: Algeria 1999, Armenia 1998, Belarus 1994, Colombia 2002, El Salvador 1984, Guatemala 1985 and 1995, Niger 1999, Sri Lanka 1994. Margin of victory Margins of victory in legislative and presidential elections. In concurrent elections, presidential margins count. Legislative 1/∑si2 fractionalization where si is the seat share of the ith party (Rae index) (see Rae 1967). Presidential 1/∑vi2 fractionalization where vi is the vote share of the ith presidential candidate (Rae index).

Appendix B Legislative opposition fractionalization Presidential opposition fractionalization

Opposition fractionalization

409

1/∑so2 where so is the seat share of the oth opposition party (Rae index). 1/∑vo2 where vo is the vote share of the oth opposition candidate (Rae index). Note: Trivially, estimating the dispersion of opposition actors presupposes the prior identification of opposition actors. Although the distinction between government and opposition is constitutive for modern democracy, in democratic regimes it is often less than razor sharp. Programmatically, government and opposition parties may be very close. Institutionally, they may be sharing power in a way that blurs the distinction between ins and outs (for instance, in parliamentary regimes under minority government or in presidential regimes under divided government). In authoritarian regimes, the division between government and opposition at times constitutes the primary cleavage of the political system (pro-regime versus anti-regime). Frequently, however, authoritarian governments manage to manipulate the party system to an extent that makes it hard to discern who is fake and who genuine in the crowded field of nominal independents and nominal opposition actors. They set up pseudo-opposition parties, interfere with genuine ones, bribe or intimidate them (see Chapter 3). One might try to cut the fog of political ambiguity and separate apparent from real opposition actors by tapping the nuanced contextual knowledge of country experts. Instead, for pragmatic reasons, I adopted a simple operational rule: I counted the party of the head of government as the ruling party, all others as opposition parties. Sources: I identified ruling parties on the basis of the World Bank Dataset on Political Institutions (dpi), Rulers, a webpage that lists heads government worldwide since about 1700 (), and country narratives in annual Freedom House reports on Freedom in the Word. Note that the index of legislative opposition fractionalization is identical to the overall index of party-systemic fractionalization in rare cases in which the incumbent party fails to gain legislative representation, as in the 2002 elections in Turkey in which prime minister Bulent Ecevit’s Democratic Left Party (dsp) did not win a single seat. Since the idea of opposition parties presupposes the existence of government parties, the calculus of opposition fractionalization is not applicable if no government party contests the election. Rare examples are interim governments run by military officers who first stage a coup and then convoke elections without participating in them (as in Niger in 1999). Opposition fractionalization in legislative and presidential elections. In concurrent elections, presidential fractionalization counts.

(Continued)

410

Appendix B

(Continued) Effective number of legislative parties Effective number of presidential candidates Constitutional majority

Regime trajectories Party alternation

Personal alternation

Electoral sequence

1/∑si2 where si is the seat share of the ith party (Laakso–Taagepera Index) (see Laakso and Taagepera 1979 and Taagepera 1999). 1/∑vi2 where vi is the vote share of the ith presidential candidate (Laakso-Taagepera Index). Ruling party holds two-thirds majority of seats or more in Lower Chamber. 0 = No. 1 = Yes. Sources: Most electoral data are taken from Nohlen (1993), Nohlen et al. (1999), Nohlen et al. (2001), Payne et al. (2002), and University of Essex (2004). These compilations of electoral information were complemented, in particular for more recent years, by Internet sources, such as Election Guide of the International Foundation for Election Systems (ifes) (), the African Elections Database (), the International Parliamentary Union Parline Database (), the Election Results Archive of the University of Binghamton, Center on Democratic Performance (), the Georgetown University Political Database of the Americas (), Keesing’s Record of World Events (), and Wikipedia Election Results (). All calculations by author. Country files with election data (in Excel), specific sources, and a variety of party-systemic indicators are available from the author upon request. Electoral alternation in the party of the chief executive. Official election outcomes designate a winning party distinct from the incumbent party. Excluded: irregular removals from power that precede or succeed an election, such as military coups or electoral rebellions. Electoral alternation in the person of the chief executive. The incumbent loses according to official election outcomes. Excluded: irregular removals from power that precede or succeed an election. Temporal location of an election in uninterrupted sequences of election. Separate counts for legislative (“legislative sequence”) and presidential contests (“presidential sequence”). The variable “electoral sequence” combines both. In case of concurrent elections in which legislative and presidential elections of different length coincide, it registers the lower values. (1) = 1st elections, (2) = 2nd and 3rd elections, (3) = 4th + elections.

Appendix B

411

(Continued) Regime change

Political institutions Parliamentarism

Majoritarian election rules

Municipal elections State elections Subnational elections Legislative weakness

Personalism Conflict data Vertical threats

Horizontal threats

Armed rebellion

Societal warfare

Registers regime trajectories between the current election and the year of the next regularly scheduled election. The possibilities are “continuity” (no regime change), “democratization” (a transition to democracy), “hegemonic opening” (a transition from hegemonic to competitive authoritarianism), and “interruption” (the abortion of the authoritarian election cycle through civil war, military coup, or mandate extensions by more than half of the original constitutional term). Parliamentary system of government. 0 = presidential system, 1 = semi-presidential system, 2 = parliamentary system. From Gerring, Thacker, and Moreno (2005). Replication dataset available at . Majoritarian electoral system. 0 = proportional representation, 1 = mixed-member majority or bloc vote, 2 = majoritarian electoral system. Inversion of “Proportional representation” from Gerring, Thacker, and Moreno (2005). Municipal governments are locally elected. 0 = No, 1 = Yes. Source: World Bank, Dataset on Political Institutions (dpi). State/province governments are locally elected. 0 = No, 1 = Yes. Source: dpi. Either municipal or state governments are locally elected. 0 = No, 1 = Yes. Source: dpi. Ineffectiveness of national legislatures. Range 0–2, with 0 = effective legislature, 1 = partially effective legislature, and 2 = ineffective (“rubber stamp”) legislature. Inversion of Legislative Effectiveness (variable s22F4) from Arthur Banks Cross-National Time–Series Data Archive (cnts) (). Regimes classified as either personalist or hybrid-personalist by Barbara Geddes (Dataset on Political Regimes). Sum of demonstrations, strikes, riots in the three years previous to the election year (p3) or in the three years following it (f3). Source: Arthur Banks Cross-National Time–Series Data Archive (cnts) (). Sum of government crises, purges, and military coups in the three years previous to the election year (p3) or in the three years following it (f3). Source: cnts. Sum of guerrilla wars and revolutions in the three years previous to the election year (p3) or in the three years following it (f3). Source: cnts. Levels of internal violent conflict. Continuous range 0–4. Average of annual magnitudes of revolutionary war and ethnic war, from State Failure Dataset, Political Instability Task Force, George Mason University () (see Marshall, Gurr, and Harff 2001).

(Continued)

412

Appendix B

(Continued) Social structure Population

Rural population Literacy Illiteracy Inequality

Linguistic fractionalization (Fl) Religious fractionalization (Fr) Linguistic polarization

Religious polarization

Phones

Military personnel Economic performance Wealth

Mid-year estimations of country population size (1,000). Missing data replaced by closest previous data. Source: United Nations Statistics Division, Demographic Yearbook (). Rural population as percentage of total population. Source: World Bank, World Development Indicators 2004 (wdi). Adult literacy rate (% people aged 15 and above). Source: wdi. Adult illiteracy rate (% people aged 15 and above). Source: Author calculations on the basis of wdi. Income distribution inequality (Gini coefficient). Missing data imputed through available data in closest available year. Source: wdi. 1 − ∑ l2 where li is the population share of the lth language group (Rae index). Source: Alesina et al. (2003). 1 − ∑ r2 where ri is the population share of the rth religious group (Rae index). Source: Alesina et al. (2003). | [|(Fl–.05) * 2|–1] | In Rae fractionalization indices, values of 0.5 denote situations of bipolarity between two groups (“polarization”) which differ from one-group dominance (Rae = close to 0) and multi-group dispersion (Rae = close to 1). If we subtract 0.5 from the Rae index of linguistic fractionalization Fl, and multiply the absolute value of the result by 2, we obtain an index of linguistic non-polarization, which I invert in order to obtain my index of polarization. | [|(Fr–.05) * 2|–1] | Anaologous to linguistic polarization. In Rae fractionalization indices, values of 0.5 denote situations of bipolarity between two groups (“polarization”) which differ from one-group dominance (Rae = close to 0) and multi-group dispersion (Rae = close to 1). If we subtract 0.5 from the Rae index of religious fractionalization Fr, and multiply the absolute value of the result by 2, we obtain an index of religious non-polarization, which I invert in order to obtain my index of polarization. Number of mainline and cell phone subscribers (per 1,000 people). Source: World Bank, World Development Indicators 2004 (wdi). Military personnel (% total labor force). Source: wdi.

Average monetary income per capita (in current Purchasing Power Parities ppp). Either as natural logarithm or in 1,000 ppp. Source: wdi.

Appendix B Growth

Inflation

Tax revenue External linkages Trade fdi oda Debt service Oil exports Oil dependency

413

Annual changes in average per capita income (election year). I also calculate average growth rates for the 5 years previous to election years (as long as data are complete or near-compete, with no more than one missing year). Source: wdi. Annual changes in consumer prices (election year). I also calculate average inflation rates for the 5 years previous to election years (as long as data are complete or near-complete data, with no more than one missing year). Source: wdi. Tax revenue (% gdp). Source: wdi. International trade (% gdp). Source: wdi. Foreign direct investment, net inflows (% gdp). Source: wdi. Official development assistance (% gni). Source: wdi. Total external debt service (% gni). Source: wdi. Relative level of mineral fuel exports (as percentage of all merchandise exports). Source: wdi. Mineral fuel export dependency (> 25% total exports). Dummy: 0 = non-oil-dependent, 1 = oil-dependent country. Source: Author calculations on the basis of wdi.

This page intentionally left blank

APPENDIX C

Coding Manipulation and Protest ELECTORAL FRAUD General Definition I define electoral fraud in a narrow manner as the manipulation of electoral administration for partisan advantage. It may take place at any stage of the electoral process (before, during, and after election day): • Voter registration and voter identification. • Preparation: distribution and location of polling places, appointment of polling station personnel, design, procurement and distribution of polling material. • Polling: access to polling stations, voting procedures, secrecy and integrity, polling observation. • Counting and vote tabulation. Categories • No fraud: No reports on irregularities or fraud; or only “minor” irregularities, excusable administrative imperfections (score 0). • Irregularities: Sporadic and unsystematic irregularities, relevant, yet not decisive, either partisan or administrative in origin (score 1). • Fraud: Widespread and systematic partisan interference with the organization of elections, even if not decisive; or decisive interference, even if not massive (score 2). The Scale and Impact of Fraud In their elections reports, observers usually conflate two issues one would ideally like to keep apart: the extent of authoritarian manipulation and the impact of manipulation. I do so as well. My category of fraud involves all instances in which election fraud is either big or decisive. Large-scale fraud

416

Appendix C

counts as fraud even if it did not affect outcomes. Small-scale fraud counts as fraud only if its reports license the conclusion that it affected outcomes in a significant manner. The category thus includes big fraud regardless of impact as well as decisive fraud regardless of scale. Decisive fraud: I take electoral manipulation to have a “decisive” impact on (a) legislative elections, when opposition parties would have won a majority or blocking minority in the legislature; and (b) presidential elections, when the winner would have failed to win either the first or the second round of the contest (in ballotage systems). Divergent judgments: In coding the scale as well as the impact of fraud, I rely on reported judgments by election observers as well as opposition parties. If they diverge, yet do not take opposing positions, I discount the statements of opposition parties. If they publicize opposite claims about the quality of elections, I give some credit to dissenting opposition complaints by “averaging” conflicting judgments (in some rough, judgmental way). Examples: In the 2001 presidential elections in Chad, observers delivered “positive reports,” while the opposition alleged “massive fraud.” In the 1993 presidential elections in Gabon, citizens took to rioting, and violent attacks took place against international observers who had issued benevolent judgments on the election. Both cases went into the intermediate category of “irregularities.” Judicial Remedies To the extent that vote rigging is corrected by competent authorities (judicial recourse), it does not count as fraud. An election is also coded as clean if competent authorities order repeat elections due to allegations of fraud and no further complaints arise afterwards. Rule of exception: Judges may well intervene in favor of authoritarian rulers, rather than upholding democratic rights. For instance, in the 1994 legislative elections in Togo, courts, alleging “opposition fraud,” ordered repeat elections for three constituencies that were crucial for the opposition to achieve a majority in the 81-member assembly. As a result of the repeat elections, the opposition lost its prospective majority. Since international observers were complacent, I coded the election as a case of “irregularities.” N-Round Elections An election is coded as fraudulent if fraud is present at any stage of the election. For instance, if ballot rigging seems decisive in the first round of presidential elections, but not in the second round (as in 2000 in Russia), the election is still coded as fraudulent.

Appendix C

417

E XC LU S I O N General definition: Exclusion is defined as formal exclusion from electoral participation of active parties and candidates, regardless of their electoral relevance (through constitutional provisions, legal rules, or administrative practices). Examples: Denial of registry on political or administrative grounds. Ad hominem rules of candidacy that exclude prominent opposition actors. Categories: • Inclusionary elections (score 0) • Exclusionary elections (score 1) Legislative elections: The binary category of exclusion fits presidential elections better than legislative elections. In legislative elections, the wholesale exclusion of parties as well as the widespread (more than sporadic) exclusion of “uncomfortable” individual candidates qualify as exclusion. Barring individual party leaders from legislative contests does not count, as in Côte d’Ivoire in 1995. Specific Coding Rules Exclusion of latent actors z exclusion: I register the exclusion of actors who are already established and active on the political scene. Formal bans on political parties that mobilize certain cleavages (like region, religion, or ethnicity) do not count as exclusion as long as they are successful in preventing the formation of such parties. Example: the constitutional ban on religious parties in Mexico. Exclusion of incumbents z exclusion: I assume exclusion to be a strategy of ruling parties. Bans on sitting presidents from re-election do not count as exclusion. Example: In Paraguay in 1992, a fraction of the ruling Colorado party allied with the opposition to introduce a constitutional ban on re-election, thus preventing President Andrés Rodríguez from running again. Exclusion of violent actors z exclusion: The exclusion of violent actors may be regarded legitimate from a democratic point of view. I do not code bans on violent actors, like former coup mongers, guerrilla organizations, regional warlords, and violent secessionist movements, as authoritarian strategies of exclusion. Example: In Guatemala in 1990, neither the military veto to left-wing guerrilla participation nor the legal barring of former dictator Rios Montt count as exclusion. Exclusion of authoritarian predecessors = exclusion: By contrast, the collective exclusion of members of the former authoritarian regime is counted as exclusion, unless they openly resort to violence. Example: the ban on members of the Duvalier regime in the 1990 elections in Haiti.

418

Appendix C

Exclusion by intimidation z exclusion: The withdrawal of parties or their refusal to register candidacies because of intimidation and fear (deterrence) does not count as exclusion. If made explicit, such “voluntary” non-participation counts as boycott. Otherwise it should show in repression data. Assassination by regime actors = exclusion: To prevent dissidents from competing in elections, electoral autocracies usually resort to legal forms. If state agents or regime-sponsored private agents kill opposition candidates, governments commonly deny responsibility. Still, in case of reasonable doubt, these assassinations should count as exclusion. Example: the assassination of Georgian opposition leader Gula Chanturia in 1994. Assassination by non-regime actors z exclusion: As I strive to capture exclusionary policies by regime actors (or their agents), the assassination of opposition candidates by opposition actors does not count as exclusion. Example: the murder of presidential candidate Abdelmajid Benhadid by Islamists in Algeria’s 1995 presidential election. Open replacement z exclusion: Bans on parties do not count as exclusion if successor parties take their place under different names yet similar programs. The same holds for allied personalities who substitute excluded candidates. Assumption: Citizens recognize successors as reliable representatives of excluded actors. Examples: Greek Albanians in 1992. Ghana in 1992: although pre-1981 parties were formally banned, their representatives nevertheless took part in the elections. Between 1998 and 2001, Turkey went through a succession of moderate Islamic parties: banned in 1998, the Welfare Party was replaced by the Virtue Party for the 1999 elections, which was banned in 2001 and replaced by the Justice and Development Party for the 2002 elections. The ironic case of 1990 Romania: Opposition parties demanded that the Communist party be banned and that the ruling nsf abstain from participating in the first post-communist elections. The nsf banned the communists to win the election itself. As the nsf constituted a veiled communist successor party, the (transient) ban on its own predecessor is not coded as exclusion. The rule of replacement does not apply to murdered candidates (or potential candidates) in presidential contests. In their absence these persons are taken to be irreplaceable. Example: The assassination of Georgian opposition leader Gula Chanturia in 1994 counts as exclusion (for the 1995 concurrent elections), even though his wife succeeded him in the presidency of his National Democratic Party. Proxy representation z exclusion: Similarly, banning some parties that represent certain interests or constituencies, while admitting others that represent the same interests or constituencies, does not count as exclusion. Example: In 1999, the Turkish Supreme Court dissolved the Democratic Party of

Appendix C

419

the Masses, a moderate pro-Kurdish party, while the pro-Kurdish People’s Democracy Party was able to take part in the election.

O P P O S IT I O N B OYC OT T General Definition Election boycott is the public refusal of an opposition party to participate in a multiparty election from which it is not formally excluded. Categories: • Participation: All major opposition parties participate (score 0). • Partial boycott: Some, but not all, major opposition parties boycott (score 1). • Full boycott: All major opposition parties boycott (score 2). Specific Coding Rules The identification of “major” opposition parties: If parties identified as “major” opposition parties withdraw, a boycott is coded as “full” boycott, even if minor parties or candidates stay in the game. Regime-sponsored opposition candidates and satellite parties do not count as “major” opposition forces, nor do genuine opposition parties that are too small to be “relevant” in the Sartorian sense (coalition and blackmail potential). Example: In the 1993 presidential elections in Togo, all renowned opposition candidates boycotted. Subsequently, the two nominal opposition candidates who remained in the game gathered a bare 3.51 percent of the vote. Independent candidates who boycott legislative races are also excluded from my registry of opposition boycotts. Official party positions: My measure grants primacy to official party positions. If individual candidates run, even though their parties call for a boycott, I register the boycott. If parties participate, while individual citizens or civic associations call for boycotts, I do not register a boycott. Non-participation of opposition parties for other motives than protest z boycott: If opposition parties renounce from entering electoral competition out of reasons which are unrelated to the democratic integrity of the election, their non-participation does not count as boycott. The failure of opposition parties to put up candidates out of weakness (as in the 1995 legislative elections in Zimbabwe) does not count as boycott. Nor does their strategic decision to abstain from fielding candidates in certain constituencies. Example: In the 1991, 1997, and 2001 legislative elections in Singapore, opposition parties were

420

Appendix C

over-cautious to appear non-threatening. They decided to field candidates for less than half of the seats to be filled, “thus guaranteeing the government an absolute majority even before voting took place” (Keesing’s World Archive, January 1991). Voter boycott = pre-electoral protest: I distinguish party boycotts from voter boycotts. The category of “opposition boycott” registers the withdrawal of opposition parties and candidates from a given election. Their logical subsequent calls on voters to boycott elections through abstention count as pre-electoral protest. The same applies to boycott calls by excluded parties and candidates. Election day boycott = active post-electoral protest: Boycotts declared any time before election day are registered as boycotts. Boycotts declared on election day count as active post-electoral protest. Second-round boycott = active post-electoral protest: In an analogous manner, if parties participate in the first round of elections, but then boycott the second round (or any further rounds) in protest over the conduct of the first round (or previous ones), I code their withdrawal as “active” post-electoral protest. Boycott by subnational secessionists z boycott: Election boycotts by territorial units or subnational authorities that aspire to secede from a country (separatist forces in nationalist conflicts) do not count as opposition boycott. Example: the boycott of the 1994 legislative elections in Moldova by the separatist Dniester region. Boycott by violent opposition actors z boycott: The refusal of violent opposition actors to participate in electoral processes does not count as boycott. Only if insurgents found a political party, with or without previous demilitarization, and their party then boycotts, I code their refusal to engage in electoral politics as boycott. Example: After its secession bid and the 1994 civil war, Yemen’s ysp returned to politics and subsequently boycotted the 1997 election.

PRE-ELECTORAL PROTEST Electoral protest = public protest by opposition actors against the integrity of national elections. Pre-electoral protest: during the six months preceding the election. Post-electoral protest: during the three months following the election. Categories: • Acquiescence: Opposition parties or candidates acquiesce to an upcoming election. They may or may not criticize prevailing rules or conditions of electoral competition and electoral governance. But they do not mobi-

Appendix C

421

lize their followers in protest against the electoral process. The category also includes opposition criticism (without mobilization) of top election authorities, or legal recourse against specific decisions (not general rules) by electoral authorities (score 0). • Active protest: Mobilization of followers by opposition parties or candidates in protest against upcoming elections (e.g. through public demonstrations, street blockades, strikes). The category also includes opposition calls on voters to boycott elections as well as legal recourse by opposition parties against the rules of electoral competition, against the general framework of electoral governance, or against electoral authorities (initiation of impeachment processes) (score 1). Specific Coding Rules Partial challenges = acquiescence: When opposition parties challenge specific parts of the electoral system only, not general rules or conditions of electoral competition or electoral governance, their criticism counts as acquiescence. Example: Protest against election dates (the strategic timing of election timetables by incumbents) (as has been occurring in most Turkish elections included in database). Only protest mobilization against the timing of first elections counts as “active protest.” In first elections, timing is crucial for opposition parties to reach the electorate. In subsequent elections, the relevance of the electoral schedule diminishes. Organized violence z protest: The category of “active protest” includes acts of civil resistance, such as hunger strikes, tax strikes, sit-ins, and street blockades. Acts of (apparently) decentralized, spontaneous violence, such as riots or violent clashes between government and opposition supporters (as in the Zimbabwe general elections of 1985), count as “active protest” only if opposition supporters protest the framework of electoral competition or governance. Instances of organized violence, by contrast, remain excluded to avoid double counting (they should be captured by cnts data on armed rebellion). Guerrilla campaigns, peasant rebellions, and acts of political terrorism thus do not count as pre-electoral protest, even if they are explicitly directed against elections. Examples: guerrilla violence in El Salvador in 1984 or Khmer Rouge efforts to disrupt elections in Cambodia 1993, or calls on voters to stay away from elections, as in Guatemala in 1990 by the left-wing guerrillas or in Algeria in 1997 by the banned fis. Election day protest = post-electoral protest: The dividing line between pre- and post-electoral protest blurs when opposition parties protest events on election day. As it cannot affect the preparation of the election anymore, election-day protest is coded as post-electoral protest.

422

Appendix C

Second-round protest = post-electoral protest: Opposition protest against the second (n) rounds of elections after the realization of the first round counts as post-electoral protest.

POST-ELECTORAL PROTEST • Acquiescence: Either explicit or tacit acceptance of defeat by losing parties or candidates, without public criticism of the electoral process. No more than low-profile criticism of irregularities. Includes also formal or de facto concession of defeat, albeit with complaints about non-decisive irregularities (score 0). • Rejection: The claim that results are falsified and thus fail to reflect the will of the electorate. Rhetorical rejection: public complaints that elections were undemocratic, that manipulation was decisive, that irregularities invalidated results. Demands for annulment. Judicial recourse: appeals to domestic or international courts. Symbolic protest: boycott of presidential inauguration or inaugural session of the parliament (score 0.5). • Active protest: Mobilization of followers in protest against election results, for instance, through public demonstrations, civic resistance, occupation of public buildings, street blockades, boycott of legislative assembly, or the spontaneous outbreak to violence (score 1). Specific Coding Rules Objectives and carriers of protest: Only protests directed against the electoral process and electoral outcomes count as post-electoral protests. Only contentious actions by opposition parties, candidates, party activists, sympathizers, or voters count as “opposition protest.” Protests by other actors directed towards other goals do not count as post-electoral protest. Partial rejection = acquiescence: The rejection of partial election results counts as acquiescence, rather than rejection. Example: demands for annulment or judicial recourse with respect to only “some” or “several” districts or seats in a parliamentary election (as in the 1995 elections in Croatia). The same applies to complaints in presidential elections that focus only on certain problematic areas (such as Albanian villages in the 1999 presidential elections in Macedonia). Exceptions: Partial objections do count as “rejection” when majority balances are at stake (in legislative elections) or when the first- or second-round victory of a candidate is at stake (in presidential elections).

Appendix C

423

Organized violence z protest: As in the case of pre-electoral protest, acts of organized violence by armed opposition actors do not count as protest actions. The same is true for coups or coup attempts by the military. By contrast, riots and other forms of (apparently) spontaneous violence attributed to “angry voters” (as in the 1990 legislative elections in Gabon) do count as active opposition protest. After boycott: Parties that refuse to participate in an election by implication refuse to recognize its results. Thus, if opposition parties boycott an election (either partially or fully) I code their post-electoral stance as “rejection” even in the absence of further information on their post-electoral behavior. If they mobilize their followers in protest, I follow established coding rules and register “active protest.”

S O U RC E S A N D C OD I N G P RO C ESS Like most similar measurement efforts, I built the Dataset on Authoritarian Elections through content analysis of international news sources and election observer reports. For all elections, I systematically revised four main sources: • • • •

online Keesing’s Record of World Events (); historic archive of the Spanish newspaper El Pais (); section “Election Watch” of Journal of Democracy; and individual-country narratives from the annual Freedom House surveys of political rights and civil liberties (print versions).

Depending on their online availability, I complemented these primary sources by election reports from half a dozen international election-monitoring organizations: • osce Office for Democratic Institutions and Human Rights (); • Commonwealth Election Law and Observer Group (); • Organization of American States (); • Carter Center (); • International Republican Institute (); • National Democratic Institute (); and • International Foundation for Election Systems (). The coding process was conceived as a mixture of content analysis and expert assessments. Content analysis assumes that coder identities do not matter. It

424

Appendix C

conceives coding as a rule-guided process in which coders are exposed to identical pieces of information, apply identical rules of data processing, and therefore move in identical ways from concrete materials to abstract categories (be they numerical or linguistic). The main challenge to the reliable measurement of authoritarian strategies across a large number of electoral regimes worldwide does not lie in the design of workable coding rules, but in the access to factual information (within the constraints of time, resources, and language proficiency that limit any individual research project). Notoriously, news sources based in Western democracies, like the ones I relied upon for this study, provide highly unequal coverage of elections in developing countries. International observer reports allow us to fill in some gaps, but not all. All variables were coded by two independent coders. In disciplines such as psychology and media research, standard operating procedures of content analysis demand that eventual differences between coders be settled by random decision. Given the undeniable insufficiencies of my documentary base, I introduced two elements of expert assessment instead: 1. Rather than arbitrating conflicting coder assessments by chance (through the random selection of scores from independent coders), I discussed divergent coding decisions with coders and then trusted my own judgment to take the final decision (deliberation and expert judgment). 2. For the purpose of external validation, I asked regional experts to evaluate my data. Dialogue with comparative political scientists specialized in different world regions led to a handful of corrections (due to additional factual information), in particular with respect to elections in sub-Saharan Africa. I thank Judith G. Kelley (Eastern Europe), M. Steven Fish (former Soviet Union), Ellen Lust-Okar (Middle East and Northern Africa), Staffan I. Lindberg (sub-Saharan Africa), and William Case (South and Southeast Asia) for their most valuable expert judgment. Of course, all responsibilities remain mine. Documentation: News reports and observer reports are available at request from the author. Individual election files that permit to establish a one-to-one relation between factual information and original coder decisions are available in FileMaker Pro 6 (.fp5 files).

I N T ER NA L A N D E XT E R NA L R E LI AB I L I T Y To ensure “internal” measurement reliability, I conducted multiple coding. To check “external” consistency, I compared my data to similar measures in emerging cross-national datasets.

Appendix C

425

Table C.1. Internal measurement reliability: agreement among coders

Regime manipulation Electoral fraud Exclusion Opposition protest Boycott Pre-electoral protest Post-electoral protest

Latin America

Other world regions

75.9 ...

65.9 81.5

89.7 89.7 79.3

86.5 82.0 75.2

Internal Reliability Content analysis for my data on electoral manipulation and opposition boycott was conducted by two independent coders: the author and two trained graduate students, one covering Latin America, the other all other regions. Although it is “open to debate” what “constitutes an acceptable level of intercoder reliability” (Neuendorf 2002: 143), a reasonable rule of thumb establishes that coder agreement should be higher than 70 percent (Neuendorf 2002: 143). Our percentages of intercoder agreement seem acceptably close to that threshold. In the coding of electoral fraud, intercoder agreement lay at 75.9 percent for Latin America, and at 65.9 percent for the rest of the world. As I measure electoral fraud as a trichotomous ordinal variable, the percentage of agreement is a demanding criterion of reliability, as any divergence among coders counts as disagreement, regardless of the distance between their judgments. With a level of agreement of 81.5 percent, intercoder reliability for opposition exclusion also reached acceptable levels. In the coding of opposition boycott, intercoder agreement consistently surpassed the conventional threshold of 70 percent as well (see Table C.1). External Reliability The development of cross-national data largely lies in the hands of private providers (mainly individual scholars and non-governmental organizations) who proceed in an uncoordinated, decentralized fashion. In this unregulated field of private data production, it is not uncommon to see scholars collecting similar data in a simultaneous fashion without mutual coordination or even knowledge. Even when their data intend to measure the same broad concepts, it is often difficult to compare them or even fuse them into integrative datasets. Being similar to one another, they are not identical. They often differ in their spatial and temporal coverage. More importantly, they usually differ in their methodological micro-choices: their formal definitions, their measurement techniques, their choice of primary sources, their units of analysis, their operational rules, their measurement scales, their coding procedures, and

426

Appendix C

their publication formats. The result is an inefficient duplication and even multiplication of measurement efforts (see Schedler 2012a). The generation of data on authoritarian elections has been a typical instance of information and coordination failures within the competitive market of private data development. When I started developing my dataset on authoritarian elections, minimally trustworthy cross-national data on the key political phenomena I wished to measure (electoral competitiveness, regime manipulation, and opposition protest) simply did not exist. In the meantime, numerous individual scholars and research teams have been developing comparative datasets on election outcomes, the integrity of elections, and the behavior of opposition parties. Most of them are limited in geographic, temporal, and substantive coverage, although two recent datasets are global in scope (iaep and nelda). Most of them diverge (in more or less subtle forms) in their general definitions and specific coding rules. Some are transparent in their sources and operational procedures, others less so. Some are public, others have not been released yet. Eventually, the scholarly community will need to think about possibilities of making these datasets mutually intelligible and integrating them in a manner that preserves their strengths and corrects their limitations. In the meantime, I limit myself to cross-validate my own data by checking their “external” consistency with roughly comparable data that have emerged over the past years (comparative data files are available from the author at request; for an overview, see Table C.2). The consistency of fraud data. The most prominent measure of electoral fraud to date has been provided by the World Bank Dataset on Political Institutions (dpi) (). It indicates whether “vote fraud or candidate intimidation were serious enough to affect the outcome of elections.” Although a number of comparative studies have used this variable (e.g. Hyde and Beaulieu 2009, Simpser 2004), it suffers from significant conceptual, methodological, and empirical flaws (see Schedler 2009c: 286–387). Conceptually, it fuses two dimensions that should analytically be kept apart: electoral fraud and candidate intimidation. Methodologically, it obviates basic requirements of transparency, as it does not reveal its sources, definitions, coding rules, and coding procedures. Empirically, the dpi measure contains numerous scores that raise serious doubts about its validity. It registers several instances of incisive fraud or intimidation that do not coincide with election dates. Most countries maintain their scores between elections, although some scores change even if no national-level elections were held (as in Colombia in 1999). In other cases, records of fraud and intimidation start one year after first elections took place (as in Nicaragua, Ghana, Kenya, and Senegal). The World Bank dataset also contains numerous false positives, that is, records of decisive fraud and intimidation in cases many observers deem fundamentally clean (like in Panama after 1990, Nicaragua

Table C.2. External measurement reliability: agreement among datasets Dataset

Coverage

Fraud Categories

Hartlyn, McCoy, and Mustillo (2008) Bratton and van de Walle (1997) and Bratton (1998) Lindberg (2006b)

World Bank Database on Political Institutions (dpi) Institutions and Elections Project (iaep) Binghamton University National Elections across Democracy and Autocracy (nelda)

Latin America, 1990s

Election quality: acceptable / flawed / unacceptable Sub-Saharan Free and fair elections: Africa, 1990s yes / no

Sub-Saharan Elections are free Africa, 1990s and fair: Entirely or somewhat / irregularities affect outcome / not at all Worldwide, Fraud and intimidation 1975–97 affect outcome: yes / no

Exclusion Agreement

Categories

Western monitors alleged significant vote-fraud: yes / no.

Note: P = Presidential elections, L = Legislative elections

Agreement Categories

Protest Agreement Categories

96.0 % (N = 25) 72.2 % (N = 18)

Boycott / participation

81.8 % (N = 18)

48.7 % (N = 74)

Full / partial / no boycott

73.4 % (N = 64)

Acceptance and criticism / rejection Acceptance / nonacceptance of results Acceptance / nonacceptance of results

Agreement 86.6 % (N = 25) 71.9 % (N = 18)

68.4 % (N = 64)

63.3 % (N = 149)

Worldwide, 1972–2005

Worldwide, 1960–2006

Boycott

69.6 % P (N = 56) 80.2 % (N = 81)

Opposition leaders prevented from running: yes / no

Boycott by major 75.7 % P party: yes / no (N = 70) 80.7 % L (N = 119) 79.8 % P Some opposition 79.2 % P (N = 72) leaders (N = 72) 82.6% boycotted 60.5 % L (N = election: yes (N = 119) 121) / no

84.0 %P Election provokes (N = 70) protest or 73.7 % L violence: yes / no (N = 118) Riots and protests 79.5 % P (N = 73) after the election: 79.3 % L yes / no (N = 121)

428

Appendix C

in 1990, and Ghana in 1996). Even more frequent are false negatives, that is, records of electoral integrity in regimes most observers would hold to be perpetrators of either fraud or intimidation or both. The long list of rather evident examples comprises Mexico in 1988, Peru in 2000, communist Poland and Romania, the Soviet Union, Iran, Iraq, Syria, Ghana, and Kenya in 1992, Zimbabwe in the 1990s, Indonesia, Pakistan, and Singapore. The degree of correspondence between my measure of electoral fraud (in a dichotomous version that collapses irregularities and fraud) and the dpi measure is very low. Due to the high number of elections without fraud, our overall agreement looks better than random (63.3 percent). Yet, within my universe of cases (N =149), dpi registers 47 instances of fraud and intimidation “serious enough to affect the outcome of elections” while I register 36 instances of decisive fraud. We agree in only 14 of these cases, no more than about one-third of our positive observations. Other recent datasets on the incidence of electoral fraud are of better quality, though most of them are limited in geographic, temporal, and substantive coverage. Jonathan Hartlyn, Jennifer McCoy, and Thomas Mustillo (2008) assess the integrity of presidential elections in Latin America since the 1990s (N = 25). They evaluate international observer reports, asking whether observers “accept,” “criticize,” or “reject” an electoral process. Assuming their categories of “acceptable,” “flawed,” or “unacceptable” elections to be roughly equivalent to mine (integrity, irregularities, and fraud), I found almost perfect agreement between our data (96 percent). Our only case of disagreement is the 1993 presidential election in Paraguay, in which I observe irregularities, while Hartlyn, McCoy, and Mustillo observe acceptant observers. Michael Bratton and Nicolas van de Walle (1997) offer dichotomous evaluations of the “free and fair” nature of first or “founding” presidential elections in sub-Saharan Africa, extended by Bratton (1998) to second elections. If I merge my categories of “irregularities” and “fraud” and take them as equivalent to their “not free and fair” category, our judgments coincide in 13 out of 18 cases (72.2 percent). In a more comprehensive effort, Staffan Lindberg (2006b) assesses the “free and fair” nature of both presidential and legislative elections in sub-Saharan Africa from 1989 to 2004 (updated through 2007 in Lindberg 2009c). He uses four categories: elections that were “entirely” or “somewhat” free and fair (which I take to be equivalent to my elections without fraud), elections in which “irregularities affected the outcome” (which I take as equivalent to my elections with irregularities), and elections that were “not at all” free and fair (which I treat as equivalent to my fraudulent elections). Our corresponding level agreement is rather low (48.7 percent). While we identify a similar number of acceptable elections (about one-quarter of all), Lindberg classifies

Appendix C

429

most of the remaining elections as intermediate cases of irregularities (70.3 percent), while I find a higher incidence of fraud (23.0 percent). Susan Hyde and Nikolay Marinov (2012) developed a comprehensive global dataset on National Elections across Democracy and Autocracy (nelda) (). Their dataset contains a categorical measure of election fraud: “Were there allegations by Western monitors of significant vote-fraud?” (variable 47). Its level of agreement with my dichotomous measure of fraud (that merges integrity and irregularities) lies at 69.9 percent for presidential fraud (N = 56) and 80.2 percent for legislative fraud (N = 81). To anticipate: nelda also contains a binary measure of exclusion: “Were opposition leaders prevented from running?” (variable 13). Here, our agreement lies at 79.8 percent for presidential elections (N = 72) and 60.5 percent for legislative elections (N = 119). Our low level of convergence in identifying legislative exclusion reflects the inherent difficulties in categorizing such a graded phenomenon (that contains infinite possibilities of “low-intensity exclusion” targeted against individual candidates and minor parties that lie below the radar of international press attention). The consistency of exclusion data. Two widely used cross-national datasets, the Polity IV dataset on Political Regime Characteristics and Transitions () as well as the Arthur Banks Cross-National Time–Series Data Archive (cnts) (), contain measures of political exclusion. Deriving its theoretical inspiration from Harry Eckstein’s work on authority patterns, the Polity dataset measures three complex dimensions of political systems: the access to executive power (institutionalization, competitiveness, and openness), the exercise of executive power (limitations on governmental decision-making), and the nature of political contestation (cleavage structures, the institutionalization and breadth of opposition). Data users tend to ignore the theoretical foundations of the Polity project and use either composite indicators of institutionalized democracy (democ) and autocracy (autoc) or combined Polity scores (polity) as simple measures of democracy, although their component parts as well as their rules of aggregation have been subject to grave methodological criticism (see especially Munck and Verkuilen 2002). In my view, the only way of preserving the validity of Polity data is through a careful analysis of its disaggregate measures. For the purpose of determining the inclusiveness of political regimes, two of its variables deserve consideration: the regulation of participation (parreg) and the competitiveness of participation (parcomp). • Regulation of participation, although treated by Polity authors as an ordinal variable, in fact represents a categorical variable that captures various dimensions of political contestation, such as the structure, stability, and depth of

430

Appendix C

political cleavages. One of its categories denotes “restricted” participation, a situation in which “significant groups, issues, and/or types of conventional participation are regularly excluded from the political process” (Marshall and Jaggers 2002: 25). • In a similar manner, Competitiveness of participation is a multidimensional categorical variable that combines information about the inclusiveness of political regimes with information about the nature of political competition. While its category of “repressed” competition denotes closed authoritarian regimes with a “demonstrated ability to repress oppositional competition” (Marshall and Jaggers 2002: 25), the category of “suppressed” competition refers to somewhat more open situations in which the regime allows for opposition activities, but “systematically and sharply limits its form, extent, or both in ways that exclude substantial groups (20 percent or more of the adult population) from participation.” Under suppressed competition, “large classes of people, groups, or types of peaceful political competition are continuously excluded from the political process.” The “banning of a political party which received more than 10 percent of the vote in a recent national election,” the prohibition of “some kinds of political organization [or] some kinds of political action,” and the systematic “harassment of political opposition” count as evidence of suppressed competition (Marshall and Jaggers 2002: 26). For the purpose of assessing the inclusiveness of electoral competition, both the parreg category of “restricted competition” and the parcomp category of “suppressed competition” suffer from excessive breadth. Both include too many items to serve the narrow purpose of measuring the legal, administrative, or judicial exclusion of political parties and candidates from the electoral arena. Both categories go well beyond my focus on the electoral arena, as they register the exclusion not only of political parties, but of societal groups, political issues, and forms of protest as well. No less importantly, the parcomp category of suppressed competition conflates exclusion and repression. The Cross-National Time–Series Data Archive registers the exclusion of political parties under the somewhat misleading title of “party legitimacy” (variable S19F6). It captures four broad situations (see cnts Codebook): • Non-party or hegemonic party regimes in which either “no parties” exist or only a “dominant party” and its “satellites” are allowed to compete (score 0). • Authoritarian regimes that practice the “significant exclusion of parties (or groups)” (score 1). • The democratically justifiable exclusion of “minor or ‘extremist’ parties” (score 2). • Fully inclusionary regimes in which “no parties” are excluded (score 3).

Appendix C

431

Given its leaner conception, this measure seems more appropriate for my current purpose than the much broader Polity variables. Still, as cnts does not reveal its definitions, coding rules, coding procedures, and sources, neither the reliability nor the replicability of its data can be taken for granted. After concluding my own coding of electoral exclusion, I double-checked my data with the Polity and cnts measures. Since my criteria were more permissive (more sensitive) than those employed by Polity and cnts, I left those cases without changes in which I had coded an election as exclusionary, while neither Polity nor cnts did. By contrast, in those cases I had classified as inclusionary, while cnts coded it otherwise, I accepted the cnts scores if, and only if, at least one of the Polity categories parreg and parcomp coincided with cnts. Through this quasi-majoritarian rule of correction, I recoded twelve cases as exclusionary I had originally classified as inclusionary: Panama 1984, Belarus 2001, Kyrgyzstan 1991 and 1995, Burkina Faso 1991, Gambia 2001 and 2002, Singapore, 1980, 1984, 1988, 1991, and 1997. Before this corrective step, the degree of agreement between my exclusion data and Polity parreg, Polity parcomp, and cnts Party Legitimacy lay at 67.8 percent, 60.9 percent, and 67.2 percent, respectively. After the correction, it lay slightly higher, at 71.1 percent, 67.6 percent, and 73.8 percent, respectively. The consistency of protest data. To assess the legitimacy of Latin American national elections in the 1990s, Hartlyn, McCoy, and Mustillo (2003) ask whether political parties “accept,” “criticize,” or “reject” an electoral process. To allow comparison, I dichotomized the data, assuming that my category of post-electoral acquiescence includes their categories of acceptance and criticism (all recoded 0), while their category of rejection covers both my categories of rejection and active protest (all recoded 1). These recodified binary data show a high level of agreement (13 out of 15 cases or 86.6 percent). Our only two cases of disagreement are the 1994 concurrent election in Mexico (which I code as protest, Hartlyn et al. as acceptance) and the 1989 concurrent election in Paraguay (which I code as rejection and Hartlyn et al. as acceptance). Bratton and van de Walle (1997), Bratton (1998), and Lindberg (2006b) offer data on opposition protest in sub-Saharan Africa. Again, our categories are not identical but with some reshuffling can be rendered roughly comparable. As Bratton contains a binary variable of opposition boycott (yes/ no), I dichotomize all boycott measures accordingly (combining partial and full boycotts). Lindberg’s trichotomous categorization of boycotts coincide with mine. Our levels of agreement range from 73.4 percent (Lindberg and Schedler) to 81.8 percent (Bratton and Schedler). With respect to the acceptance of election results by losers, Bratton and van de Walle offer a dichotomous measure (yes/no). Among the two polar possibilities of immediate acceptance by all parties and immediate rejection

432

Appendix C

by all parties, Lindberg offers an intermediate category (the delayed acceptance by all parties or the immediate acceptance by some parties). Recoding his intermediate category as post-electoral acquiescence and using a dichotomous version of my own post-electoral conflict measure (by collapsing my categories of acquiescence and rhetorical rejection) yields 68.4 percent of agreement between Lindberg and Schedler and 71.9 percent between Bratton and Schedler. The more recent Institutions and Elections Project (iaep) led by Patrick Regan and David Clark at Binghamton University offers worldwide data on political institutions and practices (). Its elections dataset offers dichotomous measures of electoral boycott and post-electoral protest. The former indicate whether national elections were “boycotted by a major party,” the latter whether the “election outcome provoked protest or violence.” Both differentiate between presidential and legislative elections. Their level of agreement with my dichotomous measure of boycott (that merges partial and full boycotts) lies at 75.7 percent for presidential boycotts (N = 70) and 80.7 percent for legislative boycotts (N = 119). Disagreements go both ways. I count 11 presidential boycotts (15.7 percent of all cases) and 14 legislative boycotts (11.8 percent) the iaep data do not register, while iaep counts 6 presidential boycotts (8.6 percent) and 9 legislative boycotts (7.6 percent) I do not register. The level of agreement between iaep and my dichotomous measure of post-electoral protest (that merges acquiescence and rejection) lies at 84.0 percent for presidential protest (N = 70) and 73.7 percent for legislative protest (N = 118). Again, disagreements cut both ways. I count 8 presidential protests (11.4 percent of all cases) and 20 legislative protests (16.9 percent) the iaep data do not register, while iaep counts 3 presidential boycotts (4.3 percent) and 11 legislative boycotts (9.3 percent) I do not register. The National Elections across Democracy and Autocracy (nelda) contains categorical measures on opposition boycott (variable 14: “Did some opposition leaders boycott the election?”) and opposition protest (variable 29: “Were there riots and protests after the election?”). Their level of agreement with my dichotomous measure of boycott (that merges partial and full boycotts) lies at 79.2 percent for presidential boycott (N = 72) and 82.6 percent for legislative boycott (N = 121). For opposition protest, our agreement lies at 79.5 percent for presidential elections (N = 73) and 79.3 percent for legislative elections (N = 121). Considering that we all looked at somewhat different factual sources through somewhat different conceptual lenses, the “external reliability” of my data seems to be reasonably high. With some startling exceptions, our levels of agreement hover around 70–80 percent. Still, though percentages of agreement are a relatively demanding standard for trichotomous data, they are a

Appendix C

433

rather soft criterion of reliability for dichotomous measures (since a simple random distribution should yield a 50 percent agreement). Besides, whether measures are categorical or ordinal, the simple fact that between about 10 and 30 percent of our data do not coincide with precision is disturbing and bound to carry substantive implications for descriptive and causal inference.

This page intentionally left blank

APPENDIX D

Descriptive Statistics Variables Hegemonic regimes N

Min.

Presidential election Concurrent election Post-Cold War

57 57 57

0 0 0

1 1 1

Regime manipulation Legislative exclusion Presidential exclusion Exclusion Exclusion (previous election) Legislative fraud Presidential fraud Fraud Fraud (previous election) Repression (election year) Repression (pre-election year) Repression (previous election) Censorship (election year) Censorship (pre-election year) Cumulative index of manipulation Censorship (previous election)

48 22 57 34 48 22 57 33 56 56 38 57 56 55 39

0 0 0 0 0 0 0 0 0 0 0

1 1 1 1 2 2 2 2 8 7 8 2.0 2.0 3.88 2.0

.5 .5 .38 .5

Max.

T tests

Competitive regimes

Mean

SD

.39 .23 .72

.49 .42 .45

.58 .32 .54 .56 .87 1.05 .89 .88 3.55 3.36 3.55 1.35 1.34 2.11 1.38

.49 .47 .50 .50 .67 .72 .72 .74 1.87 1.87 1.89 .45 .42 .77 .47

< < >

< <