Party-System Collapse: The Roots of Crisis in Peru and Venezuela 9780804783927

Citation preview

PA R T Y- S Y S T E M C O L L A P S E

PA R T Y- S Y S T E M C O L L A P S E The Roots of Crisis in Peru and Venezuela

Jason Seawright

Stanford University Press Stanford, California

Stanford University Press Stanford, California ©2012 by the Board of Trustees of the Leland Stanford Junior University. All rights reserved. No part of this book may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying and recording, or in any information storage or retrieval system without the prior written permission of Stanford University Press. Printed in the United States of America on acid-free, archival-quality paper Library of Congress Cataloging-in-Publication Data Seawright, Jason, author. Party-system collapse : the roots of crisis in Peru and Venezuela / Jason Seawright. pages cm Includes bibliographical references and index. ISBN 978-0-8047-8236-4 (cloth : alk. paper) 1. Political parties—Peru. 2. Political parties—Venezuela. 3. Voting— Peru. 4. Voting—Venezuela. 5. Elections—Peru. 6. Elections—Venezuela. 7. Peru—Politics and government—1980- 8. Venezuela—Politics and government—1974-1999. I. Title. JL3498.A1S43 2012 324.285—dc23 2012016317 Typeset by Bruce Lundquist in 10.5/15 Adobe Garamond

CON TEN TS



List of Figures and Tables

vii

Acknowledgments

ix



1. Party-System Collapse in South America

1



2. Characterizing Party-System Changes

32



3. Economics, Societal Crisis, and Anxiety

63

4. Corruption and the Collapse of Party Identification

88

5. Ideological Underrepresentation and Voter Defection

113

6. Voter Affect and the Demise of Party Systems

144

7. Explaining Parties’ Degree of Ideological Flexibility

165

8. Collapse and the Experience of Politics

201

Notes

245

Bibliography

259

Index

275

F I G U R E S A N D TA B L E S

FIGURES

1.1 Causes of party-system collapse: overall structure of the argument

5

1.2 Voters and party-system collapse: refining the argument

12

2.1 Traditional party presidential vote share, 1962–1998

54

4.1 Identification with the traditional parties in Venezuela

92

4.2 Identification with the traditional parties in Peru

93

4.3 Venezuelan party identification by ideology in 1983 and 1993

96

5.1 Venezuelans’ ideologies, 1998

115

5.2 Peruvians’ ideologies, 1990

115

5.3 Argentines’ ideologies, 1996

116

6.1 Affect and vote choice

157

TA B L E S

3.1. Economic performance of Latin American countries, 1980–2000

71

3.2. Economic models of Latin American elections, 1980–2001

77

viii

figures and tables

3.3. The economy and satisfaction with household financial situations

82

3.4. The economy and anxiety about the future

83

3.5. Anxiety and satisfaction with the political system

84

4.1. Corruption in Latin American countries, 1983–2001

99

4.2. 1993 model of Venezuelan traditional party identification

103

4.3. 1992 model of Peruvian party identification

106

4.4. Party-branded patronage

110

5.1. Survey and official vote shares in Venezuela’s 1993 elections

125

5.2. 1993 model of Venezuelan presidential vote choice

125

5.3. Survey vote shares in Venezuela’s 1998 elections

131

5.4. 1998 model of Venezuelan presidential vote choice

132

5.5. 1989 model of Lima municipal vote choice

138

6.1. Effects of anger treatment on outsider voting, as mediated through attitudes toward risk

160

6.2. Situations and predicted emotional responses

162

7.1. Average ideological distance between local leaders and the national party

186

7.2. Summary of party organizational traits

192

7.3. Regression of local/national ideological distance on party organizational traits

196

7.4. Regression of local ideological flexibility on party organizational traits

197

8.1. Citizens’ views of parties and elections

223

8.2. Citizens’ views of government’s role in resolving problems

223

8.3. Conditional differences between collapse and non-collapse countries on efficacy

233

8.4. Efficacy in Venezuela, 1983–2003

235

8.5. Conditional differences between Chavistas and non-Chavistas on efficacy

239

8.6. Conditional differences between anti-Chavistas and non-Chavistas on efficacy

240

A C K N OW L E D G M E N T S

This book is the product of many years’ research, and my efforts have been assisted in various ways by more people than I can count. Even so, let me here thank a number of people and organizations whose support has been invaluable. This project would have been impossible without the love and support of my family: Taryn Nelson-Seawright, Artemis, and Athena. Through fieldwork trips, statistical analysis, and countless revisions of drafts, you have been patient and supportive. I think it is fair to say that I could not have done this without you. Likewise, this book has been fundamentally shaped—as has my intellectual development as a whole—by Henry E. Brady, David Collier, Ruth Berins Collier, and David Freedman. All four have always been generous with their time, challenging with their feedback, and supportive beyond the call of duty. My work on this project in Venezuela, Peru, and Argentina was furthered significantly by help and support from a range of institutions. In particular, Adolfo Vargas and the Political Science Department at the Universidad Simón Bolívar in Venezuela provided suggestions, contacts, and access to survey data. The Social Science Department at the Universidad del Pacífico likewise provided contacts and an institutional home in Peru. The Political Science Department at

x a c k n o w l e d g m e n t s

the Universidad Torcuato di Tella in Argentina provided similar services and offered valuable feedback on an initial presentation of key arguments made in this volume. The associations Alternativa, Arariwa, and Transparencia in Peru provided essential help in carrying out survey and experimental research in that country. Invaluable contributions to the success of the fieldwork for this project, as well as in the overall intellectual framing of the book, were made by numerous people including Julio Cotler, Steve Ellner, Sebastian Etchemendy, Miriam Kornblith, Gustavo Mata, Martin Tanaka, Javier Tantalean Arbulu, Alfredo Torres Uribe, Adolfo Vargas, and my teams of interviewers in Argentina, Peru, and Venezuela. I was fortunate to receive feedback on this project at various stages from a wide range of brilliant scholars, established and emerging. Thanks go, in particular, to Taylor Boas, Jennifer Cyr, Pradeep Chhibber, Michael Coppedge, Henry Dietz, Thad Dunning, Zach Elkins, Sebastian Etchemendy, Natalia ­Ferretti, Daniel Galvin, Edward Gibson, Kenneth F. Greene, Kirk A. Hawkins, Maiah Jaskowski, Diana Kapiszewski, Terry Lynn Karl, Herbert Kitschelt, Loan Le, ­Steven Levitsky, James Mahoney, Scott Mainwaring, Matthew ­Marostica, ­Sebastian Mazzuca, Jana Morgan, Scott Morgenstern, David Myers, James Robinson, Ben Ross Schneider, Kathleen Thelen, participants in the 2004 Workshop on the Analysis of Political Cleavages and Party Competition at Duke University, the 2005 writing workshop in the Political Science Department at the University of California, Berkeley, my fellow students in the 2006 Latin American politics graduate research seminar at the University of California, Berkeley, participants in the 2007 Midwest Regional Workshops on Latin America at the Kellogg Institute for International Studies at Notre Dame, students in the 2009 seminar on comparative political parties at the University of Texas at Austin, participants in the 2010 Political Psychology Workshop at the University of Chicago, participants in the Conference on New Methodologies and Their Applications in Comparative Politics and International Relations, at Princeton University in 2010, students and faculty members in the Department of Political Science at Loyola University, participants in the 2011 Political Parties Working Group at Northwestern University, and discussants and audience members at various conference panel sessions. This material is based on work supported by the National Science Foundation under Grant No. 0418459, as well as a Graduate Research Fellowship. Any opinions, findings, and conclusions or recommendations expressed in this ma-

acknowledgments

xi

terial are those of the author and do not necessarily reflect the views of the National Science Foundation. This research was also supported by an International Dissertation Research Fellowship from the Social Science Research Council, a Research Grant from the University Research Grants Committee at Northwestern University, as well as funding from the University of California, Berkeley. My apologies to those I may have inadvertently omitted in this section; I appreciate your contributions, even so. As always, any mistakes are my own.

ON L I N E A PPE N DI X

An online appendix to this volume can be accessed at www.sup.org/partysystem collapse. The appendix contains additional methodological discussion, supplementary figures, and supporting statistical results related to the arguments of Chapters 2, 3, 4, 6, 7, and 8.

PA R T Y- S Y S T E M C O L L A P S E

chapter 1

P A RT Y- S Y S T E M C O L L A P S E IN SOUTH AMERICA

B

e f o r e t h e 1 9 9 0 s , Venezuela’s two-party system was among the most stable and well-institutionalized party systems in the developing world (Coppedge 1994: 174–77). One of the two traditional parties won every fully democratic presidential election in the country’s history. From the early 1970s through 1988, these traditional parties, in effect, faced no challengers, winning a combined share of at least 85 percent of the presidential vote in 1973, 1978, 1983, and 1988. Over this period, the traditional parties also dominated the legislature. In 1993, however, these established electoral patterns began to change rapidly. Both traditional parties lost roughly half of the support they had enjoyed in the previous presidential elections, and—for the first time in Venezuelan democratic history—the winner of the election was not endorsed by either of the established parties. What began as traditional-party decline in 1993 culminated, in the 1998 presidential elections, in a party-system collapse (Dietz and Myers 2007; Morgan 2007).1 Neither of the two traditional parties was able to get any traction for its selected candidate. One party endorsed a candidate from outside the party system early in the campaign cycle; the other waited until days before the election

2

pa rt y-s ys t e m col l a pse i n sou t h a m er ic a

to throw its support to that same outsider candidate. Thus the election became a contest between two candidates from outside the established party system. Both traditional parties have been electorally marginalized since that election. The same election that saw the collapse of the Venezuelan traditional parties also elevated Hugo Chávez to the presidency. Subsequently, Chávez has departed dramatically from the moderate, pro-U.S. politics that were previously traditional in Venezuela, striking out instead in the direction of a bold, confrontational populist leftism (Hawkins 2011)—an approach that regularly reaches provocative symbolic heights, memorably including the moment when Chávez used a United Nations speech (on September 20, 2006) to characterize U.S. President George W. Bush as the devil (Lapper 2007: 19–20); more substantive moments of provocation include Chávez’s repeated statements that he intended to construct “21st-century Socialism” and remake his country as a “Socialist Republic of Venezuela.” In a country that had once been a leading U.S. ally in Latin America and a model of moderate democracy, the degree of political change represented by these events is breathtaking.2 In Peru during the 1980s, a less established party system also collapsed (Cameron 1994; Tanaka 1998; Dietz and Myers 2007). Three political parties had dominated the Peruvian electoral landscape starting roughly with the 1980 presidential elections. These three parties provided all of the major presidential candidates for the elections of the 1980s. They also controlled most of the seats in the legislature and won most local elections. However, between 1985 and 1990, this three-way party system largely collapsed. From a combined 1985 presidential vote share of 85 percent, the traditional parties fell to a combined 1990 presidential vote share of only 31 percent. Indeed, neither of the two candidates who advanced to the second round of the 1990 Peruvian presidential elections came from a traditional party. In the wake of this 1990 collapse, the Peruvian traditional parties received single-digit vote shares in local and national elections for the rest of the 1990s. During that decade, outsider president Alberto Fujimori instituted a freemarket economic policy, featuring extensive privatizations and a sharp reduction in trade barriers, that substantially departed from the patterns of recent Peruvian economic history. In tandem with these economic reforms, Fujimori launched a military coup that overthrew Peru’s democratic regime and dissolved the sitting Congress. He then held a constitutional convention that refounded

pa rt y-s ys t e m col l a pse i n sou t h a m er ic a 

3

Peruvian democracy on Fujimori’s terms. At the conclusion of a turbulent decade of personalist, anti-party electoral authoritarianism, Fujimori finally lost power in the wake of a fraudulent reelection in 2000 and a series of corruption scandals involving an ally of his: intelligence operative and dirty-tricks specialist Vladimiro Montesinos.3 The rise of Hugo Chávez and of Alberto Fujimori involves many convergent series of events. The personal biography of each leader is relevant, as are the stories of their tactical, ideological, and organizational preparations for electoral victory.4 Yet, the crucial role of these factors notwithstanding, it is difficult to imagine that either man would have won power if the Peruvian or Venezuelan party systems had not been in the process of collapse. If Fujimori or Chávez had faced credible, competitive candidates from established, valued traditional political parties, then they would have faced perhaps insurmountable challenges from voters’ strategic voting calculations, citizens’ loyalty to the existing parties, and the resource and visibility asymmetries associated with major-party status. Party-system collapse significantly reduced those obstacles to outsider victory. Hence, understanding the process of party-system collapse is a vital part of thinking about the political origins of Chávez or Fujimori. Latin America is a notoriously turbulent region for political parties. Among countries where no party-system collapse occurred, net electoral volatility scores—the percentage of the overall vote that changes between two specified elections—for the period from 1982 through 1995 range from a low of 17.7 percent in Uruguay to a high of 64.3 percent in Brazil (Coppedge 2001: 175). Change in a party’s electoral strength is not at all unusual in the region. Furthermore, the experience of debt crises, economic restructuring, and neoliberal reform during the 1980s and 1990s was far from politically placid. Perhaps the Peruvian and Venezuelan party-system collapses were merely typical instances of political instability during Latin America’s neoliberal era? In fact, while party-system change of some kind has indeed been common in the region, party-system collapse has been rare. In some countries, collapse was not an issue because no identifiable party system exists; examples include Ecuador and Panama. In other countries, including Chile and Costa Rica, an established party system was relatively stable through the 1980s and 1990s. Still other countries, Argentina in particular, but also Mexico and Uruguay, have undergone extensive party-system change without experiencing party-system

4

pa rt y-s ys t e m col l a pse i n sou t h a m er ic a

collapse. Thus, even in the context of Latin America during the 1980s and 1990s, the changes observed in Peru and Venezuela stand out as extraordinary.5 How did these party systems collapse? What motivated most Peruvian and Venezuelan voters to abandon the traditional parties and instead vote for outsider candidates and parties? Why did leaders within the established parties not make strategic choices that could preempt voter alienation or bring alienated voters back? Political parties play a central role in processes of democratic representation and often profoundly shape the political experiences of citizens; hence, answers to these questions about party-system collapse are integral to understanding South American politics over the past three decades. More generally, close attention to the processes of party-system collapse in Peru and Venezuela illuminates why countries may violate the widespread expectation of partisan stability. Stability in party systems is predicted by multiple, convergent lines of research. Downsian theory regarding party decision-making predicts a stable partisan offering, down to the level of consistent ideological appeals over time, because party leaders always face the same strategic incentives in their interactions with each other and the electorate: “If the distribution of ideologies in a society’s citizenry remains constant, its political system will move toward a position of equilibrium in which the number of parties and their ideological positions are stable over time” (Downs 1957: 115). Sociological research on party systems posits linkages between parties and fundamental social groups such as classes and religions; party-system stability, then, results from slow rates of change in social structure (Lipset and Rokkan 1967; Bartolini and Mair 1990; see also Wittenberg 2006). Research on voter decision-making supports an expectation of stability in relationships of identification, relationships that either reflect hard-to-change core social identities (Campbell, Converse, Miller, and Stokes 1960; Miller and Shanks 1996; Green, Palmquist, and Schickler 2002) or the heuristic use of long-term information to facilitate short-term decisionmaking (Fiorina 1981). All three of these separate research traditions generate an expectation that party-system change should be gradual, conservative, and rare, an expectation that is only strengthened by the typical contrast in financial and organizational resources between established parties and their upstart, outsider rivals.6 Party-system collapses clearly violate this expectation; understanding their occurrence presents the opportunity to discover the conditions under which much of the established theory regarding political parties breaks down.

pa rt y-s ys t e m col l a pse i n sou t h a m er ic a 

5

1 . 1 E X P L A I N I N G PA RT Y- S Y S T E M C O L L A P S E

Perhaps in part because party-system collapse represents an anomaly from many perspectives in the broader comparative theory of parties and party systems, a number of scholars have offered hypotheses regarding the causes of collapse. These hypotheses invoke a wide range of central causes, including both attributes of electorates and features of party leadership and organization. While the existing explanations are incomplete, and in some instances misleading, many provide useful elements for the construction of this book’s explanatory account, outlined in Figure 1.1. Some scholars account for party-system collapse by reference to features of societies’ social class systems. For example, Roberts, while noting that political divisions in Venezuela during the process of party-system collapse “did not follow strict class lines,” argues that “Chávez’s appeal was especially pronounced among the unorganized subaltern sectors of the population” (Roberts 2003: 55). Thus, while arguing that party-system collapse is not a product of conflict between labor and capital, Roberts nonetheless explains it by reference to the politicization of a growing social polarization between “elites” and “the popular sectors.” In effect, he says, the lines of class cleavage have shifted since the classical populist age in Latin America—but the social and party-system crisis are nonetheless to be understood as caused by class conflicts.7 Cameron offers

f i g u r e 1 . 1 . Causes of party-system collapse: overall structure of the argument

6

pa rt y-s ys t e m col l a pse i n sou t h a m er ic a

a parallel theoretical account, focusing on party-system collapse in Peru. Here, the relevant political cleavage is between the informal sectors and participants in the formal economy: Cameron argues that “[t]he flight from the formal economy and the breakdown of the traditional party system were two sides of the same coin” (1994: 10). Economic informality weakens established party systems, in Cameron’s view, by undermining patterns of party identification, reducing ties of communication and membership between parties and society, and generating a bloc of voters uninterested in traditional ideological appeals. Cameron’s and Roberts’s accounts differ in the details: party-system collapse may be due to informality or to poor, disorganized segments of society more generally; and the mechanism linking social class with partisan developments may involve the politicization of a perhaps latent social cleavage or more direct political and organizational effects. These points notwithstanding, explanations of party-system collapse as caused by class conflict share a key implication: actors from the specified class, rather than intra-class coalitions defined by a universalist ideological position or other shared political attitudes, should provide the central electoral impetus for party-system collapse. If—as is shown in Chapter 5 using survey data—there are no strong class differences in propensity to vote against the traditional parties during the key elections, then collapse must instead be explained by multi-class factors such as ideologies or attitudes shared by citizens who vote against the established party system. Another natural approach is to account for party-system collapse as ultimately caused by citizens who engage in retrospective economic voting against the established parties as a group. Building on the well-known generalization that citizens vote against incumbents who preside over periods of poor economic performance, one might suppose that, if the traditional parties alternated in power through a period of consistent or recurrent economic crisis, voters would eventually turn against the parties as a bloc. Levitsky, for example, offers a version of this hypothesis, in conjunction with a party-organizational account to be discussed below, as an explanation for party-system collapse in Peru (Levitsky 2003: 236–38). A related hypothesis is developed by Corrales (2002). In the context of an argument regarding the causes and consequences of confrontation between presidents and ruling parties during neoliberal reform periods, Corrales highlights Pérez’s 1992 decision to accede to his party’s demands that reform be

pa rt y-s ys t e m col l a pse i n sou t h a m er ic a 

7

abandoned. This decision undermined the credibility of the state’s commitment to neoliberalism, offering “cost-bearing sectors” of Venezuelan society the freedom to oppose reform and undermine the existing political parties. This opportunity structure, Corrales suggests, accounts for “the growth of Causa-R,” the electoral victory of Convergencia in 1993, and indirectly even the political career of Hugo Chávez (Corrales 2002: 157–58). Here, it is not the economic pain produced by the failure to complete neoliberal reform that generates partysystem collapse, but rather the redistributive effects of that reform in combination with elite political turmoil. Nevertheless, this account shares an important feature with the simple retrospective voting approach sketched above. Because reform opponents are identified by the economic costs they suffer owing to new policies, it follows that those voters who have the most intense subjective experience of economic suffering should be at the center of the coalition that brings about party-system collapse. For either version of the hypothesis, the analysis of macro-level data in Chapter 3 and survey data in Chapters 4 and 5 fail to support this hypothesis: some countries passed through devastating economic crises without experiencing party-system collapse, and, in Peru and Venezuela, voters with different views about the economy during the key elections do not differ markedly in their rates of identification with or voting for the traditional parties. However, this does not imply that Corrales’s hypothesis regarding the causes of collapse is altogether unhelpful. The proposed causal connections among perceptions of the economy, redistributive preferences and ideology, and voting behavior all need to hold for the economic-voting hypothesis to be supported. If, instead, we regard redistributive beliefs as somewhat autonomous from experiences of costs due to economic crisis and change, we have the alternative hypothesis that voters who had leftist ideological commitments, and who saw the traditional parties as unresponsive to these preferences, may have served as the driving force behind party-system collapse in Venezuela. This is Morgan’s (2007, 2011) argument, in an analysis focusing on party identification rather than vote choice. The process was driven, on this account, by traditional parties’ “failure to provide adequate substantive and symbolic representation to growing sectors of society” (Morgan 2007: 84), specifically those ideologically situated toward the center and the left of society. The hypothesis that party-system collapse is driven by poor representation is central to my argument regarding

8

pa rt y-s ys t e m col l a pse i n sou t h a m er ic a

Venezuela; a modified version of this hypothesis, focusing on the center and center-right rather than the left, is important for Peru as well. Nonetheless, the ideological representation hypothesis can be improved by paying attention to the role of corruption perceptions and to the causal importance of emotion, discussed later in this chapter. Various scholars, particularly those who study Venezuela, have offered explanations of party-system collapse that highlight the probable importance of corruption and scandals in alienating citizens from the traditional parties. Coppedge, for example, argues that party-system collapse was produced by a widespread sense of “moral outrage” (2005: 311–14), a suggestion I develop further later in this volume. In Coppedge’s view, outrage was produced by a conjunction of economic crisis, corruption, and the traditional parties’ role in shielding corrupt politicians and bureaucrats from prosecution. This hypothesis suggests that the central citizen actors in the process of party-system collapse should be characterized by an interaction of two attitudes: they are both particularly concerned about the state of the economy and especially troubled by problems of corruption. Once again, the analysis of national-level and survey data in Chapters 3 and 5 shows that this expectation is not fully empirically supported; perceptions of corruption alone are not strongly correlated with vote choice; and concerns about corruption in interaction with negative attitudes regarding the performance of the economy is also a weak predictor of the decision by voters to abandon the traditional parties. Thus the roles of outrage and corruption as causes of party-system collapse need some degree of respecification. Let us turn now from voter decision-making to a discussion of partisan elites. Some scholars characterize party-system collapse as the outcome of failed elite coordination or mistaken strategy. For example, Tanaka (1998: 201–35; 2006) focuses on episodes during which congressional leaders of traditional parties adopt strategies for dealing with an outsider president that, in retrospect, are clearly ineffective. For Peru, Tanaka highlights the traditional parties’ opposition to President Alberto Fujimori’s eventually successful neoliberal economic reforms, a position that may have helped marginalize the parties through the next several electoral cycles. Regarding Venezuela, Tanaka focuses on traditional party leaders’ decision to boycott the elections leading to Chávez’s Constitutional Assembly. One may agree with Tanaka that these were counterproductive decisions from traditional party leaders’ points of view and nonetheless note

pa rt y-s ys t e m col l a pse i n sou t h a m er ic a 

9

that they constitute only a partial explanation of party-system collapse. Tanaka focuses on elite interactions after an outsider had already won the presidency, but avoids dealing with interactions between elites and voters during the electoral decline that led to the election of an outsider president—surely, at the very least, a central component of the puzzle of party-system collapse.8 In this book I focus directly on the contribution of elite-voter interactions to partysystem collapse—the elite coordination and strategy problems Tanaka analyzes as elements of the aftermath of collapse rather than causes of that outcome. In this sense, Tanaka’s work and the analysis in this volume may be regarded as complementary. Another body of theory traces traditional party leaders’ strategic failures (however characterized) during the process of collapse to features of party organization. Dietz and Myers (2007), for example, attribute party-system collapse to patterns of either excessive or inadequate party-system institutionalization, an intriguing but difficult-to-operationalize hypothesis. More fine-grained organizational hypotheses may help fill in some of the detail. In his extended discussion of the Argentine Peronist party’s survival through the repeated crises of the 1980s and early 1990s, Levitsky develops a theory of party adaptability as an inverse function of a party’s organizational routinization, or adherence to established rules, procedures, and institutional decision-making structures: “routinization limits the capacity of organization to respond quickly to environmental challenges” (Levitsky 2003: 18). Levitsky’s discussion of routinization has a family resemblance to Kitschelt’s theory of organizational entrenchment as the party-institutional explanation for politicians’ strategic failures during periods of partisan adaptation (Kitschelt 1994: 212–23). However, Kitschelt’s concept is more inclusive, treating large formal membership organizations, extensive patronage, size of the party bureaucracy, and the narrowness of the intra-party ideological distribution as indicators of organizational entrenchment. Levitsky, by contrast, explicitly characterizes mass linkages as a distinct dimension of party organization, focusing on the practical decision-making power of rules and bureaucracies within a party rather than the broader range of organizational issues highlighted by the concept of organizational entrenchment. The analysis of party organizational features in Chapter 7 finds strong support for Levitsky’s proposal to treat party organization at a more nuanced level of conceptualization and measurement, while also suggesting that some

10

pa rt y-s ys t e m col l a pse i n sou t h a m er ic a

dimensions of Kitschelt’s theory which are relatively neglected in Levitsky’s account may nonetheless be important in understanding party leaders’ strategic successes and failures during the periods of crisis that, in Peru and Venezuela, led to collapse. Thus several of the above hypotheses regarding party-system collapse contain ideas that are further developed below. Yet many of them are contradicted by aspects of the data regarding collapse in Peru and Venezuela, as subsequent chapters will show, and none of them explains the decisions of both voters and party leaders during the process of party-system collapse. Instead, each hypothesis proposes explanations at the level of voters’ decisions during elections, at the level of party strategy during periods of crisis, or at an aggregate national level. A fuller explanation of party-system collapse must provide a consistent account at all three of these analytic levels. Specifically, a theory of party-system collapse requires an account of the decision-making process that leads voters to abandon the traditional parties and an account of the factors that prevent party leaders from adjusting their party’s electoral appeals to forestall voter defection. If we adopt the familiar metaphor of democratic elections as a political market, a more complete explanation of party-system collapse must provide a demand-side account—showing how voters came to decide that outsider candidates were preferable to the traditional parties, and a supply-side account—suggesting why the traditional parties failed to anticipate and adjust to voter expectations. This study develops and tests such a joint explanation. That explanation integrates themes from many of the hypotheses mentioned above, as well as some key ideas from political psychology, to create a more complete, multivariate, and empirically rich theoretical narrative of the process of party-system collapse. 1.1.1 Voter Decision-Making and Party-System Collapse Interactions between voters and political leaders are inherently reciprocal in nature. Nonetheless, some point of entry into this process is necessary in order to make sense of the decisions by elites and the masses that created party-system collapse in Peru and Venezuela. Voter decision-making processes are an attractive theoretical starting point because voting is (temporally and causally) the final step that produces party-system collapse (see Figure 1.1). Such a collapse by definition cannot happen unless voters choose outsider candidates over those

pa rt y-s ys t e m col l a pse i n sou t h a m er ic a 

11

from the traditional parties; voters are thus the final link in any causal chain leading to party-system collapse. This book argues that voters abandon party systems because corruption scandals erode patterns of party identification, and because poor ideological representation then provides a motive for turning to outsider candidates. It is useful to start the discussion by asking why party-system collapse is an uncommon event. Why do voters so rarely decide to abandon a country’s traditional parties and support an outsider candidate? Such a decision is at least as inherently risky as supporting a party with a reputation for unreliability (Downs 1957: 105–8; Stokes 2001: 8–9; Morgenstern and Zechmeister 2001). Candidates from outside the established political system typically have little governing experience and often have a scant political reputation against which the credibility of campaign appeals may be evaluated. Because such candidates also usually have weak or nonexistent alliances with legislators and other national politicians, there is a serious risk of political crisis and deadlock if the outsider is elected. Furthermore, voting for a candidate who does not come from an established party carries a strong risk that one’s vote will be wasted. The presidency is, after all, a one-seat office. Hence the risk-averse will face severe strategic-voting pressure against opting for a candidate who does not represent a traditionally winning party (Duverger 1954; Cox 1997). For the relatively small number of voters who are very risk-acceptant, these uncertainties may not be a substantial deterrent to supporting a candidate from outside of the traditional party system, and a single source of dissatisfaction with the traditional party system may suffice to persuade such voters to support candidates from new parties or movements (Morgenstern and Zechmeister 2001).9 However, a party-system collapse cannot be produced entirely by those who are highly risk-acceptant by nature. For the more risk-averse by temperament, either a truly powerful motive or a psychological process that produces a temporary increase in risk acceptance is needed to mitigate the uncertainties associated with a vote for a candidate from a nontraditional party. In fact, as will be discussed below, risk aversion is itself endogenous to the process of party-system collapse; citizens’ attitudes toward the existing parties, and the existing social and political system more generally, may affect their broad attitudes toward the uncertainties associated with change, as suggested in the right-hand links in Figure 1.2.

f i g u r e 1 . 2 . Voters and party-system collapse: refining the argument

pa rt y-s ys t e m col l a pse i n sou t h a m er ic a 

13

What kinds of motives might outweigh the uncertainties and risks of voting for a candidate from outside the traditional party system? A rational-choice approach would suggest that any motive, if held with sufficient fervor, would be sufficient. A single issue about which a voter is particularly passionate, and for which none of the traditional parties’ views is acceptable to the voter, can make the traditional parties costly enough to her that she willingly bears the risk of voting for a nontraditional party. This insight seems valid; even in very established and stable party systems such as that in the United States during the twentieth and twenty-first centuries, socially and politically marginal third parties proliferate, and voters seem to choose those parties for a vast array of reasons (Rosenstone, Behr, and Lazarus 1996). Yet accounting for party-system collapse requires more than explaining the motives of a handful of citizens, disgruntled in a variety of sincere but unusual ways with the existing party system. Collapse is a situation in which nearly all voters decide, over a short period of time, to abandon the traditional political parties. A motive that involves unusual or only narrowly supported goals and values can lead to defections from the traditional party system, but the defections will involve a relatively small minority and thus will not substantially change the strategic-voting situation for the remaining voters. Furthermore, for many or most sources of dissatisfaction that voters may have with an administration, a set of candidates, or even the traditional parties, numerous plausible political strategies are open other than voting for an outsider candidate. For example, consider a risk-averse voter who feels ideologically distanced from all of the traditional-party candidates. Even if a nontraditional candidate appeals ideologically to a voter, uncertainty about that candidate’s electability, competence, honesty, alliances, and even the credibility of her ideological appeal itself are likely to combine to make the outsider candidate less attractive than traditional-party politicians. Similarly, for a risk-averse voter who is unhappy with the incumbent’s economic management, at least two alternatives, less uncertain in comparison with a politician from outside the traditional party system present themselves. She may vote for an opposition party from within the traditional system.10 Alternatively, she may choose to accept the inevitable assurances of the candidate from the incumbent party that the candidate’s new governing team has learned from the mistakes of the past and will offer more competent economic governance.

14

pa rt y-s ys t e m col l a pse i n sou t h a m er ic a

As a final example, consider a risk-averse voter concerned that the incumbent administration is deeply corrupt. While a nontraditional candidate is almost certain to promise a corruption-free administration, candidates from the traditional parties will probably also make such promises. In addition, living as she does in a society that is plagued by repeated corruption scandals—a condition that applies to the South American countries where party-system collapse occurred—our voter will probably expect, based on experience, that even the average politician who claims to be honest is quite corrupt. She may therefore tend to disbelieve both traditional and outsider candidates’ claims and therefore lack a motive for supporting the outsider over the traditional parties. More generally, it is important to bear in mind that, in Hirschman’s (1970) terms, dissatisfied voters have a range of strategies related to voice and loyalty, as well as the option of exit. To explain why voters choose exit, rather than voice or loyalty, the intuitive rational-choice framework in which voters select the party or candidate that provides the highest expected utility is insufficient; the central explanatory problem here involves understanding voters’ utility functions themselves, a question for which rational-choice theory offers few systematic answers. This book develops and tests a complementary perspective on the reasons why voters act in ways that produce party-system collapse, drawing on ideas about how affect and cognition interrelate. A great deal of research supports the hypothesis that emotions are intimately involved in the processes of political evaluation and judgment (e.g., Forgas 2000; Lodge and Taber 2000; Neuman et al. 2007). Affective evaluation of new political information may begin even before specifically rational evaluation takes place, and explicit rational evaluation of political information often results in emotional associations that persist long after the relevant information is forgotten. Citizens’ political thought processes thus have constant access to emotion as an implicit running tally of past political information, a prompt to engage in rational deliberation when most needed, and a heuristic decision rule for determining when to set aside habitual standing political decisions and accept riskier alternatives (Marcus, Neuman, and MacKuen 2000). It seems only sensible to look to emotional considerations in developing a theory of which categories of concerns affect mainstream voters’ decisions to abandon a traditional party system.11

pa rt y-s ys t e m col l a pse i n sou t h a m er ic a 

15

Emotion does not have uniform effects on citizens’ decision-making processes. Negative emotions, in particular, can differ in how they influence decision-making processes (Lerner and Keltner 2000, 2001). Because different negative emotions can affect decision-making, we will be able to specify which categories of affect—and potentially which associated categories of cognition and real-world situations—are most likely to motivate a citizen to abandon the traditional party system during a collapse. A key and well-studied distinction among negative emotions involves the contrast between anger and anxiety/fear. Experimental research has shown that angry individuals form more optimistic assessments of risks and are more risk-acceptant in their decision-making than their anxious or sad counterparts (­Lerner and Keltner 2000, 2001; see also Marcus, Neuman, and MacKuen 2000: 46–64). Anxious individuals, by contrast, gather more political information and base their vote choice more directly on the content of the information to which they have access (Marcus, Neuman, and MacKuen 2000: 80–94; Parker and Isbell 2010). Anger and fear are among the more common negative emotions, and may be developmental primitives from which many other negative emotions emerge (Panksepp 1998: 41–58). Hence, in seeking to understand the origins of citizens’ preference for party-system collapse, our attention should be focused squarely on the causes and consequences of these negative emotional primary colors. Which modes of affect lead citizens to turn their backs on their country’s traditional parties and accept the risks associated with voting for outsiders with no political track record or organization? Both anger and fear likely play central roles at one stage or another of the process, with anger signaling to voters that the risks of supporting an outsider candidate are more than matched by the record of pain inflicted by the traditional parties. Feelings of anxiety among voters are important for party-system collapse, because such feelings are connected with their decision to reject habit, seek new information, and revise standing decisions and commitments. Without such reevaluation, party-system collapse would be extremely unlikely. Instead, voters would probably rely on established habits and political identities as a basis for voting, a decision-making strategy that would reinforce the traditional party system. Hence, widespread anxiety helps set the stage for party-system collapse. The cause of such near-universal anxiety is most often a broad, multi­ faceted societal crisis, as suggested at the left end of Figure 1.2. When such a

16

pa rt y-s ys t e m col l a pse i n sou t h a m er ic a

crisis arises in a society, there is ample motive for a wide range of individuals to open the door to a thoroughgoing reevaluation of their political identities and habitual commitments. The analysis of cross-national survey data in Chapter 3 shows that economic crisis can play a partial role in producing the kind of anxiety that translates into doubts about the viability of the political system; presumably, a larger and more multi-faceted crisis will produce a stronger, more nearly universal sense of fear and doubt. Thus, while economic crisis cannot by itself explain party-system collapse, it can raise the stakes connected with the other issues that serve as more direct causes of collapse. Anxiety may open the door to revisiting settled decisions, but anger plays the decisive role in movements away from the traditional party system, either at the level of voting behavior or at the causally prior level of political identity. Party identification is a particularly important obstacle to party-system collapse; if a society has a substantial number of partisan loyalists, then the traditional parties have a cushion of support that will keep them competitive even during times of difficulty, crisis, and political failure. In both Peru and Venezuela, substantial numbers of voters initially reported party identifications. So it is important to understand how voters come to revise or abandon party loyalties during periods when high anxiety and make major change a possibility. When a society undergoes a persistent run of high-level corruption scandals, the result is a pervasive skepticism about, and hostility toward, politicians. On the one hand, corruption scandals involve by definition situations in which politicians act in favor of their own private interests at the expense of the interests of society as a whole. This aspect of scandals tends to undermine party identification when it functions as the result of stereotypes about the social groups that a given party represents (Green, Palmquist, and Schickler 2002). After all, corrupt politicians—and by inference the parties they belong to— represent only themselves. On the other hand, the positive emotional responses that serve as the engine of the “running tally” account of party identification (Fiorina 1977, 1981, Zechman 1979, Achen 1989, 1992) are also undermined by the anger that citizens quite reasonably experience in the face of serious problems of corruption. For these reasons, politicians’ repeated involvement in corruption scandals is a central contributor to the erosion of identification with traditional parties, and therefore a crucial ingredient of party-system collapse, as shown in the upper line of hypothesized causal linkages in Figure 1.2.

pa rt y-s ys t e m col l a pse i n sou t h a m er ic a 

17

In deciding whether to support a nontraditional party, a voter who has either lost her party identification or never formed one is confronted with the risks of supporting an outsider. What motivates such a voter to accept the uncertainties of supporting a candidate with little or no political track record who represents a party that has little or no politically relevant existence outside of its support for that candidate? As described above, anxiety is experimentally and observationally connected with risk aversion. Thus, feelings of fear and confusion related to a perceived or real decline in a country’s quality of life, in general, or economy, in particular, become an ambiguous influence on decisionmakers. Such anxiety surely motivates voters to seek change in the country’s government. Yet anxiety will also predispose voters to avoid high-risk varieties of change. Because outsider candidates and parties are inevitably high-risk modes of political and social change, anxiety is unlikely to serve as voters’ primary motivation during party-system collapse. Anger, in contrast to anxiety, increases risk-acceptance during decision-making. Therefore, anger may by particularly likely to motivate voters’ final decision to abandon the traditional parties. For collapse to take place, the fear and uncertainty connected with broad societal and economic crisis must be replaced, among a substantial number of citizens, by anger—the emotion with pride of place in accounting for the voter decisions that produce party-system collapse. The central remaining issue regarding the decisions of voters who abandon the traditional parties involves specifying the attitudes, perceptions, and issue positions that will lead them to experience political anger. Scholars have argued that the cognitions most closely associated with lasting political anger are a sense of moral injustice at the hands of specific political actors and a belief that those actors have unjustly inflicted personal harm on the voter in question (Ortony, Clore, and Collins 1988: 146–54; Lazarus 1991; Marcus 2002: 120–24; Lerner and Tiedens 2006: 117). Political discourse in both Peru and Venezuela strongly suggests which attitudes are likely to produce the widespread feelings of political anger necessary to generate party-system collapse. The two most salient and widespread accusations against the traditional parties, in the media and in the rhetoric of their nontraditional competitors, are that the traditional parties were riddled with corruption and that they were failing to represent important constituencies in society. Indeed, these claims merged in Chavez’s repeated assertions that the traditional parties had effectively sold out Vene­

18

pa rt y-s ys t e m col l a pse i n sou t h a m er ic a

zuela’s poor in order to keep more wealth for themselves and their personal networks of corrupt friends and allies, and also in Fujimori’s campaign slogan promising “Honesty, Technology, and Hard Work,” in implied contrast with dishonest and ideologically extreme traditional-party candidates. Perceptions that the traditional political elite is hopelessly corrupt and/or does not represent the voter ideologically are a potent stimulus for political anger and the belief that moral injustice has been perpetrated. The voter with these attitudes believes that her views on the good society are not being heard in government and that the reason is that politicians are too dishonest to care about the unrepresented individual. Thus, personal and social harm has resulted, not by chance or through incompetence or impersonal social forces, but because of the greed and dishonesty of named traditional-party politicians. The anger resulting from these points of view serves as a primary motive for the voters to abandon their party identifications and then take the lead in supporting outsider candidates, actions that result in party-system collapse. This argument is tested somewhat indirectly using survey data in Chapters 4 and 5, and more directly through an experimental design in Chapter 6. This book’s voter-side argument is that party-system collapse is brought about by persistent problems of corruption involving traditional-party politicians and a pattern of underrepresentation of some groups by the traditional party system as a whole. Corruption scandals are usually more of a problem for one party at a time than for the entire party system; collapse of the whole system requires a string of scandals involving each major traditional party. Underrepresentation, by contrast, is an inherently systemic problem: it arises when all of the parties simultaneously fail to speak meaningfully for an important ideological segment of the population (see also Morgan 2011). No one party can, in isolation, bring this pattern about; any movement by one party away from an important ideological constituency could simply be countered by a shift by another traditional party toward that constituency. Collapse of the system becomes a possibility when all of the traditional parties choose to neglect a major segment of the population. Thus, voter behavior during an episode of party-system collapse is produced by a conjunction of system-level and party-level explanatory factors—although the system-level pattern of under­ representation necessary for collapse can itself be accounted for by party-level organizational dynamics.

pa rt y-s ys t e m col l a pse i n sou t h a m er ic a 

19

For the remaining voters, who are not driven by the powerful combination of corruption perceptions and feelings of underrepresentation, it is helpful to distinguish between two distinct strategic situations. In some societies in the process of party-system collapse, the set of citizens who feel ideologically under­ represented may be concentrated toward one end of the spectrum of political beliefs, while in other such societies, the underrepresented may be grouped near the ideological center. When the pattern of underrepresentation is such that those who feel excluded are toward one extreme of the belief spectrum, traditional-party voters face a problem of coordination. Should they strike a bargain with a traditional rival party, supporting that party’s candidate in order to defeat the more extreme insurgent and, in the process, prevent party-system collapse by maintaining the electoral and political strength of at least one traditional party? Such a bargain may present benefits to supporters of all traditional parties, but the benefits are unevenly distributed; supporters of the traditional party that survives benefit more than do supporters of the other parties. Each party’s supporters therefore have an incentive to hold out in hopes that the other traditional parties’ voters will rally around their candidate. This coordination problem may, in the end, be best resolved by the appearance of a second nontraditional candidate representing the same general ideological space as the traditional parties. Established-party supporters can then embrace that nontraditional candidate as a strategy for defeating the outsider who represents the traditionally underrepresented ideological wing, without having to bear the costs of adopting a voting strategy that strongly benefits a traditional rival. Hence, in this political situation, there are strong pressures toward a full-scale party-system collapse. For underrepresented voters who are clustered near the ideological center, the situation is different. Supporters of traditional parties on the two ideological wings clearly have little incentive to coordinate in response to any nontraditional threat; for voters on either wing, the nontraditional candidate is clearly preferable to traditional candidates from the opposite wing. Hence, patterns of elite response to the nontraditional challenge can play a decisive role in determining whether and how party-system collapse occurs (Tanaka 2006). If traditional party elites can co-opt outsider candidates into existing organizations, they may be able to revitalize the appeal of those organizations and move the traditional parties’ ideological profiles toward the vacant center. Party-system collapse be-

20

pa rt y-s ys t e m col l a pse i n sou t h a m er ic a

comes harder to avoid if traditional-party leaders subordinate their organizations to alliances with outsider candidates or use indirect support for outsiders as a weapon against rivals at the other ideological pole. In such circumstances, voters near the center—particularly those whose risk acceptance is activated by anger at the combination of underrepresentation and corruption—are not presented with attractive traditional-party alternatives and are therefore likely to support outsiders, producing party-system collapse. Because of the tactical nature of party-system collapse in this situation, the long-term party-system disruptions produced by such a collapse may be expected to be less extreme than for a collapse toward one of the ideological wings. After all, in this scenario, the established constituencies of the traditional parties continue to exist throughout the process of collapse; these constituencies are simply forced to find temporary alternatives as a response to the shifted coalition structures generated by outsider opposition and the corresponding maneuvers of traditional elites. Hence, if the outsiders’ electoral attractions eventually fade, these voters will be likely to renew their connections with the traditional parties. The discussion up to this point has remained silent on one important issue, an issue that the study in general will largely bracket: how voters decide which nontraditional candidate or candidates to support during a process of partysystem collapse. Certainly, a party-system collapse could not occur if there were no supply of nontraditional candidates. Yet the history of electoral competition in Peru and Venezuela strongly suggests that there is no shortage of outsider candidates; most elections feature at least one potentially viable outsider. When the strategic space for outsider victory emerges through the processes of voter alienation described above, charismatic outsiders are often in ample supply. Voters who have chosen to reject the traditional parties thus need to solve a coordination problem (Cox 1997) regarding the choice of which of the available outsider candidates they should support). In both Peru and Venezuela, there is evidence voters took time to explore alternative outsider candidates before settling on one or two as the electoral focal points for party-system collapse (Cameron 1994: 114–21; Molina 2004: 169–72). Unfortunately, beyond the evidence necessary for some analysis of basic ideological assessments, the data needed to resolve the question of how voters chose to focus on specific outsider candidates in Peru and Venezuela do not exist. So for the purposes of the present study, this issue must be largely disregarded. Instead, the focus is on

pa rt y-s ys t e m col l a pse i n sou t h a m er ic a 

21

the logically prior decision to abandon the traditional parties, and the analysis focuses most closely on the first movers, those who abandon the traditional parties primarily because they are concerned about corruption (expressed by their loss of identification with the traditional parties) and ideological underrepresentation (expressed somewhat more forcefully through their votes against the traditional party system and in favor of outsider candidates). 1.1.2 Party Organizations and Party-System Collapse While the above theory provides an account of why voters abandon traditional party systems, it raises new questions about the decision-making of leaders in the traditional parties themselves. The widespread political anxiety that sets the stage for party-system collapse may often be a product of factors at least partly outside the control of parties, such as societal violence and economic performance. However, the anger that provokes the final break with the system is a product of factors the parties can more directly manage. Party leaders can, and sometimes do, take actions to alter the ideological character of their parties (Gerring 2001); they can adopt strategies, including strict party discipline or expulsion, to distance the party from the reputation effects of corruption scandals (Gillespie and Okruhlik 1991). When the very survival of the party is at stake, some explanation is needed for why traditional party leaders would fail to anticipate voters’ demands for aggressive anti-corruption politics and, especially, for more thorough ideological representation. It is worth noting that, in turning to the role of party organizations in party-system collapse, this theoretical account adopts a more streamlined view of individual decision-making than the affect-and-cognition model adopted in considering the voter side of collapse. Indeed, this section implicitly adopts an informal rational-choice model. This theoretical difference mirrors a fundamental contrast between two kinds of real-world actors. Citizens are primarily people whose attention to politics is intermittent and often secondary to other topics. These individuals usually have relatively little direct control over political outcomes. Because they are political amateurs with comparatively little innate interest in or influence over politics, citizens frequently make their decisions using simple heuristic processes (Popkin 1991; Sniderman, Brody, and Tetlock 1991). By contrast, party leaders are often or even usually political professionals. Their personal livelihood and future career trajectories are intimately related to

22

pa rt y-s ys t e m col l a pse i n sou t h a m er ic a

the flow of political tides. Such individuals have clear, self-interested motives for acquiring large amounts of information about politics and for using that information to engage in cognitive elaboration. In other words, while party leaders are unlikely to be completely rational in their political decision-making, they are substantially more likely to engage in sophisticated decision-making processes using large amounts of information, strategic thinking about the probable future consequences of present actions, and cost-benefit reasoning. For these reasons, while adhering to an unsupplemented rational decision-maker model provides an incomplete understanding of the voter side of party-system collapse, such a simplification can be a helpful way of focusing on the central issues on the party side of collapse. How, then, has party behavior during the process of party-system collapse been explained? Levitsky’s (2003) party-organizational explanations of collapse have been highlighted earlier in this chapter, as have the much more encompassing organizational hypotheses developed by Kitschelt (1994) and Myers and Dietz (2007). Although these hypotheses share the insight that too much party organization can produce strategic constraints of the kind evident in the process of party-system collapse, their details differ. In particular, whereas Myers and Dietz informally adopt a kind of unidimensional conception of party institutionalization and Kitschelt develops a unidimensional concept of organizational entrenchment that incorporates several aspects of party organization, Levitsky argues that party organizational concepts such as institutionalization need to be disaggregated, allowing separate measurement and exploration of potentially distinct causal effects of intra-party institutions and practice. In order to fully account for the patterns of party decision-making evident during the process of collapse, ideas need to be drawn from both sides of this debate. In this study I adopt Levitsky’s suggestion that organizational concepts be treated at a more fine-grained and disaggregated level. However, Levitsky’s agenda of explaining party trajectories in terms of the routinization of those parties’ decision-making procedures is supplemented with analysis of the broader range of organizational features to which Kitschelt and others are attentive. In what follows, special attention is given to avoiding the explanatory temptation to attribute party-system collapse to idiosyncratic individuals. If a handful of powerful politicians were to be responsible for the strategic shortcomings of traditional parties during the process leading up to collapse, then an account based on the

pa rt y-s ys t e m col l a pse i n sou t h a m er ic a 

23

psychological complexities of a few individuals might seem to provide a satisfactory explanation for the party’s role in the collapse. Yet this interpretation is superficial. When one individual manages to implement decisions that effectively destroy a party, scholars must ask not only why that individual chose such a counterproductive strategy, but also why other intra-party actors failed to prevent the strategic blunder. The counterfactual questions involved in this kind of causal inference all but inevitably lead in the direction of a conception of single-leader dominance as a consequence of intra-party political and organizational dynamics. Hence, this study focuses on the intra-party dynamics that enable or constrain the choices of top-level national party leaders, rather than on the actual decisions of those leaders. Beyond these theoretical considerations, cases suggest that multiple leaders, often with divergent support bases within the party apparatus, are usually involved in the intra-party contests leading to collapse. Let us consider the situation in Peru, where it is easy to emphasize the role of individual decision-makers. Each of the parties integral to the Peruvian party system of the 1980s had multiple leaders, and there are moments of strategic decision-making during the process of collapse when those leaders adopted visibly divergent positions, demonstrating that the parties were neither monoliths nor purely the patrimony of a single, supreme leader. On the Peruvian electoral left, Izquierda Unida (the United Left), Alfonso Barrantes Lingón was obviously an influential leader. However, the limits of his leadership authority within the movement became clear in the run-up to the 1990 presidential elections. Barrantes had the strategic goal of moving Izquierda Unida toward the center in order to win the presidency—a maneuver that, if successful, might have helped close the ideological gap that proved critical to Peru’s party-system collapse. Yet a substantial component of leaders within the parties involved in Izquierda Unida was hostile to such an ideological shift. These dissenting leaders eventually won the intra-party ideological contest (Roberts 1998; Cameron 1994). Barrantes, in response, split from IU, and the left finally presented two separate candidates in 1990, dividing the leftist vote and failing to appeal to the center. These developments are a central component of the Peruvian party-system collapse, and they are equally clearly not the product of a single, dominating decision-maker. A broader account that incorporates the internal organizational and political dynamics of Izquierda Unida is necessary.

24

pa rt y-s ys t e m col l a pse i n sou t h a m er ic a

On the traditional Peruvian center-right, the story is similar. The long-standing leader of Acción Popular, Fernando Belaúnde Terry, decided to subordinate his party to Mario Vargas Llosa’s neoliberal and, for the Peruvian electoral context of the time, ideologically extreme right Movimiento Libertad for the 1990 elections. Acción Popular candidates at lower levels on the ticket, and other party leaders and activists, repeatedly raised concerns about the ideological tenor of the movement’s campaign and the degree to which the traditional party was engulfed by anti-party actors. Yet these voices agitating for a shift to occupy the underrepresented Peruvian center and center-right lost their intra-party struggle (Florez 1992; Cameron 1994: 59–76). This organizational subordination of a traditional center-right party to an anti-party-system movement from the right wing freed up the ideological space subsequently taken advantage of by victorious outsider Alberto Fujimori, contributing substantially to partysystem collapse. Furthermore, as with the splintering of Izquierda Unida, it was definitely not the decision of a single actor but involved dynamics between that actor’s supporters, an outsider candidate, and a substantial range of dissenters within the party. Once again, attention to intra-party explanations is needed. Even for the traditional party of the center to center-left, the Alianza Popular Revolucionaria Americana (APRA)—the party for which the most serious case has been made that a single leader, Alan García, was responsible for the miscalculations that resulted in collapse—more than one individual was involved. After all, the party’s presidential candidate in 1990, Luís Alva Castro, had repeated conflicts with Alan García starting at least as early as 1987. These conflicts are often interpreted as involving control over the party after 1990 more than ideological direction, and Alva Castro’s electoral campaign broadly continued the shift toward the ideological left that García had initiated during his presidential term. However, the tension between the two leaders suggests that Alva Castro’s decision-making cannot be regarded as simply an extension of, or compliance with, García’s controlling leadership (Schmidt 1996: 327, 337). A more organizational and political explanation is needed for why at least two fairly independent leaders with the APRA organization made essentially similar decisions to move the party away from the largest mass of underrepresented voters in the Peruvian electorate, thereby helping produce party-system collapse. In every Peruvian traditional party (and certainly in the Venezuelan parties, for which most scholars have found evidence of a compelling role of party

pa rt y-s ys t e m col l a pse i n sou t h a m er ic a 

25

organizational factors in producing party-system collapse), the decisions that led to collapse involve multiple individuals. A broader inquiry into party organizations and intra-party politics is needed to explain party decision-making during the process of party-system collapse. Clearly, many party-organizational variables might help cause the ideological inflexibility that makes party-system collapse a possibility, and Chapter 7 reviews a variety of perspectives. In the face of a large number of competing hypotheses, theoretical reflection about the process of party ideological change helps narrow down the possibilities. In the first place, parties are unlikely to succeed in moving to a new ideological stance if they lack experienced potential candidates who can be credibly associated with that stance. Finding such ideologically diverse highlevel candidates will be easier if the base of activists and local party operatives is itself ideologically varied. A varied activist base can be used to recruit ideologically diverse local candidates, thus eventually producing a multifaceted corps of potential regional- and national-level candidates. Furthermore, a diverse base increases the probability that politicians who differ ideologically from the current stance of the central party will be able to win intra-party contests at the local or regional levels and thereby become candidates, in the process acquiring experience and a higher profile. Hence, diversity of the activist and party-operative base leads to diversity in the pool of potential candidates, which in turn facilitates ideological flexibility; when these two intra-party levels, by contrast, are dominated by a relatively homogeneous cadre of true believers in the party’s current ideological line, the party naturally becomes much less flexible. A second near prerequisite for party ideological flexibility is that the party’s decision-making apparatus be sufficiently flexible that new leadership teams have a meaningful chance of winning control of the party. If the organizational apparatus within the party is not flexible in this sense, political currents supporting new ideological appeals may never have a real opportunity to contest for intra-party power. Alternatively, if figures representing such appeals do contest and win, the party’s organization may be unable to accommodate itself to that development, with the result that the party may fragment internally. Drawing on research from other countries, as well as original survey data regarding party organization, I argue in Chapter 7 that the kind of organizational flexibility within parties needed to allow changes in ideological appeals is caused by two factors: a low degree of organizational complexity of the party’s membership

26

pa rt y-s ys t e m col l a pse i n sou t h a m er ic a

and outreach organization, and a high flow of patronage goods through the party apparatus. In Chapter 4 I tie the latter factor to another important consideration: parties’ ability to avoid the consequences of corruption scandals. A party-system collapse will occur only when (a) the voter motives described in the previous section place pressures on a country’s established parties, and (b) each of those parties has internal constraints that prevent them from effectively responding to voter pressures. In these ways this book integrates both sides of the electoral marketplace into a unified explanation of party-system collapse. 1.1.3 Party-System Collapse and Citizens’ Experience of Politics While it is important to examine the origins of party-system collapse, the consequences of collapse are equally deserving of analytic attention. When an institution that is a central pillar of democratic governance erodes, how does politics continue? Are the perhaps inevitable disruptions major or relatively minor? In sketching an answer to these questions, it is helpful to distinguish between consequences for the political regime and the functioning of elite politics and the consequences for citizens’ experience with the political process. For regimes and elite politics, the consequences of party-system collapse in South America are reasonably well known. Collapse is accompanied, almost axiomatically, by an increase in the political power of anti-party outsiders. These outsiders lack well-institutionalized groupings of elite supporters, at least initially. As a result, they have adopted the centralized, personalistic, and anti-institutional mode of governance variously described as neopopulism (Weyland 1996; Knight 1998; Collier 2001) or as delegative democracy (O’Donnell 1994; see also Cameron and Levitsky 2009). While such a leadership style has deleterious consequences for the independence and stability of political institutions, it is equally clear, although perhaps somewhat less discussed, that this mode of governance tends to polarize elite opinion and behavior. Political elites not in the president’s inner circle are, under such leadership, usually excluded from political power. They often adopt highly conflictual strategies of opposition to the government. This opposition, and the government’s predictable aggressive response to it, can degrade the quality of democracy—if, indeed, democracy survives at all. Because these patterns have been widely studied in both Peru and Venezuela (e.g., Cameron and Mauceri 1997; McCoy and Myers 2004; Carrion 2006), they receive little additional attention here.

pa rt y-s ys t e m col l a pse i n sou t h a m er ic a 

27

About the consequences of party-system collapse for citizens’ experience of politics and government, much less is known. Indeed, a wide range of consequences can be imagined. The experience of citizen power over the elite political world in the process of collapse might lead voters to see politics as far more egalitarian and open to citizen control than it was in the past. Alternatively, patterns of underrepresentation and corruption during the collapse could alienate voters from the political world altogether. Between these two extremes, there is an essentially infinite range of alternative conclusions that citizens could draw about politics from party-system collapse. However, some of those conclusions require more citizen effort and initiative than others and are therefore perhaps less likely as widespread effects of collapse. Drawing on the well-established finding that many or most citizens tend to have relatively little information about politics and seek informational short-cuts to forming attitudes and making decisions in the political sphere (Zaller 1992), we may expect that interpretations of party-system collapse that are readily available for citizen adoption should be empirically the most widespread. Hence, inquiry into the probable effects of collapse on citizens’ experience of and connections with politics turns inevitably to how political elites frame the event of collapse (Chong and Druckman 2007). Because the elites’ explanations will be among the most common and cognitively easiest to understand, they should more widely adopted among citizens than less-available competing explanations. What, then, are the prevalent elite framings of party-system collapse in Peru and Venezuela? Is there a necessary link between the process of collapse and the emergence of those framings, or is the link more contingent? In fact, the process of party-system collapse almost necessarily leads to the proliferation of a frame offered by elites from outside the party system: that the state is and has been incapable of resolving society’s biggest problems. Critiques of the performance of the state, after all, are integral to almost any anti-party-system actor’s electoral appeal: if the state is working well, then there is little reason to take the risks associated with electing a relative political unknown. This anti-state-capacity frame takes on different hues, depending on the ideological orientations of the anti-party system actors. Actors on the right and sometimes in the center are likely to promulgate this critique in its neoliberal version, claiming that all states are inherently less able than the market to resolve many categories of problems. Anti-party-system voices from the left, by contrast, are likely to claim that the state in its current constitution, filled with

28

pa rt y-s ys t e m col l a pse i n sou t h a m er ic a

corrupt bureaucrats and with actors whose interests diverge from those of the masses, either cannot or will not address society’s major problems—and that a reformulated state would be the best solution to such problems. On both sides, anti-party-system discourse is likely to emphasize the claim that the current state cannot solve the most important problems. For inattentive or moderately attentive citizens in particular, the details of that claim and the hypothetical solutions may be less important than the simple message that the state does not work. Hence, we may expect the process of party-system collapse to reduce citizens’ sense of how involved the state should be in a variety of issue areas; if the state cannot fix problems in those areas, why waste time and money trying? Other frames that may arise in the process of party-system collapse are more contingent and depend on the kind of collapse and on a society’s broader ­ideological environment. In particular, different messages will be sent if the ideological underrepresentation that produces collapse is in the center or toward the left or right wing. Collapses of the left are likely to produce mobilizing frames that emphasize the capacity of citizens to understand the political world and to act in ways that definitively change it. After all, such ideas have long been hallmarks of the left (Roberts 1998). Collapses where the anti-party actors are located toward the ideological center are far less likely to feature this kind of extensive mobilizing and capacity-emphasizing rhetoric, and collapses where anti-party actors lean toward the right may even involve explicitly demobilizing frames that stress the complexity of society’s problems and the extent to which they are best addressed through purely private, nonpolitical action. Hence, citizens who have lived through a party-system collapse toward the left, but not those who have experienced other kinds of collapse, should experience an increase in their sense of political efficacy—i.e., their belief that they are capable of understanding and acting to change politics. 1.1.4 Conclusions Party-system collapse is thus the result of an interaction between voters and party leaders. Voters’ decision to abandon the traditional parties and produce a collapse comes at the end of a complicated causal process of economic and broader social crisis that leads to greater political awareness. Voters’ change in party identification due to concerns about corruption and decision to vote for outsider candidates can best be explained by ideological underrepresenta-

pa rt y-s ys t e m col l a pse i n sou t h a m er ic a 

29

tion—and ultimately by anger, which reduces the subjective importance of the risks involved in such a large departure from political convention. Party leaders, for their part, fail to adjust ideologically to voters’ demands largely because of intra-party constraints, which may involve organizational entrenchment or narrow ideological distributions among activists within the party. The effects of collapse for citizens’ ongoing perceptions of the political world are conditional on the kinds of messages about state and citizen capacity that become prevalent during the interactions that constitute collapse. As is usually the case in the social sciences, this study is unable to test every component of the theoretical account above. In particular, systematic data regarding political affect are, to date, rare for Latin America. Given this limitation, the analysis here adopts an eclectic approach, using quantitative analysis of aggregate-level and survey data to test several observable implications of the theory of party-system collapse developed above and relying on experimental evidence to probe the key links among emotions, attitudes toward risk, and willingness to vote for a political outsider. This mixed strategy gives empirical roots to the discussion and provides evidence to support the theory’s key propositions.

1 . 2 O V E RV I E W

The evidence for this argument is developed through an analysis of political dynamics in Peru and Venezuela during the process of party-system collapse, and also by comparing the collapse countries with countries where the party system has survived. Of the countries where the traditional party system has transformed but not collapsed, Argentina is notable because it experienced many of the same pressures as Peru and Venezuela. These episodes of partysystem change are described and situated in comparison with other party-system transformations in Chapter 2. Chapter 3 compares the political and economic crises that form the context for party-system collapse in Peru and Venezuela with the often similar crises in other Latin American democracies during the 1980s and 1990s. It tests the effects of social and economic crises on party system collapse with a time-series cross-sectional analysis of change in the vote share of Latin America’s governing parties between successive presidential elections. While most scholars have argued that economic crisis deserves a central place in the explanation of

30

pa rt y-s ys t e m col l a pse i n sou t h a m er ic a

party-system collapse, this regional analysis shows that the degree of crisis as reported by aggregate statistics for each country is not a complete explanation. In comparison with the rest of Latin America, elections in Peru and Venezuela during the 1980s and 1990s produced vote shares for governing parties that are substantially different than would be expected on the basis of economic performance. However, there is some reason to believe that the broad societal crises in Peru and Venezuela did increase voter anxiety levels and expand citizens’ discomfort with the existing political system, enabling the subsequent process that more actively caused party-system collapse. Chapter 4 explores the reasons why many voters in Peru and Venezuela turned away from their standing political attachments to the traditional party system. In particular, it tests the hypothesis developed above that concerns about corruption cause the decline in voters’ identification with the traditional parties that, in turn, serves as a major causal ingredient of party-system collapse. Using survey data about elections during the process of party-system collapse and a pseudo-panel analysis, the chapter shows that corruption perceptions, rather than economic evaluations or more specific opinions about neoliberal reform, are most responsible for eroding voters’ identification with the traditional parties—although economic crisis remains a potentially important part of the story as a cause of the voter anxiety. That anxiety may set the stage for the observed reevaluation of patterns of party identification on the basis of information about corruption. The chapter concludes by offering empirical clues in favor of the hypothesis that corruption scandals are less damaging to identification with parties that channel a great deal of patronage through the party apparatus. Chapter 5 analyzes the role of ideological underrepresentation in citizens’ ultimate decisions to vote outside of the traditional party system. Using survey data, it shows that gaps in the pattern of ideological representation provided by the traditional parties, in combination with a loss of identification with those parties as analyzed in Chapter 4, explain the decision to defect from the traditional parties and vote for new alternatives. Economic perceptions and social class variables, central elements of competing theories, prove less important. The theory developed above posits a deeper causal pathway, regarding affect and attitudes toward risk, that explains why ideological underrepresentation and corruption play the central causal roles in the voter-side decision-making that leads to party-system collapse. Chapter 6 subjects this pathway to experi-

pa rt y-s ys t e m col l a pse i n sou t h a m er ic a 

31

mental test. In the research discussed there, a sample of Peruvian citizens are randomly assigned to conditions in which their emotional states are modified through viewing a scene from a film and listening to emotionally coordinated music. Then, they participate in a simulated election between a traditionalparty candidate and an outsider. The chapter shows that the manipulation is causally efficacious: anger increases voters’ probability of supporting an outsider candidate. In conjunction with the more general survey findings of the earlier chapters, these experimental results play a key role in empirically supporting the theory discussed in this chapter. Turning to the puzzle of why party leaders made poor strategic decisions when their organizational survival was at stake, Chapter 7 uses data from an original survey of local party leaders to test the theory developed above regarding the specific organizational factors to explain why the Peruvian and Venezuelan traditional parties were less strategically flexible than the Peronist party in Argentina. The analysis discusses a range of organizational dimensions that have been hypothesized to affect parties’ ideological flexibility and provides a descriptive characterization of the traditional parties in all three countries on each dimension. It then reports results from a series of statistical models to determine which dimensions best predict parties’ local degree of flexibility. The evidence supports the hypothesis that organizational entrenchment, low levels of intra-party ideological diversity, and low degrees of channeling patronage through the party apparatus lead to less ideological flexibility and thus indirectly contribute to party-system collapse. Finally, Chapter 8 considers the consequences of party-system collapse for how voters think about government and politics. Using a variety of empirical comparisons, I suggest two major ways that collapse could affect voters’ attitudes. First, the pattern of crisis experienced during the process of collapse combined with the kinds of political appeals and frames that outsider candidates need in order to succeed politically can be expected to reduce voters’ sense of the proper scope for state action. Second, the collapse toward the left in Venezuela, in conjunction with the regional prevalence of participatory ideologies among leftist political outsiders, led to an enhanced sense of political efficacy among citizens in Venezuela—an effect produced by collapse in one context but not necessarily a universal consequence of similar party-system transformations.

chapter 2

CHARACTERIZING PA RT Y- S Y S T E M C H A N G E S

D

i f f e r e n t c o n c e p t u a l i z a t i o n s of party systems and different descriptions of the recent political histories of South American countries can pose quite divergent explanatory puzzles regarding party-system collapse. Some analysts, for instance, might reject Peru as a case of collapse on the grounds that no party system existed there in the first place. Others, including Tanaka (1998, 2006), would push the date for party-system collapse at least a few years into the presidential terms of Alberto Fujimori in Peru or Hugo Chávez in Venezuela. Both of these views would result in explanatory puzzles quite different from that addressed in this volume. Hence, before presenting the evidence for this study’s explanation of party-system collapse, it may be useful to spell out the conceptual and historical considerations supporting my decision to treat the elections won by Fujimori in Peru and Chávez in Venezuela as the culminating moments leading to party-system collapse. In this chapter, the characterization of the traditional parties and party systems of Peru and Venezuela, as well as Argentina, focuses on the intensive interactions between party elites and voters. Such interactions are an essential part of the process of party-system collapse. To highlight the aspects of collapse that involve elite-voter interaction, I use a primarily electoral concept of collapse.

ch a r ac t er iz i ng pa rt y-s ys t e m ch a nge s

33

Other useful conceptions of party-system collapse have, of course, been developed; these serve to highlight different aspects of Peru’s and Venezuela’s broader political, economic, and social transitions. The purpose of this chapter is not to argue that other conceptions are incorrect or inadequate, but rather to demonstrate how elite-voter interactions support the plausibility of the conceptions I adopt here. At the same time, the chapter offers a broad historical narrative of the party-system changes that the rest of this book analyzes in more depth.

2 . 1 T R A D I T I O N A L PA RT I E S

In defining the concept of “traditional parties” as used in this study, it is useful to note that the term “traditional” has an established, and in the current analysis irrelevant, usage in modernization theory (see, e.g., Lerner 1958). In the context of modernization theory, tradition refers to societies that have not adopted the package of industrialization, urbanization, secularization, state formation, democratization, and so forth associated with “modernity.” This meaning is entirely irrelevant to discussions of Latin American party politics during the late twentieth century. By then, most Latin American countries had undergone substantial urbanization (De Oliveira and Roberts 1998: 243–53) and significant industrialization (French-Davis, Munoz, and Palma 1998: 185–88). Furthermore, the existence of a meaningful system of political parties presupposes the establishment of a reasonably democratic political regime (Sartori 1976: 3–13). Hence, in the context of modernization theory, each of the parties under consideration here exists within a largely modern society. For present purposes, the term “traditional” refers instead to those parties that have had the opportunity to develop a vibrant and extensive party tradition within a given electorate. More specifically, these parties have a substantial—i.e., multi-decade—history and have most often been electoral competitors with a serious prospect of forming the national government for multiple electoral cycles. In order for a party to have the chance to develop deep roots in a country’s democratic history, it must usually be relatively old; new or recent parties are not traditional in this sense. But parties may be traditional even if they experience some degree of volatility regarding organizational form or party name. What is most important is that the party unambiguously represent a tradition that is recognized by both voters and party leaders as a long-standing compo-

34

ch a r ac t er iz i ng pa rt y-s ys t e m ch a nge s

nent of the country’s political life. For many parties, this criterion is unambiguous. When ambiguity arises, it may be resolved through an exploration of the biographies of party leaders and the historical and ideological sources the party draws on in formulating its appeals to the population. If most of the party leaders have extensive experience within the same well-defined political movement and draw on prominent ideas, texts, and historical appeals from that movement’s past history, then the party may be regarded as traditional even if its name and current organizational incarnation are of more recent vintage. Parties that persist for a long time at the margins of politics are excluded from the traditional category; while such parties may have a committed core of supporters, they have not had the opportunity to contribute enough to the country’s institutional, electoral, and political history to qualify as traditional. The easiest way to distinguish persistent marginal movements from traditional parties is electoral relevance: traditional parties have often been central to electoral competition, and even serious competitors to form the government, at more than one point during the country’s democratic history. For parties that have been persistently excluded from competition by authoritarian and restricted democratic regimes, these electoral considerations are too restrictive. For such parties, evidence of other forms of connection with the political mainstream, including influence during authoritarian interludes and prominence during regime transitions, can substitute for electoral evidence of relevance. Thus, traditional parties, by definition, have been politically relevant for a substantial period of time and, as a consequence, have had the opportunity to establish firm organizational, political, and sociological ties to the electorate. Exactly because of this relatively high potential for the party to embed itself in society and the electorate, it is especially surprising when a traditional party collapses. 2.1.1 Argentina’s Traditional Parties Which parties, in each country, count as traditional according to this definition? This section gives a brief overview of the traditional parties of Argentina— the country that serves as the major case of party-system survival during crisis throughout the analysis; subsequent sections discuss the traditional parties of Peru and Venezuela. The emphasis is on the two criteria for traditional-party status described above: existence for a significant period of time before the

ch a r ac t er iz i ng pa rt y-s ys t e m ch a nge s

35

process of collapse, and status as a serious political force before the period of party-system collapse. Consequences of the party’s history for the development of the party’s organization and ties to society—hypothesized consequents of traditional-party status—are considered in Chapters 4, 5, and 7. Finally, the organizational and electoral trajectory of each party during the 1980s and 1990s provides context and specificity for the larger discussion of collapse. For each party, the short- and long-term circumstances that coincide with electoral gain and loss are sketched in these narratives. Argentina’s traditional parties are the Peronist party and the Radical party. The Peronists, officially known as the Partido Justicialista or PJ, originated as an electoral movement designed to support the 1946–55 government of Juan Domingo Perón. By the time of the reestablishment of Argentine democracy in 1983, the Peronist party was several decades from the moment of its formation, satisfying the first definitional requirement of traditional parties. The second requirement, which can be fulfilled by demonstrating the existence of multiple rounds of serious electoral competition to form the government, was also satisfied. Peronism won elections and formed the national government during the period from 1946 until 1955 and again in 1973. Between 1955 and 1973, and after 1976, Peronism suffered almost-constant proscription and military repression (McGuire 1997: 80–93, 145–63, 170–84; Collier and Collier 2002: 484–97, 721–42). However, during this period, Peronists won important elections under different party labels, and Perón himself was often able to broker electoral victories by striking alliances with non-Peronist politicians (McGuire 1997: 80–150). The Peronist party that was free to compete in Argentina’s new democratic political regime after democratic transition in 1983 was thus a party that had experienced decades of chaos, illegality, and institutional disruption. After the Peronists were defeated by the Radicals in the presidential election of 1983, dissidents within the party launched a movement, called renewal Peronism, aimed at increasing the transparency of party decision-making procedures and expanding its electoral appeal beyond the traditional base. When the Peronists used a primary election to select a candidate in 1988 (for the first time in party history), the winner, Carlos Menem, was a charismatic, personalist leader opposed to decision-making institutions that might reduce his personal power (O’Donnell 1994; McGuire 1997: 189–90, 207–13).

36

ch a r ac t er iz i ng pa rt y-s ys t e m ch a nge s

Menem governed Argentina from 1989 until 1999, through the period during which Peru’s and Venezuela’s party systems collapsed. During that period, he introduced extensive neoliberal reforms that moved Argentina toward integration into the international market economy. This transformation at first faced some hostility from the Peronist party—which was traditionally associated with a rather interventionist ideology (Corrales 2002: 127–31). However, Menem was able to overcome this party opposition and eventually win the support of the Peronist movement for his reelection in 1995. In the process of winning his party’s support, Menem relied heavily on three factors. First, the strength of labor unions within Peronism was on the decline during the late 1980s and the 1990s. In part, this was due to the renewal Peronism movement, which attempted to limit labor influence on party decisions. This decline in union influence also reflected a shift in party organizational strategy away from unions in the direction of local patronage machines (Levitsky 2003: 107–43). Since the union movement was particularly likely to object to Menem’s economic policy package, the marginalization of this actor within Peronism may have helped Menem regain control of the party. Second, Menem negotiated a kind of political truce with the Peronists by incorporating more party leaders into the government and by shifting toward a view of party leaders in the Congress as potential allies rather than obstacles to be overcome (Corrales 2002: 169–85). Third, Menem’s economic policies in fact won majority support in the Argentine public by mid-1991—which Menem was largely able to maintain until late 1994 (Echegaray and Elordi 2001: 202). The Peronists lost the 1999 election, in which Menem, as a second-term incumbent, was ineligible to run. Instead, the new president, Fernando de la Rúa, was elected as the candidate of a coalition between the Radicals, discussed below, and a nontraditional third party (FREPASO—El Frente País Solidario). However, when that government failed during an economic crisis in late 2001, the Peronists resumed control of the government. In the next presidential election, in 2003, Peronist candidate Néstor Kirchner won the presidency on a moderate leftist platform. The Peronist party thus survived the 1980s and 1990s without suffering any kind of electoral collapse. Argentina’s other traditional party, the Radicals (officially known as the Unión Cívica Radical or UCR), dates back to 1891, clearly meeting the first requirement for traditional-party status. Originally, the Radicals’ primary ob-

ch a r ac t er iz i ng pa rt y-s ys t e m ch a nge s

37

jective was to overcome oligarchic limitations on democracy and fraudulent interventions in elections through a strategy of insurrection and electoral abstention (Rueschemeyer, Stephens, and Stephens 1992: 178–79; McGuire 1995: 204–6). In 1912, electoral reforms had effectively ended conservative control over electoral results, allowing the Radicals to compete seriously to form the government for the first time. Under the first free and fair presidential elections, in 1916, the Radicals won a landslide victory on the basis of party organization, patronage politics, and powerful ties to Argentina’s middle classes (McGuire 1995: 206). The Radicals won two more presidential elections, in 1922 and 1928. However, economic decline associated with the beginning of the Great Depression, as well as conflict with the military, provoked a coup and the initiation of a military government in 1930—which used coercion and fraud to exclude the Radicals from power (McGuire 1995: 207–8). During Perón’s government, between 1946 and 1955, the Radicals were the primary opposition party. However, the Peronist behemoth effectively guaranteed that the Radicals would be excluded from government power until after Perón’s overthrow in 1955. The military overthrow of Perón and Peronism’s subsequent exclusion from democratic participation created electoral opportunities for the Radicals, who largely dominated electoral politics during democratic periods until the 1970s. Between this electoral success and the Radicals’ victories before 1930, the Radical party meets the criterion of seriously competing to form the government in more than one electoral cycle. How did the Radicals fare during the period of party-system collapse in Peru and Venezuela? In the 1983 presidential elections, following a turbulent Peronist government between 1973 and 1976 and a brutally repressive military regime between 1976 and 1983, the Radicals triumphed behind leader Raúl ­Alfonsín and a new, moderate, pro-democratic electoral appeal that contrasted sharply with Peronism’s reliance on divisive, class-political traditional symbols (Munck 1992: 205–7). However, during the economic crisis of the late 1980s, the Peronists once again won the presidency, as noted above. Subsequently, ­Menem’s political and economic success, as well as possible popular dissatisfaction with the Radicals’ cooperation in changing the constitution in 1994 to allow for presidential reelection, led to devastating electoral defeats for the Radicals through the mid-1990s, as the Radicals reached a 1995 low of 16.2 percent of the presidential vote.

38

ch a r ac t er iz i ng pa rt y-s ys t e m ch a nge s

In the 1999 presidential elections, the Radicals formed a coalition with a third party, FREPASO, that had split from the Peronists earlier in the 1990s. This coalition defeated the Peronists, and the Radicals became once again a governing party. However, after a wave of corruption scandals, the governing coalition fractured in 2000; furthermore, a currency crisis in 2001 (Calvo and Mishkin 2003: 100–101) led to a full-scale economic meltdown that forced the Radicals out of government by the end of the year. In the wake of this crisis, the Radical party has seen tensions at the elite and mass levels; in 2003, two of the major presidential candidates had traditional ties to the Radical party, but neither of them ran with the official party label. No other Argentine party qualifies as traditional. Other than the Peronists and the Radicals, FREPASO was the only party to have substantial electoral success or other governmental relevance since 1983. However, FREPASO never ran another independent presidential candidate, participating instead in a coalition with the Radicals in 1999 and effectively disappearing after the 2001 political and economic crisis. 2.1.2 Peru’s Traditional Parties The oldest and perhaps best known of Peru’s traditional parties during the 1980s is the Alianza Popular Revolucionaria Americana (APRA). APRA was founded in 1924 as an anti-oligarchic, indigenist, and anti-imperialist party. Like the Radical party in Argentina, APRA began its history struggling more for inclusion in the political system than for electoral victory. APRA’s founder, Víctor Raúl Haya de la Torre, was exiled from Peru for eight years after sponsoring a 1923 protest against the current dictator’s decision to dedicate the country of Peru to the Sacred Heart of Jesus (Graham 1992: 25). Rather more serious was a violent revolt against the military and the state by APRA activists in northern Peru in 1932, in response to the 1931 presidential elections that APRA claimed were fraudulent (Rojas Samanez 1987: 151–52; Graham 1992: 27–29; Collier and Collier 2002: 152). In response to this revolt, the Peruvian state banned APRA from participation in electoral politics—a ban that was enforced almost uniformly until 1962 and intermittently renewed until roughly 1978 (Cotler 1995: 328). However, government and military opposition notwithstanding, APRA frequently found ways of participating in elections and even competed seriously to form

ch a r ac t er iz i ng pa rt y-s ys t e m ch a nge s

39

the government on more than one occasion. In 1936, APRA struck a secret alliance with incumbent military president Oscar R. Benavides, a bargain that did not in the end result in legal recognition for APRA (Rojas Samanez 1987: 152–53). APRA also offered covert support for the civilian president elected in 1939, Manuel Prado (Collier and Collier 2002: 316). In the 1945 elections, candidate José Luis Bustamante y Rivero formed a coalition, the Frente Democrático Nacional, which demanded the legalization of APRA and subsequently included APRA as a junior coalition member (Rojas Samanez 1987: 153; Collier and Collier 2002: 318–19). Legalization allowed APRA to consolidate ties with the labor union movement and strengthen its nationwide party organization (Collier and Collier 2002: 325–26). However, a 1948 military coup left APRA once again proscribed and out of government. The coup, organized by conservative elites, was in part a response to actions by the APRA party in government, in part an answer to rising APRA electoral success, and also a direct consequence of an APRA decision to launch a violent revolt in a city just outside of Peru’s capital, Lima (Rojas Samanez 1987: 155–57; Collier and Collier 2002: 328–30). For the next eight years, a military regime once again adopted a policy of intense repression against APRA. However, in 1956, presidential elections were held in which APRA—though still illegal—traded its support for ex-president Manuel Prado in exchange for the legalization of the party (Rojas Samanez 1987: 157; Collier and Collier 2002: 473–74). By supporting the Prado government, APRA necessarily moved away from its leftist origins in the direction of centrist ideological appeals, producing a schism in which some leftist activists abandoned the party (Collier and Collier 2002: 477–78). In exchange, however, the party was allowed to participate in the 1962 presidential elections, in which founder Haya de la Torre won a slight plurality. Nevertheless, the military vetoed the possibility of an APRA government and required new elections in 1963, in which APRA lost to Acción Popular (Rojas Samanez 1987: 157–63; Collier and Collier 2002: 697–702). When that Acción Popular government fell to a military coup in 1968, APRA once again faced a military government and formal exclusion from the political system. This time, the military had a leftist political orientation and a desire to implement much of APRA’s traditional political platform (Graham 1992: 37–60). Between the earlier schism within APRA over support for Prado and the subsequent reformist actions of the Peruvian military during the 1970s, the APRA

40

ch a r ac t er iz i ng pa rt y-s ys t e m ch a nge s

party that existed in 1978 when the military called elections for a constituent assembly had lost much of its left wing. That change in constituency, in combination with the dislocations caused by Haya de la Torre’s death, added to the sense of crisis produced by an APRA defeat in the 1980 presidential elections. In response, APRA initiated an intra-party renovation that included a shift in ideology away from the teachings of Haya de la Torre and toward a more pragmatic electoral appeal. The renovation project also involved the emergence of new party leadership, especially Alan García, the young leader who would serve as APRA’s first successful presidential candidate, in 1985 (Graham 1992: 73–96). García’s election was APRA’s high point during the 1980s; within three years, his government had broken down amid economic disaster, Maoist insurgent violence, and policy conflict with the Peruvian economic elite (Graham 1992: 99–125). Following a third-place finish in the 1990 presidential elections, APRA fell to the electoral sidelines throughout the 1990s, never obtaining double-digit results in any national election. This decline serves as a central puzzle throughout the study. After this decade-long collapse, APRA recently experienced a remarkable revival. In both the 2001 and the 2006 Peruvian presidential elections, Alan García was the second-place finisher in the first round. In 2001, García narrowly lost the second round; in 2006, he won the second round and was elected to a new presidential term. This recent renewal for APRA is a helpful reminder that party collapse—even collapse as severe and seemingly total as that suffered by APRA—need not be irreversible or permanent. The second of Peru’s traditional parties, Acción Popular, originated in 1956—once again, several decades before the beginning of party-system collapse in the 1980s. Its founder, the architect Fernando Belaúnde Terry, had served as a legislator from 1945 until 1948, under the label of the Frente Democrático Nacional, in which he was an ally of the APRA party. However, Belaúnde was alienated by the subsequent alliance between APRA and Prado, and he therefore chose to form an alternative movement (Rojas Samanez 1987: 307–9). Belaúnde’s new party, Acción Popular, began with a leftist political program emphasizing economic planning, agrarian reform, education, and Peruvian nationalism (Belaúnde Terry 1960). Under this platform, and with support from the Peruvian military, Belaúnde was elected president in 1963, following a close, second-place finish in 1962.

ch a r ac t er iz i ng pa rt y-s ys t e m ch a nge s

41

These successful electoral showings qualify Acción Popular for traditionalparty status. The Acción Popular government implemented a moderate land reform law in 1964 (Rojas Samanez 1987: 317–18). However, a legislative opposition coalition between APRA and the authoritarian right hampered the government, eventually resulting in a military coup that ousted Belaúnde in 1968. Belaúnde maintained an anti-authoritarian posture throughout the military government, and even refused to allow Acción Popular to participate in the military-­sponsored constitutional convention of 1978 (Rojas Samanez 1987: 320–21). This posture of opposition, combined with Belaúnde’s personal appeal and perhaps other factors, allowed Acción Popular to win the presidential elections of 1980. However, the second Acción Popular government ended in economic turmoil and the rise of a Maoist guerrilla insurgent movement. These negative outcomes, in conjunction with the fact that Peru’s constitution forbade the immediate reelection of Belaúnde, coincided with a sharp decline in Acción Popular’s vote share in the 1985 elections. In the run-up to the 1990 elections, Acción Popular chose to support Mario Vargas Llosa’s outsider presidential campaign rather than run its own candidate (Cameron 1994: 59–76). As with APRA, this electoral collapse is a major part of the outcome this study seeks to explain. During the elections of the 1990s, Acción Popular was entirely unsuccessful and has not achieved the kind of subsequent electoral resurgence that has characterized APRA. The third traditional party in the Peru of the 1980s is a more complicated case. Izquierda Unida, the party of the democratic left during the 1980s in Peru, was not formed until just after the 1980 presidential elections—in which the fractured Peruvian left realized that it paid a steep electoral price for its division (Roberts 1998: 222). However, the various leftist parties and movements that formed this party had deeper roots reaching back into social mobilizations of the 1960s and especially the period of military government in the 1970s (Roberts 1998: 201–17). Thus, even though Izquierda Unida never competed in a presidential election until 1985, it represented a collection of political forces that had played an important role in Peruvian politics for decades by that point, and in that sense Izquierda Unida meets the criterion of meaningful age and especially societal roots at the beginning of the period of party-system collapse. With respect to

42

ch a r ac t er iz i ng pa rt y-s ys t e m ch a nge s

the second criterion for traditional-party status, in place of extensive experience of electoral competition for presidential power, Izquierda Unida had relatively close ties to an unelected military government that had held power for more than a decade. Because of this history of influence, and especially owing to the relatively close connections between the leftist movements that subsequently formed Izquierda Unida and the military government, it would seem inappropriate to exclude Izquierda Unida as a traditional party even though it does not meet the same definitional criteria for that status that the other traditional parties do. This conclusion is reinforced by the important role that the parties which would subsequently form Izquierda Unida played in the 1978 constitutional convention, where these leftist groups were the second-largest bloc with 30 of the 100 total seats (Dietz 1986: 147–48). Izquierda Unida was officially a coalition of leftist political parties. However, for nearly a decade after its founding, Izquierda Unida nominated candidates for national and local elected offices; coordinated campaign efforts; and cooperated, at least to some degree, in legislative politics. These features justify considering Izquierda Unida as a political party in analytic terms. For these reasons, this study henceforth classifies Izquierda Unida as a traditional party. Clearly, the classification decision is a close one, and Izquierda Unida could plausibly be regarded as just missing the criteria for traditional status by the time of the party-system collapse. In either case, it is clear that Izquierda Unida differs in many ways from the parties and movements that emerged during the process of collapse itself: Izquierda Unida consisted largely of experienced political actors, represented parties and ideological traditions with deep roots in Peru’s political society, possessed a clear image and reputation among voters, and had a significant if less than entirely institutionalized organizational base. As such, Izquierda Unida has at least a strong family resemblance to the Peruvian traditional parties. How did Izquierda Unida fare electorally during the 1980s, when the other two Peruvian traditional parties collapsed? Between the founding of Izquierda Unida and its split into two competing parties in 1989, the party obtained between 20 and 30 percent of the vote in the presidential and municipal elections of 1980, 1983, 1985, and 1986, thereby qualifying as a major force in the Peruvian party system of the 1980s. In its first election in 1980, the party won thirteen of Peru’s 188 district mayorships, nine of which were located in the

ch a r ac t er iz i ng pa rt y-s ys t e m ch a nge s

43

shantytowns surrounding the nation’s capital (Roberts 1998: 223). This initial electoral victory created a strategic dilemma that persisted through the 1980s and eventually formed the context for a major schism in the party: should ­Izquierda Unida moderate its ideological appeals in order to win elections, or should it maintain a revolutionary strategy? For most of the 1980s, the pragmatic and revolutionary components of Izquierda Unida were able to coexist, although never harmoniously. Alfonso Barrantes, a moderate and perhaps Izquierda Unida’s most electorally viable representative, was elected mayor of Lima in 1983. Barrantes implemented a range of pragmatic reforms targeted at improving the quality of life and economic position of Lima’s poor (Roberts 1998: 226). This experience served as a springboard for Barrantes’s 1985 presidential campaign, in which he came in second. This successful showing in 1985 raised the possibility of an Izquierda Unida presidential victory in 1990. However, the possibility of victory itself produced a crisis that eventually split the party. Moderates within the party advocated running Barrantes as a presidential candidate once again, and also moderating the party’s platform to appeal to centrist voters (Cameron 1994: 79–85; Roberts 1998: 247–48). Leftists, by contrast, worried that winning the election on a moderate platform might prevent the revolutionary changes in Peruvian society to which they were committed (Roberts 1998: 252–54). In 1989, when this strategic dilemma proved impossible to resolve, Izquierda Unida split into two competing parties, both of which received single-digit vote shares in the 1990 presidential elections and have been electorally marginalized since.1 2.1.3 Venezuela’s Traditional Parties Clearly Venezuela’s largest and most powerful traditional party for most of the country’s democratic history, Acción Democrática was founded in 1941 as the successor organization to a series of anti-authoritarian political movements from the 1920s onward (Martz 1966: 22–48). Hence, this party was at least fifty years from its founding moment when the Venezuelan party-system collapse began. The party was first able to participate in electoral competition after a military coup in 1945, in which Acción Democrática leaders conspired with dissident military officers to displace the existing nondemocratic political regime (Martz 1966: 60–62). A year later, in October 1946, Acción Democrática and the other participants in the military takeover held elections—usually consid-

44

ch a r ac t er iz i ng pa rt y-s ys t e m ch a nge s

ered to have been Venezuela’s first democratic vote—for a constituent assembly. Acción Democrática dominated the elections, winning 78.8 percent of the vote (Martz 1966: 69). A new constitution was promulgated in 1947, with presidential and congressional elections scheduled for December of that year. Once again, Acción Democrática won decisively, receiving 70.8 percent of the congressional vote and 74.4 percent of the presidential vote (Martz 1966: 75). The prospect of perpetual Acción Democrática electoral dominance, in conjunction with the government’s policies in favor of land reform and labor union rights, provoked a military coup against Acción Democrática toward the end of 1948 (Martz 1966: 82–85; Collier and Collier 2002: 268–70). For the next ten years, Acción Democrática was completely excluded from government power and subjected to intense repression (Martz 1966: 89–96; Collier and Collier 2002: 421–45). Both in order to regain its access to governmental power and to forestall future military coups, Acción Democrática decided to moderate its ideological position, shifting toward the political center in a process that resulted in various ideologically motivated group defections from the party throughout the 1960s (Coppedge 1994: 54–56). Furthermore, Acción Democrática signed a pact with all of the major non-Communist parties of Venezuela, called the Pact of Punto Fijo (Kornblith and Levine 1995: 44–45). With these moderating changes in place and with the initiation of a new democratic regime in 1958, Acción Democrática was able to play a leading role in Venezuelan political life. Between 1958 and 1993, Acción Democrática almost always held the largest block of seats in the Venezuelan legislature and won all but two presidential elections (Kornblith and Levine 1995: 49–53). However, in the wake of a major economic crisis, a failed effort at neoliberal economic reforms, a series of corruption scandals, and the impeachment and removal from office of an Acción Democrática president early in 1993 (McCoy and Smith 1995: 252–56), the party began to lose substantial electoral ground. From its 1988 presidential vote share of 52.9 percent, Acción Democrática’s vote fell in 1993 to a mere 23.6 percent. The party’s legislative vote share experienced an equally severe decline. Five years in the political opposition did not reverse this trend; in 1998, the party was forced to abandon its own candidate shortly before the election and endorse a partisan outsider in an unsuccessful effort to defeat leftist, anti-party-system candidate Hugo Chávez Frías. Since 1998, Acción Democrática has remained at the electoral margin. As with the Peruvian

ch a r ac t er iz i ng pa rt y-s ys t e m ch a nge s

45

traditional parties, this collapse of a once-dominant party will be a major focus of attention throughout the remainder of this study. COPEI (Comité de Organización Política Electoral Independiente), the Christian-democratic party that served as the main opposition to Acción Democrática during the period between 1958 and 1993, originated as a party of Catholic protest against the Acción Democrática-dominated government in 1946 (Kornblith and Levine 1995: 46–47; Crisp, Levine, and Molina 2003: 275–76). Hence, COPEI is almost as old as Acción Democrática—and, by the 1980s and 1990s, was several decades from its founding moment. Initially electorally marginal,2 COPEI had a major electoral breakthrough in 1968, when a splinter Acción Democrática candidacy and a strong showing by another party, the URD (Unión Republicana Democrática), created a four-way race that COPEI would win with 29.1 percent of the presidential vote (Coppedge 1994: 56; Crisp, Levine, and Molina 2003: 293). That victory cemented COPEI’s role as the primary alternative to Acción Democrática in the roughly two-party system that dominated Venezuelan electoral politics during the 1970s and 1980s. COPEI therefore meets both criteria for traditional-party status. However, following economic difficulties and a series of corruption scandals during a COPEI government between 1979 and 1984, COPEI began to lose electoral ground. Its 1978 presidential vote share of 46.6 percent stands as an all-time high; in 1983, the party’s vote fell to 33.5 percent, rebounding in 1988 to 40.4 percent before falling in 1993 to 22.7 percent. In the 1993 elections, ­COPEI’s charismatic founding figure, Rafael Caldera, left the party after personal and ideological struggles with other party leaders and launched a successful independent, anti-party-system candidacy; this division in conjunction with voter dissatisfaction may have contributed to COPEI’s electoral decline (Crisp, Levine, and Molina 2003: 296–98). COPEI has not subsequently recovered; in 1998 it was unable to field its own presidential candidate and it has received single-digit vote shares in subsequent elections. The decline of COPEI is the last major party-system change emphasized by this study. Other parties have, of course, existed during Venezuelan democratic history. For example, the URD, mentioned briefly above, was the second-place finisher in the 1958 presidential elections and received nontrivial vote shares in 1963 and 1968. This party had, however, long ceased to occupy a meaningful electoral space in Venezuela by the time of the party-system collapse; in 1988,

46

ch a r ac t er iz i ng pa rt y-s ys t e m ch a nge s

the URD received a mere 0.8 percent of the presidential vote (Kornblith and Levine 1995: 49). Likewise, the leftist party Movimiento al Socialismo is sometimes seen as an important component of the Venezuelan party system.3 Yet this party never achieved a presidential vote share higher than 5.2 percent before the beginning of the party-system collapse in 1993; likewise, it only once obtained a double-digit legislative vote share during that period (Kornblith and Levine 1995: 49–51). Hence, as with the URD and other small Venezuelan parties as well as the PPC in Peru, it seems most reasonable to regard the Movimiento al Socialismo as not meeting the criteria for traditional-party status.

2 . 2 T R A D I T I O N A L PA RT Y S Y S T E M S

The previous section’s overview of the Argentine, Peruvian, and Venezuelan traditional parties makes it clear that the Peruvian and Venezuelan traditional parties have undergone deeper and more prolonged cycles of electoral decline since the 1980s than have the Argentine traditional parties. Yet the focus of this study is not the collapse of traditional parties, but rather the collapse of traditional party systems. What, then, is a traditional party system? For present purposes, a traditional party system is defined as a party system in which the traditional parties manage to dominate electoral competition. If the traditional parties control elections to the point that no other parties appear on the presidential strategic-voting landscape, then the electorally relevant parties constitute a traditional party system. Thus, it is by no means evident that all countries during all democratic periods have a traditional party system. If short-term electoral movements consistently, if transitorily, occupy an important electoral space, then the country in question does not possess a traditional party system. Nonetheless, Argentina and Venezuela during the 1980s almost certainly had traditional party systems according to this definition; Peru’s pattern of electoral competition during the 1980s also seems to meet the definition, although perhaps somewhat less proto­ typically than the other two countries. During the 1980s, the Peronists and the Radicals (Argentina’s traditional parties) captured between 78.7 percent and 89.3 percent of the presidential vote; no other party managed a double-digit vote share during the decade. The traditional parties’ domination of the legislative arena was similar, although some-

ch a r ac t er iz i ng pa rt y-s ys t e m ch a nge s

47

what less pronounced (McGuire 1995: 241–46). It is therefore easy to conclude that only the traditional parties were electorally central enough that they had an opportunity to determine governmental majorities; Argentina’s elections in the 1980s were controlled by a two-party traditional system.4 For Venezuela, the case is perhaps even clearer. During the 1970s and 1980s, the two traditional parties jointly captured between 85.4 percent and 93.3 percent of the presidential vote, as well as between 74.6 percent and 79.5 percent of the legislative vote. During that period, no other party ever received more than 12.9 percent of either the presidential or the legislative vote (Kornblith and Levine 1995: 49–51). In light of this electoral dominance by Acción Democrática and COPEI, Molina and Pérez describe the 1973–88 period as an “attenuated two-party system,” noting that Venezuela during the 1970s and 1980s meets the definitional criteria of a two-party system but also that some small leftist movements persisted at the margins of electoral competition (1998: 11–13). However, these marginal parties did not play any significant electoral role until the 1990s; hence, Venezuela meets the definition of a traditional party system. Peru’s party system of the 1980s was substantially weaker than those of Argentina or Venezuela. The three parties identified as traditional in the previous section, APRA, Acción Popular, and Izquierda Unida, did manage to dominate electoral competition for much of the 1980s (Cotler 1995: 336). As such, they meet this study’s definition of a traditional party system. Nonetheless, both ­Acción Popular and Izquierda Unida were relatively uninstitutionalized parties, as discussed above. Furthermore, electoral competition during the 1980s did not consist of a routinized, three-way struggle among these parties. Instead, competition during the early 1980s was primarily between APRA and Acción Popular; during the mid- to late 1980s, the electoral focus shifted to a contest primarily between APRA and Izquierda Unida. These factors make the Peruvian parties of the 1980s seem less system-like, and have led some scholars to conclude that Peru during the 1980s had an inchoate party system (Mainwaring and Scully 1995b: 19). As an additional argument in favor of this study’s classification of APRA, Acción Popular, and Izquierda Unida as forming a traditional party system, it may be worth noting that the party system did possess some of the traits mentioned above as important, but not definitional, in traditional party systems. In particular, each of these traditional parties did have meaningful ties to specific

48

ch a r ac t er iz i ng pa rt y-s ys t e m ch a nge s

sectors of Peruvian society. Furthermore, as Charles D. Kenney argues, “although this party system never became well institutionalized, its almost complete collapse surprised most observers” (2004: 43). It may, of course, be useful to bear in mind that there were differences among the party systems—in terms of institutionalization, ties to social sectors, and party structure. However, in spite of substantial volatility and organizational weakness, Peru’s electoral competition during the 1980s meets this study’s definitional criteria for the existence of a traditional party system.

2 . 3 P A RT Y- S Y S T E M C O L L A P S E

Having discussed the traditional parties and party systems of Argentina, Peru, and Venezuela, one final conceptual issue remains: what is meant by “partysystem collapse”? For the purposes of this study, a party-system collapse is a situation in which all the parties that constitute the traditional party system simultaneously become electorally irrelevant. Operationally, the requirement that the decline of all parties be simultaneous is taken to mean that the collapse must occur over a period of not more than two electoral cycles. This operationalization excludes episodes of partysystem change in which the major, established parties are gradually replaced by new parties. Rather, attention is focused on moments when all parties suffer at once—a form of change that is more catastrophic and surprising, and hence a more profound violation of the expectation of party-system stability discussed at the beginning of this chapter. In addition, the criterion of electoral irrelevance, central to the idea of party-system collapse, will be regarded as met whenever a party fails to achieve either the first or second place in a presidential election. In a traditional party system with more than two parties, the parties that count as electorally relevant according to this definition will generally rotate from election to election; in two-party systems, the relevant parties are constant across electoral cycles. Presidential elections are emphasized, rather than legislative vote shares, because the president—and not the largest legislative party—forms the government in a presidential system. According to Duverger’s Law, presidential elections with only a single round of balloting will tend to produce results in which only two candidates receive a meaningful vote share (Cox 1997: 69–98); alternatively,

ch a r ac t er iz i ng pa rt y-s ys t e m ch a nge s

49

some political systems explicitly impose extra relevance on the two leading candidates, sending only these two candidates into a second round of balloting.5 Under either institutional system, it therefore seems reasonable to consider parties that fall outside of the first two slots to be less electorally relevant. In both Peru and Venezuela, the electoral decline of the traditional parties continued after the moment that this study identifies as the terminus of the party-system collapse. One might well therefore argue for a later date in both cases, highlighting the coalitional and electoral decisions that led the greatly weakened traditional parties of Peru after 1990 and Venezuela after 1998 to all but disappear in subsequent rounds of political competition. However, for present purposes, the definition offered above keeps the focus more squarely on the process of elite-voter interaction that dramatically weakened the party systems and set the stage for the traditional parties’ subsequent elite and electoral troubles. 2.3.1 Concepts of Electoral Change Party-system collapse, as defined here, is a particularly extreme form of party-system change. Indeed, party-system collapse is such an extreme form of party-system change that the Peruvian and Venezuelan collapses are two of very few clear examples of party-system collapse in recent democratic history; the breakdown of the Italian party system during the 1990s (Bardi 1996; Morlino and Tarchi 1996) probably also meets the definitional criteria advanced here, and party-system changes in Bolivia may also qualify. As an extreme category on the underlying dimension of party-system change, collapse is more important to understand than its empirical frequency would suggest, both because it is normatively important in light of the unusually extreme institutional disruption that it represents and because—as a kind of theoretical extreme—it may provide insights into the study of less intense forms of change. It is worth emphasizing that specific parties may be deeply transformed by varieties of party-system change that fall far short of collapse. For example, ­Argentina’s Peronist party changed dramatically during the 1990s (see, e.g., ­Murillo 2001; Levitsky 2003), even as the party system that it anchored survived. This study focuses on change at the system level, so such party-level transformations emerge as a potential explanation of system persistence or change, rather than a definitional component of such system outcomes.

50

ch a r ac t er iz i ng pa rt y-s ys t e m ch a nge s

This section briefly reviews a collection of important kinds of party-system trajectories along a continuum of degrees of party-system change, a continuum for which collapse represents one terminus. Empirical examples of each kind of trajectory are provided, both to illustrate the relationships between the various types and the concept of party-system collapse, and to implicitly compare well-known party-system episodes with the cases of party-system collapse discussed below. Stable Election  Even when a party system is stable, there is typically some degree of change in each party’s vote share from election to election. For example, during the 1950s, in the middle of the period when European party systems were described as “frozen” (Lipset and Rokkan 1967: 50), parties’ vote shares did change somewhat from election to election. Between 1953 and 1957, the German Christian Democrats were able to increase their vote share from 45.2 percent to 50.2 percent; likewise, between 1956 and 1958, the French Communists’ vote share fell from 25.9 percent to 19.2 percent. Even during a highly stable period, it is a mistake to expect total electoral immobility. Nonetheless, stable elections are characterized by a relative lack of change—both in the organizations that make up the party system and the relative electoral success of each. In Latin America, stable elections—even in countries with relatively established party systems—often feature higher levels of electoral volatility. For example, between 1978 and 1983, at the peak of the Venezuelan traditional party system, Acción Democrática’s presidential vote share surged from 43.3 percent to 58.4 percent. Similarly, between 1958 and 1962, the presidential vote share of Costa Rica’s Partido Liberación Nacional (PLN) increased from 42.8 percent to 50.3 percent, the generally stable character of Costa Rica’s party system notwithstanding (Yashar 1995: 82–91). Yet because neither of these elections signaled a major, permanent change in electoral alignments or relative partisan strength, it seems safe to characterize them as essentially stable elections. Party-system collapse must entail a substantially greater amount of change. Realignments  Realigning elections (also called “critical elections”) are elections in which “the decisive results of the voting reveal a sharp alteration of the pre-existing cleavage within the electorate” (Key 1955: 4).6 While the concept of realignment has been applied to a range of varieties of party-system change (Sundquist 1983: 19–34), for the present purpose of differentiating among

ch a r ac t er iz i ng pa rt y-s ys t e m ch a nge s

51

degrees of party-system change, I use the term realigning election to refer to moments in which the electoral balance shifts in a dramatic and lasting way among the parties within a given party system.7 By this definition, the changes in U.S. partisan coalitions and, to some extent, levels of support described by Carmines and Stimson (1987) as a consequence of civil rights legislation during the 1950s and 1960s would count as episodes of realignment. Likewise, the periods during the first half of the twentieth century in which the labor union movement was incorporated into the traditional party systems of Uruguay and Colombia (Collier and Collier 2002: 271–313) would seem to meet the definition. Party-system realignments certainly transform the politics of a country. Nonetheless, there is a degree of organizational continuity in a realignment that clearly differentiates it from party-system collapse. Replacement of One Party  Even more dramatic change occurs in a party system when one of its constituent parties disappears and is replaced by a new organization. This kind of change preserves the numerical format of the party system, but it typically requires substantial revision of party loyalties by voters, and it may force politicians to revisit their own party affiliations. One famous example of an episode of partisan change that resulted in the replacement of one party within the existing party is the displacement, in the United States between about 1852 and 1860, of the Whig party by the Republican party (Aldrich 1995: 126–56; Holt 1999: 726–985). The replacement of liberal parties by socialist labor parties in the United Kingdom (McKibbin 1974; Tanner 1990) and elsewhere would also seem to fit in this category. As with realignments, party-system change involving the replacement of one party entails a great deal of social, political, and electoral disruption. Never­theless, in this form of party-system change, one or more major parties do persist organizationally; hence, replacement of a single party is a less drastic form of change than party-system collapse. Expansion/Contraction of the Party System  In the discussion to this point, each form of party-system change has maintained the existing numerical format of the party system; in a two-party system, for example, one party may be replaced or the relative strength of the parties may change, but the system still has exactly two major parties. However, some episodes of party-system change involve an expansion or contraction of the party system. One or more established parties

52

ch a r ac t er iz i ng pa rt y-s ys t e m ch a nge s

may disappear without being replaced; alternatively, emerging parties may force their way into the party system without eliminating the existing parties. In Europe, the emergence of left-libertarian (see Kitschelt 1989) and radical-right (see Kitschelt and McGann 1997) political parties since the early 1980s fits into this category of party-system change. Established parties of the center-left and center-right have not generally disappeared in the face of these new partisan challengers, yet the emergence of these parties has certainly forced some restructuring of party systems. In Latin America, the emergence of the Mexican Partido Acción Nacional (PAN) and the Partido de la Revolución Democrático (PRD) as legitimate, electorally relevant competitors to the Partido Revolucionario Institucional (PRI) is a clear example of rather dramatic change by expansion of the party system (Greene 2002; Mizrahi 2003). During the 1990s, Argentina also experienced an expansion of the party system, when the Peronists and the Radicals were joined by the FREPASO movement (Seligson 2003). The scope of these party-system changes notwithstanding, expansions or contractions of a party system preserve at least some of the established parties and therefore entail a less complete change than a party-system collapse. Replacement of More Than One Party  Perhaps the most dramatic form of party-system change short of collapse is the replacement of more than one party. This occurs when, in a multiparty system, several parties crumble simultaneously and are electorally displaced by new competitors. However, at least one established party remains intact. Because of the scope of change in multiparty replacements, the numerical format of the party system often changes as well. A fascinating example of the replacement of more than one party occurred in Canada’s 1993 elections. For decades before 1993, three parties had dominated Canadian electoral competition: the Liberal, Progressive Conservative, and New Democratic parties. However, the Progressive Conservative party’s share of the national vote fell from 42.9 percent in 1988 to only 16 percent in 1993. At the same time, the New Democratic party’s national vote share fell from 20.4 percent in 1988 to a mere 6.9 percent in 1993. The Liberal party persisted through this transformation, and two new parties quickly established themselves as medium- to long-term members of the Canadian party system: the Reform and Bloc Québécois parties (Carty 2002: 351–58). Clearly, a party-

ch a r ac t er iz i ng pa rt y-s ys t e m ch a nge s

53

system crisis of this magnitude is rather close to a party-system collapse; only one Canadian party remained effectively intact through 1993. Yet that party was able to lead the government, providing a degree of stability lacking in a full-scale party-system collapse. Party-System Collapse  Party-system collapse, as defined above, is clearly distinct from, and more disruptive than, the other forms of party-system change considered here. Aside from the South American cases, one of the best examples in recent decades of a party-system collapse is the transformation of the Italian party system in 1994. That year, two of the three traditional Italian parties fell into electoral irrelevance; the Socialists fell from a national vote share of 13.6 percent in 1992 to 2.2 percent in 1994, while the more powerful Christian Democrats fell from 29.7 percent in 1992 to 11.1 percent in 1994. The third traditional party, the Communists, underwent meaningful electoral decline during roughly the same years, although that party did not reach the same depths as the other two (falling from 26.6 percent in 1987 to 16.6 percent in 1992, and then rebounding somewhat to 20.4 percent in 1994) (Bardi 1996). This simultaneous crisis of all the traditional parties seems to be a transformation of the same magnitude as that experienced in Peru during the late 1980s and in Venezuela during the 1990s. The following sections characterize those transformations in a more systematic way. 2.3.2 The Evolution of Presidential Vote Shares As discussed earlier, Peru and Venezuela suffered a party-system collapse during the 1980s and 1990s, while Argentina—in spite of substantial political, economic, and even electoral turmoil—did not. These contrasts, as well as comparisons with the more stable Latin American party systems of Chile and Costa Rica, become even clearer when shown as aggregate electoral outcomes. Figure 2.1 shows how the combined vote share of the traditional parties in presidential elections evolved in Argentina, Chile, Costa Rica, Peru, and Venezuela between 1962 and 1998. In Venezuela and Peru, party-system collapse is represented by the precipitous drop in vote share starting in the mid-1980s in Peru and in the early 1990s in Venezuela.8 For both countries, this free-fall ended in a near-zero vote share. Clearly, Peru and Venezuela underwent party-system collapse. However, none of the other three countries in Figure 2.1 suffered a party-system collapse; even

54

ch a r ac t er iz i ng pa rt y-s ys t e m ch a nge s

f i g u r e 2 . 1 . Traditional party presidential vote share, 1962–1998. source: Data drawn from the Political Database of the Americas (a data collection available online at http://pdba .georgetown.edu/) and from the electoral institutes of each country. The traditional party system in Chile consists, for the purposes of this chart, of the Christian Democrats; the Socialists and their offshoot party, PPD; and a set of parties of the right that change somewhat over time. The Costa Rican traditional parties are Liberación Nacional (PLN) and Unidad Social Cristiana (PUSC).

in Latin America during the neo-liberal period, party-system collapse was far from an everyday event. Why, then, did the party systems of Peru and Venezuela collapse? The next chapter begins the process of presenting the evidence in favor of this study’s theory of how voter and party-leader interactions, centrally involving variables related to voter affect and to patterns of party organization, can generate such dramatic party-system transformations. Before moving definitively to that argument, it will be worthwhile to briefly consider two aspects of South American politics that variously provide easily eliminated alternative explanations for collapse or an alternative framing of the outcome of this study.

2 . 4 T W O A LT E R N AT I V E A C C O U N T S O F C O L L A P S E

Two aspects of Peruvian and Venezuelan political history during the late twentieth century have received extensive attention, even though neither provides

ch a r ac t er iz i ng pa rt y-s ys t e m ch a nge s

55

a particularly compelling explanatory account of party-system collapse. Some scholars have argued that poor institutional design at the level of the political regime or electoral law causes party-system collapse; although such institutions are important for understanding any number of outcomes, the discussion below argues that they do not play a central role in collapse in these countries. It is also worth noting that both Peru and Venezuela experienced serious periods of political violence during the process of party-system collapse, a fact that might be seen as a candidate explanation for party-system collapse but is perhaps more sensibly understood as an outcome related to collapse. 2.4.1 Regime and Electoral Institutions Although some analysts argue that regime and electoral institutions play a role in determining a country’s degree of vulnerability to party-system collapse, the evidence that institutional factors are central to the causal story of South American party-system collapse is unpersuasive. Institutional discontinuities, such as significant expansions of the franchise or regime transitions, are indeed a powerful predictor of electoral volatility in general (Roberts and Wibbels 1999), but the franchise was universal in both Peru and Venezuela before the period of party-system collapse, and neither country was in the first years of a transition away from authoritarianism. Some analysts have instead proposed a more specific relationship between vulnerability to party-system collapse and some potentially problematic electoral institutions (Tuesta Soldevilla 1995: 61–79; Kenney 2004: 59–70). The influence of regime and electoral institutions should of course be taken into account in considering any party-system phenomenon, if only to rule them out as central causes. To demonstrate the problems with regarding Peruvian and Venezuelan regime and electoral institutions as central to the explanation of party-system collapse, this section briefly describes the institutional arrangements in Argentina, Peru, and Venezuela. All three countries—and indeed every other Latin American country— have presidential systems. On the one hand, presidentialism might be seen as facilitating party-system collapse because it creates a winner-take-all prize (the presidency) that allows new, outsider parties to defeat the traditional parties without having to develop legislative institutions and an extensive network of credible parliamentary candidates. On the other hand, presidentialism may

56

ch a r ac t er iz i ng pa rt y-s ys t e m ch a nge s

impede party-system collapse by increasing the electoral barriers to entry; meaningful participation in the government in strong presidential systems often requires all-out victory, not just winning enough seats to become a viable coalition partner. In any case, it is worth pointing out that the Italian partysystem collapse occurred in a parliamentary system; evidently collapse does not require presidential institutions. There are thus both theoretical and empirical arguments against presidentialism as a crucial enabling factor for party-system collapse; presidential systems may experience party-system collapse differently than parliamentary ones, but collapse seems to be possible under both institutional arrangements. Beyond the simple dichotomy of presidentialism and parliamentarism, Latin American political regimes can also be classified according to the extent of presidential powers (Shugart and Mainwaring 1997: 40–52). There is substantial variation among Argentina, Peru, and Venezuela in terms of presidential power. Argentina’s president is potentially dominant over the legislature, having decree powers and a strong veto threat. Peru’s president, at the time of party-system collapse in 1990, was able to be proactive with respect to legislation, possessing decree powers on fiscal issues and a weak veto. Venezuela’s president, by contrast, was institutionally potentially marginal, with no veto capacity and no decree powers unless specifically legislated (Shugart and Mainwaring 1997: 49; Carey, Amorim Neto, and Shugart 1997: 441–42, 456–57, 459–60). However, in spite of the limited formal powers of the presidency, Venezuelan presidents were in practice highly powerful because they enjoyed substantial informal authority and the support of highly disciplined political parties. As a result, Venezuelan presidents “are allowed to exercise great authority subject to very little oversight” (Crisp 1997: 192). Both of the countries that experienced party-system collapse thus had presidencies with weaker formal powers than the Argentine president. However, when informal powers are taken into consideration, the comparison is not as clear. It may be possible that an institutionally relatively weak presidency contributed causally to party-system collapse. However, most accounts of Peruvian and Venezuelan politics during the 1980s and 1990s give substantial weight to the decisions and actions of the presidents, suggesting that this institutional factor probably has limited value as an explanation for collapse. It can also be argued that federalism facilitates party-system collapse by providing parties and movements outside the traditional party system with local

ch a r ac t er iz i ng pa rt y-s ys t e m ch a nge s

57

governing opportunities and the chance to develop a reputation for competence and honesty. Yet degree of federalism does not line up neatly with party-system collapse; both Argentina and Venezuela have fully federal formal institutions, while Peru was essentially unitary during the 1980s (Wibbels 2000: 699–700). Moreover, practical politics in Venezuela may have been more centralized than the formal institutions would suggest. In any case, the life histories of outsider candidates during the process of party-system collapse in Peru and Venezuela suggest that the opportunity for local governing experience may not have been essential. Some important actors in party-system collapse did have experience in local or regional government before moving to the national political stage; Andrés Velásquez and Henrique Salas Romer, two important Venezuelan presidential candidates from outside of the traditional party system, had previously served as governor. On the other hand, Hugo Chávez, the anti-party-system candidate who won the Venezuelan presidency in 1998, had no prior governing experience; nor did Mario Vargas Llosa, a major candidate from outside of the Peruvian party system in 1990, or Alberto Fujimori, the victorious outsider candidate in Peru’s 1990 elections. Federalism is therefore not a central explanatory consideration for party-system collapse, although it is possible that collapse may occur differently and have divergent consequences for future political dynamics in federal as opposed to unitary states. A further institutional factor that might serve as the basis for an alternative explanation of party-system collapse is the permissiveness of a country’s legislative electoral system—but once again, this factor seems both theoretically and empirically unsuited to serve as an important part of the causal story. First, the theoretical linkage between the permissiveness of legislative electoral institutions and party-system collapse is ambiguous. Permissive electoral institutions give parties outside the traditional party system the opportunity to more easily obtain legislative representation, but they also make it more difficult for established parties to fall to a position of electoral irrelevance. Second, there is relatively little empirical variation among Argentina, Peru, and Venezuela in the permissiveness of legislative elections. The mean legislator in Argentina and Peru during the 1980s was elected from a district of nine members, while the mean legislator in Venezuela was elected from a district of eleven members (Cox 1997: 309–11). Hence, this institutional factor also fails as an important cause of collapse.

58

ch a r ac t er iz i ng pa rt y-s ys t e m ch a nge s

Finally, in this discussion of institutions the two-round Peruvian presidential election system deserves some attention. Some analysts (e.g., Tuesta Soldevilla 1995: 76–79) suggest that the rule stipulating a two-candidate run-off if no candidate receives 50 percent of the vote in the first round of elections facilitated Peru’s party-system collapse by making it easier for outsider candidates to get into the second round and win in a forced choice between the top two candidates. However, this argument is difficult to accept. In the 1980 and 1985 elections, both of the top two candidates were from traditional parties; no outsider candidate had the chance to advance to the second round until the election in which the party system collapsed, in 1990. In the 1990 elections, however, both of the candidates who advanced to the second round were from outside the traditional party system; the second round thus simply reinforced the electorate’s expressed preference for a choice among outsider candidates, rather than between an outsider and a representative of the traditional party system. Hence, it is problematic to assign a major causal role to this institutional feature in explaining party-system collapse. In light of these difficulties in accounting for party-system collapse on the basis of institutional factors, the analysis below devotes little direct attention to the institutional context. Any extension of this book’s argument to institutional contexts, such as that of the Italian party-system collapse in 1993, that do not share the broad features of the South American pattern of relatively strong presidentialism and moderately proportional representation would thus require close attention to the potential consequences of institutional difference. 2.4.2 Protests, Coups, and Political Violence Both Peru and Venezuela went through important episodes of political up­ rising, protest, and violence during the 1980s and 1990s. Causal relationships between such political disruption and violence and party-system collapse are inevitably somewhat tangled; surely the factors that caused party-system collapse must be interconnected with those that caused a substantial proportion of the Venezuelan population to support an attempted military coup, or that caused a politically meaningful, if hard to measure, number of Peruvians to support a violent guerrilla movement. Voting for a nontraditional party is certainly a less extreme expression of dissatisfaction with the political status quo than engaging in acts of protest or even political violence, but the contrasts

ch a r ac t er iz i ng pa rt y-s ys t e m ch a nge s

59

in motivation may well reflect differences in degree, rather than differences in kind. Hence, I make no attempt to speculate about the causal weight of episodes of political protest and violence in producing party-system collapse. Even so, political protest and violence were a central component of the atmosphere of crisis in both Peru and Venezuela before party-system collapse, so a brief review of the relevant events is in order. Furthermore, the patterns of political protest and violence in both countries involve a substantial emphasis on themes related to corruption, policy choices, and ideology—a fact that provides significant circumstantial evidence that such issues are quite important to at least a large minority of the population within each country, and therefore indirectly supports this book’s central argument that party-system collapse is caused by anger deriving from concerns about these issues. As part of a package of neo-liberal economic reforms, Venezuela’s Pérez government on February 27, 1989, implemented a nationwide 10 percent increase in the price of gasoline and a 30 percent increase in bus fares.9 Protests against the hike in bus fares (which may have been exacerbated by unofficial fare increases imposed by individual bus drivers) began early in the morning in downtown Caracas, in bedroom communities surrounding Caracas, and in several other major Venezuelan cities. By noon, the bus stop protests had developed into riots that involved forcible closures of roads and highways, tire burning, and throwing of stones through the windows of cars and businesses. The government failed to respond to the riots, which spread through all major Venezuelan urban areas. By the next morning, what had begun as protest against an increase in bus fares had developed into widespread looting and a total shutdown of urban transportation grids. At noon on February 28, the government made its first move in response to the riots: a television announcement calling for peace and stating that violence would not be tolerated (Sanin 1989: 27–29). When this statement proved ineffective, President Pérez and his cabinet appeared on television to announce that a curfew would be enforced from 6:00 p.m. until 6:00 a.m. for the next several days. The police and the army enforced the curfew and violently repressed the riots. This protest episode, commonly called the Caracazo, ended on about March 4; the official death toll of the riot was about 400 people, a clear majority of whom were Venezuelan citizens killed by the military (Ochoa Antich 1992). In addition to the costs of the Caracazo in economic damage and lives lost, the

60

ch a r ac t er iz i ng pa rt y-s ys t e m ch a nge s

event represents, for many Venezuelans, as a historical dividing line between the period of peace that began in the 1970s and the political and social turmoil that predominated during the 1990s. In response both to the popular discontent expressed during the Caracazo and still prevalent in broad social groups (Norden 1996, Norden 1998; Myers and O’Connor 1998) and to dissatisfaction within the military about the direction of national affairs (Aguero 1995; Trinkunas 2002), groups within the Venezuelan military attempted coups against the Pérez government in February and November 1992. When both coup attempts failed to achieve their military objectives, their leaders surrendered and were imprisoned. However, two of the leaders developed sufficient popularity via these coup attempts that they were able to play major roles in national politics after their release from prison: Oscar Arias Cárdenas, who was the second-place candidate in the 2000 presidential elections, and Hugo Chávez, who has been the president of Venezuela since the party-system collapse in 1998. Even after the major protest episode of February, 1989—i.e., the Caracazo, and these coup attempts—Venezuela continued to experience a high level of political protest throughout the 1990s.10 Protest numbers peaked temporarily at about 1,100 per year around 1993. They then surpassed 1,200 per year after the party-system collapse in 1998. Even the lower levels of protest recorded between 1995 and 1998 represent a dramatic increase from the very low levels of protest observed during the 1970s and 1980s (López Maya, Smilde, and Stephany 2002: 14–20). Hence, even though the riots and coup attempts of the 1989–92 period were never repeated, the atmosphere of political crisis and confrontation in the country persisted—and this atmosphere was a fundamental component of the experience of politics during the period leading to party-system collapse in Venezuela. Peru’s experience with political violence during the years before the partysystem collapse in 1990 was even more intense than Venezuela’s. Instead of riots and coup attempts, Peruvian political violence primarily took the form of conflict between guerrilla groups, which often adopted terrorist tactics, and the military and police forces of the Peruvian state.11 A group called the Movimiento Revolucionario Túpac Amaru played a minor but meaningful role in these conflicts (McCormick 1993), but by far the most important guerrilla movement in Peru during the late twentieth century was Sendero Luminoso,

ch a r ac t er iz i ng pa rt y-s ys t e m ch a nge s

61

a Maoist insurgent movement that, at its peak, had 10,000 full-time combatants, the support of about 15 percent of Peru’s population, and some degree of control over almost a third of Peru’s municipalities (McClintock 1998: 73–81). Sendero began as a radically militant, primarily rural splinter group within Peru’s traditionally factious political left. Starting in 1980, the group put its Maoist rhetoric regarding the importance of violence and the armed struggle into action, initiating a guerrilla war against Peru’s newly democratic regime. As the state clumsily repressed Sendero, initially killing far more people than the guerrillas did, the insurgent group increased its strength and acceptance within the central Peruvian Andes. Sendero’s strategy was to construct an insurgent, Maoist alternative state organization that could gradually win the loyalty of Peruvian citizens until the Peruvian state finally collapsed and was replaced by the guerrillas. This construction of an alternative state was accomplished in part by direct enforcement of the law in Sendero-controlled areas, with offenders often executed in dramatic and gruesome manner. Sendero also directed its violence against its enemies, broadly defined to include agents of the state, political party activists, and popular-organization leaders. In the late 1980s, Sendero turned its focus from a primarily rural struggle to a greater emphasis on attacking Peru’s capital city, Lima. The insurgents were able to establish operating bases on the outskirts of the city, and for several years succeeded in carrying out dramatic, high-profile attacks within the city. The Peruvian state during the 1980s was unsuccessful in repressing Sendero’s guerrilla onslaught. The guerrilla group was able to consistently expand its operations and area of control throughout the decade. Furthermore, army brutality in response to the insurgent threat had significant costs in terms of popular legitimacy. Not until the capture of Sendero’s leader, Abimael Guzmán, in 1992 (two years after party-system collapse) was the Peruvian state able to make meaningful progress in restraining the guerrilla threat.12 These acts of political protest and violence are central components of recent Peruvian and Venezuelan history. Furthermore, these episodes illustrate the high stakes that people in both countries attached to political outcomes during the period of party-system collapse. Themes of ideology and policy helped motivate political contention within both countries, as did excessive government responses to protest and political violence. Voting for parties from outside of

62

ch a r ac t er iz i ng pa rt y-s ys t e m ch a nge s

the traditional party system was clearly not the only avenue for opposition to the political establishment; protest and violence offered strategies that could complement or displace electoral expressions of dissatisfaction. Hence, while this study’s focus on party-system collapse captures an important dimension of political dissent within Peru and Venezuela, it is not (and perhaps could not be) comprehensive in this regard. With the descriptive and conceptual apparatus of this chapter assembled, all is now prepared for a presentation of the evidence in support of the theory of party-system collapse developed in the previous chapter.

chapter 3

E C O N O M I C S , S O C I E TA L CRISIS, AND ANXIETY

D

o p a r t y s y s t e m s c o l l a p s e because of bad economic performance? This chapter considers the extent to which party-system collapse might have been predictable given the general economic conditions of Peru and Venezuela during the 1980s and 1990s. The analysis shows that economic crisis does not satisfactorily explain party-system collapse; traditional parties in Peru and Venezuela suffered more electoral decline than would be statistically predicted on the basis of those countries’ (very poor) economic performance before collapse. Nonetheless, this chapter suggests two ways in which persistent poor economic performance sets the stage for collapse. In the first place, as the cross-national analysis unambiguously indicates, economic decline tends to weaken governing parties even though it does not guarantee their disappearance. Nevertheless, it stands to reason that weakened parties make easier prey for anti-party-system outsiders than do full-strength parties. Second, as the last section of this chapter shows through a multi-level analysis of survey and economic data, economic crisis systematically increases individuals’ anxiety about the society and the future. Anxiety augments citizens’ attentiveness to politics and their search for credible alternatives to the government, potentially helping to make party-system collapse a possibility. For these two reasons, economic crisis may be regarded as a partial causal enabler of party-system collapse.

64

economics, societal crisis, and anxiety

Peru and Venezuela both experienced deep economic crises during the 1980s and 1990s. In Peru, crisis conditions included hyperinflation, negative economic growth rates, and rising poverty levels. Venezuela’s challenges included negative economic growth rates and some inflation. How unusual are these problems in Latin America? Did other countries suffer from similar sets of problems during this period, without experiencing party-system collapse? To what extent can these problems statistically account for the party-system collapses in Peru in 1990, and in Venezuela in 1998? This chapter shows that, while Peru and Venezuela undoubtedly went through serious difficulties, some other Latin American countries experienced similar problems without party-system collapse—in particular, Argentina and Nicaragua. A statistical analysis formalizes these conclusions, showing that the economic variables considered here cannot adequately account for the extreme electoral changes involved in party-system collapse. In other words, party-system collapse emerges as a phenomenon that is distinct from other Latin American episodes of party-system change.

3.1 PERU’S ECONOMIC CRISIS

What, then, were the contours and the chronology of the economic crisis in Peru?1 The Peruvian economic crisis hit its worst point shortly before the partysystem collapse in 1990. At about that time, Peru had suffered from a few years of extremely negative economic growth, as well as massive hyperinflation. Peru’s economic difficulties during the 1980s may be seen as starting with the legacy of the military government led by General Alvaro Velasco from 1968 until 1975. This government pursued economically nationalistic goals of reducing the power of foreign capital in the Peruvian economy, as well as reformist objectives oriented toward the political strengthening of the urban and rural poor at the expense of traditional social and economic elites (Lowenthal 1975; McClintock and Lowenthal 1983). In the process, the government dramatically increased state investment, introduced economic protectionist policies that created an immense trade deficit, overvalued the currency, and relied extensively on foreign credit markets to cover shortfalls (Thorp 1991: 67–82), thereby increasing external public debt from $945 million in U.S. dollars in 1970 to $3.066 billion in 1975 (Banco Central de Reserva del Peru

economics, societal crisis, and anxiety

65

1993: 107).2 In effect, the Velasco government adopted the full economic policy package that Dornbusch and Edwards describe as the “macroeconomics of populism” (1991). As foreign markets became wary about extending additional credit and as the Velasco regime weakened domestically, the Peruvian military in 1975 changed its national leadership and economic policy direction. The new president, General Francisco Morales Bermúdez, raised prices, reduced real wages, slashed government spending, and devalued the national currency (Thorp 1991: 83–98). These policies neither removed the inflationary pressures resulting from the Velasco government’s macroeconomic imbalances nor produced strong economic growth (Sheahan 1999: 133, 138), and the military arranged a transition to democratic rule in 1980. The first elected president of the 1980s, Acción Popular’s Fernando Belaúnde Terry, pursued a more market-oriented, neo-liberal policy than either the Velasco or the Morales Bermúdez governments (Pastor 1992; Conaghan and Malloy 1994), reducing tariff rates, eliminating quantitative import restrictions, and allowing the currency to appreciate (Sheahan 1999: 139). At the same time, the Belaúnde administration continued extensive state spending and investment in infrastructure (Pastor 1992: 113–14), in the process expanding the Peruvian state’s already large external debt to a new record of $10.552 billion by 1985 (Banco Central de Reserva del Peru 1993: 107). Perhaps in part for this reason, levels of foreign private investment in Peru during the period were too low to support economic growth (Pastor 1992: 111–13). Beginning in 1983, the Belaúnde government faced an unsustainable trade deficit and a reduction in access to additional foreign credit, resulting in a sharp reduction in state spending and in the money supply (Thorp 1991: 108–17). This overall policy package, along with climate difficulties associated with the El Niño phenomenon and the effects of the regionwide debt crisis (Frieden 1991), resulted in increased inflation rates, low economic growth, and a deterioration of employment conditions (Sheahan 1999: 88–97). These failures may have resulted from a mismatch between the implementation of Belaúnde’s policy package and structural conditions in the Peruvian economy (Pastor and Wise 1992). The final presidential period before party-system collapse, Alan García Pérez’s term from 1985 until 1990, involved an experiment in heterodox economic management that produced the worst years of the Peruvian economic crisis (Thorp

66

economics, societal crisis, and anxiety

1991: 121–41). García attempted to reduce inflation and kickstart economic growth by establishing government controls over wages, prices, interest rates, and exchange rates, as well as by promoting extensive state spending to increase domestic demand (Lago 1991: 270–82). Price levels would be set via negotiation with business leaders, and funding for the entire package would be obtained by limiting foreign debt payments to no more than 10 percent of the total value of Peruvian exports (Sheahan 1999: 139–42). These policies once again created a massive trade deficit. Furthermore, with international credit markets hostile toward García as a result of his debt-repayment policies, the state had limited access to credit to support imports or domestic spending. With the effect of making economic policy problems even less manageable, García announced in July 1987 that he would pursue the nationalization of Peru’s banks and insurance companies, resulting in hostility from private business and a collapse of private investment (Gonzales de Olarte 1996: 15). An economic crisis followed, including an episode of hyperinflation, a collapse of economic growth, and a sharp increase in unemployment. Although Tanaka argues that García’s economic policy was to some extent a political success in that it managed to delay major efforts at neo-liberal stabilization until after the 1990 presidential elections (Tanaka 1998: 155–62, 186–89), it is clear that the Peruvian economy underwent a full-scale collapse during the last years of the García administration. Between 1985 and 1990, the average household in Lima, Peru’s capital city, experienced a 55 percent decline in per capita consumption, and the percentage of households at risk of starvation increased during that time period from 0.5 percent to 17.3 percent (Glewwe and Hall 1992). Indeed, the crisis was not resolved until after the party-system collapse; the next anti-party-system president, Alberto Fujimori, imposed by surprise a strict program of neo-liberal economic adjustment that controlled inflation and prevented further decline in per capita GDP, although it did not improve the labor market (Wise 1994; S. Stokes 2001; Arce 2005; Wise 2003).

3 . 2 V E N E Z U E L A’ S E C O N O M I C C R I S I S

Venezuela’s economy experienced a crisis that never reached the explosive proportions of Peru’s situation in 1989–90.3 However, while Venezuela’s crisis was not as deep as Peru’s, it was more persistent and longer lasting. Negative growth

economics, societal crisis, and anxiety

67

rates recurred periodically throughout the two decades under consideration, and the country suffered through two episodes of substantial inflation (though not hyperinflation). Venezuela’s rate of open unemployment does seem to have been somewhat worse than Peru’s; Venezuela spent the late 1980s and the 1990s varying around a mean unemployment level of about 9 percent. That figure deteriorated somewhat further after the party-system collapse. As with Peru, Venezuela’s economic difficulties during the 1980s and 1990s had their roots in political and economic decisions made during the 1970s and earlier. Perhaps the most critical period was the Acción Democrática presidency of Carlos Andrés Pérez from 1974 to 1979. Pérez significantly increased tariff protection of consumer goods and provided extensive subsidies to favored industries (Enright, Frances, and Saavedra 1996: 312–14). At the same time, the exchange rate was pegged to the dollar, resulting in an overvalued currency that encouraged imports (Karl 1997). Furthermore, the Pérez government imposed strict controls on foreign capital, requiring new foreign investments to meet state goals for employment, domestic production, regional location within Venezuela, technology transfer, and profit reinvestment. Some industrial categories, including public utilities, communications, and transportation, were strictly reserved for Venezuelan corporations (Adams 1992: 82–94). Perhaps most noteworthy were the decisions by the Pérez administration to nationalize primary extractive industries, including in particular the petroleum industry in Venezuela (Coronel 1983). These policies required extensive government spending. Upon assuming office in 1974, Pérez tripled the government budget for the year (Karl 1995: 37); increases in state spending continued for the rest of the decade. Fiscal expansion was funded in part through Venezuelan oil earnings, which were at record levels through much of the 1970s, and in part through extensive recourse to foreign debt (Karl 1995, 1997). Unfortunately, oil prices went into long-term decline starting in the late 1970s, and the subsequent governments of Luís Herrera Campíns (COPEI) from 1979 to 1984, and Jaime Lusinchi (Acción Democrática) from 1984 to 1989, were unable to continue the use of expanded government spending as an engine for economic growth. On the other hand, neither government reduced spending in proportion to the decline in petroleum revenues; in 1989, government expenditures were at 21.7 times their 1973 level (Karl 1995: 37).

68

economics, societal crisis, and anxiety

The Herrera Campíns government faced persistent opposition from organized labor (Davis and Coleman 1989), as well as from a congress controlled by the political opposition (Coppedge 1994: 68). However, economic crisis was avoided until 1983, when a sharp decline in oil prices, along with accumulated imbalances, forced the government to devalue the currency for the first time. To continue favoring selected industries while avoiding further state expenditures to defend an unrealistic exchange rate, the government created a three-tiered exchange rate, with foreign exchange available at preferential rates for targeted firms (and perhaps for political allies) (Karl 1995: 44–45). At about the same time, the state’s massive accumulated foreign debt (which was between $29 billion and $54 billion by 1982, depending on the specific government record consulted) became a topic of public debate, and reductions in state spending exacerbated economic decline (Mayorbe 1985). The Lusinchi government continued to struggle with Venezuela’s debt obligations. The existing policy package, involving multiple exchange rates, price controls, and extensive government spending, was unsustainable—particularly in light of Venezuela’s immense debt service. Yet when the Lusinchi government attempted in 1984 to adopt policy changes with a neo-liberal orientation that would ease the fiscal pressures on the state, the Venezuelan union confederation adopted a confrontational stance that, in effect, vetoed the changes (Davis and Coleman 1989). As a consequence, the Lusinchi government was all but forced to pass the economic crisis off to the following president, making no major changes in policy until the 1988 elections, which returned Carlos Andrés Pérez to office. The second Pérez administration began as the unsustainable policy package of the 1980s reached its breaking point. By early 1989, the combination of price controls and serious inflation had produced a major black market as well as rationing. The state’s budget deficit reached 9 percent of GDP. All of these economic difficulties imposed an immense social cost: the number of households below the poverty line increased by a factor of ten between 1981 and 1989 (Naim 1993: 42). In response, Pérez introduced, in early 1989, a major set of economic reforms intended to reduce state intervention, stabilize the economy, liberalize trade, introduce flexibility into labor markets, and reestablish fiscal stability (Naim 1993; Zambrano Sequin 1995; Crisp 1998). By presidential decree, trade barriers were dramatically reduced and exchange rate controls were dismantled

economics, societal crisis, and anxiety

69

(Adams 1992: 96–105). Pérez’s party, Acción Democrática, reacted with hostility, both to the substance of these reforms and to their centralized, nonpartisan mode of implementation (Corrales 2002: 121–27). As a result, a variety of laws (involving changes to labor policy, Social Security, and the tax system, among other topics) necessary to fully implement the reform package were defeated in the legislature (Corrales 2002: 126, 131–34). In combination, the early stabilizing reforms that Pérez implemented by presidential decree, and an increase in state expenditures funded in part by the temporarily rising petroleum prices that resulted from the Persian Gulf crisis of the early 1990s, produced a period of rapid economic growth, a marked decline in inflation, and a reduction in unemployment (Lander and Fierro 1996: 53–55, 57–58). However, the fall in oil prices after the end of the Gulf crisis and the failure of the legislature to implement laws completing the economic reforms led the economy to turn negative again in 1992–93. The administration of Rafael Caldera (the founder of COPEI, but elected in 1993 as an anti-party-system candidate) initially rejected the pro-market neoliberal reforms of the Pérez administration, instead emphasizing state action to stabilize the economy while improving social conditions (Maza Zavala 1996: 166–68). However, in the wake of a serious banking crisis (Krivoy 2000), Caldera in 1996 launched an economic reform project comparable to Pérez’s reforms from 1989. When the nontraditional parties that formed Caldera’s governing coalition in part rejected the reforms, Caldera sought support from the largest opposition party, Acción Democrática. With that support the Caldera government managed to carry out the privatization of some state enterprises, opened the oil sector to foreign investment, and liberalized the banking sector. However, no real changes were made to increase labor law flexibility or to improve the state’s taxcollection apparatus (Corrales 2002: 223–24). A decline in oil prices starting in late 1997 caused yet another crisis, producing a spike in inflation and a drop in growth that led into the party-system collapse election of 1998 (Corrales 2000). These differences in the nature and duration of crisis notwithstanding, both Peru and Venezuela experienced economic difficulties serious enough that they seem an intuitively plausible explanation of party-system collapse—but, as discussed earlier, so did other countries that did not experience collapse. To more definitively reject the hypothesis that economic trouble was the central cause of party-system collapse, additional questions must be asked. From a cross-

70

economics, societal crisis, and anxiety

national perspective, do Latin American voters punish incumbent parties for poor economic growth, for inflation, or for both? Is the degree of economic failure observed sufficient, in light of regional patterns, to explain the observed declines in vote share for the governing party in Peru in 1985 and 1990, and in Venezuela in 1993 and 1998? From an individual perspective, were those voters who had a more optimistic view of the Peruvian economy in 1990 and the Venezuelan economy in 1998 more likely to support the traditional parties than those individuals with perhaps more realistic views of the state of the economy? The first two of these questions are explored below, while the third—which requires individual-level survey data—is addressed in the following chapter.

3 . 3 P E R U A N D V E N E Z U E L A I N L AT I N A M E R I C A N P E R S P E C T I V E

Economic crises have not been uncommon in Latin America since 1980 (Remmer 1990: 321–27; French-Davis, Munoz, and Palma 1998: 175–77). During the 1980s, the whole region suffered from a major debt crisis, and several countries experienced episodes of hyperinflation. In the 1990s, there was economic divergence, as some countries recovered from the difficulties of the 1980s and others continued to suffer from poor growth rates and high inflation. Thus, if the period from 1980 to 2000 is considered as a whole, the economic picture in Latin America shows substantial variation. Table 3.1 shows the average annual growth, inflation, and unemployment rates in selected Latin American countries over this period.4 Seven Latin American and Caribbean countries averaged negative growth rates between 1980 and 2000: Bolivia, Ecuador, Guatemala, Honduras, Nicaragua, Peru, and Venezuela. This list suggests the possible existence of an association between negative economic growth and party-system change, while simultaneously providing important exceptions to the generalization. Party systems have collapsed in Peru, Venezuela, and possibly Bolivia. Substantial party-system instability also occurred in Guatemala. Guatemalan party politics during the period from 1980 to 2000 were complicated by that country’s process of transition to democracy; nonetheless, the major feature of partisan competition remains the volatility of the parties that occupy center stage in elections. No party has ever won the presidency twice, and parties’ legislative

economics, societal crisis, and anxiety

71

ta b l e 3 . 1 . Economic performance of Latin American countries, 1980–2000 (percent) Country

Average Annual per Capita Growth

Average Annual Inflation

Average Unemployment

Peru Venezuela

– 0.35 – 1.08

544.71 32.72

7.63 10.08

Argentina Bolivia Brazil Chile Colombia Costa Rica Dominican Republic Ecuador El Salvador Guatemala Honduras Mexico Nicaragua Panama Paraguay Uruguay

0.38 – 0.31 0.62 2.96 1.22 0.82 2.88 – 0.55 0.50 – 0.10 – 0.18 0.86 – 2.09 2.48 0.45 1.45

375.74 663.83 697.18 15.15 23.21 23.17 17.54 1.09 5.18 13.37 12.50 42.47 1387.57 4.00 18.28 49.39

8.11 5.31 5.29 8.11 11.64 5.76 16.65 8.69 9.01 2.68 5.66 3.34 10.76 13.03 5.70 10.09

vote shares are far from stable over time. For example, the two largest legislative parties in the 1999 elections (Guatemalan Republican Front and National Advancement Party) won a combined share of less than 15 percent of the vote in the 2007 legislative elections. For these four countries, trouble achieving economic growth has coincided with party-system instability, at least, and often outright collapse—although for Guatemala it is difficult to know whether economic trouble caused party-system turmoil, or whether the country has simply institutionalized a pattern of rapid party turnover. For other countries with negative average growth rates, the outcome is either more mixed or reflects marked and perhaps surprising stability. In Ecuador, substantial volatility in both vote shares and party organizations has coincided with important continuities. The largest parties of the right (Partido Social Cristiano) and the left (Izquierda Democrática) during the 1980s retained substantial vote shares as late as the 2006 legislative elections. The other two largest parties in those elections (Partido Renovador Institucional Acción Nacional and Partido Sociedad Patriótica) have the character of personal electoral vehicles for their leaders,

72

economics, societal crisis, and anxiety

a long-standing pattern in Ecuadorean politics. Honduras has been a portrait of party-system stability since 1980, with each election dominated by competition between the National and Liberal parties. Nicaragua also demonstrated a great deal of party-system stability during the 1990s, with competition centering on the Sandinista party (the dominant party of the 1980s), the Liberal party, and more peripherally the Conservative party. This pattern of relative party-system stability in Nicaragua is particularly noteworthy, given that it had the worst average economic growth rate of any country under consideration. Hence, while there are some important signs of a relationship between negative economic growth and severe party-system turmoil, there are also important exceptions that raise doubt regarding the causal meaning of the pattern. At the other extreme, three countries had average growth rates of more than 2 percent during this period: Chile, the Dominican Republic, and Panama. Chile, of course, has been characterized by party-system stability, with competition persistently focusing on stable center-left and right alliances. The Dominican Republic likewise saw stable competition among three traditional parties through the 1980s and 1990s. Although Panama experienced a great deal of regime turmoil and other political tension during the 1980s and 1990s, its party system also remained essentially intact, with competition revolving around a small collection of long-established parties and associated political families. It is worth noting that substantial party-system change or instability was also experienced, during the 1980s and 1990s, by several countries with modestly positive patterns of economic growth, including Brazil, Mexico, and Uruguay. In comparison with the finding of substantial though not universal partysystem turmoil among negative-growth countries, these facts about the party systems of relatively prosperous countries suggest the theoretical interpretation that I adhere to in the rest of this discussion. Economic problems help set the stage for party-system change, raising citizens’ anxiety level in a way that produces close scrutiny of existing and new partisan alternatives. However, such scrutiny neither guarantees that party-system change will occur nor cleanly explains why the change takes the form of collapse. The same finding is true for inflation. The data suggest a sharp distinction between a set of five countries that experienced at least one episode of hyper­inflation, on the one hand, and the other countries of Latin America where inflation rates were more moderate. Argentina, Bolivia, Brazil, Nicaragua, and Peru had aver-

economics, societal crisis, and anxiety

73

age annual inflation rates of over 300 percent for the period from 1980 to 2000; in no other country did this figure reach triple digits. Thus, whereas Venezuela had a much lower growth rate than Argentina and even Peru, Venezuela also had lower inflation than Peru; Argentina’s situation was roughly comparable to Peru’s. Hyperinflation, like negative economic growth, is associated with party-system turmoil; the five most troubled countries include two (Bolivia and Peru) that have experienced party-system collapse, as well as two (Argentina and Brazil) that have gone through substantial party-system turmoil without experiencing anything that entirely meets the definition of a collapse. The major exception to the generalization, once again, is Nicaragua, which had the worst inflation in the region yet experienced relatively little party-system change. Thus it is plausible to conclude that inflation activates citizens’ surveillance systems and enables the revision of political habits necessary for party-system change—yet, as with economic growth, this neither guarantees party-system crisis nor determines whether the crisis will take the form of a collapse. The unemployment results show even less of a pattern. Guatemala and Mexico report notably low rates; however, as discussed earlier, both countries experienced substantial party-system change during the period of interest. By contrast, Colombia, the Dominican Republic, Nicaragua, Panama, Venezuela, and Uruguay show double-digit average unemployment rates for the period under question. The countries in this group reflect a mix of severe party-system changes and high levels of stability. Peru’s average unemployment rate falls roughly in the middle of the regional distribution, as does Argentina’s. Overall, more than with inflation or economic growth, there is little evidence of a clear pattern connecting levels of unemployment with party-system change. Even when examined in conjunction, these economic indicators do not cleanly distinguish cases of dramatic party-system change from cases where party systems either evolve more gradually or persist. Consider, for example, the set of countries with average economic growth rates of 0.5 percent or less, average inflation rates of 30 percent or more, and average unemployment rates of 7 percent or higher: Argentina, Nicaragua, Peru, and Venezuela. Of these countries, two experienced party-system collapse, but the other two had far less turbulent party systems during the period under consideration. Argentina, which serves as a central contrast case throughout this study, experienced important moments of party-system turbulence during the 1990s. However, partisan com-

74

economics, societal crisis, and anxiety

petition there still revolves around the traditional Peronist party, and many of the most important opposition figures derive from the weakened Radical party. Nicaragua is an even more significant anomaly for simple economic accounts of party-system collapse: in spite of experiencing unambiguously the worst economic performance of any democratic country in the region, the traditional cornerstones of its party system have remained essentially intact. These comparisons certainly do not prove that economic crisis is irrelevant to the project of explaining collapse. Economic crisis probably sets the stage for party-system change by encouraging citizens to carefully reevaluate their established partisan ties and to consider nontraditional alternatives. However, the above comparisons suggest that macroeconomic variables do not determine the conclusions that citizens draw. Having carried out this comparative review of economic performance in Latin America, I formalize the results by conducting a statistical analysis of the extent to which economic factors predict the sharp declines in vote share that constitute party-system collapse. If the outcome of party-system collapse in some Latin American countries but not in others is to be explained by national economic performance, it is necessary (although not sufficient) that the variables in question have a large enough statistical association with decline in the governing party’s vote share to explain much or most of the decline that constituted party-system collapse. Furthermore, for a collapse to take place, the votes lost by the governing party must be transferred outside of the traditional party system; the change in vote share of the largest opposition party from the previous election is therefore also of interest. As we will see, the data do not meet these requirements, implying that the economic variables considered here are not satisfactory as an account of party-system collapse. 3.3.1 Prior Studies of Economic Outcomes and Vote Shares in Latin America Since 1990, four major studies have analyzed the relationships between aggregate national economic trends and governing-party vote shares in sets of cases including Latin American countries. In an analysis that considers the relationships between exchange rate volatility, change in GDP, and inflation, Remmer finds that, for the Latin American nations she considers, only inflation is a useful predictor of change in governing-party vote share (1991: 784).

economics, societal crisis, and anxiety

75

Using a sample that includes three Latin American countries and five other developing-world countries, Pacek and Radcliff find a relationship between declines in per capita GDP and falls in governing-party vote share (Pacek and Radcliff 1994: 752). However, growth in per capita GDP is insignificant in their model. Unfortunately, Pacek and Radcliff do not explore any economic variables other than change in GDP. In particular, they exclude inflation, rendering their results incomparable with Remmer’s. In an analysis of 328 democratic elections in Latin America, Molina finds significant relationships between the absolute level of per capita GDP and change in governing-party vote share, as well as between GDP growth and change in vote share (Molina 2001: 441). Nonetheless, like Pacek and Radcliff, Molina’s model does not consider inflation as an independent variable. Finally, in an innovative discussion that incorporates hypotheses about the influence of electoral rules on economic voting patterns, Benton (2005) finds that, in thirteen Latin American countries, more permissive legislative voting rules encourage voters to punish a broader range of parties and to consider economic failures from the relatively distant past in making voting decisions. However, this analysis considers only GDP growth as a measure of the economy, once again excluding inflation. Furthermore, it is somewhat unclear to what extent Benton’s findings about electoral rules in legislative elections could generalize to presidential elections, which inherently have quite restrictive electoral rules. In comparison with all of these studies, the analysis here allows a more flexible economic specification, including both average level and overall change in inflation and growth rates as independent variables in all models and considering unemployment in one. Thus, in addition to the distinctive focus on explaining party-system collapse, the analysis below may be seen as a modest refinement of existing aggregate models of change in governing-party vote share in Latin America. 3.3.2 Statistical Analysis The primary goal of the analysis below is to determine whether the changes involved in party-system collapse were predictable on the basis of a regionwide comparison of aggregate economic and corruption-related measures.5 It is important to be able to test, in every case, whether the elections during which party-system collapse occurred differ significantly from the regional patterns.

76

economics, societal crisis, and anxiety

If it were true that electoral results in Peru in 1985 and 1990, and in Venezuela in 1993 and 1998, fit well within the regional model, then party-system collapse could plausibly be explainable as the consequence of economic crisis. However, because, as will be seen, these elections do not fit comfortably within the regional patterns, I rule out a direct link between economic performance and party-system collapse. To test whether these elections fit the regional patterns, dummy variables for each party-system collapse election are included in all estimated models. In addition to allowing tests of the degree to which party-system collapse elections fit the model, these dummy variables in effect remove the relevant elections from the data set for purposes of estimating the other coefficients. To the extent that party-system collapse represents a distinctive phenomenon, and not merely an extreme case on a linear dimension of party-system change, these dummy variables insulate the rest of the analysis from the distinctive coefficients associated with party-system collapse. In the 1998 Venezuelan presidential election, the governing party (Convergencia) did not present a candidate, so that year is excluded from the analysis. For each model, a 95 percent confidence interval for Convergencia’s 1998 vote share is reported. Of course, no data are available for a direct test of such intervals, but the party’s legislative vote share of 2.36 percent provides a baseline sense of the party’s support during those elections. Variables measuring growth, inflation, and unemployment are the same as those discussed in the corresponding sections above. One additional variable is included in each model: following Molina’s suggestion, discussed above, that national poverty is associated with electoral volatility, logged per capita GDP in the year of the election is included for each country.6 The first model considered, with coefficient estimates presented in the column of Table 3.2 labeled “Economics Only,” shows the results of regressing governing-party vote share on the political controls discussed above, as well as a set of economic variables that are retained in all subsequent models. Specifically, the model considers average annual inflation throughout the last administration, change in annual inflation rates between the current election year and the year of the previous election, average annual growth during the administration, change in annual growth rates since the last election, and per capita GDP in the year of the election.7

economics, societal crisis, and anxiety

77

ta b l e 3 . 2 . Economic models of Latin American elections, 1980–2001 Variable Name

Economics Only

Economics + Unemployment

Intercept Last opposition vote share Candidate from main opposition

0.596 (0.297) 0.339 (0.139)* – 0.766 (0.217)**

1.044 (0.349)* 0.442 (0.145)** – 0.859 (0.216)**

Peru, 1985 Peru, 1990 Venezuela, 1993 Average inflation Change in inflation Current inflation Average growth Change in growth Current growth Per capita GDP Unemployment

– 1.384 (0.469)** 1.262 (0.533)* – 0.833 (0.460) – 0.00244 (0.00041)** – 0.0000821 (0.0000339)* 0.000551 (0.000119)** 0.0965 (0.0386)* 0.0413 (0.0164)* – 0.0760 (0.0286)* 0.00000201 (0.0000302)

– 1.407 (0.462)** 1.704 (1.049) – 0.974 (0.457)* – 0.00232 (0.000508)** – 0.0000835 (0.0000334)* 0.000460 (0.000256) 0.119 (0.0409)** 0.0390 (0.0165)* – 0.0884 (0.0288)** 0.00000201 (0.0000308) – 0.0349 (0.0153)*

R2 Degrees of freedom

0.744 48

0.763 44

notes: OLS regression estimates with all vote shares transformed via the logit function. *(p < 0.05) **( p < 0.01).

All of the economic variables other than level of GDP are statistically significant. Inflation hurts the incumbent party, although perhaps somewhat less so in the election year than at earlier points. Economic growth helps the governing party, although once again, growth during the election year may be somewhat less important. In general, there is substantial evidence that Latin American voters in the aggregate are responsive to economic signals involving both price stability and the rate of expansion of economic activity. As is evident from the data in Table 3.1, the distribution of the inflation rate in Latin America is notably skewed. Therefore, logged inflation rates may be a plausible substitute for the untransformed inflation rates used in the model; however, a test of the hypothesis that logged inflation belongs in the model has a P value of 0.61.8 Clearly, the predictive power of these variables net of the linear effects is minimal, so using the linear effects is a statistically justifiable descriptive option. In substantive terms, the logged specification collapses the cases of hyperinflation substantially toward the center of the distribution. The result that logged inflation rates are not better than the untransformed linear

78

economics, societal crisis, and anxiety

rates suggests that Latin American governing parties paid a much higher electoral price for hyperinflation crises than for routine high levels of inflation. Returning to the coefficient estimates reported in the first column of Table 3.2, it would seem that none of the party-system collapse elections closely follows the regional patterns as predicted on the basis of economic growth and inflation. Acción Popular, the Peruvian governing party from 1980 to 1985, received significantly less of the presidential vote in 1985 than the model would have predicted. APRA, the Peruvian governing party from 1985 to 1990, won significantly more of the vote than the regional patterns would suggest. Acción Democrática, the Venezuelan governing party from 1988 to 1993, lost substantially (and almost statistically significantly) more electoral ground than would be predicted on the basis of economic results. An approximate 95 percent confidence interval for the vote share of the Venezuelan governing party in 1998 is from 7 percent to 42 percent, with a point prediction at 19 percent. The observed performance of Convergencia in the Venezuelan legislative elections suggests that this interval may be somewhat too high. Hence, the party-system collapse elections are not adequately explained by inflation and growth rates, in conjunction with the political controls. Adding data on unemployment rates to the analysis, as in the model for which coefficient estimates are reported in the second column of Table 3.2, does not substantially change the findings regarding economic voting. Once again, the growth and inflation variables are broadly significant and suggest voter attention to the economy. Voters in the aggregate likewise appear to punish the incumbent party for unemployment. Dummy variables for two of the three party-system collapse elections (Peru in 1985 and Venezuela in 1993) are statistically significant. APRA’s vote share in the 1990 Peruvian elections turns out to be substantially but not quite significantly above the model’s predictions, while Acción Democrática’s vote share in the 1993 Venezuelan elections falls significantly below the predicted value, as does the vote for Acción Popular in the 1985 Peruvian elections. An approximate 95 percent confidence interval for the Venezuelan governing-party vote share in 1998 is from 9 percent to 37 percent, with a point prediction at 19 percent. Once again, this interval seems somewhat too high. While unemployment is clearly a useful predictor of governing-party vote share in Latin America, it does not improve our ability to understand the electoral outcomes involved in party-system collapse.

economics, societal crisis, and anxiety

79

A final question worth raising with respect to predicting governing-party vote shares is whether the patterns discussed in this chapter differ in partial and full democracies. In the Mainwaring, Brinks, and Pérez-Linán (2001) index of democracy in Latin America, Peru in 1990 ranks as a partial democracy, while the other party-system collapse elections occur in fully democratic regime contexts. Hence the question of whether level of democracy changes any of the patterns found above is relevant to understanding party-system collapse. However, statistical analysis suggests that regime type does not meaningfully affect the patterns studied here. A hypothesis test of an expanded model with a linear effect for level of democracy as well as all possible interactions between level of democracy and other explanatory variables has a P value of 0.29.9 Clearly, the interaction terms make an insignificant explanatory contribution; differences between partial and full democracies may be disregarded for present purposes. In summary, then, none of the statistical models considered here adequately accounts for the changes in governing-party vote share that occurred during party-system collapse in Peru and Venezuela. These explanatory failures suggest either that party-system collapse is a distinctive outcome in aggregate, crossnational terms or that the models explored above have omitted some relevant variables—or perhaps most likely, both. Additional variables, either in and of themselves or in interaction with the explanatory factors, are needed to account for the steeper-than-expected decline of governing parties in Peru in 1985 and in Venezuela in 1993 and 1998—as well as the unexpectedly good performance of APRA in the 1990 Peruvian elections. Relevant variables may be political in nature, involving concerns about corruption, party identification, and the ideological positioning of the traditional parties. Omitted variables may also include subjective perceptions of the factors considered here; voters’ perceptions of the economy need not simply reflect their country’s regional rankings, as in this comparative discussion. Furthermore, similar analysis of the vote share of the leading opposition party shows that the variables considered in this section have essentially no predictive power. For example, a regression using the political and economic variables has an R 2 of only 0.126 (as compared with an R 2 of 0.681 when the dependent variable is governing-party vote share) and none of the coefficient estimates is statistically significant. Evidently, governing parties systematically lose votes when they perform badly—but the destination of those votes is somewhat unclear.

80

economics, societal crisis, and anxiety

3 . 4 E C O N O M I C C R I S I S A N D A N X I E T Y A B O U T T H E C O U N T RY

If economic crisis does not provide a complete, direct explanation of partysystem collapse, it is nonetheless unlikely that it is causally irrelevant. The preceding analysis strongly suggests that poor economic performance weakens governing parties, even if it is usually insufficient to produce their disappearance or to guarantee that they are replaced by candidates from nontraditional parties. Poor or even disastrous economic performance plays a second relevant role in accounting for party-system collapse: as shown below, it substantially increases citizens’ anxiety levels. Research in political psychology provides guidance and a rich array of findings regarding the political effects of anxiety. Anxiety increases people’s attention to political information (Marcus, Neuman, and MacKuen 2000; Tiedens and Linton 2001; Valentino, Hutchings, Banks, and Davis 2008; Parker and Isbell 2010). It also leads citizens to devote more attention to challengers to the political status quo, and increases voters’ willingness to revisit and perhaps revise their political attachments (Marcus, Neuman, and MacKuen 2000; ­Isbell, ­Ottati, and Burns 2006). In all of these ways, anxiety about the social and political world is an important precursor to party-system collapse. How, then, is economic crisis related to rising anxiety levels? This question can usefully be subdivided into two components. First, is there an empirical relationship between bad economic performance at the aggregate level and average levels of individual anxiety about society and the future—and if so, how strong is the connection? Second, which economic variables play the strongest role in producing anxiety? Do national-level factors matter most, or are pocketbook variables more important? Does inflation or economic growth play more of a leading role? With respect to the second set of issues, the most plausible expectation is that all of the mentioned variables matter. We expect most people to become anxious when their personal or family economic situation threatens their ability to maintain their standard of living. Furthermore, individuals who are attentive to the broader economy are likely to become anxious in times of crisis because they can reasonably conclude that the crisis will eventually affect them. Poor or negative economic growth rates can threaten individuals’ sense of security by raising the probability that they will lose their jobs, and very high

economics, societal crisis, and anxiety

81

levels of inflation could wipe out savings and erode the purchasing power of an individual’s wage or salary. While theoretical reflection suggests that all of these variables can be expected to raise people’s anxiety, the same considerations imply that some should have stronger effects on average than others. For example, direct damage to a household’s economic situation requires very little political or broader economic awareness to detect; it is experienced in everyday ways. As a result, the personal experience of decline is likely to produce a higher level of anxiety than general societal economic trouble, even among people who are not politically or economically attentive. Hence, personal and household variables should matter most in producing anxiety, even though societal variables should also matter. Likewise, high levels of inflation are rather more universal in their effects than is low or negative economic growth. Economic contraction is most intensely damaging to individuals who lose their jobs as a result; such people can be expected to experience a major increase in anxiety, but that increase is produced by the individual consequences of crisis, rather than by the general societal effects. For those who do not lose their jobs, there can also be an increase in the perception of threat, but that increase is somewhat diluted by the reality that they are still employed. By contrast, hyperinflation affects society as a whole and can therefore be expected to have a broader effect than poor levels of growth. These theoretical expectations are tested using data from the 1989–93 wave of the World Values Survey. This survey wave includes a question that measures an individual’s overall level of anxiety about future changes. Respondents are asked to position themselves on a one-to-ten scale between these statements: “When changes occur in my life, I worry about the difficulties they may cause,” and “When changes occur in my life, I welcome the possibility that something new is beginning.” The first statement is a reasonable description of anxiety, and indeed includes a common synonym for that emotional state in the word “worry.” The statement forming the second pole of the scale is a general statement of both optimism and risk acceptance, both of which are reasonable affective antonyms for anxiety. In testing the relationship between this measure of anxiety and economic performance, five different economic variables are used. The first is an individual’s household income, operationalized as the decile that household falls

82

economics, societal crisis, and anxiety

into within the society’s broader income structure. Age is included as a proxy for the respondent’s position in the economic life cycle. Three macroeconomic indicators related to the country-year of each survey’s administration are also included: measures of inflation, economic growth, and per capita GDP (all drawn from the World Development Indicators). To accommodate the structure of the data, in which individual responses are nested within several country-specific surveys, a multilevel model is adopted, in which country-level random coefficients are estimated for the intercept as well as all individual-level variables (see, e.g., Gelman and Hill 2006). The analysis is presented in three parts. First, a linear model is used to predict respondents’ satisfaction with their household financial situations on the basis of objective economic indicators. Subsequently, a second linear model predicts respondents’ anxiety levels based on their financial situation and the objective indicators. Finally, anxiety and the more directly economic variables are used in a generalized logit model to predict respondents’ satisfaction with their political systems. The first stage of the analysis, reported in Table 3.3, shows that all of the economic variables matter in determining individuals’ satisfaction with their household financial situation. National wealth is a significant positive predictor, although a household’s position within the national system of inequality is an even better explanatory variable. High levels of inflation produce economic

ta b l e 3 . 3 . The economy and satisfaction with

household financial situations Variable Name Intercept Age Household income Inflation (logged) Per capita GDP (logged) Economic growth rate

Fixed Effect Estimate 2.749 (0.836)** 0.015 (0.003)** 0.870 (0.066)** – 0.255 (0.074)** 0.169 (0.080)* 0.041 (0.015)*

notes: Hierarchical linear model with random effects for the intercept and all individual-level variables (effect standard errors and covariances not reported). *(p < 0.05) **(p < 0.01) source: Data from the 1989–93 wave of the World Values Survey; 42,056 individuals are included in the analysis, grouped within thirty-four countries.

economics, societal crisis, and anxiety

83

dissatisfaction, while growth—to a somewhat lesser extent—increases individuals’ satisfaction with their situation. All of this fits with the theoretical expectations developed above, and indeed is not too surprising; it simply confirms that people’s feelings about their own finances are influenced both by macroeconomic developments and by more personal economic outcomes. Nonetheless, these findings are useful in that they suggest that pocketbook economic evaluations in fact reflect a blend of personal and broader societal outcomes. Furthermore, as will be seen below, the subjective summary of national and personal economic conditions captured in a respondent’s degree of satisfaction with her household’s financial situation captures the majority of the connection between economic indicators and the sequence of politically relevant attitudes. Hence, establishing a connection between inflation, in particular, and economic growth to a lesser extent, and household financial satisfaction is important to the overall argument. Turning to a direct analysis of individuals’ reported levels of anxiety about the future, we see that wealthier people and people who are more satisfied with their personal financial situations are meaningfully less anxious than are less wealthy or less satisfied individuals (see Table 3.4). After accounting for these individual-level economic variables, none of the macroeconomic variables are statistically significant, although economic growth approaches significance and ta b l e 3 . 4 . The economy and anxiety

about the future Variable Name Intercept Age Household income Financial satisfaction Inflation (logged) Per capita GDP (logged) Economic growth rate

Fixed Effect Estimate 3.793 (0.559)** 0.026 (0.002)** – 0.163 (0.028)** – 0.053 (0.009)** – 0.011 (0.051) – 0.024 (0.054) – 0.018 (0.010)

notes: Hierarchical linear model with random effects for the intercept and all individual-level variables (effect standard errors and covariances not reported). *(p < 0.05) **(p < 0.01). source: Data from the 1989–93 wave of the World Values Survey; 40,669 individuals are included in the analysis, grouped within thirty-four countries.

84

economics, societal crisis, and anxiety

two of the three coefficient estimates are in the expected direction. This suggests that individual economic experiences are a more powerful determinant of overall anxiety levels than are societal experiences. However, the macroeconomic variables still indirectly influence individuals’ level of financial satisfaction. Overall, then, these models suggest that economic crisis can meaningfully raise a society’s average level of anxiety. Anxiety, in turn, may play an important facilitating role for party-system collapse, as discussed above: it may increase people’s attention to political information and openness to discovering new partisan alternatives. These hypotheses cannot be given a satisfactory test using the World Values Survey data; however, an alternative dependent variable is available that may reflect a deeper level of concern about the political system as a whole. Anxiety may help establish the initial conditions for the destabilization of a country’s party system and its established political system more generally. In fact, as Table 3.5 shows, there is some reason to believe that large increases in a society’s average level of anxiety may help generate a broader crisis with respect to citizen attitudes toward the political regime. The dependent variable in this analysis is a survey question, treated here as an indicator of overall satisfaction with the political system, asking respondents whether the country they live in is “run by a few big interests looking out for themselves,” or if it, by contrast, is “run for the benefit of all the people.” Controlling for either ta b l e 3 . 5 . Anxiety and satisfaction with the political system Variable Name Intercept Anxiety Age Household income Financial satisfaction Inflation (logged) Per capita GDP (logged) Economic growth rate

Fixed Effect Estimate I

Fixed Effect Estimate II

– 0.258 (0.373) – 0.029 (0.014)* 0.001 (0.002) – 0.011 (0.014) 0.001 (0.011)

5.232 (1.655)** – 0.029 (0.014)* 0.002 (0.003) – 0.037 (0.018) * – 0.003 (0.012) 0.024 (0.143) – 0.633 (0.161)** 0.055 (0.027)*

notes: Hierarchical generalized logit model with random effects for the intercept and all individuallevel variables (effect standard errors and covariances not reported). *( p < 0.05) **(p < 0.01). source: Data from the 1989–93 wave of the World Values Survey; 7,283 individuals are included in the first analysis, grouped within twelve countries. For the second analysis, 6,859 individuals are included, grouped within ten countries.

economics, societal crisis, and anxiety

85

the individual-level economic variables or both the individual- and countrylevel indicators, anxiety has a statistically significant negative relationship with belief that the country is run for all citizens’ benefit. More broadly, this model suggests that the effects of economic crisis on attitudes toward the political regime are partly channeled through emotional states. Three economic indicators are included as predictors in the model: inflation, growth, and level of GDP. While both growth and level of GDP have direct effects on the dependent variable, inflation does not. All three variables, however, have indirect effects via anxiety—although multiplication of effect sizes across models suggests that these indirect effects are substantively modest. Overall, these results suggest that economic crisis can be an enabling cause of party-system collapse, in part by raising individuals’ levels of generalized anxiety about the future.

3.5 NEO-LIBERAL ECONOMIC REFORMS

Given these results connecting economic crisis with anxiety among voters, and because anxiety leads voters to gather more political information, to potentially revisit standing partisan decisions, and to experience lower levels of satisfaction with the political system, economic trouble should be regarded as a potentially key initiating step in the process of party-system collapse. Yet countries where economic crisis increases average levels of anxiety do not necessarily proceed to experience party-system collapse. What causes some electorates to respond to crisis by voting all but exclusively for political outsiders, while other electorates respond by turning to the opposition within the traditional party system? Some explanation other than economic performance is clearly needed. Some scholars have argued that the process and/or outcomes of neoliberal economic reform provide that explanation, but these accounts either face the same challenges as the economic-crisis explanation examined above or they become an important component of the larger set of considerations related to party ideological positioning. Aside from party-system changes, one of the highest-profile transformations in Latin America between 1980 and 2000 was the widespread adoption of neo-liberal economic reforms (Przeworski 1991; Haggard and Kaufman 1992, 1995; Weyland 2002). A continent that had once been characterized by

86

economics, societal crisis, and anxiety

interventionist, nationalist economic policies underwent a dramatic process of opening to globalization and of reducing state presence in the economy. In light of this economic transformation, it is an intuitively plausible hypothesis that neo-liberal reform is causally connected with the equally remarkable outcome of party-system collapse. In particular, party-system collapse could be seen as a consequence of the economic dislocations that result from the failure of labor-mobilizing parties (Acción Democrática in Venezuela and presumably APRA in Peru) to adjust successfully to the politics of marketizing reform and economic openness (Roberts 1996: 92–96, 1997, 2003; Levitsky 2003: 233–38). Parties may have failed either by refusing to implement neo-liberal reforms at all or by implementing them badly. Either form of failure could have generated massive economic crises. These crises, in turn, could explain why Peruvian and Venezuelan voters turned to support politicians from outside the traditional party system. At the level of individual voters, then, the hypothesis that party-system collapse is due to incumbents’ failure to produce the economic benefits of neoliberal reform involves a mechanism of economic voting. Specifically, voters move away from the traditional parties because they are dissatisfied with the country’s economic performance. There is some partial evidence in favor of this hypothesis; in a specialized sample of dwellers in the shantytowns of Lima, Peru, 37.9 percent of voters who abandoned the traditional party system in 1990 said they did so because of “bad government and/or the economy” (Dietz 1998: 212). However, the analysis above suggests some difficulties with this view, which are briefly elaborated on in subsequent chapters. Specifically, economic performance in Peru and Venezuela was not bad enough, in regional comparative perspective, to fully account for a party-system collapse, and some countries had comparable or worse performance without experiencing a collapse. Hence, an explanation that attributes party-system collapse to the economic effects of poor, failed, or delayed implementation of neo-liberal economic reforms is problematic. A second possibility, which has been relatively neglected in the literature on party-system collapse, is that economic and international pressures in favor of neo-liberal reform affected party trajectories by imposing ideological trade-offs. In this account, parties’ stances on neo-liberal reforms have political and electoral consequences in their own right, aside from their macroeconomic conse-

economics, societal crisis, and anxiety

87

quences (see, e.g., Oxhorn 1998). If parties acquiesce to the pressures in favor of marketizing reform, that pushes them toward the ideological right; by contrast, a decision to explicitly resist neo-liberal reforms may push parties toward the left. Hence, the increasing salience of neo-liberal reform as a political issue after 1980 may make a centrist (or indeed a center-right or center-left) ideological profile more difficult to maintain. The evidence presented in subsequent chapters that party ideological positioning plays a greater role than economic results in producing party-system collapse does not prove that neo-liberal reform was irrelevant to the process. However, it does suggest that the relevant variable was not successful implementation of reform but rather ideological strategy in response to the neo-liberal juncture. Chapter 5 devotes considerable attention to the role of party ideological positioning in party-system collapse; first, however, it is necessary to account for the chronologically prior decline of identification with the traditional parties of Peru and Venezuela. This puzzle requires close consideration of the effects of corruption perceptions on voters in Venezuela and Peru during roughly the decade leading to collapse.

chapter 4

CORRUPTION AND THE COLLAPSE OF PA RT Y I D E N T I F I C AT I O N

I

f , a s t h e p r e v i o u s chapter argued, party-system collapse cannot be adequately accounted for by economic crisis alone, what accounts for the changes in mass electoral decision-making that underpin the major party-system changes leading to the elections of outsiders Hugo Chávez in Venezuela and Alberto Fujimori in Peru? This chapter examines an early causal step in the process of collapse: decline in identification with the traditional parties. Party identification is not the same as vote choice, and a decline in identification preceded the Peruvian and Venezuelan traditional parties’ electoral decay. Hence, there is some reason to consider a decline in party identification as a cause of party-system collapse, not merely a component of it. Nonetheless, a decline in identification with the traditional parties is causally proximate enough to party-system collapse that in this chapter I treat change in partisanship as an outcome, as well as, in later chapters, as a potential cause. In fact, survey evidence suggests that decline in traditional party identification in Peru and Venezuela results from persistent problems of corruption, suggesting that party identification is eroded by some of the same political anger that eventually leads voters to support candidates from outside the traditional party system.

cor ru p t ion a nd col l a pse of pa rt y iden t ific at ion

89

While testing the hypothesis sketched above—in which the decline in party identification before party-system collapse is due to anger and dis­ illusionment connected with concerns about corruption—this chapter also considers several alternative hypotheses—in particular that poor economic performance or policy drove voters away from identification with the traditional parties. Variables connected with these accounts show little or no relationship with changes in patterns of party identification, conditional on views regarding corruption.

4 . 1 PA RT Y I D E N T I F I C AT I O N A S A C A U S E

As an independent variable predicting vote choice—i.e., a potential causal obstacle to party-system collapse—party identification is theoretically and empirically well established. Ever since the publication of The American Voter (Campbell, Converse, Miller, and Stokes 1960), party identification has been a required component of any attempt to explain vote choice in American elections. Discussions of partisan decline notwithstanding (Dalton, Flanagan, and Beck 1984; Wattenberg 1996; Dalton and Wattenberg 2000), the explanatory power of party identification in the U.S. context is still remarkable (Bartels 2000). Furthermore, comparative research has shown party identification to have substantial explanatory power in other political contexts (e.g., Dalton 2002: 172–94; Green, Palmquist, and Schickler 2002: 166–69; Sanders 2003). Two major theories explain the mechanism through which party identification affects vote choice. Party identification may work as a social identity; citizens may feel pressure to behave in accordance with the rest of their partisan social reference group by voting for the candidate supported by that group. Alternatively, party identification may serve as a cognitive heuristic allowing citizens to make rational electoral decisions with a minimum of mental calculation. Regardless of how party identification operates on vote choice, however, it is clear that a substantial erosion of party identification is an almost essential precursor to party-system collapse. If a sizable proportion of the electorate identifies with the traditional parties, and therefore experiences important causal pressure to keep voting for those parties, it becomes perhaps insurmountably difficult for nontraditional parties to displace the established system. Maintained partisanship is usually too powerful a force for that to happen.

90

cor ru p t ion a nd col l a pse of pa rt y iden t ific at ion

4 . 2 P A RT Y I D E N T I F I C AT I O N A S A N E F F E C T

In light of this strong theoretical expectation, as well as the conceptual proximity between decline in party identification and party-system collapse, it is reasonable to ask under what circumstances identification with the traditional parties might decline to such an extent that party-system collapse is a possibility—temporarily turning party identification into an outcome to be explained. A long-standing, successful research tradition based primarily in studies of the United States explains change in party identification largely in terms of generational replacement (Campbell et al. 1960; Carmines and Stimson 1987: 142–54; Miller and Shanks 1996). In this view, partisan loyalties are simply too strong to be severed by most kinds of political forces and events. Hence, change in the distribution of partisanship within a society mostly takes place as older, stable partisan generations are replaced by new generations socialized under different circumstances. Green, Palmquist, and Schickler (2002) have suggested an alternative formulation, within the social psychological tradition, of the conditions under which party identification might change during an individual’s lifetime. In this view, party identification is underpinned by stereotypes about which primary social groups (e.g., Catholics, unionized workers, environmentalists, mothers) support each political party. An individual’s party identification may change either if her primary social group identification changes—a relatively rare event—or if her stereotypes change about which social groups support which parties. New political issues, or new approaches by party leaders to established issues, might eventually alter stereotypes in such a way as to lead to partisan realignment. On the other hand, it seems plausible to extend the ­hypothesis by proposing that events which make the political parties look purely selfserving and disconnected from all social groups—particularly corruption scandals—might lead to the kind of across-the-board decline in identification with established parties involved in party-system collapse. The major alternative perspective on change in party identification draws on the view of party identification as a cognitive heuristic, as mentioned above. In this view, party identification is a kind of “running tally” of positive and negative aspects of each major traditional party, with identification assigned to the party whose tally is currently most desirable (Fiorina 1981; Achen 1992). Party identification should be affected by corruption scandals, but also by poor

cor ru p t ion a nd col l a pse of pa rt y iden t ific at ion

91

economic performance or by disagreement with the party’s policy efforts and issue stances. Furthermore, what is needed in order to produce a generalized decline in party identification is simply simultaneous poor performance by all of the established political parties. Arguably, this account leads to a somewhat more flexible prediction of how party identification could change during the process of party-system collapse. Yet this difference in flexibility is not especially central to the theoretical agendas of either approach; hence, testing the relative importance of corruption as against economic perceptions will not provide major leverage in comparing these theoretical traditions as a whole. In considering these three perspectives on change in partisanship, there is some modest divergence in implications for understanding party-system collapse. If party identification changes through generational replacement, then the decline in party identification before the collapse should be gradual and indeed should take place over the course of thirty or more years. Either of the other perspectives allows for much more rapid change. If partisan decline is driven by situations that make established parties look purely self-serving, then corruption perceptions should be associated with decline in party identification. Similarly, if party identification is a retrospective evaluation of party behavior more generally, concerns about corruption should also be associated with partisan decline.

4 . 3 T H E E V O L U T I O N O F PA RT Y I D E N T I F I C AT I O N IN SOUTH AMERICA

Aggregate trends in party identification, sometimes referred to as macro-partisan­ ship, have been the subject of intense scrutiny in the context of the United States,1 and have also received attention in Europe (Schmitt and Holmberg 1995; Dalton 2000). However, aggregate trends in party identification have received much less attention in Latin America. A thorough analysis of trends in macro-partisanship would be beyond the scope of this study; nonetheless, beyond their primary role in framing the analysis below, the brief descriptions in this section may serve as an introduction to the scope of the puzzle.2 Figure 4.1 depicts the evolution over time of identification with the two traditional political parties in Venezuela. Identification with COPEI dropped dramatically from 1978 until 1988, recovered somewhat by 1993, and then fell

92

cor ru p t ion a nd col l a pse of pa rt y iden t ific at ion

f i g u r e 4 . 1 . Identification with the traditional parties in Venezuela. source: Data from 1978 and 1983 Gallup pre-electoral surveys, the 1988 DATOS pre-electoral survey, the 1993 CIEPA survey, and the 1998 RedPol electoral survey, all archived at the Banco de Datos Latinoamericano de la Universidad Simón Bolívar, Caracas, Venezuela.

to near-zero levels by 1998. Identification with Acción Democrática peaked in 1983, then fell continuously from that year until the end of the time series in 1998, when only 10 percent of the electorate identified with the party. This roughly simultaneous collapse of identification with both of the Venezuelan traditional parties—at a time before the parties’ vote shares began to collapse—most probably played an important causal role in the collapse of the traditional party system, a hypothesis that is explored in later chapters. Indeed, the collapse of party identification is so striking that, in addition to considering it as an independent variable, the present chapter uses party identification as a dependent variable, arguing that corruption, more than economic outcomes or economic policy, had a major role in eroding identification with the traditional parties.

cor ru p t ion a nd col l a pse of pa rt y iden t ific at ion

93

The data on identification with the Peruvian traditional parties, presented in Figure 4.2, shows an equally impressive collapse. However, unlike the Venezuelan case, the evidence shows the Peruvian electorate as having never really identified to any substantial degree with any political party other than APRA. The most interesting trend in Peruvian party identification is almost certainly the decline in identification with APRA, from 53 percent of Peruvians in 1986 to about 7 percent in 1989. It is important once again to point out that this decline preceded, and indeed was somewhat more complete than, the party’s electoral collapse. The data presented so far on Peruvian and Venezuelan party identification may seem to support the stereotype that Latin American political parties are inadequately institutionalized and lack real connections with the electorate. Therefore, it may be useful to note that not all Latin American parties lost

f i g u r e 4 . 2 . Identification with the traditional parties in Peru. source: Data from the Apoyo surveys of public opinion in Peru, published in Informe de Opinion (Lima: Ipsos Apoyo).

94

cor ru p t ion a nd col l a pse of pa rt y iden t ific at ion

their identification base during the 1980s and 1990s. Consider identification with the Argentine traditional parties.3 Identification with the Peronist party has been remarkably stable through a decade (1992–2002) filled with crisis and turmoil of a political, social, and economic nature, always falling between 23 and 28 percent of respondents in the greater Buenos Aires metropolitan area.4 Radical party identification did largely collapse over that same period, falling from 18 percent in 1992 to less than 10 percent in 2003. Nonetheless, patterns of identification with the party system as a whole have not been as volatile as those in the Peruvian system, where identification with the strongest traditional party collapsed, or in the Venezuelan system, where identification with all traditional parties essentially vanished. As a more distant comparison case, patterns of identification with the major Chilean political parties were largely stable between 1990 and 1995,5 providing a useful reminder that some Latin American party systems are able to maintain their identifiers, even in the neo-liberal era. Thus, overly broad global or regional factors are not useful in accounting for the collapse of identification with traditional parties. In particular, the neo-liberal juncture has not obviously destroyed all social identities connected with the Peronist party in Argentina or various parties on the left, center, or right in Chile. Overall, there is a sharp contrast between the evolution of party identification in Peru and Venezuela on the one hand, and Argentina and Chile on the other. It would seem that, before party systems collapse electorally, voters’ identification with particular parties collapses. In each case, declines in identification preceded electoral declines; in Venezuela, in particular, the decline in identification with the traditional parties began at least a decade before those parties lost a significant share of the vote. Thus, collapse of party identification is not equivalent to party-system collapse. Yet the link between the two is intuitive enough—and the party identification data striking enough—that I give this variable central attention in the consideration below of survey evidence on the causes of voter defection from the traditional party system. Overall, decline in party identification functions in this study as an intervening variable, caused substantially by concerns about corruption and serving to causally connect such concerns with the ultimate decision to vote against the traditional parties.

cor ru p t ion a nd col l a pse of pa rt y iden t ific at ion

95

4 . 4 E X P L A I N I N G D E C L I N E S I N P A RT Y I D E N T I F I C AT I O N

While the prevalence of party identification in the electorate seems to be declining in many parts of the democratic world (Dalton 2002), the declines experienced by several of the collapsed parties of Peru and Venezuela have been exceptionally sharp and sudden. Why did large numbers of voters cease to identify with the predominant parties of their countries? Two hypotheses are drawn directly from the major alternative hypothesis of the book as a whole: the proposition that concerns about the economy offer the most powerful direct explanation of the phenomenon of party-system collapse. Perhaps voters ceased identifying with the traditional parties as a retrospective judgment on the economic disasters experienced in Peru and Venezuela during the 1980s and 1990s. However, the aggregate cross-national data presented in Chapter 3 showed that other countries in Latin America, including Argentina, have had economic crises of comparable magnitude without experiencing a total erosion of party identification. There is some reason to be skeptical of economic evaluations as an explanation of decline in traditional party identification. A second major independent variable in the analysis below relates to the primary alternative hypothesis: opinion with respect to neo-liberal economic policy. Partisans who disagree with neo-liberal policy may have become disenchanted with their parties when (in Argentina and Venezuela in particular) those parties took the historic step of implementing neo-liberal reforms (Morgan 2007: 81–83). Citizens may sometimes perceive a connection between neo-liberal reforms and economic outcomes, which are treated as separate independent variables in this analysis. Some individuals may see neo-liberal reforms as the cause of poor economic performance, while others may see such reforms as the solution to economic trouble. Overall, the direction of the connection between these two factors is unclear. Empirically, omitting either variable from the analysis below does not substantially change the results. Alternatively, voters may have ended their identification with the traditional parties because corruption scandals tainted the parties’ images, creating a perception that the traditional parties really only represented themselves. This is the central hypothesis defended below. In particular, it will be shown that concerns about corruption are more powerful predictors of a decision to abandon

96

cor ru p t ion a nd col l a pse of pa rt y iden t ific at ion

identification with the traditional parties than are concerns about economic policy or performance. In passing, it is worth noting that one other factor has been offered as a potential explanation for the fall of identification with traditional parties in South America: ideology. Drawing on data from the 1998 election in which Venezuela’s party system finally collapsed, Morgan (2007: 83–94) argues that ideological factors account for the drop in identification with traditional parties in that country. Unfortunately, her cross-sectional analysis cannot distinguish between those factors that structured the Venezuelan system of party identification during the period of its strength, and those that explain the collapse of that system. In particular, the left-right ideological dimension that Morgan treats as an explanation for collapse was already built into the Venezuelan system of party identification as early as 1983. As Figure 4.3 shows, even at the height of the Venezuelan traditional party system, ideological rightists were far more likely to identify with the Venezuelan traditional parties than centrists, leftists, or the nonideological. Furthermore, the decline in party identification

f i g u r e 4 . 3 . Venezuelan party identification by ideology in 1983 and 1993. source: Data from the 1983 BATOBA survey (Baloyra and Torres 1983), and the 1993 Doxa survey (Centro de Investigaciones y Estudios Politicos y Administrativos, N.d.).

cor ru p t ion a nd col l a pse of pa rt y iden t ific at ion

97

before collapse appears to have happened to more or less the same extent within each ideological category. Morgan’s analysis thus points to an important longstanding feature of how Venezuelans relate to political parties, rather than an explanation for the decline of party identification—although the ideological variable she highlights does play an important role in explaining the vote choices that ultimately produced collapse. Hence, ideology is treated in this section as a structural feature of Venezuela’s party system that helps us predict which individuals in 1993 survey data would probably have identified with a traditional party in 1983—rather than an explanation for change in identification with traditional parties. However, as will be seen below, this structural feature of the party system is a weakness that was central to explaining the process of electoral change—if not change in party identification—that finally produced party-system collapse. In summary, this section focuses on the role of economic evaluations, positions on issues related to neo-liberal reform, and perceptions of corruption in explaining decline in identification with the traditional parties as a precursor to party-system collapse. The evidence shows that corruption perceptions are a more powerful explanation than the other two factors. First, however, a more detailed descriptive understanding of the extent of corruption problems in Peru and Venezuela is needed. 4.4.1 Corruption Perceptions During the period of time just before party-system collapse, both Peru and Venezuela experienced serious corruption scandals. Scholars have proposed a connection between these corruption crises and party-system collapse in both Peru (Lynch 1999: 185, Kenney 2004: 71) and Venezuela (Coppedge 2005: 311–12). Yet these arguments are based centrally on case studies. How does the level of corruption in Peru and Venezuela compare with that in other Latin American countries? If corruption were a stand-alone cause of party-system collapse, rather than one component of a more interactive cause, then the two countries where a collapse occurred would fall toward the upper end of the observed range on indicators of corruption. By contrast, if corruption is an enabling condition rather than a direct cause of electoral breakdown, it is instead important merely that Peru and Venezuela have experienced at least a reasonable degree of corruption; other causally later components of the story

98

cor ru p t ion a nd col l a pse of pa rt y iden t ific at ion

can differentiate the two collapse countries from other countries with moderate to high levels of corruption. Four different measures of corruption are available for at least some Latin American countries during different parts of the period from 1980 to 2003. Mainwaring and Pérez-Linán (2005) have coded reports in the Latin American Weekly Report (LAWR), producing an annual count of corruption scandals. For cross-validation purposes, I have conducted a similar coding of reports in Keesing’s Record of World Events. These two publications provide independent measures of the extent of media coverage of corruption in Latin American countries. Transparency International’s Corruption Perceptions Index (CPI ) provides an aggregation of survey data on concerns about, and experience of, corruption in Latin American countries starting in 1995 (Transparency International 2005). The 1995–97 wave of the World Values Survey (WVS ) (Inglehart et al. 1999) provides a second, independent measure of corruption perceptions for Argentina, Brazil, Chile, Colombia, the Dominican Republic, Mexico, Peru, Uruguay, and Venezuela.6 As might be expected, the measures of media coverage of corruption and the measures of corruption perceptions form somewhat distinct conceptual blocks.7 The LAWR and Keesing’s scandal count indices correlate at 0.38, while the two corruption perceptions indices correlate at 0.88. No other pairwise correlation among these indices is as high. Corruption perceptions are probably of more direct interest in considering party-system change; however, the measures of media coverage are available for a greater time period. Of the two measures of media coverage, the coding based on the LAWR has a somewhat higher correlation with the two measures of corruption perceptions than does the coding based on Keesing’s. Therefore, the LAWR variable will be the primary indicator of corruption in this chapter. In spite of these problems of data availability, it is useful to look at the existing corruption perception scores before turning to the somewhat lower quality, but more widely available, measures of number of corruption scandals. The first column in Table 4.1 shows the average corruption ratings on the CPI scale (which is closely correlated with the WVS data for the points where the two overlap). The overwhelming impression from these data is that the electorates in almost all Latin American countries perceive extensive problems of corruption. Venezuela falls toward the top of the range of corruption perceptions, whereas

cor ru p t ion a nd col l a pse of pa rt y iden t ific at ion

99

ta b l e 4 . 1 . Corruption in Latin American

countries, 1983–2001 Country

CPI

LAWR

Peru Venezuela

5.8 7.4

25 17

Argentina Bolivia Brazil Chile Colombia Costa Rica Domincan Republic Ecuador El Salvador Guatemala Honduras Mexico Nicaragua Panama Paraguay Uruguay

6.7 7.5 6.3 2.9 6.9 4.9 6.7 7.5 6.3 7.2 7.8 6.7 7.3 6.6 8.3 5.2

44 16 36 9 29 9 3 27 9 10 12 20 11 7 13 3

source: CPI data are from the Transparency International Corruption Perceptions Index. LAWR data are from Mainwaring and Pérez-Linán’s coding of the Latin American Weekly Report (Mainwaring and Pérez-Linán 2005).

Peru is somewhere near the middle. The time period for these data make it difficult to draw direct connections with party-system collapse; the data for Peru in particular do not even begin until substantially after the collapse occurred. The second column in Table 4.1 shows the LAWR data for eighteen countries, presenting an overall count of corruption scandals between 1983 and 2001. In the figure, Peru seems to score somewhat higher on corruption scandals than Venezuela, but this is at least partly misleading. The data, after all, include scandals that took place after the Peruvian party-system collapse in 1990. There were eleven reported scandals in Peru between 1980 and 1990; projecting this same rate of scandals forward would give a twenty-two-year estimate of twenty-two total scandals, rather than the thirty-two actually observed. Hence, for present purposes, Peru and Venezuela can be treated as having had essentially the same number of corruption scandals before party-system collapse.

100

cor ru p t ion a nd col l a pse of pa rt y iden t ific at ion

Peru and Venezuela, the two countries where the party-system collapsed, fall in roughly the upper third of these Latin American countries in overall scandal count. Argentina, Brazil, Colombia, Ecuador, and Mexico all score higher than Venezuela, at least, and Bolivia scores very nearly as high. All of these countries have experienced party-system instability of one form or another in recent years—but in many of them instability coincided with the survival, though not always the triumph, of at least one traditional party. Thus, the cross-national pattern is consistent with this book’s argument in favor of treating corruption concerns as an enabling condition for party-system collapse. Turning briefly to the pattern of corruption scandals over time in Peru and Venezuela, both countries went through a fairly steady stream of scandals until shortly before the party-system collapse.8 During the final few years before the collapse, the rate of corruption scandals dropped somewhat in both countries. Nonetheless, in each country at least one scandal occurred during these years, so corruption concerns would not have dissipated entirely. In general, the aggregate data reviewed here are compatible with the hypothesis that corruption helps explain party-system collapse via its effects on party identification—although perhaps not with the idea of corruption as a direct or primary explanation. To resolve the question of whether corruption does in fact help explain the decline of party identification before collapse, it is necessary to analyze individual-level data. Are corruption perceptions at the individual level associated with a movement away from identification with the traditional parties?

4 . 4 . 2 I D E N T I F I C AT I O N W I T H V E N E Z U E L A’ S T R A D I T I O N A L PA RT I E S

What, then, is the evidence that corruption perceptions played an important role, especially in comparison with concerns about economic policy or performance, in bringing about the observed decline in identification with Venezuela’s traditional parties? As seen in Figure 4.1, the Venezuelan traditional parties, AD and COPEI, appear to have peaked in their ability to predominate in the domain of party identification during the late 1970s and early 1980s. During earlier periods, the parties’ combined total of identifiers was less than half the electorate (Martz 1977). But in 1978 and 1983, AD and COPEI identifiers

cor ru p t ion a nd col l a pse of pa rt y iden t ific at ion

101

combined were at least two-thirds of the electorate. By 1998, however, identifiers with these two parties accounted for merely 12 percent of the electorate. The speed and thoroughness of this change in party identifications means that standard ideas about identification changing primarily through generational replacement are incorrect or at least incomplete. After all, voters who were 30 years old in 1978, at the peak of the Venezuelan traditional party system, would have been only 50 in 1998, at the time of its demise. Hence, generational turnover seems as if it must logically be a relatively marginal aspect of this change in patterns of party identification. Furthermore, no influx of newly enfranchised voters could account for the change; Venezuela had universal suffrage and nearly universal voter turnout in 1978. A substantial number of Venezuelans must have revised their party identifications during the period under consideration. What, then, does explain this major decline? Data from the BATOBA 83 survey (Baloyra and Torres 1983), administered to 1,500 Venezuelan adults making up a nationwide sample during October of 1983, and the DOXA 93 survey (Centro de Investigaciones y Estudios Politicos y Administrativos n.d.), administered to a nationwide sample of 1,500 Venezuelan adults in May and June of 1993, permit empirical tests of the corruption perceptions hypothesis, drawn from this volume’s theoretical account of party-system collapse, as well as two alternative hypotheses, involving instead negative economic evaluations or opposition to neo-liberal economic reforms. The BATOBA 83 data provide the information necessary to impute estimates of 1983 traditional party identification into the 1993 data.9 Six variables that are stable over time are used to connect the 1983 survey wave with the 1993 wave by identifying respondents’ likelihood of having identified with a party in the earlier time period: age cohort, ideology, income category, education, housing quality, and reported parents’ party identification.10 Age cohort, defined in terms of the political regime in place when the respondent turned 18 and came of political age, is used in this analysis because reported age almost certainly meets the stability assumptions for the pseudo-panel analysis, and because there is substantial reason to believe that age cohorts differ systematically in patterns of party identification (Miller and Shanks 1996). For most adult respondents, education level should also remain stable. Furthermore, education is a powerful predictor of most political identities and attitudes (Nie, Junn, and Stehlik-

102

cor ru p t ion a nd col l a pse of pa rt y iden t ific at ion

Barry 1996). Income category and housing quality are probably less stable over time than education or age cohort. Nonetheless, in a relatively class-immobile society like Venezuela, the amount of individual change over a decade in relative social position should be quite modest. Furthermore, there is good reason to suppose that class status may be connected with patterns of identification with the Venezuelan traditional party system (Neuhouser 1992). For ideology, the case is more complex. Individuals certainly can change their ideological orientation over time, although they often do not; the question is whether there is evidence that a substantial percentage did so in Venezuela. The aggregate distributions of ideology in 1983 and 1993 are quite similar. In 1983, 30 percent of respondents classified themselves as rightists; in 1993, the figure was 31 percent. Similarly, regarding leftists, the data show a change from 17 percent in 1983 to 18 percent in 1993. Neither of these changes is statistically significant. Furthermore, both ideological categories have essentially stable relationships with other social identities and political attitudes. Hence, there is at least some reason to believe that ideological conversion involving leftists and rightists were fairly rare events in Venezuela during the decade under consideration. For centrists, there is some evidence of change, although the difference is not quite statistically significant; in 1983, 23 percent of respondents placed themselves in this category, while in 1993, 18 percent did so. This change is large enough to be somewhat worrisome. Mitigating this concern somewhat is the fact that the ideological centrist variable shows generally similar patterns of relationships with other variables in both years; for example, centrism is more common among the highly educated and those at higher social-class levels in both 1983 and 1993. Finally, and perhaps most important, repeating this analysis while dropping the variable for ideological centrism changes none of the substantive conclusions. Respondents’ reports of whether their mothers and fathers identified with traditional parties are included as a way of measuring whether respondents perceive themselves to have been deeply embedded in a partisan tradition. Some research suggests that respondent reports of parents’ partisanship are reasonably accurate, at least in teenagers (Niemi 1974: 53; Tedin 1976: 120; Westholm 1999: 531), so the measure may function in part as a meaningful measure of respondents’ objective partisan socialization. Furthermore, to the extent that responses deviate from parents’ true partisanship, the distortion is typically in

cor ru p t ion a nd col l a pse of pa rt y iden t ific at ion

103

the direction of the respondents’ own personal experience—in this case, most likely toward respondents’ own sense of partisanship when they were teenagers or young adults. Hence, whether or not they accurately measure respondents’ parents’ identities, these variables may provide some meaningful leverage on respondents’ partisanship during the peak of the traditional party system in a way that is more or less exogenous to ongoing political dynamics. Aggregate statistics on these variables generally support this interpretation. Thus, in 1983, 68 percent of respondents reported that at least one parent identified with the traditional parties; in 1993, the equivalent statistic was 71 percent, a difference that is not statistically significant. Like the other variables above, variables regarding parents’ party identification seem to be useful predictors of individuals’ 1983 partisanship in the 1993 data set. The key evidence regarding change in traditional party identification is summarized in Table 4.2, which reports the results of a pseudo-panel logit analysis of traditional party identification in 1993, using traditional party identification in 1983, corruption perceptions, economic evaluations, and opinion about the proper size of the state—a key issue involved in discourse about neo-liberal economic reform—as independent variables.11 Several important conclusions can be drawn from this analysis. First, there is evidence that party identification is meaningful and has the kind of inertia that would be expected of a real identity. This is reflected in the coefficient estimate for traditional party identification in 1983, which is large and positive. Clearly, ta b l e 4 . 2 . 1993 Model of Venezuelan

traditional party identification Variable Name Intercept Past party identification Corruption concerns Economic concerns Preferred size of state *(p < 0.05) **( p < 0.01).

Estimate (S.E.) – 1.048** (0.289) 0.790** (0.258) – 0.905* (0.428) – 0.079 (0.128) 0.074 (0.080)

104

cor ru p t ion a nd col l a pse of pa rt y iden t ific at ion

those who had a party identification in 1983 are still more likely to have one in 1993 than are others. Furthermore, in spite of the major economic depression experienced in Venezuela during the late 1980s and early 1990s, economic evaluations have essentially no relationship with traditional party identification in this model. For an individual who identified with an established party in 1983, who was happy with the current size of the state, and who saw no major problems with corruption, the estimated effect of moving from positive to negative economic evaluations is to drop the probability of identifying with the Venezuelan established parties by a mere 1.9 percent. This result suggests that revisions in party identification away from the traditional parties were not generally due to the economic results of failed neo-liberal reform.12 The comparatively small and insignificant coefficient for opinions about the optimal size of the state likewise suggests that citizens did not abandon their identification with the traditional parties because they were opposed to neoliberal reforms. For a Venezuelan who identified with a traditional party in 1983, who had no concerns about corruption, and who saw the economy as bad, the effect of moving from wanting the state to grow to wanting the state to shrink is a change of just 3.7 percent in probability of identifying with an established party in 1993. This result suggests that the sizable group of Venezuelans who preferred a larger state had patterns of party identification that were similar to those of citizens who favored a streamlined neo-liberal state. By contrast, the coefficient for corruption perceptions is relatively large and statistically significant. For a person who had identified with an established party in 1983, and with a status quo preference on the size of the state and negative economic evaluations, the effect of developing concerns about corruption is to reduce the probability of identifying with a traditional party in 1993 by fully 20.0 percent. This finding is not particularly model-dependent; effect sizes in this neighborhood arise for a wide range of specifications, including the use of different link functions, inclusion of additional control variables, and deletion of the included controls. A similar effect, of 14.0 percent, arises through a simple conditional difference in rates of party identification between those concerned about corruption and those less so, given an estimated 1983 party identification score strictly greater than zero. By way of comparison, the equivalent conditional difference for economic evaluations is only 3.4 percent, so the corruption effect is comparatively substantial. A corruption effect in the vicinity of 14 percent to

cor ru p t ion a nd col l a pse of pa rt y iden t ific at ion

105

20 percent cannot single-handedly account for the whole loss of identification with the Venezuelan established parties during the period under question, but it is clearly a major part of the story. Furthermore, it is plausible that a more fine-grained measure of corruption would have shown a stronger relationship. The overall implication of these results, then, is that the series of corruption scandals during the 1980s and early 1990s played a much larger role in eroding identification with the traditional parties than did either beliefs about the desirability of neo-liberal reform or concerns about the economic results produced by such reforms. 4.4.3 Identification with Peru’s Traditional Parties Peru’s party system during the 1980s has been described as far less institutionalized than the party systems in Argentina, Chile, or especially Venezuela (Mainwaring and Scully 1998). While this is true, there is some reason to think of the three-way political competition of the late 1970s and early 1980s as a meaningful, if volatile and weakly institutionalized, party system. The data on party identification in Peru reveal one of the sources of strength of the Peruvian party system—but they also emphasize the volatility and lack of social roots from which the party system suffered. The major strength of the party system of the 1980s, was the identification of a substantial part of urban society with the APRA party, particularly before 1987. However, a serious weakness of the party system was the fact that rates of identification with the other important parties of the system of the 1980s were usually at or below 10 percent.13 Furthermore, identification with APRA dropped sharply starting in late 1986 and concluding in January 1989, with an identification rate of about 7 percent. This weakening of the system of party identification in Peru is worthy of some analytic attention as an intervening variable in understanding that country’s party-system collapse. As with Venezuela, this large decline in party identifications raises the question of what motivated the flight from identification with APRA, the strongest Peruvian traditional party in terms of identification. Unfortunately, no Peruvian survey data from any point earlier than 1992 are available for analysis. The pseudo-panel technique used in discussing party identification in Venezuela thus cannot be used in Peru; only cross-sectional analysis is possible.14 In 1992, identification with APRA was still at 7 percent, the same level it had been at

106

cor ru p t ion a nd col l a pse of pa rt y iden t ific at ion

in April 1990. Hence, analysis at that point may provide some insights into the perceptions and attitudes that erased identification with APRA, by distinguishing between the small core of resilient supporters and the broader mass of society. Nonetheless, the results for Peru will clearly be more tentative than those for Venezuela. The 1992 survey does provide some indirect information about the reasons for APRA’s steep fall in party identification. A statistical analysis of the data, reported in Table 4.3, shows a quite strong relationship between corruption perceptions related to APRA’s administration between 1985 and 1990, on the one hand, and failure to identify with APRA, on the other. No similar relationship exists with respect to evaluations of economic policy during the APRA administration. This means that the hard core of APRA identifiers in the wake of the collapse had a similar view of the economy as other Peruvians during the García administration, but were generally less worried the level of corruption during that presidency. From these results, it may be plausible to infer that the corruption scandals plaguing the García administration, and not the devastating economic collapse during the same period, played the leading role in erasing APRA’s stock of party identifiers.

ta b l e 4 . 3 . 1992 Model of Peruvian party identification

Variable Name Intercept Economic evaluations Corruption perceptions Ideological leftist Ideological rightist Income group 1 (poorest) Income group 2 Income group 3 25–39 years old 40+ years old

Alianza Popular Revolucionaria Americana (APRA) Estimate

Acción Popular (AP) Estimate

Izquierda Unida (IU) Estimate

– 1.17 (0.66) 0.53 (0.40) – 3.00 (0.55)** 1.35 (0.54)* 0.76 (0.57) – 0.97 (0.65)

– 4.01 (1.10)** – 0.72 (0.50) – 0.13 (0.49) – 12.37 (548.93) 1.15 (0.54)* 0.62 (0.84)

– 3.12 (0.77)** 0.83 (0.43) – 0.31 (0.39) 2.19 (0.45)** – 0.35 (0.77) 0.01 (0.74)

– 0.59 (0.64) – 1.16 (0.74) – 0.03 (0.47) 0.13 (0.47)

– 0.79 (0.95) – 0.09 (0.91) 1.19 (0.81) 1.53 (0.80)

– 0.61 (0.76) 0.27 (0.75) – 0.13 (0.50) 0.27 (0.49)

notes: Maximum-likelihood estimates of the parameters in a multinomial logit model of traditional party identification. Null deviance is 688.96 on 1533 degrees of freedom; residual deviance is 578.98 on 1506 degrees of freedom. *( p < 0.05) **( p < 0.01)

cor ru p t ion a nd col l a pse of pa rt y iden t ific at ion

107

In summary, although conclusions drawn from the Peruvian analysis are necessarily weaker than those supported by the Venezuelan data, the results suggest a parallel conclusion: the decline in traditional party identification in both countries seems to be strongly connected to perceptions of corruption. 4.4.4 Identification with Argentina’s Traditional Parties If a decline in party identification is to be seen as a major ingredient of partysystem collapse, then no decline should have taken place in countries like Argentina that experienced no collapse. As discussed earlier, the Peronist party (Partido Justicialista, PJ) lost almost none of its identifiers during the period under consideration; by contrast, the traditional opposition party, the Radicals (Unión Cívica Radical, UCR) lost more than half of their identifiers. The stability of Peronist party identification may help explain why Argentina did not experience a party-system collapse during the 1990s and early 2000s. As in Chile, the presence of at least one traditional party with a substantial number of identifiers is associated with avoiding party-system collapse. In light of the repeated corruption scandals in Argentina during the Peronist government of the 1990s, it is somewhat anomalous for this chapter’s argument linking corruption and decline in party identification that Peronist party identification has been stable. Why would identification with the Peronist party be immune to corruption when identification with the Venezuelan traditional parties was significantly eroded by corruption scandals and Peru’s strongest party probably also lost many of its identifiers to concerns about corruption? The effects of corruption may depend on the kind of corruption at work (Redlawsk and McCann 2005), yet the Argentine corruption scandals during the 1990s included essentially all categories of corruption. Thus, there is no obvious voter-based explanation for the Peronist party’s ability to avoid the negative consequences of corruption. Perhaps the Peronist identity is distinctive in ways that make it less vulnerable than other parties to concerns about corruption (­Ostiguy 1998). Alternatively, the Peronist party’s well-known patronage machine (see, e.g., Auyero 2000; Stokes 2005) may distribute goods to identifiers that are party-branded in ways that make it easier for the recipients to give credit to the party for the payoff—and as a result buys the party insulation from the corrosive consequences of corruption scandals. Initial party-organizational data supporting this party branding hypothesis are discussed below.

108

cor ru p t ion a nd col l a pse of pa rt y iden t ific at ion

Turning to the decline of identification with the Radical party (UCR), we see that the same mix of hypotheses tested as potential explanations of declining traditional party identification in Venezuela are relevant. During the coalition government headed by Radical Fernando de la Rúa, which was in office from 1999 to 2001, Argentina experienced a major economic crisis, featuring negative growth rates, mass unemployment, currency devaluation, and the temporary expropriation of all savings accounts. Furthermore, the de la Rúa government saw the collapse of the previously successful convertibility policy—a policy that had been closely identified with neo-liberalism. The government was also hit by a series of corruption scandals that eventually resulted in a partial split of the governing coalition, although those scandals in part involved the government’s approach to handling corruption among Peronist politicians. It is therefore reasonable to ask whether economic perceptions, opinion about an element of the neo-liberal reform package, or corruption perceptions are useful predictors of change in party identification. Unfortunately, in the available survey data from the time, the set of time-invariant variables is insufficient to construct useful pseudo-panel indicators of Radical party identification, so no formal analysis is possible. Two major conclusions may be drawn from the information related to dynamics of party identification in Argentina. First, Argentina did not experience the kind of sharp decline in traditional party identification that was a precursor of party-system collapse in Venezuela. Identification with the Radical party dropped substantially, but Peronist party identification remains at roughly the same level as in the early 1990s, despite various economic crises. Second, the ability of the Peronist party to retain its identifiers in spite of major corruption scandals stands in need of an explanation, a task that is considered next.

4 . 5 C O R R U P T I O N A N D P A RT Y- B R A N D E D PAT R O N A G E

Why, then, were some parties’ patterns of identification apparently quite vulnerable to corruption scandals while at least the Peronist party appears to have undergone a long string of such scandals without losing its status as anchor for a collective identity? The explanation may be found in a factor related

cor ru p t ion a nd col l a pse of pa rt y iden t ific at ion

109

to party organization: the form of distribution of patronage. Patronage with a clear party brand may reduce the relationship between a voter’s observing a problem with corruption and deciding to change her party identification. Extensive distribution of patronage may make it seem as if a substantial portion of society—and not merely top political leaders—benefits from a party’s corruption. Thus, corruption within a party that distributes a great deal of patronage may be regarded both as an aspect of politics as usual and as a central component of a party-led social provisioning network. Seen in this light, corruption can be a deadweight cost to society but can also have redistributive aspects that solidify party identification among actual and potential recipients of patronage assistance. Of course, all of the parties in this study have a history of patronage politics, and the magnitude of patronage practices is difficult to measure, both for social scientists and for voters. Hence, it seems reasonable to differentiate parties not by how much patronage they deliver, but rather by how it is distributed. When patronage is distributed directly through party offices, voters receive a powerful and simple-to-understand cue that the party organization stands behind the benefits in question; the patronage unambiguously bears the party brand. This branding, in turn, may lead voters to feel less betrayed when leaders within the party prove to be corrupt. This idea regarding the effects of patronage has a heritage; for example, American reformer Jane Addams speculated that, when local politicians help the poor during times of special need, “A man who would ask at such a time where all this money comes from would be considered sinister. Many a man at such a time has formulated a lenient judgment of political corruption” (Addams 1898; see also Rosenblum 2008: 177). For parties that distribute patronage through government offices, by contrast, the branding cue is somewhat muddled. Is the benefit provided by a political party, by a specific politician or bureaucrat, or simply by the government in general as a right of citizenship? In these circumstances, voters may be less likely to view parties per se as a source of material benefits to society as a whole, and therefore may be less inclined to give parties the benefit of the doubt when a corruption scandal arises. That is to say, the brand of the patronage becomes ambiguous. The Peronist party’s relative immunity from losing party identifiers in the wake of major corruption scandals may thus be attributable to its relatively

110

cor ru p t ion a nd col l a pse of pa rt y iden t ific at ion

extensive practice of distributing clearly branded material and social patronage to voters directly through local (official and unofficial) party offices. None of the other parties, each of which seems to have lost identifiers in the wake of corruption scandals, carries out such practices as ubiquitously as the Peronists. Table 4.4 provides a first cut at thinking about the extent of party-branded patronage, showing the proportion of local organizations in each party that frequently host social activities at party offices. The Peronists rank much higher on this variable than any other party. Social activities such as birthday parties, football leagues, and clubs have long been a major component of the Peronist movement, serving as an alternative to open party activity during periods when the Peronist movement was legally proscribed (Levitsky 2003: 41–42, 49). Today, when the party is legal and in fact controls the national government, such activities supplement more explicitly political components of party participation. Common activities include football games, dances, barbecues, and parties for Eva Perón’s birthday. Sometimes these activities are “coordinated with Peronist intendentes [mayors] or governors,”15 suggesting that state resources and organization help support some of these partisan social activities. Other parties fall lower on this dimension. To the extent that partisans and community members perceive these Peronist social activities as a kind of party-branded benefit, that perception may encourage the belief that corruption benefits a wider circle than merely a handful of elite party leaders. Thus, by strengthening Peronist party identification, ta b l e 4 . 4 . Party-branded patronage Local Party Organizations That Hold Social Activities at Party Offices (percent)

Local Party Organizations That Provide Economic Services at Party Offices (percent)

Acción Democrática (AD)

29

16

Comité de Organización Política Electoral Independiente (COPEI)

24

9

Alianza Popular Revolucionaria Americana (APRA)

33

23

Acción Popular (AP)

13

2

Izquierda Unida

20

6

Partido Justicialista (PJ)

60

38

Unión Cívica Radical (UCR)

45

23

Party

cor ru p t ion a nd col l a pse of pa rt y iden t ific at ion

111

social activities may protect the party against corruption scandals. More generally, parties that play an important part in their supporters’ social lives may be more likely to retain supporters in spite of corruption scandals and other political challenges; the costs to such supporters of abandoning their party ties may be too high. The second and potentially more important of the local party activities considered here is the provision of economic goods and services (such as the sale of discounted food or the provision of health and medical care) at local party offices. Table 4.4 also shows the proportion of local organizations in each party that report frequently offering such goods and services to individuals who are not members of the party. Again, the Peronist party has a clear lead on this dimension. The relatively extensive participation of local Peronist organizations in distributing economic goods and services is consistent with many qualitative accounts of the party, such as Auyero’s discussion of Peronist “Problem-Solving Networks” (Auyero 2000: 80–118), and also with survey data about the extent of Peronist clientelism (Brusco, Nazareno, and Stokes 2004: 68–72; Stokes 2005). In many of the examples Auyero discusses, state funds and programs are the economic goods and services distributed via the local party apparatus. As with party-sponsored social activities, this spreading of the state’s wealth through local activities with a clear party brand may well reduce partisans’ outrage over corruption and as a consequence help explain Peronism’s lesser vulnerability to corruption scandals, in comparison with the Peruvian and Venezuelan parties.16 In light of this initial evidence, the proposition that party-branded patronage insulates party identifiers from the effects of concerns about corruption requires more serious testing. That is to say, voters may see parties that distribute branded patronage as “sharing the wealth” that derives from corruption, and thus as less blame-worthy, than parties that distribute patronage in other ways. This hypothesis fits the contours of the puzzle regarding corruption and Peronist party identification discussed earlier, and also corresponds with descriptive data about the patronage practices of South American political parties. While the evidence here is merely suggestive, if verified this account would explain why Peronist party identification appears relatively immune to the kinds of corruption scandals that appear to have badly damaged voters’ identification with the Venezuelan and Peruvian traditional parties.

112

cor ru p t ion a nd col l a pse of pa rt y iden t ific at ion

4.6 CONCLUSIONS

Identification with traditional parties dropped in both Peru and Venezuela early in the process of party-system collapse. Corruption deserves a central place in the explanation for this event. At the aggregate regional level, corruption scandals are (imperfectly but perceptibly) connected with party-system turmoil. At the micro level, there is evidence that voters with concerns about corruption were especially likely to cease identifying with the traditional parties in Peru and Venezuela. The same thing may have happened to the Radical party in Argentina. While identification with the Argentine Peronist party appears to be oddly immune to the effects of corruption, party-organizational evidence raises the possibility that the party’s branding of patronage was more efficient than that of other parties. Of course, declines in party identification do not necessarily imply surges in support for outsider candidates—much less the total collapse of the party system. After all, the United States and other countries have gone through periods of decline in party identification without concomitant major changes to the party system. Why did Peruvian and Venezuelan voters take the subsequent step of transferring their votes to outsider parties?

chapter 5

IDEOLOGICAL U N D E R R E P R E S E N TAT I O N AND VOTER DEFECTION

F

o r d e c a d e s, theoretical accounts of voting have emphasized ideological distance between parties and voters as a determinant of vote choice (Downs 1957) and secondarily of third-party entry into electoral competition (Palfrey 1984; Weber 1992). In empirical work, the position of voters on the left-right ideological dimension is a useful predictor of vote choice across a range of countries (Dalton 2002: 200–11). Furthermore, research on the determinants of support for third parties in the United States has found distance between the traditional parties and voters to be an important predictor of defection (Rosenstone, Behr, and Lazarus 1996: 127–32; Rapoport and Stone 2005: 26–33). Hence, it is perhaps unsurprising that ideology should play a central role in accounting for party-system collapse. Nonetheless, it is important to specify how and why ideology matters as a cause of party-system collapse, and also to consider the evidence in favor of a causal connection. Specifically, this chapter argues that the voters who are most likely to support challengers to the party system, and thus in a sense the core constituents for party-system collapse, are those who feel ideologically underrepresented and concerned about problems of corruption. The next chapter argues that these causes are particularly relevant because they produce anger, thereby reducing

11 4 ideological underr epr esentation & voter defection

voters’ aversion to risk and enabling a vote in favor of an uncertain candidate from outside the traditional party system. There is provocative aggregate-level evidence that the broad ideological positioning of parties relative to the electorate as a whole varies across the South American countries that serve as the focus of comparison here, and also that these differences align with the outcome of party-system collapse. Specifically, survey data for the periods of collapse in Venezuela (1998) and Peru (1990) show substantial portions of the electorate as falling outside the interval of ideological competition among the traditional parties. Indeed, in both contexts, the median voter appears to fall in a lacuna of traditional-party ideological spacing. In Argentina (1996), by contrast, nearly all citizens are ideologically proximate to a traditional party. These differences point to a central component of the explanation for party-system collapse: systems become vulnerable when a substantial component of the electorate feels excluded from the give-and-take of ideological competition. These points are illustrated in Figures 5.1, 5.2, and 5.3, which show the distribution of citizens’ self-reported ideology on a 1–10 scale (or a 1–7 scale, for the Argentine data), as well as the mean citizen placement of political parties on that same scale. In Figure 5.1, we find that Venezuelan citizens perceived both traditional parties as occupying virtually the same place somewhat to the right of center on the ideological spectrum. This suggests that citizens who place themselves at the center or anywhere further left may have a motive to defect to outsider challengers; this includes a substantial proportion of the Venezuelan electorate. At the same time, the ideological indistinguishability of the traditional parties also leaves the right wing of the electorate potentially vulnerable to outsider appeals. In general, a large majority of Venezuelan voters were not represented particularly well by the traditional party system of 1998. For Peruvians, the situation was substantially different (see Figure 5.2). The major remaining traditional party as of 1990, APRA, was seen by Peruvians as falling somewhat to the left of center. The two parties from the earlier Izquierda Unida coalition that participated in the 1990 elections fell further to the left than APRA. Thus, voters to the left of center were quite faithfully represented and faced a substantial range of choices in the political contest. Citizens who saw themselves as belonging to the center or the right had many fewer choices; the only party seen by Peruvians as occupying a location in the

f i g u r e 5 . 1 . Venezuelans’ ideologies, 1998. source: Data drawn from the 1998 Redpol survey, archived at the Banco de Datos Latinoamericano de la Universidad Simón Bolívar, Caracas, Venezuela. A score of 1 represents the extreme left, while 10 represents the extreme right. Bars represent the distribution of respondents’ ideological self-placement. The mean of respondents’ ideological placement of AD and COPEI are shown on the graph as two vertical lines; the lines are so horizontally close that they are visually indistinguishable.

f i g u r e 5 . 2 . Peruvians’ ideologies, 1990. source: Data drawn from the March 1990 Apoyo survey, published in the April 1990 issue of Informe de Opinion (Lima: Ipsos Apoyo). A score of 1 represents the extreme left, while 10 represents the extreme right. Bars represent the distribution of respondents’ ideological self-placement. The mean of respondents’ ideological placement of Izquierda Unida, Izquierda Socialista, APRA, and FREDEMO are shown on the graph as four vertical lines.

116 ideological underr epr esentation & voter defection

center or on the right was the Frente Democratico, or FREDEMO, a coalition that subordinated the traditional parties of the center-right to Mario Vargas Llosa’s outsider political movement. Peruvians perceived FREDEMO as having been as far to the right as Izquierda Unida was to the left; the large number of citizens on the center and center right that fell between APRA and FREDEMO might plausibly have concluded that the traditional parties had abandoned them ideologically. While not quite a majority, this group of voters with a plausible interest in outsider candidates constituted a plurality of the Peruvian electorate. Turning at last to the results for the Argentine electorate and party system, shown in Figure 5.3, we see a much smaller ideological discrepancy between voters and parties than in the other two countries. Over half of all voters placed themselves ideologically within one point of one of the traditional Argentine parties. The roughly 30 percent of voters toward the left end of the spectrum found themselves somewhat further from the established parties at that time,

f i g u r e 5 . 3 . Argentines’ ideologies, 1996. source: Data drawn from the July 1996 survey by the Center for Public Opinion at the University of Buenos Aires, archived at The Roper Center, University of Connecticut, Storrs, Conn. A score of 1 represents the extreme left, while 7 represents the extreme right. Bars represent the distribution of respondents’ ideological self-placement. The mean of respondents’ ideological placement of FREPASO, the UCR or Radicals, and the Peronists are shown on the graph as three vertical lines.

ideological underr epr esentation & voter defection 117

which corresponded with the peak of Carlos Menem’s neo-liberal Peronist administration; these voters had the temporary option of turning to a non­ traditional party, FREDEMO, for representation. In later electoral cycles, first the UCR and then the Peronists shifted slightly to the left, filling this ideological void. Indeed, over time, the Peronist party occupied virtually every imaginable position between the center left and center right on the Argentine ideological spectrum, by itself essentially preventing the emergence of a durable space for significant outsider contestation.1 Thus the aggregate picture of ideological representation in these three countries supports the argument that party systems collapse in part because of problems of ideological underrepresentation. Later in this chapter, evidence is presented to support the claim that individuals falling inside the representational lacunae identified here play a central electoral role in the process of party-system collapse, as do concerns about corruption as channeled through party identification.

5 . 1 M E A N I N G O F T H E L E F T- R I G H T IDEOLOGICAL SPECTRUM

Before turning to the evidence in support of the proposition that ideological representation, as well as concerns about corruption via party identification, play an important causal role in party-system collapse, two preliminary issues regarding the use and meaning of the left-right spectrum deserve consideration. First, there is a long-running empirical debate about whether voters use ideology to choose the party closest to them (i.e., proximity voting) or whether they instead choose the most extreme viable party on their side of the median (i.e., directional voting). Proximity voting has been the basis for most theoretical models of vote choice, although directional voting has its advocates (Rabinowitz and MacDonald 1989). Lewis and King have extensively reviewed this debate and shown that it revolves around statistical assumptions that cannot be tested using most existing data from the United States (Lewis and King 1999). This conclusion holds as well for the South American survey data considered below. Indeed, some of these data do not even allow an adequate specification of either theory, because some of the surveys lack the necessary questions about citizens’ perceptions of party position on the left-right continuum. Hence we are forced

118 ideological underr epr esentation & voter defection

to disregard debates about the nature of ideological voting and merely infer effects from the left-right self-placement of voters who choose to support specific traditional parties, or who defect to specific outsider parties. Second, it is evident that ideology is more than a unidimensional scale running from “left” to “right.” Latin America has, in recent decades, seen important political ideologies form around populism, neo-liberalism, Marxism-Leninism, Maoism, indigenous rights and identities, peasant rights, and nationalism, to name a few. This complexity of ideologies is an important part of what gives political competition in each country its own unique flavor. Yet these terms are also sometimes ambiguous and contested, varying in meaning from individual to individual, region to region, and country to country. For the purposes of systematic comparison, then, it is easier to rely on the simplified representation of ideological space provided by the left-right dimension. Nonetheless, this simplification is useful only if it in some sense represents the underlying reality. Scholars have expressed concern that Latin American parties and voters cannot meaningfully relate to the left-right spectrum because parties are opportunistic and adopt policy orientations that are extremely flexible over even the medium term (Mainwaring and Scully 1995a: 25). Concern about the applicability of the left-right spectrum is especially pronounced among analysts of Mexico, where, over the course of the 1990s “ideology became an increasingly muddled concept, representing an uncertain amalgam of classic ‘left-right’ issues, cultural values, and attitudes toward the old political establishment” (McCann and Lawson 2003: 66; see also Moreno 1998; Zechmeister 2002). At the same time, some scholars have found the left-right spectrum to be analytically useful in describing and explaining voter behavior and party system outcomes in Latin America. The left-right positioning of parties is a central component of Coppedge’s discussion of the dynamic diversity of party systems in the region (Coppedge 1998). Rosas finds that most Latin American legislative party systems can be described using a single underlying ideological dimension, although its composition varies from country to country (Rosas 2005); Colomer presents evidence that most Latin American citizens are likewise able to place themselves on the unidimensional left-right ideological spectrum (Colomer 2005). At the elite level, Kitschelt et al. (2010) find significant cross-country variation in the extent to which left-right self-placement predicts

ideological underr epr esentation & voter defection 119

Latin American legislators’ ideological and policy positions; consistent with the mass-level analysis below, their 1997 data show some meaningful economic policy structure to the left-right dimension in Venezuela and Argentina, but much less in Peru. In light of these opposing perspectives on the usefulness of the left-right ideological spectrum in Latin America, an empirical approach is needed to determine whether that spectrum currently provides a meaningful description of political space in any Latin American country. This question is especially important in light of the fact that most of the available survey data for the past twenty-five years in South America include a left-right ideological selfplacement question but do not include measures of specific issue positions and political attitudes. If the left-right spectrum can be seen as grounded in such positions and attitudes, then there is some justification for analyzing the available variables, which would then have an interpretable relationship with the actual perceptions, beliefs, and attitudes driving voters’ decision-making during party-system collapse; if not, then the existing data may be simply insufficient for the goal of understanding voter motivations. This section seeks to determine whether the left-right spectrum is analytically usable by exploring the empirical relationships between specific issue positions and value statements, on the one hand, and voters’ ideological self-placement, on the other. The analysis adopts a descriptive, as opposed to prescriptive, approach to the left-right spectrum. That is to say, the goal of the analysis is to discover how citizens think about the left-right distinction, not to determine whether they conform to some ideal-typical conception of that distinction. In addition to asking whether left-right ideology has stable connections with other political attitudes, the analysis investigates whether left-right ideology is a proxy for any of the other independent variables in this chapter. Are respondents’ left-right self-placements significantly influenced by economic evaluations, corruption perceptions, or anti-party attitudes? If so, including ideology as an independent variable may distort the coefficient estimates for those other factors. But in fact, as this section shows, left-right ideology in Argentina, Peru, and Venezuela is not strongly related to these variables. In order to determine whether left-right ideological self-placement has some meaningful connection with beliefs about issues and with general priorities for the economy, politics, and society, I run a separate regression analysis for each

120 ideological underr epr esentation & voter defection

country, using ideology as the dependent variable and various issue stances and value statements as independent variables. The data for this analysis come from the 1995–96 wave of the World Values Survey, which includes an ideological self-placement question and a wide range of forty more specific issue and value topics.2 If the left-right ideological spectrum is weakly explained by this large package of variables in a given country, that provides evidence that the leftright dimension is a category without content in that country. By contrast, if left-right ideology is descriptively related to other attitudes in such a way that the overall goodness-of-fit statistics are nontrivial, then there is clear evidence that the left-right dimension has some structure. Causal inferences about why voters classify themselves in a specific left-right slot are not the goal; rather, it is important to discover whether or not the left-right dimension is an orderly part of a larger package of attitudes. This section reports only overall fit statistics for the analysis in each country that uses the full set of independent variables. Instead, results from a restricted model including only variables with a P value of less than 0.1 in the full model are discussed for each country.3 For Venezuela, the regression with all forty independent variables has 324 valid cases and an R 2 of 0.311. This moderately high R 2 suggests that the leftright spectrum is indeed grounded in specific issues and value orientations, although it may also reflect broader social identities or measurement error. The restricted model for Venezuela includes sixteen independent variables. This model suggests a left-right spectrum grounded primarily in religion, expectations regarding the structure of employment relations, and desired extent of social transformation: leftists are easiest to find among the less religious, those with more egalitarian expectations and desires about how work should be structured, and those more desirous of radical social change. Overall, in Venezuela, the left-right spectrum approximates the classical spectrum dating back to the French Revolution. For the Peruvian data, the full model with forty independent variables has 260 valid cases and results in an R 2 of 0.361. Much of this moderately high R 2 results from the inclusion of variables with some explanatory power that do not reach statistical significance; hence, the restricted model, which includes twelve independent variables, has 556 cases and an R 2 of only 0.084. The contrast between this result and the finding in the full model suggests that the

ideological underr epr esentation & voter defection

121

Peruvian left-right spectrum is quite complex and comprises a wide range of individually minor issues and attitudes. The restricted model for Peru shows that the Peruvian left-right spectrum in the mid-1990s included numerous regime issues. Leftists were those who looked back to previous regimes, such as the democratic period of the 1980s or the military government before that, as a better time. Rightists, by contrast, tended to feel that the country was improving politically under Fujimori’s regime in the 1990s. It is likely that this particular structuring of ideological space is at least in part a product of the dramatic and turbulent political, social, and economic transformation associated with Peru’s neo-liberal transition. No similar survey data are available for the pre-Fujimori period, but less systematic evidence suggests that the left-right spectrum then featured primarily political and economic issues involving the competition among Marxist, socially and economically reformist non-Marxist, and laissez-faire economic alternatives. These three programmatic alternatives are standard characterizations for the three political blocs involved in the party system of the 1980s (Cameron 1994: 17–52; Tanaka 1998: 103–64). Evidence that the distinction among these alternatives structured the left-right ideological spectrum among voters, as well as among political elites, arises from survey data in which respondents order the representatives of each political bloc on a left-right spectrum in a way that agrees with expert rankings. For example, in Peru in 1985, the mean left-right rankings of the major parties placed Acción Popular farthest toward the right, APRA toward the center, and Izquierda Unida farther toward the left.4 Hence there is at least some reason to believe that, during the period leading up to party-system collapse, Peru’s electorate saw the left-right spectrum as significantly involving alternative political and economic plans for society—even though this spectrum became somewhat more complex and regime-centered by the middle of Fujimori’s presidency, especially after the defeat of the guerrilla group Sendero Luminoso. In Argentina, the full model has an R 2 of 0.457 on a total of 243 valid cases; this is the highest value for any of the three countries considered here, suggesting that the Argentine left-right spectrum is more firmly grounded in other political attitudes and identities than are the already reasonably well grounded spectrums in Peru and Venezuela. The restricted model, with a total of 692 valid cases, maintains an R 2 of 0.134.

122 ideological underr epr esentation & voter defection

These results suggest an Argentine mass perception of the left-right spectrum as involving fairly prototypical issues of poverty, reformism, and social transformation, as well as concerns about the environment. Corruption perceptions and other purely political issues, however, are apparently not meaningfully incorporated into the spectrum. The most important patterns involve the overall R 2 statistics for the full models in each country. These results reveal a substantial grounding of the leftright continuum in political attitudes and beliefs in each of the three countries under consideration, with that grounding being somewhat weaker but still meaningful in Venezuela and substantially strong in Argentina. Turning to the secondary task of determining the contents of the left-right dimension in each country, the data reveal somewhat different ideological spectrums in each country. In Venezuela, the left-right spectrum seems to be primarily connected with attitudes about religion, work, and social structure. In Peru, the spectrum changed over time, reflecting a classical contest of political and economic ideologies in the 1980s but a more complex collection of issues and attitudes by the mid-1990s. Argentina’s left-right spectrum seems to be connected primarily with attitudes about equality and the need for social change; approaches to the environment are also relevant. For each country, there is thus some reason to believe that the left-right ideological spectrum is a useful analytic tool. Concerns that unidimensional measures of ideology reflect different mixes of issues and concerns in each country are justified, so the distinctive meanings of “left” and “right” in each country should be kept in mind throughout the following analysis.5 Furthermore, as analysts of Mexico have suggested, regime issues are an important component of the spectrum in Peru (although these issues may have been less important before the Fujimori presidency). Nevertheless, since the spectrum seems to have a reasonably well defined meaning in each country, the analysis that follows uses left-right self-placement variables without emphasizing these country-specific variations in meaning. Another important result of the above analysis is that left-right ideology is not primarily a proxy for the other major variables used in the analysis below. Economic evaluations do not have a relationship with ideology in any of the countries; neither do corruption perceptions. While attitudes toward the democratic political regime are significant predictors of left-right ideology in Peru,

ideological underr epr esentation & voter defection 123

anti-party attitudes are not significant in any of the four countries. Thus, in using the left-right scale of ideology as an independent variable, we are most probably not distorting inferences related to the other independent variables, and, as this section has shown, we are drawing on a variable with some significant political content in each context.

5.2 EXPL AINING VOTE CHOICE IN VENEZUEL A

Having seen that the left-right ideological spectrum has a meaningful grounding in more basic issues and attitudes in the South American countries of interest in this study, we turn to the question of how ideology causally affects individual-level decisions to vote against the traditional party system. This book’s central argument is that a sense of ideological underrepresentation, as well as concerns about corruption via changes in patterns of party identification, are central individual-level motives for bringing about party-system collapse. These two attitudes produce anger, which reduces voters’ aversion to uncertainty and thereby facilitates the voter’s decision to vote for a candidate from outside the party system. The following chapter explores the affect-anduncertainty mechanism experimentally. The rest of this chapter provides evidence from survey data showing that these attitudes are in fact closely associated with voting against the traditional party system in the elections that produced collapse in Venezuela and Peru. Thus this chapter provides observational evidence of the relationship for which the next chapter develops experimental evidence of a causal mechanism. This section, which focuses on the Venezuelan elections of the 1990s, and the following section exploring vote choice in Peru, test this book’s explanation against the most important competing hypotheses about why citizens choose to vote for a nontraditional party: economic voting and social class considerations, as well as the propositions that corruption perceptions are a powerful direct explanation of party-system collapse, rather than an indirect explanation via changes in party identification. These tests use Venezuelan survey data from 1993 and 1998, as well as corroborating evidence from a 1992 survey in Peru about vote choice during the party-system collapse, to argue that the electoral decisions involved in party-system collapse can best be explained by an additive combination of decline in traditional party identification (due in part to

12 4 ideological underr epr esentation & voter defection

past corruption scandals), an ideological disconnect between a large portion of the electorate and the traditional parties, and corruption perceptions at the moment of the election. 5.2.1 The 1993 Presidential Elections The 1993 elections in Venezuela took place in the aftermath of the crisis-ridden second presidency of Carlos Andrés Pérez. Pérez had instituted neo-liberal reform by surprise shortly after his election at the end of 1988. The aftermath of these reform efforts included the riots, coup attempts, and corruption scandals described above. Pérez was impeached and removed from office in May 1993. The survey data analyzed below were collected at the moment when conflict over Pérez was at a peak. During the 1993 campaign, both the AD candidate (Claudio Fermín) and the COPEI candidate (Osvaldo Alvarez Paz) established policy positions in favor of the neo-liberal, marketizing reforms started by Pérez. By contrast, the two major outsider candidates, Andrés Velázquez of La Causa R (Radical) and Rafael Caldera of Convergencia and MAS (Movimiento al Socialismo, a long-standing but electorally marginal left party that was a junior partner in Caldera’s campaign and the first years of his government), were strongly critical of such reforms and called for much more redistributive economic policies. At the level of campaign messages, a strong economic left-right dynamic seemed to exist between outside challengers and traditional parties. A similar division existed with respect to corruption, with the outsider candidates emphasizing the need to wipe out official corruption and the traditional party candidates largely avoiding the issue.6 In May and June of 1993, the Centro de Investigaciones y Estudios Políticos y Administrativos of the University of Zulia in Venezuela commissioned a 1,500-respondent national pre-electoral survey, used as the endpoint of the pseudo-panel analysis of party identification above, in anticipation of the presidential elections to be held on December 3. In spite of the substantial time lag between the interviews and the final elections, this survey provides the best data source for considering Venezuelans’ electoral choices in 1993. Table 5.1 shows the vote share of each candidate in the survey data and in the final electoral returns. Clearly there are discrepancies between the two sets of vote shares. Most of the respondents who did not specify a vote intention likely ended up abstaining, since voter turnout in this election was only

ideological underr epr esentation & voter defection 125

ta b l e 5 . 1 . Survey and official vote shares in Venezuela’s 1993 elections Candidate Claudio Fermín

Survey Vote Share (percent)

Actual Vote Share (percent)

12

23.60 22.73

Osvaldo Alvarez Paz

21

Andrés Velázquez

17

21.95

Rafael Caldera

8

30.45

Other or None

41

1.27

source: Data from the 1993 CIEPA electoral survey of Venezuela, archived at the Banco de Datos Latinoamericano de la Universidad Simón Bolívar, Caracas, Venezuela.

58.3 percent (International IDEA 2001). Others probably shifted into one of the four major political camps, since all other parties in the end received less than 2 percent of the vote. Overall, there was a major shift in the direction of Caldera during the months after the survey. A smaller but still substantial shift toward Fermín happened during this period. Table 5.2 reports the results of a multinomial logit analysis of 1993 presidential vote choice, using the independent variables introduced above.7 The residual deviance of a model using this dependent variable with only an intercept is 3039 on 4156 degrees of freedom; the preferred model reduces the

ta b l e 5 . 2 . 1993 Model of Venezuelan presidential vote choice Variable Name Intercept Economic evaluations Corruption perceptions Ideological leftist Ideological rightist AD Party ID COPEI party ID Education Income

AD

COPEI

LCR

Convergencia 2

– 4.37 (0.58)**

– 2.14 (0.38)**

– 1.30 (0.30)**

– 3.05 (0.42)**

0.20 (0.32)

– 0.61 (0.23)*

– 0.42 (0.19)*

– 0.12 (0.25)

– 1.10 (0.51)*

– 0.14 (0.31)

0.32 (0.22)

0.23 (0.30)

0.33 (0.50) 1.19 (0.34)**

– 0.04 (0.35) 0.89 (0.25)**

0.77 (0.23)** 0.65 (0.23)*

1.13 (0.28)** 0.39 (0.34)

4.47 (0.37)** – 0.19 (1.06) 0.18 (0.08)* – 0.03 (0.07)

0.33 (0.45) 3.81 (0.27)** 0.14 (0.06)* – 0.14 (0.06)*

– 1.96 (0.74)* 1.36 (0.28)** 0.08 (0.05) – 0.04 (0.05)

– 1.77 (1.03) – 0.93 (0.75) 0.30 (0.06)** – 0.07 (0.07)

notes: Maximum-likelihood estimates of the parameters in a multinomial logit model of vote choice. Null deviance is 3039 on 4156 degrees of freedom; residual deviance is 2082 on 4120 degrees of freedom. *(p < 0.05) **( p < 0.01).

126 ideological underr epr esentation & voter defection

deviance to 2082 on 4120 degrees of freedom. A likelihood ratio test based on this information suggests that the independent variables, taken as a whole, are valuable predictors of 1993 vote choice. The model is complex; each coefficient estimate relates to the logged odds of voting for the named party as opposed to abstaining. Hence, comparisons across rows are necessary in order to evaluate relative probabilities of supporting different parties, but such comparisons are not necessarily straightforward and statistical significance is complex. To get around these issues, the discussion below introduces a set of simulations comparing estimated probabilities for a hypothetical individual with the median scores on all variables other than those of current interest (the median scores are 1 on economic evaluations, 0 on corruption, 0 on both ideological variables, 0 on both party ID variables, 4 on education, and 2 on income) and a score of 0 on the variable(s) of interest with the estimated probabilities for an otherwise identical individual with a score of 1 on those variable(s). Because turnout is not an issue of central interest, these probabilities are calculated and compared conditional on the individual in question actually choosing to vote for one of the named candidates/parties. Attention is focused primarily on relative risks—i.e., the probability of voting for the selected party given a higher score on the variable of interest divided by the probability of voting for that party given a lower score. The relative risk is more useful than raw probabilities because the simulation condition does not assign equal baseline probability to support for each party. Standard errors for these estimated relative risk ratios are calculated through bootstrapping. Perhaps unsurprisingly, voters who identify with AD are quite likely to vote for AD and substantially disinclined to vote for the outsider parties; for the simulation condition, identification with AD has a relative risk of 14.89 (with a standard error of 7.30; however, the variance in the ratio is not symmetrically spread; a bootstrapped percentile-method 95 percent confidence interval for the ratio runs from 8.35 to 38.63, suggesting that the ratio is highly significantly different from 1). AD identifiers are not likely to support COPEI, but they have nearly zero probability of supporting any nontraditional party.8 Voters with a COPEI party ID are also strongly inclined to vote in favor of COPEI (the model estimates a relative risk of 4.37 with a standard error of 0.89). COPEI identification in the simulation condition has a very small relative risk for voting for AD (ratio of 0.08 with a standard error of 0.09) or Convergencia (ratio

ideological underr epr esentation & voter defection 127

of 0.04 with a standard error of 0.03), but a somewhat higher but still generally low ratio regarding support of La Causa R (0.38 with a standard error of 0.07). In other words, identifying with either of the traditional parties sharply reduces the probability that a voter will end up supporting an outsider party. Hence, the analysis in the previous chapter presenting evidence that corruption is a central force eroding identification with the traditional parties is directly relevant to the final outcome of voting for a nontraditional party. The reason is not that the absence of a party identification necessarily determines a vote for a nontraditional party, but rather that the presence of an identification almost entirely prevents such an outsider vote. While these results are important in tying the theory and analysis from the Chapter 4 together with the work in this chapter, an otherwise similar analysis that excludes party ID variables leads to the same substantive conclusions regarding all other variables. Thus there is good reason to suppose that including party ID in the model does not significantly distort direct coefficient estimates related to corruption, economic voting, or ideology. While the causal interconnections among these variables are complex, there are evidently few statistical consequences for the present analysis related to including or excluding party identification. Indeed, the same is essentially true for each major variable or block of variables; the results presented above are not notably sensitive to the exact specification of the multivariate model. Hence, this discussion of the remaining sets of variables pays no further explicit attention to the results of omitting them from the analysis. The coefficient estimates suggest that a positive evaluation of the previous government’s economic policy performance may be connected with supporting AD rather than the other three parties, although the contrast is not statistically significant. In the simulation, the voter with otherwise median scores has a relative risk of voting for AD as against the other three parties of 1.75 (standard error of 0.58), comparing between positive and negative economic evaluations. The justification for punishing AD for bad economic performance is a straightforward application of economic voting theory; what is noteworthy here is that the other traditional party, COPEI, gains marginally statistically significant support from voters with poor perceptions of the economy (an estimated relative risk for positive economic perceptions of 0.77 with a standard error of 0.15). Among nontraditional parties, there is relatively little evidence

12 8 ideological underr epr esentation & voter defection

of a connection between vote choice and economic perceptions; Convergencia has a relative risk of 1.27 (standard error of 0.26), and La Causa R has a ratio of 0.94 (standard error of 0.10). Economic voting may thus play some role in determining which party voters supported in the 1993 elections—a finding that is unsurprising and fits the broad pattern of results showing that voters at least sometimes engage in economic voting. Yet two issues complicate this picture. First, the risk ratios in the previous paragraph do not reach standard levels of statistical significance. Second, and arguably most important, the apparent pattern of economic voting is such that voters concerned about the economy punish one traditional party (AD), while rewarding another traditional party (COPEI). While economic voting may be part of the overall picture of the 1993 elections, it is not an explanation that cleanly fits with the outcome of party-system collapse. Indeed, the data suggest that economic voting matters more as a determinant of which traditional party a voter supports than as a motive in the choice between the traditional and nontraditional parties as blocs. The results with respect to corruption perceptions are arguably somewhat simpler. Here, AD is substantially punished for perceived corruption; the model estimates a relative risk ratio of 0.28 with a standard error of 0.15; in the simulation condition, a voter who is worried about corruption is about one-fourth as likely to support AD as one who is not. COPEI may also lose some votes, but the estimated relative risk ratio is not statistically significantly different from zero (a ratio of 0.73 with a standard error of 0.19); the relative risk is also not statistically distinguishable from zero for Convergencia, which has an estimate of 1.06 (standard error of 0.24). The most nontraditional party may benefit to some extent; La Causa R has a non-significant but perhaps suggestive relative risk of 1.16 (standard error of 0.14). Overall, these results point toward a probable loss of votes for AD, and perhaps also for COPEI, among voters worried about corruption. These votes weakly appear to be transferred to the most anti-party-system option available in the election: La Causa R. Thus, in 1993 we see some evidence supportive of a direct impact of corruption concerns on traditional parties’ vote shares, as well as an indirect effect via party identification, discussed below. For the ideology variables, virtually all coefficient estimates are positive, suggesting that voters with a noncentrist ideology are more likely than either

ideological underr epr esentation & voter defection 129

nonideological or centrist voters to report some vote choice. To make as clear as possible the relationship between ideology and vote choice among those who supported one of the named parties, this paragraph deals with the simulated difference in probability of vote between a voter in the simulation condition who is a leftist and a similar voter who is a rightist. Both traditional parties are weak toward the left. AD loses ground substantially among leftists: in the simulation condition, there is a relative risk ratio related to the leftist versus rightist comparison of 0.39 (standard error of 0.21; this reaches statistical significance at the 0.05 level), suggesting that leftists are less than half as likely to support the party than rightists. This finding is important for many reasons; one interesting implication is that AD was unattractive to voters on the left even before it formed a coalition oriented toward implementing neo-liberal economic policies during the late Caldera administration. COPEI likewise has a substantial difference between rightists and leftists: the relative risk ratio for ideology in the simulation condition is 0.36 (standard error of 0.15; this is also significant at the 0.05 level). By contrast, both non­ traditional parties are at least undamaged by leftist ideology, and one is possibly stronger on the left than on the right. For La Causa R, a respondent’s ideology is connected with a somewhat weak ratio (an estimated relative risk of 1.04 with a standard error of 0.17). Convergencia gains among leftists, with a relative risk of 1.94 (with a standard error of 0.61; once again, because the variance in the ratio estimates is asymmetric, the ratio is significantly different from 1 at the 0.05 level), thus appearing to be meaningfully strengthened among leftist voters. As with corruption perceptions, as early as 1993, there is a direct—and comparatively strong—link between leftist ideology and voters’ decision to support nontraditional parties. These two collections of effects support the argument of the book regarding the causal importance of corruption perceptions and ideological underrepresentation, as do the results from the 1998 elections. The estimates for the income variable suggest that people from wealthier families are somewhat unlikely to support COPEI; the results are statistically significant but substantively smaller than those reported above, so no relative risks are shown. More educated individuals are more likely to vote and may lean somewhat more toward Convergencia then less-educated adults. Again, however, the results are substantively smaller than the important findings above. Thus these two social class variables do not seem to play an overwhelming role

130 ideological underr epr esentation & voter defection

in the electoral decision in 1993. It is particularly important to note that the lower socioeconomic status typical of those employed in the informal sector is not at all predictive of a decision to vote against the traditional party system. Overall, the three most important predictors of support for nontraditional parties are party identification, concerns about corruption, and leftist ideology. As the previous chapter showed, the dynamics of party identification substantially involve concerns about corruption; thus these findings are consistent with both a long-term and indirect effect of concerns about corruption on support for outsider parties via party identification and a smaller, short-term, and direct effect of corruption perceptions on the immediate selection of a presidential candidate. Likewise, leftist ideology is associated with voting for a nontraditional candidate, a pattern that fits well with this book’s hypothesis that ideology and corruption produce anger, which in turn serves as the immediate psychological mechanism underpinning the decision to vote against the established parties.9 5.2.2 The 1998 Presidential Elections The 1998 elections in Venezuela took place in a context of economic difficulties and charges of political corruption. The Caldera administration’s early attempts to stimulate the Venezuelan economy using heterodox economic policies ended when a banking crisis pushed the government to seek international assistance and to turn in a markedly neo-liberal policy direction. Economic performance remained inconsistent throughout Caldera’s term, and the 1998 elections were held in a period of recession. Furthermore, although Caldera was regarded by many Venezuelans as honest, many citizens came to regard other officials within his administration as corrupt. Throughout this period, Hugo Chávez Frias was an active participant in the country’s political debates, calling for revolutionary change and a boycott of elections held partway through Caldera’s administration and then organizing a political movement to participate in the 1998 campaign. AD’s candidate, Luís Alfaro Ucero, was the only traditional party candidate in the 1998 presidential election. COPEI first supported one outsider candidate, Irene Saez. When her poll numbers declined, COPEI transferred its support to another outsider, Enrique Salas Romer. In the final week before the election, AD also asked its supporters to back Salas Romer, in an attempt to defeat the

ideological underr epr esentation & voter defection

131

leftist outsider candidate, Hugo Chávez Frias (who was supported, like Caldera in 1993, by a collection of relatively marginal leftist parties, including the MAS). Thus, the final results, in which only outsider candidates receive any significant fraction of the vote, may somewhat understate the support of the traditional parties. By contrast, the RedPol survey was concluded before AD withdrew support for its candidate, providing a potentially less distorted snapshot of support for the party. A consortium of universities in Venezuela, known as RedPol, sponsored a pair of surveys before and after the 1998 presidential elections. The pre-electoral survey took place between November 13 and 27, concluding roughly a week and a half before the election on December 6. Evidence from the survey suggests that AD’s decision to withdraw support for Alfaro Ucero probably had its intended result of pushing almost all of AD’s remaining supporters to vote for Salas Romer. In the pre-electoral survey, AD’s share of the projected presidential vote was 7.5 percent. In the final election, by contrast, Alfaro Ucero received a mere 0.42 percent of the vote. In other words, in the final election results Alfaro Ucero lost roughly 95 percent of the supporters that he had at the time of the poll. On the other hand, only slightly more than 7 percent of the electorate as a whole was affected by this decision. As shown in Table 5.3, 72 percent of the respondents in the pre-electoral survey favored Chávez, Salas Romer, or Alfaro Ucero. No other candidate had significant support, although roughly a quarter of respondents either mentioned a minor candidate or reported no intention to vote. The following analysis uses ta b l e 5 . 3 . Survey vote shares in Venezuela’s

1998 elections Candidate Hugo Chávez Frias Enrique Salas Romer Luis Alfaro Ucero Other or none

Estimated Vote Share (percent) 35.9 28.2 7.5 28.4

source: Data from the 1998 RedPol pre-electoral survey of Venezuelans, archived at the Banco de Datos Latinoamericano de la Universidad Simón Bolívar, Caracas, Venezuela.

132 ideological underr epr esentation & voter defection

a combined category for respondents who did not plan to vote and respondents who mentioned a minor candidate. While there is almost certainly heterogeneity in this category, the approach of unifying it seems reasonable here both because the number of respondents in each category is small and because the real focus of analytic attention involves the distinction between the two principal outsider candidates (Chávez and Salas Romer) and the one traditional party candidate. Table 5.4 reports the results of a multinomial logit analysis of 1998 presidential vote choice, using the independent variables introduced above, with the caveat that the ideological rightism variable is excluded because of a very high negative correlation between its coefficient and that for leftism, and the COPEI identification variable is dropped because there are so few people in this category that the variable creates estimation problems.10 The residual deviance of a model using this dependent variable with only an intercept is 3831.08 on 4350 degrees of freedom; including all independent variables reduces the deviance to 3183.47 on 4320 degrees of freedom. A likelihood ratio test based on this information suggests that the independent variables, taken as a whole, are indeed valuable predictors of 1998 vote choice. The model suggests that a negative evaluation of the last government’s economic policy performance may be harmful for AD (relative risk of 0.80 with

ta b l e 5 . 4 . 1998 model of Venezuelan presidential vote choice Variable Name

Acción Democrática (AD) Estimate

PVZ Estimate 0.39 (0.53)

Movimiento V República (MVR) Estimate

Intercept

– 4.01 (1.40)**

Economic evaluations Corruption perceptions Ideological leftist AD party ID Social class “C” Social class “D” Social class “E”

– 0.18 (0.22) 0.65 (0.30)*

– 0.08 (0.11) 0.08 (0.15)

– 1.47 (0.59) 0.21 (0.10)* 0.29 (0.14)*

– 0.93 (0.56) 4.30 (0.34)** 0.29 (1.29) 1.75 (1.19) 1.60 (1.20)

– 0.23 (0.35) 0.49 (0.31) 0.12 (0.41) – 0.06 (0.39) – 0.39 (0.40)

1.02 (0.17)** – 1.71 (0.56)** 0.98 (0.50)* 0.74 (0.49) 0.79 (0.49)

notes: Maximum-likelihood estimates of the parameters in a multinomial logit model of vote choice. Null deviance is 3831.08 on 4350 degrees of freedom; residual deviance is 3183.47 on 4320 degrees of freedom. *( p < 0.05) **( p < 0.01).

ideological underr epr esentation & voter defection

133

a standard error of 0.17; the result does not reach standard levels of statistical significance) and Proyecto Venezuela (relative risk of 0.88 with a standard error of 0.04—significantly lower than 1 at the 0.05 level); by contrast a negative evaluation is beneficial for the leftist Movimiento V [Quinta] República (MVR) (relative risk of 1.20 with a standard error of 0.07—significantly higher than 1 at the 0.05 level). This seems reasonable even though the previous government, led by Rafael Caldera, was not directly associated with either AD or Proyecto Venezuela. Both parties planned to continue the generally market-oriented policies of the last half of Caldera’s administration, suggesting that voters who disapproved of the results of those policies should probably look elsewhere. MVR apparently proved somewhat of a magnet for these voters, since it offered the most anti-market message of any major party in the election. Overall, however, perhaps the most important message of these results is simply that economic evaluations do not line up terribly well with the decision to vote for or against the traditional party system in the 1998 elections. In particular, the relative risk for AD vis-à-vis the other two parties is not significantly lower than one, and indeed is almost equal with the relative risk for Proyecto Venezuela. Negative economic perceptions may play some role in driving voters to support MVR, but other factors most likely play a more central role in voters’ decision to abandon the traditional parties.11 On the direct effect of corruption perceptions, the results are relatively weak. Voters who are concerned about corruption appear to be somewhat more likely to turn out, but these voters are not obviously inclined to favor one specific party because of their attitudes about corruption. All three parties’ relative risk ratios for corruption fail to be statistically significantly different from zero (AD’s ratio of 1.58 has a standard error of 0.49, Proyecto Venezuela’s ratio of 0.89 has a standard error of 0.07, and MVR’s ratio of 1.10 has a standard error of 0.09). This does not mean that concerns about corruption are irrelevant in this election, as such concerns may play an important but indirect causal role mediated by voters’ party identification. However, these findings do suggest that voters’ contemporary perceptions of corruption, net of the perhaps important effect of their previous perceptions via changes in partisanship, are not a major predictor of vote choice. As the book’s argument predicts, AD is substantially weaker among ideological leftists than among others; the estimated relative risk of voting for AD

134 ideological underr epr esentation & voter defection

in the simulation condition, comparing leftists to non-leftists, is 0.23 (standard error of 0.14). That is to say, in the simulation condition leftists are four times more likely to vote against the traditional party system than non-leftists. Proyecto Venezuela is also hurt to some extent by leftists; the estimated relative risk is 0.46 (standard error of 0.06). Leftists are significantly more likely to vote for Proyecto Venezuela than for AD, but they are overwhelmingly most likely to vote for Chávez’s (ideologically leftist) MVR, which has a relative risk ratio for leftists vs. others of 1.60 (standard error of 0.10). In other words, individuals who fall in the space traditionally least well represented by the established Venezuelan parties are overwhelmingly unlikely to vote for those parties. They are much more likely to support Proyecto Venezuela than the remaining traditional party, AD, but they are particularly likely to support the most explicitly and emphatically anti-party-system option, MVR. These empirical patterns are strongly consistent with the hypothesis that ideological underrepresentation plays a driving role in Venezuela’s party-system collapse. Importantly, but unsurprisingly, voters who identify with AD are quite likely to vote for AD (a remarkable relative risk ratio of 17.29 with a standard error of 4.87) and almost entirely unwilling to vote for the anti-party-system MVR (relative risk of 0.04 with a standard error of 0.03). AD identifiers are also much less likely to vote for Proyecto Venezuela than similar voters without a party identification (relative risk of 0.38 with a standard error of 0.10). These results confirm that the decline in party identification, especially with AD, was an important precursor of party-system collapse; those voters who still identify with the party are much less likely than others to support outsider parties. Hence, the effect of corruption perceptions on party identification during the decade before party-system collapse, as analyzed in the previous chapter, may be an important part of the story of collapse in this crucial final election. Coefficients associated with the social class variables suggest that both AD and MVR may be more successful among the poor than among the middle class and the wealthy, although none of the results are statistically significant; this lack of significance was confirmed in the estimated relative risks, not reported here. There is thus no evidence in favor of the hypothesis that the informal sector played an especially important role in producing party-system collapse. The poorer groups, which in Venezuela are overwhelmingly part of

ideological underr epr esentation & voter defection

135

the informal sector, are if anything more likely than the working and middle classes to support the one remaining traditional party. In summary, votes in favor of parties outside the traditional system—i.e., votes in favor of party-system collapse—are best explained by concerns about corruption, directly in the earlier presidential contest as well as, in both elections, through their effects in bringing about a decline in party identification; and leftist ideology, which singles out those voters who fall in the ideological space left vacant by the traditional parties and thus experience significant underrepresentation.12 The Proportion of Leftists in Venezuela, 1998  Given the important coefficients associated with different ideological factors in the above analysis, it is critical to explore the distribution of ideologies in the Venezuelan electorate. After all, if leftists are an insignificant portion of the electorate, then it is futile to try to explain electoral outcomes in terms of their behavior. Figure 5.1 includes a histogram of Venezuelans’ ideological self-placements as of 1998 (as well as traditional-party ideological placements, not considered here). A substantial concentration of respondents can be found in the ideological center, and the right clearly outnumbers the left: 19.9 percent of respondents place themselves somewhere to the left of center, while fully 52.9 percent place themselves to the right of center. This distribution may be partly a methodological artifact; Venezuelan survey firms typically use elements of quota and convenience sampling in carrying out research, and they may also exclude some poor neighborhoods as being too dangerous to include in the sample.13 Thus it is possible that the left is somewhat underrepresented in the sample, in comparison to the population. This argument is buttressed by the discrepancy between the sample’s 36 percent support of Hugo Chávez and the figure of 56 percent that Chávez obtained in the actual elections.14 However, even if the size of the left is distorted in this sample, the fact remains that, in the Venezuelan electorate as of 1998, the left was a significant but not electorally decisive segment. The left’s 20 percent of the population may provide the initial base for a candidate, but it is incapable of electing a president or defeating the ideological right on its own. On the other hand, in a fluid campaign with multiple candidates like Venezuela’s 1998 election, solid support from 20 percent of the electorate is enough to put a candidate on the

136 ideological underr epr esentation & voter defection

strategic voting radar of centrists and of Venezuela’s large mass of nonideological voters, many of whom might be motivated to reject the traditional parties owing to concerns about corruption, as suggested by the 1993 results and the analysis of party identification in the previous chapter. The importance of the left as a solid electoral support base in favor of party-system collapse thus goes far beyond its role as roughly half of MVR’s final electoral coalition. Voters who are toward the left, are concerned about corruption, and do not identify with the traditional parties are the most likely of any relatively large category to turn to parties and candidates outside of the traditional party system; these voters can be said to be the “core constituency” for party-system collapse. Indeed, it seems reasonable to propose the counterfactual that, without voters in this and neighboring categories, party-system collapse would not have occurred. As Abramson et al. argue regarding the U.S. context: If there is a core of voters disaffected from the major parties that is continually willing to support an independent candidate, this gives that candidate a base upon which to build support gradually. On the other hand, if a candidate’s support is related to shortterm forces, and not to long-standing disaffection from the major parties, then there is little reason to believe that an anti-party movement led by an independent candidate will have much chance of success. (Abramson et al. 2000: 513)

Leftists’ ideological concerns about the two major Venezuelan parties were able to develop throughout at least the 1980s and 1990s, and corruption seems to have been an equally long-standing concern. Thus, consistent with the overall argument of the book, voters with either or both of these concerns provided a medium- to long-term core constituency in favor of party-system change, giving anti-system candidates the opportunity to build on other candidates’ past electoral coalitions to the point where such candidates were ultimately able to completely dominate the strategic-voting landscape.

5.3 VOTE CHOICE IN PERU

The analysis of Venezuela shows that ideological discrepancy, in which no traditional party represented either the left or the center-left, played a major role in producing party-system collapse, as did corruption perceptions—in substantial part via their effects over time on party identification. Unfortunately,

ideological underr epr esentation & voter defection 137

our ability to explore the Peruvian process of party-system collapse for parallel dynamics is limited because none of the surveys conducted during the course of the 1980s is available for statistical analysis. For this reason, most studies of the Peruvian party-system collapse have concentrated on economic and political aggregates (Tanaka 1998; Kenney 2004) or on ecological analysis of social-class voting (Cameron 1994; Roberts and Arce 1998). Because they draw on no information about individual beliefs and opinions, neither of these approaches is capable of shedding any direct light on this chapter’s questions about individual motivations regarding votes for non-party-system candidates. In an attempt to work around that limitation, this section analyzes a survey from 1992 with content relevant to the final stage of the party-system collapse in 1989–1990. This survey (which was also considered in the discussion of party identification above) was administered by the company Apoyo to 512 citizens of Lima in March of 1992.15 It asks about respondents’ perceptions of corruption during the García administration, economic policy performance in that period, and vote choice during the 1989 Lima municipal elections. All of the questions are retrospective and therefore probably contaminated by events and political discourse during the 1990–92 period, as well as by simple forgetfulness. These weaknesses notwithstanding, the data do provide some individuallevel evidence about connections between attitudes and the choice to vote for outsiders near the peak of the party-system collapse. This evidence echoes the results from the analysis of more directly relevant Venezuelan data, providing some evidence in favor of the cross-national generalizability of declining party identification, an ideologically underrepresented segment of the electorate, and concerns about corruption as a three-part explanation of party-system collapse. 5.3.1 The 1989 Lima Municipal Elections The 1989 municipal elections in Lima are a useful proxy for the first round of the 1990 Peruvian presidential elections. Both elections occurred during an acute societal crisis. The economy was in a process of hyperinflationary collapse, devastating the livelihoods and standard of living of millions of Peruvians at all socioeconomic levels. In addition, the conflict between the Peruvian state and the Sendero Luminoso insurgency was near its peak, with frequent terrorist attacks against civilians and compromised security in large parts of the country. Finally, as discussed earlier, the country was also facing political polarization;

138 ideological underr epr esentation & voter defection

major voices in the ongoing national debate represented a Marxist left as well as a free-market libertarian right. All of these national circumstances were present in both the 1989 municipal and the 1990 presidential elections. Moreover, there is at least some reason to think that electoral coalitions in the two elections were similar. Using the least-aggregated voting returns available, at the level of Lima’s districts, the correlation between the vote for ­Ricardo Belmont, the mayoral candidate of the outsider Obras movement, and the vote for Alberto Fujimori, the victorious outsider presidential candidate, is 0.72 (computed using data from Tuesta Soldevilla 1994). Correlations for the other parties are somewhat lower but generally comparable. Finally, the simplification of looking only at Lima is arguably acceptable. Metropolitan Lima contains about one third of the population of Peru. Furthermore, it supplied more than half of the votes for outsider candidates Alberto Fujimori and Mario Vargas Llosa in the first round of the 1990 presidential election (Tuesta Soldevilla 1994). Table 5.5 reports the results of a multinomial logit analysis of these survey data. The model is less complete in some ways than the Venezuela analysis, because the survey that provides a basis for it is substantially less rich. In particular, no measure of party identification is available. Nonetheless, the available

ta b l e 5 . 5 . 1989 Model of Lima municipal vote choice Alianza Popular Revolucionaria Americana (APRA)

Izquierda Unida (IU)

Frente Democratico (FREDEMO)

Intercept Economic evaluations

– 2.81 (0.69) 0.30 (0.44)

– 2.71 (0.69) – 0.50 (0.42)

– 2.18 (0.51) – 0.74 (0.30)*

0.66 (0.32) – 0.34 (0.22)

Corruption perceptions

– 0.26 (0.47)

0.84 (0.49)

1.15* (0.40)

– 0.08 (0.23)

Variable Name

Ideological leftist Ideological rightist Income

Obras

0.45 (0.64)

2.53 (0.48)**

– 0.76 (0.81)

– 0.58 (0.50)

– 0.70 (0.96)

– 11.94 (0.40)**

0.84 (0.82)

– 0.52 (0.64)

0.35** (0.14)

– 0.13 (0.12)

0.28 (0.20)

0.22 (0.21)

notes: Maximum-likelihood estimates of the parameters in a multinomial logit model of vote choice. Null deviance is 1434.62 on 2044 degrees of freedom; residual deviance is 1256.59 on 2012 degrees of freedom. *( p < 0.05) **( p < 0.01)

ideological underr epr esentation & voter defection 139

variables capture an important part of the book’s theoretical picture, allowing exploration of the roles of corruption perceptions, economic evaluations, ideology, and class (as proxied, to some extent, by income) in predicting vote choice. For this model as for the Venezuelan models discussed earlier, the simulation condition sets all variables other than those of current interpretive interest at their sample median values. Economic evaluations in the simulations (other than the simulation for the effects of those perceptions) are thus set at neutral. Corruption perceptions are present, in that a majority of respondents believe that President García was guilty of corruption during his administration. The median respondent is an ideological centrist and thus receives a zero for both rightism and leftism. Finally, the median respondent is in the second lowest of five income categories. As before, the model is complex, and effects require interpretation over and above the direction, magnitude, and significance of the coefficients in the table. Economic evaluations play only a limited role in explaining vote choice. Considering the relative risks for each party associated with a change from positive to negative economic evaluations, only the ratio for Obras (1.33 with a standard error of 0.17) is significantly different from 1. For APRA (0.8 with a standard error of 0.6), Izquierda Unida (0.5 with a standard error of 0.3), and FREDEMO (0.7 with a standard error of 0.2), there is a suggestion that economic voting may have hurt them all to some extent, but the evidence is weak and does not reach standard levels of significance. Concerns about the economy may push voters away from supporting the traditional parties of the center and left, but the effect is not substantively strong and the evidence is not statistically compelling; furthermore, a roughly equivalent push factor appears to exist with respect to the nontraditional FREDEMO party. Income, the rough proxy for social class available in these data, appears to play a role in voters’ decisions, but not a role especially compatible with the hypothesis that votes against the traditional party system were motivated by class concerns. One nontraditional party, Obras, appears to be stronger among lower-income than among middle-income groups (a relative risk ratio associated with the change from the lowest to the middle-income category of 0.7 with a standard error of 0.1), while the other party from outside the traditional system, FREDEMO, loses votes in relative terms among the poorest citizens (an estimated ratio of 2.1 with a standard error of 0.4). The two traditional

1 40 ideological underr epr esentation & voter defection

parties, APRA and Izquierda Unida, have estimated ratios that are statistically indistinguishable from 1 (1.5 with a standard error of 0.7 for both parties). In other words, social class considerations appear to be connected with voters’ decisions to support a particular nontraditional party, but perhaps not with decisions about whether to support the traditional party system. Corruption appears to have played a central, but slightly different, role in Peru than in Venezuela, serving primarily to damage the viability of the one remaining traditional party with direct national governing experience, APRA. That party, at the end of the contentious García administration, had a relative risk ratio connected with the difference between voters who believed that García was guilty of acts of corruption and those who rejected this claim of 0.3 with a standard error of 0.1. Obras was essentially unaffected, with a relative risk of 0.9 (standard error of 0.1) that is statistically indistinguishable from 1. The outsider-led FREDEMO party from the right gained significantly among such voters, with a relative risk of 3.0 (standard error of 1.6; because of asymmetry in the variance of the ratio, this estimate is significantly different from 1 at the 0.05 level). The much more established opposition party of the left, Izquierda Unida, may also have gained supporters among those convinced that García was guilty of corruption, although the results are not statistically significantly different from 1 (relative risk of 2.8 with a standard error of 2.8). Thus, unlike in Venezuela where concerns about corruption appear to have damaged both traditional parties, Peruvians concerned about corruption seemed to focus their discontent on the governing party. Izquierda Unida’s possible ability to capture some voters concerned about corruption was, in any case, insufficient to protect the party against collapse in light of its suboptimal ideological positioning. Rightist voters were notably unlikely to vote for the APRA municipal candidate (the estimated relative risk ratio for the simulation condition connected with a move from right to left ideology is 3.1 with a standard error of 1.5; because of asymmetries in the distribution of the relative risk ratio, this result is significantly different from 1 at the 0.05 level). This connection between ideology and municipal vote choice probably also held for APRA’s presidential candidate the following year. The other party-system candidate, from Izquierda Unida, had an even more impressive gain associated with right vs. left ideology: the estimated relative risk is 13.0 (standard error of 5.8). The candidates more clearly from outside the party system have ratios in the opposite direction. Obras has

ideological underr epr esentation & voter defection 1 4 1

a relative risk of 0.4 (standard error of 0.2), while FREDEMO has an estimated ratio of 0.2 (standard error of 0.1). These results provide some evidence that the traditional party system may have become weak toward the ideological right; separate results involving a change from centrist to leftist ideology provide statistically less significant but substantively interesting hints that the traditional party system underrepresented the ideological center. This interpretation, involving a weakness toward the center and the right, offers an interesting variation on Venezuela’s pattern of underrepresentation toward the left. In both instances, patterns of ideological underrepresentation as well as concerns about corruption appear to motivate voters to abandon the traditional parties—but the space on the ideological spectrum that is underrepresented varies between the two countries, implying that underrepresentation per se, and not a more specific pattern of underrepresentation of the left, is the variable of interest. APRA’s drift to the left during the late 1980s, in combination with corruption scandals, made center-right votes inaccessible to APRA’s 1990 candidate. Furthermore, the steep decline in APRA party identification removed an electoral cushion that would otherwise likely have advanced APRA’s candidate into the second round of the 1990 presidential election. AP’s breakdown in 1985 and subsequent decision to subordinate itself to the FREDEMO outsider movement on the right (see, again, Figure 5.2) deprived voters in the center and centerright of their traditional alternative to APRA. Clearly, the communists and socialists of IU were a poor solution to these voters’ quandary. Thus the center and center-right voters, the largest ideological bloc in the Peruvian electorate, were forced to choose between APRA and outsider candidates. Poor ideological positioning by the traditional parties, in combination with corruption scandals and the decline of traditional-party identification, thus may have produced the anger that led to party-system collapse in Peru as well as in Venezuela.

5.4 CONCLUSIONS

This chapter provides evidence about the role of corruption perceptions, ideological underrepresentation, and (for Venezuela; no Peruvian evidence was available) decline in identification with traditional parties in leading voters to abandon the existing party systems. Poor economic performance sometimes plays a role in vote choice, but there is little evidence that economic voting was

1 4 2 ideological underr epr esentation & voter defection

a central factor in the vote choices that produced party-system collapse. Instead, economic perceptions most often seem to have hurt both some traditional and some outsider parties. Likewise, there is evidence that considerations related to social class may be connected with voters’ choice of a nontraditional party, but perhaps not with the decision to support or abandon the traditional party system. Corruption, by contrast, does play a role in the choice to abandon the traditional party system, by weakening the relatively ideologically centrist APRA party and by hurting both Venezuelan traditional parties at certain moments in the process of collapse. Worries about corruption may also be an indirect cause of declines in identification with the traditional parties—and therefore may serve as both indirect and direct causes of voters’ decision to abandon the traditional party system. In both Peru and Venezuela, the ideological positioning of parties with respect to voters also seems to have played a central role in the vote choices that led to party-system collapse. In Venezuela, voters to the left of the political center had no representation within the traditional party system. When the influence of corruption concerns on some voters is combined with this pattern of ideological underrepresentation of other voters, a coalition of citizens angry with and motivated to reject the traditional party system emerges. In Peru, that coalition was provided by voters on the center and the right, as well as those who had strong concerns about corruption. This book’s theory highlights corruption and ideological underrepresentation as central causes of party-system collapse, arguing that such perceptions and situations make voters angry and therefore more willing to take the comparatively risky step of voting for candidates from outside the traditional party system. This chapter has shown the existence of a relationship between the specified attitudes and vote choice during the process of collapse, but it has not presented evidence regarding the affect-and-cognition mechanism hypothesized to tie these attitudes to the outcome of collapse. That is the task of the next chapter.

5 . 5 A P P E N D I X : VA R I A B L E S A N D S U RV E Y Q U E S T I O N S

As discussed in the main text, the analysis of the meaning of the left-right ideological spectrum in each country draws on questions from the 1995–96 wave of the World Values Survey. The questions used as independent variables ad-

ideological underr epr esentation & voter defection 1 43

dressed the environment, work issues, national economic development, specific economic issues, the quality of the political regime, poverty, religious and moral themes, and a handful of miscellaneous issues and values (political corruption, the importance of politics, willingness to fight in a war, and foreign aid). These questions are selected not because they cover all issues and values that could possibly be incorporated into the left-right spectrum, but because they address issues that are broadly representative of the universe of issue- and value-judgment content. Hence, if the ideological spectrum is grounded in such judgments, there should be a meaningful relationship between these questions and the left-right spectrum. The full list of variables and the texts of relevant questions are available from the author upon request. In order to generate the more easily interpretable restricted models reported in the main text, a regression using the full set of independent variables was run for each country. Then, the variables with a P value of 0.10 or less were selected. The final regression results reported for each country in the text use this subset of variables. This stepwise regression-like procedure is inferior to theoretically guided examination, but it is suitable for summarizing the empirical meaning of the left-right spectrum in each country. In light of this purely descriptive goal, other simplifying assumptions are adopted: all variables are constrained to have a linear, additive relationship with the dependent variable. Replacing this assumption with a more flexible functional form would add complexity to the model, which is why such an analytic move is not pursued here. In passing, it is worth noting that a more flexible functional form could only improve the overall fit of the model. The results reported in the text represent a lower bound for the extent to which the self-reported left-right spectrum is grounded in specific attitudes and issue stances.

chapter 6

VOTER AFFECT AND THE DEMISE O F   PA RT Y S Y S T E M S

T

h e c e n t r a l q u e s t i o n of this volume up to this point has been, why do voters, sometimes in very large numbers, decide to abandon a country’s traditional parties and support an outsider candidate? The preceding chapters have provided evidence regarding the perceptions, opinions, and identities that best predict the decision to support a nontraditional presidential candidate. This chapter moves to a finer theoretical level, analyzing the psychological mechanism connecting citizens’ attitudes with the decision to vote in favor of party-system collapse. Such a decision, as discussed in Chapter 1, inherently involves substantial uncertainty (Downs 1957; Stokes 2001; Magaloni 2006). Candidates from outside the political system typically have little governing experience, and often have a scant political reputation against which voters can evaluate the credibility of their campaign appeals. Furthermore, voting for a presidential candidate who does not come from an established party carries a heavy risk that one’s vote will be wasted; the presidency is a one-seat office, and hence voters who are averse to uncertainty will face severe strategic-voting pressures against opting for a candidate that does not represent a traditionally winning party (Duverger 1954; Cox 1997).

vot er a ffec t a nd t h e de m ise of pa rt y s ys t e ms

145

An explanation of the psychological mechanisms by which voters sometimes choose to accept these risks is needed, both to fill a central gap in the current theory of new-party emergence and to help distinguish between the kinds of negative events to which voters respond by supporting an outsider candidate and those for which they instead adopt other strategies, including voice or loyalty (Hirschman 1970) in conjunction with a vote for the incumbent party, support for the traditional opposition, or abstention. Empirically, these strategies for responding to disappointment with the political status quo appear to have quite different relationships with contextual variables such as ethnolinguistic fractionalization and economic performance (Powell and Tucker 2009). This finding seems odd if we regard all forms of political discontent as fungible, but it is easy to understand if diverse forms of political dissatisfaction correspond with different emotional states, which in turn activate contrasting decision-making processes in voters. This chapter tests the theoretical account developed in Chapter 1, which posits that emotional states are central to understanding how an electorate in favor of the risky outcome of party-system collapse emerges. When voters are broadly angry about the state of society, that anger reduces their aversion to uncertainty, a claim with strong roots in both theory and experimental evidence from the psychological literature on affect and decision-making. Because the risk and uncertainty connected with supporting an outsider without a political organization or track record is a central impediment to voting against the established parties, attitudes toward uncertainty have a causal effect on citizens’ choice to support a new party rather than an established one. The discussion below first presents some suggestive evidence that anger was common enough in the Peruvian and Venezuelan electorates at the time of party-system collapse to serve as a plausible component of an explanatory framework for collapse. Subsequently, the discussion turns to an experiment regarding the effects of anger on propensity to support a candidate from outside the party system. The experiment randomly exposes Peruvian subjects to one of three film clips selected to affect their emotional state and then asks them to participate in a simulated election between fictional traditional-party and outsider candidates. The results show that voters randomized to feel anger are significantly more likely to support the outsider candidate than are those assigned to the control group. Finally, suggestive

146

vot er a ffec t a nd t h e de m ise of pa rt y s ys t e ms

results from an observational study show how the emotional states considered in this chapter are linked with the political attitudes and issue positions considered in previous chapters.

6 . 1 T H E P R E VA L E N C E O F A N G E R IN PERU AND VENEZUEL A

Unfortunately, there appear to be no useful survey measures of political—or even general—affect in the Peruvian and Venezuelan electorates during the process of party-system collapse. Thus it is difficult to say with any precision how prevalent anger was, and whether it was indeed more common among supporters of outsider parties than among traditional-party voters. Even so, this chapter presents some forms of evidence that make this hypothesized causal connection more plausible. To begin with, it is easy to present qualitative evidence of high levels of anger against the traditional parties on the part of at least some Venezuelans and Peruvians during and after the period of party-system collapse. This evidence is not as useful as representative survey data would be, but it provides some reason to believe that anger was prevalent enough to explain party-system collapse. Of course, in Venezuela, a major expression of mass political anger was the set of riots in urban areas throughout the country known as the Caracazo. Smaller riots and acts of public protest persisted throughout the 1990s. The mass anger expressed through protests and political violence during the late 1980s and early 1990s might not have dissipated later in the decade. During Rafael Caldera’s second presidential administration, in the last half of the 1990s, impeached former president Carlos Andrés Pérez spent more than two years under house arrest for corruption. On one weekend evening during this period, a group of left-leaning university professors who had gathered to drink together and talk about politics decided to visit Pérez’s house and express their anger at him. They knocked on the door, waking the former president, and proceeded to shout at him for several minutes. Pérez commented that such visits were a regular part of his life, then saw them out.1 This episode demonstrates a rather high level of anger on the part of the academics who serve as protagonists; Pérez’s response to the event suggests, as well, that such anger was not altogether unusual.

vot er a ffec t a nd t h e de m ise of pa rt y s ys t e ms

147

Nor was anger during the period of party-system collapse targeted only at Pérez’s party, Acción Democrática. At least some former supporters felt similar emotions toward the other traditional party, COPEI. In late 1994, a former supporter and campaign worker for COPEI joined one of many demonstrations against the presidency of Rafael Caldera (the founder of COPEI who was at that point serving as president on an anti-party-system ticket). At that demonstration, he expressed his anger at both Caldera and the COPEI party by pouring gasoline over his collection of T-shirts from the COPEI campaigns on which he had worked and lighting them on fire.2 Although it is hard to know whether this individual’s emotional state was widely shared, the incident shows that at least some anger existed against COPEI and AD. It also raises the possibility that extreme anger can motivate even core supporters of the traditional parties to turn against the party system. High levels of anger against the Venezuelan traditional parties persisted well past the moment of party-system collapse. While conducting fieldwork in Venezuela between 2001 and 2004, I witnessed dozens of events and conversations that demonstrated a continued high level of anger at the Venezuelan traditional parties. One typical instance occurred in September 2001, in a book store, located in a middle-class neighborhood. Two men were conversing about their issues with the Chávez government. After several minutes of complaining about policy decisions, as well as about the president’s attitude toward the relatively well off, one man remarked that he sometimes wished that some other party could be put in charge of the government. The other, perhaps mistakenly, took this comment to mean that the first speaker hoped for the return of the traditional parties. In response, the second man began shouting about how the traditional parties were no alternative: “Those miserable incompetents were worse than the current one. I hope they die and go to hell!”3 These incidents provide some reason to believe that anger against the Venezuelan traditional parties was widespread, intense, and persistent in Venezuela during and even after the process of party-system collapse. There is similar evidence in Peru, although it is more difficult to point to large-scale expressions of anger such as the Caracazo. The ongoing conflict between Peru’s Maoist guerrilla insurgency, Sendero Luminoso, and the Peruvian state made large public demonstrations—let alone riots—far more dangerous in Peru than they were in Venezuela. It might be argued that the relative success

148

vot er a ffec t a nd t h e de m ise of pa rt y s ys t e ms

of that insurgency itself indicates a great deal of political anger in the country, but this claim is difficult to evaluate given the complexities of determining the motives behind and breadth of support for the guerrillas. Anger against the traditional parties was probably a partial motive for participation in the insurgency, but for other Peruvians the insurgency was probably a source rather than a consequence of anger. For the majority of Peruvians who were opposed to Sendero Luminoso, the government’s inability to eliminate the movement’s threats and violent actions was almost certainly a basis for anger.4 Such anger could also spill over to traditional parties that did not form part of the government. One activist within the leftist Izquierda Unida made public remarks in 1988 to the effect that members of Sendero Luminoso were “comrades who have taken the wrong path.” For months after making that statement, which was perceived as expressing insufficient condemnation, the activist could not appear in public without being shouted at as a traitor, a terrorist, or a secret member of the insurgency. When he entered a restaurant, he said, the other patrons would stand up and leave; some of his neighbors in an affluent Lima area moved rather than remain near him.5 Thus, the problem of political violence could generate anger against traditional politicians from various parties, not just those in government. Nor was political violence the only, or even necessarily the primary, cause of anger against the traditional parties in Peru. In 2005, a resident of Lima who had been deeply opposed to APRA leader Alan García during his presidency in the last half of the 1980s (and remained opposed to García during the 2000s) showed me an anti-García poster that he had saved from the period before party-system collapse. The poster showed images of poverty in Peru. A slogan was superimposed over these images: “Alan García, the worst human being in history.” Thus, the plight of the poor also motivated anger against the traditional parties; the APRA government’s move toward nationalizing the country’s banking system also probably provoked a great deal of anger. As in Venezuela, anger against the traditional parties did not always dissipate quickly, even after the collapse of the party system. For example, when García won a significant proportion of the vote in Peru’s 2001 presidential elections, two Peruvian brothers who owned and operated a corner grocery store in Berkeley, California, wore black arm bands to mourn the former president’s political revival. One of them complained vociferously that too many Peru-

vot er a ffec t a nd t h e de m ise of pa rt y s ys t e ms

149

vian voters had forgotten how bad things had been during the 1980s and how “wicked” García was. Examples such as these are numerous and well known. But they need not, of course, imply that most Peruvian or Venezuelan voters felt angry at the traditional political sphere during the process of party-system collapse, much less that such anger was a late step in the causal chain bringing about collapse. At the same time, these events do provide some support for those propositions. Indeed, mass protest events such as the Caracazo provide relatively strong support for the claim that anger was widely felt, even if it remains difficult to quantify the proportion experiencing that affective state. However, further evidence that anger causes people to vote against a traditional party system is still needed. For that, we turn to an experiment.

6 . 2 E X P E R I M E N TA L D E S I G N

As an empirical test of the proposed causal link between anger and votes for outsider candidates, an individual-level experimental design offers the strongest possible test—especially given the impossibility of retroactively acquiring contemporary data on affect and vote choice during the process of collapse itself. An experimental design provides a powerful test of causation because, given randomization, the only significant difference between the treatment groups should be the treatment itself. In this case, the treatment is a manipulation of subjects’ affective state. A variety of such manipulations is available (Coan and Allen 2007); the experiment discussed here relies on film clips to alter subjects’ emotional state. Emotions cannot be entirely assigned, but the component of research subjects’ affective mood that cannot be affected by the treatment should be neutralized by random assignment: the treatment and control groups should be approximately the same with respect to the mood that they would have if assigned to each treatment condition. Randomization more generally minimizes problems of confounding; psychological and other variables that might cause support for outsider parties apart from affective state should on average be equivalent for the treatment and control groups. Hence, causal inference is easier than it would be with a nonexperimental design, and the analysis generally does not include—or, indeed, require—control variables.

150

vot er a ffec t a nd t h e de m ise of pa rt y s ys t e ms

Furthermore, by ensuring that the stimuli used to induce affective states are nonpolitical, an experiment can distinguish between the affect and uncertainty theory developed above and the effects of political information. Such would not be the case for an analysis based on observational survey data, for example. In such data, respondents’ emotional state would likely be caused in part by politically relevant information such as opinions about the economy, evaluations of social policy, and so forth—information that might well have an independent causal effect on electoral decision-making. For these reasons, the experimental design discussed below offers in many ways a stronger test of the affect and uncertainty theory than an observational study would. Because nonpolitical affective manipulations are used here, the findings below contribute to the growing literature showing that nonpolitical affect, as well as affect with specifically political targets, can causally influence vote choice (see especially Healy, Malhotra, and Mo 2010). As is well known, experimental designs are intended primarily to achieve internal validity in causal inference; external validity remains a matter for conjecture and testing in future research. For the present theory, the scope for future testing is enormous, given that the hypothesized mechanism should be substantially general: it should apply wherever there is reason to regard established parties as less risky than unknown politicians. Furthermore, future testing could illuminate the ways that party-system traits alter the size of the electoral causal effect associated with anger. 6.2.1 The Mechanics of the Experiment The empirical test of the affect-and-voting theory described above relies on data generated via a computerized experiment carried out in Internet cafes in Lima and Cuzco, Peru, during July and August 2009. Subjects who consent to participate are invited to sit at a computer in the Internet cafe. A research assistant then loads up the initial Web address for the study and leaves the subject alone to watch an emotion-inducing video, participate in the simulated election, and complete a post-experimental questionnaire. Randomization and data collection are performed through the website. This approach has the practical advantage of reducing costs and errors, and it also increases the naturalism of the experiment. For most Peruvians, Internet cafes are a familiar and comfortable venue, as compared with, for example, a laboratory room in a university.

vot er a ffec t a nd t h e de m ise of pa rt y s ys t e ms

151

Upon beginning the experiment, each subject is immediately randomized to one of three affective treatments. In the anger treatment category, a clip from the film My Bodyguard is shown, in which one teenage male intimidates and attacks another teenage male and destroys his motorcycle. For the anxiety treatment, a segment of the Stanley Kubrick version of The Shining is shown, in which a child plays with toys on a carpet and is then surprised by a ball that rolls down a hallway where no one is present. These two clips have been previously validated and employed in psychological research on affect and ­decision-making (Gross and Levenson 1995); in the previous research, the anger clip was distinctly successful, while the anxiety film segment was less successful but nonetheless useful. As a preliminary check to ensure that the clips would be useful in a Peruvian cultural environment, they were shown to two focus groups before the experiment went into the field; the groups agreed that they produced the desired emotions. Nonetheless, in practice the anxiety clip was far less successful with this study’s group of subjects than it had been in previous work or in the pre-test. Finally, subjects assigned to the control condition (calm) were shown a slide show of waterfalls. After the initial emotional induction, the assigned emotions were maintained by playing selections of classical music previously validated to produce the desired emotions (Kreutz, Ott, Teich­man, Osawa, and Vaitl 2008). After viewing the randomly selected video clip, and while the music selections play, subjects learn about two fictional presidential candidates and finally vote for one or the other. The two candidates have positions on a range of important and less-important issues in Peruvian politics (corruption, crime, unemployment, inflation, free-trade treaties, indigenous communities, political experience, religious liberty and state support, poverty, agriculture, international politics, and public works), all designed and focus-grouped to be ideologically centrist. The major difference between the two candidates is that, for each subject, one is randomly assigned to be a candidate of the incumbent Partido Aprista Peruano, while the other belongs to an invented party, the Movimiento Peru y Progreso. Thus, while subjects are not explicitly guided to pay attention to partisanship in choosing between the two candidates, there is little else of substance that differentiates them. Furthermore, analysis shows no evidence that the voting patterns depend in any way on which of the two candidates is from the traditional party and which is from a new party. Hence, it is reason-

152

vot er a ffec t a nd t h e de m ise of pa rt y s ys t e ms

able to treat voters’ choice between the candidates as a choice between the relatively lower-risk alternative of a traditional-party candidate and a higher-risk but otherwise similar outsider candidate. The experiment includes only an established-party incumbent and an outsider; no established-party opposition candidate is provided. A pilot version of the study included three candidates, but subjects grew bored with the process of reading position papers before exploring all three to their satisfaction. Instead, they reported feeling dissatisfied with their decision-making because of the relatively challenging task of evaluating three brand-new politicians at once. A simpler, two-candidate campaign led participants to report less boredom and dissatisfaction with the decision-making process. Even so, this limitation in the range of candidate choice represents a potential limitation of the research design, albeit one that is ameliorated to some extent in the statistical analysis below. After voting in the simulated election, subjects complete a brief questionnaire about their current emotional state, their ideological ranking of the two candidates, their demographics, their degree of subjective risk acceptance, and their evaluation of the emotions that would most probably arise in a variety of hypothetical social, economic, and political scenarios. These variables are discussed in greater depth below. As with all experiments, of course, the key inferential leverage derives from simple comparisons of treatment group outcomes; the information derived from the post-experiment questionnaire is clearly secondary in nature. 6.2.2 Peru as the Research Context Peru was selected as the context for research for a mixture of considerations related both to experimental design and to the larger project of understanding party-system collapse. It is important, for purposes of experimental realism (Aronson et al. 1990), that participants in the experiment belong to a society in which the presence of both an important nontraditional candidate and at least one important candidate from an established party are common in presidential elections. This consideration tends to rule out as research contexts those countries with either a strong or a transient and volatile party system. Peru at present does not fall at either extreme; it has experienced substantial party-system turmoil and has recently had several nontraditional presidential candidates and even elected presidents, but it also has at least one strong, estab-

vot er a ffec t a nd t h e de m ise of pa rt y s ys t e ms

153

lished party—the Aprista Party, which controlled the presidency at the time of the experiment. Hence, an election between one party-system insider and one outsider candidate is plausible to even attentive Peruvian voters. In addition, of the set of countries that have a recent history of party-system instability and successful outsider candidates but nonetheless have at least one well-rooted traditional party, Peru was selected as a research context because of its value for this book’s larger project of understanding party-system collapse. A parallel experiment in Venezuela would be useful, but it is unclear that any traditional party in Venezuela is electorally strong enough to serve as an experimentally realistic alternative to the relatively new parties that compete in the country’s elections. In any case, if there is a general causal effect of emotion on willingness to vote for an outsider candidate via risk aversion, that effect should be as real in Peru as in any other context; finding a causal effect in Peru at least demonstrates the possibility, if not the universality, of the causal links in this book’s theory of party-system collapse that revolve around affect, uncertainty, and voting. That said, Peru’s political system has some features that are clearly relevant to thinking about the generalizability and theoretical implications of the experiment’s empirical findings. Of these, most noteworthy is the fact that Peru’s party system has been quite weak for some decades. Two of the past three presidents of Peru at the time of the experiment(Alberto Fujimori and Alejandro Toledo) were nontraditional politicians, supported by parties recently founded by the candidates themselves. Likewise, the victor of the 2011 presidential elections, Ollanta Humala, was supported by nontraditional parties. Hence, Peruvians have substantial experience with candidates, and even presidents, from previously unknown parties. This fact is convenient for the experiment in one respect: it enhances experimental realism in that relatively few participants will reject out of hand a scenario in which one of the two main presidential contenders is from a nontraditional party. However, this history of nontraditional candidates and outsider presidents probably reduces the degree of risk that most Peruvians would subjectively assign to the act of voting for an outsider candidate. This, in turn, may either reduce or increase the magnitude of the causal effect of emotion, although the affect and uncertainty theory predicts that the effect will be in the same direction in Peru as in other political contexts. Replication of this

154

vot er a ffec t a nd t h e de m ise of pa rt y s ys t e ms

analysis in countries with quite different party systems will probably be necessary in order to work through these issues. Hence, generalization in the sense of projecting the causal effect found in this experiment to voters in other country contexts or time periods must be undertaken with caution. Given the neurological nature of several of the linkages in the theory driving this study, more theoretical application of this study’s ideas in other domains may nonetheless be fruitful, but the applicability of these findings to other contexts remains a topic for further empirical testing. 6.2.3 The Sample The experiment was administered to a convenience sample of 150 subjects in Cuzco, Peru, and 300 in Lima, Peru—the much larger capital city of the country. The subjects were recruited through nongovernmental organizations (NGOs) operating in popular-sector areas within the two cities: Alternativa in Lima, and Arariwa in Cuzco. These NGOs carry out a wide variety of economic development and community-building activities that reach a diverse collection of Peruvians, both poor and middle class. Subjects were recruited by selecting a variety of different programs whose clienteles had different education and economic profiles, and inviting participants in those programs during a defined window of time to participate in the experiment. The resulting sample is not broadly representative, but it is nonetheless far more diverse and plausible as a basis for theoretical extrapolation than if the participants had been recruited on a university campus, for example. When asked what social class they regard themselves as belonging to, 11 percent identified themselves as lower class, 78 percent said they were middle class, and 11 percent self-identified as upper class, although it should be noted that respondents are disproportionately likely to self-identify with the middle class in Peru, as are citizens in many other countries. The mean age of participants was 31, with some as young as 21 and as old as 81. The median respondent had completed college-preparatory studies but had not attended college, with the group of subjects including some who had only a primary school education and others who had completed a university degree. The median respondent reported reading a newspaper a few times a week, with similar moderate levels of attention to other political information sources. These variables have statistically satisfactory balance between the anger treatment group and the other

vot er a ffec t a nd t h e de m ise of pa rt y s ys t e ms

155

two groups; P values calculated by exact randomization testing methods (Ernst 2004; Pesarin and Salmaso 2010) are mostly quite high.6 The most problematic variable is social class, which shows a small but almost statistically significant (P = 0.09) difference between the treatment and control groups; however, conditioning on social class affects none of the results discussed below. Compared with the Peruvian population in general, participants in this experiment are more likely to be middle class, more educated on average, and probably somewhat more likely to pay attention to political news. While the subjects in this experiment are thus not a representative sample of Peruvians, they are nonetheless a diverse group and may therefore serve as a starting point for theoretical inferences, if not statistical generalizations. 6.2.4 Effectiveness of Treatments For this group of subjects, the anger treatment was an effective manipulation, but the anxiety treatment was not. This unfortunate outcome limits the ability of this study to fully test the anxiety component of the hypothesis; instead, the analysis focuses on anger as the primary treatment. Receiving the anger treatment more than doubles the proportion of respondents who report feeling angry at the end of the experiment,7 from 5.6 percent among people in the anxiety group to 14.0 percent among people in the anger treatment, with the calm group falling somewhat above the anxiety group. Differences between anger and either calm or anxiety are statistically significant, with a P value of 0.02, but the difference between calm and anxiety is not significant. Of course, 14.0 percent is a low proportion in absolute terms, but this result is perhaps less problematic than it may initially appear. The treatment used in this study necessarily involves a relatively modest emotional stimulus; practical and ethical considerations preclude more powerful potential treatments. Furthermore, the indicator of emotional state used in this study is dichotomous, so subjects who felt a low level of anger may not have considered it salient enough to report. Hence, the treatment may well have had a low-level effect on a broader group of subjects than those who reported feeling angry. In any case, the results here suggest that the manipulation had at least some effect in the right direction; if the compliance rate is regarded as somewhat low, this will make the treatment effects reported in the next section all the more impressive.

156

vot er a ffec t a nd t h e de m ise of pa rt y s ys t e ms

The anxiety treatment, by contrast, was essentially unsuccessful in spite of a weak tendency in the desired direction. Of the anxiety treatment group, 40.1 percent report feeling anxiety, while 34.4 percent of subjects in the other two groups report the same experience. The difference, of course, is not statistically significant, with a P value of 0.48. The video and music used in the anxiety treatment was much less effective for this experiment’s group of subjects than it has been for prior experimental groups in the United States or for this study’s Peruvian pre-test group. A plausible explanation is that the venues where the experiment was conducted—public and often busy Internet cafes—may not have provided an ambience compatible with the relatively subtle fearful mood of the selected film clip; this aspect of the setting could account for differences in treatment efficacy between this study and prior studies that did mood induction in a laboratory context, and also the contrast between the pre-test (in a much more private university computer center) and the actual experiment. In any case, because the anxiety treatment failed for this experiment, this study cannot speak to that aspect of the theory developed above. However, the anger condition was more successful, so the discussion of experimental results below focuses on the contrast between the anger group and the combined control group made up of subjects assigned to the anxiety and calm conditions.

6.3 FINDINGS

First, there is strong evidence in favor of the existence of a causal effect by which assignment to the anger treatment group produces an increased likelihood of voting for outsider candidates. Second, the results are compatible with, but not strongly supportive of, the hypothesis that anxiety reduces the likelihood of supporting candidates from unknown parties. Third, there is evidence supporting the theoretical account postulating risk aversion as a central part of the mechanism linking citizens’ affective state to their vote choice. 6.3.1 Affective States and Vote Choice: Direct Effects The central findings of the experiment are reported in Figure 6.1, which shows the mean rates of outsider candidate voting for the anger treatment group and for the combined two other groups.8 These are simple sample proportions, with confidence intervals designed for pairwise comparison. No multivariate regres-

vot er a ffec t a nd t h e de m ise of pa rt y s ys t e ms

157

f i g u r e 6 . 1 . Affect and vote choice. source: These data collected by the author in Lima and Cuzco, Peru, July–August 2009.

sion or comparable model is used; instead, the analysis relies on the experimental design to, on average, balance confounders across the treatment groups.9 Substantial numbers of subjects in all treatment conditions vote for the outsider candidate—a result that is not particularly surprising given the Peruvian context in which non-party-system candidates routinely compete in elections at all levels and have won the presidency in recent decades. The key finding, however, is that the rate of voting for outsider candidates among people assigned to the anger treatment is substantially, and significantly, higher than the rate for people assigned to the other two treatments. For the anger group, 60.9 percent support non-party-system candidates, while the equivalent rate for the control group is 49.2 percent, a difference that has a P value of 0.01. Hence, there is support for the hypothesis that anger increases citizens’ willingness to vote for candidates from new parties.10 As a partial corrective to the design limitation discussed above, in which the structure of the simulated election provides no alternative for those who wish to vote against the incumbent but in favor of some traditional party, it

158

vot er a ffec t a nd t h e de m ise of pa rt y s ys t e ms

is useful to focus in on comparisons conditional on partisanship or attitudes toward the government. If the effect of treatment is similar among both identifiers with the governing party and others, then this becomes evidence that the effect reported here may not depend on whether the subject’s preferred party is available as part of the simulated election. Unfortunately, it is hard to draw strong conclusions in this regard, as the experiment’s subject pool contains only thirty-six self-reported identifiers with the governing party. Nonetheless, among this small group, the estimated effect of treatment is 8.0 percent, a result that is in the same direction as, and of comparable magnitude to, the overall result. A t test for interaction between treatment and identification with the governing party has a P value of 0.93; there is thus no evidence that the effect of anger on propensity to vote for outsider candidates is different for identifiers with the governing party than for others.11 Of course, subjects’ identification with the governing party may not completely capture the key issue of interest: desire to punish the governing party by voting for some opposition figure, whether from the traditional system or from outside of it. An alternative is to condition on subjects’ evaluation of the Peruvian state’s performance along a number of key dimensions. The online appendix reports the full results of a regression model estimating the effect of treatment on propensity to vote for an outsider, conditional on subjects’ evaluations of the state’s performance with respect to corruption, crime, unemployment, poverty, education, and overall government capacity.12 In this model, several kinds of negative evaluations do predict voting for an outsider candidate, sometimes with P values that approach or (in one instance) reach statistical significance. Moving from the most positive to the most negative on a scale of evaluations of government performance with respect to corruption predicts a 20 percent increase in the probability of supporting the outsider candidate; the government capacity scale has an identical effect within the model. A similar move on the unemployment scale predicts a 24 percent increase in the probability of voting for the outsider. Furthermore, moving from the most positive to the most negative on the education scale predicts a 36 percent increase in the probability of voting against the traditional-party candidate. The education and corruption coefficients are significant at the 0.05 level, while the government capacity coefficient is significant at the 0.10 level. Evaluations for poverty and crime have effects in the model that are, perhaps surprisingly, in the opposite direction.

vot er a ffec t a nd t h e de m ise of pa rt y s ys t e ms

159

However, these results are of secondary interest. What is most important is the regression-adjusted estimate of the effect of the anger treatment. The effect is essentially unchanged in comparison with the unconditional estimate; indeed, it is slightly larger (0.13) and more significant (P = 0.01). Conditioning on respondents’ evaluations of the incumbent APRA government’s performance does not alter the effect of the emotional manipulation on subjects’ propensity to support the outsider candidate. Furthermore, an ANOVA test of the hypothesis that there is an interaction between treatment and the three evaluations that have coefficients in the model which are in the expected direction and significant at the 0.10 level produces a P value of 0.17. Hence, subjects who have more negative evaluations of the incumbent government’s performance seem to be responding to treatment in essentially the same way as those with more positive evaluations. Overall, these conditioning results suggest that the effect of treatment on willingness to support the outsider candidate does not merely reflect voters’ desire to punish the incumbent party. Subjects with quite different predilections to vote against the government (either because of their party identification or their evaluations of the government) are similarly affected by the treatment. These findings support the argument that anger affects willingness to vote outside the established party system, not just propensity to punish the incumbent. A further implication of the affect-and-risk argument is that the anger-outsider voting effect should be mediated by attitudes about risk; the next section tests this implication. 6.3.2 Exploring Causal Pathways The experiment confirms the link between anger and attitudes toward uncertainty, and also supports the hypothesis that uncertainty is the major causal linkage connecting anger and willingness to vote for an outsider candidate. Of course, because aversion to uncertainty is not directly randomized, the basis for the second inference is necessarily somewhat weaker than for the anger-outsider vote causal connection discussed above. To begin with, the experiment confirms prior findings connecting anger with acceptance/aversion regarding uncertainty.13 The anger treatment group is about 8.1 points more acceptant of uncertainty than the two control groups, on average; this difference is clearly statistically significant (P < 0.01) and is

160

vot er a ffec t a nd t h e de m ise of pa rt y s ys t e ms

probably at least a substantively moderate effect on a 100-point scale. Hence, for this sample as for those used in prior experiments, anger reduces people’s aversion to uncertainty. Furthermore, there is some partial evidence based on an analysis of attitudes toward uncertainty as a mediating variable (Baron and Kenny 1986; see also Imai, Keele, and Yamamoto 2010) between the experimental treatment and the outcome of supporting an outsider candidate, that aversion to uncertainty may be causally connected with the choice to vote for a candidate from an unknown party.14 The strategy used here to test for mediation is to first use a linear regression to estimate the effect of the anger treatment on subjects’ attitudes about risk. Also, logistic regression is estimated using vote for the outsider candidate as the dependent variable and two independent variables: attitude regarding uncertainty, and assignment to the anger treatment group. A pair of causal effects are simulated for each individual in the data: the effect of switching them from the control group to the treatment group on the anger variable, and the effect of changing their attitudes about risk while holding the main treatment constant. The discussion below focuses on the sample averages of those two treatment effects, as well as on the proportion of the effect of the main treatment that can be accounted for by the causal pathway involving attitudes about risk. Three versions of the analysis are reported; Model 1 incorporates no additional control variables; Model 2 controls for subjects’ age, social class, and educational levels; and Model 3 controls for subjects’ leftright ideological self-placement, their internal and external political efficacy, and their party identification. The results for all three analyses, as shown in Table 6.1, are notably similar, and all are consistent with the proposition that attitudes about uncertainty are the major intervening variable between the experimental manipulation of emota b l e 6 . 1 . Effects of anger treatment on outsider voting, as mediated through attitudes

toward risk Result

Model 1

Model 2

Model 3

Mediation effect

0.08 (0.04– 0.13)

0.08 (0.03–0.13)

0.08 (0.02–0.13)

Total effect Mediation as share of total effect (percent)

0.10 (0.00–0.20)

0.09 (0.00–0.20)

0.09 (– 0.02–0.19)

81

81

79

vot er a ffec t a nd t h e de m ise of pa rt y s ys t e ms

161

tion and the outcome of support for an outsider candidate. In each analysis, the causal effect of the anger treatment through the risk pathway accounts for about 80 percent of the total effect of the treatment on subjects’ willingness to vote for the outsider candidate. Of course, the credibility of this inference relies on the assumption that the right control variables have been found to deal with the nonrandomized nature of subjects’ attitudes about risk, an assumption that is not fully justified by the experimental design. Nonetheless, while far from definitive, these findings are at least suggestive.

6 . 4 I M P L I C AT I O N S F O R PA RT Y- S Y S T E M C H A N G E

The experiment discussed above provides strong support for some key components of this book’s theory of party-system collapse, particularly the links among emotional states, voters’ risk calculus, and their willingness to vote for nontraditional candidates. The strongest evidence involves the link between anger and willingness to vote for political outsiders, for which the simple intentto-treat analysis found a significant causal effect. Logically weaker evidence of one form or another also supports the proposition that anxiety reduces citizens’ willingness to support candidates from unknown parties, as well as the hypothesis that attitudes of aversion/acceptance regarding uncertainty plays a key role as an intervening variable in the causal mechanism. But do these causal mechanisms really connect with the findings from earlier chapters about the importance of corruption and ideological underrepresentation in accounting for party-system collapse? This section offers some initial evidence connecting these issue concerns with anger—and therefore with the micro-level causal pathway demonstrated above. A post-experimental survey given to subjects provides some evidence that concerns about corruption and the quality of ideological representation tend to produce anger, whereas economic problems are more likely to produce anxiety. In the survey, subjects were presented with a series of anecdotes about situations that Alonso, a fictitious local taxi driver, finds himself in.15 The situations involve personal economic trouble,16 society-wide economic difficulties,17 substantial and widespread corruption,18 poor legislative representation,19 and high crime.20 For each situation, subjects are asked to identify all of the emotions that Alonso is likely to feel.21

162

vot er a ffec t a nd t h e de m ise of pa rt y s ys t e ms

The first column in Table 6.2 shows the percentage of respondents who predicted that Alonso would feel angry in each of the five situations. The predicted levels of anger for the two economic situations and the situation involving crime are comparatively quite low; for these three situations, about 20 percent of respondents expected Alonso—and by extension themselves—to feel anger. This, of course, makes sense. Each situation potentially places an individual’s survival at risk, and does so in such a way that blame is often very difficult to assign to any individual. Hence, as this book has argued, fear is a more likely response to such contexts than anger. By contrast, for the situations involving corruption and poor representation, a substantial majority of subjects predicted anger: over 70 percent for corruption and over 80 percent for poor representation. Once again, these findings are easy to explain. Both corruption and poor representation involve derelictions of duty by specific, named individuals. Thus they impose costs on the individual while making blame for those costs easy to assign. Anger is therefore the likely result for most individuals. The pattern across these five situations for anxiety is an approximate mirror image of the anger finding, as can be seen in the second column of Table 6.2. Although the baseline level of predicted anxiety across all five situations is moderately higher than the anger baseline, nonetheless the corruption and especially the underrepresentation situations elicit substantially lower levels of predicted anxiety than do the economic situations. Once again, these results fit with standard accounts of anxiety and/or fear as involving situations in which losses are imposed on an individual and blame for those losses is difficult to assign. These findings also connect this chapter’s findings—that there is a causal pathway from anger through to voting against the traditional party system—

ta b l e 6 . 2 . Situations and predicted emotional responses

Situation

Proportion Predicting Anger (percent)

Proportion Predicting Anxiety (percent)

Recession

22

88

Money trouble Crime Corruption Poor representation

22 16 74 86

85 53 38 27

vot er a ffec t a nd t h e de m ise of pa rt y s ys t e ms

163

with the findings of the previous chapters—that corruption and ideological underrepresentation, but not concerns about the economy, play a major role in explaining party-system collapse in Peru and Venezuela. Economic crisis is especially connected with anxiety, rather than anger, and as a result may bring about risk aversion as opposed to the risk acceptance that appears to promote anti-party-system voting. Thus, the previous chapter’s finding that negative economic perceptions were at most weakly related to the decision to vote against the traditional party system in Peru and Venezuela makes sense: this category of negative perceptions may well provide a motive for punishing the incumbent party, but it does not bring about the desire for radical, risky change constituted by party-system collapse. By contrast, the key explanatory variables in the survey research (corruption and poor ideological representation) are linked with anger and therefore may activate the causal chain tested in the experiment above, leading to an increased appetite for risk and thus to support for parties and candidates from outside the party system.

6.5 CONCLUSIONS

This chapter concludes the analysis of voter decision-making during the process of party-system collapse. The analysis has shown that voters who are worried about corruption are more likely to sever ties of identification with the traditional parties than are similar voters who are less troubled by, or perhaps less aware of, corruption scandals. Citizens who are poorly represented ideologically by the traditional parties are more likely to vote against the party system than those whose views are better represented. Party identification can serve as a cross-pressure, motivating even poorly represented individuals to keep voting for the traditional parties; thus corruption plays an important indirect role in accounting for partysystem collapse, through its erosion of voters’ identification with the traditional parties. In both of these analyses, corruption and problems of representation explain citizen’s role in party-system collapse better than variables connected with the economy or with social class—the two major alternative explanations. The qualitative, survey, and experimental evidence presented in this chapter explains why these relationships have this pattern. Economic difficulties, potentially including some aspects of social class, tend to produce anxiety or fear, emotional states that may not motivate votes against the party system.

164

vot er a ffec t a nd t h e de m ise of pa rt y s ys t e ms

By contrast, concerns about corruption or representation bring about anger, in turn motivating votes in favor of outsider parties. These links complete the voter-side explanation of party-system collapse. However, one major aspect of this book’s theory of party-system collapse remains untested: its account of the party-organizational reasons why Peruvian and Venezuelan traditional parties could not mitigate corruption scandals and failed to ideologically reposition themselves so as to ameliorate problems of ideological underrepresentation.

chapter 7

EXPLAINING P A RT I E S ’ D E G R E E OF IDEOLOGICAL FLEXIBILITY

I

n l i g h t o f t h e a n a l y s i s in the previous chapters, which shows that ideological underrepresentation was a central cause of voters’ decisions to abandon the traditional parties and bring about party-system collapse, a major puzzle arises. If the goal of party leaders in Peru and Venezuela was to win elections, why did they not reposition their parties on the ideological spectrum so as to win the votes of the neglected segment of citizens? Specifically, why did no traditional party in Peru adopt an ideological position on the center to center-right, where a large block of underrepresented voters was available? Likewise, why did no Venezuelan traditional party move to fill the ideological gap toward the left of center? APRA’s successful presidential campaign of 1985 emphasized mostly centrist themes with some leftist overtones; redistribution was not a focus, but using the state to manage the economy in a way that favored the socially excluded and marginalized was one of Alan García’s principal campaign messages (Graham 1992: 88–91; Cameron 1994: 42–48). However, as García’s essentially centrist, interventionist administration faced economic crisis and a seemingly rising challenge from Izquierda Unida on the left, APRA’s strategic response was to shift somewhat to the center-left. In particular, García announced an

166

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

attempt at nationalizing the country’s banks and insurance industry, and the president also switched from the centrist pro-business rhetoric that he had used, in conjunction with statements about social inclusion and the reduction of poverty, in his campaign and early administration in the direction of a more polarizing, critical rhetoric about private enterprises (Graham 1992: 111–21). This leftward shift produced the pattern discussed in the previous chapter, in which the traditional Peruvian parties had largely abandoned the center and center-right. APRA’s shift toward the center-left during the late 1980s opened a strategic opportunity for Acción Popular, which had traditionally been a centrist party, albeit with tendencies toward the center-right. However, the party failed to take advantage of the ideological opening. In the run-up to the 1990 Peruvian presidential elections, Fernando Belaúnde Terry (the founder and long-standing leader of Acción Popular) formed an alliance that, in effect, subordinated his party to the rightist candidacy of independent Mario Vargas Llosa. Throughout the campaign period, tensions repeatedly emerged between Acción Popular leaders and Vargas Llosa’s campaign team over ideological strategy (as well as more purely political concerns), with traditional-party politicians urging the more centrist approach that was electorally rational and Vargas Llosa’s team advocating the rightist, economically liberal message that they saw as necessary for Peru’s economic development.1 These conflicts notwithstanding, Acción Popular never broke out of the alliance with Vargas Llosa to run the centrist campaign that might have filled the ideological gap which outsider candidate Alberto Fujimori successfully occupied. The last Peruvian traditional party, Izquierda Unida, also faced a major strategic opportunity in the late 1980s: it could move toward the center position from which APRA had recently departed. If Izquierda Unida could moderate its message enough to criss-cross with APRA on the ideological spectrum, the traditionally leftist movement could well gain enough electoral support to win the presidency of Peru, while also closing the gap that allowed for party-system collapse. Yet this possibility was ruled out by severe intra-party resistance during Izquierda Unida’s preparations for the 1990 campaign. Alfonso Barrantes, the party’s presidential candidate in 1985 and mayor of Lima from 1984 to 1986, wanted to run again in 1990 using a more centrist appeal. However, the party apparatus’s large majority of committed leftists resisted this proposal, in the end

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

167

producing a schism and a total electoral and organizational collapse (Cameron 1994: 77–96; Roberts 1998: 246–57). In Venezuela, Acción Democrática never really tried to fill the country’s representative gap toward the center-left during the 1990s. During the early 1990s, COPEI’s traditional leader, Rafael Caldera, made an ambitious attempt to move the party toward a center-left position, and therefore to fill the Venezuelan party system’s ideological gap. Caldera’s project for the party included opposition to market-oriented economic reforms, a reemphasis on state intervention in the economy for redistributive purposes, and a renewal of the institutions of representation. However, in the ensuing intra-party struggle for control, Caldera was soundly defeated by Eduardo Fernández, who advocated a continuation of COPEI’s traditional center-right ideological strategy.2 As we saw in the previous chapter, COPEI’s subsequent decision to retain a centerright ideological position left a substantial number of voters at the center and center-left available for mobilization. As a consequence, Caldera’s independent candidacy for president in 1993 on the center-leftist strategy that he had advocated for COPEI was victorious; COPEI’s candidate was less successful. In Argentina, neither the Peronists nor the Radicals were trapped in electorally suboptimal ideological positions during the 1980s and 1990s. Somehow, the Peronist party alone, during the 1990s and early 2000s, occupied positions from the neoliberal center-right through to the center-left, including Carlos Menem’s movement toward a market-oriented reform program combined with some degree of social relief through largely party-distributed aid, and also Néstor Kirchner’s moderately anti-globalization and pro-redistribution rhetoric and economic policy. The Radicals appear to have also been reasonably flexible, and their major electoral difficulties have arisen owing to concerns about competence and the management of corruption scandals. Summarizing across the three countries, the historical record suggests that the Peronists had a particularly high level of ideological flexibility, with the Radicals scoring lower but still relatively high. APRA and Acción Popular rank somewhere in the middle, with enough flexibility to move in the wrong ideological direction but not enough to quickly recover and retake the ideological center. The Venezuelan parties, together with Izquierda Unida in Peru, fall toward the bottom in ideological flexibility. What can account for these observed differences? A party that wants to meaningfully alter its ideological profile leading into an election cycle faces a

168

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

problem of credible commitment. It is easy to announce a new political orientation, but if new statements are inconsistent with the past policy stands and governing decisions of both the party and its most prominent candidates, voters are likely to disregard them as cheap talk and instead rely on the perhaps more robust evidence of history. For example, in the 1990s and 2000s, former Peruvian president Alan García wanted to change the public profile he and the APRA party carried from one of reckless economic heterodoxy toward a more moderate position combining traces of nationalism with an image of globally oriented economic professionalism. This effort required more than a decade of work, during which García taught university courses on public and political management and issued repeated public statements regarding what he had learned from the failure of his first presidency. Indeed, even this extensive effort left García’s profile only partially altered: a substantial number of ideologically centrist Peruvian voters regarded him with suspicion even during his successful presidential campaign of 2006. How, then, do parties credibly signal a changed ideological and policy profile? A particularly compelling signal arises when the party’s leadership team is partially or completely replaced with an alternative set of leaders who have a long-standing reputation that associates them with the new profile (Panebianco 1988: 37–45, 243–50; Harmel and Janda 1994: 266–67; see also Grzymala-Busse 2002: 71–73). Thus, if an alternative ideological faction within a party takes control of the organization and nominates candidates connected with their views, it is perfectly reasonable for the electorate to regard this as a credible change in the appeals and potential future governing actions of the party. After all, for the party to renege on the change in profile in this circumstance would require the newly empowered faction either to abandon its long-standing ideological commitments or to relinquish its power within the party. Because both of these actions would typically be contrary to the interests of faction members, there is no reason that voters should expect such behaviors. Hence, policy change becomes electorally credible when it is accompanied by the right kind of personnel change within the party. In order for such ideological changes in party leadership to occur, two prerequisites are centrally important. First, and most self-evident, the party must have a supply of seasoned potential leaders who have a significant public record that establishes an ideological and policy profile different from that of the party

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

169

as a whole. If no such leaders exist, then the party cannot be taken over from within in a way that credibly signals ideological change. Instead, the party will have to either persist with its established ideological appeal or recruit a new leadership team from outside the party—a striking and unusual step, to be sure. A party is more likely to develop an ideologically diverse pool of seasoned potential leaders if the body of activists and lower-level organizational workers within the party is itself ideologically diverse. When the low-level actors within the party reflect a variety of ideological and policy stances that are in tension with the position of the party’s current central leadership, they constitute a meaningful constituency within the organization that can advocate and support political careers at the local and regional level for potential leaders with diverse ideological commitments. Without such a demand, it is unlikely that diversity in potential leaders would arise; the central party leadership will usually cultivate the careers of intra-party allies, not rivals. This study treats the ideological diversity among party activists during the process of party-system collapse as predetermined, regarding this aspect of parties’ organizations as primarily due to factors such as the prior history of intraand inter-party competition, as well as to geography and social diversity within a country. Of course, it is interesting to ask when and how parties develop an ideologically diverse corps of activists. Work by other scholars on this problem in Latin American and other contexts emphasizes the party’s historical pattern of recruitment practices (Grzymala-Busse 2002; Levitsky 2003; Greene 2007), its strategy for dealing with the government and other parties (Greene 2007), the frequency and intensity of ideologically motivated schisms (Coppedge 1994), and the importance of patronage as a motive for ongoing activist involvement (Aldrich 1995). To this list might be added the hypothesis that a country’s geographic diversity matters: a country, like Argentina, with politically highly distinctive and somewhat autonomous regions might well produce major parties whose activist body is more ideologically diverse across regions than would be the case in a less geographically differentiated context such as Venezuela. These hypotheses notwithstanding, the ideological composition of the activist corps is usually established by the time of a party-system crisis, and it is hard to quickly and significantly alter it during the crisis. Hence, this factor does not receive further explanatory attention; instead, it is treated as a cause of the ideological diversity of the experienced potential leadership cadre within a party.

170

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

Second, the party organization must be flexible enough to allow party actors currently excluded from leadership to take control of the party apparatus. If, for example, the party is so closely identified with, or so thoroughly controlled by, one leader or faction that replacing the leadership would cause the party to collapse organizationally, then ideological change cannot be accomplished via leadership turnover. Two organizational features are key in accounting for this kind of flexibility: the complexity of the party’s membership and outreach organization, and the extent to which patronage flows through the party machine.3 A complex membership and outreach organization within the party limits flexibility in leadership teams and ideological appeals because such a structure requires a large and committed team of activists in order to function. Those activists, in turn, require consistent motivation to maintain their involvement. Often, parties use ideological appeals to persuade activists to remain involved in politics. When ideology is used to motivate activists and party workers, and when the party apparatus is complex, then parties have a substantial incentive to stick close to their activists’ ideological ideal point. Indeed, the opportunity cost of moving to a substantially different ideological appeal would be the collapse of much or all of the party’s organizational apparatus and the loss of much of the party’s ability to mobilize supporters in elections and at other key political moments (Wolinetz 2002: 161–62). Parties can mitigate these problems by relying more heavily on the distribution of patronage through the channels of party organization (Aldrich 1995: 180–82; Levitsky 2003: 186–216; but see Kitschelt 1994: 222). Activists and party workers who are motivated largely by the benefits of patronage—either the direct material results of personally receiving patronage or the somewhat less direct social and economic gains that accrue to those who distribute patronage to others—are relatively, or perhaps even absolutely, indifferent to changes in the party’s ideological appeal. Thus a party that channels extensive patronage through its own organization faces little risk of losing mobilizing capacity by adopting strategic changes in ideological orientation. In summary, this theory predicts that a party will have meaningful ideological flexibility when it has (a) a pool of activists and party workers who represent a substantial range of ideological positions and (b) either a relatively streamlined set of organizations intended to mobilize members and sympathizers or a portfolio of appeals to activists and party workers that emphasizes patronage

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

171

over ideology and policy. The discussion below will present evidence for this argument in two fundamentally different ways. First, using the comparative method and party averages of survey data from a random sample of local party leaders in Argentina, Peru, and Venezuela, the chapter compares the five collapsed parties under consideration (Acción Democrática and COPEI in Venezuela; Izquierda Unida, APRA, and Acción Popular in Peru) with one significantly damaged but not entirely collapsed party (the Radicals, or UCR, in Argentina) and one party that weathered crises similar to those in Peru and Venezuela without serious electoral or organizational damage (the Partido Justicialista (PJ) or Peronist party, also in Argentina). I compare the party variables central to the theory of ideological adaptability developed above—ideological diversity of activists and party workers, complexity of the party’s mobilizing bureaucracy, and extent of patronage distribution through the party apparatus— but also other dimensions of party organization that scholars have hypothesized to be causes of capacity for ideological change. The most plausible explanations will involve dimensions on which the Peronist party falls to one extreme relative to the collapsed parties. A second and complementary approach to considering the evidence is also adopted: a regression-based approach in which local party units’ actions with respect to local-level nominations serve as dependent variables. This statistical analysis complements the comparative analysis by providing a direct link between party organizational features and ideological adaptability. The key variables highlighted in the theory of capacity for ideological change appear with the hypothesized patterns in both comparative-method analysis and regression models; no other dimensions of party organization are equally successful. Before the chapter proceeds to comparative and regression analysis, however, the next section considers a range of works on party politics that offer theories regarding the causes of the capacity for change in ideological appeals, thereby setting up the key dimensions of party organization that will serve as alternative hypotheses to the theory presented here.

7 . 1 D I M E N S I O N S O F PA RT Y O R G A N I Z AT I O N

In discussing political party strategy and adaptation to the demands of the electorate, the theoretical benchmark remains Downs’s An Economic Theory

172

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

of Democracy (Downs 1957). In the Downsian world of electoral competition, parties are free to adopt any ideological stance that seems electorally beneficial. Indeed, in Downs’s account, parties “develop ideologies as weapons in the struggle for office” (Downs 1957: 96). From this point of view, a party that adopts an electorally unhelpful ideological stance—as did the Peruvian and Venezuelan traditional parties—is behaving in an inexplicable and perhaps even irrational manner. When faced with this problem, one resolution is to turn to party-organizational factors as a way of enriching the Downsian understanding of party competition and hence of making parties’ suboptimal ideological positioning comprehensible (Strom 1990; Harmel and Janda 1994; Kitschelt 1994: 207–53; Aldrich 1995: 163–93; Grzymala-Busse 2002: 123–74; Levitsky 2003: 144–85; Greene 2007: 173–209). Party leaders, after all, operate in multiple arenas: they face legislative decisions, international pressures, and electoral challenges, but they are also situated within an organizational context in which it is necessary to fend off internal challengers, maintain the allegiance of allies and activists, and win support for electoral nominations. By adding organizational context, in particular, to our understanding of party strategy in electoral competition, we can more adequately understand the decision-making constraints that lead to electorally suboptimal ideological positioning—and therefore to party-system collapse. Across this literature, a negative relationship is posited between ideological flexibility and features of what has broadly been described as the mass-party model of organization (Kirchheimer 1966) and organizationally leaner alternatives known, for example, as catch-all (Kirchheimer 1966), electoral-professional (Panebianco 1988), or cartel (Katz and Mair 1995) parties. Within this broad tradition that uses organizational factors to explain parties’ relative capacity to shift their ideological appeals, very different factors are given explanatory emphasis. In reviewing the various hypotheses regarding causes of parties’ ideological flexibility, this section traces common themes that encompass the key ideas in each hypothesis. These common themes are broader than the concepts in the original theories in order to highlight areas of overlap and to identify a toolkit of party organizational variables that are applicable to the study of parties in a diverse range of times and places. The unifying dimensions of party organization which are developed below, and which play a central role throughout the analysis in this chapter are: the size of the party membership,

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

173

the size of the party leadership corps, party leaders’ experience, party leaders’ pragmatism, members’ contribution toward financing the party, the extent of national leaders’ control over local nominations, the overall organizational strength of local units, ties between party activists and civil-society organizations, the extent to which activists are recruited via particularistic benefits, local leaders’ ability to communicate directly with national party leaders, the party’s degree of decision-making autonomy from labor unions, ideological diversity among party activists and workers, the complexity of the party’s membership and outreach organizations, and the extent to which patronage is distributed through the party organizational apparatus. In addition to a theoretical characterization of these aspects of party organization, this section presents descriptive evidence about the organizational traits of the Venezuelan, Peruvian, and Argentine traditional parties on each dimension. The data come from a new survey of party leaders at the local level in Argentina, Peru, and Venezuela. Although this survey was administered in 2004–05, the sample drew from leaders who had been intimately involved in local party organization during the period leading up to party-system collapse in Peru and in Venezuela, and during the same years in Argentina.4 7.1.1 Size of the Party Membership A first and perhaps obvious trait of the prototypically ideologically inflexible mass party is the size of the party membership. A large official membership base can provide a party with some advantages: free campaign work from the most committed members, a fund-raising base, a core of committed electoral supporters, and so forth. However, a large membership might also constrain a party ideologically. With reference to a broader concept of “organizational entrenchment,” whose first defining attribute is the size of the party membership in relation to the electorate, Kitschelt argues that a large membership apparatus and related organizational features “make it difficult for small groups of new entrants and small shifts in the convictions of existing participants to bring about significant changes in the party’s strategic orientation” (Kitschelt 1994: 221). A large membership base may also contribute to ideological inflexibility by convincing party leaders that there is an overly steep trade-off between using new appeals to seek voters in the broader electorate and recycling existing appeals to motivate the membership (Grzymala-Busse 2002: 138–50). However,

1 74

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

it is unclear that all party members are strongly motivated by the fine details of ideology. Some may instead be motivated by patronage, tradition, or social factors, and may be relatively inattentive to ideological matters; such members would provide a useful electoral resource to the party without much loss of ideological flexibility.5 Hence, the ideological constraint sometimes associated with membership size might arguably be more directly connected with other party-organizational factors. Two survey questions directly measure the size of party membership bases at two points in time: significantly before the process of party-system collapse (1980 for Peru, the mid-1980s for Venezuela and Argentina) and later, at an early phase in the collapse (1985 for Peru, early 1990s for Venezuela and Argentina). Both ask local party workers or leaders how many official members the party had locally at the two specified points in time.6 The survey data suggest that the Venezuelan traditional parties had the largest officially registered membership base of any of the parties under consideration. Acción Democrática’s median membership of over 2,000 was larger than that of COPEI (median of 1,000–2,000), which in turn was larger than that of any other party in the study.7 The APRA party in Peru, as well as the Peronist and UCR parties in Argentina, follow in a second tier, with median official memberships in the high hundreds during the period under study. Finally, Acción Popular and Izquierda Unida never had a median local membership base larger than 200, and therefore fall toward the bottom on this dimension. To the extent that the size of a party’s enrolled membership constrains the party ideologically, these results imply that the Venezuelan traditional parties should be the most constrained, followed by APRA and the Argentine parties, with the traditional left and right parties in Peru the least constrained. These implications are considered in greater depth in the comparative analysis below. 7.1.2 Size of the Party Leadership Corps In much the same way that a large membership might constrain a party ideologically by watering down the effects of attitude change or the recruitment of new, ideologically distinctive members, so also a large leadership corps within the party may produce ideological constraint. When a party has a large group of leaders actively involved in its interior politics, it can be “more difficult to pursue coherent electoral strategies, respond to the broader electorate, or appear

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

175

united and effective . . . [I]deologues could crowd out the pragmatists, discourage new ones from joining, and, given power, squelch reform” (Grzymala-Busse 2002: 74; see also Panebianco 1988: 264; Levitsky 2003: 20). On the other hand, a large leadership corps might promote ideological flexibility if the body of leaders is internally differentiated and competitive. In such a party, the leadership group contains significant competing factions, which in turn can enhance ideological flexibility (Harmel and Janda 1994: 266–67). In other words, the ideological diversity of the leadership corps may well be more important than its size for the party’s capacity for ideological change. An open-ended question asking party leaders and workers to name the most important leaders within the party in their area since 1980 measures the size of the party leadership corps.8 For analytic purposes, the answers to this question are simply transformed into a count of the total leadership body in the local party, with any locality reporting more than ten truncated at ten to reduce the influence of outliers.9 As with the membership base, the Venezuelan traditional parties clearly have the largest leadership corps of the parties in this study. Not only do Acción Democrática (mean of 5.9) and COPEI (mean of 5.9) have larger mean numbers of leaders mentioned, but they also have by far the highest percentages of topcoded reports (38 percent and 43 percent, respectively). The three Peruvian parties as well as the UCR in Argentina fall in a middle tier, with four to four and a half leaders mentioned on average in each locality. Finally, the Peronist party falls at the bottom of the scale, with the smallest mean local party leadership corps (3.3) and in a tie with Acción Popular for the fewest top-coded responses (4 percent). 7.1.3 Party Leaders’ Experience Party leaders with greater experience of the give and take of politics may develop skills that facilitate ideological flexibility for the party as a whole. For example, experienced leaders may have developed the capacity to build coalitions around new candidates or ideas, the debate skills necessary to mobilize actors within and outside the party, the ability to read the public’s ideological mood and preferences, and so forth (Grzymala-Busse 2002: 6, 13, 76–77). Hence, parties whose leaders have more experience with democratic political competition may be more ideologically flexible. At the same time, this argument may be a better fit for its original empirical context in the post-communist countries of

176

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

Eastern Europe than in South America, where most traditional party leaders have extensive democratic experience. Nonetheless, it may be worth checking whether there is an empirical correspondence between relatively more experienced party leaders and ideological flexibility among the parties under study. A survey question asks each respondent to list her previous positions within the party.10 The two Venezuelan traditional parties have the most experienced local leaders on average, with a typical leader reporting four or more past positions, consistent with the longer recent history of democracy in Venezuela than in Peru or Argentina. APRA in Peru and the UCR in Argentina also have rather experienced local leaders on average, with a typical leader reporting about three and a half past positions. Acción Popular and the Peronists have somewhat more moderate average levels of experience among their leaders (just under three past positions), while Izquierda Unida has the least experienced leaders on average (two past positions). These patterns are considered in conjunction with those on other dimensions in the comparative analysis below, but it is worth noting that this hypothesis would predict very high levels of ideological flexibility for the Venezuelan traditional parties, and therefore appears initially problematic. 7.1.4 Party Leaders’ Pragmatism To the extent that party leaders are ideologically motivated, they may be unwilling to change their party’s appeals for the sake of winning office or better representing the demands of the population; by contrast, when leaders’ motives are more pragmatic, party flexibility may be enhanced. Greene offers a theoretical justification for these hypotheses, in work on the organizational and strategic features of opposition parties within dominant-party systems. Nonpragmatic party leaders establish tight links to core constituencies and create closed organizational structures that only recruit ideologically pure “good types.” Since their parties are populated by personnel that are committed to deep political change, they are robust against cooptation. . . However, for the same reason, their parties [tend to be] programmatically distant from the center and therefore out of step with what the average voter wants. (Greene 2007: 60–61)

In Greene’s dominant-party context, opposition parties are expected to be both non-pragmatic and ideologically non-centrist for reasons related to the

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

177

existence and strategy of the dominant party. However, the recruitment patterns he describes could be expected to operate in a similar way in any party dominated by non-pragmatic elites; while the ideological position to which the party ties itself when elites are ideological purists rather than pragmatists may be anywhere on the spectrum, such a lock-in effect would nonetheless be expected to reduce the party’s ideological flexibility. On the other hand, there may be alternative routes toward the development of a narrow ideological base of party members, activists, workers, and leaders, other than non-pragmatic elites. For example, a history of ideological schisms—perhaps even motivated by past changes in ideological appeals initiated by pragmatic, vote-seeking party leaders—could also produce a party with a narrow internal ideological distribution. Hence, it is unclear whether the pragmatism of leaders ought to be independently important, or whether it ought instead to be regarded as one of many possible pathways leading to more ideological constraint, involving low ideological variance within the party. To measure leaders’ pragmatism, the survey asks how important ideology was in their initial decision to join the party. Responses are on a four-point scale, with a 1 representing the highest degree of importance attached to ideology and a 4 representing maximal pragmatism.11 The Peronist party has the highest average level of pragmatism of any party in the study (2.1), although the margin of difference among the Peronists, the UCR (2.0), and Acción Democrática (2.0) is slender and statistically insignificant. COPEI (1.8) and Izquierda Unida (1.8) appear to fall slightly lower than the other parties, but the differences are not particularly noteworthy. 7.1.5 Members’ Contribution to Financing the Party In addition to large membership bases, mass parties are characterized by substantial reliance on party members for fund-raising. Thus, [F]rom the financial point of view, the [mass] party is essentially based upon the subscriptions paid by its members. . . In this way, the party gathers the funds required for its work of political education and for its day-to-day activity; in the same way it is enable to finance electioneering. (Duverger 1954: 63; see also Kirchheimer 1966)

Heavy reliance on the members for financing is an intuitive survival strategy if other funding sources are unavailable to the party. Nonetheless, such reliance

178

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

increases the bargaining power of members within the party, potentially limiting ideological flexibility. On the other hand, other sources of nonstate funds are likely to be equally ideologically constraining (Strom 1990: 579–81), so the effects of membership financing on ideological flexibility may well depend on the mix of funding sources that forms the counterfactual. Local party leaders were asked what percentage of the party’s local funding during the period shortly before party-system collapse came from member contributions.12 For Acción Democrática, APRA, the Peronists, and the UCR, members made a small but significant contribution to local party funds (11–25 percent). For COPEI and Izquierda Unida, members were a rather more important source of funding (26–50 percent). Finally, for Acción Popular, members contributed very little to party funds (< 10 percent). 7.1.6 National Leaders’ Control over Local Parties A major hypothesis regarding the causes of parties’ ideological adaptability involves the degree of procedural rigidity of intra-party decision-making processes. Parties with highly rigid decision-making procedures may have a harder time adopting new ideological appeals because those procedures give intra-party opponents of change a great deal of leverage. Furthermore, cognitive and organizational effects may be involved: parties with rigid decisionmaking procedures are vulnerable to external shocks, for [their decision rules] limit the speed with—and often the extent to—which they can adapt. Decision makers in a highly routinized organization will—at least initially—tend to consider a narrower range of strategic alternatives, none of which may be appropriate in a context of crisis or rapid environmental change. (Levitsky 2003: 19; see also Panebianco 1988: 265; Kitschelt 1994: 223–24)

Measuring and comparing the full extent of the procedural rigidity of party decision-making is challenging; the best approach is probably to synthetically consider many or most of the dimensions of organization considered in this chapter. A more focused measurement for a single party would require analysis of both the formal rules and the de facto process involved in a broad range of important decisions. If the set of important decisions varies across parties, then the resulting measurements would be significantly noncomparable across

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

179

cases, as any observed differences may be due either to important organizational variations or to differences in parties’ issue agendas during the window of time used in the study. A partial but potentially interesting approach, which captures an important aspect of each party’s decision-making while also focusing on similar—and similarly important—issues for each party, is to compare the decision-making procedures used in a specific, crucial decision made by all parties during the period. The decision used for these purposes in the present analysis is the selection of candidates for local office. If national party leaders are able to directly select their preferred candidates at the local level, this clearly reflects a less structured or routinized decision-making process than would be the case if national leaders have to work through party conventions, primary elections, or local party leadership councils to influence the final outcome. Hence, an examination of the process used to select local candidates can distinguish between lower and higher levels of procedural rigidity. Furthermore, this issue captures an aspect of party decision-making with clear connections to ideological flexibility: local candidates help shape the party’s overall ideological appeal, and so the candidate selection process at the local level is one of the best opportunities to signal a shift in ideology. All of the parties under consideration had to select local candidates during at least two electoral cycles just before or during party-system collapse, so this decision-making task meets criteria of universality (for the parties of interest) as well as relevance. The analysis here focuses on the proportion of local candidates who were reported, by local party leaders, to have been directly selected by national leaders during the first local elections after 1980 in each country.13 The Venezuelan parties define the higher end of this spectrum (33 percent for AD and 39 percent for COPEI), while the Peronists in Argentina fall at the lower end (18 percent). The other four parties are somewhere in between. These results suggest that the Peronists have the most rigid decision-making procedures of the parties under study, but there is an important caveat: Argentina has had an institutional (although not universally adhered-to) norm of primary elections for contested nominations throughout its current democratic period. Hence, the apparent relative routinization of Peronist party practice in this regard is an organizational feature of the party imposed by the broader political system.

180

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

To take a less institutionally driven angle on the same decision, another survey question asks local party leaders how much influence national party leaders had on the nomination decision. To at least some extent, this approach to the question sidesteps the institutional requirement for primary elections when two or more potential candidates emerge: national party leaders can often have a great deal of influence over primary elections, and also over whether multiple candidates emerge, especially in less routinized organizations. Discussion will focus on the rates at which national party leaders imposed their preferred candidates over local opposition in each party, as well as the frequency with which national leaders’ preferred candidates were selected (by imposition over opposition or not).14 On these two measures, the Peronist party is no longer the most routinized in the study: that distinction falls, by a meaningful margin, to the Argentine UCR (1 percent national imposition; 22 percent nationally preferred candidate selected). With respect to national leaders’ imposition of candidates, the Peronist party (with an imposition rate of 11 percent) falls somewhere in the middle of the pack, and regarding the overall success rate of national leaders’ preferred candidates (31 percent), the party falls second-to-last. The least routinized party by these measures is Venezuela’s COPEI (15 percent imposition and 61 percent overall success for national leaders), while the other parties show a moderate level of institutional resistance to national leaders’ desires. In addition, it may be worthwhile to explore the degree of rigidity of parties’ decision-making processes by considering the extent of local party leaders’ influence on national decisions. Evidently, the greater the inclusion of local leaders in national decisions, the higher the degree of rigidity or routinization of the process (Strom 1990: 577). After all, more people are necessarily involved, and a more complex process of information-gathering and decisionmaking is required. The survey includes a question asking local leaders how much influence they had in national decisions, on a 1 to 4 scale with 1 being the least influence.15 The differences across parties are somewhat modest; however, they do point in the expected directions. The Peronist party (1.9) is the least routinized according to this indicator, while the Venezuelan parties (2.3 for AD and 2.4 for COPEI) and Izquierda Unida in Peru (2.5) are the most routinized. Thus, while the degree of differentiation among parties may fail to reflect existing qualita-

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

181

tive knowledge, this indicator appears to provide some substantive information about the underlying dimension of interest. As a final indicator of the degree to which national party leaders exercise direct control over local leaders, or on the other hand are constrained by routinized patterns of interaction with those leaders via party bureaucracy, we consider a question that asks local party leaders to name their most important means of communication with the national-level party. The indicator is a four-point scale, with 1 representing a situation in which local leaders feel so cut off by formal layers of bureaucracy that they cannot effectively communicate with the national party and 4 representing a context in which local leaders can directly contact top-level national leaders when necessary.16 Here the most meaningful contrast is between the notoriously highly routinized parties in Venezuela, in which communication is on average mediated through the party bureaucracy (mean score of 2.6 for AD and 3.0 for COPEI), and all the other parties, which show more flexibility in communication (mean scores in the 3.2–3.4 range). The results on these indicators taken as a whole do not fully capture the informality and low routinization of Peronist party politics at the elite, national level (see, in particular, Levitsky 2003); furthermore, several aspects of that pattern of low routinization are evident in other dimensions of organization considered here. Especially relevant are the discussions of the complexity of the formal membership and outreach organization; the size of the leadership corps; the power of local units in national decisions; ties to formal civil-society organizations; and the apparent importance of patronage in activist recruitment and local party activities. Thus, this section should not be taken as suggesting that the Peronist party is highly routinized in general. Instead, the relevant findings are that, with respect to local candidate selection, electoral rules impose a significant degree of routinization on what is otherwise a highly flexible and informal organization, while the relative lack of local organizational influence on national decisions may reflect a broader pattern of low routinization in the party. The variable regarding national leaders’ involvement in local candidate selection, though ultimately not important for explaining parties’ ideological flexibility, serves as an important control variable in the regression analysis below; the communication variable plays a similar role.

182

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

7.1.7 Ties between Party Activists and Civil-Society Organizations Parties differ, sometimes substantially, in the extent to which they foster linkages and relationships with organizations in civil society. Some parties, such as Acción Democrática and COPEI during the peak of the Venezuelan traditional party system, treat civil society as a collection of “battlefields for party struggles for control” (Coppedge 1994: 27). They encourage local party leaders and activists to attempt to, in effect, colonize independent organizations. Captured organizations then become electoral resources and sources of patronage for the victorious party. Other parties, including the Spanish Socialists, develop only weak linkages with civil society (Burgess 2004: 163), or fall somewhere between these extremes. These ties between parties and civil-society organizations may also influence parties’ degree of ideological flexibility. Specifically, civil-society alliances often function analogously to other aspects of mass party organization. They offer an electoral advantage to the party through mobilization, outreach, and communication. Yet because the alliance between the party and the civil-society organization is optional and often predicated on shared ideological agendas, the electoral advantage offered by the alliance is similarly ideologically conditional. If the party suddenly alters its ideological profile, it may well lose its civil-society allies and the advantages they bring. Hence, strong civil-society ties may reduce parties’ ideological flexibility, especially when the alliances in question are based more on ideological than on social or patronage grounds. Party connections with civil-society associations are operationalized by asking party leaders about their personal involvement in different categories of associations. For each type of association, party leaders are asked whether they were a member during the past year, whether they contributed money to that association in the past year, and whether they attended at least one meeting of the association during the past year. These questions are asked about churches and religious associations; sports and recreation associations; neighborhood associations; art, music, or culture associations; associations connected with education; unions; other kinds of professional associations; environmental associations; and charitable associations. Differences across parties are largest if the associations are aggregated into an index measuring the total number of associational involvements for party leaders, and

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

183

such an aggregation also captures the overall scope of connection between parties and civil society.17 Party differences on this measure are reflected in the average number of kinds of personal civil-society involvement reported by local party leaders. The highest average belongs to Izquierda Unida in Peru (3.7); however, the most important pattern is probably the lower number of civil-society involvements of Argentine party leaders (2.8 for both parties) as opposed to others (ranging from 3.2 to 3.5). This pooled difference is highly statistically significant (the difference-in-proportions t statistic is 3.25) and suggests that the Argentine parties should be less constrained by civil-society associational involvements, to the extent that such involvements affect ideological flexibility. 7.1.8 Recruitment via Particularistic Benefits Parties use a variety of means to recruit new activists and potential party leaders, and the portfolio of recruitment tactics may have implications for ideological flexibility. When parties recruit via particularistic benefits, the resulting activists and party leaders are motivated in large part by nonideological goods and therefore may be relatively insensitive to changes in the party’s ideological profile. By contrast, when parties recruit more centrally on ideological grounds, the resulting activists and party workers may conform more closely to Wilson’s (1962) “amateur” or Wildavsky’s (1965) “purist” activist or politician types. Such activists are motivated by ideology, and specifically by the quest to have the party express their personal ideological views as explicitly as possible; if the party shifts ideologically away from them, activists may withdraw their services and cost the party in organizational terms. Hence, the party becomes ideologically constrained by the opportunity costs, as well as organizational effects, of potentially losing the services of committed activists (Aldrich 1995: 180–92; Greene 2007: 119–209). Although some discussions of the ideological orientation and intensity of activists and local party leaders focus on recruitment practices and incentives, as in this section (e.g., Kitschelt 1994: 209–12; Strom 1990: 577–78), it is of course plausible that the motivations of intra-party actors are dynamic and respond to the mix of benefits that they receive from the party over time. Hence, while motives at initial recruitment may matter to some degree, the ongoing flow of particularistic benefits to activists and local party leaders may be more impor-

184

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

tant in determining those individuals’ ongoing motives for participation, and therefore for affecting the degree of ideological constraint of the party. One measure of interest here is the extent to which ideology was important in activists’ and local party leaders’ initial recruitment; this indicator is discussed above in the section on leaders’ pragmatism. This section instead focuses on party means regarding three other potential motives for initial involvement: meeting influential people, enhancing one’s social network, and earning money, each measured on a four-point scale with 1 representing minimum and 4 representing maximum importance.18 These results may well be affected by social desirability bias. In particular, it may be regarded as undesirable to participate in party politics for financial reasons, and so economic motives for initial involvement may be underreported. In any case, the party leaders in this study show no systematic differences on this variable, with party means all in the 1.3–1.5 range. For the two sets of motives related to connections with influential leaders, the major finding is that the Argentine parties fall toward the lower end of the scale, a result that is an awkward fit with the hypothesis that such motives would enhance parties’ ideological flexibility. Specifically, for the motive of meeting influential people, the Peronist party has a mean of 2.1 and the UCR has a mean of 2.0; other parties range from 2.4 to 2.7. With respect to the motive of enhancing the activist’s social network, the Peronists’ average is 2.6 and the UCR’s is a quite low 2.2. The other parties range from 2.7 to 3.2. 7.1.9 Decision-Making Autonomy from Labor Unions A long tradition of analysis in Latin American politics emphasizes ties between labor unions and political parties as a defining characteristic of Latin American party politics (Malloy 1977; Collier and Collier 1991; Wiarda 2004). In this perspective, Latin American parties are distinctive in the intensity of the connections that they have formed to sometimes state-approved unions. Those connections are based on the policy platforms of the pre-neoliberal, importsubstituting era from the Great Depression until the early 1980s. During the neo-liberal era, however, many unions have continued to pressure their allied political parties to support the previous policy package, thus becoming an obstacle to strategic and ideological flexibility in the new international political and economic environment (Levitsky and Way 1998; Murillo 2001: 2, 66–72,

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

185

149–52; Levitsky 2003: 107–42; Burgess 2004: 100–17, 128–38). Linkages with unions during this period may also constrain ideological flexibility for parties in other regions (Kitschelt 1994: 225). Parties’ organizational ties to unions are measured with a survey question asking local party leaders how much influence unions had in party decisions during the period just before party-system collapse, producing a four-point scale with 1 representing minimum and 4 representing maximum influence.19 The Venezuelan parties score the highest, with the mean for Acción Democrática (3.7) approaching the upper bound on the scale and COPEI (2.8) scoring higher than any other party in spite of being notionally a center-right rather than a labor-based party. APRA (2.4) and Izquierda Unida (2.6) score near the middle of the scale, while Acción Popular (2.2), the Peronists (2.2), and the UCR (2.0) fall somewhat lower. It is particularly noteworthy that the traditionally laborbased Peronist party scored with the non-labor-based parties on this scale by the early Menem period. 7.1.10 Ideological Diversity among Party Activists and Workers Variation in the ideological composition of intra-party actors such as activists and local party leaders is useful in differentiating between more and less electorally successful Communist successor parties in East Central Europe (Grzymala-Busse 2002), and it plays an important role in understanding the adaptation of Social-Democratic parties to the left-libertarian challenge in Western Europe (Kitschelt 1994: 222–23). To what extent does the distribution of ideologies, in particular within a party’s leadership, help explain that party’s ideological flexibility? Parties that are exceptionally homogeneous may be expected to face severe internal resistance to strategic or ideological change. Not only is there a solid constituency for the existing strategy, but there is no significant disgruntled and therefore easily mobilizable intra-party contingent in favor of any other strategic stance. Party leaders who would change direction may find themselves with no resources in intra-party decision-making struggles other than their own personal charisma and following. If these resources prove insufficient to defeat the rest of the party machine—in primary elections or in more traditional partymachine power plays—the reform-oriented leader will have to choose between accepting defeat or leaving the party and launching a candidacy from outside

186

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

the party system. In either case, the traditional party remains strategically immobile. Parties with ideologically and socially homogeneous activist bases and leaders may be especially vulnerable to the ideological mis-positioning that is a crucial ingredient of party-system collapse. Furthermore, unlike many of the factors reviewed up to this point, the ideological composition of the party’s activist base may be a durable constraint even during periods of crisis, because changing this aspect of party organization would require the marginalization or voluntary withdrawal of what may be the majority of intra-party actors and their immediate replacement. This is a far more difficult reorganization to pull off in the short term than, say, changing decision-making rules or altering the terms of a union alliance. The distribution of ideologies within each relevant party is measured by asking party leaders about their personal ideological positions. The survey instrument contains a 1-to-10 scale of ideology, anchored with reference to well-known Latin American political leaders.20 A second question asks each leader to use the same scale in placing that leader’s national party’s ideology as of 1990. Table 7.1 shows the average absolute difference between local leaders’ self-reported ideology and the ideology of their national party organization. The major result here is that the Peronist party is substantially, and statistically significantly, more ideologically diverse in its corps of intra-party actors than are any of the other parties under study. Indeed, the Peronist party contains meaningful numbers of intra-party actors from the center-right through to the moderate left. The closest competitors on this dimension (Acción Democrática, APRA, and the UCR) have roughly half the average intra-party ideological divergence from the national party as the Peronists; the other parties fall somewhat lower. This large, significant, and probably durable organizational difference between the Peronists and the other parties may well account for much of the Peronist party’s greater observed ideological flexibility. ta b l e 7 . 1 . Average ideological distance between local leaders and the national party Dimension

AD

COPEI

APRA

AP

IU

PJ

UCR

Mean absolute ideological distance

1.2 (0.2)

0.7 (0.1)

1.2 (0.2)

0.9 (0.2)

0.8 (0.2)

2.1 (0.2)

1.1 (0.1)

Standard errors in parentheses.

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

187

7.1.11 Complexity of the Party’s Membership and Outreach Organizations The traditional mass party is characterized in part by its large, complex, and differentiated collection of membership and outreach organizations. In addition to presenting candidates for elections and perhaps coordinating corresponding campaigns, a prototypical mass party may enroll female supporters in a women’s auxiliary; provide party-branded sports, recreation, and educational associations for children and young adults; incorporate men into union-based or other occupation-related party units; organize thematic and identity-based party organizations; and maintain an active collection of local party organizations between and during major electoral seasons. Such complex membership and outreach organizations can build a party’s core “electorate of belonging” (Panebianco 1988: 267)—i.e., the group of voters who regard their fates as fundamentally and inextricably tied to that of the party. At the same time, these institutional forms have been hypothesized to reduce a party’s ideological flexibility. Strom’s (1990: 575) theory of party behavior allows for this flexibility-­limiting effect in discussing the choice between “capital-intensive” and “labor-intensive” organizational forms. The membership and outreach organizations are examples of labor-intensive party organization in that each auxiliary and party unit requires effort from a staff and members to plan the activities and programs that sustain the organization. However, large amounts of money are not essential, especially in comparison with capital-intensive activities such as media-based campaigns, polling and focus groups, hiring policy and PR experts, and so forth. Labor-intensive organizational forms tend, in Strom’s account, to constrain parties ideologically: “When leaders sacrifice organizational power or policy influence, myopic and electorally inefficient party behavior follows. Leadership concessions are particularly likely in labor-intensive party organizations” (Strom 1990: 578). In particular, labor-intensive parties are likely to replicate the portfolio of messages and ideological appeals that attracted their current activist and partyworker corps—because such appeals are among the cheapest and easiest means of retaining activists and workers. By contrast, parties with less of an investment in complex membership and outreach organizations may have a smaller activist corps and electorate of belonging, and as a result have less to lose when moving

188

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

to a different ideological profile. For similar discussions, see Panebianco (1988: 262–67); Kitschelt (1994: 216, 222); and Deschouwer (1994: 83). In response to these arguments, Levitsky (2003: 14–15) offers the important caution that parties which substitute informal, weakly institutionalized, and spontaneously or locally organized membership and outreach organizations may be able to maintain an electorate of belonging while also retaining ideological flexibility. After all, organizational units that are not formally part of the party may have a harder time gaining the kinds of leadership concessions that Strom’s theory addresses. As a result, they may affect members’ identity and electoral decision-making without substantially constraining party leaders’ ideological options. In light of the manifest importance of informal institutions in Argentine party life, and to at least some extent in many other Latin American parties, it is important to focus substantially on the complexity of relatively formal membership and outreach organizations, i.e., those that are directly and intentionally tied to the official party organization. Three pairs of questions from the party-leader survey are used to measure the extent to which, during the period of party-system collapse (the 1980s for Peru, the 1990s for Venezuela and Argentina), the various parties maintained complex formal membership and outreach organizations. The questions ask about the existence of direct party involvement with local youth organizations, local women’s organizations, and local workers’ organizations, respectively. Whenever the party was locally involved with a particular category of membership or outreach organization, the follow-up question asks party leaders to indicate the number of people active in that organization.21 The median size of each kind of organization, for each relevant party, is the focus of discussion below, with a 0 used to represent a locality where a party has no involvement with a given organizational category. The two Venezuelan traditional parties stand out for the complexity of their membership and outreach organizations. For each of the three organizational categories, the median answer of local leaders in these two parties is always higher than the median for any other party. Indeed, more than three quarters of Venezuelan local traditional party leaders report the existence of each possible category of organizational involvement, a level that no other party approaches. Furthermore, most organizations are reported as having at least 200 members.

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

189

While AD and COPEI in Venezuela define the upper end of the spectrum for these parties on this dimension, Acción Popular in Peru defines the lower empirical limit; its median local party unit is involved with a small youth organization with fewer than 100 members but has no connections with any women’s or workers’ organization. Izquierda Unida, the Peronists, and the UCR all fall close to Acción Popular. APRA in Peru arguably falls somewhere in the middle of these two clusters. While its median locality has only a small women’s or workers’ organization, it has a reasonably large youth organization with 100–200 members. Large women’s and workers’ organizations are more common in APRA than in the parties toward the lower end of the continuum. At least one quarter of APRA local leaders report a women’s and/or worker’s organization with 200 or more members; this figure is meaningfully lower for the other non-Venezuelan parties. Another survey measure provides a broader summary indicator of the complexity of parties’ membership and outreach organizations, asking local leaders to evaluate the overall strength of the local organization on a 1-to-10 scale.22 This indicator broadly replicates the findings discussed above. The major differences are that Acción Popular in Peru (7.4) and UCR in Argentina (7.4) appear to have a somewhat higher organizational score on this measure than on the measures above, while APRA (6.8) has a somewhat lower score; these may be sampling fluctuations given the more subjective nature of this second question and the higher within-party variances on it than on the measures above. Thus, to summarize, the two Venezuelan parties clearly have more complex formal membership and outreach organizations than the Peruvian or Argentine traditional parties—and might therefore be more ideologically constrained. APRA has a medium level of organizational complexity and thus would be expected to be moderately ideologically constrained if this hypothesis is correct. The other four parties would, on the basis of this dimension, be expected to show relatively substantial ideological flexibility. 7.1.12 Distribution of Patronage through the Party Organizational Apparatus As discussed earlier in the chapter, the distribution of patronage through the party apparatus may enhance parties’ ideological flexibility by reducing the sensitivity of party activists, workers, and local leaders to ideological change.

190

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

Extensive access to clientelistic resources may also enhance a party’s ideological flexibility by allowing it to partially compensate traditional allies and electorates who are hurt by a new policy or ideological appeal (Burgess 2004: 74–88), although patronage has also been hypothesized to reduce party flexibility by insulating members from societal demands (Kitschelt 1994: 222). This last hypothesis may well hold in times of relative party-system stability; however, when a full-scale party-system crisis is under way, there is every reason to believe that even recipients of extensive party patronage will become attentive to voter demands: after all, failing to meet those demands has the potential to destroy the party and halt the flow of patronage goods. The prevalence of party-channeled clientelism is measured, in this study, using a battery of survey questions about specific kinds of activities on offer at local party headquarters: distributing goods and social services to party members, distributing goods and social services to nonmembers in the community, and helping register people with government assistance programs.23 For the questions about distribution of goods and services, the indicator is a three-point scale, with 3 representing the highest degree of activity and 1 the lowest. For the question about government assistance programs, the indicator is dichotomous. On each of the three indicators, the Peronist party scores highest (2.1 on patronage to party members, 2.1 on patronage to outsiders, 71 percent on connecting citizens with government programs), although some of the differences regarding connecting citizens with government programs are small. In general, these findings confirm the many qualitative accounts of patronage within the party, such as Auyero’s discussion of Peronist “Problem-Solving Networks” (Auyero 2000: 80–118), and are also consistent with other forms of survey data about the extent of Peronist clientelism (Brusco, Nazareno, and Stokes 2004: 68–72; Stokes 2005). The relatively lower levels of patronage distribution reported in many other parties do not necessarily imply that these parties refrain from such practices; if distribution of patronage goods is carried out through government rather than party offices, for example, such practices would not appear in these data. However, patronage distribution through the party apparatus itself may be crucial for shifting activists’ and local leaders’ motives for involvement away from ideology, and therefore for enhancing parties’ ideological flexibility.

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

191

7 . 2 A C O M P A R AT I V E A N A LY S I S O F PA RT Y O R G A N I Z AT I O N A N D S U RV I VA L

In light of the wide range of potentially important dimensions along which the parties in this study differ, we require some means of focusing on dimensions that may have contributed to the relative lack of ideological flexibility of the Venezuelan parties, in particular, and also to some extent the Peruvian parties, in comparison with the greater evident flexibility of the Argentine parties, especially the Peronists. Of course, no assumption can be made that all organizational differences between the Peronists and the other parties are causally relevant. One approach would be to simply rule out all dimensions on which the Peronist party does not fall to one extreme or the other; in a univariate world, such dimensions could not possibly account for the Peronist party’s greater ideological flexibility and resistance to corruption scandals.24 But in a multivariate world in which different variables may substitute for each other in producing ideological rigidity, this simple approach could eliminate dimensions that are genuinely important for some, but not all, of the collapsed parties. Nonetheless, as an initial mode of synthesizing the complex patterns described above, this approach of differentiating dimensions where the Peronists fall at the end of the spectrum associated with maximum ideological flexibility from others has the virtue of simplicity. Table 7.2 presents a schematic summary of the data about party organizational traits from the earlier section of the chapter, providing the information necessary for an illustrative comparative analysis using Mill’s method of difference as elaborated in the more recent literature on the comparative method. The dependent variable of interest here involves parties’ demonstrated level of ideological flexibility. On that variable, of course, the Peronists rank highest, while the Venezuelan parties and Izquierda Unida from Peru rank lowest and the remaining parties fall somewhere in the middle. Treating this spectrum dichotomously, the Peronists may be regarded as a successful case and all other parties as unsuccessful. Thus, from a method-of-difference perspective, any dimension on which the Peronists score in the middle can be regarded with suspicion as a potential cause of ideological flexibility. This decision rule raises concerns about the importance of the size of the party’s official membership; the importance of leaders’ pragmatism; the relevance of membership financing; and the causal role of communication patterns. Comparing across the parties in this study, these four variables do not fit the evidence particularly well.

192

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

ta b l e 7 . 2 . Summary of parties’ organizational traits Dimension

AD

COPEI

APRA

AP

IU

PJ

UCR

Membership size

Large

Large

Medium

Small

Small

Medium

Medium

Leadership size

Large

Large

Medium

Medium

Medium

Small

Medium

Leadership experience

High

High

Medium

Low

Low

Low

Medium

Leadership pragmatism

Medium

Medium

Medium

Medium

Medium

Medium

Medium

Membership financing

Medium

High

Medium

Low

High

Medium

Medium

Leadership power in local nominations

High

High

High

Medium

Medium

Low

Low

Power of local units in national decisions

High

High

Medium

Medium

High

Low

Medium

Communication flows

Low

Medium

High

Medium

Medium

Medium

Medium

Civil-society ties

High

Medium

Medium

Medium

High

Low

Low

Particularistic benefits in activist recruitment

High

High

Medium

Medium

Medium

Low

Low

Autonomy from unions

Low

Medium

Medium

High

Medium

High

High

Intra-party ideological diversity

Medium

Low

Medium

Low

Low

High

Medium

Complexity of membership and outreach organizations

High

High

Medium

Medium

Low

Low

Medium

Importance of patronage

Medium

Medium

Medium

Low

Low

High

Medium

As a second analytical iteration, it is worth pointing out that, for some of the organizational dimensions, the Peronists fall at the wrong end of the spectrum—i.e., they fall toward the extreme that is theoretically associated with diminished, rather than increased, ideological flexibility. Such dimensions may also be viewed with some doubt about their causal relevance. The suspect aspects of party organization, following this line of argumentation,

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

193

are leadership experience, national leaders’ power over local nominations, and the prominence of particularistic benefits in the initial activist recruitment process. Seven aspects of party organization survive this initial comparative-method analysis: the size of the party’s leadership corps, the influence of local party leaders in national party decisions, ties between local activists and civil-society organizations, decision-making autonomy from labor unions, intra-party ideological diversity, the complexity of the official membership and outreach organization, and the importance of patronage as a component of the party’s local activities. At the aggregate party level of analysis, these are the most promising candidate explanations for ideological flexibility. They are the ones with patterns of scores that could help explain the observed pattern of party ideological flexibility, if the causal effect of the dimension of organization on ideological flexibility is monotonic. Having the minimally correct descriptive pattern in this sense is an important ingredient for a causal explanation, but it is not in itself sufficient. Also needed is evidence that the variable in question has the right kind of causal relationship with the outcome being explained; that is, there should be evidence that, had the score on the dimension of party organization been other than what it was, the party’s level of ideological flexibility would as a consequence have been other than what it was.25 As a very rough approximation to this kind of evidence, this chapter employs comparisons among local party units, which are presumably more directly comparable than parties taken as a whole. If these units were to be regarded as homogeneous (Holland 1986: 948), in the sense that they were identical in every causally relevant respect except for the dimensions of party organization under analysis, then differences in ideological flexibility across local units that varied organizationally would replicate the counterfactual differences necessary for causal explanation. Unfortunately, such an assumption would clearly be unreasonable; local party units operate in somewhat distinct environments, face at least moderately different policy challenges, and sometimes interact with different sets of partisan competitors. Hence, the analysis here can at best provide clues about causal patterns; in conjunction with the theory and cross-case comparisons presented above, these clues point toward intra-party ideological diversity, complexity of the membership and outreach organization, and im-

194

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

portance of patronage as the most plausible explanations of the observed differences among parties in ideological flexibility. However, the resulting inferences are more tentative, for example, than the inferences based on experimental data presented earlier in this volume. Two dependent variables will be used in the local-level analysis, paralleling the two prerequisites for ideological flexibility theorized in the introduction to this chapter: the existence of an ideologically diverse pool of experienced potential candidates, and the organizational flexibility to nominate different candidates over time. The ideological diversity of the pool of experienced potential candidates is analyzed directly, using the mean absolute ideological distance between the party’s most prominent nominee in the first local elections during the 1980s and the local informant’s rating of the national party’s ideology.26 The Peronists have the highest score (1.5), with all of the differences between the Peronists and parties other than APRA (1.2) being significant at the 0.05 level. In considering the flexibility necessary to nominate ideologically different candidates over time, the analysis focuses on the degree of ideological change between the first and second candidates offered by the party in local elections. Here, the local party unit’s record of ideological flexibility is treated as a source of insight into the national party’s flexibility. The results on this indicator suggest the existence of two clusters of parties.27 The two Venezuelan traditional parties (0.4 for AD and 0.2 for COPEI), as well as the Radicals in Argentina (0.4), get low scores. The other parties, including the Peronists (0.7) as well as the Peruvian traditional parties (0.7 for APRA, 0.6 for Acción Popular, and 0.8 for Izquierda Unida), score higher. Taken in conjunction, these indicators suggest that the Peronist party was he most ideologically flexible of the parties in this study in part because it was more flexible over time than some of the other parties, but also because it had the most internally diverse pool of potential candidates. The independent variables in the analysis are drawn from the survey questions presented above; one variable per theorized dimension of party organization is included. For several dimensions (membership size, leadership size, leadership experience, leadership pragmatism, membership financing, civil-society ties, autonomy from unions, and intra-party ideological diversity), only one

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

195

indicator was discussed above; that indicator is used in the models below. For the other dimensions, some discussion is needed. In the discussion of national leaders’ control over local parties above, several different variables were considered. Of the three indicators connected with local nomination processes, the models below retain the variable measuring whether national leaders unilaterally selected local candidates; replacing this variable with either of the other nomination variables has no substantive effect on the analysis below. The variables regarding local organizations’ power in national party decision-making and communication flows are also used; the correlations among these variables are somewhat modest, and diagnostics show little evidence of multicollinearity, so using multiple indicators from the same conceptual grouping appears statistically harmless in this instance. Turning to the relevance of particularistic benefits in the initial recruitment of activists and local party leaders, the independent variable used in the analysis below is the average of the three indicators reported in the relevant section earlier in the chapter. Likewise, the independent variable used for measuring the complexity of the membership and outreach organization is the average membership size across the three categories of party auxiliary organizations discussed above; the resulting variable was highly intercorrelated with the overall indicator of membership and outreach organization effectiveness. Likewise, the patronage indicator is an average of the variables discussed in the relevant section above. The key potential confounding variables in this analysis are the political traits of the localities in question. Parties may well adapt their organizations to local needs, and may also adjust candidate selection strategies in response to the same considerations. To address this potential problem, two strategies are adopted. First, the regressions below control for the distance between the national party’s ideological stance and the average ideological views of local voters, both as reported by local informants. This captures perhaps the most directly electorally relevant subset of ways that localities are distinctive: their differences in ideological positioning and preferences.28 Table 7.3 reports the results of the first of these models, the regression of distance between the national party and the local nominee on party organizational traits and local voters’ distance from the national party, without locality and party dummies. Consistent with the theoretical expectation developed at

196

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

ta b l e 7 . 3 . Regression of local/national ideological distance on party

organizational traits Variable Intercept Local voters’ ideological distance from the national party Membership size

Estimate (SE)

P Value

− 0.27 (0.42)

0.53

0.17 (0.05)

0.00

0.04 (0.03)

0.16

− 0.02 (0.03)

0.60

Leadership experience

0.00 (0.04)

0.96

Leadership pragmatism

0.05 (0.09)

0.59

Leadership size

Membership financing Leadership power in local nominations Power of local units in national decisions Communication flows Civil-society ties Particularistic benefits in activist recruitment Autonomy from unions Intra-party ideological diversity Complexity of membership and outreach organizations Importance of patronage R

2

N

0.03 (0.03)

0.38

− 0.11 (0.16)

0.47

0.13 (0.06)

0.04

− 0.03 (0.07)

0.65

0.03 (0.04)

0.49

0.06 (0.12)

0.63

− 0.04 (0.06)

0.50

0.51 (0.05)

< 0.01

− 0.06 (0.05)

0.22

0.00 (0.04)

0.94

0.51 193

the beginning of the chapter, the most powerful party-organizational predictor of a local candidate’s ideological distance from the national party is the distance of local informants from the national party. This is not a tautological finding. The informants whose personal ideological views are critical to the measure of intra-party ideological diversity were generally not candidates for local office. Furthermore, they usually joined the party significantly before the first local election, and indeed often held local positions within the party before the candidate was selected. Hence, this is at least evidence of a correlation between the size of a potential candidate’s ideological coalition within the local party apparatus and her probability of being selected as a candidate. At most, it may support the inference that intra-party ideological diversity in the lower and middle levels of the organization causally contributes to the ideological diversity of the experienced potential candidate

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

197

pool. Hence, those aspects of the genesis and history of a party that affect the party’s local ideological diversity (perhaps including factors such as federalism, early coalition structures, institutionalization of intra-party factions, the party’s history of ideologically motivated schisms, ideological restructuring concomitant to processes of democratic transition, and recruitment and indoctrination processes) may play a causal role in determining that party’s present and future degree of ideological flexibility. We turn next to the question of whether the organization has the flexibility to present candidates of different ideological orientations over time. Table 7.4 reports the results of regressing the absolute magnitude of change in local candidate ideology on the party organizational variables, as well as the distance between local voters and the national party. Two variables are significant at the 0.05 level in this analysis: the complexity of the membership and outreach orgata b l e 7 . 4 . Regression of local ideological flexibility on party organizational traits Variable

Estimate (SE)

P Value

Intercept

0.73 (0.41)

0.07

− 0.05 (0.04)

0.23

Membership size

0.02 (0.03)

0.57

Leadership size

0.01 (0.03)

0.68

Leadership experience

0.00 (0.03)

0.96

Leadership pragmatism

− 0.04 (0.08)

0.64

Membership financing

− 0.03 (0.03)

0.42

Local voters’ ideological distance from the national party

Leadership power in local nominations

0.18 (0.14)

0.23

Power of local units in national decisions

− 0.05 (0.06)

0.38

Communication flows

− 0.12 (0.06)

0.06

0.03 (0.03)

0.43

− 0.01 (0.03)

0.81

0.01 (0.05)

0.90

Civil-society ties Particularistic benefits in activist recruitment Autonomy from unions Intra-party ideological diversity Complexity of membership and outreach organizations Importance of patronage R

2

N

0.09 (0.05)

0.08

− 0.10 (0.04)

0.02

0.07 (0.03)

0.05

0.12 193

198

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

nization, and the importance of patronage in the party’s local activities. Specifically, a complex local organization is associated with low ideological flexibility to such an extent that the difference between the lowest observed value on this variable and the highest is correlated with a difference in flexibility that is as large as the observed difference between the most and least flexible parties in the study. Patronage, by contrast, is associated with higher levels of ideological flexibility; the coefficient is statistically significant and the substantive difference between the highest and lowest observed levels on the variable is nearly as large as that for organizational complexity. These results strongly support the argument that complex membership and outreach organizations create a substantial organizational and electoral opportunity cost to ideological change. This is so because the volunteers and low-level employees needed to operate such organizations may temporarily or permanently withdraw their services if the ideological appeals they favor are not represented by the party’s current candidates. Thus, parties with a significant investment in such organizational structures have a lot to lose if they shift ideological appeals; they can therefore be expected to shift less readily than an organizationally slenderer party. This line of argument fits the findings above; support is also found for the argument that extensive channeling of patronage through the party apparatus softens the opportunity costs of ideological shifts by providing local activists and workers with an alternative motive for continued involvement.

7.3 CONCLUSIONS

Previous chapters have shown that voters abandoned the Peruvian and Venezuelan traditional party systems essentially because of anger—rooted in the combination of a sense of ideological underrepresentation and perceptions of corruption, and facilitated by a decline in identification with the traditional parties. This chapter has shown that the parties failed to maneuver ideologically in ways that could have forestalled voter exodus for reasons rooted in party organization. Building from the hypothesis (supported with cross-national comparative evidence) that ideological change is facilitated by a diverse pool of experienced potential candidates within the party and by the organizational flexibility to select different kinds of candidates over time, the analysis has shown that three

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

199

factors contribute to parties’ ideological flexibility. First, a more ideologically diverse base of local leaders and activists increases the probability that the party will develop a pool of experienced potential candidates representing a variety of policy and ideological commitments. Second, a less complex and differentiated membership and outreach organization increases the probability that the party will be flexible enough to vary the ideological orientation of its candidates from one election cycle to another. Third, there is also some evidence that increased distribution of patronage through the party apparatus enhances flexibility in this sense. The Peronist party appears to have been well structured for ideological flexibility along each of these three organizational dimensions. The party’s local leader and activist base reflect a substantial range of variation around the national party’s ideological posture. Peronism’s formal membership and outreach apparatus is simpler and less differentiated that of several of the other parties under consideration, limiting the opportunity costs of ideological change. Finally, the Peronists channeled significant amounts of patronage through the party apparatus. Between voters’ calculations and the organizational effects on ideological flexibility explored in this chapter, the fact of party-system collapse in Peru and Venezuela, and the reasons why the Argentine party system did not collapse, become less mysterious. Voters may first feel dissatisfaction with the existing political system due to anxiety about the deep and long-lasting economic trouble. Those who identify with the traditional parties then weaken and perhaps abandon their attachment owing to concerns about corruption; such abandonment can be mitigated if the party channels a great deal of clearly party-branded patronage through its organizational apparatus, leading identifiers to conclude that the party family as a whole benefits from corruption. Finally, when the conjunction of concerns about corruption and feelings of underrepresentation makes voters angry, they abandon the traditional parties electorally and begin to support outsider parties and candidacies. Traditional parties can avoid this fate by maneuvering to minimize the set of voters who feel ideologically underrepresented, a process that is facilitated when those parties’ organizations have the flexibility-enhancing features analyzed in this chapter. The argument up to this point has focused squarely on constructing and testing the causal explanation summarized in the previous paragraph. Yet ac-

200

pa rt i e s’ degr e e of ideol ogic a l fl e x ibil i t y

counting for the causes of party-system collapse is not the only important explanatory task. Another central problem is to reconstruct citizens’ experience of the collapse. The concluding chapter makes an initial contribution to this task by exploring the messages about the state and politics that citizens absorb during a process of collapse.

chapter 8

COLLAPSE AND THE EXPERIENCE OF POLITICS

T

h e p a r t y - s y s t e m c o l l a p s e s in Peru and Vene­ zuela have produced extraordinary disruptions in the elite institutions of government. Collapse has brought teams of political outsiders (and, for the most part, novices at government) into the highest positions of political power in each country (Mayorga 2006). Furthermore, tensions between traditional economic and political elites and the new governments provided the motive, or at least the justification, for coup attempts in both countries: a successful coup led by President Fujimori against Peru’s democratic system in 1992 (Cameron 1994: 145–62), and an unsuccessful coup against Chávez and Venezuela’s democratic institutions in 2002.1 Has party-system collapse had equally striking effects on the ways rank-and-file citizens relate to politics in these two countries? This chapter explores the effects of party-system collapse on citizens’ experience of the political realm by disaggregating the phenomenon of collapse into three constituent parts: the crisis of performance and representation that produced party-system collapse; the experience of collapse itself, in which mass electoral participation remade the political landscape; and the aftermath of collapse, in which the elimination of existing elite organizational and discur-

202

collapse and the experience of politics

sive practices opened space for new and potentially different political projects. It then reviews the many possible links between each component of collapse and citizens’ evaluations and attitudes regarding government and politics, developing the hypothesis that the most important consequences of collapse are those that are highlighted in and interpreted by elite political speech, debate, and organizational practices. The content of such discourse during and after party-system collapse in Peru and Venezuela suggests that collapse should leave two major mass-attitudinal legacies in these countries. First, as a result of persistent anti-statist discourse related to the pre-collapse crisis in both countries, citizens should be expected to favor a less expansive role for the state in attempting to resolve society’s most serious problems. Second, because of the highly mobilizing, participatory nature of the leadership that has filled the political space generated by Venezuela’s party-system collapse, citizens in that country should have increased their sense of capacity to influence politics and the policy-making process. Clearly, citizens’ experience of politics is a complex and multifaceted phenomenon. This chapter cannot empirically consider every aspect of that experience. Instead, I focus on a narrower collection of important attitudes, evaluations, and identities affected by the process of party-system collapse. Comparisons between Peru and Venezuela, where collapse occurred, and Argentina and Chile, where it did not, provide an initial cut at thinking about the mass-attitudinal effects of party-system collapse. Regarding a collection of attitudes for which these initial results are of particular interest, more focused within-country comparisons are used to more closely approximate causal inference. This empirical analysis suggests that party-system collapse is associated with, and may well have produced, a broad desire for a smaller government role in many issue areas, as well as a greater sense of political efficacy among citizens. Before proceeding to the empirical component of this argument, however, some theoretical development is needed. What aspects of party-system collapse might serve as treatments that could affect mass attitudes about politics and the state? What sorts of mechanisms might connect these treatments with people’s attitudes? What barriers can we expect to encounter in attempting to make causal inferences about these relationships, and which of those barriers can be at least partially addressed through more fine-grained comparison?

collapse and the experience of politics

203

8 . 1 P A RT Y- S Y S T E M C O L L A P S E A S A C A U S E O F AT T I T U D I N A L C H A N G E

Like any large and complex event, a party-system collapse is a collection of different actions, experiences, and outcomes, each of which might have independent effects on how voters think and feel about politics and the state. Three aspects of the experience of party-system collapse seem particularly salient as treatments that might affect citizens’ experience of the political. First, citizens in a country where the party system has collapsed have lived through a process of state and partisan failure of representation. Second, citizens were involved in the collapse itself, an event in which mass politics clearly and dramatically altered the elite political world. Finally, citizens have received distinct ideological messages, been targets of patterns of mobilization, and dealt with institutions created by post-collapse presidential incumbents; if not for the fact of collapse, these experiences would probably have been confined to the margins of political discourse. Citizens’ relationship with the political world might well be shaped by what they learned from the experience of state and partisan failure during the process of party-system collapse. Many arguments in public-opinion research see people’s attitudes and beliefs about politics as shaped by evaluations of past performance (see, for example, Fiorina 1981; Lewis-Beck 1988; Achen 1992). In this account, attitudes are changed by long-term exposure to a persistent trend or pattern of failure. Citizens are not exposed to only one year of bad economic performance, only one corruption scandal, or only one instance in which political elites act in ways that are unrepresentative and unresponsive. Instead, these failures persist and repeat over time, giving even politically inattentive individuals ample opportunity to directly experience numerous instances of poor performance and adjust their attitudes, identities, and evaluations in response. To some extent, citizens need to make connections for themselves, finding interpretations that connect their personal experiences with broader national themes. However, during the process of party-system collapse, elite actors from outside the traditional party system inevitably offer interpretive frames that emphasize existing patterns of failure and provide models of how to connect individual problems and grievances with negative evaluations of and attitudes toward national parties and government institutions. Hence, the process of political competition tends to provide discourse that helps individuals translate their personal expe-

204

collapse and the experience of politics

riences of partisan and state failure into negative impressions of capacity and motives characteristic of the political world in general. Another way that party-system collapse might influence people’s relationship with the world of politics involves the event of collapse itself, which entails massive and successful collective action in which many or most voters in a country decide to abandon the existing political parties and to support one or a few challengers to the traditional party system. Such successful acts of political participation may themselves shift people’s attitudes about politics (Finkel 1985; Finkel 1987; Clarke and Acock 1989). The event of collapse may lead citizens to conclude that the balance of power in a democratic system favors mass action, and is less overwhelmingly weighted in favor of elite control, than they had supposed. Yet such a conclusion would require substantial interpretive effort on the part of individuals, since this narrative does not produce obvious political benefits for any set of elite actors—post-collapse incumbents are just as threatened by a more assertive and less compliant citizenry as were their pre-collapse counterparts. So we might expect elite messages along these lines to be relatively rare and offered in a ritualistic, pro forma way when they do exist. Instead, political elites might be expected to emphasize their own heroic roles in producing party-system collapse, a message that reinforces their own positions and power. In other words, citizens are largely left to their own resources when it comes to drawing conclusions from their collective success in overthrowing the traditional party system. Furthermore, interpretations of collapse as brought about by successful collective action on the part of an engaged public are complicated by the fact that electoral turnout fell substantially in both Peru and Venezuela during and just after the demise of the party system. In Peru, turnout fell from 80 percent in 1980 to 68 percent in 1990. Turnout in Venezuela dropped even more steeply, from 88 percent in 1983 to 53 percent in 1998. The comparison countries, where no party-system collapse occurred, also saw declining turnout during this time frame: in Chile from 95 percent in 1989 to 87 percent in 1997, and in Argentina from 83 percent in 1983 to 78 percent in 1998.2 Even so, the declines in Peru and especially Venezuela are notably large—and provide a possible counternarrative to any frames emphasizing citizens’ efficacy in producing party-system collapse. Finally, party-system collapse may change the overall distribution of citizens’ attitudes about political processes, institutions, and actors because it clears the

collapse and the experience of politics

205

ground for novel ideological messages, patterns of mobilization, and institutions created by post-collapse incumbents. By silencing or greatly reducing the voice of traditional political elites, party-system collapse may create political space for elite political messages and practices that would otherwise have been excluded from the major mass media outlets and, more generally, from the mainstream of national political debate. Citizens may in turn receive these new messages, either by paying direct attention to politics or by interacting with the state or other political institutions. Receiving new messages does not, of course, guarantee change in people’s fundamental ways of thinking and feeling about the political world—but sufficiently powerful and plausible messages could motivate such change. Of course, not all attitudes about government and politics will be affected by any of these three aspects of party-system collapse. People simply cannot be expected to make all possible interpretive links between ongoing events and their beliefs or attitudes; certain interpretations will dominate, and many others will be neglected. However, party-system collapse is a potent enough political stimulus that it would be surprising indeed if it had no effects on citizens’ impressions of politics and government. Hence, the most compelling questions are probably: which attitudes and beliefs should we expect to be most affected by party-system collapse, and which aspect or aspects of collapse should we expect to be most important in altering citizens’ perceptions and identities? One theoretical approach to these questions is to consider citizens’ use of elite framing as a route to minimizing the cognitive resources necessary to engage with the political world. Appropriating prepackaged messages and ideologies that interpret events and situations is, after all, much easier than interpreting those events and situations de novo (see, e.g., Downs 1957: 96–100). We might, then, expect those aspects of party-system collapse that are most prevalent in elite political discourse to have the largest effects in reshaping citizens’ beliefs, attitudes, and identities with respect to the political world (Zaller 1992). Furthermore, the specific impressions that should be most affected are those most clearly and directly connected with patterns of elite political debate. Applied to Peru and Venezuela, these arguments suggest that two sets of attitudinal changes should result from party-system collapse. First, anti-statist messages associated with neo-liberal ideology should lead citizens to reduce their sense of how capable government and political institutions are at resolving

206

collapse and the experience of politics

society’s problems. With less confidence in the efficacy of government action, citizens in post-collapse countries should assign a smaller leadership role to the state in resolving most of society’s problems; if the state is incapable of resolving problems in any case, why should it get involved in the effort? The anti-statist messages central to this hypothesis came to prominence in Peru particularly during the presidential campaign of Mario Vargas Llosa in 1990. Vargas Llosa vociferously interpreted the political, social, economic, and military challenges of Peru through the 1980s as evidence of state incapacity to solve major problems. His proposed solution was to sharply curtail the scope of state activity, and to encourage society to reduce its expectations regarding the efficacy of state programs (Florez 1992). Although Alberto Fujimori, the eventual winner of the 1990 elections, campaigned against Vargas Llosa’s antistatist messages, he quickly reversed course after the election, coming to echo and thus perpetuate Vargas Llosa’s analysis of Peru’s difficulties as evidence of low state problem-solving capacity (Stokes 2001: 69–73). An anti-statist elite framing regarding the meaning of Peru’s pre-collapse crisis thus persisted for years, giving citizens ample opportunity to receive and accept this message. The combination of these strong elite messages and the depth of Peru’s crisis makes it very likely indeed that citizens would respond by reducing their sense of government’s capacity to solve problems and, hence, their sense of how large a role the state should play in addressing specific problems. Similar anti-statist elite messages became publicly salient in Venezuela during President Carlos Andrés Pérez’s “gran virage” period of neoliberal reforms after 1988 (Naim 1993). Pérez’s rhetoric, like Vargas Llosa’s, emphasized the inability of state institutions to solve a wide range of societal problems and the need for society to temper its expectations for how much the state could or would do. While Pérez’s presidency ended in failure and impeachment, anti-statist neoliberal ideology did not leave its position of prominence in Venezuelan elite discourse until the election of Hugo Chávez in 1998. Candidates in both the 1993 and 1998 elections ran on anti-statist platforms, and a similar set of messages became a central part of initially pro-statist president Rafael Caldera’s response to a banking crisis midway through his presidency. Hugo Chávez’s election in 1998 initiated a major shift in the ideological tone of elite discourse in Venezuela, yet his discourse continued to emphasize the weakness and incapacity of state organizations—particularly, those still controlled by clients of the tradi-

collapse and the experience of politics

207

tional party system. The rhetoric surrounding constitutional reform during the period 1999–2000, for example, emphasized the incompetence and unresponsiveness of existing state structures and the need for institutional renovation (Alvarez 2003). Thus, as in Peru, there was an ample supply of elite discourse framing Venezuela’s social, political, and economic difficulties as evidence of a state lacking the ability to resolve major societal problems, and this discourse may well have produced a long-term decline in Venezuelans’ sense that the state should take a leadership role in addressing various categories of social problems. Anti-statist neo-liberal messages have certainly also been part of elite discourse during some of Argentina’s and Chile’s economic crises. Carlos Menem’s neoliberal reforms in Argentina during the early 1990s were accompanied by similar messages, as was the neo-liberal program of Chilean dictator Augusto Pinochet during the 1970s and 1980s. Thus, anti-statist neo-liberal elite messages are by no means always or inevitably related to a crisis that precedes partysystem collapse, as they were in Peru and Venezuela. Unlike the post-collapse countries, however, both Argentina and Chile have experienced extended periods in which anti-statist rhetoric did not occupy a central position in elite discourse; in Argentina, the period since the beginning of Menem’s second term saw a decline in such messages and even a rise in statist discourse, and Chilean politics since the end of the Pinochet dictatorship has seen a renewed elite discursive emphasis on state successes and capacity. Hence, we might expect to see differences, on average, between citizens’ views of how much problemsolving leadership the state should take in the post-collapse countries and in the non-collapse countries—differences that would be consequences of the crisis that led to party-system collapse as filtered through elite anti-statist discourse. A second set of expected attitudinal changes is more specific to Venezuela. When party-system collapse cleared away existing elites and created space for new messages and ideologies, the project that emerged to occupy the newly vacant political space was one distinctively oriented toward messages about the importance and efficacy of mass political participation. Hugo Chávez’s rhetoric of participatory democracy has been a staple of political campaigns and presidential speeches since 1998. These themes were written into the 1999 Venezuelan constitution, and they are regularly discussed in meetings and rallies of Chávez supporters at all levels. Chávez’s participatory messages emphasize two themes directly related to political efficacy. One is the openness of the state policy-­making

208

collapse and the experience of politics

process to citizens’ participatory input. Whether such openness is genuine or merely symbolic is difficult to judge; however, it seems noteworthy that a message of openness is communicated through presidential speeches, organizational meetings, and the participatory structure of many new Venezuelan social programs. A second theme is the capacity of the people to provide meaningful input into politics. Consider, for example, the following excerpt from Chávez’s May 13, 1999, speech on the accomplishments of his first 100 days in office: [I]n two hundred years of history here, what happened on the 25th of April has never happened before: a National Referendum. It is the vindication of democracy, but true democracy, participatory democracy, consultive democracy. Never before had any president who had arrived here [the presidential palace], on horseback or on foot, in uniform or in civilian garb, called on the people to consult them in a National Referendum. . . . And we’re in the middle of the process, we’re in the middle of an electoral stage. Let’s go! I deeply and fully trust the people, the Venezuelans, you, that you’ll know how to choose on the coming July 25th, when the elections are held. (Chávez Frías and Troudi 2005a, 177–78)

Note that this passage emphasizes the new openness of the state to popular participation: for the first time, Venezuela is a true, participatory democracy, and the president calls on the people for consultation in the most important national decisions. Furthermore, Chávez emphasizes his belief that the people have the capacity to act decisively and correctly in this new, more open and participatory system. Similar themes emerge in Chávez’s more general speeches about the broad direction of democracy in Venezuela and elsewhere, as for example in Chávez’s January 15, 2002, annual address to the Venezuelan legislature: Venezuela has placed a political theme of phenomenal impact on the table, and it has been taken up in various spaces. The debate about democracy has resumed, about what kind of democracy it is that we need or that our peoples need. A formal democracy, made of cardboard, hollow, and without social content. Or a democracy that, in addition to being representative, is also participatory, in which the people participate in the making of decisions. (Chávez Frías and Troudi 2005b, 77)

While couched in more universal and more theoretical terms, this kind of discourse—like the more practical quotation about referendums above—emphasizes the new openness of the Venezuelan state to popular participation,

collapse and the experience of politics

209

as well as the urgency of such participation if the country is to avoid the fate of having a politics that is “cardboard, hollow, and without social content.” Chávez’s speeches contains numerous similar exhortations. As an illustration of the other media through which Venezuelans have also received such elite messages, it may be worth reporting a few of the passages in Venezuela’s constitution that reinforce this rhetoric of state openness and popular participation. The preamble to the constitution says that the document was promulgated “with the supreme goal of refounding the Republic to establish a democratic society, participatory and protagonistic.” In a more expansive vein, Article 62 of the constitution states: All citizens, male and female, have the right to participate freely in public affairs, directly or through their elected representatives. The people’s participation in the formation, execution, and control of public management is the necessary means to achieve the protagonistic role that guarantees their complete development, both individual and collective. It is an obligation of the state and a duty of society to aid in generating the most favorable conditions for putting this into practice.

Here, citizens are promised the right to be involved in every phase of the political process. They are told that the state will consider itself obliged not only to take their participatory input seriously, but also to create conditions that make effective mass participation as easy as possible. Literally dozens of other passages in the Venezuelan constitution address similar participatory themes, promising a state attentive to popular input and promoting citizens’ self-conception as competent and integral parts of the policy-making process. This combination of messages, in various channels of communication, about state openness to participation and citizen capacity may well be a potent treatment for increasing Venezuelans’ levels of external political efficacy, or belief that they have the capacity to influence the political and policy-making process. Overall, then, this discussion generates two major hypotheses. First, citizens in both post-collapse countries are expected to be somewhat less likely to favor government leadership in resolving society’s major problems than are those in the non-collapse comparison countries. Second, Venezuelans should have higher opinions of their ability to influence government decision-making than do Argentines or Chileans. More broadly, the conceptualization of party-system collapse as a cause developed here suggests that collapse may affect citizens as a

210

collapse and the experience of politics

long-term trend of state failure during the process of collapse, as an event of effective citizen mobilization against existing political elites, or as a juncture that opens political space for the diffusion of new elite messages and ideologies. If citizens, in interpreting such a complex collection of stimuli, make the effortsaving decision to rely on elite interpretations in reshaping their attitudes, then we may expect the most causally efficacious aspects of collapse to be the trends of failure that led to collapse and new ideologies that emerged after collapse—but primarily as interpreted through the rhetoric and organization of political leaders.

8 . 2 C I T I Z E N I M P R E S S I O N S O F P O L I T I C S A N D   G O V E R N M E N T: E VA L U AT I O N S AND POSSIBLE MECHANISMS

In making the argument for the hypotheses developed above, in which the effects of party-system collapse on subsequent individual attitudes are largely consistent with major themes in elite discourse, the analysis focuses on four categories of evaluations, attitudes, and identities that, in conjunction, partially characterize citizens’ overall experience of politics and government. Attention is given to individuals’ evaluations of specific political and state institutions, patterns of party identification, attitudes regarding respondents’ capacity to understand the political world and make a difference in the policy process, and citizens’ sense of how expansive state action should be across a wide range of issue areas. Each attitude or identity can be plausibly hypothesized as an effect of at least one aspect of party-system collapse discussed above, although the data support the argument that, following the lead of elite discourse, the central attitudinal effects of party-system collapse should be a reduction in support for an expansive government role in solving society’s problems, as well as an increase in external political efficacy among Venezuelans in particular. Other aspects of citizens’ experience of government are considered both because it is intuitively plausible that they could be affected by collapse and because they serve as a useful contrast, showing that the two hypothesized effects are distinctively large in substantive terms and not merely statistically significant. Evaluations of specific institutions are often regarded as being heavily affected by citizens’ evaluations of recent performance (both economic and political) by each institution (Lewis-Beck 1988; Miller and Listhaug 1999; Lewis-Beck and

collapse and the experience of politics

211

Paldam 2000; Brady and Theriault 2001). Hence, one might expect that the crisis of performance and representation leading up to the occurrence of partysystem collapse would have produced a broad legacy of negative retrospective evaluations of government and political institutions. Yet since such evaluations are usually regarded as most sensitive to recent trends and relatively myopic regarding longer-term performance, we may expect that any negative institutional evaluations resulting from the crisis before party-system collapse would be quickly overwhelmed by evaluations drawing on post-collapse considerations. Furthermore, countries that have experienced collapse have no particular monopoly on problems of state political and economic performance, as seen in Chapter 3. Overall, then, there is perhaps a weak expectation that party-system collapse would have produced a legacy of negative evaluations of political and government institutions, though it remains possible that there is no durable effect along these lines. The data show that there is very little evidence of any difference between citizens of post-collapse countries and their non-collapse counterparts in various institutional evaluations: evaluations of the government’s ability to manage the economy; the contributions of the president, the legislature, the judiciary, and the political parties to solving society’s biggest problems; and the extent to which the party system manages to represent the upper classes, the middle classes, the working classes, and the poor. There are, of course, quite different accounts of the nature and causes of identification with a political party. The two most famous regard party identification as a heuristic that serves primarily as a running tally of past governmental performance (Fiorina 1981; Achen 1992), or as an attachment that results from the combination of primary social identities with stereotypes about which parties stand for those identities (Green, Palmquist, and Schickler 2002). If party identification is primarily a running tally of performance evaluations, then—as with institutional evaluations above—we might expect fewer citizens in post-collapse countries to report a party identification. After all, repeated corruption scandals before the party-system collapse eroded identification with the traditional parties, as discussed in Chapter 4. Since the collapse, citizens have had less time to develop positive running tallies, and parties have had less time to develop a positive performance record, than in countries that have not experienced collapse. If, by contrast, party identification is also (or perhaps instead) a result of primary social identities in conjunction with partisan stereotypes, then no clear expectation can

212

collapse and the experience of politics

be formed. If post-collapse politics has generated partisan stereo­types that link as many primary social groups to newly emerged parties as did the stereotypes underpinning the traditional party system before collapse, then there may be no difference in the distribution of party identifications between post-collapse and non-collapse countries. Of course, the newly developed partisan stereotypes in Peru or Venezuela might be either more or less inclusive than the stereotypes in the traditional party systems, in which case the distribution may indeed shift. In fact, as is shown below, levels of party identification are as high or higher in the post-collapse countries as in the countries that did not experience party-system collapse, thus generating a modest anomaly for the view of party identification in this context as a pure running tally rather than in part a linkage that emerges from the conjunction of primary group identities and partisan stereotypes. This finding probably precludes the identification of any lasting legacy of party-system collapse for the existence of party identifications in Peru and Venezuela. Regarding citizens’ preferences for the scope of state action in resolving society’s biggest problems, the situation is more ambiguous. Presumably, an individual’s opinion about whether the state should take the lead in trying to resolve problems in a given domain involves a combination of other beliefs and assessments. Specifically, it seems reasonable to suppose that the problem areas where an individual believes that the state should take the lead are problem areas in which: (a) she finds the status quo is unacceptable, (b) she believes that government action can be effective, and (c) she feels that the government institutions that would be responsible for action on the problem area in question are trustworthy and competent. If the status quo is acceptable to the respondent, then she may see little need for government action—which always carries a cost, as well as the risk of making the situation worse. If the respondent sees government action as ineffective for the problem area in question, then there is little reason to request a major government role, even if the status quo is intolerable. Finally, even if the status quo is unacceptable to the respondent and she believes that government action can be effective in the problem area in question, she may not want government action if the current holders of power in the relevant institutions of the state are seen as incompetent or dishonest: such office-holders would probably only make matters worse. The third aspect of this account, evaluations of government institutions, has been discussed above. The first two, respondents’ sense of the severity of

collapse and the experience of politics

213

various social problems and their degree of belief that government action can be effective across various problem areas, may well have been affected by the experience of the crisis that preceded party-system collapse. The possible effects of a multifaceted social and economic crisis on citizens’ sense of the severity of social problems are direct enough that they, perhaps, need little further comment. Such a crisis might also alter citizens’ mental models of the social, economic, and political worlds by teaching them that government efforts to resolve society’s problems are broadly ineffective. During the decade leading up to party-system collapse, governments in both Peru and Venezuela launched repeated programs to bring social and economic problems under control, without success. If citizens generalize these failures to a conclusion that government action has little effect, then the crisis before party-system collapse could produce a long-term decline in citizens’ sense of how big a role the state should take in a range of problem areas. The evidence shows that there is at least some reason to believe that such a decline has taken place in Peru and Venezuela in problem areas involving pollution, economic need, crime, public services, and health care—a decline that may be one of the major effects of the crisis phase of party-system collapse on citizens’ experience of the political world. Finally, political efficacy involves a complex of attitudes and evaluations that may plausibly be influenced by any of the three aspects of party-system collapse under consideration here. Internal political efficacy, or a respondent’s degree of confidence that she can understand the political world, could quite plausibly be decreased by the chaos and political crisis leading up to partysystem collapse, increased by the decisive citizen electoral action involved in producing the collapse itself, or modified in unpredictable ways by elite messages and institutional changes after the collapse. A second aspect of political efficacy considered here, respondents’ degree of agreement that greater citizen participation in party politics could help resolve society’s problems, could be reduced by the pattern of partisan unresponsiveness during the pre-collapse crisis or increased by the emergence of more acceptable parties in the wake of collapse. The third form of political efficacy analyzed below, external efficacy or citizens’ evaluation of the degree to which the political process is responsive to people like them, could be negatively affected by the crisis of representation before collapse. Alternatively, external efficacy could be increased by the successful mass political participation in electorally producing a party-system collapse

214

collapse and the experience of politics

or—especially in Venezuela—through exposure to participatory and mobilizing frames and institutions promulgated by post-collapse political incumbents. While the set of plausible relationships involving all three of these aspects of political efficacy contain ambiguities in which party-system collapse could have either increased or decreased average levels of efficacy, this ambiguity is most easily resolved for external efficacy. For internal efficacy, negative and positive considerations are available at all time periods, making it almost impossible to form any strong expectation for an overall effect. Turning to evaluations of how much difference citizen involvement in partisan politics could make, the hypothetical positive effect is more recent than the negative effect, but its existence also depends on the extent to which citizens in general see the post-collapse party system as generally superior to the pre-collapse system. If enough citizens are cynical about party politics, any effect regarding this aspect of efficacy may be swamped. Regarding external efficacy, however, the highly participatory ideology and organizing strategy of Chávez’s political leadership is more recent than any other effects, and therefore may could be expected to dominate as impressions related to earlier events fade away. Furthermore, it is a powerful and easy accessible stimulus, presented to Venezuelans through billboards, televised speeches, government social programs, campaign slogans, and so forth. Finally, the degree of cognitive sophistication and effort necessary to receive and process these direct, plainly presented, and vivid messages is quite low; a lack of internal efficacy, education, or other skills poses relatively little barrier to absorbing these stimuli, in contrast with the negative stimuli of the earlier crisis of representation, which require relatively greater historical attention and intellectual effort to consider in the context of post-collapse Venezuela. Hence, a positive effect of post-collapse mobilization on external efficacy in Venezuela may be the most theoretically plausible efficacy-related effect of party-system collapse. A variety of empirical comparisons below will provide at least partial evidence that such an effect both exists and is of substantively meaningful magnitude.

8.3 BARRIERS TO CAUSAL INFERENCE

The central strategy used below for finding initial evidence of the effects of partysystem collapse on citizens’ experience of politics and the state is to compare the reported experience, along the various dimensions discussed in the previous

collapse and the experience of politics

215

section, of individuals in Peru and Venezuela with those of people in Argentina and Chile. Not all differences will be due to party-system collapse, and it is also possible for collapse to have an effect without the creation of a cross-national difference if the effect changed what had been a distinctive distribution of attitudes into one more in keeping with regional norms. These caveats notwithstanding, it seems likely that attitudes that are significantly affected by party-system collapse would on average look quite different in Peru and Venezuela than in noncollapse countries in the region. Of course, citizens of the post-collapse countries, Peru and Venezuela, differ in many ways from citizens of Argentina and Chile. Any difference whatsoever between these two groups may serve as a confounding factor, i.e., a variable that makes a bivariate relationship appear where no true causal connection exists, that makes no bivariate relationship appear where there is a genuine causal connection, or that otherwise produces observed empirical relationships that do not accurately reflect underlying causal patterns (see, e.g., Freedman 2005: 2–4; Morgan and Winship 2007: 59–85). As always, some effort must be made to address each credible potential confounding variable or any findings will be difficult to interpret. One possible source of confounding variables involves different socialclass distributions across the four countries. For example, the percentage of respondents who have completed at least a high school education varies: the figure for Argentina is 44.3 percent, Chile 57.7 percent, Peru 33.3 percent, and Venezuela 29.8 percent. Likewise, unemployment rates differ in the four countries: 12.7 percent in Argentina, 7.8 percent in Chile, 5.9 percent in Peru, and 17.7 percent in Venezuela. These differences are not the only the possible confounders related to social class, but they are sufficient to give a flavor of the problem. Rather than attempt to consider all possible social-class confounders, which could overwhelm the data in a relatively rough-grained analysis like the present discussion, the analysis here focuses on these two factors out of the range of possible social-class confounders. Differences in media messages influence the direction and intensity of citizens’ perceptions of government and the political process. The extent of individuals’ exposure to news media may change their level of political information, and as a consequence have a range of effects on their political attitudes and identities. Media markets vary across national boundaries, and the substantive focus

216

collapse and the experience of politics

of newspaper coverage can also differ within a single market. However, there appear to be no survey data available for South America that link measures of exposure to specific media messages and themes with measures of attitudes and evaluations regarding government and politics. As a result, this analysis settles for the less satisfactory alternative of simply considering respondents’ degree of attentiveness to various categories of news media. Even on this second-best measure, there are some potentially important differences across countries. For example, 64.4 percent of Peruvian respondents report watching political news on television every day, as compared with 71.2 percent of Argentines, 77.0 percent of Chileans, and 76.5 percent of Venezuelans. There is even more reported cross-national divergence in patterns of newspaper consumption: 20.9 percent of Argentines report reading political news in a paper on a daily basis, while 13.0 percent of Chileans, 20.9 percent of Peruvians, and 35.3 percent of Venezuelans report the same behavior. In general, reported patterns of news media exposure are divergent enough across these four countries that, for attitudes that are thought to be especially sensitive to information effects, some attention to this factor as a potential confounder may be of value. Quite naturally, given objective differences in national situations and political discourse, citizens’ sense of what society’s biggest problems are varies somewhat from country to country. Yet if citizens see their countries as facing different problem agendas, their evaluation of how competent the political world is at resolving society’s biggest problems may well diverge even if underlying beliefs about the state and the political process are fundamentally similar. Respondents’ sense of their countries’ problem agendas are different enough across the four countries under analysis here to raise these concerns. For instance, in all four countries, unemployment is the modal response when individuals are asked to name the single most important problem in their society. However, the frequency with which respondents mention unemployment in this context varies: 42.3 percent in Argentina, 33.3 percent in Chile, 38.1 percent in Peru, and 32.9 percent in Venezuela. In three of the four countries, violence and crime is the second-most-reported problem, but in Venezuela education slightly outranks criminality on citizens’ problem agenda. Overall, there are enough differences in citizens’ problem agendas across countries to cause problems for attitudes that are problem-area specific. Differences in countries’ patterns of political-participatory experience might cause differing attitudes toward politics and government or be consequences of

collapse and the experience of politics

217

those attitudes, or both. This category of potential confounders thus belongs to the messy set of variables that could cause bigger problems for causal inference when they are used as conditioning variables than when they are simply disregarded. Nonetheless, cross-national differences in political participation are large enough that it may not always be reasonable to ignore this potential set of confounding variables. Thus, in three of the four countries, relatively few people report having attended a party rally during the previous national campaign: only 8.2 percent of Argentines, 6.0 percent of Chileans, and 6.1 percent of Venezuelans. However, 12.4 percent of Peruvians had attended a campaign rally—a difference that is certainly large enough to matter. Similarly, there are substantial cross-national differences in how many respondents report having personally contacted the government during the previous five years: 20.8 percent of Argentines, 20.7 percent of Chileans, 16.3 percent of Peruvians, and 9.3 percent of Venezuelans. While it is difficult to decide, for any given attitude, whether participatory experience is a meaningful confounder or whether it stands in some other relationship to the evaluations and identities of interest, it is clear that there are some substantively large cross-national differences that may be worth taking into account in comparisons across these countries. Party-system structures have historically differed substantially across the four countries under consideration here, with Venezuela traditionally having an archetypal well-institutionalized two-party system, Chile famous for its three-ideological-bloc system, Argentina having a party system built around the powerful and persistent Peronist movement, and Peru having a much more fluid and personalistic party system, although one incorporating the long-lasting APRA party. As the Peruvian and especially Venezuelan experiences discussed in earlier chapters show, none of these patterns is necessarily permanent. However, they may well have had an influence, large or small, on how current citizens see the political world in these countries. Unfortunately, the available data provide few clues as to how each individual respondent was affected by these historical party systems; hence, this potentially important factor will necessarily be disregarded. Present patterns of partisanship may also differ, an issue that is explored empirically below. To whatever extent the structure of partisanship does indeed differ across these four populations, this may be an important confounding variable if, as is sometimes suggested, citizens filter new information in line with their existing partisan dispositions.

218

collapse and the experience of politics

This analysis, based as it is on observational data, cannot definitively address any of these possible confounders; for that, a difficult and resource-intensive experimental design would probably be needed in which individuals are randomized to one of a set of pre-specified sequences of elite messages and institutional initiatives over a period of years. Yet there is value in exploring observed differences in attitudes as a way of discovering possible effects of party-system collapse, and as an aid to reasoning about which aspect of the experience of collapse is most likely to be relevant in explaining each dependent variable. For these purposes, it can be helpful to make comparisons conditioning on subsets of the potential confounding variables discussed above. When such comparisons, conditional on possibly confounding variables, are wanted, this analysis uses nearest-­neighbor matching techniques, sometimes with perfect matching but often with a propensity-score specification (Rosenbaum and Rubin 1983; Rosenbaum 2002; Morgan and Winship 2007: 87–122).3 The primary data source for this discussion is the 2002–03 CIRELA survey (Collier 2003), which provides data on random samples of individuals from the capital cities of Argentina, Chile, Peru, and Venezuela. The survey includes a useful variety of measures regarding citizens’ perceptions and experience of the political world, as well as the measures mentioned above on potential confounders. However, the restriction to capital-city respondents is a limitation. There is no particular reason to believe that citizens outside each country’s largest city have the same political attitudes and experiences as those in that city, and hence the applicability of this analysis outside those cities is an open question. Even so, for purposes of verbal simplicity, respondents are labeled in the discussion below with reference to the country they live in, rather than the city.

8 . 4 I N I T I A L C O M PA R I S O N S

What, then, are the differences between citizens of Peru and Venezuela, the post-collapse countries, and those of Argentina and Chile, the non-collapse countries, with respect to their reported experience of government and politics? This section presents initial information about the distributions of the attitudes in question, without conditioning on potential confounders. The discussion here thus produces only very preliminary ideas about possible effects of party-system collapse. However, it is helpful both to empirically high-

collapse and the experience of politics

219

light the attitudes, evaluations, and identities for which an effect of collapse is most plausible, and to provide a guarantee that the eventual findings are not mere artifacts of elaborate conditioning strategies. The discussion in this section provides an overview of attitudinal differences and similarities grouped into three major categories: retrospective evaluations of the problem-solving performance and representativeness of political institutions; citizens’ impressions regarding participation in government and politics; and citizens’ preferences regarding how much of a leading role the government should take in solving society’s problems. For the most part, in terms of direct, retrospective evaluations of state institutions, citizens in Peru and Venezuela have attitudes that fall comfortably within the range marked out by citizens in Argentina and Chile.4 When evaluating whether the legislature, the judiciary, and the political parties have made a substantial contribution to resolving the country’s biggest problems, Peruvian and Venezuelan citizens’ average attitudes fall squarely between the average attitudes of Argentine and Chilean citizens. For all three evaluations, Argentine citizens rank these institutions as having done the least, while Chilean citizens rank them as having done the most. Peruvian and Venezuelan citizens fall in between, although Peruvians are more positive toward some institutions than Venezuelans and more negative toward others. Thus these evaluative attitudes offer no a priori evidence for any effect of party-system collapse. The same holds true for evaluations of whether various social-class groups are represented by some party within the existing party system. For all four class groups considered (upper, middle, and working classes, as well as the poor), Argentine citizens are most likely to report that no party represents the group in question. Likewise, as with retrospective evaluations of the three state institutions discussed in the previous paragraph, Chilean citizens are most likely, for each class group, to report that some party does indeed represent that group. And, as before, Peruvian and Venezuelan citizens fall between Chileans and Argentines in the frequency with which they report that no party represents these four groups. While it may be interesting to note that Venezuelans are more likely than Peruvians to report that some party represents the middle and working classes and the poor, but less likely to report that some party represents the upper classes, these variations fall well within the bounds established by the two countries that did not experience party-system collapse. Hence, once again,

220

collapse and the experience of politics

there is no initial evidence that the experience of collapse shaped the distribution of these attitudes. Indeed, there is only one attitude involving the evaluation of political institutions for which both the Peruvian and Venezuelan averages fall outside the range established by Argentina and Chile: evaluation of how well the government has managed the economy over the past five years. For both countries, the average of this economic evaluation is distinctly more negative than it is for either Chile (where 50 percent say that the government has done a good or a very good job) or Argentina (70 percent positive evaluations). Furthermore, Peruvians (16 percent positive) are on average quite a bit more negative in their economic evaluations than are Venezuelans (37 percent positive). Interpreting this result requires attention to the timing of the survey, both in relation to the event of party-system collapse and in connection with economic outcomes during the five-year period of interest. As discussed above, the survey was carried out in 2002 in Peru and in 2003 in Venezuela—more than a decade after Peru’s party-system collapse and about five years after the collapse in Vene­ zuela. Hence, the question’s specified period of economic evaluation is either mostly (for Venezuela) or entirely (for Peru) during the post-collapse period; these results should thus not particularly reflect poor economic management by the state during the crisis leading up to party-system collapse. Instead, a brief consideration of economic data suggests that this difference may instead be due to citizens’ relatively accurate perceptions of the countries’ more recent economic trajectories. During the five-year period before the CIRELA survey was administered in each country, Venezuela’s real per capita GDP shrank by an average of 2.76 percent per year, while Peru’s GDP grew by a less-than-impressive average of 0.35 percent per year (Heston, Summers, and Aten 2006). By way of comparison, over the same period, Chile’s GDP grew by an average of 1.29 percent per year. Argentina’s GDP actually shrank by an astonishing average of 3.70 percent per year through this period, although this collapse seems to have been counterbalanced in Argentine citizens’ economic evaluations by the impressive economic recovery under way in 2003 when the survey was fielded.5 Hence, this cross-national comparison supports the idea that citizens’ evaluations of the economy are grounded in economic reality; there is no clear basis for suspecting this result to be a product of party-system collapse.

collapse and the experience of politics

221

For one institutional evaluation, the average attitude in Venezuela but not Peru departs significantly from the bounds established by average attitudes in Argentina and Chile: evaluations of how much the president has done to address the country’s biggest problems. Although, in an absolute sense, few citizens in any of the countries are favorable about how much their president has accomplished, there is a meaningful difference between the proportion of Venezuelans who think the president has made an important contribution (above 30 percent) and the proportions in Chile (almost 30 percent), Argentina, and Peru (around 15 percent in both cases). These evaluations probably include considerations regarding the individual presidential incumbent as well as considerations regarding broader attitudes toward the institution of the presidency. In other words, this result may reflect Hugo Chávez’s core of distinctively committed supporters, rather than any more direct effect of party-system collapse. Yet it is worth noting that the Peruvian results date from a period when ­Alberto Fujimori, the president who served immediately after the party-system collapse, was no longer in office. Considerations regarding the unpopular 2002 incumbent, Alejandro Toledo, are probably involved in the Peruvian results. It is likely that results regarding this question from 1995, when Fujimori had been in office for about the same amount of time as Chávez in 2003, would have shown a more positive evaluation of the Peruvian presidency’s contribution to resolving society’s biggest problems. After all, in 1995, the defeat of the Sendero Luminoso insurgency and the resolution of the economic crisis of the late 1980s would still have been reasonably fresh in Peruvians’ minds, and Fujimori and the Peruvian presidency might have received substantial credit for those developments. It is plausible, although not adequately empirically demonstrated by these data, that party-system collapse creates a window in which individual political leadership is seen by citizens as more efficacious than institutional action. This perception might be an accurate reflection of reality, since the process of collapse weakens the parties that provide structure to the legislative arena and also (in both Peru and Venezuela) involved institutional discontinuities that undermined the abilities of the judiciary and the bureaucracy to act independently of the president. Such a perception might also arise from the campaign rhetoric of outsider candidates during the process of party-system collapse. Because such candidates stand outside the collection of established, institutionalized parties, their rhetoric will almost necessarily be more personalistic than that of more regular elections

222

collapse and the experience of politics

during periods of party-system stability—even if those more regular elections are themselves highly personalistic in nature. In either case, if such a perception results from party-system collapse, it would account for Venezuelan citizens’ higher attribution of efficacy to the president in these data. More research is needed to determine to what extent this causal linkage is genuine, and what the dynamics of this perception might be over time. Overall, there is no clear initial evidence that party-system collapse affected citizens’ attitudes about the performance of political institutions or the representative capacity of political parties. If any aspect of collapse had an effect on these attitudes and evaluations, the effect was subtle enough to leave the Peruvian and Venezuelan publics’ attitude profiles in line with those of Argentina and Chile. On the one hand, this result may be somewhat surprising, given that party-system collapse involved important failures of performance and representation by many of the institutions discussed above. On the other hand, it may be that these attitudes and evaluations are responsive enough to short-term changes that political events and circumstances after the collapse dominate the effects of the collapse itself. While the theoretical argument developed earlier produced no expectation of an effect of party-system collapse with respect to the retrospective evaluations just discussed, theory does suggest that we should find some differences in attitudes toward participation in political processes. As with the expected non-effect regarding retrospective evaluations, the expected effects regarding participatory attitudes—primarily that external efficacy will be comparatively high in Venezuela—are confirmed by initial analysis of the data. In a somewhat surprising finding, Table 8.1 shows that citizens of Peru and Venezuela are at least as likely to report identifying with a political party as are citizens of Argentina and Chile. This result may in part reflect a somewhat lower threshold for reporting a party identification in the post-collapse countries. On the other hand, identifying with a party has a level of predictive power for various partisan forms of political participation that is similar across postcollapse and comparison countries, implying that partisanship may have similar meaning and intensity in these varying contexts (Seawright 2006: Chapter 5). More generally, this finding suggests that fundamental political orientations such as the tendency to form a party identification are relatively resilient in the face of macro-political reorganization, an interpretation that is further reinforced

collapse and the experience of politics

223

ta b l e 8 . 1 . Citizens’ views of parties and elections Percentage of citizens who…

Argentina

Chile

Peru

Venezuela

Identify with a party

50.2

48.0

64.5

51.1%

Say citizens’ involvement in party politics could help solve biggest problems

57.0

35.7

48.3

63.4

Say blank votes are a legitimate expression of political dissatisfaction

47.4

61.2

66.2

39.2

Believe citizens have low internal efficacy

50.8

51.3

55.0

Believe citizens have low external efficacy

69.2

85.7

54.8

by similar patterns of statistical relationships between having a party identity, on the one hand, and variables such as social class, civil-society engagement, and a range of political attitudes, on the other (Seawright 2006: Chapter 5). Thus, while Peru and Venezuela did experience party-system collapse, they have by no means become post-partisan political systems. On the other measures of citizens’ relationships with parties and elections presented in Table 8.1, there are some surprises and hints that party-system collapse may have shifted the parameters of mass political opinion. Consider first the distribution of attitudes regarding whether greater citizens’ involvement in party politics could help resolve society’s biggest questions. Responses to this question probably reflect a mixture of two fundamental attitudes: citizens’ external political efficacy or sense that they can make a difference in politics, and citizens’ sense that the existing political parties are underperforming and could do a better job of handling major problems. If citizens generally see themselves as lacking the capacity to change anything in the political world, then greater citizen involvement in party politics would not make any important difference. Likewise, if parties were already doing about the best that they could in managing the biggest problems in society, then additional citizen input would either make things worse or leave them the same. Hence, a positive response to this question probably requires a combination of substantial external political efficacy and at least some degree of suspicion that parties are really performing as well as they could. Results discussed earlier suggest that large majorities of

224

collapse and the experience of politics

citizens in all four countries feel that parties have not done very much about society’s biggest problems, so in practice this measure probably largely reflects citizens’ degree of belief that party politics can be affected by popular participation; that is, it is probably in large part a measure of efficacy. According to this measure, Argentine citizens show markedly higher levels of external political efficacy than do Chileans. Peruvians fall about halfway between the two non-collapse countries. Venezuelan citizens, however, appear to be more optimistic than citizens in the three other countries about their ability to improve society by participating in party politics. The fact that Peru falls within the range established by the non-collapse countries but Venezuela does not raises a range of interpretive options. Venezuela’s higher levels of external efficacy may be a long-standing Venezuela-specific phenomenon, a product of party-system collapse that wears off over time, or a result of the specific patterns of political mobilization and leadership under Hugo Chávez during the post-collapse period. More detailed analysis below begins to parse out these explanatory possibilities; for now, it is sufficient to note the pattern. The next measure in Table 8.1 captures the degree to which citizens in each country see nonparticipation in politics as legitimate. Voting is formally mandatory in each of the countries under analysis, although penalties associated with abstention vary a great deal from country to country. Hence, an item that directly asks about the legitimacy of deciding not to vote might be contaminated by respondents’ attitudes regarding obedience to the law. However, with a secret ballot, it is challenging, to say the least, to attempt to prohibit the casting of blank or spoiled ballots as an act of deliberate nonparticipation. So it may be possible to gauge citizens’ sense that nonparticipation is acceptable by asking them whether casting a blank ballot is a legitimate expression of political discontent or a dereliction of a basic duty of citizenship. About half of the sample’s Argentine respondents see nonparticipation as a legitimate mode of self-expression, and just over 60 percent of Chilean respondents agree with this assessment. In comparison with citizens in these two non-collapse countries, Peruvians are somewhat more likely to regard nonparticipation as legitimate. By contrast, Venezuelans are relatively likely to see nonparticipation as a dereliction of duty; only about 40 percent of Vene­ zuelan respondents describe casting a blank ballot as a legitimate act of civic protest. This finding is interesting in itself, and the pattern among Peruvians

collapse and the experience of politics

225

and Venezuelans corresponds well with the explicitly mobilizing and participatory political discourse of Chávez’s Venezuelan regime as well as the relatively demobilizing nature of Fujimori’s Peruvian government. However, this pattern, with one collapse country above the range established by Argentina and Chile and the other below the range, is not overly indicative of an effect of party-system collapse. Hence, for present purposes, this indicator does not receive additional attention. Unfortunately, neither of the last two indicators in Table 8.1 is available for Peruvian respondents. Consideration of these measures of political efficacy is therefore limited to a comparison among Argentina, Chile, and Venezuela. This restriction creates difficulties for interpreting the results; it becomes particularly problematic to decide whether unusual findings in Venezuela are results of party-system collapse or of other distinctive patterns in Venezuelan public opinion. More intensive analysis below attempts to find comparisons that at least partially ameliorate these restrictions. First, however, it will be useful to simply compare the three available country averages. The measure of internal political efficacy used in this analysis is a survey question that asks whether the respondent feels that politics is too complicated for a person like them to understand. In all three countries, about half of respondents agree with that assertion; while slightly more Venezuelans in the sample agree, the difference is not large enough to be taken seriously. Hence, a great many citizens of these South American countries—indeed, possibly a majority—feel overwhelmed by the informational and cognitive demands of the political world. Whether this is because of poor political communication, low education or cognitive ability, lack of interest in politics, or some mixture of the above is unclear. In any case, while the finding that a majority of respondents in all three countries feel unable to understand the political world may seem surprising, these results show a more internally efficacious public than that captured by survey research in the United States. In the ANES data for 1952 through 2000, agreement with this item ranges from a low of 58.6 percent in 1960 to a high of 73.6 percent in 1972. That is to say, in comparison with data for the United States, citizens in all three countries examined here have relatively high average internal efficacy. This general claim remains true even if the ANES data is restricted to large cities, to enhance comparability with the capital-city-only sample available for the South American countries.6

226

collapse and the experience of politics

Discussion of absolute levels notwithstanding, the central point here is that there is no compelling initial evidence of an effect of party-system collapse on internal efficacy. Finally, this analysis takes advantage of a measure of external political efficacy that asks how much influence the respondent thinks people like her can have over government decisions at the municipal, regional, or national levels.7 Across all three countries, majorities of respondents feel that people like them have little if any potential to influence government decisions. However, citizens in Argentina (almost 70 percent reporting little or no external efficacy) and Chile (where about 85 percent feel they have limited capacity to influence the government) are much more skeptical about their potential influence than are citizens in Venezuela, where a bare majority reports being essentially unable to affect government decision-making. This difference is large, and it substantively seems as if it could be connected with either of two aspects of party-system collapse: the experience of watching citizen action fundamentally change the political system, or the post-collapse experience of Chávez’s mobilizing rhetoric of participatory democracy and inclusive redistributive institutions. Further discussion in the next section attempts to parse out these two possibilities, while also adopting some comparisons over time to protect against the possibility that Venezuelans have long had higher external efficacy than their Argentine or Chilean counterparts. Let us now turn to citizens’ ideas about how much government, as opposed to corporations or the individual or collective efforts of society, ought to be involved in attempting to resolve various social problems. The findings reported in Table 8.2 show a consistent and at least somewhat surprising pattern. For every problem area, citizens in both Peru and Venezuela are always

ta b l e 8 . 2 . Citizens’ Views of Government’s Role in Resolving Problems Percentage Who Say Government Should Take the Lead in...

Argentina

Chile

Peru

Venezuela

Managing pollution

74.9

56.3

44.8

48.7

Helping citizens meet basic needs

86.6

74.1

62.0

62.9

Reducing violence and crime

91.1

88.6

64.7

85.9

Providing good public services

86.8

84.9

64.2

76.5

Ensuring access to and quality of health care

89.4

88.5

56.1

81.1

collapse and the experience of politics

227

at least marginally less likely to say that government ought to take the lead in resolving that problem than are citizens in either Argentina or Chile. Regarding problems of pollution, three quarters of Argentine respondents say the government should take the lead, and over half of Chilean respondents agree. By contrast, less than half of Peruvian and Venezuelan respondents feel that the government should have the major role in addressing pollution. The differences between citizens’ attitudes on this point in Chile and in the partysystem collapse countries are in the rough neighborhood of 10 percent and are highly statistically significant.8 One might suppose that this difference arises because Chileans, and especially Argentines are far more concerned about pollution than are Peruvians or Venezuelans, but such is not the case. In a set of questions asking respondents to state the three most pressing problems in their societies, about 5 percent of Argentine and about 14 percent of Chilean respondents mentioned pollution. In the same battery of questions, about 7 percent of Venezuelan and about 16 percent of Peruvian respondents named pollution as one of the three biggest problems in their society. Hence, the cross-national difference in the degree to which government ought to take the lead in resolving pollution is probably not due to cross-national variation in the salience of pollution problems. The difference between citizens in collapse and non-collapse countries in the proportion who believe that the government should take the lead in ensuring that all citizens can meet their basic needs is, if anything, somewhat larger. For this problem area, three quarters of Chilean and over 80 percent of Argentine respondents feel that the government should take the lead, while just over 60 percent of Peruvian and Venezuelan respondents agree. As with pollution, these differences are highly statistically significant. For the problem area of meeting citizens’ basic needs, the cross-national difference in attributed government responsibility is accompanied by a cross-national difference in problem salience, but not, perhaps, in the way one might expect. Peruvian and Venezuelan citizens are more likely to list this as one of the three biggest problems in society than are Argentines or Chileans.9 One might well presume that higher problem salience would lead to a greater attribution of government responsibility in the problem area. Yet, regarding meeting basic needs, the two post-collapse countries have higher problem salience accompanied by lower attributed government responsibility.

228

collapse and the experience of politics

Regarding the reduction of violence and crime, citizens in three of the four countries are nearly unanimous in assigning primary responsibility to the government. Nine in ten Argentine respondents assign the government primary responsibility for reducing violence and crime, as do a nearly equal proportion of Chilean respondents. Eighty-six percent of Venezuelan respondents likewise see the government as most responsible in this problem domain, while a mere 65 percent of Peruvians agree. While the difference between Chilean and Venezuelan proportions here is statistically significant,10 it is substantively rather narrow. The differences between Peru and the other three countries, by contrast, are highly significant and substantively quite large. Perhaps the most important point is that the overall pattern, in which the party-system-collapse countries assign less responsibility to the state across a wide collection of problem areas, is not violated. The results for Peru clearly fit that pattern, and the Venezuelan results also weakly fit with that generalization. These results hold true even though the salience of violence and crime as an overall problem for society is essentially identical across these four countries.11 Hence, as before, there is a contrast (if not as clear as for other problem areas) between the post-collapse countries and the other two countries in terms of how much role citizens feel the state should play in resolving a major problem. That contrast, once again, is not neatly accounted for by differences in problem salience across the countries in question. Substantial differences also arise regarding the role citizens desire the state to play in providing good public services, a problem area for which about 85 percent of Argentine and Chilean citizens want the state to take the lead, but for which only about 64 percent of Peruvian and 77 percent of Venezuelan citizens prefer the state to take primary responsibility. The differences between citizens of party-system-collapse countries and those in the other two South American democracies under analysis are large and highly statistically significant.12 As with making sure that citizens’ basic needs are met, this difference in which Peruvian and Venezuelan citizens want a smaller state role in providing public services corresponds with a pattern of problem salience in which those same citizens see public services as a bigger problem for society than do their Argentine and Chilean counterparts. Nine percent of Venezuelan and 11 percent of Peruvian respondents see this problem area as one of the three most important for their countries, while only 2 percent of Argentine and 3 percent of Chilean respon-

collapse and the experience of politics

229

dents have a similar perception. Hence, it would seem that, in the post-collapse countries, there is greater demand for improving public services than there is in Argentina and Chile—yet there is less demand for the state to play a major part in making the desired improvements. Finally, regarding the state’s role in ensuring that everyone has access to high-quality health care, the familiar pattern repeats. Respondents in both Argentina and Chile are close to unanimous in assigning the lead role for this problem area to the state: roughly nine in ten respondents give this answer. In Venezuela, while substantial numbers of citizens want the state to take the lead (80 percent), the figures are markedly lower than in the non-collapse countries. The results show sharply less enthusiasm for a state role in this problem area among Peruvian citizens: only about 56 percent want the state to take the lead. Majorities in all four countries see this as an agenda item for the state, to be sure, but the majorities are clearly larger among Argentines and Chileans than among citizens of the post-collapse countries. The contrast between Chile and the post-collapse countries may reflect a higher problem salience among Chileans; 36 percent of respondents in that country see universal access to highquality health care as one of the three biggest problems in the country, while only 22 percent of Venezuelans and 25 percent of Peruvians agree. However, this problem’s salience level in Argentina is essentially comparable with those of the post-collapse countries: 24 percent of Argentines mention health issues as one of the three biggest problems for the country. Hence, as with other problem areas, the contrast between the post-collapse countries and the other two South American nations in terms of respondents’ desired scope for state action is at best imperfectly explained by reference to problem salience. Overall, we see a persistent and somewhat surprising pattern: in both of the countries that experienced party-system collapse, citizens are somewhat more hesitant to call for a leading state role in any of a variety of problem areas than are citizens of countries that did not go through a collapse. Peruvian citizens are especially unlikely to assign the state a leading role in these domains, but even Venezuelans—citizens led by perhaps the most activist and expansionist government in early twenty-first-century Latin America—are less favorable toward state-led action than are Argentines or Chileans. Theoretically, this result seems difficult to explain as an effect of learning based on citizens’ success in changing the political world. Presumably, such

230

collapse and the experience of politics

learning would lead citizens to expect more government responsiveness, not less. After all, such a citizenry now has a credible threat to make against disliked governments. They have overthrown party systems led by unresponsive and unsuccessful leaders in the past, and presumably would be willing and able to do so again. Hence, if this aspect of party-system collapse were relevant to citizens’ views of what problem areas ought to be on the state’s agenda, one would expect the result to be a more expansive sense of that agenda, not a narrower view of the proper scope for state action. Part of the observed pattern may be due to differences in post-collapse political leadership, mobilization, and competition. Certainly Venezuelan politics during the Chávez era, in comparison with Peruvian politics through the Fujimori period and the first years of the Toledo administration, has provided voters with more messages emphasizing the legitimacy and even necessity of state action in a variety of problem areas. It would thus be entirely plausible to attribute some or all of the differences in which Venezuelans are more supportive of an expansive state role than Peruvians to events since party-system collapse. Yet Venezuelan political discourse during this period has quite probably been more statist than Argentine or (perhaps especially) Chilean discourse during the same years; a pure emphasis on this aspect of political life would lead to an empirically disconfirmed expectation that Venezuelans would be more likely to demand a large state role than Argentines or Chileans. In any case, while this kind of an explanation helps explain differences among post-collapse countries, it probably does not adequately account for the persistent gap between postcollapse countries taken as a group and the other two countries. Instead, it seems plausible that, in conjunction with the patterns of elite discourse considered earlier, the experience of state failure during the process of party-system collapse—the persistent sense that there is too much corruption and not enough representation or responsiveness discussed in previous chapters—may explain this relative hesitancy to assign the state a leading role in problem domains. While the fact of collapse and of replacement of the existing political elites may lead citizens to expect better future performance, the lengthy and painful experience of failure during the run-up to collapse, and persistent elite messages regarding prospective state problem-solving incapacity, may generate a very negative baseline expectation about how well the state will perform in any given problem domain. If so, then the crisis of representation

collapse and the experience of politics

231

during party-system collapse may have created a meaningful legacy of negative attitudes regarding the state’s overall capacity. It is worth recalling that there is no consistent pattern in which citizens of post-collapse countries retrospectively evaluate state institutions more negatively than people in the non-collapse countries (see again Table 8.2). Citizens in the two post-collapse countries do not consistently see their states as having a worse track record than do citizens of the non-collapse countries. This pattern, in combination with the results about the prospective desired state role in the problem areas discussed here, might appear to be something of an anomaly. Perhaps the retrospective evaluations of state institutions discussed earlier are somewhat malleable and reflect a relatively short time horizon, while respondents’ prospective sense of state capacity driving responses about the state’s problem-area agenda is a longer-term and less easily shifted attitude. Alternatively, respondents may draw on substantially different considerations in forming retrospective evaluations of state problem-solving and in forming prospective desires for state leadership. Additional, conditional comparisons below attempt to add some focus and nuance to this discussion, but our ability to reach firm conclusions in this regard will prove to be relatively limited. Even so, it is worth noting that initial evidence suggests that the political failures of the process of party-system collapse may leave, as a mass-attitudinal legacy, a narrowed sense of the optimum scope for state action. In summary, these initial comparisons broadly track the hypotheses developed earlier. Generally speaking, there are few obvious differences between post-collapse and non-collapse countries on attitudes other than those regarding citizens’ prospective government problem-solving agenda and external political efficacy—the attitudes on which effects of collapse were hypothesized to exist. For a handful of other attitudes, there is some initial evidence of difference between post-collapse and non-collapse countries; in particular, Peruvian and Venezuelan citizens are more negative about how the government has managed the economy, although that difference might be attributable to recent economic trends rather than to party-system collapse. Likewise, citizens of post-collapse countries are more likely to report a party identification, although this difference is much more pronounced for Peru than for Venezuela. Finally, Venezuelans are unusually likely to report that the president has made a meaningful contribution to resolving the country’s biggest problems—an outcome that could reflect

232

collapse and the experience of politics

Hugo Chávez ‘s distinctive charisma or a more general effect of party-system collapse. Overall, the initial picture is one in which the hypothesized differences regarding external efficacy and the government agenda do appear in the data, and, for many or most of the attitudes that were not expected to show effects of party-system collapse, there is little empirical evidence of difference. The online appendix reports similar comparisons conditional on the possible confounders discussed earlier. Broadly speaking, the conditioning strategies do not meaningfully alter the initial results.

8.5 FOCUSING IN ON POLITICAL EFFICACY

The comparisons considered up to this point have supported the theoretical expectation that party-system collapse in Venezuela particularly has led citizens in the direction of a more impressive sense of their ability to understand the political world and make a difference in partisan and policy processes than that possessed by their counterparts in Argentina and Chile. At a qualitative level, this is consistent with the highly mobilized role of segments of the Vene­ zuelan population during the post-collapse period, an interval during which citizens have been involved in referendums, recall elections, massive protests, strikes, and various other modes of political activism from both pro- and antigovernment points of view. Yet there are certain limitations on the analysis of these differences as conducted up to this point. First, are the findings reported so far artifacts of the decision not to condition on party identification or patterns of political participation—factors that may be partially caused by political efficacy but that may also partially cause efficacy? Second, are the higher levels of efficacy in post-collapse countries results of some aspect of party-system collapse, or are they simply long-standing differences between political orientations in Venezuela, on the one hand, and Argentina and Chile, on the other? Third, if some aspect of party-system collapse is in fact causally responsible for this difference, is the relevant component related to post-collapse leadership, as argued above? The conditional differences reported in Table 8.3 provide some insights regarding the first question. As mentioned earlier, there are some reasons to think that efficacy might be built up by experience with the political system. Engagement

collapse and the experience of politics

233

ta b l e 8 . 3 . Conditional differences between collapse and non-collapse countries

on efficacy Peru

Venezuela

Conditional on Dichotomous Party Identification Citizens’ involvement in party politics could help solve biggest problems

0.020 (0.017)

0.215 (0.017)**

Low internal efficacy

– 0.151 (0.040)**

Low external efficacy

– 0.751 (0.038)**

Conditional on Identification with Governing Party Citizens’ involvement in party politics could help solve biggest problems

0.064 (0.017)**

0.217 (0.017)**

Low internal efficacy

– 0.156 (0.040)

Low external efficacy

– 0.757 (0.038)**

Conditional on Political Participation Citizens’ involvement in party politics could help solve biggest problems

– 0.034 (0.029)

0.220 (0.023)**

Low internal efficacy

– 0.200 (0.053)**

Low external efficacy

– 0.827 (0.051)**

Conditional on All Three Citizens’ involvement in party politics could help solve biggest problems

– 0.115 (0.044)**

0.216 (0.025)**

Low internal efficacy

– 0.199 (0.057)**

Low external efficacy

– 0.848 (0.053)**

**( p < 0.01). source: Data from Collier (2003).

with party politics or other forms of participation may give citizens information and experience that helps them better understand the political world, thus increasing internal efficacy. Likewise, such experience may help citizens come to feel that participation can make a difference in the policy process and in social outcomes, either by witnessing successful efforts at political change or through a process of cognitive dissonance reduction (since citizens have participated, they may not want to believe that participation is inherently fruitless, because that would imply misguided effort on their part). Hence, it may be useful to consider differences between the post-collapse and the non-collapse countries, conditional on education, unemployment, and various combinations of party identification and political participation measures.

234

collapse and the experience of politics

Table 8.3 reports estimates employing four such combinations of conditioning variables. Estimates are reported that condition separately on having a party identification, identifying with the governing party, and involvement in a range of participatory acts. A fourth estimate conditions on all these variables simultaneously.13 Do the additional conditioning factors modify the substantive conclusion that party-system collapse in Venezuelan is associated with a rise in political efficacy? Broadly speaking, they do not. For each of the four sets of conditional differences reported in Table 8.3, Venezuelan respondents are substantially more likely to report that they could influence policy-making processes than are similar respondents from non-collapse countries. They are also somewhat more likely to state that citizens’ involvement in partisan politics could help resolve society’s problems, and that politics is not too complicated for them to understand. Thus, the additional conditioning factors explored here make no meaningful difference in the estimated differences between Venezuelan citizens and their non-collapse counterparts. Whether these conditioning factors are primarily important confounding variables, consequences rather than causes of political efficacy, or part of the causal mechanism linking party-system collapse with efficacy, may be somewhat unimportant for present purposes: conditioning on them does not really change the substantive conclusion that Venezuela’s party-system collapse may be associated with a rise in citizens’ sense of capacity to understand and influence the political process. The Peruvian results, by contrast, are inconsistent, with some sets of conditioning factors producing a positive difference on the one efficacy variable available in the Peruvian data (whether citizens’ involvement with party politics could help resolve society’s problems) while other sets of conditioning factors produce negative differences. Overall, these findings suggest that no particular estimated difference should be taken too seriously. If there is a difference between Peruvian citizens and their non-collapse counterparts on this efficacy measure, the difference is small and can easily be conditioned away. The fact that there is a much more evident difference for Venezuela than for Peru points away from, although clearly does not rule out altogether, interpretations of the difference that involve shared aspects of the experience of party-system collapse. Both Peru and Venezuela experienced a pattern of state failure and partisan unrepresentativeness during the period before collapse,

collapse and the experience of politics

235

and both experienced the potentially affirming phenomenon of citizen electoral action fundamentally altering the elite political world. Yet there is no real evidence that Peruvian citizens gained the broad boost in political efficacy that Venezuelan citizens may have received. Such a boost may have occurred but was undermined by events of the Fujimori period. Yet this interpretation seems somewhat strained by the fact that citizen demonstrations played an important role in the demise of the Fujimori regime. Alternatively, it is perhaps most plausible that such a boost occurred in Venezuela because of the mobilizing and empowering rhetoric and institutions of the Chávez period. Concluding that Venezuelans experienced a boost in political efficacy after party-system collapse requires additional evidence beyond the purely crosssectional comparisons undertaken up to this point. Did Venezuelans gain efficacy after the collapse, or have Venezuelan political culture and institutions simply always developed a greater sense of capacity and empowerment in the country’s citizens than in Argentina, Chile, or Peru? The second possibility seems somewhat unlikely, given widespread accounts in the literature of Vene­ zuelans’ co-optation by the traditional party system (e.g., Coppedge 1994) and low rates of political protest (e.g., López Maya et al. 2002) during the period before party-system collapse. Survey data suggest that this skepticism based on the existing country literature is reasonably well grounded. Table 8.4 reports the responses of residents of greater Caracas to the internal and external efficacy questions analyzed above in surveys administered during 1983,14 1993,15 and 2003. These data over time ta b l e 8 . 4 . Efficacy over time in Venezuela Efficacy Measure

Sample Proportion (percent)

1983 Low internal efficacy Low external efficacy

55.9 62.5

1993 Low internal efficacy Low external efficacy

45.6 64.3

2003 Low internal efficacy Low external efficacy

55.0 54.8

236

collapse and the experience of politics

suggest that levels of internal efficacy are somewhat volatile, with 44.1 percent of respondents characterizing themselves as capable of understanding the political world in 1983, 54.4 percent in 1993, and 45.0 percent in 2003. In light of this volatility it seems safe to say in summary that Venezuelans’ views of their own ability to understand politics have not obviously been shifted by the process of party-system collapse. However, regarding external efficacy, the results are quite different. The proportion of respondents who felt that they had a meaningful ability to affect the policy process in 1983 was 37.5 percent—a figure that is somewhat higher than the 2003 results for Argentina (30.8 percent) and especially Chile (14.3 percent), but nonetheless reasonably low. In 1993, the Venezuelan figure was a very similar 35.7 percent. These numbers suggest that Venezuelans have indeed historically had somewhat higher external political efficacy than did Argentines and Chileans in 2003. However, in 2003, after party-system collapse and after several years of a mobilizing Chavista government, fully 45.2 percent of Venezuelans report seeing themselves as having a meaningful capacity to influence government decision-making. While this number falls short of a majority, it is nonetheless higher than past Venezuelan results, and higher than results for other South American countries in 2003. That is to say, there was indeed a change during the period of party-system collapse. Of course, identifying the time period of the change does not identify the cause of the change. One the one hand, it seems perhaps reasonable to attribute this shift in political perceptions to one of the three aspects of party-system collapse that form the focus of this discussion, if only because the various aspects of collapse are the most dramatic events in recent Venezuelan political history. On the other hand, it is always possible that some other event or trend during the same period is responsible for the shift. Furthermore, it would be helpful to find at least initial empirical clues about which aspects of party-system collapse might be most connected with the attitudinal change. One approach to teasing out the aspects of recent Venezuelan experience most responsible for the ten-point shift in external efficacy is to compare identifiers with Hugo Chávez’s governing party and other Venezuelan citizens in terms of efficacy. It is somewhat difficult to imagine that the crisis of representation and government that led to the collapse could be the source of the evident rise in Venezuelans’ external efficacy. However, if that were the case, the

collapse and the experience of politics

237

resulting rise in efficacy ought to be broadly shared across all groups in society that suffered from the difficulties of that period. In particular, there should be relatively little difference between 2003 identifiers with Chávez’s party and nonpartisans; those still identifying with traditional parties or identifying with new opposition parties may have had a less negative experience during the process of collapse and so would be less informative as a comparison group. Finding a substantial difference between identifiers with Chávez’s party and the sample of nonpartisans in terms of efficacy would tend to cast some doubt on the hypothesis that the rise in external efficacy among Venezuelans was due to the crisis of representation leading to the collapse. With regard to the possibility that the rise in efficacy was due to the experience of successful participation in an electoral process that reshaped the elite political world, this factor would also seem to broadly apply to both partisans of Chávez and most nonpartisan respondents. After all, the destruction of the traditional party system was successfully accomplished—and broadly popular, among partisans of Chávez and the uncommitted, as attested by the high approval ratings that Chávez scored in opinion polling during the first several years of his presidency. Both Chávez’s partisans and the majority of nonpartisans who cheered the demise of the traditional party system thus shared the experience of successful participation in politics, and would presumably both experience any resulting rise in efficacy. Even so, this argument seems somewhat weaker than the argument regarding the experience of the crisis preceding party-system collapse, in that the sample of nonpartisans may well include some proportion of individuals who did wish to destroy the traditional party system, but who see the outcome of electing Hugo Chávez as nonetheless being a form of failure. Thus, finding a difference between partisans of Chávez and the sample of nonpartisans will not provide as much evidence against this possibility as against the hypothesis that the change in external efficacy was due to political failures before the collapse. However, discovering evidence of a substantively large difference may nonetheless tend to cast doubt on the hypothesis that the experience of successful participation alone was responsible. Finally, with regard to the hypothesis that the rise in efficacy was due to framing and mobilization by Venezuela’s political leadership in the wake of party-system collapse, a substantial difference in efficacy between identifiers with Chávez and the nonpartisan would provide some degree of empirical support.

238

collapse and the experience of politics

The argument is one of dosage. The closest supporters of Chávez are more frequently and intensely exposed to his ideological messages regarding participatory democracy and the need for citizen participation, and more directly involved in the mobilizing institutions of Chávez’s political movement, than are those who stand at a greater remove. This greater dose of participatory messages and mobilization should lead identifiers with Chávez to have experienced a greater rise in external efficacy than nonidentifiers—if this post-collapse treatment is in fact effective in increasing political efficacy. Since receiving and accepting explicit participatory messages is a process requiring less analytic effort, and probably subject to less interpersonal variation in interpretation than would be processes of reception connected with the pre-collapse crisis or the successful participatory acts that produced the collapse itself, it seems reasonable to expect this process to be at least a major component of the overall effect of party-system collapse on Venezuelans’ sense of political efficacy. In fact, empirical results are consistent with this possibility, and at least somewhat anomalous for the hypotheses involving the pre-collapse crisis or the experience of collapse itself. Table 8.5 shows estimates of differences between partisans of Chávez and Venezuelan nonpartisans on the three efficacy variables, conditional on various combinations of education and employment, and the collection of participatory acts. Regardless of the conditioning strategy, there is no difference in internal efficacy. There is a small but persistent difference in attitudes toward party politics: partisans of Chávez, on average, think citizen involvement in party politics is somewhat more effective in resolving society’s major problems than nonpartisans think it could be. This difference is likely not to be terribly meaningful, reflecting the various factors that differentially attract some individuals to involvement partisan politics more than the effects of party-system collapse. However, for external efficacy, there is a substantial difference that is not easily eliminated by conditioning. The difference here is somewhat smaller than the differences between Venezuelans and comparable individuals from Argentina and Chile—but not overwhelmingly smaller. In any case, it is large enough that it seems somewhat problematic for the hypothesis that the experience of successful participation during the process of collapse drives the increase in efficacy, and especially anomalous for the idea that the pre-collapse crisis is the relevant factor. For either of

collapse and the experience of politics

239

ta b l e   8 . 5 . Conditional differences between Chavistas and

non-Chavistas on efficacy Chavistas Conditional on Education and Unemployment Citizens’ involvement in party politics could help solve biggest problems

0.199 (0.030)**

Internal efficacy

0.119 (0.071)

External efficacy

0.657 (0.068)**

Conditional on Political Participation Citizens’ involvement in party politics could help solve biggest problems

0.193 (0.040)**

Internal efficacy

0.146 (0.097)

External efficacy

0.509 (0.104)**

Conditional on All Three Citizens’ involvement in party politics could help solve biggest problems

0.168** (0.044)

Internal efficacy

0.214 (0.101)*

External efficacy

0.594 (0.107)**

*( p < 0.05) **( p < 0.01). source: Data from Collier (2003).

these accounts, it is unclear that such a large difference would be expected between identifiers with Chávez and nonpartisan respondents. Such a large difference is expected, however, if the rise in efficacy is due in large part to participatory messages and mobilization by the main elite beneficiary of party-system collapse, Hugo Chávez. It is, of course, possible that the differences reported in Table 8.5 reflect higher efficacy levels associated with partisanship, rather than effects of party-system collapse. After all, it is impossible to condition on having a party identification when comparing identifiers with non-identifiers; party identification in this comparison is perfectly confounded with the dosage of Chavista mobilizing rhetoric and institutions to which each individual is likely to have been exposed. To address this possibility, Table 8.6 reports equivalent conditional differences, but with identifiers with anti-Chávez parties as the treatment group in place of identifiers with Chávez ‘s political movement. The estimated differences here are quite a bit smaller than those involving identifiers with Chávez’s political movement, and those for external efficacy

240

collapse and the experience of politics

ta b l e 8 . 6 . Conditional differences between anti-Chavistas

and ­non‑Chavistas on efficacy Anti-Chavistas Conditional on Education and Unemployment Citizens’ involvement in party politics could help solve biggest problems

0.136 (0.035)**

Internal efficacy

0.236 (0.085)**

External efficacy

0.333 (0.083)**

Conditional on Political Participation Citizens’ involvement in party politics could help solve biggest problems

0.112 (0.047)**

Internal efficacy

0.434 (0.110)**

External efficacy

0.230 (0.109)*

Conditional on All Three Citizens’ involvement in party politics could help solve biggest problems

0.127 (0.046)**

Internal efficacy

0.378 (0.110) *

External efficacy

0.144 (0.112)

*( p < 0.05) **(p < 0.01). source: Data from Collier (2003).

are also substantially easier to condition away. This suggests that the previous results are not merely effects of having a party identification, but may involve differences stemming from the participatory and mobilizing emphasis of Chávez’s political movement. In general, the results of this suggestion are consistent with the theoretical expectation that, in Venezuela, party-system collapse created an opening for new political discourse that increased citizens’ external political efficacy. The empirical data considered here have shown that Venezuelans are more likely than Argentines or Chileans to agree that citizen involvement in party politics could help resolve society’s biggest problems, and more likely to believe that they have the ability to influence policy-making processes. This contrast holds up for various conditioning strategies. Venezuelans in the wake of partysystem collapse also have higher external efficacy than Venezuelans before the collapse, and supporters of Chávez’s governing party have higher efficacy than nonpartisans. Direct measurement of exposure to Chávez’s rhetoric of partici-

collapse and the experience of politics

241

patory democracy would enable more fine-grained observational comparisons, and experimental work on this subject would provide a much firmer basis for causal inference. Until such additional evidence becomes available, however, we may reach a tentative conclusion that the new elite discourse which filled the political space opened by party-system collapse has produced a marked increase in citizens’ confidence that they have the capacity to influence government decisions and help resolve society’s problems.

8.6 CONCLUSIONS

The available data support the hypothesis that negative elite messages regarding state problem-solving capacity during the process of party-system collapse produced a marked decline in citizens’ prospective sense of how active a role the state should take in addressing society’s biggest challenges. Likewise, the data are consistent with the argument that new participatory messages in the wake of Venezuela’s party-system collapse have produced an increase in citizens’ sense of their own capacity to make a difference in the political world. Many of the other attitudes considered above either show no clear initial evidence of an effect of party-system collapse or can plausibly be accounted for by events and patterns other than those directly connected with the collapse. If this chapter’s interpretation of these findings is correct, then citizens’ process of interpreting the long-term implications of party-system collapse for how they experience government involves a substantial degree of elite leadership. The event of party-system collapse itself demonstrated the limits of elites’ ability to craft mass beliefs; as shown in Chapter 5, most citizens did not follow the ideological movement of traditional party leaders toward the center-right in Venezuela or toward both political extremes in Peru. Yet citizens may well have relied largely on available elite discourses in deciding how to think and feel about the events of party-system collapse—demonstrating the reliance of mass opinion on elite leadership. Hence, the overall relationship between political elites and the mass of citizens during the process of party-system collapse is better characterized as an ongoing dialogue than as a causal flow from one group toward the other. Several decades ago, E. E. Schattschneider declared that “Democracy is unthinkable save in terms of parties” (1942: 1). More recently, Aldrich has re-

242

collapse and the experience of politics

vised this claim, making the somewhat more modest assertion that “democracy is unworkable save in terms of parties” (1995: 3). Whether unthinkable or unworkable, democracy without parties emerged as a credible political project in both Peru and Venezuela during the 1980s and 1990s. For those who share Schattschneider and Aldrich’s belief that partisan democracy is best, the episodes of party-system collapse discussed above are surely dispiriting moments of institutional failure. This book has explored the causes of this kind of failure from the perspectives of both voters and party leaders, using a variety of data sources, research designs, and analytic approaches. Corruption scandals and, especially, problems of ideological representation are central to the narrative of collapse that both motivates and derives support from these analyses. Party systems collapse when the parties within them are tarnished by corruption scandals, and when large numbers of voters find that their preferences and perspectives diverge sharply from those advocated by the established parties as a group. These kinds of concerns motivate votes against the party system because they produce anger—which in turn reduces voters’ aversion to risk and thereby facilitates defection from the established parties. Concerns such as those related to economic performance do not produce party-system collapse because they bring about more fear than anger, pushing citizens to adopt the risk-averse strategy of punishing the incumbent party by voting for other parties within the established party system. Parties, for their part, become vulnerable to corruption scandals when they fail to clearly attach their brand to patronage distribution. Finally, parties become ideologically inflexible—making more likely the kinds of representational gaps that help cause party-system collapse—when they are internally homogeneous in ideological terms, and when their membership and outreach organizational apparatus is large and complex. The drastic collapse that results when these causal conditions coincide has consequences for politics at both the elite and the mass level. This argument, as developed in the preceding chapters, speaks to a number of debates in the social sciences. First and foremost, this volume has implications for understanding party-system change. As discussed in Chapter 1, the existing literature has proposed a broad menu of possible causes of party-system change. This examination of an especially acute form of change has developed theoretical and empirical tools for differentiating between factors that make a minor causal

collapse and the experience of politics

243

contribution and those that drive an episode of change. By combining behavioral analysis of mass survey data, psychological theory and experimentation, and attention to parties’ incentives and organizational patterns, it is possible to sift the central causes of transformation from the other major positive and negative events and trends existing concurrently with party-system dynamics. The theory tested above has substantive implications for other modes of party-system change as well. Economic voting, on this account, should play a causal role that is inversely related to the scale of party-system change. The reason is that deeper changes inherently involve greater risks. Since negative economic perceptions are connected primarily with anxiety and fear, which in turn reduce citizens’ tolerance for risk, voters worried about the economy should be expected to seek the smallest form of change consistent with the expression of their discontent. By contrast, corruption, problems of ideological representation, dereliction of duty, gross incompetence, and similar situations in which political leaders harm the country by palpably failing to meet their obligations should be causally connected with larger transformations of party systems. In a different vein, this volume provides an example of how deductive microfoundations for comparative arguments can be constructed on findings from social and political psychology, rather than relying purely on the streamlined model of human decision-making that drives rational-choice theory. Throughout this volume, deductive implications of voter decision-making during partysystem collapse are derived from psychological theory regarding the causal linkages between affect and risk acceptance, in conjunction with an informal rational-choice model of preferences. This small investment in psychological realism pays dividends in explaining why some kinds of negative perceptions drive party-system collapse while others merely motivate electoral turnover within the established party system. As comparativists continue to emphasize mechanism and microfoundations in their research (Elster 1989, 2007; Hedstrom and Swedberg 1998; Gerring 2007), it will become increasingly important for scholars to explore new and potentially highly productive theoretical sources of microfoundations for macro-level explanation. Finally, and perhaps most centrally, the argument developed here has implications for debates about representation. Scholars in recent decades have argued that constrained patterns of representation, in which party systems adopt a tacit or explicit bargain to exclude some ideological options from the national political

244

collapse and the experience of politics

discourse, enhance the prospects for economic performance (Conaghan, Malloy, and Abugattas 1990; Haggard and Kaufmann 1992) and democratic consolidation (O’Donnell, Schmitter, and Whitehead 1986). This study has shown that such representational constraint may also, in the right circumstances, destroy party systems and create institutional chaos. In the long run, there may simply be no substitute for responsiveness and representation. If the established party system will not provide these public goods, then outsiders will.

NOTE S

chapter 1 1.  For the purposes of this study, a party-system collapse is an event in which all the traditional parties that constitute a party system simultaneously fall to electoral near-irrelevance over a relatively short time period. A careful conceptual definition is developed in Chapter 2. 2.  For overviews of the early years of Venezuela’s experience under Chávez, see McCoy and Myers (2004) and Ellner and Hellinger (2003). 3.  For general discussions of politics during the Fujimori period in Peru, see Cameron and Mauceri (1997) and Degregori (2000). 4.  For biographical overviews of Chávez, see Marcano and Tyszka (2007) and Jones (2008). Few biographies of Fujimori exist, but some aspects of his personal history as connected with his period in government are discussed in Conaghan (2006) and Carrion (2006). 5.  Party-system changes in Bolivia during the 1990s and 2000s are arguably of comparable magnitude. 6.  For an insightful discussion of how these various factors converge to inhibit the success of presidential candidates from outside of the established party system in the United States, see Abramson, Aldrich, Paolino, and Rohde (1995). 7.  For a similar hypothesis regarding the electoral causes of party-system collapse in Venezuela, see Levitsky (2003: 235–36). 8.  Some of Tanaka’s descriptive and conceptual decisions contribute to this omission. Regarding Peru, for example, Tanaka classifies Mario Vargas Llosa’s Movimiento Libertad as a traditional party because this outsider movement incorporated two traditional centerright parties as junior partners in an alliance. Such a classification in turn leads Tanaka to conclude that the electoral strength of Peru’s traditional parties had not declined very much by 1990. This seems misleading because (a) the candidate for Movimiento Libertad was Vargas Llosa, an individual conspicuously lacking in ties with traditional party politics;

246

notes to chapters 1 and 2

(b) the two traditional parties in question had lost a great deal of electoral appeal by the 1990 elections and may not have contributed much to the electoral fortunes of the alliance; and (c) traditional-party politicians were not very successful in their efforts to influence the campaign strategy and message of the Movimiento Libertad presidential campaign. If, as seems more plausible, Movimiento Libertad is regarded as an outsider movement, then it is clear that Peru’s party system had been substantially weakened by 1990, when the first two finishers in the presidential campaign were outsider politicians. 9.  Morgenstern and Zechmeister’s analysis suggests that there may be a relatively large number of voters who are risk-acceptant. This point is in need of further consideration. Two aspects of Morgenstern and Zechmeister’s analysis cloud our ability to draw conclusions regarding the extent of relatively high risk thresholds among electorates. First, the measure used is dichotomous, providing no information about degree of risk acceptance. Second, the survey measure, asking whether it is better to rely on “the devil you know” rather than the “saint you don’t,” is likely to be contaminated by party evaluations and vote intentions. Asking such a question in the context of a very political survey inevitably provides a political coloring to the text, and in the Mexican politics of the time, “the devil you know” can really only mean politicians from the long-incumbent Institutional Revolutionary Party (PRI). Hence, respondents may have answered this question in favor of the unknown saints specifically because they disliked the PRI and intended to vote against it. Morgenstern and Zechmeister offer a range of counterarguments to this possibility, but none is particularly powerful. The one empirical test used to evaluate the possibility that people think of political parties when responding to this aphorism in the context of a political survey involves the correlation between risk acceptance and campaign information exposure among PRI supporters. The correlation turns out to be nearly zero. However, it is unclear how much attention voters would need to pay to the 1997 political campaign in order to realize that the PRI had long been the incumbent party; as a result, it is unclear that this test is decisive. While this central survey measure needs to be improved, Morgenstern and Zechmeister’s theoretical argument regarding the importance of risk acceptance is insightful. 10.  Such a movement from the governing party to a traditional opposition party is the standard hypothesis in literature on economic voting (see, e.g., Fiorina 1981; Lewis-Beck 1988; Lewis-Beck and Paldam 2000). 11.  This appeal to emotion as a source of theoretical insight is intended to supplement, rather than challenge, the informal rational-choice perspective that forms the basis for this study’s approach to understanding party-system collapse. The idea that emotional considerations can provide a foundation for predicting which kinds of preferences might matter most in voters’ decisions does not imply that those decisions fail to be strategic and goaldriven—the criteria for what is sometimes described as “thin” rationality. chapter 2 1.  The Partido Popular Cristiano (PPC), founded in the late 1960s, is a fourth Peru­ vian organization with a possible claim to traditional-party status. While this party was roughly two decades from its founding at the time of the party-system collapse, and thus might chronologically qualify as a traditional party, it had never demonstrated the substantial electoral and governing relevance necessary to meet the second requirement. The party generally won roughly 10 percent of the vote in legislative elections during the party system of the 1980s, as well as in the presidential elections of 1980 and 1985. In spite of this relatively marginal electoral performance, the PPC has something of a claim to traditional-

notes to chapter 2

247

party status due to its importance during the 1978 constitutional convention (where it held 25 of 100 seats, probably assisted by the fact that Acción Popular presented no candidates for the convention elections) and its role as junior coalition partner during the 1980–85 Belaúnde administration. These considerations notwithstanding, the party’s persistent pattern of winning roughly 10 percent of the vote—a pattern that held during the 1985 legislative and presidential elections, even after the party’s relative prominence in the constitutional convention and the Belaúnde government—argues against regarding the PPC as traditional. This decision plays relatively little role in the discussion that follows; reaching the opposite conclusion would change few of the argument and findings presented in this study, given that the PPC’s electoral strategies broadly mirrored those of Acción Popular during the period of interest. 2.  COPEI won 22.4 percent of the presidential vote in 1947, 16.2 percent in 1958, and 20.2 percent in 1963 (Kornblith and Levine 1995: 49). 3.  For an excellent study of the evolution, ideologies, and changing strategy of Movi­ miento al Socialismo, see Ellner (1988). 4.  Considering total vote fragmentation rather than power to influence governing majorities, McGuire concludes that “Argentina would fall somewhere between the ‘moderate multiparty’ and the ‘two-party’ type” (1995: 226). This conclusion, while legitimate in light of McGuire’s decision to include all parties, draws attention toward several minor parties with a relatively weak societal basis. Because it is theoretically unsurprising when electorally marginal, societally weak parties collapse, the present study does not focus on such organizations. 5.  This system will tend to produce three meaningful candidates in the first round (Cox 1997: 123–38); however, there is a clear, institutionally imposed difference in relevance between the second- and third-place finishers. 6.  The characterization of specific elections in the United States as realigning, as well as causal claims used to explain realigning elections, have been vigorously disputed (Mayhew 2002). However, the term may remain useful in a descriptive sense. 7.  This usage corresponds with the twentieth-century elections in the United States that are most often described as realigning, although not with some of the nineteenthcentury elections. 8.  The missing data for Peru between 1963 and 1980 reflects the fact that, during this period, Peru had no presidential elections and through most of this period was, in fact, controlled by a military government. The explanation for much of the missing Argentine data is the same; Argentina had several different kinds of authoritarian governments during the period reflected in Figure 2.1. An essentially democratic period from 1973 to 1976 included a presidential election in 1973 in which both of the traditional Argentine parties were allowed to compete. A single election does not show up well in a line graph; however, the 1973 election was dominated by the Peronists and the Radicals. 9.  This discussion closely follows the descriptions of the Caracazo by Míriam Kornblith (Kornblith 1998: 1–32) and Margarita López Maya et al. (López Maya, Smilde, and Stephany 2002: 16–18). 10.  A plot of protest episodes per year can be found in the online appendix to this volume. 11.  Information about the total number of politically motivated deaths in Peru, as well as about the division between killings by insurgent groups and by state forces, can be found in the online appendix to this volume. 12.  For overviews of the Sendero war in Peru, see McClintock (1998), Stern (1998), and Gorriti (1999).

248

notes to chapters 3. 4, and 5

chapter 3 1.  Basic descriptive data on Peru’s economy during the 1980s and 1990s can be found in a series of figures in the online appendix. 2.  All figures are in U.S. dollars unless otherwise indicated. 3.  Figures presenting basic descriptive data about the Venezuelan economy during the 1980s, 1990s, and early 2000s can be found in the online appendix. 4.  Data on economic growth in this chapter are taken from the Penn World Table 6.2 (Heston, Summers, and Aten 2006). Inflation and unemployment data are drawn from the World Development Indicators (World Bank 2009). 5.  Discussion of some technical details of the models and estimation process behind the results discussed below is available in the online appendix. 6.  The GDP variable is drawn from the Penn World Table (Heston, Summers, and Aten 2006). 7.  Unemployment is also a relevant economic variable, and is considered in the second model reported in Table 3.2. However, no unemployment data are available for Peru in the period from 1980 to 1985, so including this variable requires the loss of one of the crucial elections for party-system collapse. Unemployment is therefore not a part of the standard collection of economic variables. 8.  The full results of this analysis are available in the online appendix. 9.  The full results of this analysis are available in the online appendix. chapter 4 1.  Basic descriptive data on Peru’s economy during the 1980s and 1990s can be found in a series of figures in the online appendix. 2.  All figures are in U.S. dollars unless otherwise indicated. 3.  Figures presenting basic descriptive data about the Venezuelan economy during the 1980s, 1990s, and early 2000s can be found in the online appendix. 4.  Data on economic growth in this chapter are taken from the Penn World Table 6.2 (Heston, Summers, and Aten 2006). Inflation and unemployment data are drawn from the World Development Indicators (World Bank 2009). 5.  Discussion of some technical details of the models and estimation process behind the results discussed below is available in the online appendix. 6.  The GDP variable is drawn from the Penn World Table (Heston, Summers, and Aten 2006). 7.  Unemployment is also a relevant economic variable, and is considered in the second model reported in Table 3.2. However, no unemployment data are available for Peru in the period from 1980 to 1985, so including this variable requires the loss of one of the crucial elections for party-system collapse. Unemployment is therefore not a part of the standard collection of economic variables. 8.  The full results of this analysis are available in the online appendix. 9.  The full results of this analysis are available in the online appendix. chapter 5 1.  As has been discussed earlier, the Argentine party system has become more personalized and complex during the 2000s, transformations whose explanation is beyond the scope of this book. 2.  More details regarding this analysis are available in the appendix to this chapter. Lists

notes to chapter 5

249

of the specific variables used, as well as survey question texts, are available upon request from the author. 3.  Full model results for these analyses, as well as some technical details involving estimating, are available in the online appendix to this volume. 4.  Data from the Apoyo, S.A. survey of Lima, June 1985, discussed in that month’s ­Informe de Opinión (Lima: Peru). 5.  This finding parallels, at the national level, the result from survey research in developed countries that the meaning of “left” and “right” varies across age and social groups (Inglehart 1984; Fuchs and Klingemann 1989). More fine-grained analysis would doubtless uncover heterogeneity within each country as well. However, for present purposes, it is sufficient that the left-right dimension be accepted as having some grounding in specific issues and attitudes. 6.  In the aftermath of the elections, La Causa R made serious allegations that Caldera’s victory was due in significant part to electoral fraud; this issue cannot be resolved in the present analysis but nonetheless deserves mention. 7.  The multinomial logit model requires the assumption of independence of irrelevant alternatives (IIA): the probability of choosing option A rather than option B must be the same regardless of whether option C is available. The Hausman test statistic (Hausman and McFadden 1984) for the model reported in the text above, omitting votes for Andrés Velázquez, is 8.21 on 33 degrees of freedom. Thus there is no evidence that the IIA assumption meaningfully affects the results presented. The model reported in the text contains no interaction terms. Two interactions appear to be significant in an analysis of deviance test: between rightism and concern about the economy, and between leftism and identifying with COPEI. However, these interactions must be excluded because they involve empty categories and therefore produce bias in the estimation; if one case is randomly reassigned to eliminate the bias, neither interaction retains statistical significance. Because all of the independent variables are dichotomous, no other possible nonlinearities arise. 8.  AD identification in the simulation condition has a relative risk of 0.02 for La Causa R with a standard error of 0.02; the equivalent ratio for Convergencia is 0.03 with a standard error of 0.03. AD identification has a substantially higher but still very small relative risk of 0.24 for COPEI, with a standard error of 0.11. 9.  The online appendix to this volume presents an alternative analysis, with a dichotomous dependent variable representing the choice to vote for or against the traditional parties. 10.  The model reported in the text makes the simplifying assumption that the effects of different levels of the Economic Evaluations variable are linear. This assumption is helpful because of the small number of individuals with positive evaluations of the economy in the sample. Furthermore, assuming linearity seems to be more or less acceptable in light of the data; removing the assumption of linearity consumes 6 additional degrees of freedom. Yet it only reduces the deviance by 18.45. While meeting standards of statistical significance, this reduction in deviance is much smaller than the mean reduction, per degree of freedom, associated with the linear effect. Furthermore, close inspection reveals that most of the additional deviance is accounted for by a somewhat sharper distinction between people who saw the economy as significantly improved, as compared with those who saw it as only somewhat improved. The sixteen respondents who reported that the economy was significantly improved may have been less likely to vote for Salas Romer, and more likely to support Hugo Chávez, than the respondents who saw the economy as only somewhat improved. While probably substantively unimportant, this result slightly weakens the economic-voting

250

notes to chapters 5 and 6

hypothesis. Also, the model reported in the text includes no interaction terms. An analysisof-deviance comparison of the fully two-way interactive model with the single-interaction model discussed above suggests that this simplification is reasonable. Twenty-nine additional meaningful independent variables are involved in the fully interactive model; one interaction term—between working-class identity and COPEI party identification—contained no cases, and another—between ideological leftism and concerns about corruption—had to be omitted because zero individuals in this category voted for AD (randomly reassigning an individual in this category to vote for AD renders it statistically insignificant). Including these twenty-nine variables produces a reduction of 74.81 in the deviance but consumes 84 additional degrees of freedom. Hence, it is empirically acceptable to exclude the secondorder interactions. A limited exploration of higher-order interactions produced similar results: adding the six most plausible three-way interaction terms reduced the unexplained deviance by 12.71 at a cost of 18 degrees of freedom. Because all of the independent variables are dichotomous, no other possible nonlinearities arise. 11.  These findings are generally consistent with those of Weyland (2003), who argues that the economy influenced voters’ decisions regarding MVR primarily through prospective economic expectations based in Chávez’s charisma and hope for the future—and that the retrospective economic evaluations associated with Venezuela’s long crisis had a secondary, and perhaps somewhat smaller, effect by discrediting the established political elite. While these conclusions are consonant with the findings reported here, the coefficients that Weyland reports for both retrospective and prospective voting are substantially larger and more important than the coefficients considered here. The coefficients provided in this study may be preferable because the survey question behind them involves more specifically economic evaluations; the sociotropic survey questions used by Weyland ask about the national “situation” in general rather than in specifically economic terms; likewise, the pocketbook questions refer to the respondent’s perceptions of whether she is able to “live better” in general rather than in strictly economic ways. These broad questions doubtless elicit economic as well as other considerations, and therefore may be useful. However, the explicitly economic measure used in these analyses can be interpreted more straightforwardly. 12.  The online appendix contains a discussion of the assumption of independence of irrelevant alternatives vis-à-vis the above analysis, as well as the substantively convergent results of estimating a multinomial probit model for these data. 13.  These arguments are based on the author’s personal experience in negotiating with various Venezuelan survey firms about the implementation of two surveys, one in 2002 and the second in 2007. 14.  Of course, some of this discrepancy may also be due to the underreporting of abstention in the survey data. 15.  This survey is reported in the March 1992 Informe de Opinión (Lima: Peru). chapter 6 1.  Interview by the author with Gustavo Mata, Caracas, Venezuela, June 12, 2001. 2.  Interview by the author with an anonymous protester in Plaza Francia, Caracas, Venezuela, November 2, 2002. 3.  Conversation observed in Librería Centro Plaza, Los Palos Grandes, Caracas, Vene­ zuela, September 25, 2001. 4.  This reason for anger against APRA and Acción Popular during the 1980s was reported to me in literally dozens of interviews in Peru during January through May 2005. 5.  Interview, name withheld at informant’s request, Lima, Peru, February 3, 2005.

notes to chapter 6

251

6.  A table with the P values for each demographic variable is available in the online appendix to this volume. Permutation methods are likewise used to provide P values for the regression coefficients estimated below. 7.  The question asks, “How do you feel right now? (Mark all that apply)” The respondent is then presented with a list of nine emotional states, including anger and alarm/anxiety. 8.  If the data are partitioned regionally, with separate results for Cuzco and Lima, the estimated causal effect is quite similar for both regions, although outsider voting is substantially higher for all treatment groups in Cuzco than in Lima. Because there is no significant evidence of regional heterogeneity in treatment effects, the regions are pooled in all reported analysis. 9.  There is some debate regarding the desirability of analyzing experiments using multivariate regression analysis. Freedman (2008a, 2008b) shows that such analysis can produce bias in the estimate of the causal effect and can produce incorrect standard errors; Green (2009) argues that in practice the bias is very often negligible. These views suggest that a desirable procedure is to first report simple intent-to-treat results, as is done in this analysis; analysts can then report any desired multivariate regression estimates, with the caveat that some degree of skepticism may legitimately attach to any results that differ substantively from the intent-to-treat findings. In any case, the intent-to-treat results remain the baseline, and as such are the center of attention in this analysis. 10.  A range of pre-treatment variables is available for conditioning, although no adjustments produce major changes in the results. Rosenbaum (2002) suggests adjustment to improve efficiency by regressing the outcome variable of the experiment on the pre-treatment conditioning variables, and then carrying out a difference-in-means test between treatment groups using the residual from that regression as the dependent variable. For that conditioning regression, a variety of specifications were tried. The specification which resulted in the least significant results among those tried is a generalized additive model with vote for the outsider candidate as the dependent variable and age, social class self-identification, frequency of attention to television news, city of residence, party identification, and left-right self-placement as predictor variables. All but the last three are given smoothing splines. For the city of residence and party identification variables, no spline is used because the variables are measured at the nominal level. For left-right self-placement, the attempt to fit a spline creates estimation problems. Because the model is purely an auxiliary to the main task of estimating the effect of the anger treatment on outsider voting, the details of the estimation results are not reported here, although they are available on request from the author. After conditioning using this model, the estimated effect of the anger treatment on voting for an outsider party falls to 8.9 percent, and the P value for this effect rises to 0.067. This effect estimate, while at the margins of statistical significance, is basically consistent with the unconditional effect reported earlier. Thus, as is often the case for experiments, conditioning produces at most marginal changes in the overall findings. 11.  Identifiers with the governing party have a much lower baseline propensity to vote for outsiders in this experimental setup than non-identifiers. Only 21 percent of APRA identifiers in the control group vote for the outsider candidate, as compared with 52 percent of non-APRA identifiers in the control group. However, the important point here is that the difference between treatment and control groups is substantively similar, and statistically indistinguishable, for APRA identifiers and for others. 12.  Each evaluation variable is coded such that larger values correspond with more negative evaluations. Each question asks, “How would you say the government has done with respect to [performance dimension]? Very well, well, poorly, very poorly, or don’t you know?” 13.  Regression results supporting these findings are available in the online appendix.

252

notes to chapters 6 and 7

14.  The inference requires stronger modeling assumptions than those in the previous section because aversion to uncertainty is not directly randomized. In particular, the analysis assumes that, conditional on treatment, there are no missing variable problems in the relationship between attitudes about uncertainty and vote choice; it also entails linearity assumptions and an error model that are not justified by the experimental design alone. 15.  Respondents are given minimal information about Alonso. They are told that he is a 44-year-old native of Arequipa who now lives in Lima. 16.  The scenario text is as follows: “Even though many people in Alonso’s neighborhood don’t have money problems, Alonso is having a hard time finding work. His taxi broke down, and he doesn’t have the money to fix it. He doesn’t know what to do to find work, and he is running out of money.” 17.  “Alonso realizes that the country is having serious economic troubles. The price of food and gas keeps going up, and many of Alonso’s friends can’t find work.” 18.  “The news keeps making a big deal about a series of scandals involving the president and other important politicians; everyone says they stole a lot of money from the country. Alonso doesn’t know what to think about this, but he does know that he often has to pay bribes to the police so that he can keep working with his taxi.” 19.  “It seems to Alonso that the legislators who represent his city don’t listen to the people who elected them. Many of these legislators promised in their campaigns that they would improve the public schools and invest more in public works. But since they were elected, it seems that they only mind their own business and don’t do anything about the schools or public works.” 20.  “Alonso sees that more and more people are carrying guns in his neighborhood, and that some neighbors have started talking about drugs and gangs. A few weeks ago, the body of a man who had been shot turned up in a park a few blocks from Alonso’s house.” 21.  In every case, the question text is: “Which of the following phrases best describe how you think Alonso feels in this situation? Enthusiastic, angry, worried, calm, frightened, or I don’t know?” Respondents are allowed multiple responses. chapter 7 1.  For a summary of the conflict between Acción Popular and Vargas Llosa’s political team, see Cameron (1994: 63–76). The ideological component of the conflict was emphasized by Víctor Andrés García Belaunde (interview by the author, January 19, 2005, Lima, Peru) and Valentín Paniagua (interview by the author, January 21, 2005, Lima, Peru). 2.  Interview by the author with Eduardo Fernández, November 13, 2004. [AU: add location] The conflict between Caldera and Fernández also included a much less ideological struggle over the 1988 COPEI presidential nomination (Martz 1999: 647–48). 3.  For similar theoretical accounts focused on the development of U.S. party organizations, see Aldrich (1983; 1995: 163–93), and Aldrich and McGinnis (1989). 4.  In order to obtain a reasonably representative sample of local party leaders, a random sample of forty local government units was selected within each country, weighted by the number of registered voters on the assumption that the size of the party organization within an area is roughly proportional to the number of local voters. In Venezuela, the local government units selected are municipios. In Peru, they are provincias. In Argentina, they are municipios (in some areas also called partidos). For Argentina, the thinly populated areas of the far north and south were excluded from the sample for practical reasons. For each local area selected, party records (supplemented when incomplete with interviews with national and regional party leaders) were used to determine the identities of the general secretaries,

notes to chapter 7

253

secretaries of organization, and party committee members in each selected unit during the relevant time frame. From these lists, two leaders were randomly chosen. For selected party leaders who were deceased or unlocatable, a replacement was selected. For the parties considered in this chapter, the response rates varied from 64 percent (for leaders from Izquierda Unida in Peru) up to 95 percent (for leaders from the Peronist party in Argentina); between 50 and 80 data points are available for each party. Obviously, a larger sample would be desirable, but the current data set is sufficient to allow many statistically meaningful comparisons. These samples are thus not perfectly representative of party leadership during the period leading up to party-system collapse, but they are reasonably close approximations. This is especially clear when these samples are compared with the valuable but less representative and fundamentally non-random collections of interviews that make up the data base for qualitative studies of party organization. For most of the parties under consideration here, a range of valuable organizational case studies have been conducted. For Acción Democrática, see Martz (1966) and Coppedge (1994). For COPEI, see Herman (1980) and Crisp, Levine, and Molina (2003). For APRA, see Bonilla and Drake (1989) and Graham (1992). I am unaware of any major previous organizational studies on Acción Popular. For Izquierda Unida, see Roberts (1998: 201–68). Perhaps the best study of the organization of the Radical party in Argentina focuses on its nineteenth-century origins: Alonso (2000). Some highlights of the extraordinarily rich literature on the organization of the Peronist party include James (1994), McGuire (1997), and Levitsky (2003). These works rely primarily on historical records and in-depth interviews with party leaders. Such data sources provide perhaps incomparable intensity of insight and detail into the organizational dynamics of individual parties. However, the survey data considered here have two important strengths that are not as clearly present in these other approaches. The sample analyzed here is reasonably representative of the local party leadership as a whole, whereas the interviews behind most case studies instead derive from samples of convenience. In addition, the organizational measures produced from these survey data are reasonably comparable across parties and across national boundaries—because the same questions were asked in each context. Hence, the results of this analysis provide a different angle on organization within each of the parties under consideration. 5.  For a discussion of the costs and benefits to parties of large membership rolls, see Scarrow (1996). 6.  The first question asks, “How many official members of the party were there in this municipality/locality during [the first election]? Were there less than one hundred members, between one hundred and two hundred, between two hundred and five hundred, between five hundred and a thousand, between a thousand and two thousand, or more than two thousand?” A parallel question is used for the second time period. A table summarizing responses to this question in each party of interest is available in the online appendix. 7.  This finding is not due to Venezuela’s having larger local government units in terms of population than the other countries; all three countries have some very large units and some quite small ones. Furthermore, a smaller party in Venezuela, La Causa Radical, never had a mean membership base larger than 100–200 persons, although it operated in the same geographic units as the traditional Venezuelan parties. Hence there is every reason to regard the observed differences as genuine. 8.  The question asks, “Who have been the most important leaders of the party in this municipality/locality since 1980?” 9.  Means for each party are reported in the online appendix; the appendix also reports the percentage of responses that were truncated to ten.

254

notes to chapter 7

10.  The question asks, “What are the positions that you’ve held within the party?” The online appendix reports the average number of jobs listed by local leaders and party workers in each party. 11.  The text of the question is: “And when you decided for the first time to work for the party, how important was it for you to make sure that people listened to your ideas and opinions about politics? Was this the main motive for your decision, an important motive for your decision, not a very important motive, or did it have nothing to do with the decision?” The online appendix contains a table reporting the mean levels of local leaders’ pragmatism according to this measure in each of the parties in the study. 12.  The question reads: “I’d like to know a bit about the normal sources of your party’s budget in this municipality/locality. About what percentage of the budget was gathered through donations or regular contributions from local members and sympathizers of the party? Was it less than ten percent, between eleven and twenty-five percent, between twentysix and fifty percent, between fifty-one and seventy-five percent, between seventy-six and ninety percent, or over ninety-one percent?” The median response for each party is reported in the online appendix; party means would tell a similar story. 13.  The survey question asks, “How was the decision made to nominate the final candidate? Was it through a local convention of the party, through a national convention of the party, through a primary election, through a decision by local party leaders, or through a decision by national party leaders?” Proportions are reported for all parties in the online appendix. 14.  The survey question asks, “How involved were the party’s national leaders in selecting the mayoral candidate? Were they not involved at all, did they give some advice but preferred a candidate other than the one who was chosen, did they give some advice and prefer the candidate that was chosen, or were they able to impose their preferred candidate against the will of others?” Full results are reported in the online appendix. 15.  The question asks, “How much influence would you say that party leaders at the municipal/local level had on party decisions at the national level during [the first period: 1980s in Venezuela and Argentina, 1980–85 in Peru]? Did they not have any influence and only found out about decisions after they had been made, did they have the opportunity to express their opinions but couldn’t affect the decisions, did they have an important informal role in the decisions, or did they have an important formal role in the decisions?” The online appendix reports the mean response in each of the parties. 16.  The survey question measuring quality of information flows reads: “I’d like to find out a little bit about how municipal party leaders let the national party leaders know about local needs. What would you say are the main tools municipal party leaders have for sending information and requests to the national level?” This open-ended question was coded to yield four categories: no means of communication (1), indirect means of communication through multiple layers of party bureaucracy (2), indirect means of communication through only one layer of party bureaucracy (3), and direct means of communication (4). Full results are available in the online appendix. 17.  Results for each party are available in the online appendix. 18.  The survey questions measuring activist and local party leader recruitment via various kinds of particularistic benefits are: “I’m going to read you various reasons that people mention in explaining why they get involved in party politics, and I’d like you to tell me how important each reason was to you when you first decided to get involved in the party. When you first decided to work for the party, how important to you was the opportunity to mix with important people? How important to you was the opportunity to make social connections? How important to you was the opportunity to eventually earn money through

notes to chapter 7

255

the party?” Response options for each question are: “Was this the primary reason for your decision, a major reason for your decision, a minor reason for your decision, or did this not play a role in your decision at all?” Full results are available in the online appendix. 19.  Specifically, the question asks, “How much influence would you say that labor unions had in national party decisions during [the relevant time window]?” Response options are: they had no influence and only found out about decisions after they had been made; they had the opportunity to express their opinions but could not affect decisions; they had an important, informal role in decisions; or they had an important, formal role in decisions. Full results for each party are reported in the online appendix. 20.  The question text is: “In the following questions, I am going to ask you to use a metaphorical ruler to talk about the ideological positions of different individuals and groups. The ruler goes from 1, at the extreme of the political left, to 10, at the extreme of the political right. The number ‘1’ would correspond with the extreme left, like the Sandinistas in Nicaragua or like Che Guevara. The number ‘3’ would correspond with the left, but not the extreme left, like the PRD in Mexico or the PT in Brazil. The numbers ‘5’ and ‘6’ would correspond with the political center, that is, with people who accept ideas both from the left and from the right. The number ‘8’ would correspond with the right, but not the extreme right, for example Vicente Fox in Mexico. The number ‘10’ would correspond with someone on the extreme right, like Augusto Pinochet in Chile. Generally speaking, where would you put your personal point of view about politics on this scale?” 21.  The first question asks, “I’d like to find out about any special organizations that the party has in this municipality/locality. I’m thinking about party organizations for young people, women, the elderly, etc. I’m going to mention various classes of organization, and for each category, I’d like you to indicate whether your party has been involved with that kind of organization in some moment since 1980. For each organization that your party has had, I’m going to ask about the number of people who were involved and the level of activity of that organization in electoral campaigns. First, was there some organization for young people of your party in this municipality/locality?” For localities where a party leader answers in the affirmative for this question, the follow-up asks, “About how many people were involved in these organizations in this municipality/locality during [first election]? Were there less than one hundred members, between one hundred and two hundred, between two hundred and five hundred, between five hundred and a thousand, between a thousand and two thousand, or more than two thousand?” Parallel questions were used for women’s organizations and workers’ organizations. Results for each party are reported in the online appendix. 22.  The question asks, for the electoral cycle just before the beginning of party-system collapse, “For the next few questions, I’m going to ask you to use a scale from 1 to 10 to tell me how strong different party organizations were. In answering these questions, I’d like you to think not about electoral outcomes, but instead to concentrate on strength in terms of organizational resources. A ‘1’ would be an extremely weak party organization, with maybe a few people in the municipality, but no real network of volunteers. A ‘3’ would be stronger but still weak, with an established headquarters and some volunteers but little ability to mobilize a big demonstration or contact a lot of voters. A ‘5’ or a ‘6’ would be a middle level of organization, with an established headquarters, a good number of volunteers, and the ability to get in touch with pretty much everyone in the municipality during the course of the campaign. An ‘8’ would be an even stronger organization, with groups of party members established in different parts of the municipality and regular party meetings. A ‘10’ would be an extremely strong party organization, with strict dues and member lists, party clubs and activities for members, and strong efforts at making sure that every voter in the municipal-

256

notes to chapters 7 and 8

ity is given multiple chances to hear the party’s message from some other individual during each campaign. How would you classify the organizational capacity of [leader’s party] in this municipality/locality during [first period: 1980s in Venezuela and Argentina, 1980–85 in Peru]?” Full results are available in the online appendix. 23.  The questions ask: “Thinking of the party’s official headquarters in this area as well as any unofficial places where the party regularly meets, I’d like to find out about the main activities you do in those places. I’m going to read you a list of different kinds of activities, and for each I’d like you to tell me if the party in this area does them frequently, sometimes, or never. Do you offer social and community activities in the various meeting places of the party? Do you carry out social service projects, like distributing food at reduced prices or providing health care, for members in this area? Political parties often try to help people by using their official houses to provide information about government programs, to help people enroll in different kinds of government aid, or to distribute food or money from government programs to the needy. Has your party tried to help in some of these ways in this municipality/locality?” Results for each party are reported in the online appendix. 24.  This approach of eliminating variables that do not correspond with the pattern on the dependent variable would be similar to Mill’s method of difference (Mill 1891/2002: 255–56). 25.  For elaboration of a theory of causal explanation compatible with that expressed above, see Woodward (2003: 187–238). 26.  Results for each party on this indicator are shown in the online appendix. 27.  Mean levels of change for each party are reported in the online appendix. 28.  However, many other aspects of local distinctiveness remain. To address these, a second version of each model is estimated using dummies for each locality and each party. Such a model can be estimated because there is more than one party per locality; nonetheless, this approach consumes many degrees of freedom and limits the contribution of localities with nonresponse to the substantively interesting part of the model. Hence, it may be reasonable to consider variables that are significant in the model with no local dummies but not the larger model, given that the larger model essentially omits some localities; however, these effects would obviously be regarded as more fragile than those that are significant in both models. Results using locality and party fixed effects are reported and discussed in the online appendix. chapter 8 1.  For an additional discussion of both coups, and of the role of the weakening of the traditional parties in bringing them about, see Tanaka (2006). Tanaka’s discussion differs from this study in its definition of party-system collapse; as a consequence, Tanaka sees the Peruvian collapse as having happened in the wake of Fujimori’s 1992 coup and the Vene­zuelan collapse as having occurred between Chávez’s 1998 electoral victory and the subsequent constitutional convention. This alternative conceptualization of party-system collapse directs Tanaka’s attention away from the dynamics of elite-mass interaction and electoral decline explored in this study and toward the subsequent—and perhaps consequent—intra-elite dynamics of partysystem change. See also Tanaka (1998) for an extended discussion of elite processes during and following Peru’s party-system collapse. 2.  Turnout figures are from the International IDEA Voter Turnout Database, online at http://www.idea.int/vt/. Data accessed on Sept. 20, 2009. 3.  The online appendix discusses some inferential limitations of matching methods in this research context. 4.  Full results for each country on these and subsequent outcomes can be found in the online appendix.

notes to chapter 8

257

5.  If, as seems likely, these evaluations of the state’s management of the economy are heavily dominated by recent events, then Venezuela’s negative evaluations should be seen as reflecting the economic collapse connected with the anti-Chávez capital and oil strike of late 2002 and early 2003. Likewise, Peruvian pessimism about the state’s management of the economy may be connected with that country’s brief recession of 2001. 6.  Among residents of central cities, the ANES data show that agreement with the proposition that politics is too complicated for the respondent to understand runs from a low of 53.9 percent in 1960 to a high of 71.3 percent in 1978. For all but one available year, the ANES data for residents of central cities show lower levels of internal efficacy than that for Venezuela in the present data. 7.  Response options are: “much influence,” “some influence,” “little influence,” or “no influence at all.” The text dichotomizes this measure by grouping “much” and “some” influence together as a positive comparison with the two more negative responses. 8. The P value for a difference in proportions test between Chile and Venezuela is less than 0.00001. 9.  Seven percent of Argentines and 4 percent of Chileans name meeting citizens’ basic needs as one of the three biggest problems, while 17 percent of Venezuelans and 12 percent of Peruvians consider the issue a leading problem for their country. 10. The P value for a difference in proportions test is 0.029. 11.  The proportion of respondents in each country who mention violence and crime as one of the three most important problems for their society are as follows: in Argentina, 63 percent; in Chile, 65 percent; in Peru, 62 percent; and in Venezuela, 61 percent. 12. The P value for a difference in proportions test is less than 0.00001. 13.  For the differences that condition on education, unemployment, and having a party identification, or on identifying with the governing party, perfect matching is possible for both Peru and Venezuela. Perfect matching is not possible for the estimate that conditions on participation, or for the estimate conditional on all three factors. For the Venezuelan differences conditioning on education, unemployment, and political participation, all postmatch means are within 1 percentage point of each other, with three exceptions. For unemployment, the Venezuelan mean is 0.164, while the mean of the post-matching control cases is 0.141. For participation in a political rally or demonstration, the treatment mean is 0.074, while the post-matching control mean is 0.045. Finally, for working for a party, the treatment mean is 0.035 and the control mean is 0.024. When Venezuelan differences condition on participation, having a party identification, and identifying with the government, the resulting matches are predictably less precise on average. Nonetheless, the postmatching treatment and control means for all conditioning variables are within a tolerance of 3 percentage points of each other. For the Peruvian difference conditioning on education, unemployment, and political participation, all post-match means are within 3 percentage points of each other, with the exception of education: 64.9 percent of Peruvian respondents lack a high school diploma, while only 60.5 percent of post-matching control cases fall in the same educational category. The Peruvian matches when conditioning on all three factors are very poor; for example, the education variable shows a post-matching discrepancy of nearly 16 percentage points. Results from this last comparison are not to be taken seriously, as the main text notes. However, the others are likely to be acceptable. 14.  Data for 1983 are drawn from Caracas respondents in the BATOBA survey (Baloyra and Torres 1983). 15.  Data for 1993 are drawn from Caracas respondents in the Carrasquero and Welsch survey (Carrasquero and Welsch 1993).

BIBL IOGR A PH Y

Abramson, Paul R., and Charles W. Ostrom. 1991. “Macropartisanship: An Empirical Reassessment.” American Political Science Review 85: 181–92. Abramson, Paul R., John H. Aldrich, Phil Paolino, and David W. Rohde. 1995. “Third-Party and Independent Candidates in American Politics: Wallace, Anderson, and Perot.” ­Political Science Quarterly 110: 349–67. ———.2000. “Challenges to the American Two-Party System: Evidence from the 1968, 1980, 1992, and 1996 Presidential Elections.” Political Research Quarterly 53: 495–522. Achen, Christopher H. 1989 “Prospective Voting and the Theory of Party Identification.” Presented at the 1989 Annual Meeting of the American Political Science Association, Atlanta. ———.1992. “Social Psychology, Demographic Variables, and Linear Regression: Breaking the Iron Triangle in Voting Research.” Political Behavior 14: 996–1017. Adams, Francis. 1992. From Economic Nationalism to Neo-Liberalism: A Comparative Analysis of Foreign Economic Policy-Making in Venezuela and Jamaica. Ithaca, N.Y.: Latin American Studies Program, Cornell University. Addams, Jane. 1898. “Why the Ward Boss Rules.” Outlook (Apr. 2): 17–28. Aguero, Felipe. 1995. “Crisis and Decay of Democracy in Venezuela: The Civil-Military Dimension.” In Venezuelan Democracy Under Stress, ed. Jennifer McCoy, William C. Smith, Andres Serbin, and Andres Stambouli, pp. 215–35. Miami: North-South Center. Aldrich, John H. 1983. “A Downsian Spatial Model with Party Activism.” American Political Science Review 77(4): 974–90. ———.1995. Why Parties? The Origin and Transformation of Political Parties in America. Chicago, Ill.: University of Chicago Press. Aldrich, John H., and Michael D. McGinnis. 1989. “A Model of Party Constraints on Optimal Candidate Positions.” Mathematical and Computer Modelling 12(4–5): 437–50.

260 bibliogr aphy

Alonso, Paula. 2000. Between Revolution and the Ballot Box: The Origins of the Argentine Radical Party in the 1890s. Cambridge: Cambridge University Press. Alvarez, Angel E. 2003. “State Reform before and after Chavez’s Election.” In Venezuelan Politics in the Chavez Era: Class Polarization, and Conflict, ed. Steve Ellner and Daniel Hellinger. Boulder: Lynne Rienner. Anderson, Christopher J., Silvia M. Mendes, and Yuliya V. Tverdova. 2004. “Endogenous Economic Voting: Evidence from the 1997 British Election.” Electoral Studies 23(4): 683–708. Arce, Moises. 2005. Market Reform in Society: Post-Crisis Politics and Economic Change in Authoritarian Peru. University Park: Pennsylvania State University Press. Aronson, Elliot, Phoebe C. Ellsworth, J. Merrill Carlsmith, and Marti Hope Gonzales. 1990. Methods of Research in Social Psychology. 2nd ed. New York: McGraw-Hill. Auyero, Javier. 2000. Poor People’s Politics: Peronist Survival Networks and the Legacy of Evita. Durham, N.C.: Duke University Press. Baloyra, Enrique, and Aristides Torres. 1983. BATOBA 83 Survey. National sample survey of 1789 Venezuelans. Archived at the Banco de Datos Latinoamericano de la Universidad Simón Bolívar, Caracas, Venezuela. Banco Central de Reserva del Peru. 1993. “Peru: Compendio Estadistico del Sector Externo, 1970–1992.” Technical report. Lima: BCRP. Bardi, Luciano. 1996. “Anti-Party Sentiment and Party System Change in Italy.” European Journal of Political Research 29: 345–63. Baron, Reuben M., and David A. Kenny. 1986. “The Moderator-Mediator Variable Distinction in Social Psychological Research: Conceptual, Strategic, and Statistical Considerations.” Journal of Personality and Social Psychology 51(6): 1173–82. Bartels, Larry M. 2000. “Partisanship and Voting Behavior, 1952–1996.” American Journal of Political Science 44: 35–50. Bartolini, Stefano, and Peter Mair. 1990. Identity, Competition and Electoral Availability: The Stabilisation of European Electorates 1885–1985. Cambridge: Cambridge University Press. Belaúnde Terry, Fernando. 1960. “El Peru Como Doctrina.” Journal of Inter-American Studies 2: 159–69. Benton, Allyson Lucinda. 2005. “Dissatisfied Democrats or Retrospective Voters? Economic Hardship, Political Institutions, and Voting Behavior in Latin America.” Comparative Political Studies 38: 417–42. Blendon, Robert J., John M. Benson, Mollyann Brodie, Richard Morin, Drew E. Altman, Daniel Gitterman, Mario Brossard, and Matt James. 1997. “Bridging the Gap between the Public’s and Economists’ Views of the Economy.” Journal of Economic Perspectives 11: 105–18. Bonilla, Heraclio, and Paul W. Drake. 1989. El APRA, de la Ideologia a la Praxis. Lima: Centro Latinoamericano de Historia Economica y Social. Box-Steffensmeier, Janet M., and Suzanna De Boef. 2001. “Macropartisanship and Macro­ ideology in the Sophisticated Electorate.” Journal of Politics 63: 232–48. Box-Steffensmeier, Janet M., and Renee M. Smith. 1996. “The Dynamics of Aggregate Partisanship.” American Political Science Review 90: 567–80. Box-Steffensmeier, Janet M., Kathleen Knight, and Lee Sigelman. 1998. “The Interplay of Macropartisanship and Macroideology: A Time Series Analysis.” Journal of Politics 60: 1031–49. Brady, David W., and Sean M. Theriault. 2001. “A Reassessment of Who’s to Blame: A Positive Case for the Public Evaluation of Congress.” In What Is It about Government

bibliogr aphy

261

That Americans Dislike?, ed. John R. Hibbing and Elizabeth Theiss-Morse, pp. 175–92. Cambridge: Cambridge University Press. Brusco, Valeria, Marcelo Nazareno, and Susan C. Stokes. 2004. “Vote Buying in Argentina.” Latin American Research Review 39: 66–88. Burgess, Katrina. 2004. Parties and Unions in the New Global Economy. Pittsburgh: University of Pittsburgh Press. Calvo, Ernesto, and Maria Victoria Murillo. 2004. “Who Delivers? Partisan Clients in the Argentine Electoral Market.” American Journal of Political Science 48: 742–57. Calvo, Guillermo A., and Frederic S. Mishkin. 2003. “The Mirage of Exchange Rate Regimes for Emerging Market Countries.” Journal of Economic Perspectives 17: 99–118. Cameron, Maxwell A. 1994. Democracy and Authoritarianism in Peru: Political Coalitions and Social Change. Basingstoke: Palgrave Macmillan. Cameron, Maxwell A., and Steven Levitsky. 2009. Democracy without Parties? Political Parties and Regime Change in Fujimori’s Peru. In Latin American Democratic Transformations: Institutions, Actors, and Processes, ed. William C. Smith. New Jersey: Wiley-Blackwell. Cameron, Maxwell A., and Philip Mauceri, eds. 1997. The Peruvian Labyrinth: Polity, Society, Economy. University Park: Pennsylvania State University Press. Campbell, Angus, Philip E. Converse, Warren E. Miller, and Donald E. Stokes. 1960. The American Voter. New York: Wiley. Caplan, Bryan. 2002. “Sociotropes, Systematic Bias, and Political Failure: Reflections on the Survey of Americans and Economists on the Economy.” Social Science Quarterly 83: 416–35. Carey, John M., Octavio Amorim Neto, and Matthew Soberg Shugart. 1997. “Appendix: Outlines of Constitutional Powers in Latin America.” In Presidentialism and Democracy in Latin America, ed. Scott Mainwaring and Matthew Soberg Shugart. Cambridge: Cambridge University Press. Carmines, Edward G., and James A. Stimson. 1987. Issue Evolution: Race and the Transformation of American Politics. Princeton, N.J.: Princeton University Press. Carrasquero, J., and F. Welsch. 1993. Survey of Venezuela discussed in Friedrich Welsch and José Vicente Carrasquero, 1996, “¿Desconsolidación de la Democracia en Venezuela? Rendimiento y Legitimidad Normativa.” Cuestiones Políticas 16, 45–69. Carrion, Julio F. 2006. The Fujimori Legacy: The Rise of Electoral Authoritarianism in Peru. University Park: Pennsylvania State University Press. Carty, R. Kenneth. 2002. “Canada’s Nineteenth-Century Cadre Parties at the Millennium.” In Political Parties in Advanced Industrial Democracies, ed. Paul Webb, David Farrell and Ian Holliday. Oxford: Oxford University Press. Centro de Investigaciones y Estudios Politicos y Administativos. N.d. DOXA 93 Survey. University of Zulia. Chávez Frías, Hugo Rafael, and Haiman El Troudi. 2005a. 1999: “Año de la Refundación de la República” Selección de Discursos del Presidente de la República Bolivariana de Venezuela, Hugo Chávez Frías. Caracas: Ediciones de la Presidencia de la República. ———.2005b. 2002: “Año de la Resistencia Antiimperialista” Selección de Discursos del Presidente de la República Bolivariana de Venezuela, Hugo Chávez Frías. Caracas: Ediciones de la Presidencia de la República. Chong, Dennis, and James N. Druckman. 2007. “Framing Theory.” Annual Review of Political Science 10: 103–26. Citrin, Jack. 1974. “Comment: The Political Relevance of Trust in Government.” American Political Science Review 68(3): 973–88.

262 bibliogr aphy

Clarke, Harold D., and Alan C. Acock. 1989. “National Elections and Political Attitudes: The Case of Political Efficacy.” British Journal of Political Science 19: 551–62. Coan, James A., and John J. B. Allen, eds. 2007. Handbook of Emotion Elicitation and Assessment. Oxford: Oxford University Press. Collier, Ruth Berins. 2001. Populism. In International Encyclopedia of Social and Behavioral Sciences, ed. Neil J. Smelser and P. B. Baltes, pp. 11813–16. Amsterdam: Elsevier. ———.2002. “Popular Representation in Latin America’s Neoliberal Democracies: Political Participation and Associational Life.” Unpublished Manuscript. Department of Political Science, University of California, Berkeley. ———.2003. 2002–03 Survey of Comparative Institutions of Representation in Latin America. Archived at the Center for Survey Research, University of California, Berkeley. Collier, Ruth Berins, and David Collier. 1991. Shaping the Political Arena: Critical Junctures, the Labor Movement, and Regime Dynamics in Latin America. Princeton, N.J.: Prince­ ton University Press. ———.2002. Shaping the Political Arena: Critical Junctures, the Labor Movement, and Regime Dynamics in Latin America. 2nd ed. South Bend, Ind.: University of Notre Dame. Colomer, Josep M. 2005. “The Left-Right Dimension in Latin America.” Technical Report 813, Department of Economics and Business, Universitat Pompeu Fabra, Economics Working Papers. Conaghan, Catherine M. 2006. Fujimori’s Peru: Deception in the Public Sphere. Pittsburgh, Penn.: University of Pittsburgh Press. Conaghan, Catherine M., and James Malloy. 1994. Unsettling Statecraft: Democracy and Neoliberalism in the Central Andes. Pittsburgh, Penn.: University of Pittsburgh Press. Conaghan, Catherine M., James M. Malloy, and Luis A. Abugattas. 1990. “Business and the ‘Boys’: The Politics of Neoliberalism in the Central Andes.” Latin American Research Review 25(2): 3–30. Coppedge, Michael. 1994. Strong Parties and Lame Ducks: Presidential Partyarchy and Factionalism in Venezuela. Palo Alto, Calif.: Stanford University Press. ———.1998. “The Dynamic Diversity of Latin American Party Systems.” Party Politics 4: 547–68. ———.2001. “Political Darwinism in Latin America’s Lost Decade.” In Political Parties and Democracy, ed. Larry Diamond and Richard Gunther. Baltimore, Md.: Johns Hopkins University Press. ———.2005. “Explaining Democratic Deterioration in Venezuela through Nested Inference.” In The Third Wave of Democratization in Latin America: Advances and Setbacks, ed. Frances Hagopian and Scott P. Mainwaring, pp. 289–316. Cambridge: Cambridge University Press. Coronel, Gustavo. 1983. The Nationalization of Venezuelan Oil. Toronto: Lexington Books. Corrales, Javier. 2000. “Reform-Lagging States and the Question of Devaluation: Venezuela’s Response to the Exogenous Shocks of 1997–1998.” In Exchange Rate Politics in Latin America, ed. Carol Wise and Riordan Roett. Washington, D.C.: Brookings Institution. ———.2002. Presidents without Parties: The Politics of Economic Reform in Argentina and Venezuela in the 1990s. University Park: Pennsylvania State University Press. Cotler, Julio. 1995. “Political Parties and the Problem of Democratic Consolidation in Peru.” In Building Democratic Institutions: Party Systems in Latin America, ed. Scott Mainwaring and Timothy R. Scully, pp. 323–53. Palo Alto, Calif.: Stanford University Press. Cox, Gary W. 1997. Making Votes Count: Strategic Coordination in the World’s Electoral Systems. Cambridge: Cambridge University Press.

bibliogr aphy

263

Crisp, Brian F. 1997. “Presidential Behavior in a System with Strong Parties: Venezuela, 1958–1995.” In Presidentialism and Democracy in Latin America, ed. Scott Mainwaring and Matthew Soberg Shugart, pp. 160–98. Cambridge: Cambridge University Press. ———.1998. “Lessons from Economic Reform in the Venezuelan Democracy.” Latin American Research Review 33: 7–41. Crisp, Brian F., Daniel H. Levine, and Jose E. Molina. 2003. “The Rise and Decline of COPEI in Venezuela.” In Christian Democracy in Latin America: Electoral Competition and Regime Conflicts, ed. Scott Mainwaring and Timothy R. Scully. Palo Alto, Calif.: Stanford University Press. Dalton, Russell J. 2000. “The Decline of Party Identifications.” In Parties without Partisans: Political Change in Advanced Industrial Democracies, ed. Russell J. Dalton and Martin P. Wattenberg. Oxford: Oxford University Press. ———.2002. Citizen Politics: Public Opinion and Political Parties in Advanced Industrial Democracies. 3rd ed. New York: Chatham House. Dalton, Russell J., and Martin P. Wattenberg. 2000. Parties without Partisans: Political Change in Advanced Industrial Democracies. Oxford: Oxford University Press. Dalton, Russell J., Scott Flanagan, and Paul Allen Beck. 1984. Electoral Change in Advanced Industrial Democracies: Realignment or Dealignment? Princeton, N.J.: Princeton University Press. Davis, Charles, and Kenneth Coleman. 1989. “Political Control of Organized Labor in a Semi-Consociational Democracy: The Case of Venezuela.” In Labor Autonomy and the State in Latin America, ed. Edward Epstein. London: Unwin Hyman. De Oliveira, Orlandina, and C. B. Roberts. 1998. “Urban Social Structures in Latin America, 1930–1990.” In Latin America: Economy and Society Since 1930, ed. Leslie Bethell. Cambridge: Cambridge University Press. Degregori, Carlos Ivan. 2000. La Decada de la Antipolitica. Lima: Instituto de Estudios Peruanos. Deschouwer, Kris. 1994. “The Decline of Consociationalism and the Reluctant Modernization of Belgian Mass Parties.” In How Parties Organize, ed. Richard Katz and Peter Mair. Oxford: Oxford University Press. Dietz, Henry. 1986. “Electoral Politics in Peru, 1978–1986.” Journal of Interamerican Studies and World Affairs 28: 139–63. ———.1998. Urban Poverty, Political Participation, and the State, Lima, 1970–1990. Pittsburgh, Penn.: University of Pittsburgh Press. Dietz, Henry A., and David J. Myers. 2007. “From Thaw to Deluge: Party System Collapse in Venezuela and Peru.” Latin American Politics and Society 49: 59–86. Dornbusch, Rudiger, and Sebastian Edwards. 1991. “The Macroeconomics of Populism.” In The Macroeconomics of Populism in Latin America, ed. Rudiger Dornbusch and Sebastian Edwards. Chicago, Ill.: University of Chicago Press. Downs, Anthony. 1957. An Economic Model of Democracy. New York: Harper and Row. Duch, Raymond M., Harvey D. Palmer, and Christopher J. Anderson. 2000. “Heterogeneity in Perceptions of National Economic Conditions.” American Journal of Political Science 44(4): 635–52. Duverger, Maurice. 1954. Political Parties. New York: Wiley. Echegaray, Fabian, and Carlos Elordi. 2001. “Public Opinion, Presidential Popularity, and Economic Reforms in Argentina, 1989–1996.” In Public Support for Market Reforms in New Democracies, ed. Susan C. Stokes. Cambridge: Cambridge University Press.

264 bibliogr aphy

Ellner, Steve. 1988. Venezuela’s Movimiento al Socialismo: From Guerrilla Defeat to Innovative Politics. Durham, N.C.: Duke University Press. Ellner, Steve, and Daniel Hellinger. 2003. Venezuelan Politics in the Chavez Era: Class, Polar­ ization, and Conflict. Boulder, Colo.: Lynne Rienner. Elster, Jon. 1989. Nuts and Bolts for the Social Sciences. Cambridge: Cambridge University Press. ———.2007. Explaining Social Behavior: More Nuts and Bolts for the Social Sciences. Cambridge: Cambridge University Press. Enright, Michael J., Antonio Frances, and Edith Scott Saavedra. 1996. Venezuela: The Challenge of Competitiveness. New York: St. Martin’s Press. Erikson, Robert S., Michael B. MacKuen, and James A. Stimson. 1998. “What Moves Macro­partisanship? A Response to Green, Palmquist, and Schickler.” American Political Science Review 92: 901–12. ———.2002. The Macro Polity. Cambridge: Cambridge University Press. Ernst, Michael D. 2004. “Permutation Methods: A Basis for Exact Inference.” Statistical Science 19(4): 676–85. Evans, Geoffrey, and Robert Andersen. 2006. “The Political Conditioning of Economic Perceptions.” Journal of Politics 68(1): 194–207. Finkel, Steven E. 1985. “Reciprocal Effects of Participation and Political Efficacy: A Panel Analysis.” American Journal of Political Science 29: 891–913. ———.1987. “The Effects of Participation on Political Efficacy and Political Support.” Journal of Politics 49: 441–64. Fiorina, Morris P. 1977. “An Outline for a Model of Party Choice.” American Journal of ­Political Science 21(3): 601–25. ———.1981. Retrospective Voting in American National Elections. New Haven, Conn.: Yale University Press. Florez, Carlos Zuzunaga. 1992. El Arte de Perder Una Elección. Lima: PEISA. Forgas, Joseph P., ed. 2000. Feeling and Thinking: Affective Influences on Social Cognition. Cambridge: Cambridge University Press. Freedman, David A. 2005. Statistical Models: Theory and Practice. Cambridge: Cambridge University Press. ———.2008a. “On Regression Adjustments in Experiments with Several Treatments.” Annals of Applied Statistics 2: 176–96. ———.2008b. “On Regression Adjustments to Experimental Data.” Advances in Applied Mathematics 40: 180–93. French-Davis, Ricardo, Oscar Munoz, and Jose Gabriel Palma. 1998. “The Latin American Economies, 1950–1990.” In Latin America: Economy and Society Since 1930, ed. Leslie Bethell, pp. 149–240. Cambridge: Cambridge University Press. Frieden, Jeffry A. 1991. Debt, Development, and Democracy: Modern Political Economy and Latin America, 1965–1985. Princeton, N.J.: Princeton University Press. Fuchs, Dieter, and Hans-Dieter Klingemann. 1989. “The Left-Right Schema.” In Continuities in Political Action, ed. M. K. Jennings and J. van Deth. Berlin: deGruyter. Gelman, Andrew, and Jennifer Hill. 2006. Data Analysis Using Regression and Multilevel/ Hierarchical Models. Cambridge: Cambridge University Press. Gerber, Alan, and Donald Green. 1999. “Misperceptions about Perceptual Bias.” Annual Review of Political Science 2: 189–210. Gerring, John. 2001. Party Ideologies in America, 1828–1996. Cambridge: Cambridge University Press.

bibliogr aphy

265

———.2007. “The Mechanismic Worldview: Thinking inside the Box.” British Journal of Political Science 38(1): 161–79. Gillespie, Kate, and Gwenn Okruhlik. 1991. “The Political Dimensions of Corruption Cleanups: A Framework for Analysis.” Comparative Politics 24(1): 77–95. Glewwe, Paul, and Gillette Hall. 1992. Poverty and Inequality during Unorthodox Adjustment: The Case of Peru, 1985–90. Washington, D.C.: World Bank. Gonzales de Olarte, Efrain. 1996. “Inversión Privada, Crecimiento, y Ajuste Estructural en el Peru, 1950–1995.” Lima: Instituto de Estudios Peruanos. Gorriti Ellenbogen, Gustavo. 1999. The Shining Path: A History of the Millenarian War in Peru. Chapel Hill: University of North Carolina Press. Graham, Carol. 1992. Peru’s APRA: Parties, Politics, and the Elusive Quest for Democracy. Boulder, Colo.: Lynne Rienner. Green, Donald P. 2009. Paper presented at the 26th Annual Meeting of the Society for Political Methodology, Yale University, July 23–25. Green, Donald, Bradley Palmquist, and Eric Schickler. 1998. “Macropartisanship: A Replication and Critique.” American Political Science Review 92: 883–99. ———.2002. Partisan Hearts and Minds: Political Parties and the Social Identities of Voters. New Haven, Conn.: Yale University Press. Greene, Kenneth F. 2002. “Opposition Party Strategy and Spatial Competition in Dominant Party Regimes: A Theory and the Case of Mexico.” Comparative Political Studies 35(7): 755–83. ———.2007. Why Dominant Parties Lose: Mexico’s Democratization in Comparative Perspective. Cambridge: Cambridge University Press. Gross, James J., and Robert W. Levenson. 1995. “Emotion Elicitation Using Films.” Cognition and Emotion 9: 87–108. Grzymala-Busse, Anna M. 2002. Redeeming the Communist Past: The Regeneration of Communist Parties in East Central Europe. Cambridge: Cambridge University Press. Haggard, Stephan, and Robert Kaufman. 1995. The Political Economy of Democratic Transitions. Princeton, N.J.: Princeton University Press. Haggard, Stephan, and Robert Kaufman, eds. 1992. The Politics of Economic Adjustment. Princeton, N.J.: Princeton University Press. Harmel, Robert, and Kenneth Janda. 1994. “An Integrated Theory of Party Goals and Party Change.” Journal of Theoretical Politics 6(4): 259–87. Hausman, J., and D. McFadden. 1984. “Specification Tests for the Multinomial Logit Model.” Econometrica 52(5): 1219–40. Hawkins, Kirk A. 2011. Venezuela’s Chavismo and Populism in Comparative Perspective. Cambridge: Cambridge University Press. Healy, Andrew J., Neil Malhotra, and Cecilia Hyunjung Mo. 2010. “Irrelevent Events Affect Voters’ Evaluations of Government Performance.” Proceedings of the National Academy of Sciences. Washington, D.C. Hedstrom, Peter, and Richard Swedberg. 1998. Social Mechanisms: An Analytical Approach to Social Theory. Cambridge: Cambridge University Press. Herman, Donald. 1980. Christian Democracy in Venezuela. Chapel Hill: University of North Carolina Press. Heston, Alan, Robert Summers, and Bettina Aten. 2006. “Penn World Table Version 6.1, Center for International Comparisons at the University of Pennsylvania (CICUP).” Philadelphia.

266 bibliogr aphy

Hibbing, John R., and Elizabeth Theiss-Morse. 1995. Congress as Public Enemy: Public Attitudes toward American Political Institutions. Cambridge: Cambridge University Press. Hirschman, Albert O. 1970. Exit, Voice, and Loyalty: Responses to Decline in Firms, Organizations, and States. Cambridge: Harvard University Press. Holland, Paul W. 1986. “Statistics and Causal Inference.” Journal of the American Statistical Association 81: 945–60. Holt, Michael F. 1999. The Rise and Fall of the American Whig Party: Jacksonian Politics and the Onset of the Civil War. Oxford: Oxford University Press. Imai, Kosuke, Luke Keele, and Teppei Yamamoto. 2010. “Identification, Inference and Sensitivity Analysis for Causal Mediation Effects.” Statistical Science 25(1): 51–71. Inglehart, Ronald. 1984. “Changing Cleavage Alignments in Western Democracies.” In Electoral Change in Advanced Industrial Democracies, ed. Russell J. Dalton, Scott Flanagan, and Paul Allen Beck. Princeton, N.J.: Princeton University Press. Inglehart, Ronald, et al. 1999. “World Values Surveys and European Values Survey,” 1981– 1984, 1990–1993, and 1995–1997 [Computer file]. Technical report ICPSR02790-v1. Ann Arbor, Mich.: Institute for Social Research [producer]/Inter-university Consortium for Political and Social Research [distributor], 2003. International IDEA. 2001. “Voter Turnout: Venezuela.” Technical report. Stockholm: International IDEA. Isbell, Linda M., Victor C. Ottati, and Kathleen C. Burns. 2006. “Affect and Politics: Effects on Judgment, Processing, and Information Seeking.” In Feeling Politics: Emotion in Political Information Processing, ed. David P. Redlawsk. New York: Palgrave Macmillan. Iyengar, Shanto, and Donald R. Kinder. 1987. News That Matters: Television and American Opinion. Chicago, Ill.: University of Chicago. James, Daniel. 1994. Resistance and Integration: Peronism and the Argentine Working Class, 1946–1976. Cambridge: Cambridge University Press. Jones, Bart. 2008. Hugo! The Hugo Chavez Story from Mud Hut to Perpetual Revolution. London: The Bodley Head. Karl, Terry Lynn. 1995. The Venezuelan Petro-State and the Crisis of “Its” Democracy. In Venezuelan Democracy under Stress, ed. Jennifer McCoy, William C. Smith, Andres Serbin, and Andres Stambouli. Miami: North-South Center. Karl, Terry Lynn. 1997. The Paradox of Plenty: Oil Booms and Petro-States. Berkeley: University of California Press. Katz, Richard S., and Peter Mair. 1995. “Changing Models of Party Organization and Party Democracy: The Emergence of the Cartel Party.” Party Politics 1(1): 5–28. Kenney, Charles D. 2004. Fujimori’s Coup and the Breakdown of Democracy in Latin America. Notre Dame: University of Notre Dame Press. Key, V. O. Jr. 1955. “A Theory of Critical Elections.” The Journal of Politics 17: 3–18. Kinder, Donald R. 1998. “Communication and Opinion.” Annual Review of Political ­Science 1: 167–97. Kinder, Donald R., and Walter Mebane. 1983. “Politics and Economics in Everyday Life.” In The Political Process and Economic Change, ed. Kristin R. Monroe. New York: Agathon. Kirchheimer, Otto. 1966. “The Transformation of Western European Party Systems.” In Political Parties and Political Development, ed. Joseph La Palombara and M. Weiner. Princeton: Princeton University Press. Kitschelt, Herbert. 1989. The Logics of Party Formation. Ithaca, N.Y.: Cornell University Press. Kitschelt, Herbert. 1994. The Transformation of European Social Democracy. Cambridge: Cambridge University Press.

bibliogr aphy

267

Kitschelt, Herbert, and Anthony J. McGann. 1997. The Radical Right in Western Europe: A Comparative Analysis. Ann Arbor: University of Michigan Press. Kitschelt, Herbert, Kirk A. Hawkins, Juan Pablo Luna, Guillermo Rosas, and Elizabeth Zechmeister. 2010. Latin American Party Systems. Cambridge: Cambridge University Press. Knight, Alan. 1998. “Populism and Neo-Populism in Latin America, Especially Mexico.” Journal of Latin American Studies 30: 223–48. Kornblith, Miriam. 1998. Venezuela en los 90: Las Crisis de la Democracia. Caracas: E ­ diciones IESA. Kornblith, Miriam, and Daniel H. Levine. 1995. “Venezuela: The Life and Times of the Party System.” In Building Democratic Institutions: Party Systems in Latin America, ed. Scott Mainwaring and Timothy R. Scully. Stanford, Calif.: Stanford University Press. Kreutz, Gunter, Ulrich Ott, Daniel Teichmann, Patrick Osawa, and Dieter Vaitl. 2008. “Using Music to Induce Emotions: Influences of Musical Preference and Absorption.” Psychology of Music 36: 101–26. Krivoy, Ruth de. 2000. Collapse: The Venezuelan Banking Crisis of 1994. Washington, D.C.: Group of Thirty. Lago, Ricardo. 1991. The Illusion of Pursuing Redistribution through Macropolicy: Peru’s Heterodox Experience, 1985–1990. In The Macroeconomics of Populism in Latin America, ed. Rudiger Dornbusch and Sebastian Edwards, pp. 263–330. Chicago, Ill.: University of Chicago Press. Lander, Edgardo, and Luis A. Fierro. 1996. “The Impact of Neoliberal Adjustment in Vene­ zuela, 1989–1993.” Latin American Perspectives 23: 50–73. Lapper, Richard. 2007. Living with Hugo: U.S. Policy toward Hugo Chavez’s Venezuela. New York: Council on Foreign Relations. Lazarus, Richard S. 1991. Emotion and Adaptation. Oxford: Oxford University Press. Lerner, Daniel. 1958. The Passing of Traditional Society: Modernizing the Middle East. New York: Macmillan. Lerner, Jennifer S., and Dacher Keltner. 2000. “Beyond Valence: Toward a Model of Emotion-Specific Influences on Judgment and Choice.” Cognition and Emotion 14: 473–93. ———.2001. “Fear, Anger, and Risk.” Journal of Personality and Social Psychology 81: 146–59. Lerner, Jennifer S., and Larissa Z. Tiedens. 2006. “Portrait of the Angry Decision Maker: How Appraisal Tendencies Shape Anger’s Influence on Cognition.” Journal of Behavioral Decision Making 19: 115–37. Levitsky, Steven. 2003. Transforming Labor-Based Parties in Latin America: Argentine Peronism in Comparative Perspective. Cambridge: Cambridge University Press. Levitsky, Steven, and Lucan Way. 1998. “Between a Shock and a Hard Place: The Dynamics of Labor-Backed Adjustment in Poland and Argentina.” Comparative Politics 30(2): 171–92. Lewis-Beck, Michael S. 1988. Economics and Elections: The Major Western Democracies. Ann Arbor: University of Michigan Press. Lewis-Beck, Michael S., and Martin Paldam. 2000. “Economic Voting: An Introduction.” Electoral Studies 19: 113–21. Lewis, Jeffrey B., and Gary King. 1999. “No Evidence on Directional vs. Proximity Voting.” Political Analysis 8(1): 21–33. Lipset, Seymour Martin, and Stein Rokkan. 1967. “Cleavage Structures, Party Systems, and Voter Alignments: An Introduction.” In Party Systems and Voter Alignments: CrossNational Perspectives, ed. Seymour Martin Lipset and Stein Rokkan, pp. 1–64. New York: Free Press. Lodge, Milton, and Charles S. Taber. 2000. “Three Steps toward a Theory of Motivated

268 bibliogr aphy

Political Reasoning.” In Elements of Political Reason: Understanding and Expanding the Limits of Rationality, ed. Arthur Lupia, Mathew McCubbins, and Samuel Popkin, pp. 183–213. Cambridge: Cambridge University Press. López Maya, Margarita, David Smilde, and Keta Stephany. 2002. Protesta y Cultura en Vene­ zuela: Los Marcos de Acción Colectiva en 1999. Buenos Aires: Consejo Latinoamericano de Ciencias Sociales. Lowenthal, Abraham. 1975. The Peruvian Experiment: Continuity and Change under Military Rule. Princeton, N.J.: Princeton University Press. Lynch, Nicolas. 1999. Una Tragedia Sin Heroes: La Derrota de los Partidos y el Origen de los Independientes, Peru 1980–1992. Lima: Fondo Editorial de la Universidad de San Marcos. MacKuen, Michael B., Robert S. Erikson, and James A. Stimson. 1989. “Macropartisanship.” American Political Science Review 83: 1125–42. MacKuen, Michael B., Robert S. Erikson, James A. Stimson, Paul R. Abramson, and Charles W. Ostrom. 1992. “Question Wording and Macropartisanship.” American Political Science Review 86: 475–86. Magaloni, Beatriz. 2006. Voting for Autocracy: Hegemonic Party Survival and its Demise in Mexico. Cambridge University Press. Mainwaring, Scott, and Timothy R. Scully. 1995a. Building Democratic Institutions: Party Systems in Latin America. Stanford, Calif.: Stanford University Press. ———.1995b. “Introduction: Party Systems in Latin America.” In Building Democratic Institutions: Party Systems in Latin America, ed. Scott Mainwaring and Timothy R. Scully. Stanford, Calif.: Stanford University Press. Mainwaring, Scott P., and Anibal Perez-Linan. 2005. “Latin American Democratization since 1978: Democratic Transitions, Breakdowns, and Erosions.” In The Third Wave of Democratization in Latin America: Advances and Setbacks, ed. Frances Hagopian and Scott P. Mainwaring. Cambridge: Cambridge University Press. Mainwaring, Scott P., Daniel Brinks, and Anibal Perez-Linan. 2001. “Classifying Political Regimes in Latin America, 1945-1999.” Studies in Comparative International Development 36: 37–65. Malloy, James M. 1977. Authoritarianism and Corporatism in Latin America. Pittsburgh: University of Pittsburgh. Marcano, Cristina, and Alberto Barrera Tyszka. 2007. Hugo Chavez: The Definitive Biography of Venezuela’s Controversial President. New York: Random House. Marcus, George E. 2002. The Sentimental Citizen: Emotion in Democratic Politics. University Park: Pennsylvania State University Press. Marcus, George E., W. Russell Neuman, and Michael MacKuen. 2000. Affective Intelligence and Political Judgment. Chicago, Ill.: University of Chicago Press. Martz, John D. 1966. Acción Democrática: Evolution of a Modern Political Party in Venezuela. Princeton, N.J.: Princeton University Press. Martz, John D. 1977. “The Party System: Toward Institutionalization.” In Venezuela: The Democratic Experience, ed. John D. Martz and David J. Myers, pp. 93–112. New York: Praeger. ———.1999–2000. “Political Parties and Candidate Selection in Venezuela and Colombia.” Political Science Quarterly 114: 639–59. Mayhew, David R. 2002. Electoral Realignments: A Critique of an American Genre. New Haven, Conn.: Yale University Press. Mayorbe, Eduardo. 1985. “The Renegotiation of Venezuela’s Foreign Debt During 1982 and 1983.” In Politics and Economics of External Debt Crisis: The Latin American Experience, ed. Miguel S. Wionczek. Boulder, Colo.: Westview Press.

bibliogr aphy

269

Mayorga, Rene Antonio. 2006. “Outsiders and Neopopulism: The Road to Plebiscitary Democracy.” In The Crisis of Democratic Representation in the Andes, ed. Scott Mainwaring, Ana Maria Bejarano, and Eduardo Pizarro Leongomez. Stanford, Calif.: Stanford University Press. Maza Zavala, Domingo Felipe. 1996. Crisis y Política Económica: 1989–1996. Caracas: Academia National de Ciencias Económicas. McCann, James A., and Chappell Lawson. 2003. “An Electorate Adrift? Public Opinion and the Quality of Democracy in Mexico.” Latin American Research Review 38(3): 60–81. McClintock, Cynthia. 1998. Revolutionary Movements in Latin America: El Salvador’s FMLN and Peru’s Shining Path. Washington, D.C.: United States Institute of Peace Press. McClintock, Cynthia, and Abraham F. Lowenthal. 1983. The Peruvian Experiment Reconsidered. Princeton, N.J.: Princeton University Press. McCormick, Gordon H. 1993. Sharp Dressed Men: Peru’s Tupac Amaru Revolutionary Movement. Santa Monica, Calif.: RAND, National Defense Research Institute. McCoy, Jennifer L., and David J. Myers, eds. 2004. The Unraveling of Representative Democracy in Venezuela. Baltimore, Md.: Johns Hopkins University Press. McCoy, Jennifer, and William C. Smith. 1995. “From Deconsolidation to Reequilibration? Prospects for Democratic Renewal in Venezuela.” In Venezuelan Democracy Under Stress, ed. Jennifer McCoy, William C. Smith, Andres Serbin, and Andres Stambouli. Miami: North-South Center. McGuire, James W. 1995. Political Parties and Democracy in Argentina. In Building Democratic Institutions: Party Systems in Latin America, ed. Scott Mainwaring and Timothy R. Scully, pp. 200–46. Stanford, Calif.: Stanford University Press. McGuire, James W. 1997. Peronism without Peron: Unions, Parties, and Democracy in Argentina. Stanford, Calif.: Stanford University Press. McKibbin, R. I. 1974. The Evolution of the Labour Party, 1910–24. Oxford: Oxford University Press. Meffert, Michael F., Helmut Norpoth, and Anirudh V. S. Ruhil. 2001. “Realignment and Macropartisanship.” American Political Science Review 95: 953–62. Mill, John Stuart. [1891] 2002. A System of Logic: Ratiocinative and Inductive. Honolulu: University Press of the Pacific. Miller, Arthur H., and Ola Listhaug. 1990. “Political Parties and Confidence in Government: A Comparison of Norway, Sweden and the United States.” British Journal of Political Science 20(3): 357–86. ———.1999. “Political Performance and Institutional Trust.” In Critical Citizens: Global Support for Democratic Government, ed. Pippa Norris, pp. 204–16. Oxford: Oxford University Press. Miller, Warren E., and J. Merrill Shanks. 1996. The New American Voter. Cambridge, Mass.: Harvard University Press. Mizrahi, Yemile. 2003. From Martyrdom to Power: The Partido Acción Nacional in Mexico. Notre Dame: University of Notre Dame Press. Molina, Jose E. 2001. “The Electoral Effect of Underdevelopment: Government Turnover and Its Causes in Latin America, the Caribbean, and Industrialized Countries.” Electoral Studies 20: 427–46. ———.2004. The Unraveling of Venezuela’s Party System: From Party Rule to Personalistic Politics and Deinstitutionalization. In The Unraveling of Representative Democracy in Venezuela, ed. Jennifer L. McCoy and David J. Myers. Baltimore, Md.: Johns Hopkins University Press.

270 bibliogr aphy

Molina, Jose E., and Carmen Perez. 1998. “Evolution of the Party System in Venezuela, 1946–1993.” Journal of Interamerican Studies and World Affairs 40: 1–26. Moreno, Alejandro. 1998. “Party Competition and the Issue of Democracy: Ideological Space in Mexican Elections.” In Governing Mexico: Political Parties and Elections, ed. Monica Serrano. London: Macmillan. Morgan, Jana. 2007. “Partisanship during the Collapse of Venezuela’s Party System.” Latin American Research Review (February): 78–98. ———.2011. Bankrupt Representation and Party System Collapse. University Park: Pennsylvania State University Press. Morgan, Stephen L., and Christopher Winship. 2007. Counterfactuals and Causal Inference: Methods and Principles for Social Research. Cambridge: Cambridge University Press. Morgenstern, Scott, and Elizabeth Zechmeister. 2001. “Better the Devil You Know than the Saint You Don’t? Risk Propensity and Vote Choice in Mexico.” Journal of Politics 63(1): 93–119. Morlino, Leonardo, and Marco Tarchi. 1996. “The Dissatisfied Society: The Roots of Political Change in Italy.” European Journal of Political Research 30: 41–63. Morrell, Michael E. 2005. “Deliberation, Democratic Decision-Making, and Internal Political Efficacy.” Political Behavior 27: 49–69. Munck, Ronaldo. 1992. “The Democratic Decade: Argentina Since Malvinas.” Bulletin of Latin American Research 11: 205–16. Murillo, Maria Victoria. 2001. Labor Unions, Partisan Coalitions, and Market Reforms in Latin America. Cambridge: Cambridge University Press. Myers, David J., and Robert E. O’Connor. 1998. “Support for Coups in Democratic Political Culture: A Venezuelan Exploration.” Comparative Politics 30: 193–212. Naim, Moises. 1993. Paper Tigers and Minotaurs: The Politics of Venezuela’s Economic Reforms. Washington, D.C.: Carnegie Endowment for International Peace. Neuhouser, Kevin. 1992. “Democratic Stability in Venezuela: Elite Consensus or Class Compromise?” American Sociological Review 57: 117–35. Neuman, W. Russell, George E. Marcus, Michael MacKuen, and Ann N. Crigler, eds. 2007. The Affect Effect: Dynamics of Emotion in Political Thinking and Behavior. Chicago, Ill.: University of Chicago Press. Nie, Norman H., Jane Junn, and Kenneth Stehlik-Barry. 1996. Education and Democratic Citizenship in America. Chicago, Ill.: University of Chicago Press. Niemi, Richard G. 1974. How Family Members Perceive Each Other. New Haven, Conn.: Yale University Press. Norden, Deborah L. 1996. “The Rise of the Lieutenant Colonels: Rebellion in Argentina and Venezuela.” Latin American Perspectives 23: 74–86. Norden, Deborah L. 1998. “Democracy and Military Control in Venezuela: From Subordination to Insurrection.” Latin American Research Review 33(2): 143–65. Norpoth, Helmut, Michael S. Lewis-Beck, and Jean-Dominique Lafay, eds. 1991. Economics and Politics: The Calculus of Support. Ann Arbor: University of Michigan Press. Ochoa Antich, Enrique. 1992. Los Golpes de Febrero. Caracas: Fuentes Editores. O’Donnell, Guillermo. 1994. “Delegative Democracy.” Journal of Democracy 5: 55–69. O’Donnell, Guillermo, Philippe C. Schmitter, and Laurence Whitehead, eds. 1986. Transitions from Authoritarian Rule. Baltimore, Md.: Johns Hopkins University Press. Ortony, Andrew, Gerald L. Clore, and Allan Collins. 1988. The Cognitive Structure of Emotions. Cambridge: Cambridge University Press.

bibliogr aphy

271

Ostiguy, Pierre. 1998. “Peronism and Anti-Peronism: Class-Cultural Cleavages and Political Identity in Argentina.” PhD dissertation, University of California, Berkeley. Oxhorn, Philip. 1998. “Is the Century of Corporatism Over? Neoliberalism and the Rise of Neopluralism.” In What Kind of Democracy? What Kind of Market? Latin America in the Age of Neoliberalism, ed. Philip D. Oxhorn and Graciela Ducatenzeiler, pp. 195–217. University Park: Pennsylvania State University Press. Pacek, Alexander, and Benjamin Radcliff. 1994. “The Political Economy of Competitive Elections in the Developing World.” American Journal of Political Science 39: 745–59. Palfrey, Thomas R. 1984. “Spatial Equilibrium with Entry.” Review of Economic Studies 51: 139–56. Panebianco, Angelo. 1988. Political Parties: Organization and Power. Cambridge: Cambridge University Press. Panksepp, Jaak. 1998. Affective Neuroscience: The Foundations of Human and Animal Emotions. Oxford: Oxford University Press. Parker, Michael T., and Linda M. Isbell. 2010. “How I Vote Depends on How I Feel: The Differential Impact of Anger and Fear on Political Information Processing.” Psychological Science 21(4): 548–50. Pastor, Manuel. 1992. Inflation, Stabilization, and Debt: Macroeconomic Experiments in Peru and Bolivia. Boulder, Colo.: Westview Press. Pastor, Manuel, and Carol Wise. 1992. “Peruvian Economic Policy in the 1980s: From Ortho­ doxy to Heterodoxy and Back.” Latin American Research Review 27: 83–117. Pateman, Carol. 1970. Participation and Democratic Theory. Cambridge: Cambridge University Press. Patterson, Samuel C., Randall B. Ripley, and Stephen V. Quinlan. 1992. “Citizens’ Orientations toward Legislatures: Congress and the State Legislature.” Western Political Quarterly 45: 315–38. Pesarin, Fortunato, and Luigi Salmaso. 2010. Permutation Tests for Complex Data: Theory, Applications, and Software. Wiley. Popkin, Samuel L. 1991. The Reasoning Voter: Communication and Persuasion in Presidential Campaigns. Chicago, Ill.: University of Chicago Press. Powell, Eleanor Neff, and Joshua Tucker. 2009. “New Approaches to Electoral Volatility: Evidence from Postcommunist Countries.” Presented at the Annual Meeting of the American Political Science Association, Toronto, Canada. Przeworski, Adam. 1991. Democracy and the Market: Political and Economic Reforms in Eastern Europe and Latin America. Cambridge: Cambridge University Press. Rabinowitz, George, and Stuart Elaine MacDonald. 1989. “A Directional Theory of Voting.” American Political Science Review 83: 93–121. Rapoport, Ronald B., and Walter J. Stone. 2005. Three’s a Crowd: The Dynamics of Third Parties, Ross Perot, and Republican Resurgence. Ann Arbor: University of Michigan Press. Redlawsk, David P., and James A. McCann. 2005. “Popular Interpretations of ‘Corruption’ and their Partisan Consequences.” Political Behavior 27: 261–83. Remmer, Karen L. 1990. “Democracy and Economic Crisis: The Latin American Experience.” World Politics 42: 315–35. ———.1991. “The Political Impact of Economic Crisis in Latin America.” American Political Science Review 85: 777–800. Roberts, Kenneth M. 1996. “Neoliberalism and the Transformation of Populism in Latin America: The Peruvian Case.” World Politics 48(1): 82–116.

272 bibliogr aphy

———.1997. “Structural Adjustment and the Adaptation or Breakdown of Party Systems: A Comparison of Chile, Argentina, Peru, and Venezuela.” Paper presented at the Twentieth International Congress of the Latin American Studies Association, Guadalajara, Mexico, April 17–19. ———.1998. Deepening Democracy? The Modern Left and Social Movements in Chile and Peru. Stanford, Calif.: Stanford University Press. ———.2003. “Why Not Parties? The Logic of Independent and Anti-Party Political Entrepreneurship.” Paper presented at the Annual Meeting of the American Political Science Association, Philadelphia. Roberts, Kenneth M., and Moises Arce. 1998. “Neo-Liberalism and Lower-Class Voting Behavior in Peru.” Comparative Political Studies 31 (April): 217-46. Roberts, Kenneth M., and Erik Wibbels. 1999. “Party Systems and Electoral Volatility in Latin America: A Test of Economic, Institutional, and Structural Explanations.” American Political Science Review 93: 575–90. Rojas Samanez, Alvaro. 1987. Partidos Políticos en el Perú. Lima: Industrial Gráfica. Rosas, Guillermo. 2005. “The Ideological Organization of Latin American Legislative Parties: An Empirical Analysis of Elite Policy Preferences.” Comparative Political Studies 38: 824–49. Rosenbaum, P. 2002. Observational Studies. 2nd. ed. New York: Springer-Verlag. Rosenbaum, P., and Donald B. Rubin. 1983. “The Central Role of the Propensity Score in Observational Studies for Causal Effects.” Biometrika 70: 41–55. Rosenblum, Nancy L. 2008. On the Side of the Angels: An Appreciation of Parties and Partisanship. Princeton: Princeton University Press. Rosenstone, Steven J., Roy L. Behr, and Edward H. Lazarus. 1996. Third Parties in America. 2nd ed. Princeton, N.J.: Princeton University Press. Rueschemeyer, Dietrich, Evelyne Huber Stephens, and John D. Stephens. 1992. Capitalist Development and Democracy. Chicago, Ill.: University of Chicago Press. Sanders, David. 2003. “Party Identification, Economic Perceptions, and Voting in British General Elections, 1974–1997.” Electoral Studies 22(2): 239–63. Sanin, Alfredo Tarre Murci. 1989. Los Muertos de la Deuda o el Final de la Venezuela Saudita. Caracas: Ediciones Centauro. Sartori, Giovanni. 1976. Parties and Party Systems: A Framework for Analysis. Cambridge: Cambridge University Press. Scarrow, Susan E. 1996. Parties and Their Members: Organizing for Victory in Britain and Germany. Oxford: Oxford University Press. Schattschneider, E. E. 1942. Party Government. New York: Rinehart. Schmidt, Gregory D. 1996. “Fujimori’s 1990 Upset Victory in Peru: Electoral Rules, Contingencies, and Adaptive Strategies.” Comparative Politics 28(3) (April): 321–54. Schmitt, Hermann, and Soren Holmberg. 1995. “Political Parties in Decline?” In Citizens and the State, ed. Hans-Dieter Klingemann and Dieter Fuchs. Oxford: Oxford University Press. Seawright, Jason. 2006. “Crisis of Representation: Voters, Party Organizations, and Party System Collapse in South America.” PhD dissertation, University of California, Berkeley. Seligson, Amber L. 2003. “Disentangling the Roles of Ideology and Issue Positions in the Rise of Third Parties: The Case of Argentina.” Political Research Quarterly 56: 465–75. Sheahan, John. 1999. Searching for a Better Society: The Peruvian Economy from 1950. University Park: Pennsylvania State University Press. Shugart, Matthew Soberg, and Scott Mainwaring. 1997. Presidentialism and Democracy in

bibliogr aphy

273

Latin America: Rethinking the Terms of the Debate. In Presidentialism and Democracy in Latin America, ed. Scott Mainwaring and Matthew Soberg Shugart. Cambridge: Cambridge University Press. Sniderman, Paul M., Richard Brody, and Philip E. Tetlock. 1991. Reasoning and Choice: ­Explorations in Political Psychology. Cambridge: Cambridge University Press. Stern, Steve J. 1998. Shining and Other Paths: War and Society in Peru, 1980–1995. Durham, N.C.: Duke University. Stokes, Donald E. 1966. “Party Loyalty and the Likelihood of Deviating Elections.” In Elections and the Political Order, ed. Angus Campbell, Philip E. Converse, Warren E. Miller, and Donald E. Stokes. New York: John Wiley and Sons. Stokes, Susan Carol. 2001. Mandates and Democracy: Neoliberalism by Surprise in Latin America. Cambridge: Cambridge University Press. ———.2005. “Perverse Accountability: A Formal Model of Machine Politics with Evidence from Argentina.” American Political Science Review 99: 315–25. ———,ed. 2001. Public Support for Market Reforms in New Democracies. Cambridge: Cambridge University Press. Strom, Kaare. 1990. “A Behavioral Theory of Competitive Political Parties.” American Journal of Political Science 34: 565–98. Sundquist, James L. 1983. Dynamics of the Party System: Alignment and Realignment of Political Parties in the United States. Rev. ed. Washington, D.C.: Brookings Institution. Tanaka, Martin. 1998. El Espejismo de la Democracia: El Colapso Del Sistema de Partidos En el Peru, 1980–1995, En Perspectiva Comparada. Lima: Instituto de Estudios Peruanos. ———.2006. “From Crisis to Collapse of the Party Systems and Dilemmas of Democratic Representation: Peru and Venezuela.” In The Crisis of Democratic Representation in the Andes, ed. Scott Mainwaring, Ana Maria Bejarano, and Eduardo Pizarro Leongomez. Stanford, Calif.: Stanford University Press. Tanner, Duncan M. 1990. Political Change and the Labour Party, 1900–18. Cambridge: Cambridge University Press. Tedin, Kent L. 1976. “On the Reliability of Reported Political Attitudes.” American Journal of Political Science 20: 117–24. Thorp, Rosemary. 1991. Economic Management and Economic Development in Peru and Colombia. Pittsburgh, Penn.: University of Pittsburgh Press. Tiedens, Larissa Z. and Susan Linton. 2001. “Judgment Under Emotional Certainty and Uncertainty: The Effects of Specific Emotions on Information Processing.” Journal of Personality and Social Psychology 81(6): 973–88. Transparency International. 2005. “Corruption Perceptions Index.” Technical report. http:// ww1.transparency.org/surveys/index.html. Trinkunas, Harold A. 2002. “The Crisis in Venezuelan Civil-Military Relations: From Punto Fijo to the Fifth Republic.” Latin American Research Review 37(1): 41–76. Tuesta Soldevilla, Fernando. 1994. Perú Político en Cifras: Elite Política y Elecciones, 2nd Edition. Lima: Fundación Friedrich Ebert. Tuesta Soldevilla, Fernando. 1995. Sistema de Partidos Políticos en el Perú. Lima: Fundación Friedrich Ebert. Valentino, Nicholas A., Vincent L. Hutchings, Antoine J. Banks, and Anne K. Davis. 2008. “Is a Worried Citizen a Good Citizen? Emotions, Political Information Seeking, and Learning via the Internet.” Political Psychology 29(2): 247–73. Wattenberg, Martin P. 1996. The Decline of American Political Parties: 1952–1994. Cambridge, Mass.: Harvard University Press.

274 bi bl io g r a ph y

Weber, Shlomo. 1992. “On Hierarchical Spatial Competition.” Review of Economic Studies 59: 407–25. Westholm, Anders. 1999. “The Perceptual Pathways: Tracing the Mechanisms of Political Value Transfer across Generations.” Political Psychology 20: 525–51. Weyland, Kurt. 1996. “Neopopulism and Neoliberalism in Latin America: Unexpected Affinities.” Studies in Comparative International Development 31(3): 3–31. Weyland, Kurt. 2002. The Politics of Market Reform in Fragile Democracies: Argentina, Brazil, Peru, and Venezuela. Princeton, N.J.: Princeton University Press. Wiarda, Howard J. 2004. Authoritarianism and Corporatism in Latin America–Revisited. Gainesville: University Press of Florida. Wibbels, Erik. 2000. “Federalism and the Politics of Macroeconomic Policy and Performance.” American Journal of Political Science 44: 687–702. Wildavsky, Aaron. 1965. “The Goldwater Phenomenon: Purists, Politicians, and the TwoParty System.” Review of Politics 2: 386–413. Wilson, James Q. 1962. The Amateur Democrat: Club Politics in Three Cities. Chicago: University of Chicago Press. Wise, Carol. 1994. “The Politics of Peruvian Economic Reform: Overcoming the Legacies of State-Led Development.” Journal of Interamerican Studies and World Affairs 36: 75–125. ———.2003. Reinventing the State: Economic Strategy and Institutional Change in Peru. Ann Arbor: University of Michigan Press. Wittenberg, Jason. 2006. Crucibles of Political Loyalty: Church Institutions and Electoral Continuity in Hungary. Cambridge: Cambridge University Press. Wlezien, Christopher, Mark Franklin, and Daniel Twiggs. 1997. “Economic Perceptions and Vote Choice: Disentangling the Endogeneity.” Political Behavior 19(1): 7–17. Wolinetz, Steven B. 2002. “Beyond the Catch-All Party: Approaches to the Study of Parties and Party Organization in Contemporary Democracies.” In Political Parties: Old Concepts and New Challenges, ed. Richard Gunther, José Ramón Montero, and Juan J. Linz. Oxford: Oxford University Press. Woodward, James. 2003. Making Things Happen: A Theory of Causal Explanation. Oxford: Oxford University Press. World Bank. 2009. “World Development Indicators.” Technical report. Washington, D.C. Yashar, Deborah J. 1995. “Civil War and Social Welfare: The Origins of Costa Rica’s Competitive Party System.” In Building Democratic Institutions: Party Systems in Latin America, ed. Scott Mainwaring and Timothy R. Scully. Stanford, Calif.: Stanford University Press. Zaller, John R. 1992. The Nature and Origins of Mass Opinion. Cambridge: Cambridge University Press. Zambrano Sequin, Luis. 1995. “What We Have Done and What We Can Still Do in Economic Policy.” In Venezuelan Democracy Under Stress, ed. Jennifer McCoy, William C. Smith, Andres Serbin, and Andres Stambouli. Miami: North-South Center Press. Zechman, Martin. 1979. “Dynamic Models of the Voter’s Decision Calculus.” Public Choice 34: 297–315. Zechmeister, Elizabeth. 2002. “What’s Left and Who’s Right in Mexican Politics?” Paper presented at Conference of the Midwest Political Science Association, Chicago, Ill.

I NDE X

Page numbers in italics refer to figures; those followed by t refer to tables; those followed by n refer to notes, with note number. Abramson, Paul R., 136 Acción Democrática (Venezuela): activists’ ties to civil-society organizations, 182; alliance with Caldera, 69; banning of, 44; and Chávez, opposition to, 44; collapse of, 44; economic performance, and vote share, 78; and election of 1993, 123t, 127, 128, 129; and election of 1998, 130–31, 132–33, 132t, 133–34; failure of ideological repositioning, 167; history of, 43–45; ideological diversity of activists and workers, 186, 186t ; ideological flexibility, 167, 191, 194; and ideological representation, 114, 115, 124; labor union influence on, 185; leaders’ relative pragmatism, 177; membership and outreach organizations’ relative complexity, 188–89; members’ relative contribution to party financing, 178; movement toward ideological center, 44; and neoliberal reforms, 86; and normal electoral volatility, 50; and Pact

of Punto Fijo, 44; and party-branded patronage, 110t ; party leaders’ relative experience level, 176; relative rigidity of decision-making procedures, 179– 81; relative size of party leadership core, 175; relative size of party membership, 174; resistance to Pérez reforms, 69; as traditional party, 43–45; voter anger at, 147; voter identification with, trends over time, 92, 92, 100–101 Acción Popular (Peru): alliance with Vargas Llosa, 166; collapse of, 41; economic performance, and vote share, 78; failure of ideological repositioning, 166; history of, 40–41; ideological diversity of activists and workers, 186t ; ideological flexibility, 167, 194; ideological position of, 121; labor union influence on, 185; membership and outreach organizations’ relative complexity, 189; members’ relative contribution to party financing, 178; and

276 index

party-branded patronage, 110t ; party leaders’ relative experience level, 176; poor response to ideological underrepresentation, 24; as relatively uninstitutionalized party, 47; relative rigidity of decision-making procedures, 179–81; relative size of party leadership core, 175; support for Vargas Llosa, 41; as traditional party, 40–41; voter identification with, trends over time, 93, 93 activists, ideological diversity of: as base for diverse candidate pool, 25, 29, 169, 196–97, 199; and ideological flexibility of parties, 25, 169, 185–86, 186t, 193– 94, 196–97, 242; methods of establishing, 169 activists’ ties to civil-society organizations: advantages of, 182; and ideological flexibility, 182–83, 193 AD. See Acción Democrática Addams, Jane, 109 affect-and voting theory empirical test, 149–64; on anger and risk aversion, 159–61, 160t ; anger treatment, 151; anger treatment, effectiveness of, 155– 59, 157; anxiety treatment, 151; anxiety treatment, effectiveness of, 156, 161; applicability to Venezuela, 153; experimental design, 149–56; external validity of, 150; findings, 156–59, 157; implications for party-system change, 161–63; individual level design, advantages of, 149–50; mechanics of, 150–52; Peru as context for, 152–54, 157; postexperimental survey, 161–62, 162t ; randomization in, 149, 150; sample for, 154–55; sources of anger vs. anxiety, 161–62, 162t ; types of voter dissatisfaction, and likelihood of voting for outsider candidate, 158–59 affective factors: in citizen processing and retention of political information, 14, 21–22; in decision to abandon party system, 14–18, 29, 145, 156–61. 160t, 242. See also anger; anxiety/fear Aldrich, John H., 241–42 Alfaro Ucero, Luís, 130, 131, 131t Alfonsín, Raúl, 37

Alianza Popular Revolucionaria Americana (APRA; Peru): banning of, 38–39; collapse of, 40; economic performance, and vote share, 78; in election of 1985, 165–66; history of, 38–40; ideological diversity of activists and workers and, 186, 186t ; ideological flexibility, 167, 194; ideological position of, 121; and ideological repositioning, 165, 168; and ideological representation, 24, 114, 115; labor union influence on, 185; in Lima municipal elections of 1989, 138t, 139, 140, 141; membership and outreach organizations’ relative complexity, 189; members’ relative contribution to party financing, 178; movement toward center, 39, 40; and neoliberal reforms, 86; and partybranded patronage, 110t ; party leaders’ relative experience level, 176; relative rigidity of decision-making procedures, 179–81; relative size of party leadership core, 175; relative size of party membership, 174; revival of, 40; as traditional party, 38–40; violent activism by, 38, 39; voter anger at, 148; voter identification with, trends over time, 93, 93, 105–6 Alva Castro, Luís, 24 Alvarez Paz, Osvaldo, 124, 125t ANES survey, 225 anger: corruption scandals as generators of, 88, 142, 161–62, 162t, 164, 242; ideological underrepresentation as generator of, 142, 161–62, 162t, 164, 242; methods of diffusing, 242; in Peru, qualitative evidence of, 147–49; and risk acceptance, in party abandonment decision, 12, 15–18, 29, 145, 156– 61, 157, 160t, 242; in Venezuela, qualitative evidence of, 146–47. See also affect-and voting theory empirical test anti-statist rhetoric preceding party collapse, 27–28; and citizen preference for less expansive role of government, 27–28, 202, 203–4, 205–7, 241; and increased perception of individual

index 277

leaders’ effectiveness, 221–22; lack of causative link, 207 anxiety/fear: crime as generator of, 162; methods of defusing, 199; as precursor to party-system collapse, 80, 84–85, 84t ; and risk acceptance, in party abandonment decision, 12, 14, 15–17, 63, 72, 80, 84–85, 84t, 164, 242, 243; and voter attention to political issues, 80, 84. See also affect-and voting theory empirical test anxiety/fear, economic crises and, 80–85, 161–62, 162t, 163–64; affect-and voting theory empirical test on, 161–62, 162t ; personal vs. national crises, impact of, 80–81; statistical analysis, 81–85 Apoyo survey, 137 APRA. See Alianza Popular Revolucionaria Americana Argentina: anti-statist rhetoric in, 207; citizen’s perception of major national problems, 216; corruption level in, 99t, 100; currency crisis of 2001, 38; economic performance, 71; economic performance, and party stability, 72– 73, 73–74; electoral reforms of 1912, 37; media exposure levels in, 216; military governments, 37; party system changes in, 3–4; period of party-system collapse, weathering of, 36; political diversity in, 169; presidential power in, 56; relative political participation levels, 217; traditional parties in, 34– 38; traditional parties’ presidential election vote share, by year, 53–54, 54 Argentina’s government institutions, limited value in explaining party system continuity, 55–58 Argentina’s party system, 247n4; differences from other study systems, 217; as traditional party system, 46–47 Argentina’s political parties: activists’ ties to civil-society organizations, 183; broad ideological appeal of, 167; ideological diversity of activists and workers and, 186, 186t ; ideological flexibility of, 167, 191–98, 192t ; ideological underrepresentation in, 114,

116, 116–17, 219–20; labor union influence on, 185; leaders’ relative pragmatism, 177; levels of recruitment via particularist benefits, 184; membership and outreach organizations’ relative complexity, 189; members’ relative contribution to party financing, 178; party leaders’ relative experience level, 176; relative levels of party-branded patronage, 190; relative rigidity of decision-making procedures, 179–81; relative size of party leadership core, 175; relative size of party membership, 174; traditional parties, 34–38 Argentine citizens’ experience of politics, comparative analysis, 218–32; and external political efficacy, 223t, 224, 226, 236; government management of economy, 220; and government’s role in solving problems, 226–29, 226t ; ideological representation by political parties, 219–20; and internal political efficacy, 223t, 225–26; and nonparticipation in politics, legitimacy of, 223t, 224–25; and party identification, 222– 23, 223t ; president’s problem-solving efforts, 221; retrospective analysis of effectiveness of state, 219; views on effectiveness of citizen involvement, 223–24, 224t Auyero, Javier, 110t, 190 Baloyra, Enrique, 101 Barrantes Lingón, Alfonso, 23, 43, 166–67 basic needs, citizens’ views on government role in providing, 226t, 227–28 BATOBA 83 survey (Baloyra and Torres), 101 Belaúnde Terrry, Fernando: and Acción Popular, founding of, 40; economic policies, damage done by, 65; election as president, 40, 41; and election of 1990, 166; inability to move party ideological stance, 24 Belmont, Ricardo, 138 Benavides, Oscar R., 39 Benton, Allyson Lucinda, 75 Bloc Québécois party (Canada), 52–53

278 index

Bolivia: corruption level in, 99t, 100; economic performance, 70, 71; economic performance, and party stability, 72–73 Bolivia’s party-system collapse, 49; correlation with negative economic growth, 70–71 Brazil: corruption level in, 99t, 100; economic performance, 71; economic performance, and party stability, 72– 73; net electoral volatility scores, 3 Bush, George W., Chávez on, 2 Bustamante y Rivero, José Luis, 39 Caldera, Rafael: alliance with Acción Democrática, 69; anti-statist rhetoric by, 206; economic policies, 69; economic policies, and election of 1998, 130, 133; and election of 1993, 45, 123, 124, 125t, 167; failure of ideological repositioning, 167; Pérez’s house arrest under, 146; successful ideological repositioning of, 167; voter anger at, 147 Cameron, Maxwell A., 5–6 Canada’s party system, replacement of multiple parties in, 52–53 candidate pool, diverse: diverse body of activists and workers as base for, 25, 29, 169, 196–97, 199; and ideological flexibility, 25, 29, 168–69, 194, 198–99 candidates, outsider: voter fear of risks associated with, 11, 17, 144; voter selection of, 20 candidate selection process at lower levels: as indication of party decision-making rigidity, 179–80; as indication of shift in ideology, 179 capital-intensive party organizations, 187 Caracazo, 59–60, 146 Cárdenas, Arias, 60 Carmines, Edward G., 51 La Causa R (LCR; Venezuela): and election of 1993, 123t, 124, 128, 129; growth of, 7 center elites, framing of party-system collapse by, 27–28 Centro de Investigaciones y Estudios Politicos y Administrativos, 101, 124 Chávez Frías, Hugo: anti-statist rhetoric

by, 206–7; appeal of to subaltern sectors, 5; committed supporters, and levels of presidential approval, 221; confrontational populist leftism of, 2; coup attempt against, 201; dramatic political changes instituted by, 2; political parties’ opposition to, 8, 44, 130–31; rhetoric on citizen participation, and citizens’ experience of politics after party-system collapse, 202, 207–9, 214, 225, 226, 230, 235, 237–41; and Venezuelans’ evaluation of president’s problem-solving efforts, 231–32 Chávez rise to power: and corruption of traditional parties, 17–18; and election of 1998, 130–31, 131t, 134; lack of local government experience, 57; and partysystem collapse, 2, 3, 32; as product of coup attempt, 60 Chávez supporters, external political efficacy, vs. non-Chavistas, 236–41, 239t, 240t Chile: anti-statist rhetoric in, 207; citizen’s perception of major national problems, 216; corruption level in, 99t ; economic performance, 71; economic performance, and party stability, 72; media exposure levels in, 216; party system, differences from other study systems, 217; political parties in, 3; relative political participation levels, 217; traditional parties’ presidential election vote share, by year, 53–54, 54 Chilean citizens’ experience of politics, comparative analysis, 218–32; and external political efficacy, 223t, 224, 226, 236; government management of economy, 220; and government’s role in solving problems, 226–29, 226t; ideological representation of state, 219–20; and internal political efficacy, 223t, 225–26; and nonparticipation in politics, legitimacy of, 223t, 224–25; and party identification, 222–23, 223t; president’s problem-solving efforts, 221; retrospective analysis of effectiveness of state, 219; views on effectiveness of citizen involvement, 223–24, 224t

index 279

Christian Democratic Party (Italy), 53 CIRELA survey, 218 citizens’ experience of politics, party system collapse and, 26–28; Chávez’s rhetoric on citizen participation and, 202, 207–9, 214, 225, 226, 230, 235, 237–41; comparative analysis, without confounding factors, 218–32; comparative analysis method for, 202, 214–15; complexity of, 202; confounding factors, 214–18; elite framing of events, as likely largest influence, 205; experience of party system failure, and loss of faith in government action, 28, 203–4, 209, 210, 212–13, 226–31, 226t ; experience of successful mass action, and perception of external efficacy, 204, 210, 226, 229–30, 237, 238; and external political efficacy, 223t, 224, 226; and external political efficacy, conditioning on confounding variables, 232–35, 233t ; government management of economy, 220; and government’s role in solving problems, 226–29, 226t ; ideological representation of state, 219–20, 222; and increased perception of individual leaders’ effectiveness, 221; individual evaluations of political and state institutions, limited duration of, 210–11, 222, 231; and internal political efficacy, 223t, 225–26; and nonparticipation in politics, legitimacy of, 223t, 224–25; and outsider elite ideology and message, exposure to, 204–5, 207–8, 210; and party identification, 211–12, 222–23, 223t ; political efficacy, impact on, 213–14; president’s problem-solving efforts, 221; primary changes, 202; retrospective analysis of effectiveness of state, 219, 222; salient factors, identification of, 205, 209–10; salient features, 231–32; views on effectiveness of citizen involvement, 223–24, 224t citizen’s political opinions: evaluations of political and state institutions, limited duration of, 210–11, 222, 231; informa-

tion gathering methods for, 27; as product of elite interpretations, 27– 28, 29, 205, 241 class conflict, as source of party-system collapse, 5–6 Colombia: corruption level in, 99t, 100; economic performance, 71; economic performance, and party stability, 73; party system, realignments in, 51 Colomer, Josep M., 118 Comité de Organización Política Electoral Independiente. See COPEI communication flow within party, and ideological flexibility, 181, 191 Communist Party (Italy), 53 Conservative party (Ecuador), 72 contractions of party system, as form of electoral change, 51–52 Convergencia (Venezuela): and election of 1993, 123t, 128, 129; growth of, 7; and ideological representation, 124 COPEI (Comité de Organización Política Electoral Independiente; Venezuela), 45–46; activists’ ties to civil-society organizations, 182; collapse of, 45–46; corruption scandals, 45; and election of 1993, 123t, 127, 128, 129; and election of 1998, 130; failure of ideological repositioning, 167; history of, 45–46; ideological diversity of activists and workers and, 186t ; ideological flexibility, 167, 191, 194; and ideological representation, 114, 115, 124; labor union influence on, 185; leaders’ relative pragmatism, 177; as major party, 45– 46; membership and outreach organizations’ relative complexity, 188–89; members’ relative contribution to party financing, 178; and partybranded patronage, 110t ; party leaders’ relative experience level, 176; relative rigidity of decision-making procedures, 179–81; relative size of party leadership core, 175; relative size of party membership, 174; voter anger at, 147; voter identification with, trends over time, 91–92, 92, 100–101 Coppedge, Michael, 8, 118

280 index

core social identities, stability of, and party stability, 4 Corrales, Javier, 6–7 corruption: measures of, 98–99; perception of, left-right ideological spectrum and, 122–23 corruption scandals, voter responses to: alternatives to party abandonment, 14; anger as, 88, 142, 161–62, 162t, 164, 242; ubiquity of corruption, and decreased likelihood of strong reaction, 14 corruption scandals as impetus for party abandonment, 5, 8, 11, 12, 16, 18, 28, 88, 112, 163, 164, 242, 243; affect-and voting theory empirical test on, 158; in Argentina, 107–8; as indirect cause, 97–98, 133; methods of mitigating, 199; as opportunity for outsider parties, 135–36; perception of self-serving party and, 90, 95–96, 109, 111; in Peru, 17–18, 97, 100, 105–7, 106t, 140–41, 142; simultaneous involvement of all major parties, 18; in Venezuela, 8, 17– 18, 92, 97, 100–105, 103t, 123–24, 127, 128, 130, 133, 135, 136, 141, 142 corruption scandals in Argentina, as cause of party abandonment, 107–8 corruption scandals in Peru: as cause of party abandonment, 17–18, 97, 100, 105–7, 106t, 140–41, 142; and Fujimori’s loss of power, 3; level of, vs. Latin America, 97–100, 99t corruption scandals in Venezuela: in Caldera administration, 130; as cause of party abandonment, 8, 17–18, 92, 97, 100–105, 103t, 123–24, 127, 128, 130, 133, 135, 136, 141, 142; as issue in election of 1993, 124; level of, vs. Latin America, 97–100, 99t Costa Rica: corruption level in, 99t; economic performance, 71; party system, normal volatility in, 50; political parties in, 3; traditional parties’ presidential election vote share, by year, 53–54, 54 crime: citizens’ views on government role in preventing, 226t, 228; as generator of fear, 162; and voter abandonment of party system, 158

critical elections (realigning elections), 50–51 debt crises, and party system volatility, 3 decision-making procedures, rigidity of: measures of, 178–81; and party ideological flexibility, 178–81 deductive microfoundations, uses of, 243 de la Rúa, Fernando, 36, 108 democracy: consequences of party-system collapse for, 26; economic crises and party-system collapse in, 79; importance of political parties for, 4, 241–42 Deschouwer, Kris, 188 Dietz, Henry A., 9, 22 directional vs. proximity voting, 117–18 disappointment with political status quo: alternatives to party abandonment, 13–14, 144; contextual factors in voter response to, 144 Dominican Republic: corruption level in, 99t ; economic performance, 71; economic performance, and party stability, 72, 73 Dornbusch, Rudiger, 65 Downs, Anthony, 4, 171–72 DOXA 93 survey (Centro de Investigaciones y Estudios Politicos y Administrativos), 101 Duverger’s Law, 48–49 economic crises and anxiety in voters, 80–85, 161–62, 162t, 163–64; affectand voting theory empirical test on, 161–62, 162t ; personal vs. national crises, impact of, 80–84; statistical analysis, 81–85 economic crises and party identification decline, limited correlation between, 95, 104, 163, 243 economic crises and party-system collapse: anxiety/fear created by crisis and, 12, 14, 15–17, 63, 72, 80, 84–85, 84t, 164, 242, 243; crises’ weakening of government and, 63; in democracies vs. partial democracies, 79; partial causal relationship in, 63, 72, 79, 128, 132–33; possible additional factors in, 79;

index 281

prior relevant studies, 74–75; retrospective economic voting and, 6–7; statistical analysis, 75–79; in Venezuela and Peru, in Latin America context, 70–74, 71. See also neoliberal reforms and party system volatility economic crisis in Peru, 64–66; as cause of voter choice, 139, 141–42; in context of other Latin American economies, 70– 74, 71; Fujimori’s economic reforms and, 66; similarity to other Latin American crises, 69–70 economic crisis in Venezuela, 64, 66–70; as cause of party abandonment, 8, 132– 33; as cause of voter choice, 125t, 127– 28, 132–33, 132t ; in context of other Latin American economies, 70–74, 71; and party identification, limited correlation between, 104; similarity to other Latin American crises, 69–70 economic evaluations by voters: accuracy of, 220; as cause of voter choice in Peru, 139, 141–42; as cause of voter choice in Venezuela, 125t, 127–28, 132– 33, 132t, 141–42; and left-right ideological spectrum, 122–23 economic motives for party involvement, and party ideological flexibility, 184 economic restructurings, and party system volatility, 3 An Economic Theory of Democracy (Downs), 171–72 Ecuador: corruption level in, 99t, 100; economic performance, 70, 71; economic performance, and party stability, 71–72; political parties in, 3 education: and voter abandonment of party system, 158; and voter choice in Venezuela, 129–30 Edwards, Sebastian, 65 elections, realigning (critical), 50–51 electoral change, types and degrees of, 50–53 electoral institutions: discontinuities in, and electoral volatility, 55; limited value in explaining collapse, 55–58 electoral irrelevance: as criteria for party collapse, 48; definition of, 48–49

electoral rules, influence on voting patterns, 75 elite institutions, disruption of by partysystem collapse, 26, 201 elite outsiders: tensions with traditional elites, 26, 201; voter fear of risks associated with, 11, 17, 144; voter vetting of, 20 elite outsiders’ framings of party-system collapse: downplaying of successful mass action by, 204; exposure to after party system collapse, and voter experience of politics, 204–5; and perception of government incapacity, 27–28, 29, 202–7, 241 elites, consequences of party-system collapse for, 26, 201 elites’ framings of party-system collapse, influence on citizens’ perception, 27– 28, 29, 202–7, 241; Chávez’s rhetoric on citizen participation and, 202, 207–9, 214, 225, 226, 230, 235, 237–41 elite-voter interactions: as essential facet of party-system collapse, 32–33; as focal point for identifying party-system collapse, 49 El Salvador: corruption level in, 99t ; economic performance, 71 Europe’s party system: expansion of, 51–52; stable system of 1950s, 50 evaluations of political and state institutions, limited duration of, 210–11, 222, 231 expansion of party system, as form of electoral change, 51–52 external political efficacy: experience of successful mass action during partysystem collapse and, 204, 210, 226, 229–30, 237, 238; factors affecting, during party-system collapse, 213–14 fear. See anxiety/fear federalism, limited value in explaining party-system collapse, 56–57 Fermín, Claudio, 65, 124, 125, 125t Fernández, Eduardo, 167 FREDEMO. See Frente Democrático Nacional

282 index

government experience, local opportunities for, as factor in party-system collapse, 56–57 Green, Donald, 90, 251n9 Greene, Kenneth F., 176–77 growth in Latin America nations, 71; correlation with party-system change, 70–72, 75, 77, 78 growth rate, and anxiety, creation of, 80–85 Guatemala: corruption level in, 99t ; economic performance, 70, 71; economic performance, and party stability, 73; party system instability, correlation with negative economic growth, 70–71 Guatemalan Republican Front (Guatemala), 71 Guzmán, Abimael, 61

Freedman, David A., 251n9 Frente Democrático Nacional (FREDEMO; Peru): history of, 39, 40; and ideological representation, 115, 116; in Lima municipal elections of 1989, 138t, 139, 140, 141 El Frente País Solidario. See FREPASO FREPASO (El Frente País Solidario; Argentina): coalition with Radicals, 36, 38; disappearance of, 38; and expansion of Argentine party system, 52; and ideological representation, 116, 117; as minor party, 38 Fujimori, Alberto: anti-democratic coup by, 2–3, 201; anti-statist message of, 206; and citizen experience of politics, 225, 230; economic reforms by, 2, 66; and election of 1990, 138, 206; lack of local government experience, 57; loss of power, 3; as nontraditional politician, 153; political parties’ opposition to, as self-destructive, 8; successful ideological positioning by, 166; views on, left-right ideological spectrum and, 121 Fujimori, rise to power of: and corruption of traditional parties, 18; parties’ ineffective response to ideological underrepresentation and, 24; and party-system collapse, 3, 32

Haya de la Torre, Víctor Raúl, 38, 39, 40 health care, citizens’ views on government role in provision of, 226t, 229 Herrera Campíns, Luís, 67–68 Hirschman, Albert O., 14 Honduras: corruption level in, 99t ; economic performance, 70, 71; economic performance, and party stability, 72 Humala, Ollanta, 153

García Pérez, Alan: and APRA’s poor response to ideological underrepresentation, 24; and APRA’s revival, 40; corruption scandals, and collapse of APRA support, 106, 140; economic policies, damage done by, 65–66; in election of 1985, 165–66; and election of 2006, 168; and ideological repositioning, 165–66, 168; as president, 40; rise to power of, 40; voter anger at, 148–49 generational replacement, and decline in party identification, 90, 91, 101 government capacity: loss of faith in, experience of party system failure, and, 28, 203–4, 209, 210, 212–13, 226–31, 226t ; perception of, and voter abandonment of party system, 158

ideological change: as cause of party identification decline, 96, 96–97; neoliberal reforms as impetus for, 86–87 ideological conversions in Venezuela, rarity of, 101 ideological distribution within population, stability of, and party stability, 4 ideological diversity of activists and workers, and ideological flexibility, 25, 169, 185–86, 186t, 193–94, 196–97, 242 ideological diversity of candidate pool: diverse body of activists and workers as base for, 25, 29, 169, 196–97, 199; and ideological flexibility, 25, 29, 168–69, 194, 198–99 ideological exclusion, potential costs of, 243–44

index 283

ideological flexibility of parties: in Argentina, 167, 191–98, 192t ; failure of, as issue, 165, 167–68, 198; in local elections, 194, 197, 197–98; in Peru, 165– 67, 191–98, 192t ; ranking of, 167; in Venezuela, 167, 191–98, 192t ideological flexibility of parties, party organization dimensions and, 171–90; activists’ ties to civil-society organizations and, 182–83, 193; and communication flow within party, 181, 191; ideological diversity of activists and workers and, 25, 169, 185–86, 186t, 193–94, 196–97, 242; as issue, 172; labor union influence and, 184–85, 193; local leaders’ influence on national decisions and, 180, 193; local party unit comparisons, counterfactual evidence provided by, 193–98, 196t, 197t ; mass-party model of organization and, 172; members’ contribution to party financing and, 177–78, 191; membership and outreach organizations’ complexity and, 187–89, 193– 94, 197–99, 242; method of analysis for, 171; national leaders’ control over local parties and, 178–81, 193; partybranded patronage and, 189–90, 193– 94, 197–99; party leaders’ experience level and, 175–76, 193; party leaders’ pragmatism and, 176–77, 191; prerequisites for flexibility, 25–26, 168–71; recruitment via particularist benefits and, 183–84, 193; scholarship on, 172; selecting most relevant dimensions, 191–93, 192t ; significant dimensions, 198–99; size of leadership core and, 174–75, 193; size of party membership and, 173–74, 191 ideological spectrum. See left-right ideological spectrum ideological underrepresentation: as generator of anger, 142, 161–62, 162t, 164, 242; and voter choice, 113 ideological underrepresentation as impetus for party abandonment, 5, 7–8, 11, 12, 18, 28–29, 113–42, 242; ideological distribution of dissatisfied voters and,

19–20; in Lima municipal election of 1989, 136–42; methods of mitigating, 199; party responses and, 19–20; in Peru, 17–18, 136–41, 142; simultaneous neglect of all major parties and, 18; in Venezuela, 17–18, 114, 115, 142; in Venezuelan election of 1993, 124–30; in Venezuelan election of 1998, 130–36, 132t ideological underrepresentation in Argentina, 114, 116, 116–17, 219–20 ideological underrepresentation in Peru, 114–16, 115; as impetus to party abandonment, 17–18, 136–41, 142; parties’ poor response to, 23–25 ideological underrepresentation in Venezuela, and party abandonment, 17–18, 114, 115, 142; in election of 1993, 124– 30; in election of 1998, 130–36, 132t ideology, leftist: parties’ failure to represent, as cause of party collapse, 7–8; and traditional party abandonment, 128–29, 130, 133–34, 135 inflation: and anxiety, 80–85; correlation with party-system change, 72–73, 74, 77, 78; in Latin America nations, 71 informal economy, as source of party-system collapse: in Peru, 6; in Venezuela, 134–35 internal political efficacy, factors affecting, during party-system collapse, 213–14 Italy, party-system collapse in, 49, 53, 56 Izquierda Democrática (Ecuador), 71 Izquierda Unida (United Left ; Peru): activists’ ties to civil-society organizations, 182–83; as coalition, 42; in election of 1985, 165–66; and election of 1990, 166–67; failure of ideological repositioning, 165–66, 166–67; history of, 41–43; ideological diversity of activists and workers and, 186t ; ideological flexibility, 167, 191, 194; ideological position of, 121; and ideological representation, 114, 115; internal conflict and eventual division of, 43; labor union influence on, 185; leaders’ relative pragmatism, 177; in Lima municipal elections of 1989, 138t, 139, 140; mem-

284 index

bership and outreach organizations’ relative complexity, 189; members’ relative contribution to party financing, 178; and party-branded patronage, 110t ; in period of party collapse, 42– 43; poor response to ideological underrepresentation, 23; as relatively uninstitutionalized party, 47; relative rigidity of decision-making procedures, 179–81; relative size of party leadership core, 175; relative size of party membership, 174; as traditional party, 41–43; voter identification with, trends over time, 93, 93 Keesing’s Record of World Events, 98 Kenney, Charles D., 48 King, Gary, 117 Kirchner, Néstor, 36, 167 Kitschelt, Herbert, 9–10, 22, 118–19, 173, 188 labor-intensive party organizations, and ideological flexibility, 187–88 labor unions: in Argentina, declining influence under renewal Peronism, 36; as characteristic of Latin America politics, 184; and ideological flexibility, 184–85, 193; Venezuelan, opposition to reforms, 68 Latin America: economic performance, Peruvian and Venezuelan performance in context of, 70–74, 71; net electoral volatility scores in, 3; political party volatility in, 3–4, 50 Latin American Weekly Report (LAWR), 98, 99, 99t LCR. See La Causa R leadership. See party leadership leftists: framing of party-system collapse by, 27–28; importance to party system collapse in Venezuela, 136; parties’ failure to represent, as cause of party collapse, 7–8; traditional party abandonment by, 128–29, 130, 133–34, 135; in Venezuela, percent of population, 115, 135–36 left-right ideological spectrum, 249n5; analysis methodology, 142–43; applica-

bility to Latin America, 118–23; in Argentina, 121–22; corruption perceptions and, 122–23; economic evaluations and, 122–23; in Latin America, complexity of, 117–23, 118; in Peru, 120–21, 122; scholars’ use of, 118– 19; in Venezuela, 120, 122 legislative electoral system, permissiveness of, and party-system collapse, 57 Levitsky, Steven, 6, 9–10, 22, 188 Lewis, Jeffrey B., 117 Liberal party (Canada), 52–53 Liberal party (Ecuador), 72 Lima municipal elections of 1989, 137–41; as proxy for first round of 1990 presidential elections, 137–38; traditional parties’ loss of support, causes of, 138– 41, 138t local party leaders: influence on national decisions, and party ideological flexibility, 180, 193; national leaders’ control over, and ideological flexibility, 178–81, 193 Lusinchi, Jaime, 67, 68 macroeconomics of populism, 65 Mainwaring, Scott P., 98 MAS (Movimiento al Socialismo), 46, 124 mass action, successful, experience of during party-system collapse: and increased sense of external efficacy, 204, 210, 226, 229–30, 237, 238; tension with loss of faith in government, 229–30 mass-party model of organization, and party ideological flexibility, 172 McGuire, James W., 247n4 media exposure, as confounding factor in cross-nation comparisons, 215–16 meeting influential people, as goal of party involvement, and party ideological flexibility, 184 members’ contribution to party financing, and ideological flexibility, 177–78, 191 membership and outreach organizations: complexity of, and ideological flexibility, 170, 187–89, 193–94, 197–99, 242; party-branded, types of, 187

index 285

Menem, Carlos: anti-statist rhetoric by, 207; election of, 35; government of, 36; ideological repositioning by, 167; Peronist party resistance to, 36; ­successful avoidance of party collapse, 36 Mexico: corruption level in, 99t, 100; economic performance, 71; economic performance, and party stability, 72, 73; party system, expansion of, 52; party system changes in, 3–4 microfoundations, deductive, uses of, 243 model of party-system collapse, 5, 12; necessity for including supply-side and demand-side factors in, 10, 26. See also theories of party-system collapse, in current literature Molina, Jose E., 47, 75, 76 Montesinos, Vladimiro, 3 Morales Bermúdez, Francisco, 65 Morgan, Jana, 7, 96, 96–97 Morgenstern, Scott, 246n9 Movimiento al Socialismo (MAS; Venezuela), 46, 124 Movimiento Libertad (Peru), 24, 245–46n8 Movimiento Revolucionario Túpac Amaru, 60 Movimiento V República (MVR; Venezuela), and election of 1998, 132–33, 132t, 134 multinomial logit model, 125–26, 249n7 multivariate regression analysis, 251n9 MVR. See Movimiento V República Myers, David J., 9, 22 National Advancement Party (Guatemala), 71 national problems, varying perceptions of, as confounding factor in cross-nation comparisons, 216 neoliberal reforms, 85–86; labor union resistance to, 184 neoliberal reforms and party system volatility, 3, 6–7, 85–87; as nation-specific phenomenon, 94; party ideological realignment as source of volatility, 86– 87, 95, 104; poor implementation as source of volatility, 86, 95, 104

neopopulist rule, as consequence of partysystem collapse, 26 New Democratic party (Canada), 52 Nicaragua: corruption level in, 99t ; economic performance, 70, 71; economic performance, and party stability, 72–73 Obras, in Lima municipal elections of 1989, 138t, 139, 140–41 organizational entrenchment, 173 organizational flexibility: necessary components of, 170; as prerequisite for ideological flexibility, 170 outreach organizations. See membership and outreach organizations outsider elites: tensions with traditional elites, 26, 201; voter fear of risks associated with, 11, 17, 144; voter vetting of, 20 outsider elites’ framings of party-system collapse: downplaying of successful mass action by, 204; exposure to after party system collapse, and voter experience of politics, 204–5; and perception of government incapacity, 27–28, 29, 202–7, 241 Pacek, Alexander, 75 Pact of Punto Fijo, 44 Palmquist, Bradley, 90 PAN. See Partido Acción Nacional Panama: corruption level in, 99t ; economic performance, 71; economic performance, and party stability, 72, 73; political parties in, 3 Panebianco, Angelo, 188 parliamentary systems, party-system collapse in, 56 Partido Acción Nacional (PAN; Mexico), 52 Partido Aprista Peruano (Peru): in affectand voting theory empirical test, 150; as dominant Peruvian party, 153 Partido de la Revolución Democrático (PRD; Mexico), 52 Partido Justicialista (PJ). See Peronist party Partido Popular Cristiano (PPC; Peru), 246–47n1

286 index

Partido Renovador Institucional Acción Nacional (Ecuador), 71 Partido Revolucionario Institucional (PRI; Mexico), 52 Partido Social Cristiano (Ecuador), 71 Partido Sociedad Patriótica (Ecuador), 71 parties: established, financial and organizational advantages of, 4; importance to democratic processes, 4, 241–42 parties, traditional: in Argentina, 34–38; definition of, 33–34; established ties to electorate as source of stability in, 34; in Peru, 38–43; in Venezuela, 43–46 party abandonment: alternatives to, 13–14, 144; leftist ideology and, 128–29, 130, 133–34, 135. See also corruption scandals as impetus for party abandonment; ideological underrepresentation as impetus for party abandonment party abandonment, risk acceptance in, 11–18, 12, 144; affective factors in, 14– 18, 29, 145, 156–61, 160t, 242; alternatives to party abandonment, 13–14, 144; anger and, 12, 15–18, 29, 145, 156– 61, 157, 160t, 242; anxiety/fear and, 12, 14, 15–17, 63, 72, 80, 84–85, 84t, 164, 242, 243; fervent single-issue voters, small number of, 13; naturally risk acceptant voters, small number of, 11; need for model of, 144; Peruvian party system and, 153; and rational-choice model, 14. See also affect-and voting theory empirical test party abandonment in Peru: corruption scandals as impetus for, 17–18, 97, 100, 105–7, 106t, 140–41, 142; ideological underrepresentation as impetus for, 17–18, 136–41, 142 party abandonment in Venezuela: corruption scandals as impetus for, 8, 17–18, 92, 97, 100–105, 103t, 123–24, 127, 128, 130, 133, 135, 136, 141, 142; ideological underrepresentation and, 17–18, 114, 115, 124–36, 132t, 142 party-branded patronage: as cause of voter support, 107, 108–11, 110t ; and diffusing of voter anxiety, 199; and diffusion of voter anger, 242; and ideological

flexibility, 25–26, 170–71, 189–90, 193– 94, 197–99 party decision-making processes: ability to incorporate new voices, and party flexibility, 25; rational-choice model in, 21–22 party elites, poor strategy of, as cause of party-system collapse, 8–9 party identification: in Argentina, trends over time, 94; as cause of vote choice, 89, 163; causes of, 211–12; in Chile, trends over time, 94; in Peru, trends over time, 93, 93; running tally account of, 14, 16, 90–91, 211; with traditional party, and likelihood of voting for outsider party, 126–27, 130; in Venezuela, importance of, 103–4; in Venezuela, trends over time, 91–92, 92 party identification decline: causes of, 90– 91, 95–100; as essential precursor to party-system collapse, 5, 12, 88, 89, 92, 94, 123–24, 134; as nation-specific, 94. See also party abandonment party ideological flexibility. See ideological flexibility of parties party ideological repositioning: broad appeal of Argentine parties, 167; and credibility, 168; failure, causes of as issue, 165, 167–68, 172, 198; failure of, in Peru, 165–67; failure of, in Venezuela, 167; necessary prerequisites for, 168–70. See also ideological flexibility of parties party leadership: change in, as method of ideological repositioning, 168; core group, size of, and ideological flexibility, 174–75, 193; decision-making processes in, 21–22; experience level, and ideological flexibility, 175–76, 193; and ideological flexibility, multiple factors affecting, 172; national, control over local parties, and ideological flexibility, 178–81, 193; pragmatism of, and ideological flexibility, 176–77, 191. See also entries under elites party leadership, broad ideological range in: diverse body of activists and workers as base for, 25, 29, 169, 196–97,

index 287

199; and ideological flexibility, 25, 29, 168–69 party membership, size of, and ideological flexibility, 173–74, 191 party organization: labor-intensive vs. capital-intensive, 187; simple, as key to flexibility, 25–26 party organization, and ideological flexibility, 171–90; activists’ ties to civilsociety organizations and, 182–83, 193; and communication flow within party, 181, 191; ideological diversity of activists and workers and, 25, 169, 185–86, 186t, 193–94, 196–97, 242; as issue, 172; labor union influence and, 184–85, 193; local leaders’ influence on national decisions and, 180, 193; local party unit comparisons, counterfactual evidence provided by, 193–98, 196t, 197t ; mass-party model of organization and, 172; members’ contribution to party financing and, 177–78, 191; membership and outreach organizations’ complexity and, 187–89, 193– 94, 197–99, 202; national leaders’ control over local parties and, 178–81, 193; party-branded patronage and, 189–90, 193–94, 197–99; party leaders’ experience level and, 175–76, 193; party leaders’ pragmatism and, 176– 77, 191; prerequisites for flexibility, 25– 26, 168–71; recruitment via particularist benefits and, 183–84, 193; scholarship on, 172; selection of relevant dimensions, 191–93, 192t ; significant dimensions, 198–99; size of leadership core and, 174–75, 193; size of party membership, 173–74, 191 party organization, and party-system collapse, 5, 9–10, 21–26, 29; existing research on, 9–10, 22; failure to act, necessity of explaining, 21; methods of analysis in, 22–25, 29; poor response as function of multiple party actors, 22–25 party organizations in Peru, poor response to ideological underrepresentation, 23–25, 165–67, 191–98, 192t

party replacement, as form of electoral change, 51 party supporters, stereotypes about, as source of party identification, 16, 90, 211–12 party system, traditional: definition of, 46; Peru’s party system as, 46, 47–48; Venezuela’s party system as, 46, 47 party system change, implications of findings for, 243 party-system collapse: definition of, 48; differing conceptions of, 32–33; recent examples of, 49; and rise of outsider candidates, 3; as type of electoral change, 53; as uncommon event, 3–4, 11, 40. See also Peru’s party-system collapse; Venezuela’s party-system collapse party-system collapse, causes of: alternative explanations for, 256n1; dependence on theoretical and historical frames, 32; importance of understanding, 4, 49. See also model of party-system collapse; theories of party-system collapse, in current literature; entries under party abandonment party system collapse, experience of, and loss of faith in government action, 28, 203–4, 209, 210, 212–13, 226–31, 226t ; tension with citizens’ experience of successful mass action, 229–30 party system stability: electoral volatility within, 50; as expected norm, 4; forces contributing to, 4; risk aversion and, 11 party system structures, differences in, as confounding factor in cross-nation comparisons, 217 patronage, party-branded: as cause of voter support, 107, 108–11, 110t ; and diffusing of voter anxiety, 199; and diffusion of voter anger, 242; and ideological flexibility, 25–26, 170–71, 189– 90, 193–94, 197–99 Pérez, Carlos Andrés: abandonment of reforms, 6–7, 69; anti-statist rhetoric by, 206; coup attempts against, 60; economic policies, damage done by, 67,

288 index

68–69; economic reforms, protests against, 59, 124; impeachment of, 124, 206; voter anger at, 146 Pérez, Carmen, 47 Pérez-Linán, Anibal, 98 Perón, Juan Domingo: and founding of Peronist party, 35; political dealing by, 35 Peronist party (Argentina): activists’ ties to civil-society organizations, 183; adaptability of, 9; corruption scandals, limited impact of, 107–11, 110t ; diversity ideology of local candidates, 194; dramatic changes in during 1990s, 49; history of, 34–35; ideological appeal, broadness of, 167; ideological diversity of activists and workers and, 186, 186t, 199; ideological flexibility, 167, 191, 194; ideological flexibility, sources of, 191–93, 192t ; and ideological representation, 116, 117; labor union influence on, 185; leaders’ relative pragmatism, 177; levels of recruitment via particularist benefits, 184; membership and outreach organizations’ relative complexity, 189, 199; members’ relative contribution to party financing, 178; Menem and, 35–36; party-branded patronage of, as source of support, 107, 108–11, 110t, 199; party leaders’ relative experience level, 176; party organization, and ideological flexibility, 199; relative levels of party-branded patronage, 190; relative rigidity of decisionmaking procedures, 179–81; relative size of party leadership core, 175; relative size of party membership, 174; and renewal Peronism, 35. 36; as traditional party, 34–35; voter identification with, trends in over time, 94, 107–8 Peru: citizen’s perception of major national problems, 216; economic performance, 70, 71; ineffectiveness of government before party-system collapse, 213; media exposure levels in, 216; military coup of 1948, 39; military coup of 1968, 39, 41; nationalizations under García, 66; as partial democracy, 79; presiden-

tial power in, 56; relative political participation levels, 217; traditional parties in, 38–43; transition to democratic rule, 65; voter anger in, qualitative evidence of, 147–49; voter turnout after party-system collapse, 204 Peru’s government institutions, limited value in explaining party-system collapse, 55–58 Peru’s party system: differences from other study systems, 217; as mix of traditional party system and disorder, 152– 53; strengths and weaknesses in, 105; as traditional party system, 46, 47–48; weakness and disorder of, 47, 105, 153 Peru’s party-system collapse, 2; correlation with economic factors, 73, 78; correlation with inflation levels, 72–73; correlation with negative economic growth, 70–71; correlation with unemployment levels, 73; date of, definition of party-system collapse and, 49; decline in party identification preceding, 88; and democracy, impact on, 26; different perspectives on, 32; electoral institutions’ limited value in explaining, 55–58; elite anti-statist messages during, 206; existing theories on, 5–6; parties’ failure to represent ideology of voters as cause of, 8; and party identification, 212; as surprise, 48; traditional parties’ loss of presidential election vote share, 53–54, 54; as unusual, 3–4, 49 Peru’s party-system collapse, violence surrounding, 58–59, 60–61; as contributing factor in party-system collapse, 55, 61; groups responsible for, 60–61 Peru’s political parties: activists’ ties to civilsociety organizations, 183; failures in ideological positioning, 165–67; ideological diversity of activists and workers and, 186, 186t; ideological inflexibility, party organization and, 23–25, 165–67, 191–98, 192t; labor union influence on, 185; leaders’ relative pragmatism, 177; levels of recruitment via particularist benefits, 184; membership and out-

index 289

reach organizations’ relative complexity, 189; members’ relative contribution to party financing, 178; party leaders’ relative experience level, 176; poor strategies of, as cause of party-system collapse, 8; relative rigidity of decisionmaking procedures, 179–81; relative size of party leadership core, 175; relative size of party membership, 174 Peruvian citizens’ experience of politics, comparative analysis, 218–32; experience of party system failure, and loss of faith in government action, 229–31; and external political efficacy, 223t, 224; and external political efficacy, conditioning on confounding variables, 233t, 234–35; government management of economy, 220, 231; and government’s role in solving problems, 226–29, 226t ; ideological representation of state, 219–20; and nonparticipation in politics, legitimacy of, 223t, 224–25; and party identification, 222– 23, 223t, 231; president’s problem-solving efforts, 221; retrospective analysis of effectiveness of state, 219; views on effectiveness of citizen involvement, 223–24, 224t Peruvian constitutional convention, 42 Peruvian election(s): ideological underrepresentation as impetus for voter abandonment of party system, 17–18, 136–41, 142; limited availability of data on, 137; outsider candidates, voter vetting of, 20. See also Lima municipal elections of 1989 Peruvian election of 1985, failure of ideological repositioning in, 165–66 Peruvian election of 1990: and collapse of traditional political parties, 2; dominance of outsider candidates in, 58; failure of ideological repositioning in, 166; and ideological underrepresentation, inability to resolve, 23–24 Peruvian election of 2000, and Fujimori’s loss of power, 3 Pinochet, Augusto, anti-statist rhetoric by, 207

PJ (Partido Justicialista). See Peronist party political efficacy. See external political efficacy; internal political efficacy political elites, consequences of party-system collapse for, 26 political outsiders. See outsider elites political participation: Chávez rhetoric on, and citizens’ experience of politics after party-system collapse, 202, 207– 9, 214, 225, 226, 230, 235, 237–41; different levels of, as confounding factor in cross-nation comparisons, 216–17; and external ptolitical efficacy in Venezuelans, 232–35, 233t ; faith in, factors affecting, during party-system collapse, 213–14; nonparticipation, citizen views on legitimacy of, 223t, 224–25 political parties. See entries under parties pollution, citizens’ views on government role in mitigating, 226t, 227 populism, macroeconomics of, 65 poverty levels, and voter abandonment of party system, affect-and voting theory empirical test on, 158 PPC (Partido Popular Cristiano; Peru), 246–47n1 Prado, Manuel, 39 PRD. See Partido de la Revolución Democrático presidential election system, two-round, limited value in explaining party-system collapse, 58 presidential powers, limited value in explaining party-system collapse, 56 presidential systems, limited value in explaining party-system collapse, 55–56 PRI. See Partido Revolucionario Institucional Progressive Conservative party (Canada), 52 proximity vs. directional voting, 117–18 Proyecto Venezuela (PVZ), 132t, 133, 134 public services, citizens’ views on government role in provision of, 226t, 228–29 Radcliff, Benjamin, 75 Radicals (Argentina). See UCR (Unión Cívica Radical)

290 index

rational-choice model: in party decisionmaking processes, 21–22; and voter decision to abandon party system, 14, 21–22 realigning elections (critical elections), 50–51 realignments of party systems, as form of electoral change, 50–51 recruitment by political parties: ideological purism and, 176–77; via particularist benefits, and ideological flexibility, 183–84, 193 RedPol survey, 131 Reform party (Canada), 52 regimes, limited value in explaining collapse, 55–58 Remmer, Karen L., 74 renewal Peronism, 35, 36 replacement of multiple parties, as type of electoral change, 52–53 Republican party (U.S.), replacement of Whig party by, 51 right-wing elites, framing of party-system collapse by, 27–28 risk acceptance in voters, 246n9 risk aversion, and decision to abandon party system, 11–18, 12, 144; affective factors in, 14–18, 29, 145, 156–61, 160t, 242; alternatives to party abandonment, 13–14, 144; anger and, 12, 15–18, 29, 145, 156–61, 157, 160t, 242; anxiety/fear and, 12, 14, 15–17, 63, 72, 80, 84–85, 84t, 164, 242, 243; fervent single-issue voters, small number of, 13; naturally risk acceptant voters, small number of, 11; need for model of, 144; Peruvian party system and, 153; and rational-choice model, 14. See also affect-and voting theory empirical test Roberts, Kenneth M., 5, 6 Rosas, Guillermo, 118 Rosenbaum, P., 251n10 Saez, Irene, 130 Salas Romer, Henrique: and election of 1998, 130, 131, 131t ; local government experience of, 57

Sandinista party (Ecuador), 72 Schattschneider, E. E., 241 Schickler, Eric, 90 Sendero Luminoso: defeat of, and presidential approval ratings, 221; history of, 60–61; and Lima municipal elections of 1989, 137; and Peru’s ideological spectrum, 121; state’s clumsy response to, 61; strategy of, 61; and voter anger, 147–48 social class: distributions of, as confounding factor in cross-nation comparisons, 215; limited correlation with party abandonment decision, 163; and vote choice, in Peru, 139–40; and vote choice, in Venezuela, 129–30, 134–35 social group identity: and party identification, 16, 90, 211–12; stability of, and party stability, 4 Socialist party (Italy), 53 societal crisis, in process of party abandonment, 12, 15–16, 28 stable party systems: electoral volatility within, 50; as expected norm, 4; forces contributing to, 4; risk aversion and, 11 Stimson, James A., 51 Strom, Kaare, 187 Tanaka, Martin, 8–9, 32, 66, 245–46n8, 256n1 theories of party-system collapse, in current literature, 5–10; limitations of, 4, 10 third party entry, ideological underrepresentation and, 113 Toledo, Alejandro: and citizen experience of politics, 230; low popularity of, 221; as nontraditional politician, 153 Torres, Aristides, 101 Transparency International’s Corruption Perceptions Index (CPI), 98, 99t types of voter dissatisfaction, and likelihood of voting for outsider candidate, 158–59 UCR (Unión Cívica Radical; Argentina): activists’ ties to civil-society organizations, 183; coalition with FREPASO,

index 291

36, 38; corruption scandals in, 38; history of, 36–38; ideological appeal, broadness of, 167; ideological diversity of activists and workers, 186, 186t ; ideological flexibility, 167, 194; and ideological representation, 116, 117; labor union influence on, 185; leaders’ relative pragmatism, 177; levels of recruitment via particularist benefits, 184; membership and outreach organizations’ relative complexity, 189; members’ relative contribution to party financing, 178; and partybranded patronage, 110t ; party leaders’ relative experience level, 176; and period of party-system collapse, 37; relative rigidity of decision-making procedures, 179–81; relative size of party leadership core, 175; relative size of party membership, 174; as traditional party, 36–38; voter identification with, trends in over time, 94, 107, 108 unemployment in Latin America nations, 71; correlation with party-system change, 73, 77t, 78; and voter party abandonment, affect-and voting theory empirical test on, 158 Unión Cívica Radical. See UCR Unión Republicana Democrática. See URD unions. See labor unions United Kingdom’s party system, party replacements in, 51 United States’ party system: party replacement in, 51; realignments in, 51 URD (Unión Republicana Democrática; Venezuela), 45; as minor party, 45–46 Uruguay: corruption level in, 99t ; economic performance, 71; economic performance, and party stability, 72, 73; net electoral volatility scores, 3 Uruguay’s party system: changes in, 3–4; realignments in, 51 Vargas Llosa, Mario: Acción Popular support for, 41; anti-statist message of, 206; and election of 1990, 24, 166; lack

of local government experience, 57; and Movimiento Libertad, 245–46n8 Velasco, Alvaro, 64–65 Velázquez, Andrés: and election of 1993, 124, 125t ; local government experience of, 57 Venezuela: banking crisis (1996), 69, 130; Caracazo in, 59–60, 146; citizen’s perception of major national problems, 216; economic performance, 70, 71; ineffectiveness of government before party-system collapse, 213; leftists, proportion in population, 115, 135–36; media exposure levels in, 216; military coup of 1945, 43; military coup of 1948, 44; nationalizations under Pérez, 67; presidential power in, 56; privatizations under Caldera, 69; relative lack of political diversity in, 169; relative political participation levels, 217; traditional parties in, 43–46; voter anger, qualitative evidence of, 146–47; voter turnout after party-system collapse, 204 Venezuelan citizens’ experience of politics, comparative analysis, 218–32; experience of party system failure, and loss of faith in government action, 229–31; and external political efficacy, 223t, 224, 226; and external political efficacy, conditioning on confounding variables, 232–35, 233t ; and external political efficacy, in Chavistas vs. nonChavistas, 236–41, 239t. 240t ; and external political efficacy, over time, 232, 235–36, 235t ; government management of economy, 220, 231; and government’s role in solving problems, 226–29, 226t ; ideological representation of state, 219–20; and internal political efficacy, 223t, 225–26, 234; and internal political efficacy, over time, 235–36, 235t ; and nonparticipation in politics, legitimacy of, 223t, 224–25; and party identification, 222–23, 223t, 231; president’s problem-solving efforts, 221, 231–32; retrospective analysis of effectiveness of state, 219; views

292 index

on effectiveness of citizen involvement, 223–24, 224t, 234, 238, 240 Venezuelan constitution of 1947, 44 Venezuelan constitution of 1999, and citizen participation, 207, 209 Venezuelan election of 1993: ideological underrepresentation as impetus for voter abandonment of party system in, 124–30; traditional parties’ loss of support, 1, 44, 124–25, 125t ; traditional parties’ loss of support, causes of, 125–30, 125t Venezuelan election of 1998: and election of Chávez to presidency, 2; ideological underrepresentation as impetus for voter abandonment of party system in, 130–36, 132t ; traditional parties’ loss of support, 1–2, 131; traditional parties’ loss of support, causes of, 132– 35, 132t Venezuelan outsider candidates, voter vetting of, 20 Venezuela’s government institutions, limited value in explaining party-system collapse, 55–58 Venezuela’s party system: differences from other study systems, 217; stability before 1990s, 1; stable periods, electoral volatility within, 50; as traditional party system, 46, 47 Venezuela’s party-system collapse, 1–2; correlation with economic factors, 73, 78; correlation with inflation levels, 73; correlation with negative economic growth, 70–71; correlation with unemployment levels, 73; date of, definition of party-system collapse and, 49; decline in party identification preceding, 88, 94; and democracy, impact on, 26; electoral institutions’ limited value in explaining, 55–58; elites’ anti-statist rhetoric during, 206–7; existing theories on, 5, 6–7, 8; leftists’ role in, 135–36; and opening for new political discourse, 240; parties’ failure to represent ideology of voters as cause of, 7–8; and party identification, 212; traditional

parties’ loss of presidential election vote share, 53–54, 54; as unusual, 3–4, 40; violence surrounding, as contributing factor, 55, 58–60 Venezuela’s political parties: activists’ ties to civil-society organizations, 182; failure of ideological repositioning, 167; ideological diversity of activists and workers and, 186, 186t ; ideological inflexibility, party organization dimensions and, 167, 191–98, 192t ; labor union influence on, 185; leaders’ relative pragmatism, 177; levels of recruitment via particularist benefits, 184; membership and outreach organizations’ relative complexity, 188–89; members’ relative contribution to party financing, 178; party leaders’ relative experience level, 176; relative rigidity of decision-making procedures, 179–81; relative size of party leadership core, 175; relative size of party membership, 174 violence, citizens’ views on government role in preventing, 226t, 228 violence surrounding Peru’s party-system collapse, 58–59, 60–61; as contributing factor in party-system collapse, 55, 61; groups responsible for, 60–61 violence surrounding Venezuela’s partysystem collapse, as contributing factor, 55, 58–60 vote choice: economic evaluations as cause of, in Peru, 139, 141–42; economic evaluations as cause of, in Venezuela, 125t, 127–28, 132–33, 132t, 141–42; ideological underrepresentation and, 113; party identification as cause of, 89; social class and, in Peru, 139–40; social class and, in Venezuela, 129–30, 134–35 voter anxiety. See anxiety/fear voter decision-making processes, as final step of party-system collapse, 5, 10–11, 28 voter turnout, after party-system collapse, 204 voting, proximity vs. directional, 117–18

index 293

Weyland, 250n11 Whig party (U.S.), replacement of, 51 Wildavsky, Aaron, 183 Wilson, James Q., 183

World Values Survey (WVS), 81, 98, 120, 142–43 Zechmeister, Elizabeth, 246n9