FORMAL INSTITUTIONS IN INFORMAL POLITICS: THE EFFECT OF CLIENTELIST POLITICS ON PARTY SYSTEM INSTITUTIONALIZATION By Helen Hyun-Young Lee A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Political Science - Doctor of Philosophy 2014 ABSTRACT FORMAL INSTITUTIONS IN INFORMAL POLITICS: THE EFFECT OF CLIENTELIST POLITICS ON PARTY SYSTEM INSTITUTIONALIZATION By Helen Hyun-Young Lee This dissertation consists of three essays that seek to improve upon theoretical and empirical accounts of the three dimensions of party system institutionalization: (i) emergence of new parties; (ii) party nationalization; and (iii) ideological congruence between parties and voters. In explaining these aspects of the party system, my approach highlights the role of the informal mechanisms of clientelism. My dissertation argues that clientelism exerts centrifugal forces on the relationships among key actors, which in turn has negative effects on institutionalization. The first essay investigates why new parties emerge in an environment where clientelism is widespread, if clientelist practice empowers the ruling parties as conventional wisdom suggests. To answer this question, I highlight the structural aspect of clientelism. Drawing upon bargaining theory, I contend that when extensive clientelist practice is coupled with a decentralized resource structure, abundant resources actually reduce political actors’ incentives to pledge allegiance to the incumbent party, as in this situation their resource needs can be easily met by alternative resource channels. I demonstrate that the combination of clientelist resources and decentralized clientelist structure creates an incubator for the birth of new parties. In the second essay, I ask why some parties are competitive throughout the country, while others appeal to only a few specific regions. In this study, I examine if and how much informal mechanisms of clientelism mediate the impact of formal party structure on party nationalization, i.e. the patterns of territorial vote distribution of parties. To this end, I explore the relationship among three variables: party structure, effectiveness of clientelism, and party nationalization. I hypothesize that extensive party structure enhances the effectiveness of parties’ clientelist efforts, which in turn decreases party nationalization. Where clientelist efforts are effectively translated into electoral gains, elites and voters are more likely to establish relationships beyond party organizations. Therefore, I argue that effective clientelism undermines a party’s organization as a collective entity. As existing studies suggest, parties with extensive organization tend to have higher levels of party nationalization. However, I find that this positive impact is suppressed to the extent that these organizational attributes improve the effectiveness of clientelism. Finally, the third essay examines the impact of clientelism on the level of congruence. Conceptualizing clientelism as a tool of persuasion employed by political parties, I maintain that elites distribute clientelist rewards in order to alter the preferences and behaviors of voters who would otherwise hold different or no views about parties’ policies. In response to the rewards, I argue, some voters will develop an affinity to a patron party, and subsequently vote for that party despite it being located further than other parties from their ideal points. Consequently, clientelism systematically widens the gap between voters’ preferences and their vote choices, thereby undermining congruence. I suggest that voters’ utility is a function of both issue position and material inducement. I find that clientelism serves as a crucial predictor of congruence. This dissertation contributes to the understanding of party system institutionalization by emphasizing the under-explored role of informal institutions. By demonstrating how informal rules constrain and modify the way formal institutions function, this dissertation also offers important insights into the literature on the relationship between formal and informal institutions. Thus, this dissertation provides an opportunity to advance our understanding of to what extent formal institutions explain or predict political outcomes. To my parents for their constant love and support iv ACKNOWLEDGEMENTS In every aspect, my experience in graduate school has been the best years of my life so far. I have grown intellectually and personally and I found something that I am truly passionate about. For my intellectual and personal growth, I owe thanks to many people. First and foremost, I thank my dissertation chair, Eric Chang, from the bottom of my heart. He is the best adviser and mentor anyone could ever wish for. I first met him in one of my first classes at graduate school. Since then, he has been a profound influence on my intellectual growth. The conversations with him have always been intellectually inspiring and amusing. He genuinely cared about my intellectual growth and personal well-being. Throughout the years, he has never lost faith in me even when I didn’t seem to be quite on the right track, and he patiently motivated me to live up to my true potential. Without his tremendous support and guidance, I cannot imagine anything I have achieved so far in graduate school. My sincere appreciation also goes to my dissertation committee members, Michael Bratton, Jeffrey Conroy-Krutz, and Ani Sarkissian. Throughout my graduate training and dissertation writing, they have been a source of advice, encouragement and inspiration. Michael Bratton helped me to find my strength in quantitative methods. His comments and suggestions on my works have been one of the most important source of my inspirations for my research. Jeffrey Conroy-Krutz contributed enormous time and energy on offering me feedback on the drafts of my dissertation. Thanks to his invaluable comments and suggestions, I was able to hone my scholarly writing skills. Ani Sarkissian also offers me invaluable suggestions and comments on my earlier drafts of dissertation. v I owe particular thanks to Karen Battin, Melinda Hall, Steve Kautz, and Charles Ostrom for their help in every administrative need. I am also grateful for the financial support from the Department of Political Science and Graduate School at MSU, and the Afrobarometer Project. I am also thankful for Carolyn Logan for her guidance. While I was working with her at the Afrobarometer, I develop various skills in data management. Furthermore, my appreciation also goes to my mentor at Dongguk University, Youngjoong Lee, who encouraged and supported me throughout the entire graduate school years. I also would like to thank Valentina Bali, Cristina Bodea, Matt Grossmann, William Jacoby, Joongi Kim, Saundra Schneider, Corwin Smidt, for their intellectual advice and guidance. I would also like to express special thanks to my cohorts and friends, Danielle Carter, FangYu Chen, Seo-Yeon Choi, Nicholas Kerr, Sung-Min Han, Masaaki Higashijima, Shih-Hao Huang, Hsin-Hsin Pan, Chunho Park, Johann Park, Wen-Chin Wu, and Fangjin Ye. Through the dissertation writing support group, the Economist discussion group, and informal gatherings, they contributed their time and energy to provide me with great feedback on the drafts of my dissertation and with lots of encouragement. They made my years in graduate school brighter and more colorful. Last but not least, I am deeply thankful for my family who have always believed in me and provided me with love and support. Without them, I would never have completed this dissertation. My mother, Myung-Ok Lee, always reminded me of the virtue of positive thinking and grit. Throughout my graduate school years, I only had a chance to see her once. But, her positive and pleasant vibe have been strong enough to make me happy and laugh over the phone conversation. She has been the major source of my energy and peace. My father, Seung-Ki Lee, has been my role model since I was a little kid. As a professor in biochemistry, he has always been open to interesting questions and new perspectives. His passion for research motivates me to follow in his vi footsteps. My sisters, Jin-Young and Sun-Young, are my best friends and the biggest supporters. While facing similar challenges, we shared our concerns and moments. Finally, I thank my husband, Youngmin, and my parents-in-law for their patience, understanding, and support. I am truly lucky and blessed to have these wonderful people in my life. vii TABLE OF CONTENTS LIST OF TABLES .......................................................................................................................... x LIST OF FIGURES ....................................................................................................................... xi CHAPTER 1 INTRODUCTION .................................................................................................... 1 1.1 The Motivations ............................................................................................................... 1 1.2 Theoretical and Empirical Solutions ................................................................................ 5 1.2.1 Informal Institutions.................................................................................................. 5 1.2.2 Party System Institutionalization .............................................................................. 8 1.2.3 Clientelism .............................................................................................................. 10 1.2.4 Conceptualization of Clientelism ............................................................................ 13 1.3 Contribution ................................................................................................................... 14 1.4 Chapter Overview .......................................................................................................... 15 CHAPTER 2 CLIENTELIST POLITICS AND NEW PARTY FORMATION .......................... 20 2.1 Introduction .................................................................................................................... 20 2.2 Theoretical Framework .................................................................................................. 22 2.3 Empirical Analysis ......................................................................................................... 28 2.3.1 A Cross-National Investigation............................................................................... 28 2.3.2 A Cross-Party Investigation .................................................................................... 34 2.3.3 Individual-level Investigation ................................................................................. 36 2.3.4 Robustness Check ................................................................................................... 39 2.4 Conclusion...................................................................................................................... 42 APPENDIX ................................................................................................................................ 44 CHAPTER 3 PARTY EXTENSIVENESS, CLIENTELISM, AND PARTY NATIONALIZATION: THE MEDIATING EFFECT OF CLIENTELISM .............................. 48 3.1 Introduction .................................................................................................................... 48 3.2 Literature Review ........................................................................................................... 52 3.3 Theory: the mediating role of clientelism ...................................................................... 56 3.3.1 Party Structure and Effectiveness of Clientelism ................................................... 56 3.3.2 Clientelism and Party Cohesion .............................................................................. 58 3.4 Data and Measures ......................................................................................................... 61 3.4.1 Dependent Variable: Party Nationalization ............................................................ 61 3.4.2 Independent Variables: Extensiveness of a Party Organization ............................. 63 3.4.3 Mediating Variable: Effectiveness of Clientelism .................................................. 64 3.4.4 Control variables ..................................................................................................... 65 3.5 Empirical Analysis ......................................................................................................... 66 3.5.1 Analytic Strategy .................................................................................................... 66 3.5.2 Results ..................................................................................................................... 68 3.6 Conclusion...................................................................................................................... 75 APPENDIX ................................................................................................................................ 77 viii CHAPTER 4 CLIENTELISM AS AN INFORMAL DETERMINANT OF CONGRUENCE ... 81 4.1 Introduction .................................................................................................................... 81 4.2 Defining and Measuring Congruence ............................................................................ 83 4.3 Literature on Electoral Systems and Congruence .......................................................... 86 4.4 Theory ............................................................................................................................ 88 4.5 Data and Measures ......................................................................................................... 94 4.5.1 Dependent Variable: Congruence ........................................................................... 94 4.5.2 Independent Variable: Level of clientelism ............................................................ 98 4.5.3 Control Variables .................................................................................................... 98 4.6 Empirical Analysis ......................................................................................................... 99 4.6.1 Descriptive Analysis ............................................................................................... 99 4.6.2 Analysis................................................................................................................. 101 4.6.3 Robustness Check ................................................................................................. 103 4.7 Discussion and Conclusion .......................................................................................... 105 4.7.1 Implications for Core-Swing Voter Argument ..................................................... 106 4.7.2 Endogenous Preference Formation and Implication for Congruence ................... 108 APPENDIX .............................................................................................................................. 111 CHAPTER 5 CONCLUSION..................................................................................................... 116 BIBLIOGRAPHY ....................................................................................................................... 120 ix LIST OF TABLES Table 1.1 Organization of the Dissertation ................................................................................... 19 Table 2.1 Cross-National Determinants of the New Party Emergence ........................................ 33 Table 2.2 Summary of the Measurements of Variables ................................................................ 42 Table 2.3. Effect of Clientelist Level and Structure on Effectiveness of Clientelism .................. 45 Table 2.4. Effect of Clientelist Level and Structure on Voters' Supports for Ruling Parties ....... 45 Table 2.5. Robustness Checks for the Cross-National Determinants of New Party Emergence .. 46 Table 2.6 Cases Included .............................................................................................................. 47 Table 3.1. Mediation Analysis ...................................................................................................... 69 Table 3.2. Sobel Test for Mediating Effects ................................................................................. 72 Table 3.3. Bootstrap Results for Mediating Effects ...................................................................... 73 Table 3.4. Descriptive Statistics.................................................................................................... 78 Table 3.5. Case Selection by Region ............................................................................................ 78 Table 3.6. Case Selection by Country........................................................................................... 78 Table 3.7. DALP Questionnaire.................................................................................................... 79 Table 4.1. Cross-tabulation of ‘Having a party representing respondents’ views’ and ‘Having a part respondents feel close to’ ...................................................................................... 90 Table 4.2. Bivariate Regression: Relationship between the Level of Clientelism and Having a Non-Ideological Affinity to a Party ............................................................................. 91 Table 4.3. Survey Questions and Dataset ..................................................................................... 97 Table 4.4. Multivariate Analysis of Congruence ........................................................................ 101 Table 4.5. First Difference: Changes in the Expected Value of Congruence by the Changes in the Level of Clientelism under PR and SMD .................................................................. 102 Table 4.6. First Difference: Changes in the Expected Value of Congruence with the Changes from SMD to PR ........................................................................................................ 103 Table 4.7. Countries Categorized by the Level of Clientelism in 2008-2009 ............................ 115 x LIST OF FIGURES Figure 1.1. Level of Clientelism across the World ....................................................................... 12 Figure 2.1. Predicted Number of New Parties across Different Levels of Clientelism under Two Alternative Clientelist Structures ................................................................................. 34 Figure 2.2. Estimated Effects of Clientelist Level on Effectiveness of Clientelism under Alternative Clientelist Structure ................................................................................... 36 Figure 2.3. First Difference Measuring the Effect of Clientelist Level across Varying Clientelist Structure ....................................................................................................................... 38 Figure 2.4. Predicted number of new parties across different levels of clientelism under two alternative clientelist structures .................................................................................... 47 Figure 3.1 Calculation of Party Nationalization Score ................................................................. 62 Figure 3.2. Distribution of Party Nationalization Score ............................................................... 68 Figure 3.3. Pathway of a Mediation Process for a Relationship between Party Structure and Party Nationalization ............................................................................................................. 70 Figure 4.1. Conventional Measure of Congruence the absolute ideological distance between the median citizen and the governing parties ..................................................................... 83 Figure 4.2. Many-to-Many Measure of Congruence the overlap of the distribution of citizens and governing parties .......................................................................................................... 84 Figure 4.3. Three Hypothetical Scenario of Many-to-Many Congruence .................................... 85 Figure 4.4. Proportion of Respondents Having a Party Representing their Views....................... 90 Figure 4.5. Degree of an Affinity to a Party ................................................................................. 92 Figure 4.6. Congruence Patterns by the Level of Clientelism ...................................................... 99 Figure 4.7. Clientelism (measure: DALP expert survey) and Level of Distortion ..................... 112 Figure 4.8. Clientelism (measure: personal vote system) and Level of Distortion..................... 112 Figure 4.9. Clientelism (measure: corruption perception index) and Level of Distortion .......... 113 Figure 4.10. Clientelism (measures from World Economic Forum) and Level of Distortion .... 114 .. xi CHAPTER 1 INTRODUCTION This dissertation examines the relationship between formal and informal institutions. Specifically, I focus on how the nature of clientelistic politics affects the levels of party system institutionalization in light of three main questions: (i) How do clientelist practices by established parties affect the electoral fortunes of new parties? Do they create an unlevel playing field in favor of the established parties, or do they open up an opportunity for new parties to enter into the electoral game? (ii) Why are some parties competitive throughout the country, while others appeal to only a few specific regions? In other words, how do we account for the varying degrees in nationalization across party systems? (iii) Do parties’ clientelist practices lead to differentiation or convergence in what citizens want and what their representatives try to achieve? If so, what are the underlying mechanisms of differentiation or convergence? Answering these questions is crucial to improving our understanding of party system institutionalization and democratic representation processes. It also contributes to our understanding of how informal politics interacts with formal institutions. 1.1 The Motivations This dissertation is motivated by the lack of research on informal political institutions in contemporary institutional analyses. In particular, the new institutionalism brought the institutions back into the studies of politics. Rational choice theorists argue that institutions affect individuals' choices and behaviors by setting parameters to further choice (e.g. North 1990). According to the historical institutional approach, institutions shape the individual preferences and behaviors, while political outcomes are the product of the interactions among diverse interests and institutional structures (e.g. Steinmo et al. 1992). Since the emergence of the new institutionalism in the social 1 sciences, scholars have emphasized formal institutions more than informal ones. Conventionally, it is assumed that formal rules mainly shape individuals’ incentives and expectations. Institutions are often defined in terms of formal rules and officially recognized political structures. These assumptions require further scrutiny. As Sheingate (2003) maintains, out of rigid rules, unexpected patterns of interaction often emerge. Even in a tightly specified rule, uncertainty occurs due to the gaps between the written rules and their implementations (Mahoney and Thelen 2010). Thus, formal institutions neither exactly reflect collective interests nor operate in practice as initially devised. Importantly, formal political institutions are initially formulated for Western democracies. For many new democracies, therefore, formal institutions may present even greater gaps between the written stipulations and the reality. How formal institutions are implemented in practice may differ considerably according to the rules of the game, norms, and routines already embedded in a society. Surprisingly little is known about how informal rules affect political life or how extensive their effects are, although they serve a crucial role in shaping the political landscape (for exceptions see Stokes 2006; Bratton 2007; Helmke and Levitsky 2004). Informal institutions are often treated as residual variables or noise whenever formal institutions can no longer account for the variation in outcomes. When existing literature does focus on informal institutions, it often depicts them in an overly simplistic way; e.g. as the byproduct of low levels of political or economic development. This imbalance is unfortunate, as the term “institution” is conceptualized in a much broader way than just formal institutions. As North suggests, institutions refer to “any form of constraint that human beings devise to shape human interaction” (North 1990: 4). Within his theoretical formulation, individuals are assumed to have incomplete information and limited cognitive 2 capacity. To reduce risk and transaction costs, North argues, humans design institutions. In his definition, institutions include both formal and informal rules. Together they shape the preferences and strategies of the actors involved. Since the seminal work of North, the literature on new institutionalism has become increasingly interested in a broader institutional framework that incorporates informal norms and rules (Williamson 2000). More recently, Helmke and Levitsky point out that “formal institutions do not operate in a vacuum” (Helmke and Levitsky 2006: 280). Formal institutions are effective because of informally shared norms and expectations. For example, Stokes (2006) argues that in Argentina elections serve as mechanism of vertical accountability to the extent that a shared norm motivates voters to follow retrospective decision rules. Moreover, despite the extensive research on formal institutions, the debate on to what extent formal institutions explain or predict political behaviors and outcomes remains anything but settled in the comparative study of political institutions. For example, studies on the forms of government offer contradictory predictions about their effects on the quality of democracy and democratic representation. Some scholars claim the superiority of parliamentarism (Lijphart 1999; Linz 1990; Stepan and Skach 1993). They find that presidentialism inhibits accountability and increases the potential for conflicts, while undermining fair representation of interests by encouraging winner-takes-all outcomes. In contrast, others argue that parliamentarism may result in regime instability due to the difficulty of forming a stable majority government, whereas presidentialism is characterized by identifiable accountability (Shugart and Mainwaring 1998; Hellwig and Samuels 2007). Neither body of research adequately explains the contradictory outcomes. Likewise, the literature on federalism has not reached consensus on the effect of federal systems on political and economic outcomes. Some argue that in federal systems, the public sector 3 will be more efficient and the quality of government is higher. According to this argument, this system encourages competition among governments and promotes greater responsiveness to diverse local preferences (Rodden and Wibbels 2002). Others contend that federalism complicates policy-making and coordination by empowering regional politicians (Treisman 2002). Debates on the effect of institutional design implicitly assume that formal institutions will be implemented and generate expected consequences. Thus, as a good predictor of political behaviors and outcomes, this body of literature focuses on formal institutions such as constitutions, electoral systems, party systems, etc. As Knight and Sened (1995) point out, however, effectiveness of institutions varies. In spite of overwhelming emphasis on formal institutions, empirical evidence suggests that in reality many formal institutions are widely challenged and violated. We observe that the similarly designed institutions of Western democracies produce different consequences in political and economic performance in new democracies. The workings of the presidency in Chile during the 1990s are another example. The Chilean presidency was designed to be strong, but in practice the presidents systematically underutilized their legislative and agenda-setting powers during this time (Helmke and Levitsky 2004). The weak presidency in Chile is attributable to an informal institution, Concertaciones, which encouraged power sharing (Helmke and Levitsky 2004). As the anecdote implies, informal institutions have implications for the way enforcement of formal institutions occurs, the norms of behavior, and the expectations of the actors involved. Thus, informal institutions affect the outcomes of formal institutions (Grzymala-Busse 2004). Many rules and procedures that have significant influence on political life are created and enforced beyond formal channels (Helmke and Levitsky 2004). Where formal institutions are perceived to be ineffective or unfavorable, actors create or turn to informal rules to achieve their goals (Grzymala-Busse 2006). 4 Variations in the effectiveness of formal institutions is particularly salient where formal institutions have not been fully established. In the less institutionalized settings, formal institutions are subject to manipulation by key actors (Schedler 2002). As Tsebelis puts it, “the game in the principal arena is nested inside a bigger game where the rules of the game themselves are variable” (Tsebelis 1990: 8). Thus, consequences of institutions are expected to vary according to their capacity to enforce the rules and induce compliance in practice. This dissertation does not intend to solve all of these problems at one time. Instead, the purpose of this dissertation is to take a modest step toward redressing our insufficient understanding of informal institutions. To this end, I focus on a specific example of informal institutions, clientelism, and examine how it modify or constrain the workings of formal democratic institutions. Going beyond the simplistic perspective on informal institutions as a byproduct of underdevelopment, my dissertation demonstrates that informal mechanisms of clientelism have a wide spectrum of relationships with formal institutions. 1.2 Theoretical and Empirical Solutions 1.2.1 Informal Institutions Helmke and Levitsky define informal institutions as “socially shared rules, usually unwritten, that are created, communicated and enforced outside officially sanctioned channels” (Helmke and Levitsky 2004: 727). As an important source of constraints devised by people to regulate their interactions, an increasing body of literature considers informal institutions such as sanctions, customs, traditions, or codes of conduct (North 1990). In many polities, informal rules are deeply entrenched in political life. They constrain and change formal institutions and political behavior (Brinks 2003). Therefore, although unwritten and 5 unofficial, informal institutions often have meaningful influence on individuals’ expectations and behavioral patterns. As Bratton points out, in reality it is social and power relations revolving around informal rules that determine “who gets what, when, and how” (Bratton 2007: 96). Importantly, formal and informal institutions are closely intertwined. Helmke and Levitsky (2004) identify the four types of relationships. First, formal and informal institutions can be complementary. In this case, informal institutions make the formal institutions more effective by facilitating the pursuit of individual goals within the formal institutional framework (Helmke and Levtisky 2004). For example, Grzymala-Busse (2004) maintains that informal financing and mobilization techniques used in electoral campaigns strengthen the role of elections as the means of elite competition. These informal techniques reduce the transaction costs of elections by providing information for strategic decisions (Grzymala-Busse 2004). Second, informal institutions may be substitutive to formal institutions. Substitutive informal institutions achieve what formal institutions fail to do. According to Taylor-Robinson (2006), for instance, informal institutions of clientelism fill the representation gap in Costa Rica, where the formal electoral rules do not provide an incentive for elected officials to represent local needs. Third, formal and informal institutions may have a competing relationship. Competing informal institutions provide incentives incompatible with the formal rules. Cases in point include corruption, tax loopholes, or subcontracting, which take advantage of the gaps in formal rules (Grzymala-Busse 2004). Finally, formal and informal institutions can have an accommodating relationship. Accommodating informal institutions reduce conflicts, while finding solutions within existing settings (Helmke and Levitsky 2006: 15). For example, Langston (2006) argues that in Mexico the informal rule of executive succession, called Dedazo, enables the outgoing executive to appoint the next president, thereby keeping intra-regime conflict to a minimum. 6 My dissertation moves beyond a simplistic perspective on informal institutions as a byproduct of underdevelopment. Thereby, I demonstrate a more nuanced relationship between formal and informal institutions than previously explored within the literature on party system institutionalization. For the purpose of my dissertation, I focus on one example of informal institutions, clientelism, which influences crucial aspects of institutional order, power relations, and allocation of resources in many polities. Specifically, Chapter 2 demonstrates that clientelism shapes inter-party relationship in new democracies where formal electoral rules and procedures are not yet fully established. According to the classification of Helmke and Levitsky (2004), this type of informal institutions have substitutive relationship with formal institutions. In this chapter, I argue that, in new democracies, the way clientelist exchange occurs significantly affects the rate of new party emergence, which is mainly considered as the product of formal electoral rules and social constructs in advanced democracies. In Chapter 3, I argue that informal institutions of clientelism have competing relationship with formal institutions. Clientelism creates incentives for elites to behave in ways that are incompatible with the formal institutions. To follow informal rules of clientelism, elites choose particularistic interests over programmatic and public-oriented goals. Through clientelist practices, individual elites gain at the expense of the party organization’s shared purpose. Chapter 4 illustrates accommodating nature of informal institutions. Parties and elites participate in electoral competition without directly violating formal rules of electoral competitions. However, by distributing clientelist rewards, patron parties and elites seek to alter the preferences and behaviors of voters in favor of them. Thereby, clientelist practices enable the patron parties to assume competitive advantage over other parties. In other words, the parties and elites abide by 7 the formal rules, while modifying their substantive effects through informal mechanisms of clientelist practices. 1.2.2 Party System Institutionalization A large body of literature has investigated party and party system institutionalization. 1 These studies argue that institutionalized parties and party systems are crucial to democratic functioning and consolidation because they effectively channel social demands (Hicken and Kuhonta 2011). Specifically, they demonstrate that institutionalized parties tend to have more accountability, greater stability of interests, and more broadly targeted policy programs. According to Huntington (1968), organizations gain value and stability through institutionalization. Mainwaring and Scully (1995) conceptualize four dimensions of party system institutionalization. In institutionalized systems, (i) the pattern of party competition is stable and regular, (ii) parties have strong roots in society, (iii) political actors see parties as a legitimate and necessary part of the democratic process, and (iv) party organizations are not the instrument of only a few ambitious elites. In this dissertation, I utilize three approaches to the study of party system institutionalization common in the literature on party competition and voting behaviors: 2 (i) 1 For example, see Huntington 1968; Kuenzi and Lambright 2001; Mainwaring and Scully 1995; Mainwaring and Torcal 2006 2 Other approaches include the spatial model of voting behavior (Downs 1957; Cox 1990; Hinich and Munger 1996; Budge 1994), social cleavage theories on party systems (Lipset and Rokkan 1967), and theories on party realignments (Inglehart 1990). 8 electoral volatility3 and new party formation, 4 (ii) party nationalization, 5 and (iii) congruence. 6 First, the studies of new party formation enhance our understanding of party system institutionalization by examining the stability and regularity of party competition. The rise and fall of new entrants to electoral competitions prevent voters from developing party loyalties, thereby hindering system institutionalization (Tavits 2008). Second, party nationalization captures the institutionalization of individual parties, which in turn play an important role in party system institutionalization (Lupu 2008). Cohesive party organizations with nation-wide appeal enhance the stability and regularity of interparty competition. Finally, the literature on congruence also provides important insights into party system institutionalization by examining how parties develop roots in society and how voters develop attachments to parties (Mainwaring and Torcal 2005). The existing explanations for new party formation, nationalization, and congruence are unsatisfactory in that they mainly rely on formal institutions as key explanatory factors. Informal institutions are prevalent in how political parties actually function, such as in recruiting activists, selecting candidates, managing party finance, maintaining societal linkages, and delivering votes. For instance, a politician’s career path may be determined by which patron-client networks she is involved in. Also, organizational boundaries of a party may be ambiguous due to informal networks which constitute a crucial part of party organizations. Freidenberg and Levitsky (2006) argue that if decision-making, resource allocations, or career paths are determined by informal 3 Przeworski 1975; Pedersen 1983; Coppedge 1998; Roberts and Wibbel 1999; Mainwaring and Zoco 2007 4 Cox 1997; Hug 2001; Tavits 2008 5 Cox and McCubbins 1993; Stokes 1967; Jones and Mainwaring 2003; Chhibber and Kollman 2004; Cox 1997; Kasuya and Moenius 2008; Bochsler 2010; Schattschneider 1960 6 Powell 1994; McDonald, Mendes, and Budge 2004; Golder and Stramski 2010 9 institutions, analyses that focus only on formal structures will produce a misleading understanding of how a party works. The extent to which formal institutions affect party systems might be spurious because of alternative incentives generated by informal institutions. Thus, party competition and voting behavior may not be merely motivated by ideological or programmatic rationales as existing literature assumes. Likewise, if informal institutions are the main sources that citizens rely on to get things done and to process political information, analysis of their behaviors and attitudes should consider the effects of informal institutions. Bratton (2007) echoes this proposition and argues that where the performance of formal institutions does not meet popular expectations, people seek to make up for perceived institutional deficiencies by counting on the informal ties of clientelism. For example, he contends that informal rules of clientelism continue to affect Africans' political behaviors and attitudes (Bratton 2007). 1.2.3 Clientelism Clientelism is an informal mechanism through which patrons and clients engage in strategic alliances for mutual benefit. These relationships are established based on exchanges. Using their power and resources, patrons offer selective benefits to clients. In return, clients provide support and assistance to patrons (Scott 1972; Eisenstadt and Roniger 1984; Hopkin 2006; Kitschelt and Wilkinson 2007; Stokes 2007). Recent studies argue that clientelism is a ubiquitous and enduring feature of contemporary political parties.7 In many polities, clientelism is deeply embedded in key political functions by creating behavioral patterns “to which all political actors are acutely attuned” (Bratton 2007: 98). As van de Walle (2007) maintains, clientelism can be found in all polities, although its forms, 7 Katz and Mair 1995; Kitschelt 2000; Piattoni 2001; van Biezen 2000 10 extent, or political consequences vary. As Figure 1.1 illustrates, clientelism prevails throughout the world, including economically and politically advanced polities such as Greece, Spain, Belgium, Italy, and Japan. Scholars maintain that contemporary clientelism needs to be distinguished from traditional clientelism in that the former is characterized by less personalized, more equal, and more instrumental relationships than the latter (Hopkin 2001; Kitschelt and Wilkinson 2007; Piattoni 2001). Considerable variation exists across countries and parties in terms of the extent and the patterns of clientelist practices. However, little attention has been paid to different patterns of clientelist practices and their political consequences. My dissertation examines the implications of the extent and patterns of clientelist exchanges for elite fragmentations and ultimately for party system arrangements. 11 Figure 1.1. Level of Clientelism across the World Note: I have created the map based on DALP dataset (2008) using GunnMap. The level of clientelism is presented in color in which yellow presents the lowest level and red indicates the highest level of clientelism. The gray color indicates the missing data. 12 1.2.4 Conceptualization of Clientelism I argue that clientelism plays a crucial role in shaping the nature of relationships among political parties, politicians, and voters. To explain variations in political parties and party systems, each of the three chapters in this dissertation explores how clientelism influences different aspects of the following relationships: inter-party relationships, intra-party relationships, and party-voter relationships. Specifically, in order to explain the dynamics of inter-party relationships, Chapter 2 conceptualizes the patron-client exchange as a form of a bargaining. I argue that decentralized clientelist structures shift the bargaining power away from existing parties, while motivating ambitious challengers to establish new political parties. In Chapter 3, I focus on the mediating effects of clientelism on intra-party coordination. Based on party-level analysis, I argue that clientelism provides incentives for individual politicians to gain at the expense of the party organization they are involved in. As an obstacle to intra-party coordination, clientelism facilitates unequal vote distributions across districts, thus lower party nationalization. Finally, with regard to the outcome of party-voter relationships, Chapter 4 describes clientelism as a tool of persuasion employed by political elites and parties. I maintain that political elites use clientelistic rewards to alter the preferences and behavior of voters who would otherwise hold different or no views about parties’ policies. Therefore, clientelism persuades voters to vote further away from their own ideal points on the policy spectrum, which in turn undermines congruence at the aggregate level. Taken together, my dissertation examines both vertical and horizontal accountability. Chapter 2 and Chapter 3 discuss horizontal accountability, which considers the degree to which elites are responsible to other agencies and institutions. Chapter 4 enhances our understanding of the determinants of vertical accountability, which refers to the degree to which citizens hold officials 13 accountable (Stokes 2001). Taken together, I argue that clientelism increases electoral volatility and undermines levels of party and party system institutionalization. 1.3 Contribution The three essays in this dissertation examine the destabilizing and fragmentary effect of clientelism on party system institutionalization in terms of intra-party, inter-party, and party-voter relationships. Specifically, I examine how clientelism reinforces centrifugal tendencies within party organizations and party systems. Centrifugal forces disperse and divide elites and voters in pursuit of divergent objectives. The literature on clientelist politics often relies on the assumption that parties have an incentive to restrain competition for the sake of effective coordination in resource allocation, and presumes that political elites will remain committed to the clientelist arrangement. In this context, clientelism serves as a means to control intra-elite relationships, allowing the leaders to punish potential challengers and reward loyal supporters. Accordingly, existing studies tend to focus on the role of clientelism in attracting supporters and in enforcing patrons’ positions. For example, Huntington argues that clientelist rewards distributed by parties promotes the coordination of office seekers and, thereby, the formation of stable majorities. In turn, it contributes to effective and enduring political parties (Huntington 1968). Likewise, Arriola (2009) argues that leaders facilitate intra-elite accommodation and discourage oppositional challenges by recruiting more elites into their clientelist coalitions. By doing so, they achieve stability for their political regimes. Finally, Katz and Mair (1995) propose that major parties will develop cartel-like behavior over time. With a common interest in preserving their privileged positions, major parties restrain competition based on collusion and increase intra-party cooperation. Consequently, Katz 14 and Mair (1995) argue that parties will eventually come to adopt each other’s positions and strategies and begin to resemble one another. This conventional argument raises the questions of how patrons manage to keep clients from defecting and how patrons maintain a reliable cooperative arrangement in the absence of enforcement mechanisms. The ideal case of trouble-free agreement without a tedious bargaining process is exceptional rather than commonplace (Laver 1999). The literature seldom considers the destabilizing and fragmentary effects of clientelism because it focuses exclusively on the relationship between those who benefit from and are involved in clientelist relationships. By going beyond the narrow relationship between patron and client, this dissertation demonstrates that clientelism may have centrifugal effects on party organizations and party systems as a whole. This dissertation contributes to the understanding of party system institutionalization by emphasizing the under-explored role of informal institutions. By demonstrating how informal rules constrain and modify the way formal institutions function, this dissertation also offers important insights into the literature on the relationship between formal and informal institutions. Thus, this dissertation provides an opportunity to advance our understanding of an important question: to what extent do formal institutions explain or predict political outcomes? 1.4 Chapter Overview This dissertation is organized as follows. In Chapter 2, the first essay, Clientelist Politics and New Party Formation in New Democracies, addresses the question of why new parties emerge in an environment where clientelism is widespread if clientelism empowers the ruling parties as conventional wisdom suggests. To answer this question, I present and test a theory explaining the varying extent to which clientelism helps or hinders ruling parties to dominate party systems and 15 its consequences for the formation of new parties. To explain this variation, I highlight the underexplored structural aspect of clientelism. Specifically, I argue that the effect of clientelist practices on new party formation critically depends on the way clientelist exchange takes place. Drawing upon bargaining theory, I posit that when extensive clientelist practice is coupled with a decentralized resource structure, these abundant resources actually reduce political actors’ incentives to pledge allegiance to the incumbent party, as their resource needs can then be easily met through alternative resource channels. Under such circumstances, the combination of clientelist resources and decentralized clientelist structure creates an incubator for the birth of new parties. The analysis of new party formation is conducted at three levels. First, I examine new party formation from a cross-national perspective using an aggregate measure of clientelist level and structure created from the expert survey dataset. Second, I evaluate the implications of my theory at the party level. I show that parties’ clientelist efforts are less effective in producing actual votes if clientelist structure is decentralized. Finally, I conduct an individual-level analysis drawn from cross-national mass survey data in order to examine whether clientelist level and structure affect citizens’ votes for new parties. In Chapter 3, the second essay, Party Extensiveness, Clientelism, and Party Nationalization: The Mediating Effect of Clientelism, asks why some parties are competitive throughout the country, while others appeal to only a few specific regions. This study examines if and how much informal mechanisms of clientelism mediate the impact of formal party structure on party nationalization (i.e. the patterns of territorial vote distribution of parties). To this end, I explore the relationship among three variables: party structure, effectiveness of clientelism, and party nationalization. I hypothesize that extensive party structure enhances the effectiveness of parties’ clientelist efforts, which in turn decreases party nationalization. Where clientelist efforts 16 are effectively translated into electoral gains, elites and voters are more likely to establish relationships beyond party organizations. Therefore, I argue that effective clientelism undermines a party’s organization as a collective entity with a common goal. As existing studies suggest, parties with extensive organizations tend to have higher levels of party nationalization. However, this positive impact will be suppressed to the extent that these organizational attributes improve effectiveness of clientelism. This chapter takes the party as the unit of analysis. To evaluate my hypotheses, I conduct mediation analysis. Finally, my third essay in Chapter 4, Clientelism as an Informal Determinant of Congruence, examines the impact of clientelism on the level of congruence. Conceptualizing clientelism as a tool of persuasion employed by political parties, I contend that elites distribute clientelistic rewards in order to alter the preferences and behaviors of voters who would otherwise hold different or no views about parties’ policies. In response to the rewards, I argue, some voters will develop an affinity to a patron party and vote for that party despite its being located further than other parties from their ideal points. Consequently, clientelism systematically widens the gap between voters’ preferences and their vote choices, thereby undermining congruence. I challenge the conventional assumptions that individual voters have one-dimensional and single-peaked political preferences. Instead, I suggest that voters’ utility is a function of both issue position and material inducement. In sum, my dissertation examines the varying patterns of party system institutionalization in light of informal mechanism of clientelism. Table 1.1 presents the organization of this dissertation. Each chapter investigates different aspects of relationships among key actors, which have important implications for party system institutionalization. In explaining the frequency of new party emergence, Chapter 2 conceptualizes the exchange between patron and client as a form 17 of bargaining. Empirically, my model illustrates that there is an interaction effect between the structure and the level of clientelism. In my analyses at the country, party, and individual levels, I find that formal electoral rules and informal mechanism of clientelism have substitutive effect. In new democracies where formal electoral rules are weak, clientelist logic plays a crucial role in the dynamics of electoral competition. Chapter 3 describes clientelism as an obstacle to intra-party coordination. My analysis focuses on the mediating effect of clientelism on the relationship between party structure and party nationalization outcomes. My findings on party nationalization demonstrates the cases in which informal rules of clientelism are incompatible with formal rules in that elites' clientelist practices undermine formal electoral rules and party organizations. Finally, in Chapter 4, the analysis of congruence theorizes that clientelism serves as a persuasion tools employed by parties and elites. The cross-national analysis illustrates that informal logic of clientelism has accommodating relationships with formal democratic rules. In electoral competition, parties and elites abide by the formal rules of the game, while being managed to stay ahead of the game by means of clientelist rewards. Taken together, my findings demonstrate that the diverse relationship between formal and informal institutions together shapes the political landscapes. 18 Table 1.1 Organization of the Dissertation Chapter 2 Outcomes New party emergence Relationship of Inter-Party Relationship Interest Theoretical Role of Patron-client exchange Clientelism as a form of bargaining Empirical Role of Interaction between the Clientelism structure and level of clientelism Relationship with substitutive formal institutions Unit of Analysis country; party; individual Methodology Regression analysis (Poisson, linear, logit) Chapter 3 Party nationalization Intra-Party Relationship Chapter 4 Ideological congruence Party-Voter Relationship Clientelism as an obstacle to intra-party coordination Clientelist rewards as a tool of persuasion The mediating effect of clientelism on the relationship between party extensiveness and party nationalization Level of clientelism competing accommodating party country Mediation analysis; simultaneous equation model Multivariate analysis 19 CHAPTER 2 CLIENTELIST POLITICS AND NEW PARTY FORMATION 2.1 Introduction As the wave of democratization continues to spread throughout North Africa and the Middle East, the emergence of new political parties in nascent democracies has recently regained scholarly attention. Research on new party formation originates from Duverger’s study, and it has subsequently branched into two parallel literatures. The first sociological approach emphasizes the role of socioeconomic conditions and values orientation of a society (Lago and Martinez 2010, Clark and Inglehart 1998). According to this perspective, new parties are more likely to form when established parties fail to address new social cleavages. The institutional approach, on the other hand, attributes new party formation to the cost of entry and the benefits of office induced by institutional rules (Cox 1997; Hug 2001; Tavits 2008). Nevertheless, neither the sociological nor the institutional approach adequately accounts for new party formation in young democracies. In new democracies social cleavages rarely change as often and fast as party systems. Meanwhile, formal institutions are less institutionalized and often subject to manipulation by ruling elites. More bluntly, the relationship between formal institutions and new party formation might be spurious, as political institutions themselves are nested inside the strategic consideration of political actors. Lust (2009) echoes my proposition, arguing that elections as formal institutions matter much less than informal institutions like clientelism in shaping the interaction between voters and politicians in developing democracies. Towards a fuller understanding of how new parties emerge in new democracies, this paper moves beyond the current literature and presents a theory that highlights the importance of informal rules of clientelist politics. I emphasize the distinction between two intertwined yet different concepts of clientelist practice: one of level of clientelism and one of structure of 20 clientelism. The former is referred to as the amount of discretionary public resources available to the incumbent (such as public jobs, contracts, and services), whereas the latter is defined as the extent to which decision-making power over public resource distribution is centralized and monopolized by the incumbent. Importantly, I argue that the effect of clientelist resources on new party formation depends on the way clientelist exchange takes place. Specifically, drawing upon bargaining theory, I argue that when state resources are monopolized by the incumbent, an increase in the size of clientelist resource will reinforce the dominance of the incumbent party and hinder new party formation. However, when extensive clientelist practice is coupled with decentralized resource distribution, abundant state resources actually reduce political actors’ incentives to pledge allegiance to the incumbent party as their resource needs can be easily met by alternative resource channels. Under such circumstances, the combination of clientelist resources and decentralized clientelist structure creates an incubator for new parties. My analysis is conducted at three levels: First, I employ a cross-national analysis to directly examine whether new party formation depends on clientelist levels and structures. Second, I assess the implications of my theory at the party level, and I show that parties’ clientelist efforts are less effective in producing actual votes when clientelist structure is decentralized. Finally, I conduct an individual-level analysis to investigate whether clientelist levels and structures affect citizens’ electoral support for new parties. I find supportive evidence consistent across different levels of analyses. My results also survive a series of robustness checks. Before proceeding, it is useful to briefly use the case in Peru to illustrate my theory. Peru has a long history of clientelist politics, however, new parties constantly emerge to replace existing parties during elections. For instance, the major parties in the 1980s, such as C-90, FREDEMO, and APRA, degenerated into the minor parties with less than 2 percent of the vote in 2000 21 (Levitsky and Cameron 2003). Recently in the 2011 general election, 8 out of 10 candidates came from new parties and alliances (Lupu 2012). In Peru, clientelist exchange occurs in an atomized and temporary way among national politicians, mediators, and local citizens. During elections politicians “improvise” political parties to establish privileged links with local powers. In turn, autonomous local organizations continuously change their affiliations to new political parties (Munoz Chirinos 2013: 2). According to Levitsky and Cameron (2003), this interesting dynamics between new party formation and clientelist politics in Peru can be attributed to the growth of informal urban economy in the late 1980s that make resources for clientelism available from multiple channels. In other words, alternative resources from the informal sectors created a decentralized clientelist structure, which consequently contributed to the growth of new political parties in Peru. The rest of the paper is organized as follows. Section 2 briefly reviews the existing literature and highlights the gap in the theory about new party formation in new democracies. In addition, it provides a theoretical framework of the interaction effect between size and structure of clientelism on the emergence of new parties, and it derives a set of hypotheses. Section 3 empirically tests my hypotheses at three different levels. Finally, Section 4 concludes with a discussion regarding the interaction between informal and formal institutions. 2.2 Theoretical Framework Conventional wisdom holds that clientelist resources provide an unfair playing field for ruling elites to dominate party systems. This branch of scholarship particularly emphasizes the patron’s ability to induce the clients’ compliance by punishing potential challengers and rewarding loyal supporters (Stokes 2005; Hicken 2011). For instance, in his study on how traditional patrons 22 pressured villagers in central India, Krishna argues that the patron’s authority is “inherited and unquestioned, met with fear and subservience on the part of ordinary villagers” (Krishna 2007: 146). Yet, if the conventional wisdom is true, then how do we explain the frequent emergence of new parties in new democracies where clientelist politics is the norm rather than the exception? If the cost is high for new entrants to enter the electoral arena, then how are new parties ever established in developing democracies? To solve this puzzle, I suggest that in countries with pervasive clientelist practices, the real challenges for the incumbent often come from their own parties. Indeed, defection of elite members from the incumbent party to form new parties has characterized the political landscape of many new democracies. In India, a country notorious for prevalent clientelist practice, elite defection has become a common phenomenon since the general election in 1967. 8 The extensive clientelist practice also fails to prevent the continuous elite defections and the eventual breakdown of ruling parties in Senegal,9 Lesotho,10 Nigeria,11 Mexico,12 just to name a few. 8. Between 1967 and 1970, 148 party members defected; between 1976 and 1977, one of every five members of the legislature was a defector, and 15 of these defectors became chief ministers. See International Referred Research Journal, April, 2011. pp.15-16. 9. In 1998, an extensive defection within the ruling Socialist Party of Senegal was initiated by Djibo Ka, who later established a new party, Union for Democratic Renewal. Later in the 2000 election, another wave of massive defection from the Socialist Party contributed to the historic victory of the opposition party led by Abdoulaye Wade. 10. Before general election in 2007, Lesotho’s parties faced extensive intra-party splits in the Parliament. In 1997, a faction leader of the ruling Basotho Congress Party (BCP), Mokhehle, broke with the party and established the Lesotho Congress for Democracy (LCD). In 2001, in turn, the leading LCD members left the party to form the Lesotho People’s Congress. Meanwhile, the Basutoland African Congress was formed following a split within the BCP in 2001. 11. Many Nigerian politicians had defected from the ruling People’s Democratic Party to form the new parties since 1999: e.g. Comrade Chukwuemerijie to the Progressive People Alliance, Chief Sergeant Awuse to Democratic People’s Party, Chief Kalu to the Progressives People’s Alliance, a senator Osakwe to Accord Party. 12. The elite defections also occurred in the long dominant Mexican party, PRI. For example, the prominent members of the PRI faction, Corriente Democra´tico (CD), left the PRI and joined the National Democratic Front (FDN) in early 1988. 23 Contrary to what the existing literature suggests, from these illustrative cases we can see that the emergence of new parties in many new democracies is not a manifestation of new issues or demands from the civil society. Rather, it is often the result of internal power struggle within the ruling parties. Levitsky and Way (2010) support my observations, showing that the major opposition movements draw most of their support from former ruling elites who defected from governing coalitions in their case studies of Georgia, Kenya, Malawi, Senegal, Ukraine, and Zambia. The next logical question is under what conditions political elites are more likely to defect from the dominant ruling parties, especially in countries with abundant state resources? In other words, what drives elite politicians and voters away from the incumbent patrons with state recourses to form and support new parties? To answer these questions, I highlight the structure of clientelism in which resources are mobilized and exchanged and distinguish the concept of clientelist structure from the traditional focus on clientelist size in the literature. Drawing from Worsnop (2013), I define the structure of resources in terms of directness and exclusiveness in resource provision. Directness refers to the distance between the national leadership and the local citizens with regard to resource provision. In other words, this concept captures the degree to which responsibilities of resource provision are delegated, or the vertical dispersion of power over resource distribution. On the other hand, exclusiveness examines if patrons are the sole providers of these resources. Put differently, exclusiveness refers to the number of dispensing resources, or horizontal distribution of authority over resource provision. Following Worsnop's insight, I argue that the structure of resource provision can be decentralized in two ways. First, decentralization of clientelist practices occurs when multiple patrons compete for the same group of clients. For example, Cornelius (1977) shows that clients 24 in Mexican urban slum enjoy greater autonomy than their counterparts in rural regions because urban clients had a wider choice of patrons. Second, decentralization of clientelism takes place when a single patron delegates the task of clientelist management to multiple intermediaries such as local elites. In such extended clientelist networks, ordinary citizens have little direct contact with their patron from the summits of national politics and only are linked to their patron through a chain of mediating relationships (Muno 2010). 13 Meanwhile, local mediators often play a bridging role connecting national politicians at the top and ordinary voters at the bottom.14 In sum, the first scenario captures the horizontal fragmentation of authority over resources, while the second scenario represents the vertical fragmentation. Both cases represent the proliferation of the alternative resource providers in clientelist exchange. The extent to which clientelist structure is decentralized has substantial implications on the power balance between the client and the patron. At one end of the continuum where the clientelist structure is highly centralized, patron-client relationship becomes highly repressive. Weingrod (1977) presented an unequal relationship between a community leader and the residents in Sardinian villages. He shows that because it was hard for the villagers to build any other relationships outside the village, they had to put up with unreasonable treatment by their patron otherwise they would lose resources. At the other end of the continuum where clientelist structure is decentralized, individuals are in a better position to make demands with greater freedom to leave the network on their own terms. Krishna (2007) describes the highly flexible patron-client 13. Given the complex reality of the relationship among patrons, clients, and mediators, the dyadic depiction of clientelism has limitations in explaining the cause and consequences of clientelist practice. For the discussions on the extended structure of clientelism, see Kitschelt and Wilkinson (2007), and Stokes (2007). 14. For example, Hilgers (2011) shows how the leaders of Francisco Villa Popular Front party from Mexico City exchange political support for help in accessing social housing credits with the urban poor and use their followers as leverage in negotiations for public sector jobs, political candidacies, and public resources with the Federal District government. 25 relationship between multiple political entrepreneurs (known as naya netas) and the individual villagers in central Indian village. Unlike patron-client relationship in Sardinian villages, the villagers are not locked into an unequal relationship with a single patron. They enter patron-client network with one naya netas who offer them “most resources to finance development works in their village, and they switch allegiance easily from one set of politicians to another” (Krishna 2007: 148). Upon a closer look at the contrasting cases of the Sardinian and the Indian villages, I see that the presence of alternative resource providers contributes to the greater equality between patrons and citizens. From the perspective of bargaining theory, the availability of alternative patrons in the Indian villages provides citizens with more exit options, which in turn allows them to strike a better deal with their patrons. In addition, players with exit options are more likely to leave the bargaining game since they are less tolerant of unfavorable terms and have no incentive to settle for smaller rewards than they can get elsewhere. Therefore the presence of exit options reduces a range of mutually acceptable agreements (Kato 1998). If we conceptualize a patron-client relationship as a bargaining game, we can easily see that decentralized clientelist structures provide multiple exit options by allowing greater numbers of resource suppliers. Importantly, when the patron is not the only supplier, he can no longer dictate the terms of clientelist deals or exercise unilateral coercion to induce compliance of the client. To the contrary, the clientele groups can shop around for the patron who offers them the best deal. Meanwhile, a decentralized clientelist structure can encourage ambitious politicians to create their own support base without entirely relying on ruling parties. With greater resource autonomy, the mediators and the local patrons can easily “break with or withhold support from national 26 politicians” to promote their own interests and those of their clients (Beck 2008: 16).15 Hence, contrary to the conventional expectations, clientelism may not always encourage compliance to dominant parties if the resource channel is decentralized. On the contrary, I expect that decentralized clientelist structure will encourage defection from ruling party and new party formation. Note that a decentralized clientelist structure can also reduce the durability of the existing patron-client relationship because of the accompanying information asymmetry. In a bargaining game when one party has more information about the key parameters pertinent to the negotiation, it can exercise greater bargaining power by exploiting her informational advantage. In decentralized clientelist settings, the national patron rarely has enough knowledge about the actual value of the exit options to the clients or the level of loyalty towards the current patron as opposed to alternative patrons. Given the difficulty for the national party leaders to get private information that may be relevant for the bargaining outcome, the patrons are in a disadvantageous position against the clientele group. Weinstein (2005) echoes this proposition and argues that, conditional on resource endowment, clientelist practices result in a separating equilibrium with low-committed individuals being more likely to assemble in a new selected group of patronage. Hence, as clientelist structures become increasingly decentralized, they inadvertently undermine the cohesion of party organization and result in party splits. In sum, drawing on these theoretical frameworks, I hypothesize that the effect of clientelism on new party formation depends on the way clientelist practices are managed and 15. According to Nash (1950), the bargaining problem can be defined as any contractual interaction between two parties “who have the opportunity to collaborate for mutual benefits” (Nash 1950: 155). The bargaining theory assumes that the two parties are rational in calculating their payoffs and that no party will enter into a contract unless they expect it to be beneficial (Rubinstein 1982: 97). In other words, bargaining represents the situation where rational actors have common interest in reaching an agreement on exchange terms, while they also have conflicting interests. 27 distributed. While in a highly centralized setting, clientelism will undermine the formation of new parties as the conventional wisdom suggests, I argue that if the clientelist structure is decentralized then this negative effect of clientelism on the new party formation decreases. 2.3 Empirical Analysis Given the causal complexity and the difficulty in measuring clientelist structures, I test my hypotheses using three different units of analysis. First, I conduct a cross-national analysis to examine if the number of new parties is a function of the aggregated measures of clientelist level and structure. Second, I examine the implications of my theory at the party level and see if parties’ clientelist efforts are less effective in yielding votes where clientelist structure is decentralized. Finally, at the individual level, I analyze whether clientelist politics induce citizens’ electoral support using public survey data. My research design is guided by the notion of triangulation (Denzin 1970). By drawing evidence from multiple measures and data sources at three different levels of analysis, I hope to increase the credibility of my findings and reduce potential biases and deficiencies from using only a single method of inquiry. As Webb et al. (1966: 3) persuasively argue, “…once a proposition has been confirmed by two or more independent measurement processes, the uncertainty of its interpretation is greatly reduced.” 2.3.1 A Cross-National Investigation I start my cross-national analysis by building on Tavits (2008)’ data on new party entry to parliamentary elections in East European countries, and I expand the data to cover 45 third wave 28 democracies around the world.16 Following Tavits, I only include countries that score at least 4 on both Freedom House scales to control the level of democratic development. To measure the formation of new parties, I count the number of new parties that have appeared in a given election for the first time and received a vote share of at least 0.3%. Note that literature disagrees on the definition of a new party. Some studies show that a significant portion of new parties are derived from a split from an existing party and hence consider the splits and mergers as new. They argue that in new democracies where individual politicians frequently reshuffle along the party lines, the difference between the splinters and the new parties is quite slim (Mair 1999; Lago and Martinez 2010). On the other hand, others limit the new party to the genuinely new organization (Bakke and Sitter 2005). Following Tavits, I consider a party new either if it splits from an existing party or is genuinely new.17 To measure the level of clientelism, I use a recent dataset on clientelism from the Democratic Accountability and Linkage Project (DALP) (Kitschelt et al. 2009).18 The DALP asks the respondents to evaluate how much effort parties make in their country to their electoral target constituencies. To facilitate the comparative study across parties and countries, the DALP introduces the five most commonly used “currencies” of clientelistic exchange, including gifts of consumer goods, preferential access to social policy entitlements, public sector employment, government contracts or procurement opportunities for business, and client influence over regulatory procedures. For each item, parties’ clientelist efforts are evaluated based on four grading 16. To compile the data, I referred to the following online sources: http://africanelections.tripod.com; http://psephos.adam-carr.net; http://www.eisa.org.za; http://electionguide.org; www.electionresources.org 17. I exclude the electoral alliances formed mainly by existing parties. 18. The expert survey has been conducted in 89 countries between 2008 and 2009. The respondents were asked to evaluate political parties in their countries regarding the ways in which parties attempt to connect with voters, ranging from programmatic to clientelistic efforts. For more details, see Kitschelt and Freeze (2011) 29 schemes from “a negligible effort or none at all” to “a major effort.” As Kitschelt (2011) posits, by framing the questions about clientelist practices based on concrete actions, the survey avoids the politically charged notion of clientelism. Operationally, I follow Kitschelt (2011) and create a composite index at the country level based on the five average measures of clientelist efforts of individual parties (weighted by their vote share in the most recent national legislative election) within a country. In addition, the DALP also provides useful information on the structure of clientelism by gauging the extent to which political authority is decentralized. Specifically, the DALP asks its respondents to describe the candidate nomination process for a given party and see whether national legislative candidates are chosen by national party leaders, a bargaining process between the different levels of party organization, regional/state-level organizations, or local/municipal level actors. The literature suggests that candidate nomination has substantial implications for power balance within a party organization and control over resources. Schattschneider (1942: 64) forcefully argues that “…the nature of nominating procedure determines the nature of the party; he who can make the nomination is the owner of the party.” Other studies report similar findings, showing that when nominations are centrally controlled, candidates are more likely to remain loyal to the party on whom the trajectory of their political careers depends (Rahat and Hazan 2001). Importantly, the literature considers the nomination structure as a crucial determinant of the way resources are distributed across the different organizational levels of the party. For example, Kemahlioglu (2005) compares the degree of controls over resources by local party organizations in Argentina and Turkey in light of the candidate selection structure. She argues that in Argentina the local organs of the parties hold greater control over resources than its Turkish counterparts because the sub-units of the Argentine party are engaged in the candidate selection process. Hence, 30 I expect that clientelist resources will be more centrally managed and distributed when a national party tightly controls the candidate nomination process and most decentralized when candidate selection process is more open and participatory. Finally, I follow the literature and take into account several institutional and structural factors that would affect the cost of entry, the benefits of office, and the potential of electoral support when elites decide whether to form a new party (Cox 1997; Hug 2001; Tavits 2008). First, my model controls for the costs of entry. The entry cost is expected to increase with the number of signatures required to put a new party name on the ballot. I measure it using the variable Petition available from the Institutions and Elections Project, which equals 1 if petition signatures are required and 0 otherwise. Secondly, the availability of public financing for new parties alleviates the entry cost for new parties, thus promoting their emergence. This variable is coded as 1 if public funding is provided and 0 otherwise.19 In addition, district magnitude accounts for how permissive the institutions are to the new challengers; for example, Cox (1997) shows that larger district magnitude encourages new party formation.20 Lastly, following Tavits (2008), I also control for the type of electoral system.21 The intuition is straightforward: the more proportional the system, the more likely new parties emerge as the cost of winning a seat is lowered. Next, the model controls the benefits of office. Studies have shown that when the elected position allow considerable power over policy making and the spoils of office, new parties have greater incentives to take on challenges to enter (Tavits 2008). To measure the decision making powers of chief executives, I consider the variable executive constraints available from Polity IV. This variable 19. The cross-national data on these variables are available for the year 2008 from the International Institute for Democracy and Electoral Assistance (IDEA). 20. I use the data on district magnitude from Database of Political Institutions (2009). 21. I use the measure of electoral systems from the Democratic Electoral Systems around the World compiled by Bormann and Golder (2013). 31 refers to the extent of institutionalized constraints on the decision-making powers of chief executives. I also take into account the length of democracy. The notion is that electoral history determines the level of uncertainty about electoral performance, thereby influencing the entry decision by the elites. Lastly, the literature suggests that disappointing economic performance works in favor of new entrants. Thus, I include inflation, unemployment rate, and GDP growth from the World Development Indicators. To reiterate, I argue that the effect of clientelism on new party formation is contingent on its structure. In a highly centralized setting, higher level of clientelism is negatively associated with the number of new parties. However, I expect that this negative effect declines as the clientelist structure decentralizes. To evaluate the hypothesis, I estimate the following interaction model at the country level: Number of New Parties = β0 + β1 𝑃𝑎𝑡𝑟𝑜𝑛𝑎𝑔𝑒 𝐿𝑒𝑣𝑒𝑙 + β2 𝑃𝑎𝑡𝑟𝑜𝑛𝑎𝑔𝑒 𝑆𝑡𝑟𝑢𝑐𝑡𝑢𝑟𝑒 + β3 (𝑃𝑎𝑡𝑟𝑜𝑛𝑎𝑔𝑒 𝐿𝑒𝑣𝑒𝑙 ∗ 𝑃𝑎𝑡𝑟𝑜𝑛𝑎𝑔𝑒 𝑆𝑡𝑟𝑢𝑐𝑡𝑢𝑟𝑒) + 𝛽𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑠 + 𝜀 Since the dependent variable is a nonnegative count variable, I use a Poisson regression with robust standard errors to model the data generating process.22 Table 2.1 shows the estimation results. The coefficients on the key explanatory variables are statistically significant with the expected signs, corroborating my argument that the effect of clientelist level on the formation of new parties is contingent upon the clientelist structure. Given the non-linearity nature of my model, I illustrate the conditional effect of clientelist politics in Figure 2.1. As we can see, as clientelist level increases, the predicted numbers of new party decrease. However, the downward slope for 22. My dependent variable does not have an excessive number of zeros. Nor does the variance of our dependent variable is greater than that of a true Poisson. 32 the decentralized clientelist structure is flatter than the one for the centralized clientelist structure, indicating that the negative effect of clientelist level on new party formation is smaller under decentralized clientelist structures. In addition, we can also see that at a higher level of clientelism, the predicted numbers of new parties are actually greater under decentralized clientelist structures. Together, these results highlight the importance of clientelist structure in shaping the dynamics of new party formation in new democracies. Table 2.1 Cross-National Determinants of the New Party Emergence DV: Number of New Parties Model 1 Clientelist Level Clientelist Structure Interaction Petition Public financing District Magnitude Electoral System Executive Constraints Inflation GDP growth Unemployment _cons N ll -0.8067*** (0.1967) -1.3241** (0.4739) 0.5857** (0.2101) -0.0981 (0.2438) -0.3608 (0.2737) 0.2458*** (0.0672) 0.2474 (0.1806) -0.1037 (0.0751) -0.0423* (0.0196) 0.0269 (0.0200) 0.0757*** (0.0102) -0.0102 (0.6966) 45 -110.7366 33 Table 2.1 (cont’d) chi2 97.5882 Note: Poisson regression model; each country’s exposure to democracy is considered in the model using the number of consecutive years of democracy. * p<0.05, ** p<0.01, *** p<0.001. Figure 2.1. Predicted Number of New Parties across Different Levels of Clientelism under Two Alternative Clientelist Structures 0 10 20 30 Predictive Margins of Patronage Structure .5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5 8 Patronage Level Centralized Decentralized Note: The figure is generated using Model 1 of Table 2.1. 2.3.2 A Cross-Party Investigation The cross-national evidence presented so far provides top-down support of my theory that intense clientelist efforts may be associated with increasing number of new parties under decentralized clientelist structure. In essence, I interpret this finding as evidence that decentralized clientelist structure reduces the credibility of threat to withhold clientelist resources and consequently undermines the durability of the existing patron-client relationship. Under such circumstances, efforts to provide clientelistic benefits may not necessarily translate to greater compliance and loyalty from the clients. In other words, I argue that parties’ clientelist efforts to 34 mobilize voters and sustain the cohesion of party organization would be less effective under decentralized clientelist environments. To further corroborate my argument, I test the implications of my theory at the party level using the DALP data. The DALP taps directly to the effectiveness of parties’ clientelist efforts by asking respondents to evaluate how effective parties are in their efforts to mobilize voters by targeted benefits. Operationally, I aggregate the score of perceived effectiveness from respondents at the party level, and regress the variable of clientelist effectiveness on the variables of clientelist size, clientelist structure, and their interaction term. Again, if my theory holds, clientelist efforts should be most effective when clientelist structure is centralized. As the clientelist structure decentralizes, clientelist practices become ineffective in inducing compliance and support from voters. This is indeed what I have found. As Figure 2.2 clearly demonstrates, clientelist efforts are indeed useful in inducing electoral support as the conventional wisdom suggests. More importantly, the upward slope for the decentralized clientelist structure is less steeper than the one for the centralized clientelist structure, indicating that the clientelist effort is less effective in mobilizing voters under decentralized clientelist environment. 23 The difference between centralized and decentralized clientelist environment is statistically significant, as evident in the non-overlapping 95% confidence interval. In sum, the results show that the effectiveness of clientelist efforts in securing electoral support critically depends on the clientelist structure. 23. The numerical results are shown in Appendix, Table 2.2. 35 5 4 3 2 1 Effectiveness of Clientelist Efforts Figure 2.2. Estimated Effects of Clientelist Level on Effectiveness of Clientelism under Alternative ClientelistEffectiveness Structure of Targeting 5 10 15 20 Patronage Level Centralized Decentralized Note: The figure is generated using Model 2 of Table 2.3 in Appendix. The vertical lines present 95% confidence interval. 2.3.3 Individual-level Investigation Finally, I test the implications of my theory on voters’ electoral behavior using the most recent Afrobarometer survey data. Following my theory, I argue that decentralized clientelist structure provides the clients alternative providers of state resources, which in turn makes it less likely for them to stay in the existing clientelist network. Hence, by implications, I expect that voters will be less likely to support the ruling party when they perceive more diversified channels of resource management and distribution. The dependent variable in this individual-level analysis is voters’ support for ruling parties. This binary variable is coded from the survey question “If a presidential election were held tomorrow, which party’s candidate would you vote for?” I identified a full list of ruling parties at the time of the survey for each country included in dataset, and I coded as 1 if respondents would vote for the ruling party and 0 otherwise. To measure the level of clientelism at the individual level, 36 I follow Young (2009) and utilize a question on whether and why respondents contacted formal or informal leaders. The choices are “did not contact any”, “community problem”, “personal problem”, which I recode into 0, 1, 2, where 2 indicate the highest level of clientelism. According to Young (2009), these questions on the Afrobarometer survey best serves as “a test for an effect of personal, clientelistic networks on voting behavior”. Although a contact with leaders in and of itself is not necessarily equivalent to clientelism, direct and interpersonal interaction has been considered the most fundamental characteristic of clientelistic exchange. Hicken (2011: 291) correctly notes that “…as long and complex as this broker network may become, most scholars stress that at the core are personal, dyadic relationships between individuals.” Thus, to trace clientelism, the literature has increasingly used survey questions on whether individuals have contacted elected officials for help (Kitschelt 2011). As to the measure of clientelist structures, I capture the decentralization of discretionary power over clientelist resources by examining how many different sources of authorities the respondents rely on to meet their needs. Specifically, this measure taps into the perception of respondents regarding who they believe has the power and resources to help them. If the respondents contacted more diverse authorities, I expect that they consider the authorities and resources are more decentralized. Hence, I created the composite index based on whom respondents contact about important problems. The choices include “local government councilor”, “a member of parliament”, “official of a government agency”, “religious leader”, “traditional ruler”, and “other influential persons”. I recoded each variable as 0 if the respondent chose “Never” and 1 if they contacted more than once. Then, I created the index by summing up the six questions. Consequently, the variable ranges from 0 to 6, where 6 represent highest level of (perceived) decentralization. Lastly, I control the respondents’ perception on the previous and future economic condition. Building upon the insights from the economic voting 37 literature, I expect voters to base their vote on whether incumbents are responsible for economic outcomes. I also control for the level of education of the respondents. The finding from the logistic regression provides strong empirical support to my theory.24 As we can see in Figure 2.3, the relationship between clientelist level and voters' support for ruling parties is highly contingent on clientelist structure. When clientelist structure is highly centralized, an increase of clientelism from the median level to the 75% level would lead to voters’ incumbent support by 10%. However, as voters recognize greater dispersion of the authority over resources, the effect of increasing in clientelist level on voters’ incumbent support starts to reduce substantially. This result supports my hypothesis that clientelist efforts do not have uniform effects on voters' support for ruling parties; instead, incumbents' clientelist efforts pay off electorally only when the clientelist structure is centralized. 40 20 0 -.02 -.01 0 % of observation .01 60 .02 Figure 2.3. First Difference Measuring the Effect of Clientelist Level across Varying Clientelist Structure 0 Low 2 4 6 High Patronage Structure Note: Pr(Vote for Ruling Party=1 | Clientelism= 75 percentile) - Pr(Vote for Ruling Party=1 | Clientelism= 50 percentile). This figure is generated using CLARIFY based on Model 3 of Table 2.4 in Appendix. 24. Again, the numerical results are shown in Appendix, Table 2.4. 38 2.3.4 Robustness Check To buttress my empirical analysis further, I now undertake a series of robustness checks to ensure the validity of my previous results. To begin with, I employ a variety of alternative indicators for my key variables, i.e. the level and the structure of clientelism, to ensure that my results are not driven by arbitrary choice of measurements. First, in terms of the level of clientelism, I use a different question from the DALP survey to measure the level of clientelism. This alternative question asks country experts to evaluate the extent to which parties mobilize electoral support by delivering targeted material benefits. The responses are scaled on a 1-4 metric, where 1 indicates “not at all” and 4 “a great extent.” In a nutshell, the question captures whether parties establish relationship between politician and voters through a programmatic or through a clientelistic linkage. Unlike the previous measurement, this question is not based on parties’ performance in actually targeting clientelist resources, but on “an explicit or implicit statement of intent” (Kitschelt and Freeze 2011: 23). Note that empirically these two measurements correlate only modestly (77%). In additions, one might question the subjective nature of these two measurements since both of them are drawn from survey data and hence might capture only the perception of the clientelist politics rather than the reality of it. To alleviate this concern, I cross-check my results by employing a commonly used, objectively-based measurement on the clientelist spending: the government expenditure on wages and employer contributions as the percentage of total government expenditure. In the absence of a direct measurement, the existing literature considers the spending on personnel as the most widespread manifestation of clientelist practice (Grzymala-Busse 2008). As Chubb (1982) puts it, “…politics revolves around the job or position and a job signifies a vote and vice versa.” 39 Second, regarding the structure of clientelism, the DALP also provides valuable information regarding the extent to which clientelism is decentralized by examining whether the patron delegates the task of clientelist management to intermediaries. In particular, it ask whether the parties “…have local intermediaries (e.g. neighborhood leaders, local notables, religious leaders) who operate in local constituencies on the parties’ behalf, and perform a variety of important tasks such as maintaining contact with large groups of voters, organizing electoral support and voter turnout, and distributing party resources to voters and supporters?” The responses fall into three categories, ranging from “they have local representatives in most constituencies,” “they have local representatives in some constituencies,” to “they have almost no local representatives.” Using this information, I create an alternative indicator of clientelist structure based on the presence of intermediaries. I argue that while parties may benefit from strong informal network of intermediaries to reach out to local voters, the capacity of a national party to control the clientelist network will inevitably diminish as network grows. Moreover, I use Vanhanen’s (2003) indicators of power resource distribution to capture the structure of clientelism in a more objective manner. The Vanhanen indicator ranges from 0 to 100, where 100 represents high relative distribution of power resources. This measure is calculated by taking into account the diversification of occupation, the education level of the population, and the level of decentralization in non-agricultural economic resources. Many studies have used this indicator to measure the various dimension of inequality (Amendola et al. 2013), and I argue that this indicator can serve as an indirect yet useful measurement for the power distribution within the clientelist network. Model 4 and Model 5 in Table 2.5 (shown in Appendix) reports the results from using these alternative subjective and objective measurements respectively. For both models, the coefficients 40 of clientelist level and structure as well as the interaction term are statistically significant with the expected signs. These results corroborate that my previous findings that increases in clientelism can lead to more new party formation when the clientelist structure is decentralized. Finally, recall from my individual-level analysis that voters are more likely to withdraw their support for the incumbent in decentralized clientelist environments. To ensure the validity of this finding, I further collect data on the vote share of the new parties and examine whether they depend on the interaction between clientelist level and the clientelist structure. As expected, I find that as clientelist structure decentralizes, higher level of clientelism is associated with higher level of vote share for new parties. The result is in line with my micro-level evidence. Jointly, these results represent the micro-foundation for the electoral outcome of new parties. In sum, all the analyses presented so far underscore my theory: the effect of clientelist level on new party formation depends on the way clientelism is structured. The finding that clientelist resources have a conditional effect on new party formation is shown to be remarkably resilient, surviving rigorous robustness tests. As Table 2.2 summarizes, my results hold across three different units of analysis and various alternative measurements. 41 Table 2.2 Summary of the Measurements of Variables Unit of Analysis Countrylevel Dependent Variable Number of New Parties Independent Variable: Clientelist Level 1: Clientelist efforts with five specific goods and services Independent Variable: Clientelist Structure 1: Nomination structure 2: Presence of intermediaries 2: Intensity of parties’ clientelist efforts 3: Vanhanen’s indicators of power resource distribution 3: Government expenditure on wages and employer contributions Party-level Individuallevel 2.4 Effectiveness of Clientelist Efforts Voters' Support for Ruling Parties Clientelist efforts with five specific goods and services Nomination structure Whether respondents contacted formal or informal leaders Number of authorities respondents rely on Conclusion As a fundamental means of building political support, I concur with the literature that clientelism plays a crucial role in shaping the party system. As Medina and Stokes (2007: 69) puts it, “…no democracy on earth has found its way to preventing all forms of patronage”. However, I disagree with conventional wisdom on how clientelism shapes the party system, and I emphasize the importance of structural aspect of clientelist resources. I conceptualize the clientelist exchange as a form of a bargaining, and I argue that increasing clientelism accompanied with a decentralized resource control leads to greater numbers of alternative suppliers. This, in turn, creates a competitive environment empowering voters with greater freedom to leave the clientelist network and join new parties. In addition, decentralized supply channels also increase information asymmetry between patrons and clients. Taken together, they shift the bargaining power away from the existing parties and undermine their organizational integrity, while facilitating challengers to form new political parties. 42 Some might wonder why party leaders decentralize the clientelist structure in the first place if it encourages new challengers. Although a comprehensive investigation of the origin of decentralization is beyond the scope of this paper, I expect that the decision of decentralization is partially driven by the trade-off between holding control over resources and extending support base. As Slater (2004) shows, even Suharto, one of the most powerful authoritarian rulers, had to put a lot of efforts to constantly create and nurture supports for the regime. However, given its inherently non-universalistic nature, the extent to which clientelist practices can induce the support is bound to be limited. Therefore, politicians have incentives to delegate their redistributive authority to those who can attract more supports on behalf of them. At the same time, this process inevitably leads to the surrender of power to intermediary authorities. As Baldwin (2011) suggests, African politicians often give up power over the allocation of critical resources to local leaders to receive more electoral support beyond their core ethnic group. I note that there are other types of clientelism aside from public resources. Arriola (2012) argues that private sources such as support from business sectors can also facilitate the formation of opposition coalitions. Due to the data constraints, my empirical analyses mainly focus on the structure of control over public resources. However, conceptually I expect that the existence of private sources would strengthen my theoretical prediction as they further decentralize the clientelist structure and hence facilitate new party formation. Finally, this paper contributes to the growing literature on the relationship between formal and informal institutions. As Helmke and Levitsky (2004) show, the formal and informal institutions can be mutually reinforcing, supplementary, conflicting, and accommodating. By showing how clientelist politics shapes party systems in new democracies, this paper demonstrates how the informal rules constrain and modify the way formal rules evolve. 43 APPENDIX 44 Table 2.3. Effect of Clientelist Level and Structure on Effectiveness of Clientelism DV: Effectiveness of Clientelism Model 2 Clientelist Level (b15) 0.1606*** (0.0232) Clientelist Structure (a5n) 0.3045* (0.1315) Interaction (b15Xa5n) -0.0285* (0.0122) _cons 0.7241* (0.2794) N 506 r2 0.4372 F 49.4247 Clusters 88 ll -330.9603 Note: Linear regression model. Robust standard errors clustering at country level in parenthesis; * p<0.05, ** p<0.01, *** p<0.001 Table 2.4. Effect of Clientelist Level and Structure on Voters' Supports for Ruling Parties DV: Support for Ruling Parties Clientelist Level Clientelist Structure Interaction Education Retrospective Voting Prospective Voting _cons N ll chi2 Model 3 0.0475* (0.0208) 0.1964*** (0.0362) -0.0369* (0.0156) -0.1393*** (0.0099) 0.1202*** (0.0172) 0.3052*** (0.0159) -0.5317*** (0.1064) 16796 -9322.7891 3767.7168 45 Table 2.4 (cont’d) Note: Logistic regression model. Country dummies are not reported on the table. Robust standard errors in parentheses. ***p<0.01; **p<0.05; *p<0.1. Table 2.5. Robustness Checks for the Cross-National Determinants of New Party Emergence DV: Number of New Parties Clientelist Level Clientelist Structure Interaction Petition Public financing District Magnitude Mixed (Electoral System) Presidency Executive Constraints Log(Length of Democracy) Inflation GDP growth Unemployment lnalpha _cons _cons N N cluster ll chi2 Model 4 -0.2808*** (0.0762) -4.7512*** (1.2071) 0.1092*** (0.0311) -0.6476* (0.2936) -0.6275* (0.3143) 0.0041*** (0.0009) 0.4186 (0.2224) -0.0246 (0.1203) -0.0801 (0.0882) -0.2474 (0.2917) -0.0929*** (0.0193) 0.0013 (0.0200) -0.0604* (0.0255) Model 5 -0.0227* (0.0105) -0.0564*** (0.0171) 0.0015* (0.0007) 0.9149* (0.4271) 0.3859 (0.2906) -0.0036 (0.0045) 0.2262 (0.1338) -0.0622 (0.1094) -0.1334* (0.0555) 0.1645 (0.1394) 0.0002* (0.0001) -0.0335 (0.0228) -0.0081 (0.0111) -1.6371*** (0.4139) 14.9899*** 1.8448** (3.3172) (0.7117) 32 101 35 -62.3588 -236.0991 101.5963 136.2759 46 Table 2.5 (cont’d) Note: Model 4 is produced based on the subjective measures from DALP dataset. Model 5 is estimated based on the objective measure from World Development Indicators and Vanhanen index, which is used to create the proxy for clientelist level and clientelist structure, respectively. Table 2.6 Cases Included Albania Egypt Macedonia Romania Argentina El Salvador Mali Russia Bangladesh Estonia Mauritius Slovakia Benin Georgia Moldova Slovenia Bolivia Ghana Mozambique Tanzania Brazil Guatemala Nicaragua Turkey Bulgaria Honduras Niger Ukraine Chile Hungary Nigeria Uruguay Colombia Jamaica Panama Zambia Croatia Kenya Paraguay Czech Rep. Latvia Peru Ecuador Lithuania Poland 17 countries in Eastern Europe and post Soviet Union 14 countries in Latin America 2 countries in North Africa & the Middle East 10 countries in Sub-Saharan Africa 1 countries in South Asia, 1 countries in The Caribbean 0 5 10 15 Figure 2.4. Predicted number of new parties across different levels of clientelism under two Predictive Margins with 95% CIs alternative clientelist structures 4 14 24 34 44 54 Patronage Level Centralized Decentralized Note: The figure is generated using Model 5 of Table 2.5. The vertical lines present 95% confidence interval. 47 CHAPTER 3 PARTY EXTENSIVENESS, CLIENTELISM, AND PARTY NATIONALIZATION: THE MEDIATING EFFECT OF CLIENTELISM 3.1 Introduction Why are some parties competitive throughout the country, while others appeal to only a few specific regions? Recent years have seen an increasing interest in understanding the pattern of territorial vote distribution of parties, which is called party nationalization.25 From the supply side perspective, party nationalization captures the number of districts in which parties and candidates are competing in a given country (Kasuya and Moenius 2008) and the degree to which copartisans coordinate to run under a common party label (Cox 1997). 26 From the demand side, it presents voters’ electoral behaviors across districts in terms of the degree to which they offer similar levels of support to the parties throughout districts (Morgenstern et al. 2009). Highly nationalized parties would be equally strong in every district. In contrast, marginally nationalized parties gain most of their votes in a few specific districts (Bochsler 2010). To achieve a consistent level of support across districts, i.e. high party nationalization, the literature recognizes the importance of efficient coordination among elites and voters on a common party label. As a crucial determinant of elites' incentives and capacities to coordinate, prominent explanations for the level of party nationalization have focused on how political 25 Note that party nationalization is different from another commonly used term, nationalization, which is defined as the process of taking a private assets into public ownership by a national government. 26 Literature defines nationalization in terms of several different aspects (Schattschneider 1960; Morgenstern et al. 2009, Schakel 2013): First, static (or distributional) nationalization (Morgenstern et al. 2009) refers to the degree of homogeneity in the geographic distribution of a party’s vote or the consistency of a party’s support across a country; Second, dynamic nationalization is about the degree to which a party’s vote in the various districts changes uniformly over time (Stokes 1967; Brady 1985; Bawn et al. 1999). Third, multilevel nationalization takes into consideration the multilevel dimension of party systems (Schakel 2013). Finally, party linkage nationalization focuses on the extent to which party candidates come together under common party labels (Chhibber and Kollman 2004; Cox 1997). This study focuses on the static nationalization and the party linkage nationalization. 48 authority is distributed, which is manifested by organizational structures. For example, parties with territorially comprehensive structures are associated with a greater institutionalization (Mainwaring 1999), greater party cohesion (van Biezen 2000), and greater capacity to effectively perform and shape political processes (Szcerbiak 2001). At the party system level, likewise, as national governments exert more control over local areas, the literature demonstrates that voters and candidates are likely to have greater incentives to coordinate on common party labels (Chhibber and Kollman 1998) and achieve higher nationalization (Harbers 2010). While prior scholarship on party nationalization predicts exclusively positive effects from extensive political authority, the studies on intra-party politics suggest that a potentially contradictory mechanism might be at work. For example, Boucek (2012) argues that the extended dominance of a party can threaten a party organization by generating intra-party differences. Having a lack of external challenges, she argues, politicians are less concerned about obtaining votes for the party and more devoted to pursuing individual or factional ambitions. In a similar vein, Warner (1997) maintains that a monopoly leads politicians to drop their ‘competitive guard’ and channel their efforts into clientelism. In this situation, individual politicians gain at the expense of the party organization’s common purpose. I argue that the key to understanding the contradictory effects of extensive party structure lies in the informal mechanism of clientelism. This is an informal contractual relationship in which patrons distribute resources to supporters in order to strengthen their political positions (Piattoni 2001).27 Literature on clientelism argues that clientelist practices are more effective for parties that 27 Clientelist resources include a wide variety of monetary and non-monetary rewards such as jobs, goods, licenses, and other public decisions. These benefits can be distributed by a variety of agents such as local notables, government or party bureaucracies, and party activists; whoever has access to the resources (Warner 1997). 49 have extensive organizations because these organizational attributes expand parties’ monitoring and distributing capacities (Kitschelt and Kselman 2010). In turn, another strand of literature on the consequences of clientelism proposes that this practice tends to undermine institutionalization of parties and party systems and cultivates more particularistic linkages with voters and politicians (Warner 1997; Kitschelt 2000; Hicken 2011). The findings from these two bodies of literature on clientelism imply that extensive political structure might have negative effects on party nationalization: extensive organizational structure facilitates the effectiveness of clientelism, which in turn undermines parties’ abilities to coordinate on the common party label. In spite of its potential implications, the role of informal mechanisms of clientelism has largely been overlooked in the party nationalization literature. The goal of this study is to examine if and how much informal mechanisms of clientelism mediate the impact of formal party structure on party nationalization. To this end, I explore the relationship among three variables: party structure, effectiveness of clientelism, and party nationalization. Effectiveness of clientelism is defined as the extent to which parties turn their clientelistic efforts into electoral success (Kitschelt and Kselman 2010). In terms of party structure, I focus on extensiveness, which is commonly seen as crucial determinants of party performance. Drawing on Harmel and Janda (1982), extensiveness of a party is defined in terms of its structural capacity to reach voters across a country. The concept is manifested by parties’ geographic reach or territorial presence through local offices, affiliated organizations, and established relationships with local notables (Kitschelt and Kselman 2010; Tavits 2012). I posit that extensive party control structure enhances the effectiveness of parties’ clientelist efforts, which in turn decreases party nationalization. This implies that party nationalization partially depends on the informal mechanism of clientelism. As existing studies 50 suggest, I admit that extensive political authority does have a positive effect on party nationalization. However, I insist that this positive impact will be suppressed to the extent that these organizational attributes improve effectiveness of clientelism. Where clientelist efforts are effectively translated into electoral gains, elites and voters are more likely to establish the relationships beyond party organizations. Therefore, I maintain that effective clientelism threatens a party's existence as a collective entity with a shared goal. My study takes a party as the unit of analysis.28 In my analysis, 256 political parties from 49 countries are included. 29 To obtain the party-level data on party nationalization, I use the Constituency-Level Elections Archive (CLEA) datasets which contain the measure of party nationalization for a particular party in a given country-year dyad. In addition, I obtain the data on effectiveness of clientelism and extensiveness of party structure from the Democratic Accountability and Linkages Project (DALP). My hypotheses suggest a causal chain of relations where an independent variable (i.e. extensiveness of party) affects a mediating variable (i.e. effectiveness of clientelism), which then affects an outcome variable (i.e. party nationalization).30 To empirically evaluate my hypotheses, 28 Party system nationalization is a weighted sum of party nationalization of all parties in a given country. My analysis focuses on party nationalization of individual parties. 29 In my case selection, I focus on the established parties with more than 5% of vote share. I follow the criteria of case selection of DALP dataset outlined by Kitschelt (2013). Political parties included are from the countries with at least two million inhabitants with at least two recent rounds of national elections under at least semi-democratic conditions with the Freedom House Index of at least 4.0. Regionally, my cases are distributed as follows: 51 parties in Eastern Europe and post Soviet Union, 46 parties in Latin America, 5 parties in North Africa and Middle East, 20 parties in Sub-Saharan Africa, 90 parties in Western Europe and North America, 45 parties in Asia (East, South-East, South). 30 The term ‘causal’ is used because the path diagram represents a prediction about possible causal influence. However, given that my data come from non-experimental designs, I can only test whether a hypothesized causal model is plausible and consistent with the data. 51 I conduct mediation analysis based on three sets of linear regressions.31 In addition, I conduct two additional significance tests on the mediating effects of clientelism: i.e. the Sobel test and nonparametric bootstrap sampling analysis. The results from these different methods provide consistent support for my hypothesis that clientelism mediates the effect of party structure on party nationalization. My study contributes to research on party nationalization by situating this underexplored informal mechanism within the context of current theory on the determinants of party nationalization. 3.2 Literature Review Building on earlier studies of party system development and institutionalization, recent scholars have investigated the causes and consequences of party nationalization, as well as methods of measurement (Chhibber and Kollman 1998; Jones and Mainwaring 2003; Brancati 2008; Lago and Montero 2009; Morgenstern et al. 2009; Bochsler 2010).32 The literature maintains that party nationalization has important implications for key aspects of political outcomes. First, where a party draws most of its support from a few specific districts, i.e. low party nationalization, 31 Hypotheses regarding mediated effects are common in political science research. For example, Nelson et al. (1997) maintain that the effect of news frames on tolerance for the Ku Klux Klan is mediated by the perceived importance of public order values. In the study of political behavior in post-communist countries, Powers and Cox (1997) argue that understandings of the past mediate the relationship between personal circumstances and satisfaction with economic reforms. Using mediation analysis, Brader et al. (2008) present that anxiety mediate the effect of racialized news coverage on public opinion about immigration. Arceneaux (2012) proposes that the effect of political argument on public opinion is mediated by anxiety. Johnson et al. (2012) claim that campaign receipts mediate the effect that race has on the decision of a House member to seek a Senate nomination. Newman et al. (2013) demonstrate that an acculturating context exerts a mediating effect on the preferred amount of immigration through its impact on cultural threat. 32 The concept of institutionalization captures the collective stability of inter-party competition and parties’ relationships with society at the party system level (Mainwaring and Scully 1995). However, within the party system, individual parties may vary in terms of the level of institutionalization. The patterns of party nationalization capture this underexplored aspect of intra-party dynamics. Specifically, a low level of party nationalization represents a weak shared identity of a party’s members and constituents. 52 it may favor its stronghold over other areas in policymaking decisions (Jones and Mainwaring 2003). In Ethiopia, for example, 80% of the parties in the House of People’s Representatives have regional bases. As Chege (2007) finds, their policy agendas are largely limited to region-specific issues. Second, studies argue that a more nationalized party cultivate highly coordinated settings which facilitate executives’ efforts to build legislative coalitions based on national issues (Jones and Mainwaring 2003). Thus, scholars propose that nationalized parties and party systems tend to generate more responsive government (Lupu 2008). Finally, the literature contends that the nationalization of politics makes elections meaningful by making power alternation more likely (Schattschneider 1960). In new democracies, the foundations of opposition parties are often limited to ethnic and regional constituencies. In the absence of an established nationwide membership, representation remains local and they are less likely to become a viable alternative to incumbents. According to the party nationalization literature, the degree of vote dispersion of a party is crucially determined by coordination of candidates and voters across districts (Cox 1999). When electoral competitors refrain from entering the race independently and converge on a common nation-wide party platform, similar patterns of electoral competitions emerge across districts (Chhibber and Kollman 1998). Conversely, when elites fail to ally with other groups or candidates, competitors and parties are likely to proliferate at the regional level (Hicken 2009). In this setting, electoral competitions occur among different sets of parties across districts and the electoral behavior of voters and candidates is likely to diverge from district to district. Thus, vote shares obtained by candidates from the same party vary considerably across districts. To explain variations in party nationalization, therefore, the literature has investigated institutional determinants that might affect incentives for elites and voters to coordinate. Among 53 others, institutions defining the structure of political authority and control have been considered as one of the most important determinants of party nationalization (Brancati 2008). Prominent studies on party nationalization share the view that the distribution of political authority is a crucial factor determining incentives to coordinate (Hicken 2002; Chhibber and Kollman 1998). At the party system level, for example, Chhibber and Kollman (1998) contend that as governments centralize power and their policies affect local areas, candidates tend to engage in national-level organizations and voters support nationally competitive parties over regionally-based ones. As national governments exert more political or economic control over local areas, the distribution of votes becomes more similar across districts because candidates and voters coordinate on the national level to influence policy outcomes (Harbers 2010). Likewise, at the party level, the internal distribution of authority is expected to affect parties’ capacities to develop more coherent electoral strategies, broaden their support bases and induce politicians and voters to coordinate under common party labels (Szcerbiak 2001; Scott 2004). Specifically, Mainwaring (1999) associates territorially comprehensive parties with greater system institutionalization. In a similar vein, Tavits (2012) defines organizational strength and success in terms of the extensive structure of political parties. She argues that these parties are electorally advantageous because they can effectively address environmental challenges and formulate policy. Perceived as reliable and capable, a party with extensive structure is capable of inducing elites and voters to coordinate. The discussion from existing studies leads to the following hypothesis about the internal distribution of authority at the party level: H1: As party organization becomes more extensive by exerting greater political control over local areas, the level of party nationalization increases. 54 Although existing studies on party nationalization found that formal institutions play a significant role in shaping party nationalization, their explanations are unsatisfactory in that they have rarely considered the challenges of coordination derived from informal institutions. The extant literature assumes that political elites and voters are motivated to coordinate across districts by programmatic or ideological causes. In particular, the goal to influence policy is considered as the key motivation to coordinate on parties at the national level (Chhibber and Kollman 2004). As Mainwaring (1999) points out, however, party competition rarely occurs solely over policyoriented issues. In a setting where programmatic issues play a limited role, informal rules and particularistic relationships matter more than formal party structure to the patterns of party competition. In particular, clientelist practices create intra-party groupings which add another dimension to elite competition within a party (Levitsky 2003). Therefore, clientelist practices carry implications for elites’ incentives to coordinate, which have been overlooked in existing party nationalization literature. More importantly, the literature on clientelism suggests positive effects deriving from extensive party structure, which is conventionally considered to strengthen national party organizations. However, my study argues that there are negative effects as well. Extensive party structure not only strengthens national parties, but it also offers a favorable environment for effective clientelist practices by enhancing monitoring and distributing capacities. In the latter case, a decrease of party nationalization is expected because clientelist practices primarily serve individual or factional purposes rather than party goals. Given the potential effects of clientelism, the end results of party nationalization may not be solely a product of formal institutions as identified in the extant literature. In spite of its potential implication for party nationalization, existing party nationalization literature has rarely looked into the role of informal mechanisms of clientelism. To fill these gaps in the existing literature, the next 55 section provides theoretical explanations that account for variations in party nationalization in light of the mediating role of clientelism. 3.3 Theory: the mediating role of clientelism The primary interest of my study is to examine the extent to which effectiveness of clientelism mediates the impact of party structures on party nationalization. There are a lack of theoretical studies explicitly focusing on the relationship among the three variables within a common framework. Thus, I build upon two related bodies of literature to guide my expectations: specifically, they include previous works on (a) party structure and effectiveness of clientelism and (b) clientelism and party cohesion. 3.3.1 Party Structure and Effectiveness of Clientelism To account for the mediating effect of clientelism, I first investigate the relationship between party extensiveness and clientelism. Clientelist practices pose challenges to parties and political elites in distributing resources and monitoring clients’ behaviors. As Okun points out, the loss of efficiency is inherent in the distribution process, which he called a ‘leaky bucket problem’ (Okun 1975: 91). He argues that some of the resources will inevitably disappear in transit, so the recipients of interest will not receive all the resources as patrons initially intended. In addition to addressing inefficiency in distribution, parties also need to ensure that recipients of targeted benefits honor their promises to cast their vote for the party. However, due to the secret ballot and the informal rules dictating clientelist exchange, enforcement of clientelist terms is difficult (Helmke and Levitsky 2004). 56 In the face of these challenges, literature on clientelism argues, parties’ organizational structures has important implications for the effectiveness of clientelism. Stokes (2005) argues that successful clientelist parties have extensive organizations deeply involved in voters’ social networks. Extensive organizations translate to broad formal membership and networks of local branches, informal associations with external networks, and affiliations with civil society organizations (Kitschelt and Kselman 2010). Examples of this type of extensive party organizations include Peru’s Peruvian Aprista Party, Mexico’s Institutional Revolutionary Party, and Argentina’s Peronists Party. Based on extensive organizational networks, these parties maintain intimate relationships between elites and citizens. With these relationships, they are in a better position to extract valuable information about voters’ preferences and behaviors (Kitschelt and Kselman 2010; Stokes 2005). In turn, these parties and elites can exercise firmer control over clientelist transactions by denying noncompliant clients further access to clientelist benefits. By reducing inefficiencies inherent in clientelist transactions, extensive organizations are conducive to effective distribution of targeted benefits and controlling opportunism. A party’s marginal return from its clientelist effort increases with greater organizational extensiveness. In other words, increases in organizational extensiveness are more likely to turn parties’ clientelistic efforts into tangible electoral gains (Kitschelt and Kselman 2010). As a result, parties and elites have greater incentives to divert their efforts and resources to the distribution of targeted benefits as opposed to programmatic policy campaign (Kitschelt and Kselman 2010). This leads to the following proposition: H2: As party organization becomes more extensive, it is more likely to be effectively clientelist. 57 3.3.2 Clientelism and Party Cohesion In addition, I explore the relationship between clientelism and party nationalization. While earlier studies considered parties to be homogeneous entities, more recent works have recognized that there are varying levels of diversity in party members’ preferences (Boucek 2009). A coherent party platform and electoral strategy are not always easily acquired. Rather, in order to achieve coordination across districts and form a nationalized party, a party needs to be able to mediate intra-party conflict and to encourage various intra-party cliques to compromise on policy priorities (Hicken 2009). I argue that intra-party coordination gets more difficult and costly with increasing clientelism. While clientelism attracts political elites and voters to the provider party, it also offers resources and opportunities for elites to cultivate their own support groups within and beyond a party organization, which are largely temporary and informal. It can be seen as ironic that increasing resources will undermine parties’ governability. As Panebianco (1988) insightfully points out, however, the nature of power is reciprocal, if asymmetrical, which opens up the possibility of the balance of power being shaken up: One can exercise power over others only by satisfying their needs and expectations. One thereby submits oneself to their power. In organizations, every organizational actor controls at least a small zone of uncertainty, i.e. possesses resources which can be capitalized on in power games (Panebianco 1988: 22). By nature, clientelism strengthens distinctive relationships between patrons and clients, where each relationship serves particularistic interests. Thus, clientelism is likely to encourage independent chains of command to develop within and apart from the national party organization (Bridges 1984). Specifically, clientelism promotes proliferation of factions and blurs party lines by developing alliances with opponents. By widening the gap between the de facto power structure 58 and the formal structure of a party, clientelism creates an unfavorable environment for a party to sustain as a cohesive organization. Where clientelism is prevalent and effective, I insist that centrifugal forces are at work which draw supporters and participants away from the party in which they were initially involved. In terms of intra-party relationships, clientelism promotes intra-party rivalry, thereby making horizontal coordination costly. Within a party, politicians often strive to cultivate allies and support groups consisting of influential political actors including fellow politicians, party members, and activists (Kemahlioglu 2011). It is important for them to obtain sufficient support to secure nominations or control policy making processes. Patron-client networks provide the support necessary to achieve these goals.33 Importantly, the membership of patron-client networks is not confined to the boundary of a national party organization. The zero-sum nature of the clientelism increases the level of intra-party competition to the extent that factional leaders enter into alliances with the party’s external opponents to undermine intra-party rivals (Weiner 1967). In addition, a patron-client relationship starting within a party is often extended vertically down to local followings. Beyond their immediate followings, politicians build vertical alliances with local patrons (Scott 1972).34 Upon receipt of clientelist benefits from higher levels of the party’s hierarchy, in turn, local patrons promise to draw support from their own local followings (Sandbrook 1972). Because these relationships are established based on the reputations of the individual politicians, electoral competition is likely to revolve around the individual elites rather than around the party. To the extent that the vertical alliance constitutes an independent unit, it can 33 Within the network, things such as campaign funds, favorable list placement, internal party posts, and jobs can be exchanged (Benton 2001). 34 The patron serves his clients as "a short cut through the maze of authority" in their dealings with regional or national officials (Kenny 1960: 19). The local patrons, i.e. mediators, are of significant interest because of his role in linking the local community to the nation (Wolf 1965). 59 enjoy the flexibility to provide or withdraw support for the national party. For example, Warner (1997) shows that among politicians in Brazil who maintain their own political machines and alliances there is a broad tendency to break away from their parties. The existence of multiple power centers within and beyond a party organization introduces conflicting party goals, campaign strategies, and member preferences over policy platforms because each group can serve as a veto player. Conflicting goals and preferences within a party increase the cost of cross-district coordination (Hicken and Stoll 2011). In sum, clientelism facilitates emergence of intra-party groupings, which restrain parties’ abilities to develop coherent party strategies or to coordinate across districts in electoral competitions. Consequently, candidates and voters are less likely to converge on a few coherent national-level party brands. Taken together, I hypothesize the following: H3: Effective clientelism is negatively associated with the level of party nationalization due to increasing cost for the party to coordinate across districts. To summarize, the theoretical discussions from the three bodies of literature yield the following predictions: Party extensiveness increases party nationalization. In addition, party extensiveness enhances effectiveness of clientelism, which in turn undermines party nationalization. Thus, the positive impact of party extensiveness on party nationalization will be suppressed to the extent that party extensiveness improves effectiveness of clientelism. 60 3.4 Data and Measures 3.4.1 Dependent Variable: Party Nationalization The unit of analysis in my study is a political party. To measure the party nationalization, I draw on the Party Nationalization Score (PNS) proposed by Jones and Mainwaring (2003), which builds on the Gini coefficient of inequality in the vote shares across districts. Gini coefficient has been widely employed to calculate heterogeneity in the distribution of income (Deininger and Squire 1996) and electoral outcomes (Taagepera and Shugart 1989). According to the Bochsler’s review on the existing measures of party nationalization, the Gini coefficient-based measure is one of the most ideal alternatives among the existing measures because it has the least methodological problems (Bochsler 2010: 2-6). Figure 3.1 illustrates the calculation of the Gini coefficient. The x-axis represents the electoral districts, placed in the order of their vote shares for the party Pi. The y-axis displays the cumulative function of votes across regions. In the first district, a party Pi obtains the lowest vote share. The vote shares from the first district are added up to the ones from the second district. This way, a district i carries the information about the sum of all previous vote shares up to the district i. Thus, the last district presents the total number of votes a party Pi received across the whole district (Jones and Mainwaring 2003; Bochsler 2010). Finally, the Gini coefficient is calculated by the area between the graph and the 45-degree perfect homogeneity line, which ranges from 0 to 1. With heterogeneous vote distribution (i.e. low party nationalization), this area becomes close to 1. To have the higher value represent higher party nationalization, Jones and Mainwaring (2003) subtract this value from 1. 61 Figure 3.1 Calculation of Party Nationalization Score Note: (Bochsler 2010: 161) In my analysis, I use the revised version of Party Nationalization Score (PNS) with two modifications suggested by Bochsler (2010). First, the revised measure accounts for the variations in the number of electoral districts across countries. Statistically, a country with smaller number of districts presents less variation in party nationalization because extreme values disappear by aggregation. To allow comparison across countries with different number of districts, the revised measure standardized the Gini-coefficients. Second, the measure also adjusts for the unequal sizes of districts within a country (Bochsler 2010: 161-164). In order to correct for the fact that electoral behaviors of smaller districts are overly represented in the nationalization outcome, the revised measure weight the importance of districts by their population size. Taken together, the formula of the revised Party Nationalization Score (PNS) is as follows: 𝑃𝑁𝑆𝑠𝑤 = (𝑃𝑁𝑆𝑤 ) 1 log(𝐸) 𝑃𝑖 𝑖 ∑𝑑 1 (𝑣𝑖 ∗(∑1 𝑃𝑗 − )) , where 𝑃𝑁𝑆𝑤 = 2 ∗ 2 𝑑 ∑𝑑 1 𝑣𝑖 ∗∑1 𝑃𝑖 and the variable E is a constant calculated at the national level as follows: 𝐸= (∑𝑑 1 𝑣𝑜𝑡𝑐𝑠𝑡 ) ∑𝑑 1 𝑣𝑜𝑡𝑐𝑠𝑡 2 , where 𝑣𝑜𝑡𝑐𝑠𝑡 is the number of votes cast in constituency i. 62 I use the Constituency-Level Elections Archive (CLEA) datasets which contain the measure of party nationalization. At the party level, the measure indicates the level of nationalization for a particular party in a given country-year dyad. The measure is calculated based on the electoral outcome at the lower house legislative elections. To minimize the possibility of reverse causality, I construct my cross-sectional data on the dependent variable that is measured temporally after the independent variable and the mediating variable, which is the year 2008. 3.4.2 Independent Variables: Extensiveness of a Party Organization To obtain the data on party structure, I use the Democratic Accountability and Linkages Project (DALP), which asks countries’ experts to evaluate major parties in a given country (Kitschelt 2013). The data is collected in 2008 and 2009 on parties’ organizational structures and policies.35 I use the dataset aggregated at the party level. Building on Kitschelt and Kselman (2010), party extensiveness is measured by a battery of items which are designed to represent a diverse “transmission belts through which political parties may reach voters” (3). Specifically, to tap into the extensiveness of a party organization, I create a composite index based on three variables on party structure: (i) the formal extensiveness of a party organization; (ii) the informal extensiveness of a party organization; and (iii) the extensiveness of linkages with civil society organization. First, the formal extensiveness is captured by the experts’ evaluation on the portion of districts where a party has a physical presence such as local branch office. The answer consists of four-level scale ranging from 1 (no local offices) to 4 (permanent local offices in most districts).36 In addition, the informal extensiveness measures the parties’ capacity to reach voters through 35 See Appendix for the list of countries and the number of parties included in the analysis. 36 I recoded these variables so that higher scores indicate greater extensiveness. 63 informal ties such as local intermediaries. The variable ranges from 1 (no local representatives) to 3 (local representatives in most constituencies). Finally, the extensiveness of linkages with civil society organization is assessed by using the following survey question: “Do the following parties have strong linkages to one or more of the following civil society organizations?” I create an index on the number of organizations that a party maintains explicit linkages. If there is no linkage to any groups, 0 is assigned. If there is any linkage, I count the number of organizations that a party maintains relationship with from the following lists: unions; business and professional associations, religious organizations; ethno-linguistic organizations; urban or rural associations; and women’s organizations. Using these three variables, I create the index by summing up the three variables. Larger value represents greater extensiveness of a party organization. I weight the measure of party extensiveness by party size. Existing studies point out that small parties are statistically more nationalized than larger ones because of little variance in their vote shares. For the measure of party size, I use an average of a party’s vote shares in the most recent two national legislative elections before 2009 available from the DALP dataset. 3.4.3 Mediating Variable: Effectiveness of Clientelism To measure effectiveness of clientelism, I draw on the survey questions from DALP, where the experts are asked to estimate how successful parties are in translating clientelistic effort into actual vote gains. Level of effectiveness is assessed based on four grading schemes from “not at all” to “to a great extent.” 64 3.4.4 Control variables My analysis also includes other covariates that the existing literature found relevant to party nationalization. First, existing studies suggest that federalism can undermine party nationalization by reducing party cohesion (Chhibber and Kollman 2004; Desposato 2004; Brancati 2008). With independent decision making authority at the subnational level, politicians have incentive to serve local preferences deviating from their party. The measure of federalism is coded as 1 if a country has a federal political structure and 0 otherwise. Second, the literature maintains that single member district (SMD) system tends to generate lower party nationalization (Hicken 2009). Due to the majority rule to win a seat, parties only invest in the districts where they have a chance to win. This results in unequal vote distributions across districts. Using the classification scheme by Norris (2009), majoritarian systems are coded as 0, mixed systems as 0.5 and proportional systems as 1. Third, parliamentary systems are expected to present higher level of party nationalization than its presidential counterparts (Schattschneider 1960; Morgenstern and Swindle 2005). In a parliamentary system, it is less costly to achieve intra-party coordination due to interdependent relationship between executive and legislative candidates. Based on the Database of Political Institutions, presidential system is coded as 0, strong president elected by assembly as 0.5, and parliamentary system as 1. Finally, I control for ethnic fractionalization with the expectation that it undermines party nationalization because diverse ethnic settings pose greater challenges to aggregate the interests (Linz and Stepan 1996; Cox 1997; Brancati 2003; Caramani 2004). The measure of ethnic fractionalization ranges from 0 to 1, where higher values indicate greater heterogeneity (Alesina et al. 2003). 65 3.5 Empirical Analysis 3.5.1 Analytic Strategy To test my hypotheses, I conduct a mediation analysis. This analysis seeks to explain the mechanism through which an independent variable influences an outcome variable by means of a mediating variable (Mackinnon, Fairchild, and Fritz 2007). In examining a mediational proposition, the relationship among variables is decomposed into two causal paths. One of these paths carries the direct effect of the party extensiveness (i.e. independent variable) on party nationalization (i.e. dependent variable) while controlling for effectiveness of clientelism (i.e. the mediator). The other path links these variables through a mediator, called the mediating (or indirect) effect. The total effect is the sum of the direct and mediating effects: Total effect (c) = Direct effect (c’) + Indirect effect (ab) The path coefficients (a, b, c′) estimate the strength of hypothesized associations. When a path includes other variables, the strength of the relationship for this path is estimated by multiplying the coefficients for each leg of the path. Thus, the strength of the mediated relationship is estimated by the product of a × b coefficients. Mediation occurs if the effect of an explanatory variable on the outcome variable is partly or entirely transmitted by the mediating variable (Holland 1988; Sobel 1990).37 The most common way to obtain coefficient estimates for these paths is to run the following series of regression analyses, i.e. the causal steps approach. 38 First, regressing the dependent variable on the independent variable; second, regressing the mediator on the independent variable; 37 A mediation effect differs from an interaction effect in that no causal sequence is implied by an interaction effect. 38 Another way to estimate the path coefficients is the structural equation model. 66 and third, regressing the dependent variable on the mediator (Hayes 2013; Zhao et al. 2010).39 My mediation model for the ith party is given by Effectiveness of Clientelismi = β0 + βxz*Party Extensivenessi + εzi, Party Nationalization ln (1−Party Nationalizationi )= γ0 + γzy*Effectiveness of Clientelismi + γxy*Party Extensivenessi + εyi i Note that my measure of party nationalization draws on a Gini coefficient which is bounded between 0 and 1.40 As Figure 3.2 demonstrates, this violates the assumptions of normality and linearity, which makes linear regression inappropriate. In addition, the assumption of homoscedastic errors is violated because the variance tends to decrease when the mean gets closer to 0 or 1.41 To handle this data, I take a logit transformation of my dependent variable as I conduct the regression analyses. 39 These steps are similar to causal step approach to establish a mediation effect, recommended by Baron and Kenny (1986). Although widely used, it is not the most recommended procedure because mediation may occur even when the conditions are not met such as in competitive mediation (MacKinnon et al. 2007; MacKinnon, 2008; Zhao et al. 2010). Given its limitations, I use it only to estimate the path coefficients. To evaluate the statistical significance of the mediating effects, I employ two additional methods. 40 Party nationalization literature indicates there is an increasing heterogeneity of the vote with increasing number of districts, but with a decreasing marginal effect of the number of districts (Morgenstern et al. 2014). Alternative solutions to the violation of heteroskedasticity, normality, and linearity include modeling the distribution of the dependent variable with a beta distribution or fitting a generalized linear model with binomial distributions and the logit link function. 41 Strictly, logit p cannot be determined for the extreme values of 0 and 1. Since my data does not have any 0 and 1, this practice will not cause any problem. 67 0 5 10 Percent 15 20 25 Figure 3.2. Distribution of Party Nationalization Score 0 .2 .4 .6 .8 1 Party Nationalization In addition to the causal steps mediation analysis, I corroborate the estimation results using the structural equation models. Then, to evaluate the significance of the mediating effects, I conduct the Sobel test and the non-parametric bootstrap analysis. 3.5.2 Results Table 3.1 displays results for the causal steps mediation analysis. Model 1, 2, and 3 are estimated without control variables, while Model 4, 5, and 6 control for the variables. In what follows, I focus on the interpretation of the latter three models, but both cases present consistent results for the relationship among party extensiveness (X), effectiveness of clientelism (M), and party nationalization (Y). Figure 3.3 shows the corresponding path diagram with the path coefficients estimated from this mediation analysis. Standardized coefficient estimates are reported in the brackets. 68 Table 3.1. Mediation Analysis DV: Party Nationalization (Y) Model 1 XY Party Extensiveness (X) 0.677*** (0.405) Effectiveness of Clientelism (M) Model 2 X M 0.204*** (0.035) Model 3 X+M Y 0.868*** (0.069) -0.934*** Model 4 X Y 0.668*** (0.068) Model 5 X M 0.204*** (0.035) (0.115) Ethnic fractionalization Federalism Parliamentary System Majoritarian System _cons 0.405*** (0.102) N 256 F F(1, 254)=85.81 Prob > F p<0.0001 R2 (adjusted) 0.25 2.435*** (0.049) 256 F(1, 254)=33.27 p<0.0001 0.11 2.678*** (0.295) 256 F(2, 253)=86.18 p<0.0001 0.41 Fit Statistics for the chi2_bs(3)=164.52, p > chi2=0.000, RMSEA=0.000, Structural Equation Model CFI=1.000, TLI=1.000, SRMR=0.000, CD=0.453 -1.107** (0.333) -0.286 (0.168) 0.104 (0.163) -0.652*** (0.168) 1.004*** (0.213) 256 F(5, 250)=30.42 p<0.0001 0.37 2.435*** (0.049) 256 F(1, 254)=33.27 p<0.0001 0.11 Model 6 X+M Y 0.831*** (0.067) -0.768*** (0.115) -0.683* (0.313) -0.226 (0.156) 0.079 (0.151) -0.570*** (0.156) 2.700*** (0.321) 256 F(6, 249)=37.25 p<0.0001 0.46 chi2_bs(3)=217.89, p > chi2=0.000, RMSEA=0.134, CFI=0.911, TLI=0.756, SRMR=0.060, CD=0.503 Note: 1. X: Party Extensiveness, M: Effectiveness of Clientelism, Y: Party Nationalization; 2. Coefficients are ordinary least squares regression estimates. Standard errors are in parentheses;* p<0.05, ** p<0.01, *** p<0.001; 3. Fit Statistics: RMSEA (Root Mean Squared Error of Approximation); CFI (Comparative Fit Index); TLI (Tucker-Lewis index); SRMR (Standardized Root Mean Squared Residual); CD (Coefficient of Determination). Generally, CFI ≥0.9, TLI ≥0.9, SRMR ≤ .08, and RMSEA ≤0.06, and larger CD represent a good fitting model. 69 Figure 3.3. Pathway of a Mediation Process for a Relationship between Party Structure and Party Nationalization Effectiveness of Clientelism (M) a= 0.204***/[0.340***] (0.035) b= - 0.768***/[-0.342***] (0.115) Mediating Effects: a x b = -0.157 Party Extensiveness (X) Party Nationalization (Y) c’=0.831***/[0.616***] (0.067) Note: Standardized coefficients are reported in brackets and standard errors are in parentheses. Proportion of total effect that is mediated by Effectiveness of Clientelism: 𝐼𝑛𝑑𝑖𝑟𝑒𝑐𝑡 𝐸𝑓𝑓𝑒𝑐𝑡 −0.157 = ≈ −0.243 𝑇𝑜𝑡𝑎𝑙 𝐸𝑓𝑓𝑒𝑐𝑡 0.831 + (−0.157) As shown in Model 4 in Table 3.1, first regression predicts the level of party nationalization from party structure, while controlling for the mediating variable. The coefficient from this regression corresponds to path c in Figure 3.3: the total effect. The estimation results show that the total effect of party extensiveness on party nationalization is c = 0.668. This effect is statistically different from zero, t(df=250)=9.79, p=.000, with a 95% confidence interval from 0.534 to 0.803. The party with greater organizational extensiveness tends to achieve a higher level of party nationalization. Since I perform a logit transformation of my dependent variable (i.e. party nationalization), the results can be interpreted as the percentage change in the ratio of nationalization: [exp(0.668)-1]*100≈95. One unit increase in party extensiveness is associated with a 95% increase in the proportion of votes obtained by this party across districts, i.e. increase in the level of party nationalization. 70 Next, a second regression predicts the mediating variable, effectiveness of clientelism, from the independent variable, party extensiveness. Model 5 in Table 3.1 demonstrates that the coefficient of path a is 0.204, with t(df=254) = 5.77, p = .000. For a party with greater structural extensiveness by one unit, I predict a 0.204 unit increase in effectiveness of their clientelist efforts turning into electoral gains. This implies that more extensive parties tend to enjoy greater effectiveness in their clientelistic practices than their less extensive counterparts. Finally, as Model 6 in Table 3.1 shows, a regression is performed to predict party nationalization from both party extensiveness and effectiveness of clientelism. This regression provides coefficient estimates for path b and path c’ in Figure 3.3. The coefficient c′ estimates the strength of the direct effect of party extensiveness on party nationalization that is not mediated by clientelism. The estimated direct effect of party extensiveness on party nationalization, controlling for clientelism, is c′ = 0.831, t(df=249) = 12.31, p = .000. I expect about a 129.6% increase in party nationalization for each additional unit of party extensiveness.42 The coefficient on the effect of clientelism on party nationalization is b = -0.768. This effect is statistically different from zero t(df=249)= -6.69, p=.000, with a 95% confidence interval from -0.994 to -0.542. For each unit increase in effectiveness of clientelism, I predict a 53.6% decrease in party nationalization.43 Party nationalization is predicted reasonably well by party extensiveness and effectiveness of clientelism, with adjusted R2 = 0.46 and this is statistically significant, F (6,249) = 37.25, p < .0001. Based on the coefficient estimates of these paths, I find that the magnitude of mediating effect is 0.204 × (-0.768) = -0.157. The percentage change in the ratio of party nationalization can be obtained by [exp(-0.157)-1]*100 ≈ -14.5. Thus, for each one unit increase in a party’s 42 [exp(0.831)-1]*100≈129.6 43 [exp(-0.768)-1]*100≈53.6 71 extensiveness, the proportion of votes obtained by this party decreases by 14.5% across districts. This decrease in party nationalization is explained by the effects of party extensiveness on clientelism. This result demonstrates that more effective clientelism is likely to decrease party nationalization. Overall, the estimation results presented in Figure 3.3 and Table 3.1 indicate that the effectiveness of clientelism mediates 24.3% of the total effect of party extensiveness on party nationalization, while about 75.7% of the total effect is direct.44 To evaluate the significance of mediating effect of clientelism, I use the two most widelyused methods. First, I conduct the Sobel test (Sobel 1982). The null hypothesis of interest is H0: ab = 0, which evaluate if the mediating path is significantly different from zero.45 The Sobel test statistics in Table 3.2, z = -4.341, p = 0.000, suggest that the mediating effect of clientelism is significantly different from zero. Thus, the effectiveness of clientelism significantly mediates the effect that party structure has on party nationalization. Table 3.2. Sobel Test for Mediating Effects Effectiveness of Clientelism (M) Note: H0: a x b = 0; *** p<0.001 Mediating Effect -0.157*** Test statistics (z) -4.341 Std. Error 0.036 P-Value 0.000 Secondly, I conduct a non-parametric bootstrap sampling analysis. This process provides an empirical sampling distribution that can be used to derive a confidence interval, a p value, and 44 The proportion of total effect that is mediated by effectiveness of clientelism is computed by dividing the indirect effect by the total effect (MacKinnon 2008). 45 The Sobel test provides an approximate estimate of the standard error of ab as follows: 𝑆𝐸𝑎𝑏 = √𝑏 2 𝑠𝑎2 + 𝑎2 𝑠𝑏2 where a and b are the unstandardized regression coefficients that represent the effect of the independent variable on the mediating variable and the effect of the mediating variable on the dependent variable, respectively. The test of the indirect effect is given by z = ab/𝑆𝐸𝑎𝑏 72 a standard error for the estimates of the coefficients for indirect effect, axb (Imai et al. 2010). In general, bootstrap confidence intervals are preferred over the Sobel test because they do not rely on a priori assumptions about distributions (Preacher and Hayes 2008; Zhao et al. 2010). 46 As Table 3.3 presents, the bootstrap estimated mediating effect is -0.157 (p = 0.000) with a standard error of 0.032. The 95 % bootstrap confidence interval for the value of the indirect effect ab is from -0.231 to -0.102, and because zero is not in the confidence interval, it is concluded that the indirect effect is statistically different from zero. The results of the nonparametric bootstrap sampling mediation analysis are consistent with the findings of the causal step analysis and the Sobel test: the mediator significantly predicts party nationalization. Table 3.3. Bootstrap Results for Mediating Effects Mediating Effect Std. Error 95% CI Effectiveness of Clientelism (M) -0.157*** 0.032 -0.231 -0.102 Note: Non-parametric bootstrapping was performed with 5000 samples. I use the SPSS script for the PROCESS procedure (Preacher and Hayes 2008). A bias-corrected and accelerated confidence interval (CI) was created for the coefficient of mediating effect. Statistical significance of mediating effect is determined by whether a confidence interval include zero or not (Fritz and MacKinnon 2007). *** p<0.001 Taken together, these results provide consistent evidence to support for my hypothesis that clientelism mediate the effect of party extensiveness on party nationalization. In addition, the direct path from party extensiveness to party nationalization is also statistically significant. Therefore, the total effect of party extensiveness is partially mediated by clientelist practices. Importantly, although the direct effect of party extensiveness is to increase party nationalization, the mediational path has the opposite effect: party extensiveness increases effectiveness of clientelism, 46 The z tests used for the Sobel test assume that values of this ab product are normally distributed across samples from the same population. Studies have demonstrated that the sampling distribution of ab is highly skewed and the normality assumption is likely to be violated (Fritz and MacKinnon 2007). 73 which in turn decreases party nationalization. When the direct and mediated effects have opposite signs, the mediator acts as a suppressor variable which produces a small or non-significant total effect (MacKinnon et al. 2000). Such models are known as competitive (or inconsistent) mediation models (MacKinnon et al. 2007).47 To corroborate the results, I use the structural equation model (SEM) to estimate the mediating effect. The mediation analysis literatures argue that the SEM approach is statistically superior to the conventional causal step approach because it estimates the series of regressions simultaneously instead of assuming that each equation is independent (Zhao et al. 2010). In general, it is argued that fitting a single SEM model offers more efficient and consistent estimation than the three regression pieces (Iacobucci et al. 2008). The path coefficients are identical to the ones from the causal step analysis because the correlation matrix translates to path coefficients (Iacobucci et al. 2008). The last row in Table 3.1 shows that the SEM model presents reasonably good model fit as indicated by the selected overall goodness-of-fit statistics: SRMR = .06; χ2(df=249) = 217.89, p≤0.001; RMSEA=0.042; CFI=0.911; TLI=0.756.48 47 Drawing on the example of MacKinnon et al. (2000), competitive mediation effects can be found in the interrelationships among workers’ intelligence, level of boredom, and the number of errors made. All else being equal, the more intelligent workers would make fewer errors. However, they would also exhibit higher levels of boredom and boredom can increase the number of errors. Thus the direct effect of intelligence on errors would be negative, while the indirect effect mediated by boredom would be positive. Combined, these two hypothetical effects may cancel each other out, resulting in a small total effect of intelligence (MacKinnon et al. 2000). 48 Higher values of CFI (Comparative Fit Index) demonstrate greater improvement over the baseline model in fit. In contrast, lower values of SRMR (Standardized Root Mean Square Residual) and RMSEA (Root Mean Squared Error of Approximation) indicate better model-data fit. SRMR and RMSEA measure the extent to which the model reproduces the sample covariance matrix. Similar to an R2, CD (Coefficient of Determination) shows how well the data fits a model. Rule of thumb guidelines are that CFI ≥0.9, TLI ≥0.9, SRMR ≤ .08, RMSEA ≤0.06, and larger CD represents a well-fitting model. 74 3.6 Conclusion This study asks why political parties have varying patterns of vote distributions across districts. To answer this question, I examine the underexplored informal mechanism of clientelist practices. I propose that clientelism act as a mediator in the relationship between party structure and party nationalization. An extensive party structure enhances the chance that parties’ clientelist efforts turn into electoral gains. In turn, effective clientelism prohibits the emergence of a nationalized party by increasing incentive for politicians to cultivate their own support base within and beyond party organizations. Given that a nationalized party is mainly a product of efficient coordination among candidates across districts, proliferation of factions increases the cost of coordination. Using the party-level datasets, I conduct the mediation analysis. In line with my hypotheses, the empirical results suggest that party extensiveness increases effectiveness of clientelism, which in turn undermines party nationalization. In addition, the direct effect of party extensiveness on party nationalization is positive and statistically significant. Overall, clientelism exerts a suppression effect on the relationship between party extensiveness and party nationalization, which would otherwise present stronger positive associations. In this study, a focus on the party level allows me to find the dynamics of intra-party groupings. Specifically, I demonstrate that clientelist practices provide incentives for the copartisan to employ divergent electoral strategies deviating from their national party platforms. Thus, my findings imply that with respect to intra-party relationships, clientelism can be a source of competition and conflict rather than of solidarity. My study opens the way for additional research. By nature, mediation analysis takes a confirmatory approach to the investigation of the causal relations among key variables. Thus, the 75 results can verify whether a hypothesized theoretical model is consistent with the data. To determine whether any alternative causal relations can be ruled out, evidence from micro-level case studies or experimental works can be supplemented. In addition, this study focuses on crosssectional analysis due to the limited availability of the data on informal mechanisms of clientelism. However, increasing body of party nationalization literature point out the importance of understanding how a party’s vote distribution pattern changes over time. Likewise, the clientelism literature suggest that over an increasing time span clientelist transactions may provide different incentives to the relevant actors. If it is a repeated practice, it is important for participants to build reputations and credibility. Consequently, informal rules can be institutionalized, while their centrifugal effects on relevant actors might be attenuated. Further study is called for to analyze the incentive structures of actors engaging in coordination efforts to form nationalized parties and party system over a longer time span. 76 APPENDIX 77 Table 3.4. Descriptive Statistics Variable Party Nationalization Effectiveness of Clientelism Formal extensiveness (local offices and paid staff) Organizational Informal extensiveness (local intermediaries) Extensiveness Centralization of electoral of a Party strategy Linkage with civil society Ethnic Fractionalization Federalism Majoritarian System Party Size Parliamentary System Obs 256 256 Mean 0.71 2.64 Std. Dev. 0.237 0.593 Min 0.005 1 Max 0.975 3.857 256 3.12 0.654 1 4 256 2.21 0.458 1 3 256 2.37 0.353 1 2.8 256 256 256 256 256 256 0.33 0.34 0.22 0.33 0.16 0.22 0.170 0.234 0.419 0.837 0.143 0.414 0.023 0.002 0 0 0 0 0.795 0.850 1 1 0.678 1 Table 3.5. Case Selection by Region Region Eastern Europe and post Soviet Union Latin America North Africa & the Middle East Sub-Saharan Africa Western Europe and North America East Asia South-East Asia South Asia The Caribbean Total Frequency 51 45 5 20 89 6 12 27 1 256 Table 3.6. Case Selection by Country Country Albania Argentina Austria Bangladesh # of parties 4 5 5 4 # of parties 1 4 2 8 Country Jamaica Japan Korea Latvia 78 Table 3.6 (cont’d) Bolivia Botswana Brazil Canada Colombia Costa Rica Croatia Czech Republic Denmark Dominican Republic Estonia Finland France Germany Ghana Greece Honduras Hungary India Indonesia Ireland 4 3 11 5 7 4 10 5 8 3 3 8 6 6 3 4 2 4 17 6 5 Mauritius Mexico Netherlands New Zealand Nigeria Norway Pakistan Peru Philippines Poland Portugal Romania Russian Federation South Africa Spain Sweden Turkey UK US Zambia Total 3 5 8 7 3 7 6 4 6 6 5 5 6 3 3 7 5 3 2 5 256 Table 3.7. DALP Questionnaire Mediating Variable Effectiveness of Clientelism Please assess how effective political parties are in their efforts to mobilize voters by targeted benefits. [1] Not at all, [2] To a small extent, [3] To a moderate extent, [4] To a great extent Formal Extensiveness Do the following parties or their individual candidates maintain offices and paid staff at the local or municipal-level? If yes, are these offices and staff permanent or only during national elections? Independent [1] No, the party does not maintain local offices Variable [2] Yes, the party maintains local offices, but only during national elections [3] Yes, the party maintains permanent local offices in SOME districts [4] Yes, the party maintains permanent local offices in MOST districts Informal Extensiveness 79 Table 3.7 (cont’d) Do the following parties have local intermediaries (e.g. neighborhood leaders, local notables, religious leaders) who operate in local constituencies on the parties’ behalf, and perform a variety of important tasks such as maintaining contact with large groups of voters, organizing electoral support and voter turnout, and distributing party resources to voters and supporters? [1] No, they have almost no local representatives [2] Yes, they have local representatives in SOME constituencies [3] Yes, they have local representatives in MOST constituencies Linkage with Civil Organization Political parties often have more or less routine and explicit linkages to civil society organizations such as unions, business or professional organizations, and cultural organizations based on religion, language, or ethnicity. The linkages might include leadership and membership overlap, mutual financial support, reserved positions for representatives of these organizations at National Conventions, etc. Do the following parties have strong linkages to one or more of the following civil society organizations? o No linkages to all the groups o Unions o Business associations and professional associations o Religious Organizations o Ethnic/ linguistic organizations o Urban neighborhood or rural associations/movements o Women’s organizations 80 CHAPTER 4 CLIENTELISM AS AN INFORMAL DETERMINANT OF CONGRUENCE 4.1 Introduction Representative democracy requires heterogeneous public preferences to be embodied by a small group of representatives. A large body of literature has studied whether and how diverse individual preferences can be aggregated into a coherent choice. Mandate theory sees representatives as delegates who act on behalf of their constituency rather than as those who act on their own independent judgment (McDonald, Mendes, and Budge 2004). Following the mandate theory, democratic representation means that the actions of representatives are expected to be responsive to what the people wish. Accordingly, the quality of representation is assessed by the degree of a match between what citizens want and what the representatives try to achieve (Powell 2013). Starting with the seminal study of Miller and Stokes (1963), a growing body of literature has investigated the mechanisms that connect citizens and their elected representatives based on the notion of congruence. Congruence is an election-based linkage between citizens and their representatives (Powell 2004). Through voters’ choice of parties and candidates in elections, voters’ preferences are linked to preferences and behavior of representatives. Literature emphasizes that a simple correspondence of interests and policy is not enough in that a benevolent dictatorship is not a representative democracy even though it serves citizens’ interests just as well as a democracy (Pitkin 1967: 232–234). What makes democracy different from autocracy is an institutional mechanism that ensures regular opportunity for citizens to express their interests (McDonald, Mendes, and Budge 2004: 3). For this reason, studies on representation predominantly focus on the effect of electoral institution that will reliably connect citizens to representatives. 81 The fact that literature mainly examined electoral rules as a determinant of congruence indicates that it sees citizens’ votes as an accurate guide to underlying preferences. However, few studies have asked what happens to congruence if citizens’ preferences are not expressed at the ballot box. If voters do not vote according to their sincere preferences, congruence can still be undermined in spite of the functioning electoral rules. As a serious impediment to congruence, this paper examines the underexplored effect of clientelism on congruence. Clientelism is an informal mechanism through which parties obtain political support in exchange for selective benefits to voters or groups (Hopkin 2006).49 In this paper, I conceptualize clientelism in terms of a tool of persuasion employed by political elites and parties. By means of clientelistic rewards, elites try to alter the preferences and behavior of voters who would otherwise hold different or no views about parties’ policies. In response to the rewards, I expect that voters vote further from their ideal points, although the degree of responsiveness varies by their strength of partisanship, information, or needs. In spite of its ability to shape the party-voter relationship, clientelism has rarely taken center stage in the study of opinion congruence, which predominantly focuses on advanced democracies. Conventional wisdom says that programmatic commitment will replace clientelistic practices with political and economic development. Although previous studies have broadly 49 Hopkin (2001) argues that there is significant differences between the old and the new clientelism. The new clientelism is characterized by more equal and less personalized relationship between patron and clients. Piattoni (2001) also maintains that "democracy strengthens the clients' bargaining leverage vis-avis brokers and patrons". Literature increasingly observes that modern clientelism takes the form of a more "complex pyramidal exchange network of exchange" (Kitschelt and Wilkinson 2007:8). The examples of rewards offered in clientelistic exchange include tangible goods such as food, clothing, medicine, mattresses, construction materials, and utility bill payments (Schaffer 2007: 1–2) and services such as transporting voters to polling places, haircuts, teeth cleaning and vasectomies (Nichter 2014). In a longer term, clientelistic transaction also involves public sector jobs and promotions or preferential, discretionary access to subsidized goods such as land, public housing, education, utilities, or social insurance benefits, and specific procurement contracts to private enterprises (Kitschelt and Wilkinson 2007). 82 assumed that clientelism has a negative impact on policy representation (Stokes 2006), few studies have offered empirical evidence on this assumption in the context of opinion congruence. To fill this gap, this paper examines whether and how clientelism is systematically associated with the level of congruence between political parties and their voters. Specifically, I argue that countries with higher levels of clientelism show higher levels of incongruence in policy preferences between the public and the governing parties. 4.2 Defining and Measuring Congruence The influential work by Huber and Powell (1994) initiated the study on congruence based on the comparison of the position of the median voters with that of the governing parties. Since then, the median position in public opinion serves as a common yardstick of representation. Congruence is considered to be enhanced when the parties’ position move closer to the median position. Scholars have considered that representation of the median citizen produces the most favorable outcome of representation. Assuming that citizens always favor the position closer to them, the median position is the optimal location which minimizes the average distance between voters’ ideal positions and the government’s policy (Cox 1997), thereby decreasing dissatisfaction with collective choice (Dalton et al. 2011). As Figure 4.1 illustrates, empirical studies have predominantly conceptualized and measured congruence in terms of the absolute ideological distance between the median citizen and the governing parties or government. Figure 4.1. Conventional Measure of Congruence the absolute ideological distance between the median citizen and the governing parties Note: Golder and Stramski (2010) 83 Recent studies point out the limitations of this approach to congruence in that it invalidly assumes constituencies to be homogenous and it ignores the ideological configuration of the collective body of representatives. The theorists of democratic representation emphasizes that what makes it representation is not a single action by one participant, but the overall structure and functioning of the system, which is shaped by the compound behaviors of people involved (Pitkin 1972: 221). As demonstrated in Figure 4.2, this vision of representation has been conceptualized in an alternative measure, labeled ‘many-to-many congruence’, which is proposed by Golder and Stramski (2010). Instead of the absolute distance between the median citizen and the party, they take into consideration the distribution of both preferences. The level of congruence is assessed by the degree of similarity in both the shape and location of citizen and representative preferences. Congruence is maximized when the two distributions are identical (Golder and Stramski 2010). Given that the many-to-many measure of congruence considers both shape and location of preferences, it is a better way to understand the pattern of congruence in a country where voters are heterogeneous and are not strongly linked to a party.50 Figure 4.2. Many-to-Many Measure of Congruence the overlap of the distribution of citizens and governing parties Note: Golder and Stramski (2010) 50 For the studies employing many to many measure of congruence, see Golder and Stramski 2010; Eerd 2011; Espana Najera and Martınez-Roson 2012; Bengtsson and Wass 2012; Boas and Smith 2014; Ruth 2011; Andeweg 2011; Corral 2013 84 To explain the concept of many-to-many congruence, I introduce the Figure from Golder and Stramski (2010). As Figure 4.3 illustrates, it offers three hypothetical countries, A, B, and C. The top row presents the probability distributions on a left-right issue dimension for the citizens and representatives in each country, whereas the bottom row demonstrates their cumulative distributions. The measure of many-to-many congruence is captured by the shaded area between the cumulative distributions. The larger shaded area indicates greater dissimilarity between the preferences, thus lower congruence. In countries A and B, the shapes of probability distributions for the citizens and representatives are identical, while the position of the representatives in country B is located further away from that of the citizens, in comparison to country A. The larger shaded area in country B indicates lower level of congruence than in country A. In country C, the shapes of the two probability distributions differ, whereas the location of the distribution is similar. Figure 4.3. Three Hypothetical Scenario of Many-to-Many Congruence Note: Figure from Golder and Stramski (2010) 85 4.3 Literature on Electoral Systems and Congruence Powell (2004) sees democratic responsiveness as a series of intertwined linkages. The first linkage connects the preferences of citizens to their behavior in elections. The second linkage indicates the process in which citizens’ voting behavior influences election outcomes and government formation. Finally, the third linkage concerns the relationship between the policy makers and the policy outcomes that get implemented.51 The quality of democracy is determined by institutional arrangements that support each linkage of responsiveness. As the mandate theory claims, elections are the key mechanism through which popular preferences are translated into policy (McDonald, Mendes, and Budge 2004). For this reason, congruence literature has closely examined the effect of electoral rules on the level of congruence.52 However, empirical studies have produced mixed findings so far. Some suggests that proportional representation electoral rules (hereafter PR) promotes greater congruence (Huber and Powell 1994; Wessels 1999; McDonald and Budge 2005; Powell 2000). Studies also maintain that high district magnitude and a large number of political parties also contribute to better congruence (Thomassen 1999). At elections, they argue that PR offers a wider range of party choices so that voters have more options closer to their preferences (Powell 2000). At the government formation stage, PR further promotes congruence by pulling the government toward the center of the policy spectrum. That is because PR systems generally require a coalition government, and the median parties play a key role in this process (Huber and Powell 1994). In 51 In this article, I focus on the congruence achieved at the election because clientelistic practice is more relevant to mobilization of citizens during election. In contrast, government formation entails bargaining and coordination among the political elites. 52 Some of the representative studies include Huber and Powell (1994), Powell (2000, 2006, 2009, 2013), Powell and Vanberg (2000), Kim and Fording (2006), Kim et al. (2010), McDonald, Mendes, and Budge (2004), McDonald and Budge (2005), Blais and Bodet (2006), Golder and Stramski (2007). 86 contrast, other scholars found that PR tend to produce less centrist parties, thereby decreasing the level of congruence. When more parties are competing, each of them wants to distinguish itself from the others (Cox 1997). Thus, the distance between parties and voters increases (Blais and Bodet 2006; Wessels 1999). Moreover, proportionality of vote-to-seat conversion is associated with multi-member districts and low electoral thresholds, which creates diverse outcomes within proportional systems (Sirinic 2014). Also, these studies found it questionable that the median parties pull the government toward the center in coalition government formation because their inclusion depends on elites’ decision, which may prefer alternative parties located further from the median voter position (Ferree et al. 2013). As to the majoritarian electoral system (hereafter SMD), its effect on congruence is not straightforward. SMD promotes congruence by producing identifiable governments because a single party often wins the majority of the seats in the parliament. When elections create single party majority government, the majority support the party and, therefore, the majority’s preferences will approximate the party’s policy position. On the other hand, majoritarian systems offer incentives for strategic voting because minor parties are less likely to pass electoral thresholds. In this setting, voters may choose the party that is not necessarily closest to their preference in order to not waste their votes. Yet others find no consistent advantage of either PR systems or SMD systems in enhancing congruence between governing parties and voters (Blais and Bodet, 2006; Golder and Stramski 2010; Golder and Lloyd 2014). The contradictory results suggest that the electoral systems may not be sufficient to explain congruence level as a main determinant, which motivates this study to explore beyond the formal institutional determinants. To help explain the lack of a conclusive answer, I turn to an alternative linkage mode called clientelism. 87 Recent research expands the basis of comparison to include new democracies (Blais and Bodet 2006; Golder and Stramski 2010). However, little empirical research considers clientelism as the crucial determinant of congruence even though the role of clientelism is expected to be more prominent in new democracies where volatile party system offers an ideal incubator of clientelism.53 4.4 Theory Congruence literature argues that functioning electoral institutions ensure the right representatives to be selected and voters’ preference to be reproduced by them. This argument assumes that citizens’ voting behaviors reflect their genuine policy preferences. I challenge this assumption by asking: what if voters’ electoral behaviors do not express their genuine preferences? Clientelistic practices are essentially a persuasion strategy employed by parties that seek to broaden their base of supporters. By means of material benefits, parties and individual elites attempt to influence voters’ preferences and behaviors (Kitcshelt 2000; Stokes 2005).54 For example, in Argentina the Peronist party distributes clientelist rewards to weakly opposed voters to induce them to switch their vote choices (Stokes 2005).55 In contrast to existing work on congruence, thus, I argue that political elites do not only replicate existing voter preference, but they also proactively influence voters’ choice. Consequently, I expect that clientelism undermines congruence by nudging voters in favor of the clientelist parties, which 53 Some of the exceptions include Luna and Zechmeister (2005); Kitschelt et al. (2010); Mainwaring, Bejarano, and Pizarro (2006). 54 The effectiveness of clientelism to change the voting behavior depends on the monitoring and enforcement mechanism on the part of the clientelist parties. 55 The effectiveness of effort depends on enforcement mechanism. Stokes argues that the Peronist party uses its “deep insertion in voters’ social networks” to violate the secret ballot, and is therefore able to enforce compliance when buying citizens’ vote. 88 often requires voters to choose options further away from their preferred positions. If a party’s clientelist efforts induce individual voters to vote for the party that do not closely represent their views, then even institutionalized electoral arrangements cannot reliably create close connections between voters’ preferences and policy outcomes. In this case, citizens’ voting behaviors may be endogenously determined by their clientelist relationship with parties and elites. In reality, we often observe that parties’ clientelistic efforts generate positive attitudes and political support among recipients. The case in point is Bolsa Familia, a large scale social welfare program of the Brazilian government. It offers financial aid to poor families, through which 26% of the Brazilian population were covered. The Washington Post reports the political effect of Bolsa Familia as follows: the poor strongly associates the program with Lula, … , and surveys indicate that the two-thirds of the population that lives on less than $500 per month provide overwhelming support for Lula. Lula is expected to gain 63 percent of total valid votes versus 36.8 percent for the opponent Alckmin, according to a poll (Washington Post 2006).56 As the electoral consequence of the welfare program illustrates, some recipients base their preferences on experiences of clientelistic exchange. Voters develop an affinity with a party regardless of whether or not the party represents their views. Table 4.1 shows that about 9.7% of the respondents feel close to a party even if they do not think any party represent their views. Figure 4.4 illustrates the cross-national distribution of this phenomenon. The figure demonstrates the differences between the percentage of respondents feeling close to a party and the percentage of respondents feeling that their views are represented by a party. The countries with positive values indicate that there are respondents who feel close to 56 Reel, M. (2006, October 29). Cash Aid Program Bolsters Lula's Reelection Prospects. Washington Post. Retrieved May 30, 2014, from http://www.washingtonpost.com/wpdyn/content/article/2006/10/28/AR2006102800823.html?nav=rss_world/southamerica 89 the party even if there is lack of representation. In other words, the gap between voters’ affinity to a party and the representativeness of the parties suggests that voters’ partisan affinity is not solely a function of issue positions as congruence literature assumes. Table 4.1. Cross-tabulation of ‘Having a party representing respondents’ views’ and ‘Having a part respondents feel close to’ Are you close to any political party? Yes No Don't know Total Is there a Yes 41.14 18.15 0.6 59.89 party that No 8.77 26.24 0.51 35.52 represents 0.49 4.59 your views? Don't know 1.13 2.97 Total 51.04 47.36 1.6 100 Note: Data from CSES Module 3. The number represents the cell percentages. Total number of cases are 60,328. The number is the percentage. Figure 4.4. Proportion of Respondents Having a Party Representing their Views Have an Affinity to a Party (%) - Having a Party to Represent (%) 10 0 -10 -20 -30 Australia Austria Belarus Brazil Canada Chile Croatia Czech Republic Denmark Estonia Finland France Germany Greece Hong Kong Iceland Ireland Israel Japan Korea, South Mexico Netherlands New Zealand Norway Peru Philippines Poland Portugal Romania Slovakia Slovenia South Africa Spain Sweden Switzerland Taiwan Thailand Turkey United States Uruguay -40 Note: Data from CSES Module 3 90 What motivates some people to feel close to a party even if a party does not represent their views? In Table 4.2, I examine if parties’ clientelist efforts are associated with this case. I report the coefficients on the measure of clientelism from a bivariate logit regression where ideological congruence is the dependent variable. I employ robust standard errors clustered by country to take account of potential heteroskedasticity. The results indicate that the level of clientelism is positively and significantly associated with the case in which voters have an affinity to a party without being represented by the party. Table 4.2. Bivariate Regression: Relationship between the Level of Clientelism and Having a Non-Ideological Affinity to a Party Model 1 Clientelism 0.1234*** (0.0320) _cons -3.7064*** (0.4117) N 63854 N cluster 38 chi2 14.8358 Note: DV: the percentage of respondents who do not have a party representing their views, but at the same time have an affinity to a party. This case is summarized in Table 4.1 highlighted in gray color. Robust standard errors clustered by country in parentheses; substantively similar results are obtained if I do not employ robust or clustered standard errors; *** p<0.001 The next question is then how voters’ affinity to a clientelist party lead to an actual vote choice. Recall that a negative effect of clientelism on congruence should occur only when voters actually vote for the party that is further from their ideal preferences. Literature emphasizes the importance of monitoring to actually get voters to vote for clientelist parties. However, as the Brazilian Bolsa Familia case illustrates, sometimes voters voluntarily respond to the material benefits by offering support even without explicit monitoring. Studies of preference formation argue that people often base judgments on the considerations that are immediately salient or most accessible to them at the time of expression (Zaller 1992). Due to memory limitations, it is costly 91 for people to examine exhaustive information whenever they form a preference (Druckman and Lupia 2000). Following from this argument, clientelism may influence voters’ choice at the ballot box in that it is a tangible action that increases awareness about the patron party. If an individual has recently engaged in clientelistic exchange, then a party or a candidate may come to the top of the individual’s head. As a result, an individual not only develops an affinity to the party that previously would not exist, but she may also choose them at the election. Figure 4.5. Degree of an Affinity to a Party Are you Close to Any Party? Degree of Closeness Don't Know No Yes 0 Very Close 10 20 30 40 Somewhat Close 50 Not very Close 60 70 80 90 Don't Know Note: Data from CSES Module 3; Total sample size: 43,232 My argument relies on the assumption contradictory to the conventional description of voters’ incentive structure. Drawing on the classic Downsian spatial model of political competition, existing studies of congruence presumes that voters can place all election options along a single dimensional political spectrum in light of their own ideological preferences. Among these options, voters are expected to find and select a single optimal choice of a party closest to their own ideological position according to their single-peaked preference structure. In reality, however, 92 voters’ incentive structure seems much more complicated. As Figure 4.5 shows, even among those who have a preferred party, the degree of attachment to the party varies. Specifically, only 23% of them felt “very close” to the party, 41% of them responded “somewhat close”, and 13% of them said “not very close.” Although the literature treats the party identifier as a homogeneous group, it is likely that the voting behavior of those who feel very close to the party is likely to be different than those who do not feel very close. Specifically, I expect that their behavioral differences become most salient when they face alternative incentives such as material benefits from other parties.57 In sum, I conceptualize clientelism in terms of a tool of persuasion that political parties and elites employ to broaden and consolidate their support bases. As a result of clientelist rewards, individual voters develop an affinity to patron parties and vote for the parties that do not necessarily represent their policy preferences. Thus, I argue that clientelism systematically increases the gap between voters’ preferences and their vote choices, which in turn decreases the level of congruence. Taken together, my theoretical discussion leads to the following hypothesis about the relationship between clientelism and congruence: H: Countries with higher levels of clientelism show higher levels of incongruence in policy preferences between citizens and governing parties. 57 The reasons why voters respond to non-programmatic redistributive benefits vary, ranging from low partisanship to lack of political information (Dixit and Londregan 1996; Grossman and Helpman 2001). 93 4.5 Data and Measures 4.5.1 Dependent Variable: Congruence My dependent variable evaluates the degree of congruence between the policy positions of the governing parties and the substantive preferences of the citizens. I use the many-to-many measure of congruence by Golder and Stramski (2010), which compares the similarity of the distributions of citizen and representative preferences on the left-right issue dimension. Specifically, the measure is calculated by the following formula: 𝐶𝑜𝑛𝑔𝑟𝑢𝑒𝑛𝑐𝑒 = ∑ 𝐹𝑣𝑜𝑡𝑒𝑟 (𝑥) − 𝐹𝑝𝑎𝑟𝑡𝑦 (𝑥) 𝑥 , where 𝐹𝑣𝑜𝑡𝑒𝑟 (𝑥) and 𝐹𝑝𝑎𝑟𝑡𝑦 (𝑥)are the cumulative distribution functions (CDFs) for the citizen and party preferences. This measure calculates the area between the CDFs for the citizens and parties. As the discrepancies between the parties’ position and the citizens’ position increase, so thus the area between the CDFs. The larger value of the measure indicates lower congruence (Golder and Stramski 2010).58 In calculating my measure, I modified one thing from the many-to-many measure of congruence suggested by Golder and Stramski (2010). Instead of averaging them out, I take into consideration the full range of party position scores assigned by individual experts on a given party. By using an average score, Golder and Stramski assumes that a party is a unitary actor sharing the ideological position and policy preferences (2010: 97). In reality, this is rather a stringent assumption to meet. Political parties are often volatile alliances of politicians with heterogeneous interests and views rather than behaving as unitary actors (Kitschelt and Freeze 2011). If experts’ evaluation on a given party presents greater variations, it indicates that the party fails to provide 58 Note that my measure of congruence compares the policy preferences of parties from that of citizens in a single point in time and it does not take into consideration the actual policy output. 94 coherent and unambiguous position. In addition, it also implies that individual representatives vary from one another within a party or their mobilization efforts cut across the party line. 59 The variation can provide an important information about coherency of party’s position, thus I will take advantage of this variation in calculating my congruence measure. Since the measure of manyto-many congruence intend to captures the preferences of the collective representative body as a whole, I believe my modification of measure also suits the purpose of the original measure. To construct the measure of congruence, I need data on both the distribution of citizen preferences and the distribution of parties’ preferences. Following the conventional approach, I obtain data on the distribution of citizen preferences from the public opinion survey, while the distribution of parties’ preferences from the expert surveys on democratic accountability mechanisms.60 Specifically, to evaluate the distribution of parties’ preferences I use the expert survey dataset from the Democratic Accountability and Linkage Project (DALP) compiled by Kitschelt (2013). The expert survey has been conducted in 89 countries between 2008 and 2009. The respondents were asked to evaluate political parties in their countries regarding the ways in which parties attempt to connect with voters, ranging from programmatic to clientelistic efforts (Kitschelt and Freeze 2011). The advantage of using this data is that the data also includes the measure of my key independent variable, i.e. the level of clientelism. To evaluate voters’ positions I use the World Value Survey (WVS) wave 5. When the same questions are available, I use two 59 To further illustrate my point, let me provide a brief hypothetical example. Let’s say 10 experts are asked to place the two parties A and B of their country on the same left-right scale from 0 to 10, where 0 means the left. The score of party A’s position ranges from 2 to 8 with standard deviation of 2 and mean 4.5, while the score of party B ranges from 4 to 6 with standard deviation of 0.5 and mean 4.5. Although their mean positions are the same, experts’ perceived position of party A present greater variations than that of party B. 60 Some of the studies using the combination of public opinion data and expert survey data includes the following: Espana (2010); Powell (2006); Eerd (2011); Boas and Smith (2014); Luna and Zechmeister (2005); Rohrschneider and Whitefield (2007). 95 additional public opinion survey to fill in the missing data: Latin American Public Opinion Project (LAPOP) (2008) and AsiaBarometer (2005). Although it is a widespread practice to compare the positions of the voters and the party from the different survey sources, some may consider this practice problematic due to the differences in the scale in some cases. Since my measures compare overall distribution patterns between citizens and parties rather than exact positions on the issues, this problem is less of a concern. To avoid potential limitations, I further corroborate my result using an alternative congruence measure on the overall left-right placement, which is created from a single dataset, i.e. Comparative Study of Electoral Systems (CSES) project.61 Based on these datasets, I create the two different measures of congruence as follows. Table 4.3 presents the complete wording and associated dataset. (1) Congruence on the Overall Left-Right Placement I construct the measure of congruence on left-right placement because it is the most common aspect of congruence which existing studies have examined. I do not insist that most voters perceive political competition in terms of sophisticated ideological concepts, such as socialism or liberalism. However, it has been used as a scale that represents the major conflicts in the political system (Inglehart 1990: 273). (2) Congruence on the Issues regarding the Redistribution In addition to comparing the left-right position, I also examine congruence in an additional substantive issue area, which is the common proxy for governmental policy output in the literature 61 This approach has its own limitation in that the public opinion surveys may capture citizens’ perception on the positions of both citizens and parties. However, the accuracy of public perceptions of parties’ positions is not a critical problem because voters will choose parties at the elections based on these perceptions (Dalton, Farrell, and McAllister 2011). Since my study mainly concerns with the extent to which voters feel they are represented, the measure based on citizens’ perceptions of the parties serves the purpose. 96 (Hobolt and Klemmensen 2008). Because left-right does not summarize all the issues and cleavages that define political competition in many polities, congruence on other substantive issues offer more complete picture of congruence patterns. I create the measure of congruence on redistribution issues. It compares the preferences of the parties and voters on whether they advocates extensive social spending and redistributing income to benefit the less well-off in society. Table 4.3. Survey Questions and Dataset [Congruence] Overall Left-Right Placement 1: Expert survey + Mass survey Parties' DALP [1] Party is best located at the “left” of the national political spectrum based position upon its overall policy positions and ideological framework. [10] Party is best located at the “right” of the national political spectrum based upon its overall policy positions and ideological framework. Voters' WVS In political matters, people talk of "the left" and "the right." How would you position place your views on this scale, generally speaking? (Code one number): Left 1 2 3 4 5 6 7 8 9 10 Right LAPOP On this card there is a 1-10 scale that goes from left to right. Nowadays, when we speak of political leanings, we talk of those on the left and those on the right. In other words, some people sympathize more with the left and others with the right. According to the meaning that the terms "left" and "right" have for you, and thinking of your own political leanings, where would you place yourself on this scale? Indicate the box that comes closest to your own position: Left 1 2 3 4 5 6 7 8 9 10 Right [Congruence] Social spending on the disadvantaged: Expert survey + Mass survey Parties' DALP [1] Party advocates extensive social spending redistributing income to position benefit the less well-off in society. [10] Party opposes extensive social spending redistributing income to benefit the less well-off in society. Voters' WVS How would you place your views on this scale? 1 means you agree position completely with the statement on the left; 10 means you agree completely with the statement on the right: Incomes should be made more equal 1 2 3 4 5 6 7 8 9 10 We need larger income differences as incentives for individual effort LAPOP Please tell me to what extent you agree or disagree with the following statements: The (Country) government should implement firm policies to reduce income inequality between the rich and the poor. To what extent do you agree or disagree with this statement? AsiaB Please indicate how much you agree or disagree with each statement: It is desirable that the people are equal, even if the economy is stagnant, rather than unequal but developing fast (AsiaBarometer 2005) 97 4.5.2 Independent Variable: Level of clientelism To measure the level of clientelism, I use the same expert survey dataset from the Democratic Accountability and Linkage Project (DALP) that is used to obtain the distribution of parties’ preferences. The survey asks the respondents to evaluate how much effort parties make in their country to their electoral target constituencies. Specifically, the question asks about the five most commonly used “currencies” of clientelistic exchange, including gifts of consumer goods, preferential access to social policy entitlements, public sector employment, government contracts or procurement opportunities for business, and client influence over regulatory procedures. For each item, parties’ clientelistic efforts are evaluated based on four grading schemes from “a negligible effort or none at all” to “a major effort.” Operationally, I follow Kitschelt (2011) and create a composite index at the country level based on the five average measures of clientelistic efforts of individual parties weighted by their vote share in the most recent national legislative election within a country. 4.5.3 Control Variables I take into account the institutional and socio-economic factors that might affect the level of congruence. On the one hand, my model controls for the electoral system. Some literature argues that PR promotes a better fit between voters and parties in their positions (Powell 2009), while others find no effect of electoral institutions on congruence level. In addition, I incorporate GDP growth to control for the level of economic development. Literature expects that socio-economic development is positively associated with the level of congruence (Kitschelt 2000). 98 4.6 Empirical Analysis 4.6.1 Descriptive Analysis In light of the many-to-many congruence measure, I present the patterns of congruence in some countries categorized by the level of clientelism.62 Figure 4.6 shows that the distributions of citizens and parties are more similar in countries with lower level of clientelism. With increasing level of clientelism, the figures present greater discrepancies between the distributions both in terms of location and shape.63 Figure 4.6. Congruence Patterns by the Level of Clientelism A. Lowest Level of Clientelism: 25th percentile SLOVENIA 40 30 20 10 0 0 1 2 3 4 voter_cses3 5 6 7 8 9 10 party_cses3 62 To produce the figures, I used the questions on the left-right placement available from CSES wave 3. The level of clientelism has been obtained from the expert survey dataset, DALP. 63 See Table 4.7 in Appendix for the complete list of countries in my analysis categorized by the level of clientelism. 99 Figure 4.6 (cont’d) B. Lower Level of Clientelism: 50th percentile CHILE 30 25 20 15 10 5 0 0 1 2 3 4 5 party_cses3 6 7 8 9 10 9 10 voter_cses3 C. Higher Level of Clientelism: 75th percentile ROMANIA 20 15 10 5 0 0 1 2 3 4 5 voter_cses3 6 7 8 party_cses3 D. Highest Level of Clientelism: 100th percentile MEXICO 25 20 15 10 5 0 0 1 2 3 4 voter_cses3 5 6 7 8 party_cses3 100 9 10 4.6.2 Analysis To test the hypothesis on the relationship between clientelism and congruence, I use the two measures of congruence as the dependent variables. In the analysis, I am interested in how the level of clientelism is related to the two measures of congruence. Instead of using separate regression analyses for each outcome variable, therefore, I estimate the multivariate linear regression model. Multivariate results allow for testing of coefficients across equations. Table 4.4 presents the estimation result of the multivariate analysis. As expected, the level of clientelism is strongly related to both the congruence on left-right placement and the congruence on redistribution issues. The sign is consistently positive, which means that increasing clientelism is associated with an increasing gap between parties’ preferences and voters’ preferences.64 None of the control variables are significant. In particular, the variable on proportional representation does not reach statistical significance, although it presents a negative sign as conventional studies expect. On the whole, this analysis provide support for my hypothesis that clientelism is negatively associated with congruence. Table 4.4. Multivariate Analysis of Congruence Clientelism PR Polarization GDP Growth Model 2 DV: Position on Left-Right Placement 0.1793** (0.0564) -0.1389 (0.0968) 0.0694 (0.0430) -0.0063 (0.0088) Model 3 DV: Position on Redistribution 0.2066*** (0.0566) -0.1614 (0.0972) 0.0515 (0.0432) -0.0050 (0.0088) 64 Recall that my measure of congruence captures the area between the Cumulative Distribution Functions of parties’ and that of voters’. The greater the area, the larger the discrepancies between parties’ preferences and voters’ preferences. In other words, the larger value of the measure indicate lower congruence. 101 Table 4.4 (cont’d) _cons N F r2 0.4843** (0.1756) 47 2.613 0.199 0.2880 (0.1764) 47 3.606 0.256 To better understand how much clientelism is associated with congruence under various scenarios, Table 4.5 reports the predictions from several simulations for each congruence measure. These simulations predict the changes in the expected value of congruence level caused by changes in the level of clientelism by 20th percentile increments, which indicates the strength of the effects of clientelism. The probabilities are predicted by the separate linear regression model with the robust standard errors. All control variables are held at the mean, except the electoral institutions measure, which is held at 0 (SMD) and 1 (PR), respectively. The results show that, in all cases, the first differences are positive with the increase in the level of clientelism, which indicates that clientelism increases the value of congruence measure. In other words, clientelism decreases the level of congruence. Table 4.5. First Difference: Changes in the Expected Value of Congruence by the Changes in the Level of Clientelism under PR and SMD Level of Clientelism First difference PR min-p20 p20-p40 p40-p60 p60-p80 p80- max SMD min-p20 p20-p40 p40-p60 Congruence Left-Right Placement 0.165 ( 0.078 0.248 ) 0.157 0.076 ( 0.036 0.114 ) 0.100 0.090 ( 0.042 0.136 ) 0.080 0.062 ( 0.029 0.093 ) 0.046 0.059 ( 0.028 0.088 ) 0.060 0.169 ( 0.085 0.259 ) 0.154 0.078 ( 0.039 0.119 ) 0.098 0.092 ( 0.047 0.142 ) 0.079 102 Redistribution ( 0.067 0.247 ( 0.043 0.157 ( 0.034 0.126 ( 0.020 0.073 ( 0.026 0.094 ( 0.067 0.245 ( 0.043 0.156 ( 0.034 0.125 ) ) ) ) ) ) ) ) Table 4.5 (cont’d) p60-p80 0.063 ( 0.032 0.097 ) 0.046 ( 0.020 0.072 ) p80- max 0.060 ( 0.030 0.092 ) 0.059 ( 0.026 0.093 ) Note: This simulation result is produced using the CLARIFY statistical software package (Tomz et al. 2003). Entries are First Differences for each measure of congruence. I report the change in the expected value of congruence level caused by increasing the level of clientelism by 20 percentile increment: e.g. from minimum to 20th percentile, from 20th percentile to 40th percentile, from 40th percentile to 60th percentile, from 60th percentile to 80th percentile, and finally from 80th percentile to maximum value. 95% confidence interval in parentheses. Probabilities are based on linear regression estimates with robust standard errors. Control variables are held at mean except for the electoral institutions. In turn, Table 4.6 presents the changes in the expected value of congruence with changes in the electoral institutions from SMD to PR. As the 95% interval value indicates, this change does not have any statistically significant impact on the level of congruence. In sum, the simulation also provides the supportive evidence in favor of my hypothesis that clientelism is associated with a lower level of congruence. Table 4.6. First Difference: Changes in the Expected Value of Congruence with the Changes from SMD to PR Electoral System Congruence First difference Left-Right Placement Redistribution SMD ->PR -0.108 ( -0.289 0.077 ) -0.087 ( -0.260 0.097 ) Note: This simulation result is produced using the CLARIFY. Entries are First Differences for each measure of congruence. I report the change in the expected value of congruence level caused by changing electoral system from SMD to PR. 95% confidence interval in parentheses. Probabilities are based on linear regression estimates with robust standard errors. Control variables are held at mean. 4.6.3 Robustness Check To ensure that my finding is not contingent on the choice of measurements, I estimate the model using a conventional measure of congruence, i.e. the absolute ideological distance between the median citizen and the governing parties. Based on the dataset compiled by Kim, Powell, and Fording (2010), I create the measure of congruence capturing the distance between the position of 103 the median voter and the median parties, which is labelled as distortion.65 Parties’ positions are estimated based on a cross-national left-right scale constructed from the mentions of a number of topics compiled by Comparative Manifesto Project. The position of the median voter is estimated based on the distribution of party votes (Kim and Fording 2003). As a measure of my key independent variable, i.e. clientelism, I first use my original measure derived from DALP expert survey data. In addition, I estimate the models based on proxies for clientelism. First, I use the measure of the personal vote system as a measure of level of clientelism using the dataset by Johnson and Wallack (2006). This variable ranks countries in increasing order of incentives to cultivate a personal vote. The variable ranges from 1 to 13, which takes into account four dimensions of the electoral system: ballot, vote, pool, and district magnitude (Carey and Shugart 1995). In personal vote system, where a party label helps little in electoral competition, individual politicians are more likely to be engaged in clientelistic practice to distinguish them from other candidates. Second, I use the measure of the corruption level from the dataset by Transparency International (2012). 66 Corruption Perception Index is another commonly used proxy for clientelism. The CPI focuses on corruption in the public sector and defines corruption as the abuse of public office for private gain. The CPI Score ranges from 0 to 1, where 0 represents highest level of corruption. Finally, I use 4 measures from the Global Competitive Index constructed by the World Economic Forum, which include (i) diversion of public funds, (ii) favoritism in decisions of government officials, (iii) irregular payments and bribes, and (iv) wastefulness of government spending. These measures range from 1 to 7, where 1 represents a higher level of clientelism. 65 Distortion is the opposite concept of congruence. 66 These measures of clientelism are coded such that lower scores represent higher level of clientelism. 104 Figure 4.7-4.10 in Appendix present the estimation results of the linear regression. The coefficient of clientelism is statistically significant and positively associated with the distortion, thus negatively correlated with congruence. As presented in the figures, this result is corroborated across alternative measures of clientelism. In sum, the results are in line with my main findings that a higher level of clientelism is associated with higher distortion and lower congruence. 4.7 Discussion and Conclusion This paper presents that clientelism is negatively associated with the level of congruence. Drawing on the conceptualization of clientelism as a tool of persuasion employed by political parties, I argue that elites distribute clientelistic rewards in order to alter the preferences and behavior of voters who would otherwise hold different or no views about parties’ policies. In response to the rewards, I expect that some voters develop an affinity to and vote for a party located further from their ideal points. Consequently, clientelism systematically widens the gap between voters’ preferences and their vote choices, thereby undermining the congruence. The empirical result supports my expectation that countries with higher levels of clientelism present higher levels of incongruence in policy preferences between the public and the governing parties. To advance an argument, I challenge the conventional assumptions that individual voters have one dimensional and single peaked political preferences. Instead, I suggest that voters’ utility is a function of both issue position and material inducement. In response, parties also employ both programmatic and clientelistic linkage strategies to maintain sustainable electoral support. Regarding the relationship between the two mechanisms, Robinson and Verdier (2013) describes it as “two faces of the same political game.” My finding contributes to the literature on congruence by demonstrating the overlooked effect of an informal determinant on congruence level. In existing works of congruence, the debate 105 on the determinant of congruence has predominantly revolved around formal electoral mechanism. I concur that the formal electoral rules is at the core of democratic system. However, in order to fully understand how congruence is achieved, I maintain that we need to consider an informal realm in which parties and voters interact. Clientelistic rules are informal and unwritten institutions, but they may not necessarily be of trivial value. As Stokes (2006) point out, it may have an important impact on how democracy works when large number of citizens use it. 4.7.1 Implications for Core-Swing Voter Argument My findings suggest that clientelism is negatively associated with the level of congruence. This result has implications for the debate over the type of voters which parties channel their material resources into. Specifically, it provides a potential evidence in favor of the swing voter argument. Given the limited resources, parties have to determine how to allocate the resources to make the most out of them. To explain how parties behave in the face of conflicting incentives, the swing voter model predicts that parties distribute selective goods to the swing voter (Lindbeck and Weibull 1987). In this scenario, parties seek to create a heterogeneous but broader support base. The main purpose of resource allocation is to persuade the voters to switch their vote choices away from their initial preferences. On the other hand, the core voter theory argues that parties allocate their resources to core supporters (Cox and McCubbins 1986; Cox 2005). Parties target narrower but coherent groups of voters. By definition, the core voters are already predisposed in favor of a party. Thus, what parties try to achieve through resource allocation is mainly coordination and mobilization rather than persuasion.67 67 Shaw (2008) defines core voters as those who have voted the same party in three consecutive elections. 106 Following this logic, core voters experience less tension between their preferences and their vote choice in response to clientelistic benefit, than their swing counterparts. Regarding the outcome of congruence, the swing voter hypothesis is strengthened if parties use clientelism as persuasion strategy which induce the beneficiary of clientelistic benefit to change their vote choices further from their ideal points. The negative association between clientelism and congruence suggests that clientelism undermine congruence by inducing voters to choose the option further from their original preferences. On the other hand, the core-voter hypothesis is supported if parties use clientelism as a mobilization strategy which affects core supporters to vote but not demand any changes in vote choices further from their ideal points. In this scenario, the relationship between clientelism and congruence should be either positive or null. The fact that clientelism undermine congruence implies that parties are more likely to direct their efforts to the swing voters to persuade away from their ideal position.68 Swing voters are generally political moderates who are located near the political center of the electorate (Dixit and Londregan 1996: 1144). They are more likely to compromise their party affinities or ideological position in response to particularistic benefits (Dixit and Londregan 1996), and, therefore, they are susceptible to the persuasion effect of clientelism that creates a gap between voters’ original preferences and their vote choices. 68 Studies show that information also affects the stability of preferences. Specifically, the best-informed and worst-informed voters have highest stability. Whereas the best-informed voters are knowledgeable enough to limit the extent to which new information influence them, the worst-informed persons have little feedback and stimulation for preference change (Druckman and Lupia 2000). 107 4.7.2 Endogenous Preference Formation and Implication for Congruence This paper argues that voters’ informal relationship with parties and candidates shapes voters’ voting behavior. This raises a fundamental question about whether parties’ clientelistic efforts only influence voters’ behavior in short term or whether they will eventually influence and shape voters’ preferences. My study cannot answer this question because most of the public opinion datasets have short time span. With limited time points in the dataset, it is difficult to track how clientelism affect the changes in ideological and policy preferences of voters over time. For the sake of future research, however, the answer to this question has important normative implications for the study of congruence. If voters’ preference is not necessarily exogenous and fixed as congruence literature assumes and if it is susceptible to the influence from political elites, can the match between voters’ preference and elites’ preference still serve as a valid yardstick to evaluate the quality of democracy? Is it enough to look into the formal electoral rules that translate voters’ preference into policy? Two increasing body of literature speaks to these questions. On the one hand, research shows that individual citizens’ relationship with parties affect how they process political information (Zaller 1992). Specifically, insider and outsider of clientelist network develop systematically different attitudes and preferences in all aspects of political life, be it vote choice, ideological self-placement, or attitudes toward the government policy (Anderson and Tverdova 2001). The beneficiaries of either formal or informal systems interpret the workings of the political system in a more positive way. 108 On the other hand, the cognitive dissonance theory demonstrates that an action can change preferences, not the other way around. 69 The theory states that individuals try to maintain consistency in their behavior, beliefs, and attitudes, as being inconsistent generates psychological discomfort (Festinger 1957). Thus, when they have behaved in a way contradictory to their beliefs or preferences, the individual unconsciously alters his beliefs in line with their behaviors so as to alleviate the discomfort of having inconsistent attitudes and actions (Mullainathan and Washington 2009). For example, Mullainathan and Washington (2009) find that the mere act of voting for a specific candidate boosts a more positive attitude toward the candidate. 70 The individual voters adjust their partisan attitudes toward a party to alleviate the internal discomfort, which arises from having voted for a candidate whom they used to have a poor opinion of. Contrary to the common assumption in the voting behavior literature, the dissonance theory implies that voting behaviors may shape preferences and beliefs, not the other way around (Denny and Doyle 2007).71 After all, it is easier to change preferences and attitudes than to go back and undo the behavior that has already occurred (Dinas 2013; Festinger 1957). The idea of the two theories point to the possibility that voters’ preference may be compromised to the interest of parties and political elites who have power and resources to exercise influence. My findings present that clientelism influence citizens’ voting behaviors. As the cognitive dissonance theory suggest, if the vote choice itself can create a sense of commitment and get the voters increasingly attached to their choice, i.e. the patron party, clientelist efforts by parties 69 Elster, 1983. Sour Grapes: Studies in the Subversion of Rationality; Festinger, 1957. A Theory of Cognitive Dissonance. 70 The studies on the impact of voting behavior on party identification dates back to several decades ago (Dobson and St. Angelo 1975; Howell 1980; Jennings and Markus 1984; Knoke 1976; Markus and Converse 1979). However, these studies are mainly framed in the context of short-term partisan volatility (Dinas 2013). More recent studies, also see Beasley and Joslyn (2001) 71 Also, see Bandura (1989). 109 and elites can manufacture the citizen preference to their favors. In this scenario, the conventional studies on congruence that merely comparing voters’ preferences and parties’ preferences may only reflect preferences of parties and elites. It may not be the kind of representative democracy that normative theorists envisioned. 110 APPENDIX 111 Figure 4.7. Clientelism (measure: DALP expert survey) and Level of Distortion Level of Clientelism Proportional Representation Polarization N: 22 F: 2.8508 Log-likelihood: -45.3081 R2: 0.4015 GDP Growth (%) -10 -5 0 Average Marginal Effects 5 Note: The figure plots the estimation results of linear regression model. Congruence dataset from Kim, Powell, and Fording (2010); Clientelism Dataset from Kitschelt (2013) Democratic Accountability and Linkages Project. Figure 4.8. Clientelism (measure: personal vote system) and Level of Distortion Personalistic Tier Electoral System Polarization N: 149 # of clusters: 21 R2: 0.2091 GDP Growth (%) -6 -4 -2 0 Average Marginal Effects 2 Note: The figure plots the estimation results of linear regression with panel-corrected standard errors; Congruence dataset from Kim, Powell, and Fording (2010); Clientelism Dataset from Johnson and Wallack (2006: Electoral Systems and the Personal Vote). 112 Figure 4.9. Clientelism (measure: corruption perception index) and Level of Distortion Corruption Perceptions Index Electoral System Polarization N: 51 R2: 0.2074 GDP Growth (%) -6 -4 -2 0 2 Average Marginal Effects 4 Note: The figure plots the estimation results of linear regression model with robust standard error clustered at the country level. Congruence Dataset from Kim, Powell, and Fording (2010). Corruption Perception Index is commonly used as a proxy for clientelism. The CPI focuses on corruption in the public sector and defines corruption as the abuse of public office for private gain. The CPI Score ranges between 10 (highly clean) and 0 (highly corrupt) 113 Figure 4.10. Clientelism (measures from World Economic Forum) and Level of Distortion Lack of Clientelism Lack of Clientelism Electoral System Electoral System Polarization Polarization GDP Growth (%) GDP Growth (%) -15 -10 -5 0 5 Diversion of public funds -10 -5 0 5 Favoritism in decisions of government officials Lack of Clientelism Lack of Clientelism Electoral System Electoral System Polarization Polarization GDP Growth (%) GDP Growth (%) -10 -5 0 5 Irregular payments and bribes -15 -10 -5 0 5 Wastefulness of government spending Note: The figure plots the estimation results of linear regression model. Congruence Dataset from Kim, Powell, and Fording (2010); Clientelism Dataset from (Schwab 2012); Each model is estimated using 4 alternative measures of clientelism. The measure of clientelism ranges from 1 to 7, where 1 represents higher level of clientelism. Years 2011–12 weighted average. 1. Diversion of Public Funds: "In your country, how common is diversion of public funds to companies, individuals, or groups due to corruption?" [1 = very common; 7 = never occurs]. N: 24, F: 3.8686, Log-likelihood: -68.0186, R2: 0.4489 2. Irregular Payments and Bribes: Average score across the five components of the following Executive Opinion Survey question: "In your country, how common is it for firms to make undocumented extra payments or bribes connected with (a) imports and exports; (b) public utilities; (c) annual tax payments; (d) awarding of public contracts and licenses; (e) obtaining favorable judicial decisions." In each case, the answer ranges from 1 (very common) to 7 (never occurs). N: 24, F: 3.8686, Loglikelihood: -68.0186, R2: 0.4489 3. Favoritism in Decisions of Government Officials: "To what extent do government officials in your country show favoritism to well-connected firms and individuals when deciding upon policies and contracts?" [1 = always show favoritism; 7 = never show favoritism]. N: 24, F: 3.8686, Log-likelihood: -68.0186, R2: 0.4489 4. Wastefulness of Government Spending: "How would you rate the composition of public spending in your country?" [1 = extremely wasteful; 7 = highly efficient in providing necessary goods and services]. N: 24, F: 3.8686, Log-likelihood: -68.0186, R2: 0.4489 114 Table 4.7. Countries Categorized by the Level of Clientelism in 2008-2009 Very low Australia Austria Canada Czech Republic Denmark Estonia Finland France Germany Ireland Latvia Netherlands New Zealand Norway Slovakia Slovenia Spain Sweden Switzerland United Kingdom United States Low Chile Costa Rica Greece Israel Italy Japan South Korea Malaysia Poland Portugal Russia South Africa Thailand Uruguay High Bangladesh Bolivia Brazil Croatia Georgia Hungary Indonesia Moldova Morocco Pakistan Peru Romania Serbia and Montenegro Taiwan Ukraine Zambia 115 Very High Argentina Bulgaria Colombia Dominican Republic Ecuador Egypt El Salvador Ghana Guatemala Honduras India Jamaica Mali Mexico Mongolia Nicaragua Panama Paraguay Philippines Turkey Venezuela CHAPTER 5 CONCLUSION This dissertation examines patterns of party system institutionalization in terms of informal rules of clientelism. Motivated by the lack of focus on informal institutions in the literature on party system institutionalization, my dissertation focuses on how clientelist practices affect three types of relationships that are crucial to understanding this process: (i) inter-party relationships, (ii) intra-party relationships, and (iii) party-voter relationships. This dissertation argues that clientelist practices shape the incentives motivating individual actors and thereby have significant impacts on the party organizations and party systems in which they are involved. Specifically, my dissertation argues that clientelism exerts centrifugal forces on the relationships among relevant actors, which in turn has negative effects on party system institutionalization. With regard to inter-party relationships, Chapter 2 examines the mechanism of new party formation by focusing on the structures in which clientelist exchanges occur. I review the conventional argument that clientelism empowers the ruling parties, and I suggest that the effect of clientelism is contingent on the specific clientelist structure. Conceptualizing clientelist exchange as a bargaining process, I argue that clientelist practice under a decentralized structure facilitates new party formation. Three levels of empirical analysis offer consistent evidence supporting my hypothesis. From the cross-national analysis, I find that the effect of clientelism on emergence of new parties depends upon the details of the clientelist structure. In centralized settings, the number of new parties decrease with increasing levels of clientelism. However, this negative effect of clientelism is alleviated under decentralized structures. Under decentralized structures, the number of new parties increase with increasing clientelism. In the party-level analysis, I show that parties’ clientelist efforts are less effective in producing actual votes if the clientelist structure is decentralized. This result corroborates my theoretical prediction that 116 decentralized clientelist structures reduce the credibility of threats to withhold clientelist resources, and consequently undermines the durability of the existing patron-client relationship. Finally, the results from the individual-level analysis demonstrate that clientelist levels and structures affect citizens’ votes for new parties. Specifically, I find that as voters recognize greater dispersion of a party’s authority over resources, the effects of increasing clientelist levels on voters’ support of incumbents decreases substantially. These results suggest the significance of clientelist structure in determining the frequency of new party formation in new democracies. With special emphasis on intra-party relationships, Chapter 3 investigates the determinants of varying patterns of territorial vote distribution of parties, i.e. party nationalization. Based on a mediation analysis using the party-level datasets, I find that clientelism mediates the effect of party extensiveness on party nationalization. Although the direct effect of party extensiveness is to increase party nationalization, the mediational path has the opposite effect: party extensiveness increases the effectiveness of clientelism, which in turn undermines party nationalization. Therefore, clientelism exerts a suppression effect on the relationship between party extensiveness and party nationalization, which would otherwise present stronger positive associations. Finally, Chapter 4 focuses on the dynamics of party-voter relationships. I posit that clientelism systematically widens the gap between voters’ preferences and their vote choices, thereby undermining congruence. Using a multivariate linear regression analysis based on the many-to-many congruence measure, I find that the level of clientelism is negatively associated with congruence on left-right placement and congruence on redistribution issues. This finding has implications for understanding parties’ and voters’ behaviors. On the one hand, in terms of voters’ behaviors, it suggests that voters respond to both issues and material inducement. This raises a question about the validity of existing arguments in the congruence 117 literature. Specifically, my study asks whether a match between voters’ preferences and elites’ preferences is a valid yardstick to evaluate the quality of a democracy, if voters’ preference is susceptible to influence from political elites. Given that voters’ preferences may be compromised to the interest of parties and political elites who have the power and resources to exercise influence, I conclude that it is not sufficient to focus on formal electoral rules as a determinant of congruence. On the other hand, in terms of parties’ behavior, the result that clientelism undermines congruence implies that parties are more likely to direct their efforts to persuading swing voters away from their ideal positions. Swing voters are more likely to compromise their ideological position in response to particularistic benefits. Therefore, they are more susceptible to the persuasion effect of clientelism that creates a gap between voters’ original preferences and their vote choices. My findings suggest some possible extensions for future research. First, researchers may improve upon the method for conceptualizing and measuring patron-clientelism. This dissertation draws on expert surveys and proxy data to capture key dimensions of clientelist practices. More direct and objective measures of clientelism can enhance our understanding of the relationship between clientelism and party system institutionalization. Importantly, given the significance of temporal elements in party system institutionalization, collecting data on changes in the patterns of clientelism over time is desirable. The traditional perspective on clientelism assumed its eventual disappearance in the course of democratization. However, recent phenomena show the resilience of clientelism, which has adapted to changing environments. Still, we know little about how the forms, extents, and functions of clientelistic practices change over time, because existing literature conceptualizes it as largely static and there is few data available that captures these temporal dynamics. The ever- 118 changing political landscape constantly rearranges the balance of power and creates new political spaces. Thus, the scope and nature of clientelism is expected to vary over time. In addition, research can be further extended to different types of clientelism. For example, private sources of clientelist benefits, such as various business sectors, can also facilitate the formation of patron-client relationships. With enhanced data availability, many questions about the dimensions of clientelist practices can be addressed, such as: How are clientelistic network organized across different levels of government? And how much autonomy do key actors (e.g. patron, clients, and brokers) have in a given patron-client network? Finally, the enforcement of the informal rules of clientelism is another issue which needs to be examined. Potential questions regarding this topic include: Why have patron-client relations remained intact in some cases, while they have been repeatedly challenged in others? What explains variation in the strength and level of stability in patron-client relations? When and how do the key actors manage to institutionalize clientelistic flows of resources? And how do the patrons or the brokers induce compliance from clients? This dissertation ultimately aims to motivate the study of the political functions of diverse informal institutions. By demonstrating the significant role of informal mechanisms in shaping political outcomes, my study suggests that any future research in comparative political institutions needs to consider the interactions of informal and formal institutions. 119 BIBLIOGRAPHY 120 BIBLIOGRAPHY Adams, James, Jane Green, and Caitlin Milazzo. 2012. “Who Moves? Elite and Mass-Level Depolarization in Britain, 1987–2001.” Electoral Studies 31 (4): 643–55. doi:10.1016/j.electstud.2012.07.008. African Elections Database. 2013. A Database of Election Results in Sub-Saharan Africa. http://africanelections.tripod.com. Alemán, Eduardo, and Marisa Kellam. 2008. “The Nationalization of Electoral Change in the Americas.” Electoral Studies 27 (2): 193–212. doi:10.1016/j.electstud.2007.10.005. Alesina, Alberto, Arnaud Devleeschauwer, William Easterly, Sergio Kurlat, and Romain Wacziarg. 2003. Fractionalization. Working Paper 9411. National Bureau of Economic Research. http://www.nber.org/papers/w9411. Álvarez-Rivera, Manuel. 2013. “Election Resources on the Internet.” Retrieved from http://www.electionresources.org. Amendola, Adalgiso, Joshy Easaw, and Antonio Savoia. 2013. “Inequality in Developing Economies: The Role of Institutional Development.” Public Choice 155 (1-2): 43–60. Anderson, C. J., and Y. V. Tverdova. 2001. “Winners, Losers, and Attitudes about Government in Contemporary Democracies.” International Political Science Review 22 (4): 321–38. doi:10.1177/0192512101022004003. Andeweg, Rudy. 2011. “Approaching Perfect Policy Congruence: Measurement, Development, and Relevance for Political Representation.” In How Democracy Works: Political Representation and Policy Congruence in Modern Societies : Essays in Honour of Jacques Thomassen, edited by Bas Denters, Martin Rosema, and Kees Aarts. Amsterdam University Press. Arceneaux, Kevin. 2012. “Cognitive Biases and the Strength of Political Arguments: Cognitive Biases and Argument Strength.” American Journal of Political Science 56 (2): 271–85. doi:10.1111/j.1540-5907.2011.00573.x. Arriola, L. R. 2009. “Patronage and Political Stability in Africa.” Comparative Political Studies 42 (10): 1339–62. Arriola, Leonardo R. 2012. Multiethnic Coalitions in Africa: Business Financing of Opposition Election Campaigns. New York: Cambridge University Press. Bakke, Elisabeth, and Nick Sitter. 2005. “Patterns of Stability Party Competition and Strategy in Central Europe since 1989.” Party Politics 11 (2): 243–63. Baldwin, Kate. 2011. “When Politicians Cede Control of Resources: Land, Chiefs and CoalitionBuilding in Africa.” Afrobarometer Working Paper Series No. 130. 121 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. 1998. “Electoral Continuity and Change, 1868 – 1996” 17 (3): 301–26. Bawn, Kathleen, Gary W. Cox, and Frances Rosenbluth. 1999. “Measuring the Ties That Bind: Electoral Cohesiveness in Four Democracies.” In Elections in Japan, Korea, and Taiwan under the Single Non-Transferable Vote: The Comparative Study of an Embedded Institution, edited by Bernard Norman Grofman, Sung-Chull Lee, Edwin Winckler, and Brian Woodall. Ann Arbor: University of Michigan Press. Beasley, Ryan K., and Mark R. Joslyn. 2001. “Cognitive Dissonance and Post-Decision Attitude Change in Six Presidential Elections.” Political Psychology 22 (3): 521–40. Beck, Linda J. 2008. Brokering Democracy in Africa: The Rise of Clientelist Democracy in Senegal. New York: Palgrave Macmillan. Belchior, Ana Maria. 2009. “Left-Right Party Congruence across Europe.” In XXI World Congress of Political Science, Santiago, Chile. Vol. 12. Bengtsson, Asa, and Hanna Wass. 2012. Congruence between MPs’, Non-Elected Candidates’ and Citizens’ Preferences for Representational Roles. SSRN Scholarly Paper ID 2108227. Rochester, NY: Social Science Research Network. http://papers.ssrn.com/abstract=2108227. Benoit, Kenneth, and Michael Laver. 2007. “Estimating Party Policy Positions: Comparing Expert Surveys and Hand-Coded Content Analysis.” Electoral Studies 26 (1): 90–107. doi:10.1016/j.electstud.2006.04.008. Benton, Allyson Lucinda. 2001. “Patronage Games: Economic Reform, Political Institutions, and the Decline of Party Stability in Latin America”. PhD Dissertation, University of California Los Angeles. http://catalog.hathitrust.org/Record/007253222. Best, Robin E., Ian Budge, and Michael D. Mcdonald. 2012. “Representation as a Median Mandate: Taking Cross-National Differences Seriously: Representation as a Median Mandate.” European Journal of Political Research 51 (1): 1–23. Binzer Hobolt, S., and R. Klemmensen. 2007. “Government Responsiveness and Political Competition in Comparative Perspective.” Comparative Political Studies 41 (3): 309–37. Blais, A., and M. A. Bodet. 2006. “Does Proportional Representation Foster Closer Congruence Between Citizens and Policy Makers?” Comparative Political Studies 39 (10): 1243–62. doi:10.1177/0010414005284374. Blais, André, and Ignacio Lago. 2009. “A General Measure of District Competitiveness.” Electoral Studies 28 (1): 94–100. doi:10.1016/j.electstud.2008.07.007. 122 Boas, Taylor C., and Amy Erica Smith. 2014. “Looks Like Me, Thinks Like Me? Descriptive Representation and Opinion Congruence in Brazil.” In Annual Meeting of the Midwest Political Science Association, Chicago, IL. Bochsler, Daniel. 2010. “Measuring Party Nationalisation: A New Gini-Based Indicator That Corrects for the Number of Units.” Electoral Studies 29 (1): 155–68. Bormann, Nils-Christian, and Matt Golder. 2013. “Democratic Electoral Systems around the World, 1946-2011.” Electoral Studies 32: 360–69. Boucek, F. 2009. “Rethinking Factionalism.” Party Politics 15 (4): 455. Boucek, Françoise. 2012. Factional Politics: How Dominant Parties Implode or Stabilize. New York, NY: Palgrave Macmillan. Brader, Ted, Nicholas A. Valentino, and Elizabeth Suhay. 2008. “What Triggers Public Opposition to Immigration? Anxiety, Group Cues, and Immigration Threat.” American Journal of Political Science 52 (4): 959–78. Brady, David W. 1985. “A Reevaluation of Realignments in American Politics: Evidence from the House of Representatives.” The American Political Science Review 79 (1): 28–49. doi:10.2307/1956117. Brancati, Dawn. 2003. “Design over Conflict: Managing Ethnic Conflict and Secessionism through Decentralization”. PhD Dissertation, Columbia University. ———. 2008. “The Origins and Strengths of Regional Parties.” British Journal of Political Science 38 (01). doi:10.1017/S0007123408000070. Bratton, Michael. 2007. “Formal versus Informal Institutions in Africa.” Journal of Democracy 18 (3): 96–110. Bridges, Amy. 1984. A City in the Republic: Antebellum New York and the Origins of Machine Politics. Cambridge U.K.; New York: Cambridge University Press. Brinks, Daniel M. 2003. “Informal Institutions and the Rule of Law: The Judicial Response to State Killings in Buenos Aires and São Paulo in the 1990s.” Comparative Politics 36 (1): 1–19. Budge, Ian. 1994. “A New Spatial Theory of Party Competition: Uncertainty, Ideology and Policy Equilibria Viewed Comparatively and Temporally.” British Journal of Political Science 24 (4): 443–67. Budge, Ian, Hans-Dieter Klingemann, Andrea Volkens, Judith Bara, and Eric Tanenbaum. 2001. Mapping Policy Preferences: Estimates for Parties, Electors, and Governments 1945-1998. Oxford: Oxford University Press. Caramani, Daniele. 2004. The Nationalization of Politics. Cambridge, UK ; New York: Cambridge 123 University Press. Carey, John M., and Matthew Soberg Shugart. 1995. “Incentives to Cultivate a Personal Vote: A Rank Ordering of Electoral Formulas.” Electoral Studies 14 (4): 417–39. Carr, Adam. 2013. “Psephos Election Archives.” Retrieved from http://psephos.adam-carr.net. Celis, R., A. Torres, J. M. Gatell, M. Almela, R. Rodrà Âguez-Roisin, and A. Agustà Â-Vidal. 1988. “Nosocomial Pneumonia. A Multivariate Analysis of Risk and Prognosis.” CHEST Journal 93 (2): 318–24. Chege, Michael. 2007. Political Parties in East Africa: Diversity in Political Party Systems. Stockholm, Sweden: International IDEA. Chhibber, P., and K. Kollman. 1998. “Party Aggregation and the Number of Parties in India and the United States.” American Political Science Review 92 (2): 329–42. Chhibber, Pradeep, and Ken Kollman. 2009. “Party Aggregation and the Number of Parties in India and the United States.” American Political Science Review 92 (2): 329–42. Choi, Syngjoo, Douglas Gale, Shachar Kariv, and Thomas Palfrey. 2011. “Network Architecture, Salience and Coordination.” Games and Economic Behavior 73 (1): 76–90. doi:10.1016/j.geb.2011.01.001. Chubb, Judith. 1982. Patronage, Power and Poverty in Southern Italy: A Tale of Two Cities. Cambridge: Cambridge University Press. Claggett, William, William Flanigan, and Nancy Zingale. 1984. “Nationalization of the American Electorate.” American Political Science Review 78 (1): 77–91. Clark, T. N, and R. Inglehart. 1998. “The New Political Culture: Changing Dynamics of Support for the Welfare State and Other Policies in Postindustrial Societies.” In The New Political Culture, edited by T. N Clark and V. Hoffmann-Martinot, 9–72. , CO: Westview. Collyns, Dan. 2010. “Peru Sees Rise of Personal Politics.” BBC, October 27, sec. Latin America and Caribbean. Retrieved from http://www.bbc.co.uk/news/world-latin-america11636035#story_continues_2. Cornelius, Wayne A. 1977. “Leaders, Followers, and Official Patrons in Urban Mexico.” In Friends, Followers, and Factions: A Reader in Political Clientelism, edited by Steffen W Schmidt, 337–53. Berkeley: University of California Press. Corral Gonzalez, Margarita. 2013. “Uneven Representation? Analysis of Democratic Responsiveness in Latin America”. PhD Dissertation, Vanderbilt University. Cox, Gary W. 1990. “Centripetal and Centrifugal Incentives in Electoral Systems.” American Journal of Political Science 34 (4): 903–35. doi:10.2307/2111465. 124 ———. 1997. Making Votes Count: Strategic Coordination in the World’s Electoral Systems. Cambridge University Press. ———. 1999a. “Electoral Rules and Electoral Coordination.” Annual Review of Political Science 2 (1): 145–61. doi:10.1146/annurev.polisci.2.1.145. ———. 1999b. “Electoral Rules and the Calculus of Mobilization.” Legislative Studies Quarterly 24 (3): 387. doi:10.2307/440350. ———. 2005. “Electoral Institutions and Political Competition Coordination, Persuasion and Mobilization.” In Handbook of New Institutional Economics, edited by Claude Menard and Mary M. Shirley, 2005 edition, 69–89. Dordrecht; Great Britain: Springer. ———. 2010. “Swing Voters, Core Voters, and Distributive Politics.” In Political Representation. Cambridge University Press. Cox, Gary W., and Mathew D. McCubbins. 1986. “Electoral Politics as a Redistributive Game.” The Journal of Politics 48 (02): 370–89. doi:10.2307/2131098. Cox, Gary W., and Mathew D. McCubbins. 1993. Legislative Leviathan: Party Government in the House. University of California Press. Crisp, Brian F., Santiago Olivella, and Joshua D. Potter. 2012a. “Electoral Contexts That Impede Voter Coordination.” Electoral Studies 31 (1): 143–58. ———. 2012b. “Party-System Nationalization and the Scope of Public Policy: The Importance of Cross-District Constituency Similarity.” Comparative Political Studies 46 (4): 431–56. Crisp, Brian F., Joshua D. Potter, and John J. W. Lee. 2012. “Entry and Coordination in MixedMember Systems: A Controlled Comparison Testing the Contamination Hypothesis.” Journal of Politics 74 (02): 571–83. doi:10.1017/S0022381611001769. Dalton, Russell J., David M. Farrell, and Ian McAllister. 2011. “The Dynamics of Political Representation.” How Democracy Works: Political Representation and Policy Congruence in Modern Societies, 21–38. Deininger, Klaus, and Lyn Squire. 1996. “A New Data Set Measuring Income Inequality.” The World Bank Economic Review 10 (3): 565–91. doi:10.1093/wber/10.3.565. Denny, Kevin, and Orla Doyle. 2009. “Does Voting History Matter? Analysing Persistence in Turnout.” American Journal of Political Science 53 (1): 17–35. Denzin, Norman K. 1970. The Research Act: A Theoretical Introduction to Sociological Methods. Chicago: Aldine Transaction. Desposato, Scott W. 2004. “The Impact of Federalism on National Party Cohesion in Brazil.” Legislative Studies Quarterly 29 (2): 259–85. doi:10.3162/036298004X201177. 125 Dinas, E., and K. Gemenis. 2010. “Measuring Parties’ Ideological Positions with Manifesto Data: A Critical Evaluation of the Competing Methods.” Party Politics 16 (4): 427–50. doi:10.1177/1354068809343107. Dinas, Elias. 2014. “Does Choice Bring Loyalty? Electoral Participation and the Development of Party Identification: DOES CHOICE BRING LOYALTY?” American Journal of Political Science 58 (2): 449–65. doi:10.1111/ajps.12044. Dixit, Avinash, and John Londregan. 1996. “The Determinants of Success of Special Interests in Redistributive Politics.” The Journal of Politics 58 (4): 1132–55. doi:10.2307/2960152. Dobson, Douglas, and Douglas St. Angelo. 1975. “Party Identification and the Floating Vote: Some Dynamics.” The American Political Science Review 69 (2): 481–90. doi:10.2307/1959081. Downs, Anthony. 1957. An Economic Theory of Democracy. New York: Harper and Row. Druckman, James N., and Arthur Lupia. 2000. “Preference Formation.” Annual Review of Political Science 3 (1): 1–24. Egel, Daniel. 2009. “Tribal Diversity, Political Patronage and the Yemeni Decentralization Experiment”. University of California, Berkeley. Eisenstadt, S. N., and Luis Roniger. 1984. Patrons, Clients and Friends: Interpersonal Relations and the Structure of Trust in Society. New York: Cambridge University Press. Electoral Institute for Sustainable Democracy in Africa. 2013. “Country Profiles.” http://www.eisa.org.za/WEP/countryindex.htm. Elster, Jon. 1985. Sour Grapes: Studies in the Subversion of Rationality. New York: Cambridge University Press. Erikson, Robert S, Michael MacKuen, and James A Stimson. 2002. The Macro Polity. New York: Cambridge University Press. España-Nájera, Annabella, and Mar Martínez Rosón. 2010. Political Representation in Central America. LASA. Ferrara, Federico. 2004. “Electoral Coordination and the Strategic Desertion of Strong Parties in Compensatory Mixed Systems with Negative Vote Transfers.” Electoral Studies 23 (3): 391–413. doi:10.1016/S0261-3794(03)00028-3. Ferree, E. Karen, G. Bingham Powell, and Ethan Scheiner. 2013. How Context Shapes the Effects of Electoral Rules. Task Force on Political Science, Electoral Rules, And Democratic Governance. Washington, D.C.: American Political Science Association. Festinger, Leon. 1957. A Theory of Cognitive Dissonance. Stanford, Calif.: Stanford University Press. 126 Freidenberg, Flavia, and Steven Levitsky. 2006. “Informal Institutions and Party Organization in Latin America.” In Informal Institutions and Democracy: Lessons from Latin America, edited by Gretchen Helmke and Steven Levitsky, 178–200. Baltimore: Johns Hopkins University Press. Fritz, Matthew S., and David P. Mackinnon. 2007. “Required Sample Size to Detect the Mediated Effect.” Psychological Science 18 (3): 233–39. Funston, N. John. 2000. Election Fervour: Political Contest in Thailand and Malaysia. Institute of Southeast Asian Studies. Gantner, A. 2008. “Bargaining, Search, and Outside Options.” Games and Economic Behavior 62 (2): 417–35. Golder, Matt, and Gabriella Lloyd. 2014. “Re-Evaluating the Relationship between Electoral Rules and Ideological Congruence.” European Journal of Political Research 53 (1): 200–212. Golder, Matt, and Jacek Stramski. 2010. “Ideological Congruence and Electoral Institutions.” American Journal of Political Science 54 (1): 90–106. Grossman, Gene M., and Elhanan Helpman. 2001. Special Interest Politics. Cambridge, Mass: The MIT Press. Grzymala-Busse, A. 2006. “The Discreet Charm of Formal Institutions: Postcommunist Party Competition and State Oversight.” Comparative Political Studies 39 (3): 271–300. doi:10.1177/0010414005284216. Grzymala-Busse, Anna. 2004. “Informal Institutions and the Post-Communist State.” The National Council for Eurasian and East European Research, Title VIII Programme. http://www.ucis.pitt.edu/nceeer/2004_818-02f_Grzymala-Busse.pdf. ———. 2008. “Beyond Clientelism: Incumbent State Capture and State Formation.” Comparative Political Studies 41 (4-5): 638–73. Harbers, I. 2009. “Decentralization and the Development of Nationalized Party Systems in New Democracies: Evidence From Latin America.” Comparative Political Studies 43 (5): 606– 27. doi:10.1177/0010414008330285. Harmel, R., and J. D Robertson. 1985. “Formation and Success of New Parties.” International Political Science Review 6 (4): 501–23. Harmel, Robert, and Kenneth Janda. 1982. Parties and Their Environments: Limits to Reform? Longman. Hayes, Andrew F. 2013. Introduction to Mediation, Moderation, and Conditional Process Analysis: A Regression-Based Approach. New York: The Guilford Press. Hellwig, Timothy, and David Samuels. 2008. “Electoral Accountability and the Variety of 127 Democratic Regimes.” British Journal of Political Science 38 (01): 65–90. Helmke, Gretchen, and Steven Levitsky. 2004. “Informal Institutions and Comparative Politics: A Research Agenda.” Perspectives on Politics 2 (4): 725–40. ———. , eds. 2006. Informal Institutions and Democracy: Lessons from Latin America. Baltimore: Johns Hopkins University Press. Hicken, A. 2011. “Clientelism.” Annual Review of Political Science 14 (1): 289–310. Hicken, A., and Martinez Kuhonta. 2011. “Shadows From the Past: Party System Institutionalization in Asia.” Comparative Political Studies 44 (5): 572–97. doi:10.1177/0010414010396460. Hicken, a., and H. Stoll. 2011. “Presidents and Parties: How Presidential Elections Shape Coordination in Legislative Elections.” Comparative Political Studies 44 (7): 854–83. doi:10.1177/0010414011401231. Hicken, Allen. 2002. “Parties, Pork and Policy: Policymaking in Developing Democracies”. PhD Dissertation, University of California San Diego. ———. 2009. Building Party Systems in Developing Democracies. 1 edition. Cambridge ; New York: Cambridge University Press. Hicken, Allen, Ken Kollman, and Joel Simmons. “Party System Nationalization and the Provision of Public Health Services.” Hilgers, Tina. 2011. “Clientelism and Conceptual Stretching: Differentiating among Concepts and among Analytical Levels.” Theory and Society 40 (5): 567–88. Hinich, Melvin J., and Michael C. Munger. 1996. Ideology and the Theory of Political Choice. University of Michigan Press. Hirschman, Albert O. 1971. A Bias for Hope: Essays on Development and Latin America. New Haven: Yale University Press. Holland, Paul W. 1988. “Causal Inference, Path Analysis and Recursive Structural Equations Models.” ETS Research Report Series 1988 (1): i–50. Hooghe, Marc, Bart Maddens, and Jo Noppe. 2006. “Why Parties Adapt: Electoral Reform, Party Finance and Party Strategy in Belgium.” Electoral Studies 25 (2): 351–68. doi:10.1016/j.electstud.2005.05.006. Hopkin, Jonathan. 2006. “Conceptualizing Political Clientelism: Political Exchange and Democratic Theory.” Paper presented at American Political Science Association Annual meeting, Philadelphia, August 31 – September 3, 2006. Hopmann, P. T. 1995. “Two Paradigms of Negotiation: Bargaining and Problem Solving.” The 128 Annals of the American Academy of Political and Social Science 542 (1): 24–47. Howell, Susan E. 1980. “The Behavioral Component of Changing Partisanship.” American Politics Research 8 (3): 279–302. doi:10.1177/1532673X8000800302. Huber, John D., and G. Bingham Powell. 1994. “Congruence between Citizens and Policymakers in Two Visions of Liberal Democracy.” World Politics 46 (03): 291–326. Hug, Simon. 2001. Altering Party Systems Strategic Behavior and the Emergence of New Political Parties in Western Democracies. Ann Arbor: University of Michigan Press. Huntington, Samuel P. 1968. Political Order in Changing Societies. New Haven: Yale University Press. ———. 1991. The Third Wave: Democratization in the Late 20th Century. Norman: University of Oklahoma Press. Iacobucci, Dawn, Neela Saldanha, and Xiaoyan Deng. 2007. “A Meditation on Mediation: Evidence That Structural Equations Models Perform Better than Regressions.” Journal of Consumer Psychology 17 (2): 139–53. Inglehart, Ronald. 1990. Culture Shift in Advanced Industrial Society. Princeton, N.J: Princeton University Press. Inoguchi, Takashi, Akihiko Tanaka, Shigeto Sonoda, and Timur Dadabaev. 2005. AsiaBarometer Survey Data. AsiaBarometer Project. www.asiabarometer.org. International Foundation for Electoral Systems. 2013. “ElectionGuide.” http://electionguide.org. Jennings, M. Kent, and Gregory B. Markus. 1984. “Partisan Orientations over the Long Haul: Results from the Three-Wave Political Socialization Panel Study.” The American Political Science Review 78 (4): 1000–1018. doi:10.2307/1955804. Johnson, Gbemende, Bruce I. Oppenheimer, and Jennifer L. Selin. 2012. “The House as a Stepping Stone to the Senate: Why Do So Few African American House Members Run?: the house as a stepping stone to the senate.” American Journal of Political Science 56 (2): 387–99. Johnson, Joel W., and Jessica S. Wallack. 2006. “Electoral Systems and the Personal Vote [Dataset].” http://hdl.handle.net/1902.1/17901. Jones, M. P., and S. Mainwaring. 2003. “The Nationalization of Parties and Party Systems: An Empirical Measure and an Application to the Americas.” Party Politics 9 (2): 139–66. doi:10.1177/13540688030092002. Kasuya, Yuko, and Johannes Moenius. 2008. “The Nationalization of Party Systems: Conceptual Issues and Alternative District-Focused Measures.” Electoral Studies 27 (1): 126–35. Kato, J. 1998. “When the Party Breaks up: Exit and Voice among Japanese Legislators.” American 129 Political Science Review 92 (4): 857–70. Katz, R. S. 2001. “The Problem of Candidate Selection and Models of Party Democracy.” Party Politics 7 (3): 277–96. Katz, R. S., and P. Mair. 1995. “Changing Models of Party Organization and Party Democracy: The Emergence of the Cartel Party.” Party Politics 1 (1): 5–28. Kawato, Sadafumi. 1987. “Nationalization and Partisan Realignment in Congressional Elections.” American Political Science Review 81 (4): 1235–50. Kazi, Z B. 2011. “Defection Politics in India.” International Referred Research Journal 3 (27): 15–16. Kemahlioglu, Ozge. 2011. “Jobs in Politicians’ Backyards: Party Leadership Competition and Patronage.” Journal of Theoretical Politics 23 (4): 480–509. Kemahlioglu, Ozge. 2005. “Intra-Party Competition and Clientelism : An Analysis of Public Sector Employment in Argentina and Turkey.” Paper presented at American Political Science Association Annual meeting, September 1 - September 4, 2005. Kenny, Michael. 1960. “Patterns of Patronage in Spain.” Anthropological Quarterly 33 (1): 14– 23. doi:10.2307/3316536. Kim, Heemin, and Richard C. Fording. 1998. “Voter Ideology in Western Democracies, 1946– 1989.” European Journal of Political Research 33 (1): 73–97. ———. 2003. “Voter Ideology in Western Democracies: An Update.” European Journal of Political Research 42 (1): 95–105. doi:10.1111/1475-6765.00076. Kim, HeeMin, and Richard C. Fording. 2005. “Party Manifesto Data and Measures of Ideology in Western Democracies.” Florida State University and University of Kentucky. http://garnet.acns.fsu.edu/~hkim/Ideology_Measures_Paper-Final.pdf. Kim, HeeMin, G. Bingham Powell Jr, and Richard C. Fording. 2010. “Electoral Systems, Party Systems, and Ideological Representation: An Analysis of Distortion in Western Democracies.” Comparative Politics 42 (2): 167–85. Kitschelt, Herbert. 2000. “Linkages between Citizens and Politicians in Democratic Polities.” Comparative Political Studies 33 (6-7): 845–79. doi:10.1177/001041400003300607. ———. 2007. “Party Systems.” In The Oxford Handbook of Comparative Politics, edited by Carles Boix and Susan C. Stokes. Oxford ; New York: Oxford University Press. ———. 2011a. “Clientelistic Linkage Strategies. A Descriptive Exploration.” In Workshop on Democratic Accountability Strategies. Duke University. ———. 2011b. Democratic Accountability and Linkages Project (version July 20, 2011). Durham, 130 NC: Duke University. ———. 2013. Democratic Accountability and Linkages Project. Durham, NC: Duke University. Kitschelt, Herbert, and Kent Freeze. 2011. “Programmatic Party System Structuration : Developing and Comparing Cross-National and Cross-Party Measures with a New Global Data Set.” Kitschelt, Herbert, Kent Freeze, Kiril Kolev, and Yi-Tang Wang. 2009. “Measuring Democratic Accountability: An Initial Report on an Emerging Data Set.” Revista Ciencia Política 29 (3): 741–73. Kitschelt, Herbert, and Daniel M. Kselman. 2010. “The Organizational Foundations of Democratic Accountability: Organizational Form and the Choice of Electoral Linkage Strategy.” Washington, D.C. Kitschelt, Herbert, and Steven I. Wilkinson, eds. 2007. Patrons, Clients and Policies: Patterns of Democratic Accountability and Political Competition. Cambridge University Press. Knight, Jack, and Itai Sened, eds. 1995. Explaining Social Institutions. Ann Arbor: University of Michigan Press. Knoke, David. 1976. Change and Continuity in American Politics: The Social Bases of Political Parties. Johns Hopkins University Press. Krishna, Anirudh. 2007. “Politics in the Middle: Mediating Relationships between the Citizens and the State in Rural North India.” In Patrons, Clients, and Policies: Patterns of Democratic Accountability and Political Competition, edited by Herbert Kitschelt and Steven I. Wilkinson, 141–58. Cambridge, UK: Cambridge University Press. Kuenzi, Michelle, and Gina Lambright. 2001. “Party System Institutionalization in 30 African Countries.” Party Politics 7 (4): 437 –468. doi:10.1177/1354068801007004003. Lago, Ignacio, and F. Martínez. 2011. “Why New Parties?” Party Politics 17 (1): 3–20. Lago, Ignacio, and José Ramón Montero. 2009. “Coordination between Electoral Arenas in Multilevel Countries.” European Journal of Political Research 48 (2): 176–203. doi:10.1111/j.1475-6765.2008.00832.x. ———. 2013. “Defining and Measuring Party System Nationalization.” European Political Science Review, no. March (March): 1–21. doi:10.1017/S1755773913000027. Langston, Joy. 2003. “The Formal Bases of Informal Power: Mexico.” University of Notre Dame: Kellogg Institute for International Studies. Laver, Michael. 1999. “Divided Parties, Divided Government.” Legislative Studies Quarterly 24 (1): 5–29. doi:10.2307/440298. 131 Levitsky, Steven. 2003. Transforming Labor-Based Parties in Latin America: Argentine Peronism in Comparative Perspective. Cambridge, U.K. ; New York: Cambridge University Press. Levitsky, Steven, and Maxwell A. Cameron. 2003. “Democracy without Parties? Political Parties and Regime Change in Fujimori’s Peru.” Latin American Politics and Society 45 (3): 1– 33. doi:10.1111/j.1548-2456.2003.tb00248.x. Levitsky, Steven, and Lucan Way. 2010. Competitive Authoritarianism: Hybrid Regimes after the Cold War. New York: Cambridge University Press. Lijphart, Arend. 1999. Patterns of Democracy: Government Forms and Performance in Thirty-Six Countries. New Haven: Yale University Press. Lindbeck, Assar, and Jörgen W. Weibull. 1987. “Balanced-Budget Redistribution as the Outcome of Political Competition.” Public Choice 52 (3): 273–97. Lindberg, Staffan I. 2009. “Accountability: The Core Concept and Its Subtypes.” Africa Power and Politics Programme Working Paper 1. Linz, Juan J. 1990. “The Perils of Presidentialism.” Journal of Democracy 1 (1): 51–69. Linz, Juan J., and Alfred Stepan. 1996. Problems of Democratic Transition and Consolidation: Southern Europe, South America, and Post-Communist Europe. Baltimore: Johns Hopkins University Press. Lipset, Seymour Martin, and Stein Rokkan. 1967. Party Systems and Voter Alignments: CrossNational Perspectives. Free Press. Luna, Juan, and Elizabeth Zechmeister. 2005. “Political Representation in Latin America: A Study of Elite-Mass Congruence in Nine Countries.” Comparative Political Studies 38 (4): 388– 416. doi:10.1177/0010414004273205. Lupu, Noam. 2008. “Nationalization and Party Institutionalization in Twentieth-Century Argentina.” Retrieved from http://www.noamlupu.com/nationalization.pdf. ———. 2012. “The 2011 General Elections in Peru.” Electoral Studies 31 (3): 621–24. doi:10.1016/j.electstud.2012.03.004. Lust, Ellen. 2009. “Competitive Clientelism in the Middle East.” Journal of Democracy 20 (3): 122–35. MacKinnon, David P. 2008. Introduction to Statistical Mediation Analysis. 1 Pap/Cdr edition. New York: Routledge. MacKinnon, David P., Amanda J. Fairchild, and Matthew S. Fritz. 2007. “Mediation Analysis.” Annual Review of Psychology 58 (1): 593–614. MacKinnon, David P., Jennifer L. Krull, and Chondra M. Lockwood. 2000. “Equivalence of the 132 Mediation, Confounding and Suppression Effect.” Prevention Science : The Official Journal of the Society for Prevention Research 1 (4): 173. Mahoney, James, and Kathleen Ann Thelen. 2010. “A Theory of Gradual Institutional Change.” In A Theory of Gradual Institutional Change, edited by James Mahoney and Kathleen Ann Thelen. Cambridge; New York: Cambridge University Press. Mainwaring, S., and E. Zoco. 2007. “Political Sequences and the Stabilization of Interparty Competition: Electoral Volatility in Old and New Democracies.” Party Politics 13 (2): 155–78. doi:10.1177/1354068807073852. Mainwaring, Scott. 1999. Rethinking Party Systems in the Third Wave of Democratization: The Case of Brazil. Stanford, CA: Stanford University Press. ———. 2003. “The Political Recrafting of Social Bases of Party Competition: Chile, 1973–95”, 55–84. doi:10.1017/S000712340300003. Mainwaring, Scott, Ana Maria Bejarano, and Eduardo Pizarro Leongomez, eds. 2006. The Crisis of Democratic Representation in the Andes. 1 edition. Stanford, CA: Stanford University Press. Mainwaring, Scott, and Timothy Scully. 1995a. “Introduction: Party Systems in Latin America.” In Building Democratic Institutions: Party Systems in Latin America. Stanford, CA: Stanford University Press. ———. , eds. 1995b. Building Democratic Institutions: Party Systems in Latin America. Reprint edition. Stanford, CA: Stanford University Press. Mainwaring, Scott, and Mariano Torcal. 2006. “Party System Institutionalization and Party System Theory After the Third Wave of Democratization.” In Handbook of Party Politics, 204– 27. London: SAGE Publications Ltd. Mainwaring, Scott, Arturo Valenzuela, and Michael Coppedge, eds. 1999. “The Evolution of Latin American Party Systems.” In Politics, Society, And Democracy Latin America, 171–206. Boulder, Colo.: Westview Press. Mair, Peter. 1999. “New Political Parties in Long-Established Party Systems: How Successful Are They?” In Elites, Parties and Democracy: Festschrift for Professor Mogens N. Pedersen, edited by Erik Beukel and Poul Erik Mouritzen, 207–24. Odense: Odense University press. Mazzoleni, M. 2009. “The Saliency of Regionalization in Party Systems: A Comparative Analysis of Regional Decentralization in Party Manifestos.” Party Politics 15 (2): 199–218. doi:10.1177/1354068808099981. McDonald, Michael D., and Ian Budge. 2005. Elections, Parties, Democracy: Conferring the Median Mandate. Oxford: Oxford University Press. McDonald, Michael D., Silvia M. Mendes, and Ian Budge. 2004. “What Are Elections For? 133 Conferring the Median Mandate.” British Journal of Political Science 34 (01): 1. doi:10.1017/S0007123403000322. Medina, Luis Fernando, and Susan Stokes. 2007. “Monopoly and Monitoring: An Approach to Political Clientelism.” In Patrons, Clients and Policies: Patterns of Democratic Accountability and Political Competition, edited by Herbert Kitschelt and Steven Wilkinson. Cambridge University Press. Miguel, Carolina De. 2013. “Democratization and Party System Nationalization : Electoral Incentives and Party Discipline”, no. April: 1–30. Miguel, Carolina G De. 2011. “The Geography of Economic Inequality, Institutions and Party System Territorialization”, 1–59. Miller, Warren E., and Donald E. Stokes. 1963. “Constituency Influence in Congress.” American Political Science Review 57 (01): 45–56. doi:10.2307/1952717. Moenius, J. 2004. “Measuring Party Linkage across Districts: Some Party System Inflation Indices and Their Properties.” Party Politics 10 (5): 543–64. Morgenstern, S. 2005. “Are Politics Local?: An Analysis of Voting Patterns in 23 Democracies.” Comparative Political Studies 38 (2): 143–70. doi:10.1177/0010414004271081. Morgenstern, Scott, John Polga-Hecimovich, and Peter M. Siavelis. 2014. “Seven Imperatives for Improving the Measurement of Party Nationalization with Evidence from Chile.” Electoral Studies 33: 186–99. Morgenstern, Scott, Stephen M. Swindle, and Andrea Castagnola. 2009. “Party Nationalization and Institutions.” The Journal of Politics 71 (04): 1322–1322. Morris, Stephen D. 1995. Political Reformism in Mexico: An Overview of Contemporary Mexican Politics. Boulder: Lynne Rienner. Mullainathan, Sendhil, and Ebonya Washington. 2009. “Sticking with Your Vote: Cognitive Dissonance and Political Attitudes.” American Economic Journal: Applied Economics 1 (1): 86–111. doi:10.1257/app.1.1.86. Muno, Wolfgang. 2010. “Conceptualizing and Measuring Clientelism Conceptualizing and Measuring Clientelism.” In Neopatrimonialism in Various World Regions, GIGA German Institute of Global and Area Studies. Hamburg. Munoz Chirinos, Paula. 2013. “Campaign Clientelism in Peru : An Informational Theory”. PhD Dissertation, University of Texas at Austin. Nash Jr, John F. 1950. “The Bargaining Problem.” Econometrica 18 (2): 155–62. Nelson, Thomas E., Rosalee A. Clawson, and Zoe M. Oxley. 1997. “Media Framing of a Civil Liberties Conflict and Its Effect on Tolerance.” American Political Science Review, 567– 134 83. Network of Democracy Research Institutes (NDRI). 2010. Political Clientelism, Social Policy, and the Quality of Democracy: Evidence from Latin America, Lessons from Other Regions. Quito, Ecuador. Nichter, S. 2008. “Vote Buying or Turnout Buying? Machine Politics and the Secret Ballot.” American Political Science Review 102 (01): 19–31. Nichter, Simeon. 2014. “Conceptualizing Vote Buying.” Electoral Studies 35 (September): 315– 27. doi:10.1016/j.electstud.2014.02.008. Norden, D. L. 1998. “Party Relations and Democracy in Latin America.” Party Politics 4 (4): 423– 43. doi:10.1177/1354068898004004002. Norris, Pippa. 2014. “Democracy Time-Series Dataset, Release 3.0.” Accessed May 30. http://www.pippanorris.com. North, Douglass C. 1990. Institutions, Institutional Change and Economic Performance. Cambridge ; New York: Cambridge University Press. Okun, Arthur. 1975. Equality and Efficiency: The Big Tradeoff. Washington D.C.: Brookings Institution Press. http://www.brookings.edu/research/books/1975/equalityandefficiency. Otero, Patricia, and Juan Rodriquez-Zepeda. 2010. Measuring Representation in Latin America: A Study of the Ideological Congruence between Parties and Voters. SSRN Scholarly Paper ID 1642532. Rochester, NY: Social Science Research Network. http://papers.ssrn.com/abstract=1642532. Panebianco, Angelo. 1988. Political Parties: Organization and Power. Translated by Marc Silver. 1 edition. Cambridge, England ; New York: Cambridge University Press. Pedersen, Morgens. 1983. “Changing Patterns of Electoral Volatility in European Party Systems, 1948-1977.” In Western European Party Systems: Continuity and Change, edited by Hans Daalder and Peter Mair, 29–66. Beverly Hills, CA: SAGE Publications. Pellicer, Miquel, and Eva Wegner. “Southern Africa Labour and Development Research Unit Electoral Rules and Clientelistic Parties : A Regression Discontinuity Approach”, no. 76. Piattoni, S. 2001. Clientelism, Interests, and Democratic Representation: The European Experience in Historical and Comparative Perspective. Cambridge University Press. Pitkin, Hanna Fenichel. 1967. The Concept of Representation. University of California Press. Powell, G. Bingham, Jr. 2000. Elections as Instruments of Democracy: Majoritarian and Proportional Visions. New Haven, CT: Yale University Press. ———. 2004. “Political Representation in Comparative Politics.” Annual Review of Political 135 Science 7 (1): 273–96. ———. 2006. “Election Laws and Representative Governments: Beyond Votes and Seats.” British Journal of Political Science 36 (02): 291. doi:10.1017/S0007123406000160. ———. 2009. “The Ideological Congruence Controversy: The Impact of Alternative Measures, Data, and Time Periods on the Effects of Election Rules.” Comparative Political Studies 42 (12): 1475–97. doi:10.1177/0010414009332147. ———. 2013. “Representation in Context: Election Laws and Ideological Congruence Between Citizens and Governments.” Perspectives on Politics 11 (01): 9–21. Powell, G. Bingham, Jr., and Georg S. Vanberg. 2000. “Election Laws, Disproportionality and Median Correspondence: Implications for Two Visions of Democracy.” British Journal of Political Science 30 (03): 383–411. Powers, Denise V., and James H. Cox. 1997. “Echoes from the Past: The Relationship between Satisfaction with Economic Reforms and Voting Behavior in Poland.” American Political Science Review, 617–33. Preacher, Kristopher J., and Andrew F. Hayes. 2008. “Asymptotic and Resampling Strategies for Assessing and Comparing Indirect Effects in Multiple Mediator Models.” Behavior Research Methods 40 (3): 879–91. Przeworski, Adam. 1975. “Institutionalization of Voting Patterns, or Is Mobilization the Source of Decay?” The American Political Science Review 69 (1): 49–67. doi:10.2307/1957884. Rahat, G., and R. Y. Hazan. 2001. “Candidate Selection Methods: An Analytical Framework.” Party Politics 7 (3): 297–322. Randall, V., and L. Svasand. 2002. “Party Institutionalization in New Democracies.” Party Politics 8 (1): 5–29. doi:10.1177/1354068802008001001. Reel, Monte. 2006. “Cash Aid Program Bolsters Lula’s Reelection Prospects.” Washington Post, October 29, sec. World. Retried from http://www.washingtonpost.com/wpdyn/content/article/2006/10/28/AR2006102800823.html?nav=rss_world/southamerica. 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, 575–90. Robinson, James A, and Thierry Verdier. 2013. “The Political Economy of Clientelism.” Scandinavian Journal of Economics 115 (2): 260–91. Rodden, Jonathan, and Erik Wibbels. 2002. “Beyond the Fiction of Federalism: Macroeconomic Management in Multitiered Systems.” World Politics 54 (04): 494–531. Rohrschneider, Robert, and Stephen Whitefield. 2007. “Representation in New Democracies: 136 Party Stances on European Integration in Post-Communist Eastern Europe.” Journal of Politics 69 (4): 1133–46. Roniger, Luis. 2004. “Political Clientelism, Democracy, and Market Economy.” Comparative Politics 36 (3): 353–75. Rubinstein, Ariel. 1982. “Perfect Equilibrium in a Bargaining Model.” Econometrica 50 (1): 97– 109. Rushing, William. 1976. “Profit and Nonprofit Orientations and the Differentiations-Coordination Hypothesis for Organizations : A Study of Small General Hospitals.” American Sociological Review 41 (4): 676–91. Ruth, Saskia Pauline. 2014. “Clientelism, Policy Responsiveness, and Inequality.” Accessed November 28. http://paperroom.ipsa.org/papers/paper_26083.pdf. Samuels, Dj. 1999. “Incentives to Cultivate a Party Vote in Candidate-Centric Electoral Systems - Evidence from Brazil.” Comparative Political Studies 32 (4): 487–518. doi:10.1177/0010414099032004004. Sandbrook, Richard. 1972. “Patrons, Clients, and Factions: New Dimensions of Conflict Analysis in Africa.” Canadian Journal of Political Science/Revue Canadienne de Science Politique, 104–19. Scarrow, Susan. 2005. Political Parties and Democracy In Theoretical And Practical Perspectives. Washington, D.C.: National Democratic Institute for International Affairs. Schakel, Arjan H. 2013. “Nationalisation of Multilevel Party Systems: A Conceptual and Empirical Analysis.” European Journal of Political Research 52 (2): 212–36. doi:10.1111/j.1475-6765.2012.02067.x. Schattschneider, E. E. 1942. Party Government. New York: Rinehart and Company, Inc. Schedler, A. 2002. “The Nested Game of Democratization by Elections.” International Political Science Review 23 (1): 103–22. Scheiner, Ethan. 2005. “Pipelines of Pork Japanese Politics and a Model of Local Opposition Party Failure.” Comparative Political Studies 38 (7): 799–823. Scott, James C. 1972. “Patron-Client Politics and Political Change in Southeast Asia.” American Political Science Review, 91–113. Scott, W. Richard. 2004. “Reflections on a Half-Century of Organizational Sociology.” Annual Review of Sociology 30 (1): 1–21. doi:10.1146/annurev.soc.30.012703.110644. Shaw, Daron. 2008. “Swing Voting and U.S. Presidential Elections.” In The Swing Voter in American Politics, edited by William G. Mayer, 75– 101. Washington, D.C.: Brookings Institution Press. 137 Sheingate, Adam D. 2003. “Political Entrepreneurship, Institutional Change, and American Political Development.” Studies in American Political Development 17 (02): 185–203. Shugart, Matthew S., and Scott Mainwaring. 1997. “Juan Linz, Presidentialism, and Democracy: A Critical Appraisal.” Comparative Politics, 449–71. Singer, Matthew M., and Laura B. Stephenson. 2009. “The Political Context and Duverger’s Theory: Evidence at the District Level.” Electoral Studies 28 (3): 480–91. doi:10.1016/j.electstud.2009.06.002. Širinić, Daniela. 2014. “Inequality in Political Representation and Electoral Systems.” Paper presented at Midwest Political Science Association Annual Meeting, Chicago, IL. Slater, Dan. 2006. “The Ironies of Instability In Indonesia.” Social Analysis 50 (1): 208–13. Sobel, Michael. 1982. “Asymptotic Confidence Intervals for Indirect Effects in Structural Equation Models.” Sociological Methodology 13: 290–312. Sobel, Michael E. 1990. “Effect Analysis and Causation in Linear Structural Equation Models.” Psychometrika 55 (3): 495–515. doi:10.1007/BF02294763. Stepan, Alfred, and Cindy Skach. 1993. “Constitutional Frameworks and Democratic Consolidation: Parliamentarianism versus Presidentialism.” World Politics 46 (01): 1–22. Stokes, Donald E. 1967. “Parties and the Nationalization of Electoral Forces.” In The American Party System: Stages of Political Development, edited by William N. Chambers and Walter D. Burnham, 1 edition, 182–202. New York: Oxford University Press. Stokes, Susan. 2005. “Perverse Accountability: A Formal Model of Machine Politics with Evidence from Argentina.” American Political Science Review 99 (3): 315–25. ———. 2007. “Political Clientelism.” In The Oxford Handbook of Comparative Politics, edited by Charles Boix and Susan Stokes, 604–27. New York: Oxford University Press. Stokes, Susan C. 2001. Mandates and Democracy: Neoliberalism by Surprise in Latin America. Cambridge ; New York: Cambridge University Press. Stoll, Heather. 2004. “Social Cleavages, Political Institutions, and Party Systems”. PhD Dissertation, Stanford University. ———. 2010. “Elite-Level Conflict Salience and Dimensionality in Western Europe: Concepts and Empirical Findings.” West European Politics 33 (3): 445–73. Szczerbiak, A. 2001. “Party Structure and Organizational Development in Post-Communist Poland.” Journal of Communist Studies and Transition Politics 17 (2): 94–130. Taagepera, Professor Rein, and Matthew Soberg Shugart. 1989. Seats and Votes: The Effects and Determinants of Electoral Systems. New Haven: Yale University Press. 138 Tavits, Margit. 2008a. “On the Linkage between Electoral Volatility and Party System Instability in Central and Eastern Europe.” European Journal of Political Research 47 (5): 537–55. doi:10.1111/j.1475-6765.2008.00782.x. ———. 2008b. “Party Systems in the Making: The Emergence and Success of New Parties in New Democracies.” British Journal of Political Science 38 (1): 113–33. ———. 2012. “Organizing for Success: Party Organizational Strength and Electoral Performance in Postcommunist Europe.” The Journal of Politics 74 (01): 83–97. Taylor-Robinson, Michelle. 2006. “The Difficult Road from Caudillismo to Democracy, or Can Clientelism Compliment Democratic Electoral Institutions? Exploration of the Honduran Case.” In Informal Institutions and Democracy in Latin America, edited by Gretchen Helmke and Steven Levitsky, 106–24. Baltimore, MD: Johns Hopkins University Press. Teorell, Jan, Nicholas Charron, Stefan Dahlberg, Sören Holmberg, Bo Rothstein, Petrus Sundin, and Richard Svensson. 2013. The Quality of Government Dataset (version 15May13). University of Gothenburg: The Quality of Government Institute. www.qog.pol.gu.se. The AmericasBarometer. 2005. The Latin American Public www.LapopSurveys.org. Opinion Project (LAPOP). The Comparative Study of Electoral Systems, Module 3 Full Release [dataset]. 2013. www.cses.org. Thelen, Kathleen. 2004. How Institutions Evolve: The Political Economy of Skills in Germany, Britain, the United States, and Japan. Cambridge University Press. Thomassen, Jacques. 1999. “Political Communication between Political Elites and Mass Publics: The Role of Belief Systems.” In Policy Representation in Western Democracies, edited by Warren Miller, Sören Holmberg, and Roy Pierce, 33–58. Oxford: Oxford University Press. Thorlakson, L. 2009. “Patterns of Party Integration, Influence and Autonomy in Seven Federations.” Party Politics 15 (2): 157–77. doi:10.1177/1354068808099979. Transparency International. 2012. Corruption Perception http://www.transparency.org/research/cpi/overview. Index. Retried from Treisman, Daniel. 2000a. “Decentralization and Inflation: Commitment, Collective Action, or Continuity.” American Political Science Review, 837–57. ———. 2000b. “Decentralization and the Quality of Government.” Manuscript UCLA. http://www.sscnet.ucla.edu/polisci/faculty/treisman/Papers/DecandGovt.pdf. Tsai, C.-H. 2005. “Policy-Making, Local Factions and Candidate Coordination in Single NonTransferable Voting: A Case Study of Taiwan.” Party Politics 11 (1): 59–77. doi:10.1177/1354068805048473. 139 Tsebelis, G. 1990. Nested Games: Rational Choice in Comparative Politics. University of California Press. Van Biezen, I. 2000. “On the Internal Balance of Party Power: Party Organizations in New Democracies.” Party Politics 6 (4): 395–417. doi:10.1177/1354068800006004001. Van de Walle, N. 2007. “The Path from Neopatrimonialism: Democracy and Clientelism in Africa Today.” Cornell University, Center for International Studies, Working Paper. Van Eerd, Jonathan. 2011. “Elite-Mass Congruence in African Dominant Party Systems: The Cases of Botswana and Lesotho.” In IPSA-ECPR Joint Conference, University of Sao Paulo. Retried from http://paperroom.ipsa.org/papers/paper_26094.pdf. Vanhanen, Tatu. 2003. Democratization: A Comparative Analysis of 170 Countries. London: Routledge. Voeten, Erik. 2001. “Outside Options and the Logic of Security Council Action.” American Political Science Review 95 (4): 845–58. Warner, C. M. 1997a. “Political Parties and the Opportunity Costs of Patronage.” Party Politics 3 (4): 533–48. doi:10.1177/1354068897003004005. ———. 1997b. “Political Parties and the Opportunity Costs of Patronage.” Party Politics 3 (4): 533–48. doi:10.1177/1354068897003004005. Webb, Eugene J, Donald T. Campbell, Richard D. Schwartz, and Lee Sechrest. 1966. Unobtrusive Measures: Nonreactive Research in the Social Sciences. Chicago: Rand McNally. Weingrod, Alex. 1968. “Patrons, Patronage, and Political Parties.” Comparative Studies in Society and History 10 (4): 377–400. ———. 1977. “Patrons, Patronage, and Political Parties.” In Friends, Followers, and Factions: A Reader in Political Clientelism, edited by Steffen W Schmidt, 323–36. Berkeley: University of California Press. Weinstein, J. M. 2005. “Resources and the Information Problem in Rebel Recruitment.” Journal of Conflict Resolution 49 (4): 598–624. Wessels, Bernhard. 1999. “System Characteristics Matter. Empirical Evidence from Ten Representation Studies.” In Policy Representation in Western Democracies, edited by Warren Miller, Sören Holmberg, and Roy Pierce. Oxford: Oxford University Press. Whitefield, S. 2006. “Mind the Representation Gap: Explaining Differences in Public Views of Representation in Postcommunist Democracies.” Comparative Political Studies 39 (6): 733–58. doi:10.1177/0010414005277221. Williamson, Oliver E. 2000. “The New Institutional Economics: Taking Stock, Looking Ahead.” Journal of Economic Literature, 595–613. 140 Wolf, Eric. 2001. “Aspects of Group Relationships in a Complex Society: Mexico.” In Contemporary Cultures and Societies of Latin America: A Reader in the Social Anthropology of Middle and South America, edited by Dwight B. Heath and Richard Adams. New York: Random House. World Value Survey (version Wave 5 2005-2008). 2005. Madrid Spain: Aggregate File Producer: Asep/JDS, World Values Survey Association. Worsnop, Alec. 2013. “Not All Fragmentation Is Equal : Insurgent Organizational Structure and Control of Collective Violence.” In International Studies Association. San Francisco, CA. Wu, Jaushieh Joseph. 1995. Taiwan’s Democratization: Forces behind the New Momentum. Hong Kong; New York: Oxford University Press. Young, Daniel. 2009. “Is Clientelism At Work In African Elections? A Study of Voting Behavior In Kenya And Zambia.” Afrobarometer Working Paper Series No. 106. Zaller, John R. 1992. The Nature and Origins of Mass Opinion. Cambridge England ; New York, NY, USA: Cambridge University Press. Zhao, Xinshu, John G. Lynch Jr., and Qimei Chen. 2010. “Reconsidering Baron and Kenny: Myths and Truths about Mediation Analysis.” Journal of Consumer Research 37 (2): 197–206. doi:10.1086/651257. 141