Democratization and the Emergence of Responsive Party Systems in Latin America

Democratization and the Emergence of Responsive Party Systems in Latin America A Research Project Simon Bornschier University of St. Gallen/ University of Zurich Prepared for the Workshop The Formation of Ideologically Based Party Systems IPSA-ECPR Joint Conference, São Paulo, February 16 to 19, 2011

1. Summary of the Research Project The point of departure of this project is the failure of the major theories of democratization to explain differences in the consolidation of democracy across Latin America. I argue that an important determinant for the successful democratization is the emergence of a party system that is responsive to the citizenry. Because many parties in Latin America and elsewhere use clientelistic appeals to mobilize voters, they fail to represent the programmatic preferences of their electorate, which would be one of democracy s most central goals according to democratic theory. Following recent theories of accountability (Kitschelt 2000, Kitschelt and Wilkinson 2007), programmatic, clientelistic, and charismatic linkages between citizens and policymakers compete with one another in mobilizing voters, and the prevalence of programmatic linkages is a necessary condition for responsive party systems. In this research project, I therefore study under which conditions programmatic linkages replace clientelistic linkages. The project combines comparative historical analysis and quantitative statistical methods to empirically analyze this process. I postulate two routes to programmatic party competition. One is historical and took place during the first wave of democratization in the early 20 th century, similar to the formation of ideological cleavages in Western Europe. The other is more recent and depends on the presence of parties that actively seek to overcome clientelistic patterns of mobilization. I study the first route by adopting a Rokkanian perspective that focuses on critical junctures and historical legacies that set countries apart (Lipset and Rokkan 1967, Rokkan 1999, Collier and Collier 1991). In some countries in Latin America, such as in Chile and Uruguay, strong ideological cleavages emerged during the early steps towards democracy, and this resulted in an early prevalence of programmatic party competition. In other countries, a different historical sequencing allowed the established political elites to maintain clientelistic modes of mobilization. As a result, party systems may remain unresponsive to the demands of the citizenry for decades. The first aim of the project is thus to develop a cleavage account of party system formation that contrasts the development in twelve Latin American countries with that of Western Europe and North America. In order to apply the cleavage approach to contexts outside the old democracies, it must, however, be adapted. In particular, I will extend 1

the approach by integrating the deliberate attempts of political elites to prevent sociostructural conflicts from manifesting themselves politically. The insights generated from this historical analysis are then used to derive predictions concerning contemporary patterns of interest representation across the continent. My hypothesis is that the way conflicts were mobilized early on affects the long-term balance between clientelistic and programmatic mobilization strategies employed by parties. These predictions are tested in a quantitative analysis of contemporary linkage practices in twelve Latin American countries. For this analysis, I develop a new analytical model to identify and measure different types of linkages between parties and voters. I then analyze these modes of party mobilization and the quality of representation by combining data on the programmatic position of parties with data on the preferences of voters, measuring the congruence between parties and voters at various points in time between the mid-1990s and the 2000s. In addition, I draw on innovative data from an expert survey to assess the linkages employed by parties to mobilize voters. In this quantitative analysis of contemporary linkage practices, I also address the second, alternative route to programmatic party competition. This route is open even to those countries that lack the favourable historical circumstances of the forerunners in terms of democratic accountability. In line with recent research pointing to the role of agency in cleavage formation, the early historical experience cannot be expected to fully determine contemporary patterns of party competition. Consequently, I study the extent to which clientelistic linkages are successfully being crowded out in a gradual process that started with the most recent wave of democratization in Latin America in the 1980s. Indeed, in a number of countries such as Brazil, new parties of the left have appeared, which explicitly aim at changing the dominant ways of voter mobilization. There is some evidence that in refusing to make clientelistic appeals, parties opting for programmatic linkages can force the other, established parties to adopt more clear-cut programmatic profiles as well. In other countries, recent movements of the left centred on charismatic personalities have succeeded in breaking into party systems where historically rooted parties failed to respond to the preferences of voters. Venezuela is a case in point. I hypothesize that these two forms of left-wing mobilization have diverging consequences for the emergence of responsive party systems, and consequently for the quality of democracy. While programs play the dominant role in the first case, they are amalgamated with 2

charismatic linkages and selective incentives in the latter case. Presumably, this undermines the institutionalization of a party system and the establishment of firm mechanisms of accountability. This project will make an important contribution to the study of the role of party systems in democratization by innovatively linking historical and quantitative approaches. On the one hand, it engages in comparative historical analysis to develop a cleavage account of party system formation in a context where it has not been widely used. On the other hand, it combines mass-level surveys and expert survey data on party positions and linkage practices to assess the limits of path dependency and the possibilities of political agency in shaping political representation. 2. Detailed Presentation of the Project 2.1 State of research in the field 2.1.1 The failure to explain differences in democratic consolidation The first euphoria over the unprecedented diffusion of democratic rule around the world in the Third wave of democratization has given way to more gloomy assessments of the prospects for the consolidation of democracy in many of these countries. Starting with O Donnell s (1994) famous article on delegative democracy, the democratic quality of many of the new regimes was questioned. This has resulted in a proliferation democracies with adjectives or defects of democratic regimes that are used to characterize regimes lying in the grey zone between democracy and authoritarianism (e.g., Collier and Levitsky 1997, Diamond 2002, Merkel et al. 2003, 2006, Merkel 2004). In particular, this literature testifies large problems regarding the rule of law, horizontal or inter-institutional accountability, and in the responsiveness of party systems to voter preferences. Most of the standard approaches to democratization fail to adequately explain the differences between countries in their progress in democratic consolidation. The process by which authoritarian rule succumbed to democracy, emphasized in the transition approach 3

(O Donnell and Schmitter 1986, Linz and Stepan 1996), was found insufficient to explain why many democracies failed to consolidate after having completed their transition to electoral regimes (O Donnell 1996). The modernization approach (Lipset 1957), on the other hand, fared equally badly in predicting democracy on this continent (Mainwaring and Pérez- Liñán 2003, 2005). Even Inglehart and Welzel s (2005) re-formulation of modernization theory, which introduces political culture as the hitherto missing link between economic development and democratic institutions, proves unable to explain the stark contrasts within Latin America. Figure 1 reproduces the impressive relationship found by the authors between values favouring self-expression and the level of what they call effective democracy. At the same time, the highlighting of Latin American countries, reveals that fairly similar levels of commitment to democracy in the populace can produce strongly divergent levels of democracy. Figure 1: Relationship between self-expression values and effective democracy Source: Re-analysis of data from Inglehart and Welzel (2005: Figure 7.2) The third major school of democratization, the structural or macro-historical sociology approach (Rueschemeyer et al. 1992, Collier and Collier 1991), proves more fruitful. 4

Rueschemeyer et al. (1992) show that differences in societal power relations since the end of colonial rule in Latin America impinged on the viability of democracy. What is striking about their results is that those countries possessing societal structures favourable to early democratization are also those that display the highest quality of democracy according to Freedom House (www.freedomhouse.org) or as measured by Inglehart and Welzel (2005), as Figure 1 indicates. Hence, a strong contrast persists between Chile and Uruguay on the one hand, and countries such as Venezuela, Brazil, Argentina, and Peru, on the other hand. But how can these long-standing differences be explained? And do these findings imply a structural determinism that is impossible to overcome? Recent research on political parties has come to recognize the neglected role of parties in the process of democratization. 2.1.2 Political parties and democratic linkage As the actors linking citizens and the political system, parties play a central role in guaranteeing democratic governance. In a path-breaking book, Mainwaring and Scully (1995) have argued that democratization entails not only the building of formal democratic institutions, but also of party systems that represent the interests of voters in the political process. 1 The most important contribution of these authors has been to reveal vast differences in party system institutionalization, and thus in the degree to which they constitute veritable systems in Sartori s (1976) terms. The stability of the basic patterns of opposition or conflict impinges heavily on whether or not party systems structure the expectations of political actors and introduce predictability in politics. This, on the other hand, is considered a central prerequisite of democratic accountability and of the congruence between citizens and their representatives (Mainwaring and Scully 1995, Mair 1997, 2001, Tóka 1998, Mainwaring and Torcal 2006). Furthermore, strong parties that are representative of society can be hypothesized to be one of the most important ingredients of successful power sharing, which, in turn, has been shown to be a central determinant of democratization (Linder and Bächtiger 2005). While not necessarily being a remedy for all defects of new democracies, then, the representative function of party systems makes them a key element of democracy (Pitkin 1967, Dahl 1971, Powell 2000). Empirically, Luna and Zechmeister (2005) show that party system institutionalization is an important predictor of the congruence of citizen preferences and party positions in Latin America. 1 In an earlier analysis, Dix (1992) had shown large differences in the degree of institutionalization of individual parties in Latin America. 5

Party systems, once institutionalized, seem to remain remarkably stable, and the pathdependent nature of their evolution is able to account for the continuing differences in the quality of democracy across Latin America. In Europe, the conjuncture of large-scale processes of social change and extensions of the suffrage resulted in the early emergence of cross-local functional conflicts that displaced the clientelistic and local politics prevalent before (Caramani 2004). Latin American trajectories have proven much more varied. While some party systems, such as the Brazilian one, have emerged more or less from scratch after every disruption of democratic rule, the party systems in Colombia, Uruguay and Chile still carry the imprint of the conflicts prevalent in the early decades of the 20 th century, when democracy was first established (Dix 1989, González 1995, Scully 1995, Coppedge 1998, Mainwaring 1999). In these cases, strong linkages between parties and voters seem to have frozen party systems into place much like their European counterparts. Mainwaring and Zoco (2007) have recently shown that lasting differences between party systems worldwide result from specific historical sequences, highlighting that there is nothing automatic in the stabilization of interparty competition over time. The degree to which party systems have stable roots in society and become institutionalized thus seems to be a crucial aspect of the more general path dependency of democratic trajectories. One of the reasons why party systems may not institutionalize is the prevalence of clientelistic practices. While clientelistic forms of political mobilization have been brought in conjunction with the specific social structure of Latin American societies early on (Chalmers 1977), the study of clientelism has been reinvigorated by some recent theorizing by Kitschelt (2000) and Kitschelt and Wilkinson (2007). From this perspective, programmatic linkages between citizens and policymakers compete with clientelistic and charismatic linkages in mobilizing voters. While providing valuable theoretical tools and data that will be used in this research project, the more recent contributions to the concept of political linkage have not yet found satisfactory answers regarding the central question of when programmatic linkages are capable of ousting clientelism. Kitschelt and Wilkinson (2007) argue convincingly that at a certain point of development, clientelism simply becomes too expensive to be worthwhile for politicians. At the same time, the level of development would have led us to expect programmatic parties long ago in many parts of Latin America, as Dix (1989) has stated. The imperfect relationship between the correlates of modernization and the level of democracy across Latin America put in evidence in Figure 1 underscores the presence of other factors hindering the establishment of full democratic accountability. 6

2.1.3 Historical party system formation and citizen-party linkages Taken together, what we know from the earlier approaches to clientelism concurs with theories of path dependency (Pierson 2000) in suggesting that the vertical networks typical of developing societies are not easily displaced by the horizontal type of political organization that characterizes programmatic party-voter linkages. To the degree that the established parties are able to prevent new competitors from entering the system, clientelistic practices are likely to remain unaltered. As Shefter (1977, 1993) and Geddes (1994) have argued and empirically shown, in the absence of a universalistic state bureaucracy that precludes such practices a condition that does not apply in Latin America the established parties are able to secure their position by distributing particularistic benefits. Hagopian s (1996) case study of Brazil reveals that clientelism is an instrument of long-established political elites to hold on to their positions of power and privilege. Only externally mobilized parties, in Shefter s terminology, which do not have access to the ruling circles of power, push for programmatic competition because programs are all they have to offer. By the same token, it can be hypothesized that once ideological party competition has been established and is maintained, clientelistic appeals will not prove very successful. For voters that are sufficiently informed and are offered clear programmatic options, selling their vote for a particularistic benefit is unlikely to be an attractive option. Consequently, the initial emergence of a party system that is responsive to the preferences of the citizenry emerges as a decisive moment in the evolution of party systems. The first step in my approach to the research question at hand is thus to focus on this decisive moment. The Western European historical experience testifies that functional conflicts resulting from large-scale processes of nation state formation and industrializing are capable of forming party systems rooted in specific social groups and based on powerful ideologies (Lipset and Rokkan 1967, Rokkan 1999, Caramani 2004, Bartolini 2005). In Europe, this dynamic was assisted, however, by the temporal coincidence of societal processes with extensions of the suffrage. Where party systems are based on cleavages, clientelism plays a negligible role, at least when compared to party systems without clear roots in social structure. I apply a historical cleavage approach in order to explain the different kind of linkages parties forge to mobilize voters in Latin American party systems. Few attempts have been made to apply a cleavage framework outside Western Europe (Randall 2001), but where they have been undertaken, they demonstrate the applicability of the concept to Latin America 7

(Dix 1989, Coppedge 1998, Collier and Collier 1991). Nonetheless, the applications of the cleavage concept remain limited and have the shortcoming of not adequately addressing the trade-off between the different possibilities of linking citizens and their representatives. Furthermore, I intend to analyze explicitly to which degree the historical trajectory of Latin American countries corresponds to or differs from that of the Western European cases and the US. In the following, I discuss in detail how this analysis differs from these studies. Collier and Collier s (1991) study certainly represents the most ambitious and theoretically elaborate application of the concept of critical junctures, which they consider the essence of the cleavage approach, to regime dynamics to Latin America. However, their analysis does not explain the genesis of the party system as such, but focuses on the incorporation of the labour movement and its long-term impact on the party competition and the viability of democracy. My approach is both broader and more modest. On the one hand, I do not pretend to undertake a similarly detailed and comprehensive historical analysis of the incorporation of specific societal groups into politics. Instead, the analysis will be more concerned with highlighting the central differences the Western European and US-American path to programmatic party competition which is similar to that followed by certain Latin American countries, but not others. In those countries that followed different historical trajectories, I focus on the factors accounting for the absence of stable links between parties and voters in others and that allow clientelism to survive. Thus, I do not focus on the representation of specific interests such as those of workers, but on whether or not the party system reflects ideological divisions prevalent in society. While Dix (1989) and Coppedge (1998) have adopted a cleavage approach to explain party system formation in Latin America more generally, they do not take into account that parties frequently use clientelistic means of mobilization. As a consequence, parties may not only be unrepresentative of narrower interests defined by class and the like, but also lack the internal pluralism that has allowed parties in the US to perform representative functions in the absence of strong cleavages (cf. Epstein 1980). This results in two shortcomings. First, in those cases where parties have firm roots in society, we cannot be entirely sure whether programmatic, clientelistic, or charismatic linkages account for the anchorage. Mainwaring and Scully (1995) face the same difficulty of having to assume that the institutionalization of party systems reflects congruent representation between citizens and policymakers. However, the stability of a party system may also be an indication that strong parties are able to control and 8

administer clientelism, contrary to the individualistic practices prevalent in Brazil, for example. Consequently, the party system is firmly rooted in clientelistic networks, and not in ideological divisions. It also seems possible for charismatic personalities such as Peron in Argentina to forge stable citizen-party alignments that are only vaguely rooted in ideology (Madsen and Snow 1991). Furthermore, the historical analysis must be extended to the more proximate past since the initial institutionalization of a cleavage-based party system cannot in all cases be expected to preclude clientelistic practices emerging later on. With respect to recent transformations in Western European party systems, it has been shown that when the established conflicts are pacified, they no longer restrain the mobilization space for new conflicts (Kriesi and Duyvendak 1995, Bornschier 2010). In the Latin American context, where particularistic benefits are less costly and no universalistic public administration exists, reverting to clientelistic practices may be an alternative for established parties representing out-dated conflicts. The second shortcoming resulting from the neglect of parties linkage practices in existing applications of the cleavage approach to Latin America lies in its failure to illuminate the political dynamics in all those countries that lack strong ideological cleavages. In Coppedge s (1998) analysis, Brazil, Bolivia, Peru and Ecuador form a residual category, where party systems did not become institutionalized. The fundamental difference between these cases and Argentina and Venezuela is not all that clear, however. Furthermore, the residual category as such is unsatisfactory. In order to shed light on what unites these diverse cases, we need to determine the mechanisms that have prevented social structural conflicts to manifest themselves in politics in these countries. Is this due solely to political sequencing in liberalization and democratization (Dahl 1971), or do parties in these contexts simply integrate and reflect broad interests, as Dix (1989) assumes? Or did political elites take advantage of these historical factors in order to prevent the representation of interests that were harmful to their interests, as the case of Brazil suggests (Bornschier 2008)? Only a comparative analysis will be able to show the necessary conditions both for cleavages to manifest themselves, as well as to remain repressed. Besides developing a cleavage account of party system formation that contrasts the Latin American development with that of Western Europe and North America, this part of the study is fundamental to the second step I seek to undertake, namely, to explain contemporary patterns of interest representation. Thus, from the comparative historical analysis, I derive 9

predictions concerning linkage practices and the responsiveness of party systems in contemporary Latin America. This combination represents one of the major innovations of this approach. In a similar vein, but based on a more limited typology specifying the character of communist rule, Kitschelt et al. (1999) likewise have made predictions concerning the types of linkages that are likely to emerge in Central and Eastern Europe. 2.1.4 Contemporary linkage practices and political agency in cleavage formation In the contemporary period that is studied using quantitative data, I analyze the interaction of historically rooted party system features and political agency. Recently, a wide range of applications of the cleavage approach outside Western Europe have begun to put more emphasis on the role of political actors in shaping voter alignments. This line of research stresses that parties are capable of forging connections between social structure and party allegiances even in the absence of the large-scale transformations of social structure characteristic of the Western European path to cleavage formation (Enyedi 2005, Deegan- Krause 2006, Chhibber and Torcal 1997, Torcal and Mainwaring 2003). I adopt this perspective by focusing on the possible emergence of parties that are responsive to social groups in those countries that lacked clear cleavages until recently. Recent research on Brazil testifies that the Workers Party mobilizes an electorate whose common desire is to end the traditional politics of patronage and to establish responsible party government (Samuels 2006). Recent research indicates that the other Brazilian parties are becoming more programmatically oriented as well (Hunter 2007, Hagopian et al. 2009). I thus hypothesize that, despite a continuing practice of corruption under the new government, the Workers Party seems to exert a pressure on the established parties to develop a more clear-cut ideological profile. By focusing on quantitative data on party positions and voter orientations, it is possible to assess to which degree parties are responsive to the policy preferences of their voters. Luna and Zechmeister (2005) have presented one of the rare analyses of this kind for nine Latin American countries. Even if this approach does not explicitly assess whether party-voter linkages are programmatic or clientelistic, it does so indirectly because policy positions should play no role in clientelistic mobilization. A shortcoming of this analysis is that it relies on an a priori definition of the relevant dimensions of political conflict that are assumed to be 10

identical for all countries under study. An alternative, pursued by Moreno (1999), is to empirically determine the dimensions of conflict underlying voter orientations by use of factor analysis. In a second step, Moreno then investigates which of the conflict dimensions underlying voter preferences have actually been mobilized by parties by analyzing what distinguishes their electorates in ideological terms. Based exclusively on data from voters, this approach does not tell us whether the dimensions revealed are actually those mobilized by political parties and is limited by the items available items in the cross-national surveys employed. I address these shortcomings in two ways. First of all, we use the knowledge generated in the comparative historical analysis to assess the relevant dimensions of conflict in each of the twelve party systems. My measure of responsiveness thus focuses explicitly on those dimensions that we know to be relevant in the specific country. Secondly, I assess whether the party system still reflects these historical conflicts by empirically determining the dimensions underlying parties programmatic offer. The analysis thus moves beyond Moreno s (1999) account by integrating demand side and supply side data and differs from Luna and Zechmeister s (2005) approach by focusing on the historically and currently relevant dimensions of conflict in each party system. Furthermore, by analyzing at least three timepoints between the mid-1990s and 2008 in each country, I track the dynamic process of programmatic linkage formation in those countries where clientelism has recently come under pressure. As far as the factors are concerned that account for responsible party government, and preclude clientelistic practices, Luna and Zechmeister (2005) find empirical evidence that greater congruence between citizens and voters is achieved by party system that feature strong left-wing parties. This finding is in line with expectations, but needs to be qualified. In line with Castañeda (2006) and Castañeda and Morales (2008), I hypothesize the existence of two lefts in contemporary Latin America. Those left-wing parties that emerged in the struggle for democratization, such as the Workers Party in Brazil, the Frente Amplio in Uruguay, or the Socialist Party in Chile, establish durable links with society and crowd out clientelistic linkages where they existed. New parties of the left in Venezuela, Bolivia, and Ecuador, on the other hand, seek to establish charismatic linkages and also revert to clientelistic practices. Consequently, the implications for vertical and horizontal accountability of these two types of parties are quite different. 11

Finally, data that is currently collected in the Democractic Accountability and Linkages Project under the direction of Herbert Kitschelt at Duke University (http://www.duke.edu/web/democracy/) will allow me to assess the nature of citizen-party linkages directly. The expert survey covers 85 countries, among them the twelve countries that I plan to study. It features detailed information on party organization, campaign methods, the programmatic profile of parties, and the prevalence of selective incentives used by them in mobilizing voters. This will allow me to shed light on the nature of linkages in those contexts where we do not find parties to represent the preferences of their voters. Furthermore, it will be possible to integrate the information on linkage practices with that regarding party positions for the last time-point of the analysis of the congruence of voters and parties. I have been involved in the collection of data for the Swiss expert survey and will cooperate further with this project, participating in conferences that are planned among the project collaborators. 2.3 Detailed research plan 2.3.1 Objectives The project aims at enhancing our understanding of why party competition centres on ideological conflicts in some Latin American countries, and on particularistic benefits in others. It seeks to single out the impact of critical junctures and historical processes on the one hand, and of the strategies of political actors on the other hand in explaining different types of linkage between citizens and representatives in these countries. In order to attain these objectives, I seek to employ innovative methods and to fruitfully combine the historical and the statistical approaches. In the historical analysis, the project will adapt the cleavage approach and apply it to a region where it has so far not been widely used. By drawing explicit comparisons between the old democracies and twelve Latin American countries, the project contributes to an understanding of the similarities and the differences both between the two contexts, as well as within Latin America. In developing the approach, I aim to make it amenable to explaining not only which 12

types of conflict emerge in different countries, but also how specific historical trajectories lead to programmatic or clientelistic party competition. By confronting the comparative historical analysis with quantitative data from Latin America s most recent experience of re-democratization, this project studies the long-term impact of historical processes. By drawing on a variety of data sources, the project aims at contributing a more nuanced and theoretically elaborate explanation of the types of conflict that prevail in contemporary Latin America. Ultimately, the aim is to combine historical and contemporary factors into a common framework that contributes to an explanation of the vast differences in the quality of democracy not only between the old and more recent democracies, but also within Latin America (e.g., Diamond and Morlino 2005, Munck and Verkuilen 2002, Altman and Pérez-Liñán 2002, Merkel et al. 2003, 2006, Bühlmann et al. 2008). 2.3.2 Hypotheses regarding the relationship between historical conflicts, the responsiveness of party systems, and linkage practices My basic contention is that programmatic linkages and clientelistic and/or charismatic linkages crown one another out. Both historical factors, as well as the current behaviour of political parties impinge on the balance between these three types of linkage. Figure 2 tentatively sketches a model that will be used to study the congruence of representation. Furthermore, it relates the congruence of representation to linkage practices and to the impact of historical cleavage formation on these linkages. This grid of expectations thus integrates the historical and quantitative parts of the analysis. In studying the correspondence between party positions and voter orientations, it goes beyond prior research in not only assessing the responsiveness of the party system, but in also differentiating various types of representation that have varying consequences for clientelistic and populist modes of mobilization. Most centrally, it matters not only whether parties reflect voter positions, but also how polarizing and therefore salient a division is. 2 2 As more than one division is likely to be relevant in each country, this analytical schema is applied to each of them separately. For analytical purposes, we have dichotomized the two dimensions in Figure 2, but empirically, they create a two-dimensional space in which countries can be positioned at each point in time we study. 13

Figure 2: Responsiveness of party systems and linkage practices Does party system reflect the preferences of voters? Polarization of parties Yes match No mismatch high Responsive party system: SEGMENTED CLEAVAGE BASED ON STRONG PROGRAMMATIC LINKAGE Clientelism strongly constrained: Both parties and voters highly polarized and durably aligned along cleavage Unresponsive party system: OUT-DATED CLEAVAGE (REMINISCENT OF HISTORICAL PROGRAMMATIC LINKAGE) Strong party control of clientelism; medium-level constraints on new outsider parties employing charismatic forms of mobilization low Responsive party system: COMPETITIVE POLITICAL DIMENSION BASED ON PROGRAMMATIC LINKAGE Parties responsive to citizenry as they compete to represent position of median voter; clientelism constrained by practice of congruent representation Unresponsive party system: PERVASIVE CLIENTELISM AND/OR CHARISMATIC LINKAGES PREVALENT New conflicts may be mobilized by outsider parties that engage in: - Populist anti-establishment mobilization (charismatic linkages) - Long-term establishment of programmatic linkages based on advocacy of political accountability and/or concrete policy positions We find two basic situations in Figure 2, one in which the party system is responsive to the policy preferences of voters (cells to the left), and one where it is not (the two right-hand cells). In the top left cell, segmented cleavages feature firm programmatic linkages between parties and voters that are based on strongly contrasting ideologies. Such a situation reflects the historical mobilization of cleavages that have proven highly enduring. This is also a possible end state of the gradual formation of new cleavages that may be in the process of emerging since the late 1980s. On the other hand, when the party system is weakly polarized, but parties nonetheless reflect voter preferences, we find a mode of competition as described by theorists such as Schumpeter (1942) or Downs (1957): A competitive political dimension where parties compete for the median voter. While clientelism is also strongly constrained by the responsiveness of the party system to conflicts at the voter level, competitive political dimensions are unlikely to embody the same degree of collective identification of social 14

groups with political parties, and are therefore more easily pacified. Nonetheless, in a system where parties are accustomed to taking and representing the preferences of voters, clientelism is unlikely to re-emerge. I now turn to the two situations where the party system is unresponsive to the preferences of voters. In this case, clientelism is likely to play a major role in forming citizen-party alignments. Either parties represent an out-dated cleavage that has been pacified on the voter side, with parties reverting to clientelistic practices to hold on to voters. If parties that have resulted from the historical conflicts persist, they are likely to possess strong party organizations and control the distribution of clientelistic benefits. In this situation, clientelism freezes the party system in the medium term. As a result, the party system may appear institutionalized despite the absence of programmatic linkages. My approach thus overcomes Mainwaring and Scully s (1995) problematic assumption that clientelism is absent by definition in institutionalized party systems. Despite the firm hold of clientelism, however, popular discontent with the established parties may create a window of opportunity for populist anti-establishment mobilization against the existing cartel of parties. Venezuela can be hypothesizes to fall into this category before the rise of Chavez. The same lack of accountability results in those cases where cleavages were never able to materialize, which are situated in the bottom right cell. Here, parties have always relied primarily on particularistic benefits and pervasive clientelism, possibly combined with charismatic leadership, to secure votes. The party system is hardly anchored in society, and the weakness of party organizations makes clientelism work in a highly personalized fashion. As a result, either flash-parties or political outsiders with populist messages may emerge from scratch and attract considerable voter shares. Without parliamentary majorities, they are unlikely to be able to govern, however, resulting in what O Donnell (1994) has termed delegative democracies lacking horizontal accountability between institutions. Collor de Mellos victory in the 1991 Brazilian elections is a case in point. The other possibility is that new parties emerge that try to introduce substantive policy conflicts in order to change the dominant mode of politics. At first, such parties are likely to mobilize on the issue of accountability, as many of the former outsider parties do that pushed for re-democratization in the late 1970s and 1980s. The most obvious example would be the Workers Party in Brazil. In the long run, accountability divides are likely to be complemented or superseded by new political conflict such as an economic left-right or a value divide. Once programmatic 15

competition is established, however, clientelistic linkages are progressively displaced programmatic linkages throughout the country. Empirically applying the analytical schema set out in Figure 2 on the one hand serves to assess the impact of historical processes on contemporary patterns of linkage. On the other hand, it allows us to track the evolution of these patterns of opposition over the past two decades. Despite the path dependency constituted by historical party system formation, political actors have the capacity to forge new links with voters. Thus, the model is employed to assess the differing impact of the two lefts in Latin America on cleavage formation and party system institutionalization. 2.3.3 Selection of cases My objective is to include as many Latin American countries as possible with differing historical experiences in order to develop typologies of these trajectories as well as testable predictions for current party competition. Furthermore, the analysis should cover those countries commonly featured in comparisons of the continent (O Donnell, Schmitter, Whitehead 1986, Mainwaring and Scully 1995, Coppedge 1998, Dix 1989, Collier and Collier 1991, Collier 1999). This will allow me to follow up on and engage in discussion with prior research. On the other hand, the number of countries to be considered faces constraints in terms of the manageability of the project and data availability. In particular, a sufficient amount of secondary sources is needed for the historical part of the analysis and the countries must be covered by in the survey and expert data used in the quantitative analysis. For these reasons, I have decided to focus on twelve Latin American countries, omitting the Central American and Caribbean countries, which are generally analyzed on their own. However, I follow the common practice to include Costa Rica, one of the oldest democracies in the region, as well as Mexico. These twelve cases provide some important variation in their historical development. Chile, Uruguay, Costa Rica, and Colombia at least partially established a pluralist political order early on in the late 19 th and early 20 th centuries, and were able to develop institutionalized party systems when the suffrage was extended thereafter. The initial party system has remained largely intact in all of these cases but Uruguay, which saw a strong party of the left emerge as a counter-pole to the two long-established parties. The 16

Uruguayan case is a particularly interesting one for the study of path dependency since the liberals and conservatives to some degree reverted to clientelistic practices after the pacification of the original conflict that had led to their emergence. However, the Frente Amplio forced them back onto a programmatic track since the 1980s. Among those cases that lacked even a moderately pluralist order, an institutionalized party system nonetheless emerged in Argentina, Venezuela, Mexico, and Paraguay. From there, the trajectories of these countries have diverged, however. Mexico saw the erosion of its dominant party and the slow emergence of a pluralist order. In Venezuela, the pacification of the original conflict and the major parties reliance on clientelism and patronage led to the rather sudden collapse of the party system and to the rise of Chavez as an anti-establishment candidate. In Argentina, a division between Peronists and non-peronists has survived and continues to structure the party system despite widespread clientelistic practices. In Brazil, Bolivia, Ecuador and Peru, on the other hand, the party system did not become institutionalized in the early phases of democracy, and frequent interruptions of democratic rule, in conjuncture with a highly personalistic practice of clientelism, have also prevented later institutionalization. Nonetheless, these countries show some important variation in their recent development. In fact, as discussed above, Brazil represents a highly interesting case of a slow but gradual evolution towards greater institutionalization and a receding of clientelistic practices under the pressure of the Workers Party. In Bolivia and Ecuador, on the other hand, charismatic leadership has played a much more important role than grass-roots mobilization in the rise of left-wing movements in the past years. In the historical part of the analysis, the Latin American cases will be contrasted with the distinctive paths to establishing programmatic party competition in Western European and the North America. Programmatic linkages displaced clientelism in two different ways in the old democracies. The simultaneity of suffrage extensions and large-scale processes of societal change created cleavages that disrupted the vertical networks of clientelism throughout Western Europe, as well as in Chile. Compared to Latin America, the Western European path to responsive party system formation appears rather homogeneous. In North America, the gradual erosion of vertical networks in the modernization process, and the emergence of a pluralist pattern represent an alternative path away from clientelism (Chalmers 1977, Epstein 1980). Although the US and Latin America differed considerably in terms of inequality and the power of the upper classes in the relevant time periods, it has been argued that the North 17

American model of party system formation corresponds more closely to Latin American patterns (von Mettenheim 1995). North America is thus included in the historical analysis. In terms of the time period that is studied, the analysis starts with Latin America s national revolution constituted by the independence from colonial rule in the 19 th century. In this period, a common conflict between liberals and conservatives that mainly centred on issues of centralization, free trade, and the relationship between the state and the Church was prevalent throughout the continent. From this common starting point, the analysis then follows the differing trajectories up to the present. The beginning and the end points of the quantitative part of the analysis are determined largely by data availability. Since the aim is not only to test predictions developed in the historical analysis, but also to analyze a dynamic process in which new actors of the political left push the party system to more closely represent societal interests, as many points in time as possible are needed. Ideally, there would be data from the period of re-democratization in the 1980s or even earlier, but for lack comparative data, the analysis will start in the mid-1990s. I cover at least three points in time for which both data on party positions and voter preferences is available, the latest being 2008 (see next section). The quantitative study focuses solely on the twelve Latin American countries, since similar analyses for Western Europe have already been undertaken (Kriesi et al. 2008, Bornschier 2010). 2.3.4 Data and methods The first step in this project makes use of comparative history (Mahoney and Rueschemeyer 2003). I engage in a secondary analysis of existing literature in history and the social sciences. Similarly to Collier and Collier (1991), I re-interpret this literature from the perspective of cleavage theory, but with special emphasis on those critical junctures that have put countries either on programmatic or clientelistic tracks, or that have assigned an important role to charismatic leadership. Contrary to many approaches that focus mainly on those social structural antagonisms that have left their imprint in party systems, I also address the question why some conflicts have not manifested themselves in politics. This requires the study of a substantial amount of literature and then synthesize historical trajectories into accounts of alternative paths to party system development. For the comparison with Western Europe and 18

the US, on the other hand, the project can rely on already highly synthesized works (Rokkan 1999, Mann 1986, 1993, Luebbert 1991, Caramani 2004, Bartolini 2005, Epstein 1980). In a second step, I draw on quantitative data from a number of sources to study parties political offer, as well as the political demand constituted by voter preferences. The combination of expert survey data and public opinion surveys allows us to position parties and voters on substantive policy dimensions. The analysis will begin by determining the relevant dimensions of competition in a party system. I proceed in two steps. For one thing, I operationalize the differing policy options that result from the dominant, historically formed dividing lines in each country and use confirmatory factor analysis to assess the degree to which these divisions remain salient. For another, I determine the main dimensions of party competition empirically using exploratory factor analysis or Multidimensional Scaling (MDS). I then follow the same procedure for the demand side. By reconstructing the dimensions found in the political offer I establish how closely the parties reflect the position of their electorate. This allows us to operationalize the analytical model set out in Figure 2 above. By focusing on various points in time between the 1990s and the 2000s, the analysis tracks the evolution of the patterns of competition prevalent in the party system in every country. To integrate data on parties political offer and voters demands, identical or roughly comparable questions in both types of surveys employed are necessary. Prior research shows that this is analytically challenging, but generally feasible (Kitschelt et al. 1999, Luna and Zechmeister 2005, Kriesi et al. 2008, Bornschier 2010). Table 1 gives an overview of the quantitative data available for Latin America. The main constraint is the limited availability of data on party positions. A first analysis limited to a subset of four countries can be carried out in 1995. Unfortunately, there is no adequate data for the full sample of twelve countries before 1997. Three further points are available for which it is possible to match data on the supply and demand sides of political competition, and I will analyze all four points in time. The temporal proximity of the 2005 survey of parliamentary elites and the 2006-2007 expert survey on party positions will allow us to compare the results derived from these two different sources. In order to include all twelve countries, I will rely on Latinobarómetro surveys. Because the quality of this data, mainly in terms of representativity, is not perfect, it will be complemented and validated with data from the World Values Survey for subsets of the 19

sample. These surveys satisfy highest standards and cover some of the most important countries studied. As the most recent expert survey that can be drawn on, data from the Democratic Accountability and Linkages Project allows me to combine the analysis of the responsiveness of the party system with assessments of the kind of linkage parties cultivate with their voters. In particular, this will provide valuable insights into those cases in Figure 2 where parties fail to adequately reflect the preferences of voters. The detailed information on the way parties mobilize will help establish whether clientelism indeed plays the dominant role in these contexts, as the model assumes, or whether there are different reasons for the misrepresentation of voters. A first rival explanation would be that parties lack information on voter preferences and for this reason present an inadequate offer. A different hypothesis is that voters are unable to discern party positions due to a lack of political sophistication, despite parties attempts at developing clear-cut profiles. The data also make it possible to distinguish the two basic types of clientelistic exchange: Party-controlled strategies and individualized practices centred on individual representatives. In a final step, I assess the impact of different modes of representation on the quality of democracy in Latin America. Here, we draw on various sources such as the annual surveys administered by Freedom House (www.freedomhouse.org) and the Polity Project (http://www.systemicpeace.org/polity/polity4.htm). Furthermore, data combining quantitative and qualitative assessments of various dimensions of democratic performance in Latin America is currently being collected by a team directed by Leonardo Morlino at the Istituto Italiano di Scienze Umane in Florence (Morlino n.d.). 20