Grassroots Democracy? Movements and Party Organization in Bolivia

Transcription

1 Grassroots Democracy? Movements and Party Organization in Bolivia Santiago Anria University of North Carolina at Chapel Hill Department of Political Science Hamilton Hall, CB 3265 Chapel Hill, NC Phone: (919) DRAFT VERSION: PLEASE DO NOT CITE! Abstract: The rise to power of movement-based parties is a new phenomenon that is clearly spreading. Political scientists expect these parties to become increasingly oligarchic as they govern nationally. The Bolivian MAS deviates from the conventional wisdom on this type of parties, as it has followed a sui generis organizational trajectory that has facilitated grassroots impact and constrained elite control. Using a within-case research design and original data collected through fieldwork, this paper explains conditions and mechanisms facilitating this outcome in the crucial area of candidate selection. Although institutional elements are important, the key to understanding how these parties operate lies within the organizational context in which they are embedded. Where civil society is strong and has mechanisms to arrive at collective decisions, it can play an important role in resisting the oligarchization of allied parties. Movement-based parties operate differently in different localities depending on the structuration of the political space.

2 Political parties have undergone profound changes in recent years. Increasingly, as mass partisanship seems to be a relic of the past (Levitsky & Cameron, 2003), parties founded to sustain the support of a single charismatic leader have become common prominent examples include the United Socialist Party of Venezuela under Chávez (Ellner, 2013), and the Italian Forza Italia and the Popolo Della Libertá under Berlusconi (McDonnell, 2013). In these parties, the locus of organizational power is squarely at the top. By contrast, other parties, like the Green movement parties in Western Europe, reject personalism in the interest of boosting participation and resisting oligarchic pressures. However, success in achieving grassroots party democracy, particularly after assuming governmental responsibilities, has generally proven to be elusive (Jachnow, 2013). New parties have been particularly important in Latin America (Levitsky, Loxton, & Van Dyck, n.d). One of the most salient developments in the region is the recent emergence and access to power of left parties representing the interests of those on the political margins (Cameron & Hershberg, 2010; Levitsky & Roberts, 2011). Usually described as movementbased parties (Madrid, 2012), they draw their organizational strength from connections to grassroots social movements, like the Brazilian PT (Workers Party), the Uruguayan FA (Broad Front), the Salvadorian FMLN (Farabundo Martí National Liberation Front), the Venezuelan LCR (La Causa R), or the Bolivian MAS (Movement Toward Socialism). Movement-based parties are not just a Latin American phenomenon; they are also emerging in Africa (LeBas, 2011), Western Europe (Kalyvas, 1996), and Eastern Europe (Glenn, 2003). Despite their increasing importance, we know little about how these parties work. Research has tended to focus on the origins of movement-based parties (Kitschelt, 1989a; Keck, 1992; Bruhn 1997; 2

3 Chandra, 2004; Van Cott, 2005), but their internal politics remain both under-examined and under-theorized. In Latin America, the ascension of movement-based parties to national-level power generated some optimism about the prospects for building internally democratic organizations that encourage wide and substantive participation of organized civil society in the making of collective decisions (Van Cott, 2005; Handlin & Collier, 2011; Madrid, 2012; De la Torre & Arnson, 2013). Extending direct grassroots participation is a historic goal of the left in Latin America, and it has been associated with the post-cold War notion of deepening democracy (Roberts, 1998, p. 3; Goldfrank, 2011, p. 162). Scholars such as Levitsky and Roberts (2011, p. 13) and Pribble (2013) have shown that new left parties and political movements in power in Latin America vary in the extent to which their internal structures disperse authority, but a more fundamental question still remains unanswered: What are the sources of variation between and within parties? This question has a long tradition in the study of political parties. It goes as far back as Michels s oligarchy theory (1962[1911]), which, in its short version, predicts the inevitable rise of elite-dominated hierarchical structures that concentrate power and de-emphasize bottom-up participation in procedures of social choice. 1 Michels s theory is either the argument influential analyses explicitly confront, or the conclusion at which such studies generally arrive. 2 Framed as a fundamental sociological law of political parties, it denies the very possibility of democratic modes of governance within parties, particularly as they contest elections and access power. Although the idea of internal party democracy is the subject of an age-old debate, it regained increasing attention in the comparative study of political parties (Hazan & Rahat, 2010; Cross & Katz, 2013; Cross & Pilet, 2014), in response to the almost-universal crises of representation and 3

4 the decline of mass party membership. Under what conditions can parties escape what Michels believed was their inevitable oligarchic fate? This question is important for both substantive and theoretical reasons. Substantively, understanding the internal life of an increasingly common phenomenon like movement-based parties, I argue, adds to one of the most strikingly under-developed literatures in comparative politics: the debate about what happens inside the black box of parties (Levitsky, 2001). Even though parties remain weakly organized in much of the developing world (Cyr, 2012), the era of party building is far from over, and movement-based parties are well-equipped to build strong organizations, all else equal (Levitsky, Loxton, & Van Dyck, n.d.). Moreover, differences in the nature and organization of these parties affect important public policy outcomes (Pribble, 2013; Huber & Stephens, 2012). For example, the politics of redistribution is better understood by looking inside movement-based parties, since more internally democratic parties generally lead to more redistributive reforms. Though research on new parties is growing, most of it focuses on the conditions giving rise to stable parties (Lupu, 2013; Van Dyck, 2013; Tavits, 2013). Whether and how parties can escape oligarchy, particularly when they are in power, is a theoretically important question for understanding political representation, participation, and related phenomena (Hazan & Rahat, 2010). When democracy within governing parties is deficient, the voices of citizens or even the party s own constituencies may not be heard, hindering their participation in political life. When governing parties are internally democratic, by contrast, they may generate opportunities and incentives for empowering social groups and individuals that have traditionally been at the margins of political life. 3 Arguments about party democracy are, in effect, arguments about democratizing democracy (Santos, 2005). The goal here is not to refute Michels s iron law 4

5 of oligarchy, nor to prove that it does not apply everywhere. It is rather to explain with original, systematic evidence the conditions and mechanisms under which broader and more substantive grassroots participation can be promoted in contemporary governing parties. 4 This article addresses this puzzle by using a within-case research design with theorytesting and -building purposes. Specifically, it explains variation in patterns of candidate selection within the case of the Bolivian MAS. The value of this approach is twofold. First, the selection of candidates for electoral office is a central aspect of internal party democracy and political interest aggregation, and we know little about how movement-based parties operate in this realm. 5 Second, MAS is one of the most successful movement-based parties in power in Latin America, and yet the case deviates from the conventional wisdom on this type of party: defying theoretical expectations, it has followed a sui generis organizational trajectory that has facilitated grassroots impact and constrained elite control, even after assuming power at the national level. Conventional accounts of these parties focus on cases that develop a strong organizational infrastructure of collective action before assuming national power, such as the Brazilian PT (Samuels, 2004; Hunter, 2010; Ribeiro, 2014). Consequently, their conclusions tend to emphasize the normalization of these parties, and the difficulties of sustaining bottom-up participation when they govern at the national level. 6 As a deviant case, studying MAS can serve to advance and refine theory by explaining party-movement relationships that can empower the grassroots and, in so doing, lead to the incorporation of popular sectors in the political arena. 7 Concretely, this article identifies conditions and mechanisms that provide opportunities for broad and substantive grassroots impact on candidate selection procedures: the presence of strong grassroots organizations that can agree on selection. When such organizations are absent or do not agree on selection, elite decision-making is more likely to occur. In other words, where 5

6 civil society is strong and has mechanisms to arrive at collective decisions, it can play an important role in affecting internal party governance and resisting the oligarchization of allied parties. This paper contributes new theory and empirical findings to debates among scholars of parties and party strategy in Latin America and beyond. The political science literature on parties is dominated by rational choice approaches that see parties as unitary actors detached from their social bases. 8 Yet, we have learned from the more sociological and comparative historical approaches of Panebianco (1988), Kitschelt (1994), Gibson (1997), and Levitsky (2003) that we cannot think of parties simply as unitary actors dominated by a unified leadership. I present new data demonstrating that just as parties operate differently in different localities in terms of how they organize their electoral appeals to different social bases (Gibson, 1996; Kitschelt et al., 2010; Luna, 2014), they also work differently in the realm of candidate selection depending on the structuration of the political space in which they operate. Although there is a significant amount of new literature on party strategy, 9 the question of how different patterns of party-civil society relations matter for the internal politics of a party remains under-theorized. I show with original data that scholars of parties can benefit by asking how these relationships vary across geographical constituencies, and by studying the impact this has for party politics. I proceed as follows. First, I define what I mean by movement-based parties, and situate the argument in leading literatures. Second, I describe the empirical strategy and research design. Third, drawing on 170 in-depth interviews with key actors, I examine the mechanics of candidate selection within the Bolivian MAS, and explain how sources of within-party variation are related to different patterns of party-civil society relations. Fourth, I conclude with some thoughts on the analytical implications of escaping party oligarchy. 6

7 MOVEMENT-BASED PARTIES, EXPLAINING THEIR INTERNAL POLITICS This section does two things: first, it defines more precisely the concept of movementbased parties. Second, it develops a theory of within-party variation in candidate selection in which institutional and structural factors interact. Institutional elements comprise aspects of a country s electoral system that influence candidate selection by creating the space for the grass roots to shape the process, and elements related to the strength of the local party apparatus. Structural elements refer to elements associated with the strength of civil society. Strongly organized societies, as the empirical literature shows, serve as a potential power base for parties. They can play an important party-building role, and they can shape internal party governance. Movement-Based Parties Movement-based parties are parties whose core constituency is comprised of grassroots social movements. This definition parallels Levitsky s (2003, p. 4) definition of labor-based parties. The central difference between labor-based parties and movement-based parties lies in the nature of their core constituency. Movement-based parties are also different from Kitschelt s (2006) analytical characterization of movement parties. Movement parties are almost always the electoral vehicles of a social movement mobilized around a single issue (Kitschelt, 2006, p. 283). By contrast, movement-based parties are broader alliances of various movements and, as such, they are better prepared to incorporate a broader set of issues, actors, and demands. Such parties follow what Roberts (1998, p. 75) calls the organic model of party development, in the sense that they are organizationally hybrid: they engage in extra-institutional social mobilization, and at the same time compete for office. Members and leaders who run for electoral office, however, tend to be drawn directly from social movements rather than from the 7

8 ranks of a separate, professional political caste. While these parties vary in terms of ideology (Kitschelt, 2006, pp ), they almost always share a rejection of hierarchical control as well as an explicit commitment to maximizing democratic participation at the grassroots. Movement-based parties are often seen as transitional phenomena (Kitschelt, 2006, p. 288), but the into what question is not settled. One salient argument suggests that the more participatory and bottom-up decision-making patterns that are generally present at early stages in the life of a movement-based party are only viable for a short time, as the logic of electoral competition will inevitably push parties to subordinate their mobilizational strategies to the logic of political power. However, movement-based parties may not evolve in a unilinear way. It is also theoretically plausible that such parties follow contingent structural and strategic incentives that make the return to the organizational patterns common in the early phases possible. In short, there are no a priori reasons to presuppose that parties based on movements will transition to a form of party that is hierarchical, exclusive, and centralized. That these parties are moving targets offers unique opportunities to examine them as forms of political organization that may challenge conventional notions of how parties work internally. 10 Their genesis on grassroots mobilization and their hybrid nature may encourage democratic control from below by allowing for the existence of opposition among allied groups that check power from within and generate pressures from below in ways that constrain the decision-making power of the party leadership. The realization of this potential depends on the organizational strength and unity of allied movements. A key aspect of the trajectory of movement-based parties is candidate selection. How parties select candidates is critical for the dynamics of internal interest aggregation. Thus, in the rest of the paper I use candidate selection as a tool for understanding movement-based parties. 8

9 Institutional Elements My analysis will focus on two institutional elements electoral rules and the nature of the local party organization. Just as electoral systems affect party outcomes like discipline in Congress (Samuels, 1999), electoral systems also affect candidate selection (Siavelis and Morgenstern, 2008b, pp ) by creating space for civil society actors to shape the process. Mixed-member electoral systems (MPPs), like the one used in Bolivia, force parties to produce individual district candidates alongside a party list. It is more likely that strong civil society organizations will be able to agree on nominating single-member district candidates than proportional representation candidates. This is largely because the former districts are smaller, which eases coordination among the organizations that are present in a given district. MPPs also create at least two distinct incentives for the party s top leadership, predisposing it toward the selection of different types of candidates. On the one hand, for districts with single-member district candidates, the key for electoral success is the candidate s personal reputation and support within the district. There are at least three possible scenarios for the leadership. First, where a candidate emerges with strong organizational backing, the leadership can accept the proposed candidate of the organizations, as this would increase the probability of getting out the vote. Second, where there are strong organizations but contested nominations, the leadership can take on a role of arbitration to maximize the chance of electoral success. In these cases, the option that maximizes the likelihood of success is if the leadership chooses a candidate that is most acceptable to a majority of local organizations. Third, where there are few social organizations linked to the party or where these are weakly organized, the central leadership can use nominations to build alliances to existing organizations or to attract the 9

10 support of specific social sectors. For proportional representation candidates, on the other hand, the key to electoral success is the overall strength of the party ticket. Parties often use these candidacies to diversify their lists and attract maximum electoral support. Candidates can emerge with strong organizational backing, but they are likely to be acceptable to multiple organizations or attractive to particular social groups. Recent studies also highlight the relevance of the organizational strength of party subunits for party outcomes (Van Dyck, 2013; LeBas, 2011; Levitsky, 2003). Tavits (2013) argues that local organizational strength helps parties survive. 11 In addition to mobilizing supporters and delivering votes, as the empirical literature shows, a strong local party organization can play a role in the aggregation of political interests, as in the selection of candidates. It can either nominate candidates directly, or serve as an arena for resolving conflicts among competing groups. However, variance in the organizational strength of party subunits does not fully explain variation in candidate selection patterns, nor how internal conflicts are resolved. This paper shows that the party s top leadership can serve as an arbiter in chief in conflicts where the local party organization is either weak or strong. It is more likely that the leadership performs such a role in contexts with a strong and heterogeneous civil society that is not fully aligned with the party, however. In other words, it is the failure of coordination among grassroots actors that, in general, creates an organizational opportunity for the party leadership to centralize power. This finding points to the importance of examining broader structural elements associated with the strength of civil society, and more broadly, with the organizational contexts in which parties are embedded. 10

11 Structural Elements Indeed, structural factors associated with the strength of civil society are central in my explanatory framework. Classic works in political sociology going as far back as Duverger (1954), as well as more recent empirical research on civic engagement and party-building, have established that densely organized civil societies can serve as a potential power base for parties. 12 The expectation in that literature is that organizational infrastructures of politically oriented associations may contribute to the building of strong parties by reducing costs and coordination problems associated with those efforts (LeBas, 2011). This organizational inheritance, as Levitsky, Loxton, and Van Dyck (n.d.) argue, can provide invaluable resources to new political parties and contribute to their long-term empowerment. For these authors, new parties are more likely to take root, and also to persist over time, where politicians build upon the infrastructure of pre-existing organizations. Movement-based parties, even if initially loosely organized and held together, are well positioned to build strong parties, all else equal. I take this literature a step further by adding that variation in social and organizational density can also play an important party-building role by shaping internal party governance. I argue that a well-organized civil society can generate dispersed or coordinated pressure from below that can in turn place limits on the centralized decision-making power and self-regarding political objectives of the party leadership. 13 Specifically, organized civil society may play an important role in resisting the oligarchization of allied parties in the critically important realm of candidate selection, by nominating desired candidates in spite of the leadership preferences. This is largely a function of how the political space is structured in a given electoral district, and of the political alignments between civil society and the party. At least four combinations of party-civil society are possible: 11

12 (1) strong civil society aligned with the party, (2) strong civil society aligned with opposition parties, (3) strong civil society with different political alignments, and (4) weak civil society (in which case political alignments are less relevant). Systematic evidence from candidate selection within a movement-based party demonstrates that oligarchic decision-making by the party leadership is less likely to take place in districts where grassroots organizations aligned with the party are well organized, have mechanisms to arrive at decisions, and can agree on selection. In turn, in contexts where strong grassroots organizations aligned with the party are absent, or where they are strong but have multiple alignments and do not agree on selection, oligarchic decision-making in the hands of a small party elite is much more likely to occur. Similarly, contexts of weak civil society create organizational opportunities for power concentration in the hands of a few. Thus, the evidence highlights the importance of both the strength of civil society organizations and political alignments of civil society on candidate selection outcomes. These findings are consistent with recent developments in the social movement literature inspired by the organizational ecology tradition (Soule, 2012). Though remaining disconnected from the literature on political parties, organizational ecologists point to the centrality of looking at the broader organizational field in which parties operate (Robertson, 2011). The findings presented in this article have interesting implications regarding the consequences that relevant contextual conditions, such as diverging patterns of party-civil society relations, have for the internal politics of parties, an area that could spark further research. Just as parties deploy different linkage strategies to attract different electoral constituencies in unequal societies (Luna 2014), they operate differently in different localities 12

13 depending how the political space is structured. Before turning to Bolivia s MAS, a discussion of relevant variables and their operational indicators is in order. Variables and Operational Indicators To understand how patterns of candidate selection might contribute to resist oligarchic pressures and promote grassroots participation, one must analyze the levels of centralization, inclusiveness, and bureaucratization of the selection processes. Most of the recent political science literature on candidate selection highlights the importance of centralization as an indicator for the distribution of power within a party (Bille, 2001, pp ; Gallagher, 1988, pp. 4-8), and most studies tend to focus on the actors driving the selection processes, and specifically on the geographic and functional location where candidates are selected (Hazan & Rahat, 2010, pp ; Siavelis & Morgenstern, 2008b, pp ). In short, they focus on the where question. Candidate selection can be decentralized in two senses, territorial and social. It is territorially decentralized when local or regional party selectorates nominate party candidates, as opposed to national party selectorates (Hazan & Rahat, 2010, p. 55). It is socially decentralized when it ensures the representation for representatives of groups that are not defined territorially, such as trade unions, women, minorities, or even subgroups within these groups (Hazan & Rahat, 2010, p. 56). At one extreme, when candidates are selected exclusively by a national party body with no procedure to ensure territorial or social representation, then the candidate selection process centralizes power. At the other extreme, when exclusively local-level bodies and/or corporate actors within the party select candidates, then the candidate selection process decentralizes power. 14 Although it is generally assumed that more decentralized 13

14 candidate selection offers greater influence to members and activists, the question of how widely these members and activists are involved in the selection process cannot be addressed exclusively from the point of view of the degree of centralization (Gallagher, 1988, p. 4). The second variable, inclusiveness, refers to the number of people and level of participation of those involved in the selection of candidates at each site of selection (Field & Siavelis, 2008, p. 624). This is a question of who is involved in selection. The who selects is referred to in the more recent literature as the selectorate. The selectorate can range from one person or several people up to the entire electorate of a given nation (Hazan & Rahat, 2010, p. 33). At one extreme, then, when only one leader or a small party elite are involved in the selection of candidates, the selectorate is exclusive. At the other extreme, when a greater number of people are involved in the selection activities, the selectorate is more inclusive. But there are different forms of exclusivity and inclusiveness. In some exclusive systems, a single leader or a few party elites can dominate the selection process. In other cases, self-nomination can be the norm. In more inclusive cases, grassroots and corporate groups or local-level actors can determine who is selected. Or even a system of primary elections can be the norm. The third variable, the level of bureaucratization, refers to the degree to which a party s rules constrain the behavior of key actors involved in selection (Siavelis & Morgenstern, 2008a, p. 16). Essentially, what distinguishes bureaucratic and non-bureaucratic selection systems is the degree to which internal party rules matter in resolving conflicts among competing actors. At one extreme, a bureaucratic selection system involves following elaborate rules for candidate selection. At the other extreme, when party rules do not matter and a small party elite controls de facto the selection process, the selection process resembles an appointment system (Hazan & Rahat, 2010, p. 73). And a group-based selection process, which can be more or less rule-bound, 14

15 is one in which corporate or functional groups have the last word. Corporate or functional groups do more than select candidates; they can also exert veto power over potential candidates, with different levels of bureaucratization and formality (Siavelis & Morgenstern, 2008b, p. 41). Table 1: Candidate Selection in Electoral-Professional and Movement Parties Electoral-professional Movement-based parties parties Centralization Centralized Decentralized; territorially and socially Inclusiveness Low to high High participation of corporate/grassroots actors Bureaucratization Bureaucratic; rule-based patterns of selection; party organs play a central role Non-bureaucratic; rules (where they exist) are not effectively enforced; grassroots actors play a central role Table 1 summarizes the three dimensions in the study of candidate selection for idealtypes of electoral-professional parties and movement-based parties, which are often seen as polar opposites. Parties are remarkably diverse within both categories, however. In an archetypical movement-based party, the degree of centralization is low, the levels of participation of corporate and grassroots groups are high, and these groups play a central role in the selection of candidates. It might be useful to note that the different dimensions in the table can be in conflict and negate each other. While the first two characteristics of movement-based parties indicate possible decentralization of power, the third one indicates possible centralization. This can occur if a small group of influential individuals bypass established practices and exert nomination or veto power over candidacies. When this occurs, and a small group controls the selection process, 15

16 then selection concentrates power. In short, it would be challenging to say in the abstract whether decentralization is higher in movement-based parties than in electoral-professional parties. RESEARCH DESIGN In the rest of this paper I focus on one case, the Bolivian MAS, to develop a theory of within-party variation in candidate selection in which institutional elements interact with civil society characteristics. The experience of MAS is particularly interesting when thinking about movement-based parties because it deviates from the conventional wisdom on this type of parties. 15 Though there are tendencies pointing in the direction of power concentration, grassroots groups have retained considerable bottom up influence in processes of candidate selection in districts where such organizations are strong, united, and have mechanisms to arrive at decisions. The result has been salient variation in candidate selection outcomes across different geographical constituencies. Examining subnational variation in a highly diverse country that uses a mixed-member electoral system (MPP), like Bolivia, is also useful because it allows for a systematic comparison of candidate selection strategies. 16 It allows us to see how the same party selects candidates for different electoral lists, while keeping constant all other contextual factors. Existing empirical studies offer important insights on how MAS selects candidates; however, they are generally descriptive and leave critical questions unanswered (Zuazo, 2008; Do Alto & Stefanoni, 2010). Specifically, although they acknowledge variation within MAS and provide a wealth of qualitative information, they fail to explain what the main sources of that variation actually are. 16

17 My empirical analysis addresses this issue. It examines subnational variation in the selection of candidates for national office in the electoral process leading to the 2009 election. This is a key moment to look at because it allows us to see MAS in its most expansive phase, when it adopted a catch-all strategy of electoral recruitment. As such, the period leading to that election can bee seen as s most likely environment for oligarchic decision-making. I rely on systematic empirical evidence collected through interviews with over fifty MAS representatives from the districts of La Paz, Cochabamba, Oruro, and Santa Cruz (see Table 2), 17 in addition to 120 interviews with key informants, to explain how candidates are selected under different partycivil society constellations. Interviewees include leaders of allied grassroots organizations, nonelected regional party brokers, unsuccessful aspirants, members of the executive branch, representatives of opposition parties, experts, journalists, as well as candidates nominated for local office in rural and urban districts. 18 Data from these interviews are supplemented by a close reading of newspapers of the process and its aftermath, and of the existing secondary literature. Table 2. Number of Interviews with MAS Representatives in Congress, by Department Department Number of Total elected Total Seats Interviews under MAS La Paz Oruro Santa Cruz Cochabamba Other 11** Total * * This count only includes the total seats for the departments systematically studied in this paper, which accounts for a clear majority. The total number of seats in the Bolivian Congress is 172. Of these, the MAS has 114. ** This number includes interviews with MAS representatives from Bolivia s remaining departments (Chuquisaca, Potosi, Tarija, Beni, and Pando). 17

18 Parties are complex systems that develop uneven organizational structures across the territory and relate differently to their constituencies in multiple settings depending on how the political space is structured. Therefore, subnational analyses are particularly promising for studying internal party organization and behavior. The analysis developed here reveals tremendous variation across different localities. Conditions where organizations allied to MAS have a near monopoly of organization are more likely to be observed in the rural areas in Bolivia s western departments. These include La Paz, Cochabamba, and Oruro. There, single-member district candidates are more likely to emerge from social organizations and be accepted by the party leadership. At the same time, this pattern of candidate selection reflects a de facto diffusion of power, in the sense that it mirrors the balance of power between MAS and territorially grounded grassroots organizations, and also between these organizations themselves. Bolivia is no exception to the rule that urban areas are generally more heterogeneous than rural areas. In some cities or urban districts the diffusion of power among organizations leads to a situation where agreement on candidates is difficult, while in other cases the organizational density (i.e. the presence of organizations aligned with MAS) is significantly lower. In both types of situations, the candidate selection process exhibits a combination of oligarchic decisionmaking with grassroots participation and consensus building. Specifically, where there are strong organizations but no consensus among them, the leadership is likely to choose a candidate acceptable to a majority of local organizations; where organizations are weak, however, the leadership is likely to select candidates that will help to build alliances to existing organizations or to attract support from particular groups that may increase electoral returns. These patterns are similar to the one observed in the case of most proportional representation candidates. Finally, in 18

19 Bolivia s eastern departments, which represent new arenas of competition for MAS, the social organizations linked to the party are weak. In the absence of strong organizations that can agree on candidate selection, elite decision-making is more likely to occur. EMPIRICAL SETTING Bolivia s bicameral Congress consists of a Chamber of Senators with 36 seats, and a Chamber of Deputies with 130 seats. All elected representatives serve five-year terms, and reelection is permitted. Members of the Chamber of Senators are elected through a closed-list proportional representation. Deputies are elected by a mixed-member proportional (MPP) electoral system. The introduction of this system created two different types of seats in that chamber, plurinominal (proportional representation) and uninominal (single-member district), forcing parties to produce individual district candidates alongside a party list. In addition to those, the country s 2009 constitution established a few special seats for indigenous peoples and Afro-Bolivians. Seventy uninominal representatives are elected by plurality vote in single-member districts, fifty-three plurinominal representatives are elected in through a closedlist proportional representation system, and the seven special representatives are elected by plurality vote in single-member constituencies. In the 2009 general election MAS gained 26 of 36 seats in the Chamber of Senators, 33 of the 53 plurinominal representatives, 49 of the 70 uninominal representatives, and 6 of the 7 special seats. Because MPPs force parties to produce individual district candidates alongside a party list, it is useful to make an analytical distinction between several types of candidates. This distinction should be seen as reflecting a general pattern, to which exceptions exist. On the one hand, uninominal candidates tend to emerge from below through various participatory 19

20 mechanisms and traditions that are not codified on a single party norm. Here grassroots organizations have significant influence on the selection process, helping to ward off oligarchy. On the other hand, plurinominal candidates tend to emerge from agreements reached by the leadership and by invitation from the top. These candidacies reflect power struggles and agreements reached between the leadership and grassroots organizations, or between the leadership and specific individuals. Here the influence of grassroots organizations is reduced. UNINOMINAL CANDIDATES Generally, in selecting these candidates, MAS delegates responsibilities and control to the grassroots organizations that are present in a given electoral district. In these cases, then, candidate selection consists of decentralized and inclusive procedures that provide significant opportunities for the grassroots to influence decision-making. The process relies on nonbureaucratic patterns of selection, and candidates emerge based on the strength of the social organization they represent. Prior to an election, the National Directorate of MAS issues a call for nominations to allied grassroots organizations throughout the country. These organizations then are in charge of conducting screening, pre-selection, and candidate nomination processes, and they do so by electoral district and according to the norms and procedures they themselves deem adequate. In most cases, the leadership of MAS respects the decisions by grassroots organizations. With some exceptions, these organizations have the last word on nominations. This is evidence against the operation of an iron law. Formal membership in MAS is not a condition for candidacy. Instead, aspiring candidates need to be approved by the people in their territory. 19 The only hard-and-fast rule 20

21 that MAS respects is that each district has to ensure rigorous gender equality: men s and women s names have to alternate on electoral lists. Although there is no rule stipulating that uninominal candidates must have experience as a leader of a grassroots organization, this is almost always the case. As former deputy Dionicio Núñez commented, it is practically impossible to become a uninominal candidate for MAS if you do not have experience as a [social organization] leader. Interviews with multiple uninominal representatives in La Paz, Cochabamba, Oruro, and Santa Cruz, confirm this observation. Analyses based on the survey of Latin American parliamentary elites conducted by the University of Salamanca show that most of them, indeed, came from a grassroots organization (Zegada & Komadina, 2014). The process of selection is territorially and socially decentralized. The key actors are grassroots organizations with a territorial base. These gained legal status as Territorial Grassroots Organizations (OTBs) with the 1994 Law of Popular Participation, and they generally include neighborhood associations, traditional indigenous organizations (the ayllus), and modern peasant unions (the sindicatos campesinos). Candidate selection within MAS ensures the representation of the OTBs that decide to join the party. 20 Below I examine subnational variation in the nomination of uninominal candidates according to the four party-civil society constellations outlined on pp Strong Civil Society Aligned with MAS Where civil society is strongly organized and aligned with MAS, as in the western departments of La Paz, Cochabamba, Oruro, and Potosí, this decentralized way of participation tends to be the norm. The process begins at the lowest organizational level of the union structure, the sindicatos campesinos, and then moves up to the higher organizational levels that structure the territory, the subcentrales and the centrales. Figure 1 depicts a stylized version of the 21

22 functional levels of the peasant union organization in Bolivia. 21 In general, the subcentral aligns with the territory of the electoral district, meaning that there is generally one subcentral per electoral district. The more specific mechanics of the process described below can be only described as a general tendency, and the following account portrays an ideal typical model through which MAS selects uninominal candidates. 22 The selectorate for such candidacies is highly inclusive, and there is a clear emphasis on extending grassroots participation. As Vice President Álvaro García Linera commented in an interview, because these candidates are not handpicked, they are not the candidates of the party in a strict sense. They are selected by grassroots organizations as a function of their territorial power, and they are the representatives of those organizations. In other words, in contexts where civil society is strong and aligned with MAS, the participation and decision-making capacity of grassroots organizations with a territorial base is high in the case of uninominal candidates. The mechanics of selection can be summarized in three steps. First, each sindicato and other OTBs in a given district organize meetings to make a preliminary screening of potential candidates and then select their nominees. These meetings are referred to as ampliados or cabildos; they are crowded events that ensure broad grassroots participation. The individuals who are elected at this level will then represent their organization and in the competition at the next highest level of organization, the subcentrales. At this level, each subcentral holds an ampliado or a cabildo to choose among the sindicato-level nominees. The winner of each subcentral contest then goes on to compete for representation at the next highest level of organization, the central. The candidates for each electoral district are defined at this level, as seen below. 22

23 Figure 1: Organizational structure of Bolivia s peasant unions 23 Central Subcentral Subcentral Subcentral Subcentral Sindicato Sindicato Sindicato Sindicato Sindicato Sindicato Sindicato Sindicato Sindicato Sindicato Sindicato Sindicato Sindicato Sindicato Sindicato Sindicato Third, the central organizes an ampliado or a cabildo with all of the nominees presented by the subcentrales. The winners at this level typically emerge as uninominal candidates if they receive the support of all the organizations involved since the first step of the process just described. This support is determined at a large congress with the broad participation of the authorities of all the levels described so far, and it is obtained by mechanisms of union democracy that are not codified under a single written norm. The runners-up serve as substitute candidates for the district. Strong Civil Society Aligned with Opposition The expansion to the east, where MAS was historically weak, pushed decision-making structures into a more oligarchic direction. 24 In Santa Cruz, for example, the selection of 23

24 uninominal candidates followed the familiar bottom up pattern described above, particularly in rural districts; in the city of Santa Cruz, where there are strong organizations aligned with opposition forces, a local party structure also played an influential role and nominated several candidates from its ranks. 25 However, selection at the level of the department also included alliance and coalition building with other parties, with politically influential groups and nontraditional organizations, and with a wide array of ad hoc urban organizations. These alliances, which guaranteed representation for members of those groups, were created by the national leadership; they were created neither by the local party, nor by the grassroots organizations that control the Regional Directorate. 26 The composition of the electoral list reflected an internal balance of power that favored those urban groups over peasant organizations, reflecting a more centralized and exclusive selection pattern, whereby the influence of a small party elite was strengthened. 27 It also revealed a strong pragmatism by the top party leadership. Evidence from other eastern departments, where major civil society organizations are aligned with the opposition, reveals a similar pattern of candidate selection that combines participation from below with oligarchic decision-making. Strong Civil Society with Multiple Alignments The fact that MAS has grown fast and in a decentralized manner has given significant flexibility to newly incorporated local organizations regarding the selection of candidates. In general, their decision to be a part of MAS implies mutual benefits. The logic is the following. MAS opens its electoral lists and gives control over the selection process to these organizations. Thus, MAS benefits from the social networks and organizational infrastructure of these organizations, which are familiar with the electoral terrain and able to organize campaigns and 24

25 mobilize resources more efficiently. In turn, grassroots organizations benefit from the image of MAS, which generally increases their likelihood of electoral success. This symbiotic relationship is different in rural and urban environments, however. Grassroots organizations with a territorial base are central to selection in both settings, but in urban areas there are usually no clearly identifiable organizations that exert dominance over the territory. Rather, there is a multiplicity of neighborhood associations, professional associations, cooperatives, unions and the like, and they usually are in competition during the selection process and competing organizations often have difficulty agreeing on a preferred candidate. When conflicts arise and competing organizations cannot reach agreement, a small party elite that often includes the president himself acts as an arbiter and has the last word. As I then elaborate further, it is failures of coordination among grassroots actors that create an organizational space for the leadership to centralize power, pushing internal decision-making structures into a more oligarchic direction. Weak Civil Society In the rural areas, particularly where grassroots organizations have a dominant control over the territory, MAS has not invested much in the building of a party branch independent of these organizations. 28 In the urban areas, by contrast, and particularly in places where grassroots organizations are not strong, or where they do not have dominant control over the territory, MAS has constructed territorial party organizations of varying strength. For the most part, however, these structures lack independent decision-making power, creating an organizational opportunity for the party leadership to centralizer power in contexts where civil society is weak. 25

26 Summary To summarize, systematic evidence from the departments of La Paz, Cochabamba, Oruro, and Santa Cruz, complemented by less systematic observations of other departments, suggests that the crucial variable determining the nature of candidate selection is the strength of the grassroots organizations, which is defined as a function of its territorial power. Here, MAS diffuses power among territorially grounded grassroots actors, which generally have the last word on selection. Once these organizations nominate a candidate, this person becomes a candidate for MAS. However, when conflicts emerge among competing organizations, MAS tends to concentrate decision-making power in the hands of a small party elite and even in the hands of Morales himself. These disputes are rarely resolved through formal channels, or by the local party organization. These dynamics are more commonly observed in urban areas, where no single organization controls the territory, and in the eastern departments, where MAS expanded by means of a catch-all strategy of recruitment. PLURINONMINAL CANDIDATES Districts for plurinational candidates are larger, meaning that there are different partycivil society constellations within each district. It also means that coordination among competing organizations is generally more complicated than in uninominal districts. Conflicts among such organizations create an organizational opportunity for the party leadership to centralize power. Indeed, plurinominal candidates typically emerge from agreements between the leadership of MAS and specific social sectors, or are directly selected by Morales. These candidacies help to generate balances territorial, corporate, urban/rural, and male/female after the list of uninominal candidates is approved. 29 Here the selection process is more centralized; 26

27 the principal actors are either national party leaders, members of the national-level government, or brokers with access to patronage resources in departmental governments, and ultimately Morales himself. In comparison to the uninominal candidates, grassroots influence is reduced. Although the selection of plurinominal candidates is more centralized and exclusive, and therefore more oligarchic, it can be seen as a part of a deliberate strategy of addition that allows for the incorporation of sectors and groups that do not have a territorial or an institutional corporate base. As Leonida Zurita commented, the idea is to include everyone that is, the professionals, the non-professionals, the intellectuals, the non-intellectuals, the indigenous and nonindigenous middle class, the women, and so on. It is in that sense that our project is one of inclusion and not of exclusion. This view is akin to the view of Concepción Ortiz, MAS s Vice President, who said that this mechanism allows MAS to balance its electoral lists, and it is seen by the leadership as an inclusionary way to give representation to urban middle classes. Formal membership and having a background as a movement leader are not conditions for candidacy. As a result, some plurinominal candidates, particularly those without such a background, are generally seen as unwelcome competitors by rural and peasant organizations and by candidates that emerge from these organizations. They consider themselves the authentic representatives of MAS. Indeed, these plurinominal candidates are referred to as invited. That characterization is used to contrast organic rural-based rank and file with invited urban and middle classes. This form of nomination was not widespread during the early days of MAS. Rather, as Do Alto and Stefanoni (2010, pp. 312, 325) suggest, the invitation of candidates only became common in preparation for the 2002 election. Here MAS developed an expansive strategy of electoral recruitment and coalition building in order to compete successfully for the presidency 27