Clientelism and Electoral
Volatility in Brazil
Helloana Medeiros
A thesis presented for the degree of Doctor of Philosophy
Escola Brasileira de Administração Pública e de Empresas Fundação Getúlio Vargas
Brazil July 2019
Ficha catalográfica elaborada pelo Sistema de Bibliotecas/FGV
Medeiros, Helloana
Clientelism and Electoral Volatility in Brazil
/ Helloana Medeiros.
– 2019.165 f.
Tese (doutorado) - Escola Brasileira de Administração Pública e de Empresas, Centro de Formação Acadêmica e Pesquisa.
Orientador: Cesar Zucco Júnior Inclui bibliografia.
1. Clientelismo.2. Democratização. 3. Deputados federais. Eleições - Brasil. I. Zucco Júnior, César. II. Escola Brasileira de Administração Pública e de Empresas. Centro de Formação Acadêmica e Pesquisa. III. Título.
CDD – 324.981
Helloana Medeiros
Clientelism and Electoral Volatility in Brazil
Dissertation presented at Escola Brasileira de Administração Pública e de Empresas for the degree of Doctor of Philosophy. Advisor: Prof. Cesar Zucco Jr.
Rio de Janeiro 2019
Toda volição, ainda que seja uma atividade do espírito, relaciona-se com o mundo das aparências no qual seu projeto deve realizar-se [...].A ten-são daí resultante [...] causa uma espécie de inquietação na alma que beira facilmente a confusão, uma mistura de medo e esperança que se torna in-suportável quando se descobre que, na formulação de Santo Agostinho, querer e ser capaz de realizar, velle e posse, não são a mesma coisa. A tensão pode ser superada somente pelo fazer.
Acknowledgments
I want to express my gratitude towards Prof. Cesar Zucco Jr. for his supervi-sion, support, and assistance throughout the planning and execution of this research. During my graduate years, Brazilian politics experienced great turmoils that have shaken many of the beliefs established about its political system. Studying difficult topics such as clientelism and electoral strategies in Brazil during those transforma-tive years have been challenging. Finding solutions to each specific obstacle posed by such topics in an unstable political context was only possible with the careful as-sistance of an experienced and generous professional such as Prof. Cesar. From EBAPE, I also would like to thank Prof. Fabio Caldieraro for having commented on an earlier version of this work and Prof. Octavio Amorim Neto for having changed my perspective about qualitative methods. From the university staff, I am grateful for Roberta Guimarães for having provided support to my academic pursuits.
I am grateful to Prof. Jairo Nicolau for his support during this project, and for the friendship, he has provided along the years. In 2014, Prof. Jairo invited me to collaborate in a project and at the end of this work, he became the main supporter of my candidacy to the Ph.D. Five years later, I am still an inexperienced learner -but hopefully, hold now a couple of interesting stories to share. I also would like to thank Prof. Glauco Peres da Silva, for the many insightful comments he shared as a committee member. Likewise, I am also grateful to Prof. Lucas Novaes and Prof. Alisha Holland for the time dispensed writing inspiring and detailed comments. I also would like to express my gratitude to Prof. Simeon Nichter for the time kindly dispensed listening to my research interests and the thoughtful comments he shared with me about this topic.
I included several adjustments in the final design of this research to address the concerns and critical assessment provided by such generous and willing profes-sionals that I had the privilege to learn from. I am certain that if any contribution can be found in the following pages, it was achieved through their careful revision, and the attention the project has kindly received. Remaining mistakes, of course, reflect my limitations solely.
From my Alma mater, Universidade Federal do Ceará, I would like to express my very great appreciation of Prof. Valmir Lopes for his friendship, and the advises he has provided along the years for my work. Prof. Valmir’s inspiring comments have always been a guide in my way of doing research, questioning theories, and analyzing problems. I have never found another limitless source of kindness and
Nillin and Prof. Geísa Mattos for the conversations, and many laughs we shared whenever they visited me in Rio. It was in the corridors of this lovely Social Sciences department that I learned how to be curious and inquisitive. To you and many others I owe inspiring lessons about knowledge, ethics and respect that have guided my path like stars in the darkest of hours.
During the survey application, the assistance provided by Lara Saraiva and Antônia Gabriela de Araújo was also greatly appreciated. Working as survey appli-cants can be tough, as most sociologists are aware. Lara and Gabriela did an excellent job, going through performance pressures and squeezed time schedules. Their ded-ication was crucial to the execution of this work, and therefore, I am in terrible debt with them as well.
Finally, I wish to thank my family for their extensive support and encour-agement throughout my study. My parents, Rita and Hugo, have provided the best safety net a student could wish (delicious free meals included). I am grateful for their love and for all the emotional support they have kindly offered. To my sisters, Lara and Raquel, for being the best companionship for conversations about “life, the uni-verse, and everything else”. To my baby nieces, Gabriella and Giovanna, for giving me the most lovely hugs in the world. And to my son, Caio, for having become even more funny and loving than I could have aspired for. The completion of this work has demanded many hours of my attention and care away from them. Unable to reverse clocks, I am forever grateful for their understanding.
In this dissertation, I investigate some of the theoretical and empirical puzzles posed by the persistence of clientelist practices in Brazil. As explored in the following In-troduction, the competitiveness of the Brazilian electoral system and reports of in-creased vote buying contrasts with most theoretical explanations for the phenomenon. Chapter Two presents theoretical arguments for why competitiveness changes the nature of clientelist relationships, classifying the phenomenon in traditional and short-term clientelism. Chapter Three analyzes literature about clientelism in dif-ferent institutional periods of Brazilian history. Chapter Four analyzes electoral volatility; using this new measurement to understand how stable are the connec-tions between federal deputy candidates in Brazil and their city-level electorates. A qualitative analysis offer support to short-term clientelism theory: three selected candidates associated with clientelism and with elevated electoral volatility in 2018 increased their vote shares between elections in municipalities where their precinct-level concentration are higher, suggesting the mechanism of professional brokerage operates as predicted. Chapter Five presents theoretical implications and methods for the Re-election Candidates Survey. Results show that candidates in a sub-group of participants with high predictability and elevated electoral volatility reported po-litical strategies and displayed voting patterns in accordance with short-term clien-telism theory. These results support the claim that this electoral strategy could have been used by a broader set of federal deputy candidates in Brazil during the 2018 elections.
List of Figures
2.1 Probability of Punishment and Number of Candidates - Different
Dis-trict Magnitude and Number of Clientelist Offers. . . 15
3.1 Civic Organizations Created in Brazil Per Year and State (1950-2017) . . 56
3.2 Civic Organizations Created in Ceará Per Year - Before and After De-mocratization . . . 60
3.3 Civic Organizations Created in Ceará per week - Pre and Post PSJ (1994-1996). . . 61
3.4 Comparison of civic organizations created per day. . . 62
3.5 Types of Electoral Strategies by Two Attributes . . . 74
4.1 Candidates’ Electoral Volatility in Brazil per State (2002-2018) . . . 97
4.2 Histogram: Electoral Volatility in Brazil (2002-2018) . . . 100
4.3 Distribution of Electoral Volatility per Group - Elected and Non-elected. 101 4.4 Federal Deputy Domingos Neto and Domingos Filho . . . 111
4.5 Domingos Neto’s Precinct Concentration per Group - Different cutoff points. . . 114
4.6 Federal Deputy André Fufuca and Fufuca Dantas . . . 116
4.7 André Fufuca’s Precinct Concentration per Group - Different cutoff points. 119 4.8 Federal Deputies Zé Silva and Áurea Carolina . . . 122
4.9 Zé Silva’s Precinct Concentration per Group - Different cutoff points. . 123
4.10 Áurea Carolina’s Precinct Concentration per Group - Different cutoff points. . . 124
5.1 Actual versus Predicted Vote Shares and Rank Position Outcomes per Question Groups. . . 137
5.2 Survey Participants’ Answers per Sample Groups - Main Strategy Re-ported. . . 140
5.3 Survey Participants’ Region per Sample Groups . . . 143
5.4 Survey Participants’ Electoral Status per Sample Groups . . . 144
5.5 Survey Participants’ Rank Accuracy per Sample Groups . . . 145
5.6 Survey Participants’ Precinct Concentration per Cities’ Vote Share In-crease Group - Different Cutoff points. . . 146
List of Tables
3.1 Linear Regression and Fixed Effects Model - Number of Associations per Muniicipality in Ceará- Pre and Post PSJ (1934-2017) . . . 65 3.2 Linear Regression and Fixed Effects Results: Effective Number of Cands.
and Assocs. per Type - Ceará. . . 67 4.1 Number of Cities, Number Effective Candidates per State and Cities and
Electoral Volatility of Candidates in Brazil - Absolute and Median Values Federal Deputies 2018 . . . 95 4.2 Liner Regression and Fixed Effects Model - Candidate’s Vote Shares and
Electoral Volatility per City. . . 99 4.3 Linear Regression with Fixed Effects Models - Volatility and Average
Dominance . . . 104 4.4 Domingos Neto’s 10 most inconsistent electorates Number of Votes,
Per-cent Change and Rank Position in 2014 and 2018. . . 113 4.5 André Fufuca’s 10 most inconsistent electorates Number of Votes,
Per-cent Change and Rank Position in 2014 and 2018. . . 117 4.6 Zé Silva’s 10 most inconsistent electorates Number of Votes, Volatility
and Rank Position in 2014 and 2018. . . 122 5.1 Fixed Effects Linear Regression Models - Comparing respondents’
an-swers within cities. . . 137 5.2 Linear Regression Models for DV measures - Percentage Accuracy and
5.3 Linear Model Results: Distinct Accuracy Measures and Electoral Volatility.141 5.4 Frequency Table - Volatility and Accuracy Levels . . . 142 A.1 Desciptive Statistics: Values per individual . . . 155 A.2 Desciptive Statistics: Values per individual and city . . . 156
Contents
1 Introduction 1
2 The two clientelisms. 7
2.1 Punishment, monitoring and uncertainty . . . 10
2.2 Traditional patrons as monopolists: PRI in Mexico and Peronists in Ar-gentina . . . 17
2.3 Short-term clientelism . . . 23
2.4 Short-term clientelism and vote buying: a necessary distinction . . . 33
3 Clientelism and electoral strategies in Brazil: a qualitative approach. 39 3.1 Clientelism before democratization . . . 43
3.1.1 Enlisting voters and buying votes (1946-1964) . . . 47
3.1.2 Dominance and traditional clientelism (1964-1988) . . . 51
3.2 Civic associations as professional brokers: clientelism after 1988 . . . 55
3.3 Electoral strategies: a framework . . . 70
4 Electoral volatility and short-term clientelism: three case studies. 84 4.1 Electoral volatility . . . 87
4.1.1 What is electoral volatility? . . . 89
4.3 Volatility and Short-term Clientelism. . . 105
4.3.1 Methods: Case studies and a smoking-gun test . . . 106
4.3.2 Domingos Neto: family tradition and the two clientelisms . . . 110
4.3.3 André Fufuca: highly unstable votes in Maranhão . . . 115
4.3.4 Zé Silva: vote buying and concentration patterns compared in Minas Gerais . . . 120
4.4 Implications . . . 125
5 Predictability, volatility and short-term clientelism 129 5.1 Accuracy and volatility . . . 131
5.2 Methods: Re-election Candidates Survey . . . 133
5.3 Results . . . 136
6 Conclusion 150
1. Introduction
In April 2015 the former Brazilian federal deputy Pedro Corrêa1was arrested in one
of the phases of Operação Lava Jato - an investigation to uncover money launder-ing.2,3,4During police investigation, one of Corrêa’s former employees was arrested
with many payment receipts from accounts that allegedly belonged to voters. The receipts ranged from electric power and water bills, to gas purchases, construction materials and vehicle parts, which, according to this employee, Corrêa payed for vot-ers in former elections to cast a ballot for him.5 Months later, the former deputy
ex-ecuted a plea bargain in Federal Court and was condemned to almost three decades in prison for 72 acts of passive corruption.6A couple of years later, the Court allowed
Corrêa to be confined to house arrest due to health problems.
It is worth noting that Corrêa had already been accused and found guilty in another famous corruption scandal in Brazil - the Mensalão Case. However, such episodes did not seem to shake Corrêa’s confidence as a politician. In 2017, during an interview with the press from his home prison, the former deputy revealed: “I am a politician who carried out elections with money, arranging vote buying with mayors and councilors”, and guaranteed that he would still be elected to Camara Federal if the justice system allowed - regardless of his wrongdoings.7
1Partido Progressita - Pernambuco
2Operacação Lava Jato (or Car-Wash operation) is an ongoing investigation coordinated by the
Brazil-ian Ministerio Publico Federal, Justiça Federal and Polícia Federal that aims to reveal and punish money laundering schemes related to corruption practices inside government companies and institutions. With-out precedent in Brazilian history, it has returned more than 11 billion reais and demanded almost 200 arrests so far.
3Lava Jato: MPF recupera R$ 11,9 bi com acordos, mas devolver todo dinheiro às vítimas pode levar décadas
(https://www.bbc.com/portuguese/brasil-43432053)
4Lava Jato completa 3 anos e soma 198 prisões; 5 políticos se tornaram réus no STF
(https://g1.globo.com/politica/operacao-lava-jato/noticia/lava-jato-completa-3-anos-e-soma-198-prisoes-5-politicos-se-tornaram-reus-no-stf.ghtml)
5Documentos indicam ‘compra de votos’ por Pedro Corrêa
(https://politica.estadao.com.br/blogs/fausto-macedo/documentos-indicam-compra-de-votos-por-pedro-correa/).
Termo de declarações de Jonas Aurélio Lima Leite:
(https://politica.estadao.com.br/blogs/fausto-macedo/wp-content/uploads/sites/41/2015/04/depoimento-jonas-aurelio.pdf)
6Tribunal aumenta em 9 anos pena do ex-deputado Pedro Corrêa
(https://www1.folha.uol.com.br/poder/2017/09/1918294-tribunal-aumenta-em-9-anos-pena-do-ex-deputado-pedro-correa.shtml/)
7Condenado, Pedro Corrêa diz que política se fazia com corrupção e compra de
vo-tos
Corrêa’s voters are not alone. According to a 2015 study, commissioned by Tribunal Superior Eleitoral (TSE) for the company Checon Pesquisa/Borghi, selling and buying votes in Brazil are still diffuse practices, as at least 28% of the interviewees reported witnessing or having knowledge of vote buying in the 2014 elections. The research used focus group methodology and included almost two thousand partici-pants from eighteen to sixty years old distributed throughout seven Brazilian capitals and from all social classes.8 The study reveals that the voter perception of vote
buy-ing as a criminal act is not strongly rooted in the society. As one interviewed voter explained, she needed the money and candidates had enough resources to share.9
Motivated by such evidence, this dissertation is focused on understanding why vote buying and clientelism are efficient electoral strategies for Brazilian politi-cians and candidates. Despite elevated poverty rates and the vulnerability of poor individuals, Brazil has one of the most fragmented political systems in the world. It is extremely competitive and politicians more often disperse their voting pat-terns, while clientelism is usually associated with long-term exchanges and rigid local control of resources. How have such practices, related to authoritarianism and geo-graphic concentration, adapted to the democratic period? How have they changed over time? And how do they compare to other electoral strategies used by candi-dates?
Although its long-documented history of patronage and the recurrence of political scandals have led most observers to naturalize the existence of clientelism and vote buying in contemporary Brazil, the persistence of such practices should not be taken for granted, at least theoretically. Both the elevated competitiveness of Brazilian politics and the bulk of democratic rules established in the country in the 1980’s were expected to undermine the effectiveness of clientelism, especially considering that poverty and inequality levels have decreased significantly since then -but clientelism doesn’t appear to (Nichter, 2018; Medeiros, 2013; Lopes, 2005; Torres, 2018; Pessoa Jr., 2011; Gay, 2016, 1999, 2006; Montero, 2010). This doctoral research aims to explore some of the reasons behind and possible explanations for this puzzle. Under many instances, the fact that Brazilian politicians could use clientelist
8Pesquisa revela que compra de votos ainda é realidade no país
(http://www.tse.jus.br/imprensa/noticias-tse/2015/Fevereiro/pesquisa-revela-que-compra-de-votos-ainda-e-realidade-no-pais)
9A voter from Goiânia was quoted in the final report of the research admitting that “I made a lot of
money from politicians who are not worth a thing and I was in need of it to pay my power and water [bills], I took from all, received from many, and didn’t vote for one. I accepted from eight different politicians, R$ 30.00 from each, state deputy, federal deputy. (Could you do that?) You couldn’t, but I was in need and they have the money to deliver to others.”
strategies with some success in elections is completely at odds with the political sys-tem’s low-entry costs and its extremely fragmented political landscape - especially for non-majoritarian offices. In the face of competitive elections, and a large num-ber of candidates seeking office, several candidates could use clientelist strategies to attract voters. Nonetheless, if many politicians (and their brokers) could approach the same set of voters with clientelistic offers, we should expect increased costs for monitoring and greater defection from individual agents. Instead of the classic mo-nopolistic political market (more traditionally associated with clientelism) the land-scape in Brazil approaches more closely an auction for votes: expensive and mostly ineffective (Nichter, 2018), as the principal-agent problems and defection issues that arise in clientelistic exchanges (Stokes, 2005) are not easily solved in competitive clientelist systems.
Moreover, clientelism is thought to be an ongoing political relationship (Hicken, 2011; Kitschelt, 2000; Stokes et al., 2013). Thus, it should require that the personal linkages between voters, brokers and politicians remain stable over time. Though this seems to be the case at the local level (Nichter, 2018), this stability is hardly re-producible in state and federal elections (Novaes, 2017), as alliances between politi-cians and party coalitions can be considered fuzzy at best. Its yarn-ball resemblance leads us to believe that party leaders and upper-tier politicians are constantly fight-ing each other above the shiftfight-ing sands of volatile and inconstant connections, in a system where voters’ party identification is considerably low (Zucco and Samuels, 2015; Novaes, 2017). Paradoxically, despite a greater increase in party system com-plexity and reports of clientelism, democracy has strengthened and flourished: the country has achieved great political and economic advancements in the last decades and is considered a successful case of a completed transition to democracy in Latin America.
During the democratic consolidation period in the 1980’s, Brazil’s most im-portant political institutions were shaped following prescriptions of what scholars thought should make a vibrant democracy. The electoral system adopted propor-tional representation (PR) method for the Legislature - a body of electoral institu-tions that is believed to be less permeable to clientelism and patronage than majori-tarian systems (Pellicer et al., 2013). Additionally, the country has experienced more than thirty years of free and regular elections and its democratic institutions have also been able to endure the test of time, resisting several economic and political crises without succumbing (at least so far) to authoritarian interventions.
Brazil, it didn’t seem to matter how much democratic institutions and governments accomplished, party politics and elections never became completely vertically or-ganized; party roots in civic society were loose at best, and their ideological iden-tities blurred. Furthermore, political fragmentation in the country rose to alarm-ing levels (Rebello, 2012). After droppalarm-ing from 8.5 effective parties in 1990 to 7.8 effective parties in 1998, the fragmentation in the Camara dos Deputados (Brazil’s Lower Chamber) has increased steadily since 2002, getting to above 13 the num-ber of effective parties - the world’s largest - in the 2014 elections (Gallagher, 2017). Politicians seemed inept at keeping their constituencies loyal, increasing the issues of electoral accountability in the country (Samuels, 2002). They were also more or less unable to keep themselves on a single partisan track for very long and changed par-ties frequently (Melo, 2000). When it comes to coalition and party alignment, you could hardly follow a single candidate’s alliances over time, let alone whole parties’ alliances.
In relation to local politics, major changes have taken place during democ-ratization. Municipalities gained the status of federal entities with the 1988 Constitu-tion, and several policies have been decentralized to the municipal level since. Partic-ularly, the health (SUS) and social services (SUAS) systems have structured decision-making processes and implementation organizations in which the municipalities have become crucial. Evidence from qualitative studies points out that although candidates still used clientelism broadly, the importance of mass communication strategies and the relevance of the policies and services delivered at the local level also increased (Lopes, 2005; Accioly de Carvalho, 2009; Medeiros, 2013), represent-ing a major change relative to the prevalence of former patron-client relationships. The objective of this dissertation is to investigate some of the reasons why, despite democratic advancements and elevated competitiveness, clientelism has not disappeared in Brazil. I assume that a short-term type of clientelism has been risen, while traditional-patron connections crumbled. Short-term clientelism has chal-lenged both party theory and the literature on clientelism itself as it is practiced broadly by several parties, is more instrumental and flexible, and seemingly does not exert rigid control over voters or clearly provide stable and vertical relationships between politicians. Existing in a country where only thirty years ago democratic institutions were thought to be prepared to leave clientelism behind altogether, and in which programmatic provision of services has advanced greatly in the last decade, the existence of this political chimera is even more baffling and complex.
clientelism is advanced in order discuss if clientelism can be categorized in distinct types, according to monopolistic or competitive environments. Although this does not intends to be an exhaustive classification, I argue that it can assist researchers in understanding different mechanisms through which clientelist exchanges are en-acted. My theory of the two clientelisms thus contributes to the current literature as it highlights more properly differences between expected time horizon of clientelist exchanges. By explaining the mechanisms through which distinctive types of clien-telism might emerge, a step towards addressing this significant gap in the current literature is initialized, especially in what is concerned with short-term clientelist in-teractions. It is conjectured that if competitiveness changes the nature of clientelist relationships, we should find in the same country differences in clientelism follow-ing authoritarian and democratic regimes. Additionally, if candidates can increase their voting shares through short-term clientelism, their electoral volatility should be elevated and their precinct concentration should be greater in places where they increase more their voting shares between elections.
Chapter Three presents a qualitative assessment of clientelism across differ-ent political regimes in Brazil. A historical perspective of local politics in Brazil is advanced, highlighting the changes in the relationships between voters and elected officials at the local level. This comparative-historical account is important because it shows more clearly how the use of clientelist strategies has changed over the course of different institutional rules and competitive environments, especially after Brazil-ian democratization in the 1980s. I also present a broader system of classification for electoral strategies, focused in distinguishing how candidates and parties delegate the task of eliciting support to different types of organizations, in monopolists or competitive environments while employing or not conditional exchanges to lever-age votes.
In Chapter Four I propose the study of an innovative measure of candidates’ bailiwicks inconsistency. I investigate in detail the electoral volatility of candidates to federal deputy in Brazil (1998-2018), review interpretations of spatial voting in the country and test theoretical assumptions. It is sustained that this measure can help us identify candidates who employ the short-term type of electoral strategy. Following, I select three case studies of re-elected federal deputies associated with clientelism and with elevated volatility, exploring their voting patterns through a smoking-gun test. In what is related to precinct aggregation, results show that these candidates display patterns consistent with predictions of short-term clientelism theory.
analyze the cross-section between electoral volatility and candidates’ accuracy when predicting their voting patterns just prior elections. Results suggest that in the cat-egory of high volatile and highly precise candidates, precinct aggregation patterns are compatible with the short-term clientelism theory, supporting the claim that this practice has been an electoral strategy used by candidates more broadly.
2. The two clientelisms.
The continued prevalence of clientelism in the political and electoral life of Third Wave democracies has given rise to renewed scholarly interest in the topic. Formerly thought to disappear as authoritarian states vanished and economic modernization ensued, as the practice persisted, novel theoretical and methodological tools were proposed. Nevertheless, building a theoretical framework for clientelism compatible with the resilience of such practices in democratic environments has proven difficult, despite many endeavors Stokes et al. (2013); Holland and Palmer-Rubin (2015); Diaz-Cayeros et al. (2016); Nichter (2018). In this Chapter, I point to an overlooked gap in this discussion: conditional exchanges between voters and politicians could be interpreted as at least two types, systematically chronicled as empirically distinct.
Auyero (2000) describes clientelism in Argentina as a long-term party busi-ness. The author cites, for instance, the case of Lucia, a party broker to the Peronist machine connected to a patron named Matilde:
Lucia considers herself a friend of Matilde: “She always lends you a hand.” Lucia has known Matilde since 1984 [for more than a decade] and is a manzanera [block delegate] of the state-funded Plan Vida. Matilde also provides her with food [to distribute among her sup-porters].(Auyero, 2000, p.03).
Focusing on Peru, Muñoz (2014) provides a contrasting overview of clien-telism. For example, the author describes the handout of cash in matchboxes during rallies as an effective informational strategy during electoral campaigns:
Because there are a lot of vehicles, people are wandering around the cars. It is there where the compañeros ask them to climb in and give them a matchbox [with money inside]. Then, seeing this, a lot of people struggle to get into a car and as a lot of people gather, they start saying “this party is going to win.”(Muñoz, 2014, p.46).
The building of decades-long political connections possesses institutional and organizational implications that diverge completely from receiving on-the-spot payments, despite that both might be behaviors induced by conditional exchanges with voters and brokers. While Muñoz (2014) illustrates the universe of clientelist relationships in Peru as flexible, ephemeral, and leveraged by uncertainty, Auyero (2000) portrays its Argentinean counterpart as a closed and hierarchically structured rigid system. However, it is commonly accepted that clientelism operates differently in particular contexts – as social phenomena regularly do. What has been surpris-ingly unnoticed is that clientelism has been described to operate following similar
patterns of contrasting nature even within the same polities.
For example, describing clientelist relationships in India, Wilkinson (2007) cites that
At the local level, Congress politicians in the 1950s and 1960s typi-cally contracted votes through upper-caste local intermediaries, who used their social status and control of land, credit, and muscle power to contract to deliver local upper as well as lower-caste votes to the Congress candidate. [...] Patrons [...] invoked norms of reciprocity and obligation, and stressed the necessity for villagers to have con-tinued access to the power of the Congress dominated state. (Wilkin-son, 2007, p.114)
However, in the following decades the author states that inter- and intra-party com-petition in India grew, rendering a very different political landscape in the country: [t]he mere presence of stronger opposition parties after the late 1960s had an immediate effect in increasing the supply of goods delivered to voters. [...]. In villages where the Congress candidates now faced strong competition from the communists or Jana Sangh, the very low-cost gifts or threats of retaliation that had been used in the past no longer worked. Instead, candidates had to promise more in terms of cash, food, and promises of economic development for the village. (Wilkinson, 2007, p.114)
Similarly, a distinction between old and new clientelisms can be found in earlier studies (Mouzelis, 1985; Fox, 1994; Hopkin, 2006). Although India is a spe-cific case in which electoral laws enacted at the national level constricted competi-tion while leveraging it at state and local eleccompeti-tions (Wilkinson, 2007, p.117), in Peru (Muñoz, 2014), Colombia (Holland and Palmer-Rubin, 2015) and Romania (Gherghina and Volintiru, 2017) a similar pattern of transformation in clientelist relationships can be observed after democratization and competition increased, leading researchers to create distinct classifications for clientelist practices.
In Brazil, for example, Gay (2006) describes the strategy of Partido Democrático Trabalhista (PDT) post-democratization, which had elected as state governor of Rio de Janeiro Leonel Brizola in 1982. In Vidigal, an autonomous neighborhood asso-ciation focused on eradicating conditional exchanges with politicians is depicted by the author as only accepting public funds and projects that did not require recipro-cation. The author observes, though, that despite that PDT representatives seemed to comply with the association requirements, this was more or less empty rhetoric: politicians effectively asked for votes when left alone with voters. And voters, despite the efforts of association leaders, rewarded the PDT with their votes.
While the PDT’s strategy described by Gay (2006) resembles clientelism, benefits were distributed to a collective entity, and it did not necessarily depend
on individual compliance. Clientelism was exerted in a very thin relationship with pork barrel politics in this case, and distanced from the traditional clientelist type of practices depicted in the literature. Given these differences in empirical implemen-tations of clientelism, is the phenomenon the same? Or is there a conceptual line to be drawn? Could we take advantage of this variation for building explanatory models about clientelism. The objective of this Chapter is to investigate these concerns.
I review literature on clientelism to argue that the phenomenon is better framed as two different classes of conditional exchanges that vary according to the competitiveness1 of political systems. This discussion contributes to the literature
by highlighting why in many instances monitoring and threats of punishment are not credible in the presence of open competition. I use this distinction on compe-tition levels to classify two types of clientelism: traditional and short-term. Here I conceptualize and review both types separately, and discuss the implications of this classification for the concepts of clientelism and vote buying.
Political machines seem great at determining electoral outcomes before vot-ing booths open, but poor at resistvot-ing competition. They can constrict electoral competition by exerting dominance over other organizations and controlling re-sources, rendering greater incentives for their brokers and voters to support them. In this Chapter, I assume that in the typical embodiment of clientelist relationships – denominated traditional clientelism - society organizations enhance the efficiency of clientelist strategies because competition is low and patrons exert control over these organizations, structuring vertical relationships inside bailiwicks that repel contes-tants, strengthening monitoring mechanisms and reinforcing the ongoing nature of reciprocal relationships and benefits distributed to the community. When compet-itiveness increases, traditional patrons suffer from disputes inside their bailiwicks, impeding (or diminishing the relevance of) control mechanisms. Eventually, this could lead towards the dismantling of their political machines. Thus, in my theory of the two clientelisms, traditional patrons would tend to opt out of politics when they lack mechanisms to control competition.
In competitive environments, the traditional clientelist strategy is compli-cated. Without the vertical control of organizations, patrons lose their ability to se-cure electoral success for certain. I argue that when facing electoral uncertainty, patrons’ monitoring of voters might become a less-optimal mechanism to sustain 1I use the concept of competitiveness to embody not only the presence of more than one effective
con-testant at elections (parties or candidates), but also the availability of participatory mechanisms for voters, such as secrecy of ballot, freedom of civil society organization and individual political liberties.
political power. Additionally, competitiveness possibly increases the price of votes in electoral disputes, leading to greater costs (Corstange, 2017) or a “war of attrition”, in which candidates’ offers increase until their resources are depleted (Maskin and Tirole, 1988). In such contexts, civic society organizations can act as professional bro-kers, helping candidates to structure conditional exchanges of the short-term type. A market for votes emerges when fragmented organizations work to distribute con-ditional exchanges for distinct candidates. Despite competitiveness, these networks are able to find a temporary equilibrium in the short-term that leverages candidates political support. Thus, short-term clientelism is likely to spread and thrive more as competitiveness increases (Muñoz, 2014).
If this intuition is correct, scholarly research might be focusing on questions that do not help us solve the puzzle of clientelism in competitive and uncompetitive systems alike. To consider which voters are targeted and what mechanisms parties and brokers implement for monitoring is valid and important. However, without un-derstanding the level of competitiveness candidates face in different territories, how clientelist networks are structured, and how they overcome issues related to elec-toral uncertainty and price increases, the enlightenment provided by these former questions may become limited. If the relationship between clientelism and compe-tition is inherently endogenous, voters can be monitored and targeted differently in distinct types of environments, for instance (Corstange, 2017). More importantly, clientelist strategies of the traditional or the short-term type could prevail supported by distinct sets of institutions in competitive and monopolistic settings.
2.1 Punishment, monitoring and uncertainty
In this section, I review the literature and explore theoretical relationships between clientelist threats of punishment, monitoring, and uncertainty. While most of the scholarly debate has focused on the issue of machines’ monitoring of voters, it over-looks the implied assumption that voters - when monitored - will get punished even if the clientelist candidates they compromised for are not elected. I believe this as-sumption is unrealistic: in most places, patrons depend on government revenues themselves to punish or grace their followers. Nichter (2018) for instance finds that in local elections in Brazil “citizens who declared support for victorious candidates are widely viewed as experiencing favoritism with health care, employment, and water.” (Nichter, 2018, p.127).
Although clientelism harms voters, the practice involves a certain degree of voluntarism (Kitschelt, 2000), which becomes greater as electoral surveillance and accountability systems improve and ballot secrecy is enforced in democracies. Monitoring of voters’ behavior by parties, even if imperfect, has been found to ex-plain clientelist compliance, especially in the context of continuous exchanges and long-term interactions (Stokes, 2005). Other mechanisms pointed to include re-ciprocal cultural ties (Schaffer and Schedler, 2006; Hicken, 2007), repeated inter-actions (Schaffer and Baker, 2015; Greene, 2015), declared support and vulnerability (Nichter, 2018).
When it comes to financing clientelist exchanges, Roninger notes that clien-telism is likely to be financed either by state or private resources (Roniger, 2011). Clientelism financed by public expenditures includes the usage of public jobs, public goods construction, public contracts, and others in exchange for votes, while private resources configure those sponsored directly by the candidates and parties through their electoral campaign finances, usually gathered from private sector companies. Though clientelist politicians use both types of revenues interchangeably, in most cases, the advantages of clientelism seem better grasped by those parties and can-didates who dispose of a larger pool of public resources. Usually, privately financed clientelism encompasses the offers of money, consumer goods, and services, while publicly financed is distinct in what goods can get distributed through public money. For example, patrons can offer their clients access to public jobs and privileges within the state bureaucracy only if they hold office.
Examples of of clientelism in which dominant parties use some combina-tion of discrecombina-tionary public funds and competicombina-tion restriccombina-tion to enhance their elec-toral support include the People‘s Action Party (PAP) in Singapore (Hicken, 2007), the Partido Revolucionario Institucional (PRI) in Mexico (Fox, 1994; Magaloni et al., 2007; Greene, 2007, 2015) and the Peronist Party in Argentina (Stokes, 2005; Auyero et al., 2009; Zarazaga, 2014; Weitz-shapiro, 2016). Party patronage, another form of publicly funded clientelism, has also been found to be prevalent in contemporary Austria (Ennser-Jedenastik, 2014).
Medina and Stokes (2002) assumes that incumbents “exercise [exclusive] monopoly control over a good that is useful to voters”(Medina and Stokes, 2002, p.03), citing a factory owner who distributes jobs as an example. From this baseline, their model evolves to “a result of reduced competition from an assumption of initial monopoly control by one competitor”(Medina and Stokes, 2002, ibidem). This assumption and its corollary implication impose complicated challenges. It is possible that reduced
political competition establishes first in certain areas, and after it patrons could use the monopolistic control of resources provided by government offices in exchange for votes. There is a chance that patrons do not have much to offer voters individu-ally - only through their control of public resources, their power becomes effective. This situation occurs whenever government resources are fundamental for parties’ clientelist strategies.
When clientelist conditional exchanges are funded mainly with public re-sources, the implicit puzzle is: why, if other parties or candidates could have access to these resources (after winning elections), does the power of incumbent patrons remain unchallenged? Although the authors state that monitoring of voters is suf-ficient to guarantee voters’ compliance, if a patron loses office, her monopoly over resources vanishes. Thus, threats to punishing defection might become unfeasible. Answering the former becomes less problematic when we conjecture Medina and Stokes (2002)’s assumption the other way around: competition reduction establishes first, and later, dominant parties or politicians achieve a monopoly of resources.
If elections are competitive, at each electoral period, the incumbent is subject to lose her monopoly and thus the exclusive usage of resources, as election outcomes are probabilistic. When political competition is not restricted, chances exist that in-cumbents will not be able to remain in office in the future - even if they use clien-telism. If the ongoingness of clientelism is necessary for voters to comply (Nichter, 2014; Stokes et al., 2013; Hicken, 2011), parties and candidates have to signal effectively every electoral turn to these voters that they will be winners of the contest -otherwise, how could they accomplish their promises of ongoing benefits? And how could they threaten to punish defective voters? Thus clientelist incumbents’ ulti-mate compliance enforcement strategy - the monitoring of voters’ behavior - might become uncertain or insufficient if these candidates are not expected to win office.
Spending on dominant areas is crucial for hegemonic party rulers (Corstange, 2017; Greene, 2015). It helps to cultivate their dominance by producing an “aura of invincibility” and deterring elite divisions, opposition candidates’ entry, and clients’ defections, while also preventing future competition (Diaz-Cayeros et al., 2003; Blay-des, 2011; Corstange, 2017). Moreover, the greater parties and candidates exert con-trol over competition and exit options of voters, the less they distribute in clientelist contracts to the needier of voters. As Corstange (2017) study underscores, in monop-sonic politics, the absence of competition in the usage of clientelist practices “enables the machine to focus on poor and politically disinterested voters whose low reser-vation prices make them most likely to change their electoral behavior in exchange
for modest payoffs”(Corstange, 2017, p.72). Thus, machines who walk the paths alone towards control of the safe’s keys are more likely to target poorer voters – and to buy them with less effort.
Monopolizing resources and institutions is advantageous to traditional clien-telist patrons because threats of punishment tend to become more effective in polit-ically monopolized contexts. As expressed by Stokes (2009)’s model, the probability of punishment is part of the equation that affects the behavior of voters in clientelist systems. Especially for traditional clientelist strategies funded with public resources, monitoring may not always guarantee a sufficient mechanism for parties and can-didates to assure conditional exchange efficiency as an electoral strategy. If a clien-telistic candidate is not expected to win (or at least be competitive) during elections, how can she credibly threaten to punish her defective voters afterward?
The so-called "Down’s paradox" correlates with this issue, in the sense that if no particular vote is likely to be decisive, agents can take electoral results as exoge-nous. If the expected returns from voting for one candidate are negligible in large electorates, why would the expected costs (or punishments) be significant - even if monitoring is high?
[U]nless his vote actually decides the election, it does not cause the “right” party to be elected instead of a “wrong” party; whether or not the “right” party wins does not depend on how he votes. Therefore, voting "correctly" produces no gain in utility whatsoever; he might as well have voted “incorrectly”. [...] The probability that any one citizen’s vote will be decisive is very small indeed. It is not zero, and it can even be significant if he thinks the election will be very close; but, under most circumstances, it is so negligible that it renders the return from voting “correctly” infinitesimal. (Downs, 1957, p.146) However, when a voter knows for certain which candidate will be elected, and that this candidate has an imperfect mechanism for checking how she voted, then it is to her benefit to cast a ballot for the candidate – even if the electoral results are not defined by her single vote. Voters get more benefits during and after elections by supporting this clientelist candidate (Nichter, 2018), and defecting imposes non-negligible threats of punishment. It is also true that the candidates’ monitoring can enhance voters beliefs about their chances of success at the polls.
If the candidate holds a monopoly over certain resources to distribute re-gardless of government, then monitoring becomes the most salient mechanism for assuring support. But in most contexts of traditional clientelist strategies, punish-ment threats are credible because in ere campaigns clientelist candidates display their strength, controlling competition and endorsing the electorates’ beliefs about their
likelihood of success while monitoring how voters have behaved in voting booths. Inside bailiwicks, from the political monopolies established, patrons can prevent the entrance of competitors and diminish incentives for voters to defect. Never-theless, if no previous control exists, elections can be viewed as completely proba-bilistic events; thus the chances of a defective voter being punished diminish with the number of candidates in dispute - regardless of the monitoring level performed. A simplistic example should suffice to stress this reasoning in detail.
Consider that a voter receives an offer from every clientelist candidate who approaches her (or that she voluntarily approaches). This voter is always a defector: she never fulfills the promise, voting otherwise or not voting at all. How much her expected probability of punishment is, given electoral results are probabilistic and that the candidate (or candidates) she defects from can only punish her if they hap-pen to win the dispute? Assume there exist a different number of candidates, clien-telist offers and electoral district magnitudes, such as the following set of constraints demonstrates:
M = {1, m}
C = {2, m + 1}
M ≤ C
O = {1, C − 1}
Where M is the district’s magnitude, C is the number of candidates in dispute and O is the number of clientelist offers the voter accepts. If M = 1 or O = 1:
P (P unishment|Def ection) = O C∗ p
where p is the probability of monitoring. Imposing that M 6= 1, O 6= 1 and that O 6= C, considering the set of constraints above, it is always true that M > C − O and thus:
P (P unishment|Def ection) = P(PO∗ PC−O∗ p
−N)
PC
Where PO = ON
is the probability of electing O clientelist candidates to N offices; PC−O = M −NC−O
is the probability of electing C − O non-clientelist candidates to M −N offices and PCis the probability of electing any given candidate to office, MC
. Additionally, N = {1, . . . , O} if M >= O and N = {1, . . . , M} if M < O. Assume p = 0.9 and that it is equal for all candidates - i.e., all clientelist candidates can monitor with
almost complete certainty the choices of their voters. Figure 2.1 depicts partial results for this analysis2.
Figure 2.1:Probability of Punishment and Number of Candidates - Different District Magnitude and Number of Clientelist Offers.
0.00 0.25 0.50 0.75 0 25 50 75 100 Number of Candidates Expected Probability 0.25 0.50 0.75 1.00 Client. Ratio (a)Magnitude of 1 5 8 10 2 3 4 25 50 75 100 25 50 75 100 25 50 75 100 0 25 50 75 100 0 25 50 75 100 0 25 50 75 100 0.25 0.50 0.75 1.00 0.25 0.50 0.75 1.00 0.25 0.50 0.75 1.00 0.25 0.50 0.75 1.00 0.00 0.25 0.50 0.75 0.25 0.50 0.75 1.00 Number of Candidates Expected Probability 0.25 0.50 0.75 Client. Ratio
(b)Distinct Electoral Magnitudes. Source: Authors’ elaboration.
We can see from Figure 2.1 that the more candidates there are in any given election, the lower the expected punishment voters experience is. This probability decreases less rapidly in districts of greater magnitude, when compared to lower-magnitude districts. It occurs because the probability of any candidate to be elected is increased with the number of seats. Punishment probability of our hypothetical 2Results were calculated using an algorithm in R statistical software, following the formulas presented
voter also increases with higher proportions of clientelist offers in comparison to the total number of candidates - which is an implication of the imposition of consider-ing a voter who always defects. Considerconsider-ing these restrictions, results thus exem-plify the principle that the more candidates there are in a dispute, the worse-off any clientelist candidate would be in making credible threats of punishment - assum-ing this punishment depends on their probability of beassum-ing elected. This model thus evidences the point that if candidates cannot provide means for voters to believe they will be elected, monitoring alone is not sufficient to guarantee that their threats provide sufficient mechanisms for enforcing voters’ behavior, regardless of levels of monitoring.
It might seem contrasting that the less competitive elections are, the more voters would be inclined to participate and vote following their patrons’ agreements. However, in a restricted political environment, showing support for a dominant pa-tron is an effective way voters encounter to obtain resources distributed selectively, to be protected from state authority (Holland and Palmer-Rubin, 2015) or to get in-cluded in the reciprocal chain of recurrent exchanges that regularly protects them against vulnerability and poverty (Auyero, 2000; Magaloni et al., 2007). In a compet-itive environment where elections are probabilistic, informational costs are greater for voters and candidates. Patrons can lack particular mechanisms for threatening and monitoring efficiently, and this challenges their prospects of being successful. In this settings, other mechanisms such as those proposed by Muñoz (2014) and Nichter (2018) might become more efficient.
Another problem for traditional patrons is that, when competitiveness in-creases, they may face contesting opposition from other clientelist candidates inside their bailiwicks. In market situations, wars of attrition occur whenever firms engaged in a highly competitive dispute start pricing at marginal cost, hoping that the com-petitor will relent and step-away from competition. In clientelist politics, this situa-tion would occur whenever two or more candidates cyclically increased their offers to voters, expecting to attract their opponents’ electorate with marginally greater monetary (or other) incentives.3 Traditional patrons hold a competitive
disadvan-tage in attrition wars: their cost caps include not only the payments dispensed during elections, but also the costs of keeping organizations and an ongoing flow of benefits to their supporters. Even if short-term clientelist candidates and traditional patrons had the same initial amount of resources to spend (which we do not know is true), the former would be at an advantage during electoral periods. Short-term clientelist
candidates do not deal with costs of selective distribution in the long-term,4 which
facilitates over-spending during elections, for instance.
Furthermore, in competitive elections, candidates and voters alike update their expectations about probable winners at each instant, following the natural in-coming flow of information from political campaigns. Thus, those who spend more could be viewed as those more inclined to be successful (Muñoz, 2014). While tradi-tional clientelist patrons are unlikely to survive in this scenario (at least unless they can repel contestants), short-term clientelism and vote buying could be one of sev-eral strategies candidates employ for effectively informing audiences about their chances of victory (Muñoz, 2014). In my theory of the two clientelisms, traditional patrons are defined by the control exerted over competition and their monopoly over resources, while short-term clientelists typically use independent organizations to bargain block votes through one-shot conditional exchanges. In the following sec-tions, I define and review each concept in detail.
2.2 Traditional patrons as monopolists: PRI in Mexico
and Peronists in Argentina
Following Scott (1972); Weingrod (1968); Eisenstadt and Roniger (1984); Graziano (1976) and the long tradition of studies that conceptualized clientelist relationships as asymmetrical, iterative and conditional, Magaloni et al. (2007) develops a the-ory of electoral investment focused on different portfolios delivered by parties, fur-ther demonstrating its usefulness in the context of the erosion of PRI rule in Mex-ico. The authors’ intuition is that, relative to programmatic public goods provision, clientelism’s attractiveness depends on the extent of poverty levels and the degree of electoral risk experienced by politicians. In places where low levels of economic development make it harder for parties to develop loyal supporters, the return of electoral strategies is riskier, and thus parties opt for clientelist exchanges, as condi-tionality guarantees the return of their investments. Therefore, “clientelism is a po-litical investment strategy designed to deter voter exit and, simultaneously, to hedge electoral risks when more investment in public goods is required to win elections” (Magaloni et al., 2007, p.72).
For the authors, the fiscal decentralization introduced in Mexico during the 4It is unclear how short-term clientelists deal with price pressures empirically. I conjecture that
1990s was essential to give PRI’s incumbent candidates an electoral advantage, by monopolizing state resources that allowed them to expand the party machine (Ma-galoni et al., 2007). More generally, the authors have found that even within the context of its power erosion, PRI could keep its dominance in localities with mod-erate levels of economic modernization and lower levels of political competition (Magaloni et al., 2007). Nevertheless, the authors claim that endogeneity between development and competition controlled for in their model is likely to be overesti-mated, as no causal or quasi-experimental design was considered in the analysis.
However, the theory is innovative in that they explore the connection be-tween clientelism, monopoly, and voter exit (Hirschman, 1970). It seems intuitive that in places where poverty rates are high, and incumbents have access to pub-lic resource control, they will use such resources to diminish their electoral risk by offering individuals selective benefits, in contrast to public goods. The incumbent party’s decision is then either to invest in clientelist goods or programmatic goods (in which access to citizens occur regardless of their voting behavior). The condi-tional character of clientelist exchanges imposes that the electoral benefits parties extract from public goods allocation are always riskier than the electoral benefits ex-tracted from clientelism. As per definition, the handout in clientelism would always be compensated by electoral support (Magaloni et al., 2007). Moreover,
[i]n smaller and isolated localities there was often not even a menu of electoral choices. Even in the presence of some opposition, often all that the PRI really needed to do to get peasants to cast a vote for the party was to pay for their transportation, because peasants’ “sincere” preference was the PRI.(Magaloni et al., 2007, p.186).
By “sincere preference” the authors mean that “by threatening, whether ex-plicitly or not, to suspend or withdraw the transfers that peasants needed, the PRI thus managed to deter rural voters from supporting another party or engaging in any form of open confrontation with the regime” (Magaloni et al., 2007, p.197). We can see clearly from this statement that, although the authors highlighted earlier that the fiscal concentration of resources was crucial to advance the PRI machine, there was an absence of opposition and real competitiveness between parties. Wherever some level of competition existed, PRI likelihood of being elected was so high that it rendered its threats of conditional transfer withdraws to prevent voters from voting otherwise effective.
Magaloni et al. (2007) also show that, as incumbents become more concerned with their electoral risks and the return of their investments, they tend to invest more in clientelist strategies in poor places, whereas they would invest more in public
goods in richer localities. Implicit in their theory of clientelism is the role of electoral uncertainty on clientelist practices. Because compliance of voters is never certain, to keep their hegemony over electoral results, dominant political parties must control competition, promoting a continuous flow of handouts that, by itself, might keep voters from defecting. Thus, making the chances of opposition victory a very thin promise for voters, individual compliance is assured. Ultimately, the authors argue that the monopoly of resources and control of competition are different mechanisms that drive the association found between poverty and clientelism (Magaloni et al., 2007).
It is important to notice that the country’s fiscal rules and decentralization had given the incumbent party a significant level of monopoly over political life, only threatened by the larger, richer and more heterogeneous cities. Additionally, as Weitz-Shapiro (2012) has put it, the measurement strategy used by the authors does not capture clientelism directly, as it focuses on the classification of private versus public goods distributed by the Mexican PRONASOL policy program. As the authors state, “clientelism, [. . . ] is a method of distribution and does not map directly onto conventional distinctions between private and public goods” (Weitz-Shapiro, 2012, p.02) .
Moreover, in contexts where party systems are fragmented rather than dom-inated by a single party-machine and clientelism is used by different parties it is not straightforward whether the results observed in their study of Mexican Prona-sol would hold (Corstange, 2017). It is rather tautological to state that the incumbent party monopoly at the local level sustained because a credible threat of punishment inhibited exit from the relationship when the monopoly over the threat mechanisms was assumed in the first place (Medina and Stokes, 2002). We could further ask: what explains this threat? And why are parties other than the incumbent not allowed to use it?
When we look deeper into Magaloni et al. (2007)s’ model of clientelism, what it seems to suggest is a situation in which, at state and federal levels, an authoritarian party that remained in power for decades controlled economic and fiscal resources, using clientelist exchange channels to keep opponents at bay in local elections - a tactic especially efficient in poor places. Control of fiscal resources is also a major explanatory variable of Calvo and Murillo (2004)s’ model of clientelism, where they explain returns of patronage by highlighting differences in political parties’ access to resources and voters’ dependence on public sector jobs in Argentina (Calvo and Murillo, 2004).
The fiscal advantage was found, in the Argentinean case, to be a consequence of economic institutions that favored the clientelist party dominating provinces and, at the same time, majoritarian rules that favored an over-representation of these po-litically controlled units (Calvo and Murillo, 2004). Thus, controlling resources is a fundamental tool for hegemonic parties. As handouts were distributed continuously, and no other party threatened their dominance over government resources, voters had incentives to support the machine to keep receiving benefits. At the exact mo-ment in which opponents menaced their hegemony over electoral results, it seemed that the Mexican PRI, as well as the Argentinian Peronist party, had incentives to in-vest more in public goods (Magaloni et al., 2007; Weitz-Shapiro, 2012; Greene, 2015). According to Auyero (2000), in Argentina “unidades básicas [(UB)], political brokers, and state-funded programs have become the sources of resources that cir-culate in the informal networks of survival”(Auyero, 2000, p.61), where “unidades básicas” are grass-roots level organizations under the control of the Peronist party since the 1950’s (Aelo, 2016). Aelo (2016) shows that “[i]n the early years of Peronism, the “unidad básica” - together with the athenaeums, centers, etc. - grouped indi-viduals by ideological-political affinity. Given these possibilities, it was common for leaders to “open‘unidades básicas’ at their own expense.””(Aelo, 2016, p.617).5
Even-tually, “the Superior Council would try to “bring order” [to the organizations’ system] by arranging the center to be called “unidad básica”, and that would only be so once “authorized” by the respective district interventions.”(Aelo, 2016, p.610). Thencefor-ward, the author claims that these civic organizations’ autonomy diminished.
This possibility [of autonomously creating unidades básicas] was grad-ually canceled after 1952, as the “unidades básicas” had a defined ter-ritorial scope, and individuals could [only] join in the one closest to their home. It seems, also, that the number of authorized “unidades básicas” tended to be reduced. These organizations also underwent functional modifications. Although the Peronists had understood, at least since 1949, that the “unidades básicas” were not mere “commit-tees” and should not be devoted solely to political action, during this period they emphasized cultural, sports, social assistance and chan-neling of neighborhood demands [...].(Aelo, 2016, p.618)
What the author calls “los métodos verticales adoptados por la cúpula par-tidaria” (Aelo, 2016, p.618) seem to have been effective in creating a political control of conditional exchanges inside poor Argentinean territories that were found func-tioning almost five decades later by Auyero (2000), centered at “unidades básicas”. Considering the Argentinean case, Weitz-Shapiro (2012)’s analysis is of particular in-terest. The author stresses that clientelism has the implicit costs of diminishing elec-toral support from non-poor constituencies, while it generates votes from the poor. This argument further helps to explain why larger and richer municipalities were
more competitive and less clientelist than smaller ones. Considering the costs and political opposition of mayors in local assemblies, Shapiro finds that the interaction between poverty and competitiveness explains variation in the usage of clientelist strategies by politicians. Clientelism is measured here as the likelihood of a mayor to personally influence the beneficiaries list of a food program in Argentina, as cap-tured by an open survey question performed with informants. In the words of the author:
[w]here political competition is high, each vote gained or lost from clientelism will be felt quite acutely. High competition should be compatible with clientelism when voters are mostly poor, and the electoral benefits of clientelism outweigh the costs. When high levels of political competition are coupled with a large middle class, how-ever, the electoral costs of clientelism should outweigh the benefits. In contrast, this trade-off between the electoral costs and benefits of clientelism should be less relevant in contexts of limited politi-cal competition. Where a politician believes his security in office is assured, the votes lost and gained because of clientelism are unlikely to determine a politician’s future in office. Under circumstances of limited political competition, then, the demographic makeup of his constituents should have a more muted effect on a politician’s deci-sion about whether or not to use clientelism. (Weitz-Shapiro, 2012, p.08)
The author predicts that “nonpoor voters are likely to punish politicians who engage in it [and that] [t]he salience of these costs should vary with the share of the population that is nonpoor and a politician’s security in office” Weitz-Shapiro (2012, p.36). Nevertheless, what the author finds in her analysis is that while only popu-lation size had a significant and negative effect on clientelism in the absence of the interaction term, poverty levels of municipalities were only significant predictors of clientelism when moderated the effect of competition. The fact that poverty did not have a significant effect on clientelism per se contrasts with the underlying im-plications of the author’s theory, and it is left unexplained. If nonpoor voters induce politicians to opt out of clientelism, then we should expect a significant effect of poverty on the definition of clientelist strategies.
Both Magaloni et al. (2007)s’ and Weitz-Shapiro (2012)’s analyses of clien-telism involved the usage of clientelist strategies by dominant and centralized single-party machines that kept their electoral dominance through clientelism and control of fiscal resources in poor municipalities and verticalized organizations at the grass-roots level. Clientelism, in these contexts, could become more a tool in the exercise of furthering political control than a persuasion strategy. Thus, clientelism is not only affected by the levels of political competition, it could also be a mechanism to control it (Hicken, 2011; Calvo and Murillo, 2004) – which suggests an endogenous relationship between competition and clientelism. Furthermore, if we consider that
the usage of clientelist strategies in the long run also impacts poverty levels, the paths through endogeneity in this relationship could be further complicated (Schaffer and Baker, 2015).
Although both papers deliver very significant theoretical contributions, they seem to point to the difficulty of disentangling the effects of competition control and poverty over clientelism, rather than presenting a unanimous test of such the-ories. In contexts where clientelism is a major strategy of a single-party machine, it has incentives to restrict competition, verticalize organizations, and control fiscal resources. In combination, these strategies support machines’ strategies to main-tain political support through ongoing conditional handouts of benefits to voters. In such contexts, clientelism is both a regime legitimization and a control strategy (Blaydes, 2011). Both depictions of clientelism suggested the establishment of “cap-tive” constituencies and sustained long-term interactions within clientelist politics, where dominant patrons keep their bailiwicks by controlling resources and keeping opponents at bay - probably with the aid of carefully designed institutional rules and territorial dominance.
When democratization and competitiveness challenge political monopolies, the certainty once established over electoral results is fickle. We can thus conclude that traditional clientelism should become a less feasible strategy, more costly and even altogether impossible, as candidates would have incentives to target richer con-stituents (Corstange, 2017), with more uncertain gains and less credible threats. Both in Argentina and Mexico, the PRI and the Justicialistas (PJ), respectively, have lost several important electoral disputes at the national and sub-national levels in recent years. However, what remains active of their former single-party machines orga-nizations in the contemporary politics of both countries – and why – is a question for further research. In Chapter Three of this dissertation, I analyze descriptions of clientelist relationships across different regime types in Brazil. It is conjectured that if the hypothesis of this study is reasonable – and clientelism changes with the level of competitiveness – we should find traditional clientelism as a common strategy used by candidates in authoritarian periods and short-term clientelism arising as an electoral strategy in democratic contexts.