Electoral Studies 35 (2014) 315–327

Contents lists available at ScienceDirect

Electoral Studies journal homepage: www.elsevier.com/locate/electstud

Conceptualizing vote buying Simeon Nichter* Department of Political Science, University of California, San Diego, Social Sciences Building 301, 9500 Gilman Drive, #0521, La Jolla, CA 92093-0521, USA

a r t i c l e i n f o

a b s t r a c t

Article history: Received 19 September 2012 Received in revised form 2 March 2013 Accepted 25 February 2014 Available online 16 March 2014

This study investigates the concept of vote buying, with a particular focus on its usage in research on clientelism. Vote buying is often poorly defined. Such conceptual ambiguity may distort descriptive findings and threaten the validity of causal claims. Qualitative analysis suggests that researchers often employ the concept of vote buying differently, and regressions from Nigeria and Mexico suggest that using alternative definitions can yield divergent empirical results. This diverse usage also poses the risk of conceptual stretching, because scholars often use vote buying to describe other phenomena. To improve future research, analysts should pay close attention to the conceptualization of vote buying. Ó 2014 Elsevier Ltd. All rights reserved.

Keywords: Vote buying Clientelism Machine politics Political machine Conceptualization

1. Introduction Use of the term “vote buying” has increased sharply in recent decades. Its mention in published books has quintupled since 1980 (see Fig. 1), and over 10,000 recent academic articles and unpublished manuscripts mention the term.1 This study investigates the concept of vote buying, with a particular focus on its usage in research on clientelism. We build on influential qualitative work that reveals how conceptual ambiguity can undermine scholarly research (e.g., Sartori, 1970; Collier and Levitsky, 1997; Levitsky, 1998). Although many recent studies continue to advance our understanding of clientelism, they are often imprecise about what constitutes vote buying. This lack of conceptual clarity may distort descriptive findings and threaten the validity of causal claims. Our qualitative analysis suggests that researchers often employ the concept of vote buying differently, and regressions from

* Tel.: þ1 858 354 1854; fax: þ1 858 534 7130. E-mail address: [email protected]. 1 The caption of Fig. 1 describes the data about the growth in usage of the term “vote buying.” The number of scholarly works using the term is from Google Scholar (March 2014). http://dx.doi.org/10.1016/j.electstud.2014.02.008 0261-3794/Ó 2014 Elsevier Ltd. All rights reserved.

Nigeria and Mexico suggest that using alternative definitions can yield divergent empirical results. Diverse use of the term “vote buying” also poses the risk of conceptual stretching (Sartori, 1970). Scholars employ the term to describe various political phenomena, such as paying cash to voters on Election Day (Lehoucq, 2007), inducing legislators to support NAFTA (Evans, 2004), increasing pensions for all elderly citizens (Thames, 2001), and paving roads in co-ethnic districts (Burgess et al., 2012). We develop a typology of four distinct ways in which vote buying is used in the scholarly literature, and argue that two of these categories involve conceptual stretching. Studies should clarify how they use vote buying in order to reduce conceptual ambiguity, as well as to improve descriptive and causal inference. They should also pay close attention to potential heterogeneity, as predictions and findings do not necessarily apply across categories of vote buying. The present article aims to alleviate conceptual ambiguity about vote buying, with a particular focus on how the concept is used in the field of clientelism. To this end, we: (1) identify key differences in how recent studies define clientelist vote buying; (2) examine how these different definitions can affect empirical results; (3) develop a typology of the broader usage of vote buying in political

316

S. Nichter / Electoral Studies 35 (2014) 315–327

additional examples of studies examining clientelist vote buying are discussed below. Before investigating how other studies define clientelist vote buying, we first emphasize two key points of this paper: (1) scholars should make conscious decisions about what attributes to include and exclude in their own definitions of clientelist vote buying, and (2) they should provide clear and explicit definitions. At the outset, we follow this guidance by providing our own systematized concept3: Fig. 1. Relative frequency of “vote buying” mentions in books.

science; (4) discuss the risk of conceptual stretching and potential heterogeneity; and (5) emphasize the need to differentiate clientelist vote buying from other forms of clientelism. 2. Clientelist vote buying: unpacking the concept Many studies employ the concept of vote buying when discussing clientelist linkages between elites and citizens. During campaigns in many countries, clientelist parties (or political machines) deliver material inducements to individual or small groups of citizens in exchange for political support. As Kitschelt and Wilkinson (2007) emphasize, a defining characteristic of clientelistic exchanges is that “the politician’s delivery of a benefit is contingent upon the actions of specific members of the electorate” (10, italics in original). Studies of clientelism that use the term “vote buying” often underscore that the contingency of benefits requires parties to overcome the threat of opportunistic defection by voters (e.g., Schaffer and Schedler, 2007: 20; Vicente and Wantchekon, 2009: 301).2 In order to ensure that recipients actually comply with vote-buying agreements, analysts frequently contend that machines engage in monitoring and enforcement once they distribute selective benefits. An example of a study that focuses on clientelist vote buying is Stokes (2005). Stokes argues that during elections in Argentina, the Peronist party distributes rewards to weakly opposed voters in exchange for switching their vote choices. She closely examines mechanisms that facilitate these contingent exchanges. Stokes argues that the Peronist party uses its “deep insertion in voters’ social networks” to violate the secret ballot, and is therefore able to enforce compliance when buying citizens’ votes (315). Similarly, Lehoucq’s (2007) study of clientelist vote buying in nine countries highlights the importance of enforcing contingent exchanges. Lehoucq argues that “due to the principalagent problems inherent in vote buying,” parties will only engage in vote buying if they can monitor how citizens vote in order to “ensure that bargains are kept” (42). Numerous

2 It should be emphasized that the present article focuses exclusively on studies that explicitly refer to vote buying. Some scholars who study clientelism, but eschew the concept of vote buying, do not have a strong focus on opportunistic defection. Examples include Auyero (2000) and Levitsky (2003), who never employ the term “vote buying.”

Clientelist vote buying is the distribution of rewards to individuals or small groups during elections in contingent exchange for vote choices. Rewards are defined as cash, goods (including food and drink), and services. Post-election benefits, employment, public programs, and transportation to the polls are not considered rewards. As explored below, the attributes included in our definition are mentioned in many – but by no means all – existing studies. Future studies on the topic would ideally share a common systematized concept of clientelist vote buying. If scholars do not adopt a mutually agreed-upon systematized concept, they should at least be explicit about how they define vote buying and discuss potential implications. Unfortunately, conceptual ambiguity is common in the study of clientelist vote buying. At a most basic level, studies of clientelism often report survey or fieldwork evidence about the relative prevalence of clientelist vote buying, but it is sometimes unclear what specific attributes of benefits are considered. For example, does clientelist vote buying refer strictly to the payment of cash, or does the contingent distribution of goods, services, public program benefits and even employment qualify? With the goal of improving future research on the topic, this section examines the existing literature and highlights key similarities and differences in how researchers define the term. Fig. 2 provides an overview of the defining attributes employed by 15 studies. Given that vote buying is only considered “clientelist” if recipients agree to deliver votes in exchange for selective benefits, studies are only included if their definition of vote buying involves contingent exchange. Although these studies are not in any rigorous sense a representative sample of research on clientelist vote buying, they include works by all contributors to Frederic Schaffer’s (2007) edited volume on the topic, as well as other recent and/or frequently cited works. Given that these authors’ definitions of clientelist vote buying are sometimes explicit and sometimes implicit, the total counts for each row and column should not be taken as a precise summary. Rather, they are intended to provide an overall sense of the number of attributes considered by these particular authors, and the approximate importance of different attributes in the broader discussion of

3 A “systematized concept” refers to “the specific formulation of a concept adopted by a particular researcher or group of researchers” (Adcock and Collier, 2001: 530).

S. Nichter / Electoral Studies 35 (2014) 315–327

317

Fig. 2. Attributes included in explicit or implicit definitions of clientelist vote buying.

clientelist vote buying. While many other studies are analyzed in this article, the research included in this figure involves relatively explicit conceptualizations of clientelist vote buying. A key defining attribute of clientelist vote buying is its timing. Scholars typically emphasize that exchanges are not only ex ante in that benefits are distributed prior to voting, but also that exchanges occur on or soon before Election Day. All 15 studies in Fig. 2 concur that delivering contingent benefits during electoral campaigns constitutes vote buying. For example, Lehoucq (2007: 33) argues that “(b)uying a vote is trading something of value – usually, but not only, cash – for someone’s choice on election day.” Similarly, Schaffer (2007: 5–6) explains that “vote buying

is a last-minute effort to influence electoral outcomes, typically taking place days or even hours before an election, or sometimes on election day itself.” In fact, some researchers argue that timing is a key element differentiating vote buying from other concepts such as pork or patronage (Hicken, 2007: 51; Schaffer, 2007: 5–6). At a basic level, researchers concur that vote buying includes the delivery of selective benefits during electoral campaigns. In contrast with the broad consensus about contingent benefits during campaigns, there is more contention about whether promises of future benefits constitute clientelist vote buying. On the one hand, twelve studies in Fig. 2 exclude future benefits in their conceptualization of

318

S. Nichter / Electoral Studies 35 (2014) 315–327

clientelist vote buying. Such scholars often explain that the very fact that citizens receive benefits before voting is why opportunistic defection threatens clientelist vote buying (e.g., Stokes, 2005: 315; Lehoucq, 2007: 33, 42; Vicente and Wantchekon, 2009: 301). They argue that because citizens vote after receiving benefits, political machines must monitor citizens in order to ensure that they follow through with their side of the agreement. By contrast, three studies in Fig. 2 explicitly state that promises of future rewards constitute vote buying. For example, Desposato (2007: 103) views vote buying to include instances in which candidates promise postelection benefits to individuals. Likewise, Schaffer and Schedler (2007: 24) agree and refer to this tactic as “deferred delivery,” which involves “postponing payment and dispensing it conditionally upon the right candidate winning.” Overall, while most scholars do not consider promises of future rewards to constitute clientelist vote buying, some contend that they do. Scholars disagree not just about the timing of benefits, but also about what is distributed. As Fig. 2 indicates, all authors agree that offering cash at election time in exchange for political support constitutes clientelist vote buying. Beyond this unsurprising result, however, stark differences emerge between authors’ defining attributes of the concept. While the term “vote buying” may well conjure up images of candidates surreptitiously offering money to voters in dark alleys, few authors restrict the concept to such a narrow definition. Clientelist vote buying is typically viewed to include not just cash, but also particularistic goods and services. As Fig. 2 demonstrates, 12 of the 15 studies include goods and services, in addition to cash, in their conceptualizations of clientelist vote buying. For example, Brusco et al. (2004: 69) discuss a wide variety of handouts used in Argentina for vote buying, including food, clothing, medicine, mattresses, construction materials, and utility bill payments. Similarly, Schaffer (2007: 1–2) lists cash and a broad range of goods distributed in exchange for votes, as well as services such as haircuts, teeth cleaning and vasectomies. By contrast, three studies in Fig. 2 explicitly or implicitly conceptualize clientelist vote buying more narrowly as distributing cash in exchange for votes. For example, Vicente and Wantchekon (2008: 3) explicitly define “vote buying as votes-for-cash (before the election).”4 In addition, Wang and Kurzman (2007) only mention cash offers to citizens – about $10 each, and “this price was set by the campaign and was nonnegotiable” (66) – in their analysis of the Kuomintang in Taiwan.5

4 However, in a later published version Vicente and Wantchekon (2009: 293) expand their conceptualization beyond cash, defining “vote buying as votes-for-cash, or votes for other fungible goods, before the election.” Also note that the authors make an unusual distinction between vote buying and clientelism, defining the latter “as the exchange of votes for favors conditional on being elected (e.g. jobs in the public sector)” (2009: 293). 5 It should be noted that Wang and Kurzman (2007: 66, 68) mention some brokers (not voters) who received non-monetary commissions for their vote-buying efforts (i.e., buying citizens’ votes with cash).

Likewise, Cox and Kousser (1981) only mention cash rewards in their prominent study of the historic U.S. (but see the mention of rum in Cox, 2009: 344). It should be underscored, however, that neither Wang and Kurzman (2007) nor Cox and Kousser (1981) explicitly state how clientelist vote buying is defined, so it is unclear whether their research findings actually stem from a stricter conceptual definition, if other goods and services simply were not offered in the relevant context, or if in fact they were offered but fall outside the scope of their study. Nevertheless, there remains some conceptual ambiguity with respect to what constitutes clientelist vote buying – although some scholars explicitly include much more than cash outlays, others do not. Another related source of conceptual variation is that some researchers explicitly or implicitly exclude specific types of goods and services from their definitions of clientelist vote buying. For example, some scholars emphasize the delivery of food and alcohol, including Schaffer (2007: 2), Stokes (2007: 84) and Lehoucq (2007: 38). But while some studies highlight food as one of the most frequently distributed benefits (e.g., Brusco et al., 2004: 69), five studies in Fig. 2 make no mention at all of food or alcohol. Of course, it is unclear whether this exclusion actually stems from a stricter conceptual definition, or if there are context-specific reasons that edible items are not delivered in some areas (e.g., citizen preferences or ease of distribution). In order to avoid such confusion, authors should clearly identify what is included and excluded in their definitions of rewards. Another example in which some authors are relatively more inclusive in their definitions of clientelist vote buying involves the “service” of transporting voters to polling places. For instance, Heckelman (1998: 437) considers offering transportation to voters to be a form of vote buying, explaining that buses used to take voters to the polls “simply act as a more efficient (cheaper) means of bribery for the parties.” Likewise, Cornelius (2004: 49) and Valenzuela (2002: 10) include transportation as a form of vote buying in their studies on Mexico and Chile, respectively. This conceptualization is more inclusive than that of most other scholars, including Cox and Kousser (1981: 656fn), who distinguish transportation to the polls from vote buying and offer it as an example of “more legitimate forms of activity.” More broadly, while many authors agree that offering goods and services to voters in exchange for political support constitutes clientelist vote buying, some disagreements persist. Scholars also differ with respect to whether they include employment and public program benefits in their definitions of clientelist vote buying. With respect to the former, most scholars suggest that the employment of voters in exchange for political support should be considered “patronage” instead of vote buying (Stokes, 2009: 605– 606). Yet four of the 15 studies in Fig. 2 perceive some types of employment as forms of vote buying. For instance, Schaffer and Schedler (2007: 21) discuss one form of vote buying in which candidates offer temporary employment to voters (especially during campaigns) for “rendering some nominal service,” when their real objective is to

S. Nichter / Electoral Studies 35 (2014) 315–327

obtain those citizens’ votes.6 Baland and Robinson (2007, 2008) examine employment as a form of “indirect” vote buying, highlighting the role of employers as intermediaries. They argue that candidates offer selective benefits to employers, who in turn induce (or in some instances coerce) their workers to vote in a particular way. In addition to employment, another contentious defining attribute is contingent public program benefits. As indicated in Fig. 2, three studies include such benefits in their definition. For example, while Brusco et al. (2004) do not analyze public programs, they “acknowledge that the manipulation of ostensibly public programs for electoral support is indeed a form of vote buying” (67). Similarly, Cornelius’s (2004: 52) study of vote buying in the Mexican election of 2000 provides as an example of vote buying “strategically timed distribution of checks to beneficiaries of federal government social programs.” By contrast, 12 studies in Fig. 2 do not discuss public program benefits as forms of clientelist vote buying. Overall, the above discussion highlights diverse and often conflicting definitions of clientelist vote buying in the academic literature. While researchers concur that the contingent distribution of cash for political support during campaigns constitutes clientelist vote buying, other defining attributes prove more contentious. Future studies on the topic would ideally share a common systematized concept of clientelist vote buying, such as the one offered earlier. At the very least, they should make conscious decisions about what attributes to include and exclude in their definitions, and should offer explicit explanations in their research. 3. Implications for empirical analyses Divergent definitions of clientelist vote buying affect not only conceptual clarity, but also empirical analyses. We now explore this important point using surveys from Nigeria and Mexico. These data are particularly useful because unlike most other surveys, they specifically asked respondents what types of rewards they received or were offered. Although the cases of Nigeria and Mexico are not necessarily representative, our findings serve as a warning to researchers who seek to test theoretical predictions about clientelism with survey data. Studies often use regressions to test, for example, whether political machines target citizens on the basis of factors such as political preferences, propensity to turn out, and poverty (e.g., Stokes, 2005; Nichter, 2008; Finan and Schechter, 2012).7 Specifications below suggest that the definition of rewards can influence the results of quantitative analyses. Scholars must therefore ensure that findings are robust to alternative definitions of vote buying, or alternatively posit

6 Note that Schaffer and Schedler’s (2007: 21) discussion suggests the goal of providing such employment may be to “generate gratitude” toward vote buyers, which raises the question of whether this phenomenon should be classified as vote buying (see Sections 4.3 and 5). 7 Different strategies of electoral clientelism, such as turnout buying, are discussed below in Section 7. Though not the focus of this paper, such strategies often yield different theoretical predictions (see Nichter, 2008; Gans-Morse et al., 2014).

319

compelling theoretical explanations why findings only hold with a subset of definitions. Fig. 3 investigates an Afrobarometer survey in Nigeria that provides information about several categories of rewards.8 The Nigeria study asks about whether an individual received a vote buying offer, not whether he or she actually received the reward. Of those responding about vote buying offers during the 2003 election, 259 reported offers of cash, 93 reported offers of goods (including food/ drink), and 27 reported employment offers.9 The first three columns of Fig. 3, which employ a linear probability model, provide initial evidence that the correlates of receiving a vote buying offer in Nigeria differ depending on how one defines rewards. For example, when rewards include cash, goods and jobs, citizens who are unlikely to vote as well as supporters of the incumbent PDP party are disproportionately likely to receive rewards (significant at the 95% and 90% level, respectively), but neither relationship is statistically significant when rewards are more restrictively defined as cash only (see columns 1 and 3). Conversely, three covariates – poverty, education and living in a rural area – are statistically significant when using cash only, but not with the more inclusive definition. In order to test differences across definitions more rigorously, columns 4–6 employ a seemingly unrelated regressions (SUR) model (Zellner, 1962). The SUR model analyzes these three regressions – which are identical except that their dependent variables use alternative definitions of vote buying – as a system, given correlation in the error terms across these regressions.10 Cross-equation tests reject the null of equality of coefficients between regressions with the most inclusive and those with more restrictive definitions of rewards. More specifically, chisquared tests reject the joint hypothesis of equal coefficients between columns 4 and 6 at the 95% level, and between columns 5 and 6 at the 99% level.11 Turning to specific variables, chi-squared tests using the SUR model reject the null that the coefficients on Likely Voter are equal between columns 4 and 6, as well as between columns 5 and 6 (both at the 99% level). In addition, chi-squared tests reject the null that the coefficients on poverty – as well as living in a rural area – are equal between columns 5 and 6 (both at the 95% level). Overall, the survey data from Nigeria suggest that findings from regression analyses may depend crucially on how one defines what constitutes a reward. Just as the Afrobarometer data suggest that empirical findings can depend on one’s definition of rewards, so too does survey evidence from Mexico. Fig. 4 examines data

8 Analyses are based on Afrobarometer Nigeria dataset 3.5, released in September 2007. Technical details and data are available at http://www. jdsurvey.net/afro/AFBTechnical.jsp. 9 Of 2339 citizens responding to questions about vote buying during the 2003 election, the percentage receiving these categories of rewards are: 11.1% cash, 4.0% goods, and 1.2% jobs. 10 A Breusch–Pagan test of independence rejects, at the 99% level, the null hypothesis of independent errors across these regressions. 11 By contrast, cross-equation tests fail to reject the null between the definitions in columns 4 and 5.

320

S. Nichter / Electoral Studies 35 (2014) 315–327

Fig. 3. Nigeria: correlates of rewards using different definitions.

from the 2006 Mexico Panel Study.12 The third round of this survey asks whether individuals received specific types of particularistic benefits during the previous few weeks of the campaign. Of 1582 individuals responding to this question, only 6 reported receiving cash, 42 reported receiving goods, and 26 reported receiving food or drink.13 The survey does not ask whether individuals received jobs in exchange for their votes. Although the results are somewhat less striking than in the Nigerian case, the correlates of vote buying remain dependent on how one defines rewards. For example, “Likely Voter” is significant at the 99% level in column 2 but insignificant in columns 1 and 3, while “PAN Supporter” is significant at the 95% level in column 1 but insignificant in columns 2 and 3. Once again, to test differences more rigorously, columns 4–6 employ an SUR model, and cross-equation tests reject the null of equality of coefficients between regressions. More specifically, chi-squared tests reject the joint hypothesis of

12 Lawson, Chappell et al., 2007. The Mexico 2006 Panel Study. Wave 3. http://web.mit.edu/clawson/www/polisci/research/mexico06/index. html. 13 Of 1582 responding, the percentages receiving rewards are: 0.4% cash, 2.7% goods, and 1.6% food/drink. Goods reflect the survey’s “gifts” category.

equal coefficients between columns 5 and 6 at the 99% level, and between columns 4 and 5 at the 90% level.14 Turning to specific variables, chi-squared tests using the SUR model reject the null that the coefficients are equal across some equations for the following four variables: PRD Supporter, Likely Supporter, Poverty and Age.15 All in all, evidence from Mexico also suggests that regression results can depend on one’s definition of rewards. Overall, the survey data from both Nigeria and Mexico suggest that regression findings can be sensitive to the particular definition of vote buying employed. Unfortunately, researchers who seek to test theoretical predictions about clientelism with survey data may conduct regressions using only one definition – and their findings may hinge on which definition they employ. For example, a

14 By contrast, cross-equation tests fail to reject the null between the definitions in columns 4 and 6. 15 In particular, chi-squared tests reject the null of equality of coefficients for Poverty between columns 5 and 6 (at the 99% level), for PRD Supporter between any two columns (at the 90% level for columns 4/5, 99% level for columns 4/6, and 95% level for columns 5/6), for Age between columns 4/6 as well as columns 5/6 (at the 90% and 95% level, respectively), and for Likely Supporter between columns 4/5 (at the 90% level).

S. Nichter / Electoral Studies 35 (2014) 315–327

321

Fig. 4. Mexico: correlates of rewards using different definitions.

4. Typology of “vote buying” usage in studies

as we argue, often inappropriately – in political science. We develop a typology of four distinct categories of usage of the term vote buying in the scholarly literature, and argue that two of these categories involve conceptual stretching (Sartori, 1970). Fig. 5 provides a typology of how academic studies utilize the concept of vote buying, with representative examples. As shown in the typology, the literature on vote buying can be categorized according to two key dimensions: (1) whether or not selective benefits are delivered to individual or small groups of citizens; and (2) whether selective benefits are contingent on political support.16 Clientelist vote buying – a central focus of this article discussed extensively above – distributes contingent benefits to individual or small groups of citizens in exchange for political support. Another focus of many researchers is legislative vote buying, which also involves contingent benefits but targets legislators. With non-

Thus far, we explored alternative definitions of vote buying in the clientelism literature and examined how different definitions can influence empirical findings. Scholars must also be careful to ascertain whether or not their topic of study should even be called vote buying. Vote buying is an evocative term that is used rather loosely – and

16 The present study focuses exclusively on how the concept of vote buying is used in the context of politics at the national or subnational level. Thus, we do not consider vote buying in corporations (e.g., Clark, 1978; Cole, 2001), or as mentioned in a few studies, in international organizations (e.g., Gillespie, 2001; Kuziemko and Werker, 2006).

study claiming that a given clientelist party targets swing voters may be vindicated with one conceptualization of rewards, but not with another. Researchers should thus be cautious that quantitative results do not depend on what attributes are included or excluded in their definitions. The choice of what constitutes a vote-buying reward should be explicit, and survey questions should be carefully worded in order to capture the different types of benefits included in a given definition. Ideally, surveys should ask openended or closed-ended questions about what particular benefits were given, rather than just asking yes/no questions about whether rewards were received. When time constraints make such a detailed questions impossible (e.g., when a single clientelism question is included in a general election survey), enumerators should be instructed to read specific examples of items that constitute rewards.

322

S. Nichter / Electoral Studies 35 (2014) 315–327

Fig. 5. Common usage of “vote buying” in academic studies.

excludable vote buying, parties or candidates target districts with benefits such as infrastructure projects in an effort to generate political support. Finally, non-binding vote buying delivers benefits to individual or small groups of citizens, without requiring their commitment to provide political support. We argue in Section 5 that studies that fall in the non-excludable and non-binding categories inappropriately use the term “vote buying” due to conceptual stretching. But we first investigate these conceptual distinctions more deeply by examining academic studies that fall in the three newly introduced categories (legislative, non-excludable and non-binding vote buying). 4.1. Legislative vote buying Scholars often employ the concept of vote buying when studying the legislative arena. Whereas clientelist vote buying targets individual or small groups of citizens, legislative vote buying targets legislators. A wide literature examines how vote buyers – who may include interest groups, presidents, governors, party leaders or committee leaders (Groseclose and Snyder, 1996: 304) – deliver selective benefits to legislators in exchange for their votes on a particular bill. Scholars who employ the term “vote buying” understand such exchanges to be contingent: selective benefits (often called “payments” in the literature) are written into a bill before roll-call voting, and in return legislators who receive benefits agree to support the legislation. For instance, an influential formal study by Groseclose and Snyder (1996) argues that vote buying results in supermajority coalitions in legislatures under most conditions. They focus exclusively on contingent payments to legislators, explaining that “payments for votes on a bill are typically written into the bill itself (as special conditions, allowances, exemptions, transition rules, and so on)” (304). Likewise, Dekel et al. (2009) also formally examine vote buying of legislators using contingent payments. The authors assume that “vote buying is an ordinary transaction: the lobbyist gets full control of the vote in exchange for an up-front payment to the legislator” (106).

Evans (2004) provides various examples of legislative vote buying. Evans explains that vote buying – which she defines as adding projects for a legislator’s district to a bill, in exchange for obtaining that legislator’s vote – allows leaders to obtain legislators’ support more easily than making substantial, broader revisions to the bill (36). Overall, she argues that “efforts at vote buying with distributive benefits were successful in three major instances in the House of Representatives. On the two highway reauthorization bills and NAFTA, policy coalition leaders evidently succeeded in changing members’ voting intentions by giving them distributive benefits for their districts” (159). Legislative vote buying is also frequently discussed outside the US context. For example, Brazil’s mensalão scandal involved illegal side-payments from the executive office to individual legislators, and was labeled as vote buying by the press, official investigative committees, and academics (Pereira et al., 2008: 4). Other studies of legislative vote buying (i.e., distributing contingent benefits to legislators) include Snyder (1991), Diermeier and Myerson (1999), Ansolabehere et al. (2001), King and Zeckhauser (2003), and Snyder and Ting (2005). 4.2. Non-excludable vote buying Non-excludable vote buying is another way in which studies use the concept of vote buying. Such studies investigate the allocation of local public goods, such as hospitals and roads, across political districts. Whereas clientelist vote buying targets individual or small groups of citizens and legislative vote buying targets legislators, nonexcludable vote buying targets districts. Scholars who employ this usage do not argue that the phenomenon involves contingent exchange. According to such analysts, neither citizens nor legislators commit to vote in a particular way so that their districts obtain benefits. Rather, it is argued that politicians distribute local public goods as a political investment, with the aim of generating future electoral support from citizens who vote in targeted districts. Observe that once local public goods are delivered to targeted districts, members of those districts cannot be excluded from benefits (Estevez et al., 2002: 4) – by definition, local public goods are non-excludable within localities. This characteristic of non-excludability is a key reason that vote buying with local public goods cannot rely on contingent exchange: citizens in targeted districts will receive benefits regardless of whether or not they agree to vote in a particular way. Beatriz Magaloni’s (2006) excellent book on hegemonic party survival and demise in Mexico offers an example of non-excludable vote buying. In a chapter entitled “The Politics of Vote Buying,” Magaloni argues that PRONASOL, a national poverty relief program, “was an effective votebuying program” (146). By vote buying, she refers to the PRI regime channeling local public goods disproportionately to municipalities governed by the PRI rather than the opposition (122, 149–150). Magaloni is particularly careful to emphasize that the chapter does not focus on “individual vote buying through excludable benefits, which the literature equates with clientelism,” but instead on “the

S. Nichter / Electoral Studies 35 (2014) 315–327

geographic allocation of total PRONASOL resources” (123, italics in original).17 Similarly, Herron and Theodos (2004) employ the concept of non-excludable vote buying in their econometric study of how the Illinois state government allocated “member initiative grants” across districts. The member initiative grants provided community benefits to targeted districts, including ambulances, fire trucks, and playground improvements (288). Herron and Theodos find that before the 2000 election, “Illinois decision makers who allocated member initiative funds sought to distribute them in a way that would be most beneficial in the sense of vote buying,” such as by targeting politically competitive districts (287). Another example of non-excludable vote buying is a recent study by Burgess et al. (2012) on the political economy of road placement in Kenya. The study finds that the Kenyan president employs wide discretion when building roads, and channels substantially more paved roads to districts where his ethnicity is dominant (16–17). The authors conduct various analyses to examine whether the president’s targeting of co-ethnic districts with roads is motivated by vote buying, and explain that the “votebuying hypothesis consists of politicians targeting (or committing to target) public spending so as to maximize their probability of being elected” (4, 17). Overall, these studies provide examples of non-excludable vote buying, which distributes local public goods to districts without contingent exchanges.

4.3. Non-binding vote buying Yet another way in which scholars utilize the concept of vote buying involves distributing selective benefits to individual or small groups of citizens without contingent exchanges. With such non-binding vote buying, parties distribute particularistic benefits in the hopes of generating goodwill that will yield electoral returns during the next election. Some scholars argue that non-binding vote buying targets benefits using partisanship or other political criteria, while other researchers suggest that this form of vote buying distributes benefits programmatically but nevertheless aims to generate political support among recipients. Given that non-binding vote buying involves benefits distributed to individual or small groups of citizens, special care must be taken not to confuse this subtype with clientelist vote buying. Unlike clientelist vote buying, recipients do not commit to voting in a particular way in exchange for benefits. Hence, this form of vote buying does not involve commitment problems because recipients do not promise to deliver electoral support. As a result, in sharp contrast with literature on clientelist vote buying, studies on non-binding vote buying do not discuss opportunistic defection. To explore how scholars employ the concept of nonbinding vote buying, consider Diaz-Cayeros et al. (2006). Their study finds that two programs distributing individual

17

More specifically, the chapter focuses on the geographic allocation of approximately 70 percent of PRONASOL funds that consisted of “public works targeted to towns, municipalities or regions” (Magaloni, 2006: 122–3).

323

benefits to the poor, Oportunidades and Seguro Popular, boosted Calderón’s political support and contributed to his victory in the 2006 Mexican presidential election. On the one hand, they contend that it is “highly implausible” that clientelism affected the distribution of benefits (30). However, Diaz-Cayeros et al. indicate that the programs reflect a relatively benign form of vote buying because individuals who received benefits responded favorably and chose to provide political support for Calderón. As they explain: “Favorable beneficiary reaction to the incumbent party, where it occurred, may indicate successful votebuying by the party in power, but it is likely to be votebuying of the good sort” (30).18 Another example of non-binding vote buying is Chen (2008), which examines the Federal Emergency Management Agency’s (FEMA) distribution of disaster assistance to individuals in the wake of the 2004 Florida hurricane season. Chen employs the concept of vote buying throughout his research, but never argues that FEMA benefits were contingent on individuals agreeing to vote in a particular way. Rather, he finds a “partisan asymmetry in voter responsiveness to FEMA aid,” in that Republicans were more likely than Democrats to deliver electoral support for Bush after receiving individual disaster assistance (21). Chen concludes that “a political party’s most efficient vote-buying strategy is to target pre-election monetary awards to its core partisan supporters in the hopes of increasing their turnout rate, thus enhancing the party’s vote share” (24). Another illuminating example of non-binding vote buying is a contentious editorial written by Cristovam Buarque (2006), a Brazilian Senator and chief architect of the pilot program that evolved into Bolsa Familia, the largest conditional cash transfer in the world. He claims that Bolsa Familia “has become a vote-buying scheme” because its payments to the poor increased political support for former President Lula. Buarque does not suggest that the program provides cash benefits to individuals as part of contingent (clientelist) exchange; rather, he asserts that “it turned into a program with a strong electoral appeal since it is seen as aid without counterpart, a kindness received from the government.” In addition, Thames (2001: 67) argues that Yeltsin engaged in vote buying in Russia when he expanded overall expenditures on benefits that accrue to individuals (such as increasing pensions and raising student stipends) in order to attract the political support of specific types of voters. In sum, these studies provide examples of non-binding vote buying, which distributes selective benefits to individual or small groups of citizens without contingent exchanges.

5. Risk of conceptual stretching These various ways that researchers use vote buying raises the issue of conceptual stretching. Employing vote

18 To provide broader context, it should be noted that the important ongoing work of Diaz-Cayeros et al. (2010: 222) seeks to “measure the vote buying potential of various forms of government transfers – public versus private goods, delivered through clientelistic versus nonclientelistic programs.” See also Estevez et al. (2002).

324

S. Nichter / Electoral Studies 35 (2014) 315–327

buying to describe diverse phenomena involves both potential benefits and risks. Vote buying is an evocative term that can be effectively employed in a variety of contexts, so long as that usage reflects the true underlying meaning of the concept. The disaggregation of vote buying above helps to distinguish between distinct uses of the concept in academic studies, but the question remains whether each category represents a “full instance” of the concept (Collier and Levitsky, 1997). Of course, the answer to this question depends on what a given analyst considers to be his or her “root definition” of vote buying. Scholars may choose to include or exclude particular attributes in their own root definitions; in fact, an important point of this article is to emphasize that such decisions have important consequences for research and should thus be made explicitly and consciously.19 However, we argue that such flexibility should not be taken too far, and that contingent exchange should be deemed an essential component of any root definition of vote buying. Vote buying involves contingent exchange because selective benefits are provided on a quid pro quo basis.20 That is, recipients agree to vote in a particular way in return for selective benefits. Given this root definition, we argue that studies captured by the bottom two categories of Fig. 5 – non-excludable and non-binding “vote buying” – engage in conceptual stretching because the phenomena they describe do not involve contingent exchange. Contingent exchange is missing with non-excludable vote buying because citizens who reside in targeted districts cannot be excluded from local public goods, and it is missing with non-binding vote buying because citizens (or small groups) who receive benefits make no commitments to deliver electoral support. As such, we argue that scholars writing about these two categories should employ more accurate terms and eschew the use of “vote buying” altogether. For example, studies investigating the allocation of local public goods (i.e., “non-excludable vote buying” in the typology) should more appropriately use terms such as “pork” rather than “vote buying.” While avoiding use of the term would be preferable, we provide another option for scholars who remain committed to using “vote buying” to describe the bottom two categories of Fig. 5. They should clearly identify them as “diminished subtypes” (Collier and Levitsky, 1997).21 Collier and Levitsky underscore two key aspects of diminished subtypes: (1) they “are not full instances of the root definition” of the concept; and (2) they typically highlight attributes of the root definition that are either missing or present (1997: 437–8, italics in original). Following their approach, one could employ the labels “non-excludable vote buying” and “non-binding vote buying” to convey their diminished nature. The adjectives “non-excludable”

19 As Collier and Levitsky (1997: 435fn) explain, the term “root definition” does “not imply that it is the ‘correct’ definition of the relevant concept . It is simply the definition that, for a particular author, is the point of departure in forming the subtype.” 20 Note that other related concepts, such as clientelism, may or may not involve quid pro quo exchange. 21 Collier and Levitsky (1997) also discuss several alternatives to the strategy of using diminished subtypes.

and “non-binding” provide information about the missing defining attribute, the absence of contingent exchange. Overall, this discussion underscores the importance of paying close attention to conceptual issues when studying vote buying. Avoiding conceptual stretching is crucial. Scholars would ideally refrain from using the term “vote buying” to describe other phenomenon such as pork, and at the very least should acknowledge when cases they discuss do not represent full instances of the concept.

6. Heterogeneity of findings Distinguishing between the uses of the term vote buying in Fig. 5 not only provides insight about conceptual stretching, but also has the potential to improve descriptive and causal inference. Theoretical predictions and empirical findings from one form of vote buying need not necessarily apply to another. Given this study’s particular focus on the field of clientelism, we now emphasize that findings from studies on legislative vote buying need not apply to the concept of clientelist vote buying. Some studies apply findings between categories of vote buying without explicitly recognizing the differences between them. For example, findings from Groseclose and Snyder’s (1996) excellent study on legislative vote buying is cited by several studies on clientelist vote buying, such as Vicente (2008: 5, 7), Vicente and Wantchekon (2009: 295), and Morgan and Vardy (2012: 3–4). In addition, some insightful work on legislative vote buying extends findings from legislative politics to the study of clientelism, without emphasizing important distinctions. For example, Dal Bó develops a formal model of legislative vote buying and suggests that findings have “implications for lobbying, for clientelism, for decisions in legislatures, boards, and central banks, and for the efficiency of democracy” (2007: 789, emphasis added). While findings about legislative vote buying may well travel to the concept of clientelist vote buying, there are many reasons one might not want to assume blindly that they do. At least five key differences suggest that scholars should be cautious when applying insights from legislative vote buying to clientelist vote buying:

(1) Probability of a vote being pivotal: One reason that vote buying may involve a different logic in legislative settings is that each legislator is substantially more likely to be pivotal than each citizen in general elections. Whereas millions of citizens vote in many elections, legislative voting rarely includes more than a few hundred. (2) Visibility of vote choices: It is often less difficult to observe roll-call voting than voting in general elections, which has important implications for the difficulty of monitoring vote-buying transactions. Even though clientelist parties in practice engage in a wide variety of tactics to violate the secret ballot, at least de jure ballot secrecy exists in the vast majority of elections across the world. By contrast, legislative voting is frequently done openly, and roll-call voting, on which studies of legislative vote buying typically focus, is the

S. Nichter / Electoral Studies 35 (2014) 315–327

most observable stage in the lawmaking process (Snyder, 1992: 16). (3) Frequency of voting: Roll-call voting occurs much more frequently than general elections. This fact may have important consequences for the enforcement of votebuying transactions, because there are more frequent interactions to punish defectors with legislative vote buying than with clientelist vote buying.22 (4) Level of voter information: In many cases, legislators have substantially more information about alternative policy choices than citizens. Information asymmetries are especially likely when clientelist vote buying targets poor citizens (e.g., Stokes, 2005). In part due to greater information, legislators may in some cases be more likely to act strategically than citizens. (5) Legality of vote buying: Whereas clientelist vote buying is illegal in virtually all contemporary societies, some forms of legislative vote buying involve trading of favors that is either permitted by law or subject to unenforced laws (Hasen, 2000: 1339–1340). One example of such a legal exchange is provided by several studies on legislative vote buying: Marjorie Margolies-Mezvinsky explicitly agreed to vote for President Clinton’s 1993 tax bill, and in exchange Clinton spoke at a conference held by Margolies-Mezvinsky in her district (cf Groseclose and Snyder, 1996: 312). As this non-exhaustive discussion of differences between voting in legislatures and general elections suggests, there are many reasons why vote buying may not be identical across these two arenas. The point is not that studies on legislative vote buying cannot provide useful insights for scholars examining clientelist vote buying. Rather, the point is that scholars must pay close attention to potential heterogeneity, and should be careful not to assume that predictions and findings apply across forms of vote buying. 7. Clientelist vote buying vs. other clientelism Thus far, we have explored several reasons why it is important to pay careful attention to the conceptualization of vote buying. These reasons include improving conceptual clarity, ensuring that results are not overly sensitive to included or excluded attributes, clarifying whether the topic of discussion is actually vote buying at all, and highlighting whether findings from other studies travel across types of vote buying. We now turn to another important conceptual problem that threatens research on clientelist vote buying. The failure to distinguish clientelist vote buying from other forms of clientelism can have serious implications for causal inference. For example, Nichter

22

Of course, in some cases, vote buying is combined with forms of “relational clientelism” (see Section 7) that involve more frequent interactions beyond electoral campaigns. The present discussion focuses exclusively on vote buying.

325

(2008: 20) argues that “much of what scholars interpret as vote buying (exchanging rewards for vote choices) may actually be turnout buying (exchanging rewards for turnout).” But turnout buying is only one of numerous strategies frequently conflated with vote buying. Given that article only briefly mentions other strategies, this section aims to clarify the distinction between clientelist vote buying and other forms of clientelism. First, a fundamental distinction should be made between clientelist vote buying and forms of “relational clientelism” (Nichter 2010). Clientelist vote buying is a form of “electoral clientelism,” in which elites deliver all payoffs during campaigns. Because strategies of electoral clientelism deliver all benefits to recipients before voting occurs, they involve the threat of opportunistic defection by citizens (or small groups of citizens). The party or politician that buys votes during the campaign remains uncertain that recipients will actually follow through with the agreement to deliver their vote choices on Election Day. By contrast, the type of commitment problem involved in relational clientelism is quite different. Relational clientelism involves ongoing benefits that extend beyond electoral campaigns. Because at least some benefits are delivered to recipients after voting, relational clientelism – unlike clientelist vote buying – involves the threat of opportunistic defection by both citizens and elites. In other words, not only is the party unsure that the citizen will deliver his or her vote choice, but also the citizen is uncertain that the party will provide promised benefits after the election. In sum, clientelist vote buying is a form of electoral clientelism, and should be distinguished from forms of relational clientelism. Another key point is that clientelist vote buying must not be confused with other forms of electoral clientelism. To clarify this point, Fig. 6 helps to distinguish clientelist vote buying from other forms of electoral clientelism.23 Reward targeting plays a central role in distinguishing strategies, so the typology emphasizes two key attributes of citizens: (1) inclination to turn out, and (2) political preferences. Clientelist vote buying rewards opposing (or indifferent) voters for switching their vote choices. Although studies on clientelism typically assume – either implicitly or explicitly – that political machines distribute benefits to voters in exchange for voting against their preferences, machines also often engage in other strategies of electoral clientelism.24 Vote buying should be distinguished from forms of electoral clientelism that do not aim to influence vote choices. One strategy mentioned above, turnout buying, provides benefits to unmobilized supporters in exchange for showing up at the polls. During the 2004 US election, five Democratic Party workers in East St. Louis were convicted for offering cigarettes, beer, medicine, and $5 to $10 rewards to increase turnout of the poor (Nichter, 2008). In Argentina, Nichter contends that while both strategies

23 The typology is adapted from Nichter (2008: 20), and the discussion below is adapted from Nichter (2008, 2010), and Gans-Morse, Mazzuca and Nichter (2014). 24 Rewarding loyalists is shown in Fig. 6 but not discussed below, as it is a form of relational not electoral clientelism.

326

S. Nichter / Electoral Studies 35 (2014) 315–327

Fig. 6. Clientelist vote buying vs other clientelist strategies during elections.

coexist, Stokes’s (2005) survey data are more consistent with turnout buying than vote buying. Turnout buying has also been identified in Argentina, Mexico and Venezuela (Dunning and Stokes, 2008; Rosas and Hawkins, 2008). Another strategy of electoral clientelism that must be distinguished from clientelist vote buying is abstention buying, which provides benefits to indifferent or opposing citizens for not voting (Cox and Kousser, 1981; Schaffer, 2002; Cornelius, 2004).25 This demobilizational strategy decreases votes received by the opposition. For instance, Cox and Kousser (1981) investigate newsarticles in New York state from 1870 to 1916 and find that operatives rewarded many rural voters for staying home on Election Day. Likewise, candidates employ benefits to demobilize opposition voters when busing them away from voting localities in the Philippines and buying their voter identification cards in Guyana (Schaffer, 2002). Double persuasion delivers rewards during elections to influence vote choices and induce electoral participation. The clientelism literature suggests many citizens have few ideological preferences or reasons to vote, other than handouts offered by machines (e.g., Chubb, 1982, 171). With double persuasion, clientelist parties dispense rewards to such nonvoters who do not inherently prefer them for ideological or programmatic reasons. This strategy should be distinguished from clientelist vote buying as it targets different citizens and has a mobilizational as well as a persuasive dimension. To sum up, we have discussed two important sources of conceptual ambiguity about clientelist vote buying that are important for researchers to address. First, studies must be sure to differentiate clientelist vote buying from other forms of vote buying. And second, studies must be sure to differentiate clientelist vote buying from other forms of clientelism. These two crucial distinctions heighten conceptual clarity and may also serve to improve descriptive and causal inference. 8. Conclusion Over the past few decades, use of the term “vote buying” has increased sharply. This study investigates the concept

25 Although abstention buying is frequently called “negative vote buying,” a more precise alternative term would be “negative turnout buying” as it affects turnout, not vote choices.

of vote buying, with a particular focus on its usage in research on clientelism. We have shown that researchers employ many different definitions of vote buying. These differences can undermine conceptual clarity and empirical analyses. Evidence from Nigeria and Mexico suggests that different definitions can yield divergent empirical results. The findings of this paper suggest that scholars should make conscious decisions about what attributes to include and exclude in their own definitions of vote buying. And most crucially, they should offer clear, explicit explanations of their definitions. Unpacking the concept of vote buying contributes to scholarly research by clarifying its usage in different contexts. The diverse usage of vote buying poses the risk of conceptual stretching, because scholars often use the term to describe other phenomena. Academic usage of the term “vote buying” tends to fall in four distinct categories: clientelist, legislative, non-excludable, and non-binding. Although studies captured in the first two categories (clientelist, legislative) focus on what may be accurately called vote buying, studies captured in the latter two categories (non-excludable, non-binding) inappropriately use the term. Neither involves contingent exchange, which by most standards is a defining attribute of vote buying. Such distinctions are important not only for conceptual clarity, but also because findings from descriptive and causal inference may not entirely correspond across categories. An additional problem in the study of machine politics is that clientelist vote buying is often conflated with other forms of clientelism. Overall, this study challenges scholars to improve their conceptualization of vote buying. Future research should pay closer attention to how the concept is employed, in order to mitigate conceptual ambiguity and potentially improve descriptive and causal inference.

Acknowledgments The author thanks the following people for their comments and suggestions: David Collier, Gustavo Bobonis, Henry Brady, Maysa Eissa, Jordan Gans-Morse and Marco Gonzalez-Navarro. Nichter also acknowledges support from the National Science Foundation, the Harvard Academy for International and Area Studies, and the Center on Democracy, Development and the Rule of Law at Stanford University.

S. Nichter / Electoral Studies 35 (2014) 315–327

References Adcock, R., Collier, D., 2001. Measurement validity: a shared standard for qualitative and quantitative research. Am. Polit. Sci. Rev. 95 (3), 529–546. Ansolabehere, S., Snyder, J.M., Stewart, C., 2001. The effects of party and preferences on congressional roll-call voting. Legis. Stud. Q. XXVI (4), 533–572. Auyero, J., 2000. The logic of clientelism in Argentina: an ethnographic account. Lat. Am. Res. Rev. 35 (3), 55–81. Baland, J.-M., Robinson, J.A., 2007. How does vote buying shape the economy? In: Schaffer, Frederic C. (Ed.), Elections for Sale: the Causes and Consequences of Vote Buying. Lynne Rienner Publishers, Boulder, CO. Brusco, V., Nazareno, M., Stokes, S.C., 2004. Vote buying in Argentina. Lat. Am. Res. Rev. 39 (2), 66–88. Buarque, C., 2006. Brazil’s school stipend has become a vote-buying scheme. Accessible Online at: http://www.brazzil.com/articles/162march-2006/9537.html. Burgess, R., Jedwab, R., Miguel, E., Morjaria, A., 2012. Ethnic Favoritism and Democracy. Typescript. Chen, J., 2008. Buying Votes with Public Funds in the US Presidential Election: Are Swing or Core Voters Easier to Buy Off?, Typescript, Stanford University. Chubb, J., 1982. Patronage, Power, and Poverty in Southern Italy: A Tale of Two Cities. Cambridge University Press, New York. Clark, R.C., 1978. Vote buying and corporate law. Case West. Law Rev. 29, 776–807. Cole, D.R., 2001. E-proxies for sale – corporate vote-buying in the Internet age. Wash. Law Rev. 76, 793–858. Collier, D., Levitsky, S., 1997. Democracy with adjectives: conceptual innovation in comparative research. World Polit. 49 (3), 430–451. Cornelius, W., 2004. Mobilized voting in the 2000 elections: the changing efficacy of vote buying and coercion in Mexican electoral politics. In: Dominguez, I. Jorge, Lawson, Chappell (Eds.), Mexico’s Pivotal Democratic Elections: Candidates, Voters, and the Presidential Campaign of 2000. Cox, G.W., Kousser, J.M., 1981. Turnout and rural corruption: New York as a test case. Am. J. Polit. Sci. 25 (4), 646–663. Cox, G., 2009. Swing Voters, Core Voters and Distributive Politics. Ed. Ian Shapiro, Susan Stokes, Elisabeth Wood and Alexander S. Kirshner, Political Representation. Cambridge University Press, New York. Dekel, E., Jackson, M.O., Wolinsky, A., 2009. Vote buying: legislatures and lobbying. Q. J. Polit. Sci. 4, 103–128. Desposato, S., 2007. How does vote buying shape the legislative arena? In: Schaffer, Frederic C. (Ed.), Elections for Sale: The Causes and Consequences of Vote Buying. Lynne Rienner Publishers, Boulder, CO (Chapter 7). Diaz-Cayeros, A., Estevez, F., Magaloni, B., 2006. Buying off the Poor: Effects of Targeted Benefits in the 2006 Presidential Race. Typescript, Harvard University. Diaz-Cayeros, A., Magaloni, B., Estevez, F., 2010. Strategies of VoteBuying: Poverty, Democracy, and Social Transfers in Mexico. Typescript, Stanford University. Diermeier, D., Myerson, R.B.,1999. Bicameralism and its consequences for the internal organization of legislatures. Am. Econ. Rev. 89 (5), 1182–1196. Dunning, T., Stokes, S.C., 2008. Persuasion vs. Mobilization. Paper presented at the American Political Science Association Conference, Boston, MA. Estevez, F., Diaz-Cayeros, A., Magaloni, B., 2002. A Portfolio Diversification Model of Electoral Investment. Typescript, Stanford University. Evans, D., 2004. Greasing the Wheels: Using Pork Barrel Projects to Build Majority Coalitions. Cambridge University Press, New York. Finan, F., Schechter, L., 2012. Vote buying and reciprocity. Econometrica 80 (2), 863–881. Gans-Morse, J., Mazzuca, S., Nichter, S., 2014. Varieties of clientelism: machine politics during elections. Am. J. Polit. Sci. in press, http:// onlinelibrary.wiley.com/doi/10.1111/ajps.12058/abstract. Gillespie, A., 2001. Transparency in international environmental law: a case study of the International Whaling Commission. Georgetown Int. Environ. Law Rev. 14 (333), 333–348. Gonzalez-Ocantos, E., de Jonge, C.K., Melendez, C., Osorio, J., Nickerson, D.W., 2012. Vote buying and social desirability bias: experimental evidence from Nicaragua. Am. J. Polit. Sci. 56 (1), 202–217. Groseclose, T., Snyder, J.M., 1996. Buying supermajorities. Am. Polit. Sci. Rev. 90 (2), 303–315. Hasen, R.L., 2000. Vote buying. California Law Rev. 88 (5), 1323–1372. Heckelman, J.C., 1998. Bribing voters without verification. Soc. Sci. J. 35 (3), 435–443. Herron, M.C., Theodos, B.A., 2004. Government redistribution in the shadow of legislative elections: a study of the Illinois Member Initiative Grants Program. Legis. Stud. Q. XXIX (2), 287–311.

327

Hicken, A., 2007. Institutional incentives: do candidate-centered electoral rules encourage vote buying? In: Schaffer, Frederic C. (Ed.), Elections for Sale: the Causes and Consequences of Vote Buying. Lynne Rienner Publishers, Boulder, CO. King, D.C., Zeckhauser, R.J., 2003. Congressional vote options. Legis. Stud. Q. XXVIII (3), 387–411. Kitschelt, H.P., Wilkinson, S.I., 2007. Patrons, Clients, and Policies: Patterns of Democratic Accountability and Political Competition. Cambridge University Press. Kuziemko, I., Werker, E., 2006. How much is a seat on the Security Council Worth? Foreign aid and bribery at the United Nations. J. Polit. Econ. 114 (5), 905–930. Lehoucq, F.E., 2007. When does a market for votes emerge? Historical and theoretical perspectives. In: Schaffer, Frederic C. (Ed.), Elections for Sale: the Causes and Consequences of Vote Buying. Lynne Rienner Publishers, Boulder, CO. Levitsky, S., 1998. Institutionalization and Peronism: the concept, the case and the case for unpacking the concept. Party Polit. 4 (1), 77–92. Levitsky, S., 2003. Transforming Labor-based Parties in Latin America: Argentine Peronism in Comparative Perspective. Cambridge University Press. Lippert-Rasmussen, K., 2011. Vote buying and election promises: should democrats care about the difference? J. Polit. Philos. 19 (2), 125–144. Magaloni, B., 2006. Voting for Autocracy: Hegemonic Party Survival and its Demise in Mexico. Cambridge University Press, New York. Morgan, J., Vardy, F., 2012. Negative vote buying and the secret ballot. J. Law Econ. Organ. 28 (4), 818–849. Nichter, S., 2008. Vote buying or turnout buying? Machine politics and the secret ballot. Am. Polit. Sci. Rev. 102 (1), 19–31. Nichter, S., 2010. Politics and Poverty: Electoral Clientelism in Latin America. Doctoral Dissertation. University of California, Berkeley. Pereira, C., Power, T.J., Raile, E.D., 2008. Coalitional Presidentialism and side payments: explaining the Mensalão Scandal in Brazil. Typescript. Rosas, G., Hawkins, K., 2008. Turncoats, True Believers, and Turnout: Machine Politics in the Absence of Vote Monitoring. Typescript. Sartori, G., 1970. Concept misformation in comparative politics. Am. Polit. Sci. Rev. 64 (4), 1033–1053. Schaffer, F., 2002. Might cleaning up elections keep people away from the polls? Historical and comparative perspectives. Int. Polit. Sci. Rev. 23 (1), 69–84. Schaffer, F.C., 2007. Elections for Sale: The Causes and Consequences of Vote Buying. Lynne Rienner Publishers, Boulder, CO. Schaffer, F.C., Schedler, A., 2007. What is vote buying? In: Schaffer, Frederic C. (Ed.), Elections for Sale: The Causes and Consequences of Vote Buying. Lynne Rienner Publishers, Boulder, CO. Snyder, J.M., 1991. On buying legislatures. Econ. Polit. 3 (2), 93–109. Snyder, J.M., 1992. Long-term investing in politicians; or, give early, give often. J. Law Econ. XXXV, 15–43. Snyder, J.M., Ting, M.M., 2005. Why roll calls? A model of positiontaking in legislative voting and elections. J. Law Econ. Organ. 21 (1), 153–178. Stokes, S.C., 2005. Perverse accountability: a formal model of machine politics with evidence from Argentina. Am. Polit. Sci. Rev. 99 (3), 315–325. Stokes, S.C., 2007. Is vote buying undemocratic? In: Schaffer, Frederic C. (Ed.), Elections for Sale: the Causes and Consequences of Vote Buying. Lynne Rienner Publishers, Boulder, CO (Chapter 6). Stokes, S.C., 2009. Pork, by Any Other Name...Building a Conceptual Scheme of Distributive Politics. In: Paper presented at the American Political Science Association Conference, Toronto, Canada. Thames, F.C., 2001. Did Yeltsin buy elections? The Russian political Business cycle, 1993–1999. Commun. Post-Commun. Stud. 34, 63–76. Valenzuela, J.S., 2002. Futile or effective? The Right’s ‘Gifts’ to voters in the Second Round of the Chilean Presidential Elections of January 2000. In: Paper presented at the Comparative Politics of Vote Buying Conference at the Massachusetts Institute for Technology, Center for International Studies, Cambridge, MA. Vicente, P.C., 2008. Is vote buying effective? Evidence from a field Experiment in West Africa. Typescript. Vicente, P.C., Wantchekon, L., 2009. Clientelism and vote buying: lessons from field experiments in African elections. Oxf. Rev. Econ. Policy 25 (2), 292–305. Vicente, P.C., Wantchekon L., 2008. Clientelism and Vote Buying: Lessons From Field Experiments in African Elections. Wang, C.-S., Kurzman, C., 2007. Logistics: how to buy votes. In: Schaffer, Frederic C. (Ed.), Elections for Sale: the Causes and Consequences of Vote Buying. Lynne Rienner Publishers, Boulder, CO. Zellner, A., 1962. An efficient method of estimating seemingly unrelated regressions and tests for aggregation bias. J. Am. Stat. Assoc. 57 (298), 348–368.