Electoral Clientelism and Redistribution: H ow Vote Buying Undermines Citizen Demand for Public Services 20 23 Working paper series 2023:10 Monika Bauhr Mogens K. Justesen Electoral Clientelism and Redis- tribution: How Vote Buying Undermines Citizen Demand for Public Services Monika Bauhr Mogens K. Justesen WORKING PAPER SERIES 2023:10 QoG THE QUALITY OF GOVERNMENT INSTITUTE Department of Political Science University of Gothenburg Box 711, SE 405 30 GÖTEBORG September 2023 ISSN 1653-8919 © 2023 by Monika Bauhr and Mogens K. Justesen. All rights reserved. Electoral Clientelism and Redistribution: How Vote Buying Undermines Citizen Demand for Public Ser- vices Monika Bauhr Mogens K. Justesen QoG Working Paper Series 2023:10 September 2023 ISSN 1653-8919 Abstract Does electoral clientelism limit public demand for programmatic redistribution? While recent studies sug- gest that distribution of pre-electoral clientelist transfers can divert resources from post-electoral program- matic redistribution, fewer studies have investigated how citizens’ experiences of clientelism shape demand for programmatic redistribution. We suggest that electoral clientelism undermines citizens demand for pro- grammatic redistribution, since it establishes norms of reciprocity, signals commitment to particularistic payoffs, and may alienate people who would otherwise be supportive of redistribution. Using a nationwide survey fielded in South Africa following the 2016 municipal elections, we show that citizens’ who experience electoral clientelism express lower demand for government redistribution related to health, education, un- employment, pensions, and social grants. We also find that support for redistribution decreases most among citizens with high levels of political trust and who believe the political system is relatively uncorrupt. This suggests that electoral clientelism reinforces demand for particularistic as opposed to programmatic redis- tribution. Key words: clientelism, vote buying, redistribution, demand for public goods, South Africa, taxation. Monika Bauhr Mogens K. Justesen The Quality of Government Institute Copenhagen Business School Department of Political Science mkj.egb@cbs.dk University of Gothenburg Monika.Bauhr@pol.gu.se 2 Electoral clientelism is prevalent in new democracies, serving as instruments of political parties in their efforts to mobilize electoral support from broader constituencies (Mares and Young 2019; Jensen and Justesen 2014; Nichter et al. 2014; Stokes et al. 2013; Hicken 2011; Markussen 2011). In theory, the intro- duction of multiparty competition and universal suffrage should increase the political power of poor and low-income groups in society, and result in broad-based public services, more extensive programmatic re- distribution, and a more equal distribution of income in society (Son and Bellinger 2021; Acemoglu and Robinson 2006; Boix 2003; Meltzer and Richard 1981). In contrast, clientelist redistribution entails that private or club goods – e.g., money, jobs, or access to public services – are targeted to voters and small groups in return for their vote or political support (Stokes et al. 2013). In this sense, electoral clientelism – distributing material benefits in exchange for votes during election campaigns – incorporates elements of bribery, where parties or candidates try to bribe people to vote for them. A growing number of studies point to the adverse distributional effects of electoral clientelism. Studies suggest that electoral clientelism affects programmatic policies and pro-poor services by distorting the supply and allocation of public funds (Khe- mani 2015; Larreguy 2013; Robinson and Verdier 2013; Baland and Robinson 2012; Markussen 2011). This, in turn, reduces access to public services among the poor and ultimately increases inequality and poverty. Without denying the direct costs of clientelism, our argument emphasizes a different way through which clientelist parties may affect the provision of public goods – by shaping public demand for programmatic redistribution. Specifically, we suggest that electoral clientelism activates norms of reciprocity, fuels demand for post-electoral particularistic payoffs – including preferential access to public service delivery – and thereby undermines demand for programmatic redistribution. For citizens, this effectively gives rise to a trade-off between experiences of electoral clientelism and demand for universal public service delivery. Using data from a nationally representative survey conducted after the 2016 municipal elections in South Africa, we construct a measure of voters’ experience of electoral clientelism – vote buying, turnout buying and abstention buying – for a large nationally representative sample of the adult population. Our results document a strong negative association between electoral clientelism and voter preferences for program- matic redistribution on issues such as education, health, and social welfare transfers. Furthermore, we show that the negative association between electoral clientelism and demand for programmatic redistribution is strongest among citizens’ expressing a high level of political trust and beliefs that the political system is relatively free from corruption. These results lend support to the idea that electoral clientelism may not only contribute to undermine popular demand for programmatic redistribution but also reinforce an equilibrium where clientelism fuels demand for particularistic payoffs and continues to co-exist with weak public service provision. Our article makes two interrelated contributions to the literature. First, to the best of our knowledge, this study is one of the first to provide evidence of a direct link between electoral clientelism and voter prefer- ences for programmatic redistribution in young democracies. Much in line with recent work (e.g., Keefer and Khemani 2009; Robinson and Verdier 2013; Khemani 2015; Person et al 2019; Carlsson 2021), we suggest that clientelist transfers can substitute for programmatic public service provision. Electoral clien- telism does not only distort the allocation of resources on the supply-side, but also voters’ demand and expectations of the type of benefits the state and public sector can and will provide. Personal experiences of electoral clientelism may, in other words, reinforce a vicious cycle where citizens with limited expectations of benefitting from public service delivery (Holland 2018) may be an easier target of electoral clientelism and vote buying which, in turn, further diminishes expectations of benefiting from any wider and more 3 programmatic redistributive efforts. Experiences of electoral clientelism thereby add to the potential expla- nations as to why citizens fail to express demand for broad-based government redistribution where it is needed the most. Second, our study contributes to shed light on whether the association between electoral clientelism and demand for programmatic redistribution is shaped by perceptions of widespread corruption and low polit- ical trust – signaling disillusionment about the functioning of the democratic system writ large – or whether electoral clientelism serves to signal commitment to the distribution of future particularistic or club goods, thereby providing perceptions of some level of material security. Our findings suggest that experiences with electoral clientelism diminishes demand for programmatic redistribution particularly among citizens who have high levels of trust and believes that the political system is relatively free from corruption. For these groups – whose beliefs in the political system are relatively sanguine – being confronted with clientelist practices may induces a particularly strong sense of disengagement with programmatic forms of redistribu- tion and thereby a weaker demand for broad-based public services. However, electoral clientelism may well be perceived as a credible commitment to alternative modes of redistribution and the distribution of future particularistic or club goods. This may set societies into a suboptimal trajectory of high clientelist–low public service delivery equilibrium, resulting in failure by governments to deliver broad-based public goods. Electoral Clientelism and Demand for Public Services Most research on electoral clientelism has focused on the strategies parties use to target voters – and whether such strategies are useful for generating political support and mobilizing voters during elections. Particular attention has been given to whether parties target swing or core voters (Stokes et al. 2013; Nichter 2008; Stokes 2005), the impact of poverty (Kao et al. 2017; Mares and Young 2016; Jensen and Justesen 2014; Stokes et al. 2013), and how parties use local brokers to monitor voter compliance (de Kadt and Larreguy 2018; Rueda 2017; Szwarcberg 2015; Stokes et al. 2013) or to shape the opinions of local networks of voters (Schaffer and Baker 2015). In fact, a number of studies find that vote buying and clientelist ties may be successful strategies for mobilizing voters and generating political support – particularly for incumbent par- ties with access to government resources (Kramon 2016; Vicente and Wantchekon 2009; Bratton 2008; Wantchekon 2003; Brusco et al. 2004; Bauhr & Charron. 2018).1 These findings are challenged by studies showing that the secret ballot gives voters autonomy to accept vote bribes from any clientelist party while simultaneously voting for their preferred party (van de Walle 2007; Lindberg 2013). Other studies similarly question the effectiveness of vote buying in changing electoral outcomes (Guardado and Wantchekon 2017; Conrad-Cruz and Logan 2012). This implies that there is not necessarily a trade-off between accepting a clientelist offer and demanding programmatic public services. Existing work also investigate the consequences of vote buying for the economy and the supply of public policies. While a number of arguments (Hicken 2011; Vicente and Wantchekon 2009; Stokes 2007) and formal models (Aidt and Jensen 2017; Robinson and Verdier 2013; Keefer and Vlaicu 2008; Baland and Robinson 2012) emphasize that clientelism and vote buying have adverse effects on the economy and public goods provision, empirical evidence is limited. Keefer (2007) provides indirect cross-country evidence that clientelism is associated with under provision of programmatic redistribution and public goods. Using his- torical data from Chile, Baland and Robinson (2012) show that land prices responded rapidly to electoral 4 reforms that increased voter autonomy, thereby weakening the political power of landed elites. In a related paper, Larreguy (2013) argues that where the PRI party in Mexico is better able monitor their political brokers – thereby putting pressure on them to mobilize their clientelistic network – the supply of public schooling is lower. Drawing on evidence from India, Anderson et al. (2015) shows that clientelist relation- ships and elite dominance lead to under-provision of pro-poor public services, and Khemani (2015) finds a negative association between vote buying and pro-poor policies for a selection of municipalities in the Phil- ippines. Most of the existing literature, therefore, emphasizes that clientelism affects supply-side of pro- grammatic policies and pro-poor services by creating incentives for parties to divert resources away from those areas and towards clientelist modes of redistribution (Khemani 2015; Larreguy 2013). We know less, however, about the link between electoral clientelism and public demand for public services. Why Electoral Clientelism Shapes Preferences for Redistribution There are at least two, widely divergent, reasons to expect that electoral clientelism may undermine voter demand for public service delivery. First, to the extent that voters perceive electoral clientelism as a form of corruption (Weitz-Shapiro 2014), exposure to electoral clientelism may reduce citizens expectations of ever benefiting from government service. Thus, if electoral clientelism signals to voters that corruption is widespread, it may lead to alienation and mistrust and thereby lower expectations of benefiting from gov- ernment services (Bauhr and Grimes 2014). The growing literature on why citizens fail to demand public services in contexts where they would presumably benefit from it the most point to this potential alienation effect, and that the mere exposure to inequality may lead the poor to perceive widespread inequality to be natural state of affairs (Ansell and Samuels 2015; Cramer and Kaufman 2011).2 In contexts where the elite is corrupt or fail to represent voters, citizens may be uninformed about their material interest, for instance when unions and left-wing parties are non-existent (Bradley et al. 2003; Morgan and Kelly 2017). Further- more, in contexts where access to welfare systems excludes the poor, citizens may not support them simply because they do not expect to be beneficiaries of the welfare system (Holland 2018). If electoral clientelism signals to voters that politicians are corrupt, experiences of electoral clientelism will reduce trust in govern- ment and thereby lower demand for public services. An alternative explanation for the link between electoral clientelism and citizen demand for public services emphasizes that clientelism may be used by parties to obfuscate the undesirable features of clientelism (Mares and Young 2019), such that voters may perceive electoral clientelism as an implicit signal of com- mitment to post-electoral redistribution (Kramon 2016). In a study on Kenya, Kramon (2016) suggest that voters determine a candidate’s credibility along three dimensions: competence, trustworthiness, and elec- toral viability. In an experimental setting, Kramon found that information about electoral handouts in- creased both voter perceptions of a candidate’s electoral viability and expectations that politicians will pro- vide resources and benefits to poor voters after the election. If voters expect that politicians will provide particularistic payoffs once they assume office, this may effectively lower citizens’ expected material de- pendence on more programmatic forms of redistribution. Even in established democracies – but especially in newer ones – the state is present to varying degrees, with some areas lacking government infrastructure (transportation and communication), services (schools and health clinics), and the organizational capacity needed to enforce laws and regulations. The reach of the state (Herbst 2000; Mann 1984; Migdal 1988; Soifer 2008) may affect demands for accountability (Bauhr and Grimes 2021). When citizens have no access 5 to and do not benefit from state-provided public goods such as education and health care, they are less likely to expect the state to provide these kinds of services. Moreover, where state reach and informational capacity are weak, the state will be unable to collect taxes, which undermines the social contract – along with citizens’ interest in politics. In the context of the African region, paying taxes is a strong and robust predictor of individuals’ interest in politics (Broms 2015; Persson and Rothstein 2015). Experiences of elec- toral clientelism may reinforce this dynamic, by reducing the material incentives to seek and demand access to public goods. In other words, experiences of electoral clientelism establishes or (re)activates norms of reciprocity between elected officials and citizens. It may thereby create an effective linkage between citizens and politicians (Kitschelt and Wilkinson 2007). While studies suggest that voters may indeed link electoral clientelism to perceptions of political corruption (Weitz-Shapiro 2014), experiences of clientelism in combination with deficient public service delivery does not necessarily make citizens feel indifferent towards politics or make them “withdraw even further from the orbit an already marginal state” (Bratton and Chang 2006, 1063; see also Kitschelt et al. 2007, 2010). Quite to the contrary, clientelism may coexist with a relatively high level of trust in politicians as brokers of access to material goods and public services. In contexts where public services are weak or limited, voters may also have greater trust in politicians using clientelist strategies. Indeed, clientelism may make voters more forgiving of corrupt politicians (Bøttkjær and Justesen 2021). This gears voters toward seeking and demanding particularistic payoffs as opposed to programmatic public goods. We therefore expect that experiences of electoral clientelism undermine demand for programmatic public services. Thus, experiences of electoral clientelism may lead to both disengagement with programmatic forms of public service provision, and raise expectations on future particularistic payoffs, thereby undermining de- mand for programmatic public goods. In the following we investigate this relationship in the South African context. Data and measurement We test our hypothesis using survey data from a national representative survey collected in South Africa following the 2016 municipal elections (Justesen and Schulz-Herzenberg 2018). While South Africa is cate- gorized as an upper-middle income country, it is also one of the most unequal countries in the world (World Bank 2016). Access to public services is far from evenly distributed across different localities and segments of the population. Although government cash transfer programs have been expanded significantly through a relatively encompassing system of social grants (Plagerson et al. 2019), the country continues to struggle with significant problems of poverty and destitution – creating fertile grounds for clientelism (Stokes et al. 2013; Jensen and Justesen 2014; Scott 1969). In addition, corruption has increasingly become a problem at all levels of government with ramifications for citizens and public service delivery. Post-apartheid reform of the South African economy and political system has been spearheaded by the African National Congress (ANC) – the dominant party in government party since the inaugural democratic election in 1994. However, the uninterrupted dominance of the ANC in South African politics has also enabled the party to develop into a ‘party-state’ where the boundary between the ANC party and the South 6 African state is blurred or altogether fused (Southall 2016). This places the ANC in a privileged position in terms of control of both programmatic and non-programmatic modes of redistribution. Indeed, as part of the effort to govern South African society, the ANC has developed an elaborate network of local party branches throughout the country, which allows party operatives to serve as brokers of access to government services for citizens at the local level. While party mediation may facilitate access to goods and services for citizens, it also enables the ANC to use and develop clientelist modes of distribution in their interactions with citizens. In fact, qualitative evidence suggests that the ANC does rely on clientelist strategies as part of its efforts to mobilize political support – particularly in poorer areas and townships (Wegner et al. 2019; Dawson 2014; Darracq 2008). The survey data we use was collected in collaboration with Citizen Surveys, a South African research con- sultancy. The survey was fielded in all nine provinces in South Africa in the wake of the August 2016 mu- nicipal elections. The sample size is 3,210 and is representative of voting-age citizens in South Africa. To ensure a national representative sample, sampling of respondents was stratified based on province, ur- ban/rural area, and race, which ensures sufficient coverage of different subgroups. Census data from Sta- tistics South Africa was first used to draw a sample of enumeration areas (EA) – the smallest geographical unit for which demographic data are available. Within EAs, enumerators performed random walks in order to select households for the sample, and – finally – within the randomly selected household, a randomization algorithm pre-coded onto the tablet was used to select the respondent being interviewed. All interviews were performed by trained enumerators in face-to-face interviews using a standardized, tablet-based ques- tionnaire available to respondents in one of six languages. The key dependent variable in our study is demand for programmatic redistribution. To measure support for redistribution, we create an index scaled from 0 to 100 consisting of five questions asking about re- spondents support for government redistribution on key welfare issue such as education and social grants. High values of the index indicate support for more government redistribution. The question asked was the following: “Thinking about government spending on the following areas, should there be much more spending than now, more spending than now, the same spending as now, less spending than now, or much less spending than now?” Before reading the question, respondents were told that answering “more” could require a tax increase, while answering “less” could require a reduction in those services. For our index, we use responses pertaining to government spending on a) health, b) education, c) unemployment benefits, d) old age pension, and e) social grants (Cronbach’s alpha for the redistribution index is 0.87). For each item, responses are given on five-point scale from ‘much more than now’ to ‘much less than now’. To avoid sequencing effects in respondent answers – for instance, effects of always asking about health spending first – the order of the different spending categories were randomized for each interview. Figure 1 shows the distribution of the dependent variable – the index of public demand for redistribution. It is obvious that the index is remarkably skewed towards more spending on key welfare issues – and with a mean of 78 suggests significant popular support for increased government redistribution in South Africa. 7 The independent variable of interest to our study is experiences of electoral clientelism – that is, whether people have been targeted by a political party with offers of material benefits in exchange for a specific type of political behavior at the elections. We take a broad view of electoral clientelism to include strategies involving vote buying, turnout buying, and abstention buying, or combinations of these (Gans-Morse et al. 2014). To measure electoral clientelism, we rely on the following three questions: Vote buying: How often (if ever) did a candidate or someone from a political party offer you something, like food, or a gift or money if you would vote for them in the elections? Turnout buying: How often (if ever) did a candidate or someone from a political party offer you something, like food, or a gift or money if you would show up to vote in the elections? Abstention buying: How often (if ever) did a candidate or someone from a political party offer you something, like food, or a gift or money if you would not go and vote in the elections? We combine responses to these questions into a single binary measure that equals one (1) if a respondent answered ‘once or twice’ or ‘often’ to any of questions, and zero (0) otherwise. Using this metric, around 8.5 percent of respondents in the sample report experiences with being offered material benefits in exchange for their vote, electoral participation, or abstention in the run-up to the 2016 municipal elections. While our design does not allow us to estimate causal effects, we include a large number of control variables that could confound the relationship between electoral clientelism and demand for public redistribution. These include standard socio-economic variables like poverty, education, social grant receipt, unemploy- ment status, age, gender, racial classification, respondents’ partisan attachments, as well as ethnic and reli- gious group affiliation. We also include a full set of region or municipality fixed in the regressions. All results are obtained using OLS regression with standard errors that are clustered at the level of enumeration areas 8 – the smallest geographical unit for which demographic data are available and within which households are sampled. Detailed variable descriptions and summary statistics are available in appendix A. Results Table 1 shows the main results for testing whether electoral clientelism is negatively associated with demand for government spending on public services. To begin with, recall from Figure 1 that – on average – there is significant public support for increased redistribution on key welfare issues like health, education, and social grants among the South African population. Model 1 in Table 1 starts by showing results from a simple bivariate regression with electoral clientelism as the sole regressor. The constant, therefore, shows the average level of support for government redistribution among people who were not approached with offers of electoral clientelism during the 2016 municipal election campaign. The coefficient of electoral clientelism shows that support for government redistribution among respondents who did report experi- ences with offers of clientelist distribution is, on average, almost 14 points lower (and relatively precisely estimated) on a scale from 0 to 100 compared to the baseline group who did not receive clientelist offers in the run-up to the elections. This difference is not only highly significant but also remarkably large in sub- stantial terms, suggesting a strong relationship between electoral clientelism and redistributive preferences. TABLE 1. ELECTORAL CLIENTELISM AND PREFERENCES FOR PROGRAMMATIC REDISTRIBUTION Model (1) (2) (3) (4) (5) (6) (7) (8) Dependent variable Redistribut- Redistribut- Redistribut- Redistribut- Redistribut- Redistribut- Redistribut- Redistribut- ion ion ion ion ion ion ion ion Electoral clientelism -13.81*** -12.97*** -12.77*** -12.53*** -11.83*** -12.62*** -11.34*** -11.80*** (-7.12) (-6.76) (-6.55) (-6.51) (-5.86) (-6.64) (-6.16) (-6.42) Poverty -0.02 -0.06** -0.06** -0.06** -0.04* -0.07*** (-0.73) (-2.19) (-2.07) (-2.27) (-1.70) (-2.62) Education -1.22*** -1.38*** -1.19*** -1.21*** -0.57 (-2.95) (-3.38) (-2.93) (-2.98) (-1.54) Social grant 3.26*** 2.84*** 3.22*** 3.43*** 3.66*** (3.00) (2.62) (2.98) (3.19) (3.55) Unemployed 1.65 1.50 1.51 1.65 2.71*** (1.51) (1.35) (1.40) (1.56) (2.60) Age 0.06* 0.05 0.06* 0.07** 0.07** (1.89) (1.57) (1.71) (2.06) (2.25) Female -0.22 -0.16 -0.43 -0.34 0.08 (-0.23) (-0.17) (-0.44) (-0.35) (0.09) Constant 79.67*** 86.05*** 86.57*** 85.56*** 86.97*** 87.35*** 86.44*** 70.63*** (116.25) (66.22) (55.32) (19.07) (19.32) (18.10) (19.94) (10.18) Province fixed effects No Yes Yes Yes Yes Yes Yes No Municipality fixed effects No No No No No No No Yes Racial classification No No No Yes Yes Yes Yes Yes Urban-rural location No No No Yes Yes Yes Yes Yes Partisanship No No No No Yes No No No Ethnicity fixed effects No No No No No Yes No No Religion fixed effects No No No No No No Yes No Municipality fixed effects No No No No No No No Yes Observations 2,853 2,853 2,794 2,751 2,603 2,751 2,751 2,751 R2 0.027 0.064 0.068 0.090 0.092 0.106 0.125 0.266 Note. OLS regressions. Robust t-statistics in parentheses. Standard errors are clustered by enumeration area. *** p<0.01, ** p<0.05, * p<0.1. 9 In models 2-8, we probe the robustness of the negative association between electoral clientelism and public support for government redistribution. Model 2 includes a full set of fixed effects for the nine provinces of South Africa. This focuses attention to ‘within-province’ variation in the data but does not change the rela- tionship between electoral clientelism and support for redistribution. Model 3 includes a control for poverty which – in theory – should be strongly related to preferences for redistribution (Acemoglu and Robinson 2006) and is widely considered to be one of the most important source of selection into clientelist networks (Jensen and Justesen 2014; Stokes et al. 2013; Scott 1969).3 However, in contrast to expectations of standard models of electoral politics and redistribution (Acemoglu and Robinson 2006; Meltzer-Richard 1981), pov- erty is not related to support for redistribution in model 3. In fact, in models 4-8, results for the poverty variable suggest that support for increased government redis- tribution is not highest among the poorest groups in the population of South Africa, although being a recipient of social grants is positively associated with an increase in support for government spending on social services. The lower demand for public services among the poor – who presumably would be the ones that would benefit from such services the most – may be explained by both informal and formal barriers to access public services and the often-exclusionary nature of public policies that may not reach or benefit the very poorest. This, in turn, limits not only real but also expected benefits from public service delivery and thereby stifles demand (cf. Holland 2018). By contrast, citizens who do benefit from government program, such as in the case of social grant recipients, may also expect more from the state and place greater demands on government redistribution.4 In model 4, we include a large set of socio-economic controls, which also does little to change the relation- ship between electoral clientelism and support for redistribution. Model 5 controls for respondents’ partisan affiliation5, while models 6 and 7 include a full set of ethnicity and religion fixed effects, respectively. Finally, model 8 substitutes the provincial fixed effects with a full set of municipality fixed effects – utilizing ‘within- municipality’ variation in the data. In all cases, the association between electoral clientelism and support for government redistribution remains substantially large and statistically significant. Overall, the results in Table 1 shows a strong and remarkably robust negative relationship between reported experiences of electoral clientelism and public support for government expenditures on public services. Throughout all models, the coefficient is large – in the neighborhood of 11-14 points on scale from 0-100 – and highly significant, and the relationship survives all controls and all fixed effects without major changes in the magnitude of the coefficient or the level of statistical significance. While our estimates do not identify the causal effect of clientelism on support for redistribution, they are consistent with our proposed theory of how electoral clientelism may shape the demand-side of government redistribution. These results, we believe, suggest that clientelist distribution may, in certain contexts, be perceived as a substitute for pro- grammatic redistribution or may alienate people to such an extent that they lose confidence in the ability of the state to deliver public services. In this way, electoral clientelism way serve as a mechanism for political parties to stay in power – without working towards channeling voter preferences for programmatic redis- tribution into the political system. While previous work on clientelism and redistribution has chiefly focused on how clientelism may distort the supply of funds allocated for programmatic redistribution (Khemani 2015; Larreguy 2013; Robinson and Verdier 2013), our story offers a new element to the explanation of how clientelism may shape government redistribution – but one that highlights how voter demand for re- distribution is adversely shaped by parties’ use of clientelist strategies. 10 Alienation or Trust? How can we understand the link between experiences of electoral clientelism and lower demand for public services? As discussed in the above, electoral clientelism tends to reduce public goods provision because it diverts resources away from productive to more unproductive sectors. Electoral clientelism also fuels cor- ruption since political actors seek access to illicit sources of funds to finance vote buying, including the selective implementation of laws and regulations (Holland 2016; Keefer and Vlaicu 2008; Schmidt et al. 1977). To the extent that electoral clientelism is seen as a form of corruption we could expect electoral clientelism to fuel perceptions of corruption. Much in line with studies on how visible corruption produces disengagement (Bauhr and Grimes 2014), electoral clientelism may produce disengagement with the political system writ large, and lower expectations on public goods provision. This would also be in line with recent experimental evidence suggesting that politicians who distribute electoral handouts are seen as less likely to fight corruption (Kramon 2016).6 However, the extent to which electoral clientelism is seen as a form of corruption may vary across subsets of the population, and partly depending on factors such as whether citizens directly benefit from it and initial expectations on government performance. We therefore investigate if electoral clientelism have di- vergent effects among citizens that perceive the system as corrupt and express low levels of trust in political institutions, and among citizens that instead perceive the system as relatively well functioning (i.e., express a high level of political trust and perceive low levels of corruption). Citizens that experience electoral clien- telism, but still believe that corruption levels are relatively low, and institutions are trustworthy, may perceive that electoral clientelism signals trustworthiness and commitment to redistribution and thereby also com- petence and electoral viability (Mares and Young 2019; Kramon 2016). Indeed, electoral clientelism can be used by politicians and their brokers to obfuscate the normatively undesirable features of clientelism (Mares and Young 2019; Kramon 2016) and signal politicians’ future intent to redistribute material resources, and in particular preferential access to scarce public services during as well as after the election (Diaz et al. 2016; Weitz-Shapiro 2012). In a high-corruption environment, voters may even perceive electoral clientelism as a form of “tax refund”, in the sense that corrupt politicians are “paying back” part of the funds they have previously stolen (Lindberg 2003; Banegas 1998). This would be in line with evidence on voters supporting corrupt politicians when lacking “clean” alternatives or when they believe they benefit from the system at hand (Bauhr and Charron 2018; Chang, Golden, and Hill, 2010; Manzetti and Wilson, 2007; Agerberg 2019). In other words, among citizens that express relatively sanguine beliefs about the political system (high trust and low levels of corruption), experiences of electoral clientelism may reduce demand for programmatic redistribution since citizens perceive that there is a trade-off between clientelist and programmatic forms of redistribution. On the other hand, if electoral clientelism produces drops in support, particularly among low trusting individuals, this may suggest that electoral clientelism produces disengagement with all forms of redistribution. To examine how political trust and beliefs about corruption moderate the association between electoral clientelism and demand for programmatic redistribution, we depart by constructing two indices: One for political trust and one for corruption perceptions. The index of political trust consists of seven items: Trust in Jacob Zuma (president of South Africa at the time of the survey), trust in the National Parliament, trust in the Electoral Commission, trust in the municipal council, trust in the mayor of the municipal council, trust in traditional leaders, and trust in the ANC – the governing party. Before constructing the index, we 11 conduct principal components factor analysis, which shows that all items load onto the same latent factor (Cronbach’s alpha for the political trust index is 0.89). Details of the factor analysis are available in appendix B. The corruption perceptions index consists of nine items measuring how respondents assess the likelihood that the following actors or institutions are involved in corruption: President Jacob Zuma, officials in the office of Jacob Zuma, representatives of the National Parliament, the mayor of the municipality, elected members of the municipal council, local government officials, tax officials, traditional leaders, and the po- lice. For the corruption index too, we start by running a principal component factor analysis which likewise shows that all items load onto one factor (Cronbach’s alpha for the corruption index is 0.92). Details are available in appendix C. Subsequently, we construct two additive indices, which for ease of interpretation are scaled from 0 to 100. For the political trust index, low/high values indicate low/high trust. For the corruption perceptions index, low values indicate that respondents believe there is little corruption in the political system, while high values indicate beliefs that corruption is widespread. To examine how beliefs about political trust and corruption moderate the relationship between electoral clientelism and demand for programmatic redistribution, we run regressions similar to model 4 in Table 1 – with demand for programmatic redistribution as the dependent variable – where electoral clientelism is interacted with political trust and corruption perceptions, respectively. In figure 2, we show the result for the first regression where electoral clientelism is interacted with political trust (full results are available in appendix D). Figure 2 shows the marginal association between electoral clientelism and demand for programmatic redistribution, conditional on levels of political trust (along the 12 x-axis). The solid line shows the marginal association between electoral clientelism and demand for pro- grammatic redistribution and how it changes as political trust increases from low to high levels. The dotted lines show 95 percent confidence intervals. The plot clearly shows that experiences with electoral clientelism depresses demand for public services, and that the effect is particularly pronounced among citizens with a high level of political trust. In contrast, for people who have low levels of political trust, experiencing elec- toral clientelism does not change their demand for public services. This suggests that encounters with po- litical parties and brokers using clientelist strategies during election campaigns tend to lower demand for programmatic forms of redistribution most strongly among people who have relatively high levels of polit- ical trust. This finding is consistent with the idea that electoral clientelism indeed signals a trade-off between clientelist and programmatic redistribution in that electoral clientelism can coexist both with high trust and low demand for redistribution. This may also signal a commitment to an alternative, more particularistic redistributive system. Figure 3 shows an equivalent plot where the electoral clientelism variable is interacted with the index of corruption perceptions (full results are available in appendix E). The plot shows how the association be- tween electoral clientelism and demand for programmatic redistribution changes as perceptions of corrup- tion change from low to high along the x-axis. Figure 3 shows that experiences with electoral clientelism tend to depress demand for government redistri- bution among people who believe political corruption is moderately common or relatively uncommon. If people hold beliefs that corruption is widespread (equivalent to high values of the corruption perceptions index), encounters with parties or brokers engaging in clientelist distribution during elections make little 13 difference to their preferences for programmatic redistribution. In other words, electoral clientelism seems to lower demand for government redistribution the most among people who think of the world of politics as being relatively free from corruption – or at least not embroiled in corruption. Demand for redistribution is less affected among people who are already inclined to associate political actors and institutions with corruption. This too suggests that electoral clientelism may induce a sense of disengagement with program- matic (but not necessarily particularistic) modes of redistribution in that citizens who hold relatively san- guine beliefs about the political system– are more likely to express lower expectations on public service provision. Discussion and Conclusion The results of the analysis above show that experiences of electoral clientelism are strongly associated with lower demand for programmatic public service provision in sectors such as health, education, and unem- ployment. The negative association is stronger for people with high levels of political trust and beliefs that the political system is relatively uncorrupted – suggesting that these groups become particularly alienated when faced with political parties engaging in clientelist practices and perceive a trade-off between these two modes of redistribution. We suggest that both expectations of post-election particularistic benefits and norms of reciprocity may explain why experiences of electoral clientelism reinforces the high electoral clientelism-low public service delivery equilibrium. Thus, the distribution of electoral handouts does not only distort the allocation of public funds, it also shape the very demand for how those funds should be used. These results have im- portant implications for explaining why democratic systems fail to direct resources into universal public goods as opposed to particularistic payoffs, and ultimately prospects of securing access to public services where they are the most needed. What are the prospects of breaking this vicious cycle? Some contend that clientelism operating in a demo- cratic context may have more benign effects. For instance, van de Walle (2014) suggests that democratiza- tion in Africa has made clientelism more broadly redistributive, focusing less on elite patronage and more on transfers to broader masses of voters. Nichter and Peress (2016) move on to develop a demand-side explanation of electoral clientelism emphasizing how clientelist exchanges arise and endure because voters often request goods and services from politicians and threaten to cast their vote for the opposition if their demands are not fulfilled. This argument would be consistent with an interpretation of clientelism where parties are more responsive to voter demands (Khemani 2015, 85). However, if such demands are particu- laristic rather than programmatic, this may not be conducive to moving societies towards the development of universal public service systems or more encompassing welfare states that might contribute to alleviate poverty and counter the economic inequalities arise in political systems characterize by systemic clientelist distribution. The main policy implication of our findings is therefore that building public services that ben- efit broad groups of citizens require that we take a broader view of anti-corruption campaigns. Specifically, anti-corruption campaigns need to address corruption perpetrated by politicians and bureaucrats in office – but they also need to focus on combatting corruption in the process of getting elected for public office, because such electoral corruption may contribute to undermine citizens’ confidence in and demand for public services. 14 Notes 1 A related, but distinct, literature examines whether conditional cash transfer programs are effective at generating political support (De La O 2015; Magaloni et al. 2015; Zucco 2013). 2 Sands and de Kadt (2020), however, find that exposure to inequality generates increased support for re- distribution in low-income areas. 3 The poverty measure captures people’s experiences with lack of access to basic household necessities like food and clean water (Bratton et al. 2005). The index is based on six items measuring how often, if ever, over the past year respondents have experienced lack of food, clean water, medicines or medical treatment, fuel for cooking, a cash income, or electricity (because of inability to pay electricity bills). High values indi- cate poverty. Note that low values on the index does not indicate that people are rich, but simply that they are ‘not poor’ in the sense of not lacking access to basic household necessities. 4 Our results also show that older people are more likely to demand greater levels of government redistri- bution, but that education is associated with a lower level of demand for more government expenditures on public services. 5 Partisan affiliation is measured as a four-way classification between non-partisans (reference), people iden- tifying with the ANC, or the DA, or another political party in South Africa. 6 In Kramon’s study electoral clientelism reduced the share of respondent who believe that the candidate would “help the economy grow and make policies benefit all Kenyans” (2016, 487). It is important to note that these are voters that hear about vote buying rather, as in our case, voters that have real world personal experiences with vote buying. 15 References Acemoglu, D., and Robinson, J. A. (2005). Economic Origins of Dictatorship and Democracy. Cambridge University Press. https://doi.org/10.1017/CBO9780511510809 Anderson, S., Francois, P., and Kotwal, A. (2015). “Clientelism in Indian Villages”. The American Economic Review 105(6), 1780–1816. Ansell, S and D. Samuels (2015). Inequality and Democratization: An Elite Competition Approach. Cambridge: Cambridge University Press. Bauhr, Monika, and Marcia Grimes. 2014. “Indignation or Resignation: The Implications of Transparency for Societal Accountability.” Governance 27 (2): 291–320. https://doi.org/10.1111/gove.12033 Bauhr, Monika, and Nicholas Charron. 2018. “Insider or Outsider? Grand Corruption and Electoral Accountability.” Comparative Political Studies, June, 0010414017710258. https://doi.org/10.1177/0010414017710258 Bauhr, Monika and Marcia Grimes. 2021. “Democracy and Quality of Government” in Bauhr, Monika, Andreas Bågenholm, Marcia Grimes and Bo Rothstein (eds). The Oxford Handbook of Quality of Government. Oxford: Oxford University Press. https://academic.oup.com/edited-volume/33431 Balafoutas, L. (2011). “How much income redistribution? An explanation based on vote-buying and corruption”. Public Choice 146(1/2), 185–203. Baland, J-M., and Robinson, J. (2012). “The political value of land: Political reform and land prices in Chile”, American Journal of Political Science 56(3), 601-619. Banegas, Richard. (1998). “Marchandisation du Vote, Citoyenneté et Consolidation Démocratique au Benin”, Politique Africaine 69(1), 75–88. Berenschot, W., and Aspinall, E. (2020). “How clientelism varies: Comparing patronage democracies”, Democratization 27(1), 1–19. https://doi.org/10.1080/13510347.2019.1645129 Bradley, D., Huber, E., Moller, S., Nielsen, F., and Stephens, J. D. (2003). “Distribution and Redistribution in Postindustrial Democracies”, World Politics 55(2), 193–228. https://doi.org/10.1353/wp.2003.0009 Brusco, V., Nazareno, M., and Stokes, S. C. (2004). “Vote Buying in Argentina”, Latin American Research Review 39(2), 66–88. https://doi.org/10.1353/lar.2004.0022 Bøttkjær, Louise T. and Mogens K. Justesen (2021). “Why do voters support corrupt politicians? Experimental evidence from South Africa”, Journal of Politics 83(2), 788-793. https://doi.org/10.1086/710146 Carlson, E. (2021). “Distribution or Diversion? Distribution of Local Public Goods in the Presence of Clientelist Brokers”, Political Research Quarterly 74(2), 274-287. 16 Casas, A. (2018). “Distributive Politics with Vote and Turnout Buying”, American Political Science Review 112(4), 1111–1119. https://doi.org/10.1017/S0003055418000291 Cramer, B. D., and Kaufman, R. R. (2011). “Views of Economic Inequality in Latin America”, Comparative Political Studies 44(9), 1206-1237. Cruz, C. (2019). “Social Networks and the Targeting of Vote Buying”, Comparative Political Studies, 52(3), 382–411. https://doi.org/10.1177/0010414018784062 Cruz, C., and Keefer, P. (2015). “Political Parties, Clientelism, and Bureaucratic Reform”, Comparative Political Studies, 48(14), 1942–1973. https://doi.org/10.1177/0010414015594627 Cruz, C., Labonne, J., and Querubín, P. (2020). “Social Network Structures and the Politics of Public Goods Provision: Evidence from the Philippines”, American Political Science Review 114(2), 486–501. https://doi.org/10.1017/S0003055419000789 Darracq, V. (2008). “The African National Congress (ANC) Organization at the Grassroots”, African Affairs 107(429), 589–609. Dawson, H. J. (2014). “Patronage from below: political unrest in an informal settlement in South Africa”, African Affairs 113(453), 518–539. De la O, A. L., and Rodden, J. A. (2008). “Does religion distract the poor?: Income and issue voting around the World”, Comparative Political Studies, 41(4-5), 437-476. https://doi.org/10.1177/0010414007313114 de Kadt, D., and Larreguy, H. A. (2018). ”Agents of the Regime? Traditional Leaders and Electoral Politics in South Africa”, Journal of Politics 80(2), 382–399. https://doi.org/10.1086/694540 Diaz-Cayeros, Alberto, Federico Estévez, and Beatriz Magaloni. 2016. Strategies of Vote Buying: Democracy, Clientelism, and Poverty Relief in Mexico. New York: Cambridge University Press. Gans-Morse, J., Mazzuca, S., and Nichter, S. (2014). “Varieties of Clientelism: Machine Politics during Elections”, American Journal of Political Science 58(2), 415–432. Gough, I., Wood, G., Barrientos, A., Bevan, P., Davis, P., and Room, G. (2004). Insecurity and Welfare Regimes in Asia, Africa and Latin America: Social Policy in Development Contexts. Cambridge University Press. https://doi.org/10.1017/CBO9780511720239 Guardado, J., and Wantchékon, L. (2018). “Do electoral handouts affect voting behavior?”, Electoral Studies 53, 139–149. https://doi.org/10.1016/j.electstud.2017.11.002 Herbst J (2000). States and Power in Africa. Comparative Lessons in Authority and Control. New Jersey: Princeton University Press. Hicken, A. (2011). “Clientelism”, Annual Review of Political Science 14(1), 289–310. https://doi.org/10.1146/annurev.polisci.031908.220508 17 Holland, A. C. (2016). “Forbearance”, American Political Science Review, 110(2), 232–246. https://doi.org/10.1017/S0003055416000083 Holland, A. C. (2018). “Diminished Expectations: Redistributive Preferences in Truncated Welfare States”, World Politics 70(4), 555–594. Jensen, P. S., and Justesen, M. K. (2014). ”Poverty and vote buying: Survey-based evidence from Africa”, Electoral Studies 33, 220–232. https://doi.org/10.1016/j.electstud.2013.07.020 Justesen, M. K., and Bjørnskov, C. (2014). “Exploiting the Poor: Bureaucratic Corruption and Poverty in Africa”, World Development 58, 106–115. https://doi.org/10.1016/j.worlddev.2014.01.002 Justesen, M. K., and Schulz-Herzenberg, C. (2018). “The Decline of the African National Congress in South Africa’s 2016 Municipal Elections”, Journal of Southern African Studies 44(6), 1133–1151. https://doi.org/10.1080/03057070.2018.1539376 Kao, K., Lust, E., and Rakner, L. (2017). Money Machine: Do the Poor Demand Clientelism? (SSRN Scholarly Paper ID 3581811). Social Science Research Network. https://doi.org/10.2139/ssrn.3581811 Keefer, P., and Khemani, S. (2009). “When Do Legislators Pass on Pork? The Role of Political Parties in Determining Legislator Effort”, The American Political Science Review 103(1), 99–112. https://doi.org/10.2307/27798488 Keefer, P., and Vlaicu, R. (2017). “Vote buying and campaign promises”, Journal of Comparative Economics 45(4), 773–792. https://doi.org/10.1016/j.jce.2017.07.001 Khemani, S. (2015). “Buying votes versus supplying public services: Political incentives to under-invest in pro-poor policies”, Journal of Development Economics 117, 84–93. https://doi.org/10.1016/j.jdeveco.2015.07.002 Kitschelt, H., Hawkins, K. A., Luna, J. P., Rosas, G., and Zechmeister, E. J. (2010). Latin American Party Systems. Cambridge: Cambridge University Press. https://doi.org/10.1017/CBO9780511750311 Kitschelt, H., Wilkinson, S. I., and Wilkinson, P. (2007). Patrons, Clients and Policies: Patterns of Democratic Accountability and Political Competition. Cambridge: Cambridge University Press. Kramon, E. (2016a). “Electoral Handouts as Information: Explaining Unmonitored Vote Buying”, World Politics 68(3), 454–498. https://doi.org/10.1017/S0043887115000453 Kramon, E. (2016b). “Where is vote buying effective? Evidence from a list experiment in Kenya”, Electoral Studies 44, 397–408. https://doi.org/10.1016/j.electstud.2016.09.006 Kramon, E. (2018). Money for Votes: The Causes and Consequences of Electoral Clientelism in Africa. Cambridge: Cambridge University Press. Lindberg, S. I. (2013). “Have the cake and eat it: The rational voter in Africa”, Party Politics 19(6), 945–961. https://doi.org/10.1177/1354068811436030 18 Mares, I., and Young, L. (2019). Conditionality and Coercion: Electoral Clientelism in Eastern Europe. Oxford: Oxford University Press. Mares, I., and Young, L. E. (2018). “The Core Voter’s Curse: Clientelistic Threats and Promises in Hungarian Elections”, Comparative Political Studies 51(11), 1441–1471. https://doi.org/10.1177/0010414018758754 Mares, I., and Young, L. (2016). “Buying, Expropriating, and Stealing Votes”, Annual Review of Political Science 19(1), 267–288. https://doi.org/10.1146/annurev-polisci-060514-120923. Markowski, R. (2019). “Creating Authoritarian Clientelism: Poland After 2015”, Hague Journal on the Rule of Law 11(1), 111–132. https://doi.org/10.1007/s40803-018-0082-5 Markussen, T. (2011). “Inequality and political clientelism: Evidence from South India”, Journal of Development Studies 47(11), 1721-1738. Meltzer, A. H., and Richard, S. F. (1981). “A Rational Theory of the Size of Government”, Journal of Political Economy 89(5), 914–927. JSTOR. Morgan, J., and Kelly, N. J. (2010). ”Explaining Public Attitudes toward Fighting Inequality in Latin America”, Poverty and Public Policy 2(3), 79–111. https://doi.org/10.2202/1944-2858.1091 Nichter, S. (2008). “Vote buying or turnout buying? Machine politics and the secret ballot. American Political Science Review 102(1), 19–31. Pellicer, M., Wegner, E., Benstead, L. J., and Lust, E. (2021). “Poor people’s beliefs and the dynamics of clientelism”, Journal of Theoretical Politics 33(3), 300-332. https://doi.org/10.1177/09516298211003661 Persson, A., & Rothstein, B. (2015). It’s My Money: Why Big Government May Be Good Government. Comparative Politics, 47(2), 231–249. http://www.jstor.org/stable/43664141 Persson, A. L., Rothstein, B., & Teorell, J. (2019). Lost in Transition: A Bottom-Up Perspective on Hybrid Regimes. Annals of Comparative Democratization, 17(3), 10-12. Plagerson, S., Patel, L., Hochfeld, T., and Ulriksen, M. S. (2019). “Social policy in South Africa: Navigating the route to social development”, World Development 113, 1–9. https://doi.org/10.1016/j.worlddev.2018.08.019 Remmer, K. L. (2010). “Political Scale and Electoral Turnout: Evidence From the Less Industrialized World”, Comparative Political Studies 43(3), 275–303. https://doi.org/10.1177/0010414009352638 Robinson, J. A., and Verdier, T. (2013). “The Political Economy of Clientelism”, The Scandinavian Journal of Economics 115(2), 260–291. https://doi.org/10.1111/sjoe.12010 Sands, Melissa and Daniel de Kadt (2020). ”Local Exposure to Inequality Raises Support of People of Low Wealth for Taxing the Rich”, Nature 586, 257-261. 19 https://doi.org/10.1038/s41586-020-2763-1 Soifer, H. (2008). “State Infrastructural Power: Approaches to Conceptualization and Measurement”, Studies in Comparative International Development, 43(3/4), 231–251. https://doi.org/10.1007/s12116-008-9028-6 Soifer, H. D. (2012). “Measuring state capacity in contemporary Latin America”, Revista de Ciencia Política, 32(3), 585–598. https://doi.org/10.4067/S0718-090X2012000300004 Son, B. and N. Bellinger (2021). “The health cost of autocratization”, Forthcoming in Journal of Development Studies. https://doi.org/10.1080/00220388.2021.2017891 Stockemer, D., LaMontagne, B., and Scruggs, L. (2013). “Bribes and ballots: The impact of corruption on voter turnout in democracies”, International Political Science Review, 34(1), 74–90. https://doi.org/10.1177/0192512111419824 Stokes, S. C. (2005). “Perverse Accountability: A Formal Model of Machine Politics with Evidence from Argentina”, American Political Science Review, 99(3), 315–325. https://doi.org/10.1017/S0003055405051683 Stokes, S. C., Dunning, T., Nazareno, M., and Brusco, V. (2013). Brokers, Voters, and Clientelism: The Puzzle of Distributive Politics. Cambridge: Cambridge University Press. https://doi.org/10.1017/CBO9781107324909 Szwarcberg, M. (2012). “Uncertainty, Political Clientelism, and Voter Turnout in Latin America: Why Parties Conduct Rallies in Argentina”, Comparative Politics 45(1), 88–106. Van de Walle, S. (2016). “When public services fail: A research agenda on public service failure”, Journal of Service Management, 27(5), 831–846. https://doi.org/10.1108/JOSM-04-2016-0092 Vicente, P. C., and Wantchekon, L. (2009). “Clientelism and vote buying: Lessons from field experiments in African elections”, Oxford Review of Economic Policy 25(2), 292–305. Wantchekon, L. (2003). “Clientelism and Voting Behavior: Evidence from a Field Experiment in Benin”, World Politics 55(3), 399–422. Weghorst, K. R., and Lindberg, S. I. (2013). “What Drives the Swing Voter in Africa?”, American Journal of Political Science 57(3), 717–734. https://doi.org/10.1111/ajps.12022 Weitz-Shapiro, Rebecca. (2012). “What Wins Votes: Why Some Politicians Opt Out of Clientelism”, American Journal of Political Science 56(3), 568–83. 20 Appendix Appenix 1: Suitable name for the appendix Appenix 2: Suitable name for the appendix 21