Accepting or Resisting? Citizen Responses to Corruption Across Varying Levels of Competence and Corruption Prevalence

Abstract

The literature studying citizen responses to exposed political corruption is rapidly growing. While some studies explore how information credibility and group identities can reduce the electoral impact of the exposure of corruption, this article addresses different mechanisms for weak electoral accountability for corruption: candidate competence in public works provision and corruption prevalence. It uses a vignette experiment embedded in a national survey in Peru to isolate the causal effect of political corruption on electoral support. The results suggest that even types of corruption with side benefits would be harshly punished when attributed to incompetent politicians. They also indicate that while voters punish corruption more leniently when a candidate is competent, they respond negatively to corruption regardless of the prevalence of corruption, which casts doubt on the idea that voters in highly corrupt environments are acceptant of corruption.

Keywords

corruption survey experiments elections public works provision Latin America

Introduction

A rapidly growing literature on citizen responses to corruption has advanced our understanding about how information credibility and group identities can reduce the electoral impact of the exposure of corruption (Anduiza et al., 2013; Chong et al., 2015; Weitz-Shapiro and Winters, 2017). In comparison, we know less about how some contextual conditions undermine informed voters’ ability to punish corrupt candidates. While we know that economic gains are one prominent factor explaining why voters sometimes overlook corruption (Fernández-Vázquez et al., 2016; Manzetti and Wilson, 2007; Rosas and Manzetti, 2015; Zechmeister and Zizumbo-Colunga, 2013), we are still unclear about the nuances of the relationship between corruption accusations and candidate competence in public works provision. Empirical studies on the corruption-competence trade-off use different causal logics to explain why economic welfare leads voters to put up with corruption. Moreover, our existing knowledge about the trading of corruption for economic well-being has not yet offered guidance about how this exchange plays out in highly corrupt environments. Cross-national studies suggest that the normalization of bribery might increase tolerance to corruption, but we need new studies to shed light on whether widespread corruption does in fact translate into electoral impunity for politicians suspected of corrupt behavior.

This article makes several contributions that advance our understanding of the micro-foundations of political accountability for corruption. First, in order to understand the discrepancies in the empirical literature on corruption, this article theoretically distinguishes between different ways in which economic well-being drives voters’ tolerance of corruption. The exciting empirical literature about the long-standing idea that tolerance of corruption is a function of the material benefits that voters associate with corrupt politicians has produced mixed evidence. While some studies have found that economic side benefits could help protect corrupt politicians from electoral penalties (Klašnja and Tucker, 2013; Zechmeister and Zizumbo-Colunga, 2013), other research reach different or even opposing conclusions (Esaiasson and Muñoz, 2014; Winters and Weitz-Shapiro, 2013).

The present article builds on the literature’s disagreement and analyzes the relationship between corruption and economic welfare with additional nuance. I focus on the specific ways in which economic well-being drives voters’ tolerance of corruption. One version of the link that is often found in the trade-off literature suggests that certain types of corruption bring side benefits that would render corruption innocuous to voters. Instead, another version of the link, less often found in the experimental literature, would argue that economic performance induces voters to evaluate corruption of competent politicians in a fundamentally different way from that of incompetent ones. This is an important theoretical distinction between two trade-off accounts, one based on corruption types and another based on candidate types. This subtle distinction matters because if voters are more lenient with corrupt candidates who are competent, even types of corruption that carry side benefits will be harshly punished when attributed to incompetent politicians.

A second important contribution of the present article is to contrast the trade-off explanation to another more explicitly related to the social context in which corruption information is disseminated: perceptions of widespread corruption. Existing cross-national studies showed that voters sometimes overlook corruption to preserve economic well-being in highly corrupt environments, but they did not test the normalization mechanism directly (the process by which the normalization of bribery reduces electoral accountability for corruption). Therefore, this article separates the trade-off effect from the “they are all corrupt” effect by manipulating voters’ expectations of corruption and leveraging within-country variation of corruption perceptions. In particular, the article assesses a normalization mechanism through an experimental design that allows for the measurement of voter responses to corruption under different but comparable corruption level settings.

Finally, this article sheds light on the politics of electoral impunity for corruption in an understudied country, Peru, a low-middle income nation with a long history of corruption and weak rule of law (Quiroz, 2008). Like other young democracies, this developing country has enacted reforms aiming at increasing government transparency and strengthening political institutions, but these reforms have been implemented unevenly (Levistsky, 2013; Vergara and Watanabe, 2016). More importantly, although corruption has become one of the most critical problems for Peruvians in recent years, corruption concerns vary considerably across the country. Only a handful of regions see corruption as the top political concern, while the rest are primarily worried about the more pressing matters of crime and unemployment.

The article will proceed as follows. First, I review the existing literature on the interaction between corruption, competence, and voting. Second, I discuss the existing literature suggesting the voters’ expectations of corruption might be driving the limits of electoral accountably. Third, I describe the survey experiment I designed for Peru and compare it to other candidate vignettes in corruption experiments. The fourth section presents the results of the experimental study. The final section discusses some of the key implications of the findings and concludes.

Corruption, Competence, and Voting

In spite of the prevailing view that elected officials who are charged with wrongdoings are able to hold on to office or retain popular support, recent experimental studies find that exposing corruption can help prevent the persistence of dishonest officials in government (Banerjee et al., 2010; Bobonis et al., 2010; Ferraz and Finan, 2008). This divergence has renewed the interest in understanding precisely how and when voters use corruption evidence to inform their candidate evaluations and decide their vote (for good reviews of this literature see De Sousa and Moriconi (2013) and De Vries and Solaz (2017)).

An important strand of this emerging literature focuses on information credibility and group identities. For example, some experimental studies show that the effect of exposing corruption depends on the severity of the malfeasance (Chong et al., 2015) and on the credibility of the sources of information (Botero et al., 2015, 2017; Weitz-Shapiro and Winters, 2017). Moreover, other noteworthy studies show that voters are more likely to overlook corruption cases that affect their own party, their own ethnic group, or male politicians (Anduiza et al., 2013; Banerjee and Pande, 2009; Barnes et al., 2017).

In contrast, another strand of the literature focuses on how certain contextual and institutional factors may influence voter responses to exposed corruption. For instance, they focus on the role that features of proportional representation (PR) electoral systems play on creating incentives for corrupt rent-seeking and on reducing opportunities for monitoring, which has inspired a debate about whether electoral rules, and which features, can serve as constrains on corruption or not (Chang, 2005; Chang and Golden, 2007; Kunicová and Rose-Ackerman, 2005). Party composition of government has also motivated important studies that point out that citizens vote corrupt incumbents out of office more often under unified government than under divided government (Schwindt-Bayer and Tavits, 2016; Tavits, 2007). Similarly, good economic performance is believed to protect corrupt governments from public disapproval (Zechmeister and Zizumbo-Colunga, 2013), but our understanding of why and how these political contexts are more (or less) conducive to electoral accountability for corruption is still incomplete.

Among the contextual factors that may influence the decision to punish a corrupt candidate, the economy is one of the most prominent (Manzetti and Wilson, 2007; Rosas and Manzetti, 2015; Zechmeister and Zizumbo-Colunga, 2013). Several studies suggest that tolerance of corruption is a function of economic gains voters associate with corrupt politicians, but those studies reach divergent conclusions. For example, some experimental studies find support for the idea that citizens are pragmatic with regards to corruption, showing that they would tolerate corruption as long as the state of the economy was good (Klašnja et al., 2017; Klašnja and Tucker, 2013). Similarly, other studies show that corruption cases that ensure jobs, service delivery, or other economic gains are less severely punished than those that do not (Botero et al., 2017; Klašnja et al., 2017; Konstantinidis and Xezonakis, 2013). However, still other studies find no support for the “tradeoff hypothesis,” showing that voters are unlikely to support a corrupt politician even if this politician delivered public goods (Winters and Weitz-Shapiro, 2013) or that competent politicians will in fact pay higher costs for corruption than incompetent ones (Esaiasson and Muñoz, 2014). How can we reconcile these diverging findings?

Trade-off Hypothesis: Two Different Accounts

One step toward making sense of the mixed evidence is to consider the channels through which economic benefits might motivate voting for a corrupt politician. A standard account of the trade-off hypothesis attributes tolerance of corruption to differences in corruption types because certain corruption practices may bring side benefits that would render corruption acceptable. Demanding bribes in exchange for public contracts, for example, could be seen as less reprehensible than receiving illegal campaign donations from large corporations. The first type of corruption would produce identifiable side benefits such as jobs and services, whereas the second type would be obscure about how the average voter might benefit from it and would reveal instead a corrupt leader gaining undue political advantage. While all types of corruption involve some form of private gain from public office, only some types of corruption have evident welfare consequences for the electorate (Fernández-Vázquez et al., 2016).¹

The notion that voters deem some corruption types more acceptable than others is consistent with the long-standing idea that voters might reward corruption if they materially benefited from it (Rundquist et al., 1977). Several empirical studies have examined this interpretation of the trade-off hypothesis based on corruption types. For example, Klašnja et al. (2017) finds that corruption that brings construction jobs is punished less harshly than corruption that does not. Botero et al. (2017) finds that corrupt behavior described as clientelism would cost candidates less than corrupt behavior that is mainly seen as private enrichment.² Nonexperimental studies also highlight the role of welfare consequences of certain corruption types. Fernández-Vázquez et al. (2016) for instance finds that welfare-enhancing types of corruption are less detrimental to electoral success than welfare-reducing types. That is, citizens see some forms of offenses as less reprehensible when they have the potential to improve the material well-being of their communities.

However, another plausible instance of the relationship between economic gains and corruption would be one in which voters overlook corruption because the allegedly corrupt politician can sometimes be seen as a competent representative as well. By focusing on candidate type rather than on corruption type, this second interpretation of the trade-off hypothesis allows us to separate the acceptability that certain corruption practices carry from the economic benefit that voters may associate with certain politicians. This distinction is important, because while corruption might bring side benefits, it is still possible that some corrupt mayors were less successful at revitalizing the economy than others. An important implication is that clientelistic politicians would still be punished if they performed worse than their non-clientelistic peers. That is, among candidates who are accused of comparable dishonest behavior, some politicians could be more effective in approving advantageous policies and programs than others. Good economic performance, therefore, would indicate that this authority figure is well-suited for the job, motivating voters to evaluate corruption qualitatively differently among competent politicians than among incompetent ones.

Noticing the subtle difference between two types of channels through which economic benefit shields corrupt politicians (corruption type or candidate type) is particularly relevant for the study of accountability in contexts of uninstitutionalized party systems, in which partisanship is generally weak and non-programmatic linkages dominate (Kitschelt, 2000; Roberts, 2013). When voters cannot rely on party or policy cues to make judgments about electoral choices, they may find signals of quality in candidates’ revealed attributes. Candidate type becomes a very important piece of information to guide voting. In fact, candidate type can be such a strong signal that it may motivate voters to overlook corruption evidence. Indicators of candidate competency and efficiency would, therefore, assist voters in better evaluating a candidate’s fitness for office and electing highly skilled politicians. While acknowledging the importance of previous findings that voters are indeed less likely to disapprove of corruption types that carry side benefits, this article examines another mechanism for the rouba mas faz (“he steals but gets things done”)³ based on candidate type: the role of a candidate’s reputation of being an efficient public manager. We should expect to see that citizens punish corruption less harshly in a competent candidate who has shown the ability to deliver collective benefits while in office than in an incompetent candidate.

The understanding of this theoretical nuance, which is often overlooked in studies of the corruption-competence trade-off, has important consequences for the experimental design. Ideally, we would design an experiment that holds the type of corruption constant, thereby allowing us to test how candidate competence has a distinct effect on citizens’ responses to an identical type of corrupt behavior. With a few exceptions (Esaiasson and Muñoz, 2014; Winters and Weitz-Shapiro, 2013), research on the trade-off hypothesis has mostly examined the failure of political accountability as a result of voters’ acceptance of certain corrupt acts. Here, I propose to examine it as result of the value that voters assign to candidate competence traits. Again, this focus on candidate type is theoretically relevant because, even if corruption brought side benefits, voters would still punish it if corruption came from incompetent politicians. As a result, clientelistic candidates, or candidates engaging in welfare-enhancing types of corruption, might still receive electoral penalties if they lacked a good track record.⁴

Hypothesis 1a. The electoral penalty for corruption is smaller for politicians who are competent in delivering public works than for those who are incompetent in delivering public works.

Another step toward making sense of the mixed evidence in previous work is to consider the type of economic gains for which the literature has suggested that voters might reward competent authorities. Performance associated with these gains has ranged from competence in promoting local economic development (Klašnja and Tucker, 2013) to competence in delivering selective goods such as public works or patronage jobs (Klašnja et al., 2017; Winters and Weitz-Shapiro, 2013).⁵ While subnational authorities often claim credit for infrastructure projects that have the potential to directly impact the material well-being of constituents, public works provision is particularly susceptible to corruption (Kenny, 2009; Locatelli et al., 2017). The public procurement system is often plagued with secretive arrangements between public officials and construction contractors. In fact, the central actor in one of the largest recent corruption scandals in Latin America is the international giant Odebrecht, a construction company that has admitted to paying some $800 million in bribes to public officials across the region. If politicians competent in public works provision were generally perceived as lacking integrity, this perception could be one of the reasons why some studies did not find support for an explicit exchange of corruption for public works. It is possible that if a community receives outstanding public works provision under an authority accused of taking bribes, citizens might take competence as a sign of greater opportunities for corruption.⁶ This discussion yields the following hypothesis:

Hypothesis 1b. The electoral penalty for corruption is larger for politicians who are competent in delivering public works than for those who are incompetent in delivering public works.

Why Should Widespread Corruption Matter?

Another explanation for the mixed evidence in favor of the explicit exchange idea would be that voters condone corruption in competent politicians in part because they expect corruption to be generally widespread among the political elite. In high corruption environments, the comparative disadvantage of a corrupt candidate would be underestimated because of the large probability that other electoral alternatives would also be corrupt (Pavão, 2018) and because of the prospect that any honest candidate would not remain upstanding for a long time (Bauhr and Charron, 2018). Therefore, when voters’ prior expectations are that corruption is a normalized practice, new information about one official’s honest or dishonest behavior might be overlooked. In other words, we should expect that corruption information would not affect voting behavior uniformly across populations with different prior beliefs about corruption (Arias et al., 2016).

The literature has paid attention to the impact of social norms on the likelihood that individuals would engage in bribing (Corbacho et al., 2016), but there is less systematic evidence regarding how beliefs about societal corruption may affect citizens’ readiness to punish it in the electoral arena. In a cross-national study, Klašnja and Tucker (2013) indirectly attribute differences in voters’ attitudes toward corrupt but efficient governments to country levels of widespread corruption. Similarly, Pavão (2018) finds that prompting individuals to think that corruption is very high in their country makes them more tolerant of political corruption.

Putting together these studies, the emerging evidence suggests that the environmental degree of corruption affects citizens’ electoral reactions to new cases of corruption. Voters would overlook corruption accusations when they perceived that the practice of corruption was so widespread that all candidates were likely to be corrupt or that throwing one rascal out would do little to reduce the normalization of corruption in society. The present article, therefore, extends the proposition that prior beliefs about societal levels of corruption matter for tolerance of corrupt acts (Cameron et al., 2009; Pavão, 2018), to the study of how the pervasiveness of corruption in society influences the individual decision to electorally punish corrupt acts of officeholders. This discussion yields the following hypothesis:

Hypothesis 2a. The electoral penalty for corruption is smaller when a voter expects corruption to be prevalent than when a voter expects corruption to be limited.

However, there are several reasons to doubt that widespread corruption should necessarily undermine electoral accountability. The reasoning behind the idea that the normalization of corruption makes citizens indifferent to corrupt politicians is incomplete. First, it is not clear that the acceptance of bribing as a normalized practice should readily translate into tolerance of corrupt politicians. Citizens who expect a large proportion of politicians to be involved in corrupt dealings may place high value on candidates’ integrity credentials. Second, recurrent exposure to political corruption, may make the issue of corruption an important national problem and lead voters to use a politician’s record of corruption to evaluate which politician is in fact able to enact anticorruption policies. Therefore, rather than losing all hope of finding clean candidates, citizens would actively search for and reward them.

Finally, some of the claims about the irrelevance of corruption in settings of widespread corruption seem to be out of sync with contemporary affairs. A recent wave of citizen upheaval and public denunciation of corruption in Latin America indicates that people are in fact willing to speak out against corruption (Avenburg, 2017). Moreover, the emergence of parties using anti-corruption rhetoric in countries with high levels of corruption, especially in southern and eastern Europe (Polk et al., 2017), also casts doubt on the idea that voters in these settings are pardoning of corrupt politicians. Recent electoral results in highly corrupt countries also show that candidates from the right and the left of the ideological spectrum (e.g. Jair Bolsanaro in Brazil or Manuel Lopez Obrador in Mexico) are successful in using the rhetoric of cleaning up politics of corruption.

Hypothesis 2b. The electoral penalty for corruption is not smaller when a voter expects corruption to be prevalent than when a voter expects corruption to be limited.

Experimental Design

This study uses a candidate vignette to randomize the information about not only a hypothetical candidate’s corrupt record but also their competence in delivering public works and their district’s level of corruption. Similar factorial designs have been used in other survey experiments of corruption as a way to test multiple hypotheses by randomly modifying components of a vignette (Hainmueller et al., 2013). The vignette was embedded in a nationally representative survey in Peru and fielded during a 3-week period from 2 November 2015 to 23 November 2015 by a local survey firm, Ipsos-Apoyo.⁷ A sample of 1308 Peruvians was randomly drawn using a stratified two-stage cluster sampling with replacement. The sample was first stratified into 5 regions: Center, North, South, Lima, and Amazon. Then, districts were randomly sampled from within each region with replacement, and from each district, neighborhoods were randomly sampled. Face-to-face interviews were conducted using electronic devices to record the data.

The variation of the three experimental factors of interest of two levels each resulted in eight vignettes. Respondents were randomly exposed to one of these vignettes and then immediately asked to judge the hypothetical politician in the vignette by stating how likely someone like them would cast a vote for this candidate.⁸ Overall, this experimental approach was adopted as an unobtrusive way to measure the socially desirable attitude of rejecting a corrupt candidate.⁹ First, it does not expose a respondent to multiple vignettes that would make them aware of the experimental variations. Second, it presents a vignette with a rather complex hypothetical candidate that has multiple qualities, making it difficult for the respondent to notice the key factors manipulated in the vignette. And, finally, it asks for the attitudes of a third person, allowing the respondent to answer honestly without explicitly stating their preference.¹⁰

Holding constant the type of corruption, this vignette varied whether the hypothetical candidate was accused of wrongdoings or not. In the corruption condition, therefore, respondents were given information that the fictional politician had been criticized for receiving bribes in exchange for public contracts, and in the control condition the hypothetical politician had instead been praised for performing these contracts in an honest and transparent manner. The type of corruption was receiving bribes for public contracts, which, unlike illegal campaign funding or private enrichment, can have positive economic consequences for the public (e.g. a concession for a large infrastructure project could generate jobs and promote economic development). In the competent condition, respondents were exposed to a candidate that had enacted more public works than the majority of mayors, whereas in the incompetent condition the candidate had completed fewer works than the majority of mayors. Notice that all candidates were described as completing some amount of public works for the benefit of their communities, but some candidates were outstanding and others were mediocre. Finally, in the prevalent corruption condition, the former mayor was running for office in a district known for its high levels of corruption, and in the limited corruption condition, this district was instead known for its low levels of corruption.¹¹

The vignette was the following (Figure 1):

Figure 1.

Vignette.

The experimental setting and the wording of the vignette were carefully selected to increase success in eliciting honest responses. The Peruvian case helps maximize the chances of properly testing the effects of perceptions of widespread corruption as they vary greatly within the country, while allowing for realism in the experimentally manipulated contextual conditions. Corruption has become one of the most important concerns of Peruvians, but this concern is nevertheless more acute in certain regions than in others. In some regions, the percent of respondents who think corruption is “very common” among public officials reaches as high as 80%, but in others, corruption is considered “very common” only for 20% of respondents (see Figure A3 in appendix).

To ensure the authenticity of the manipulations, which is a major challenge of such experiments (Druckman et al., 2011), the language of the vignette was adapted to Peruvian politics. By presenting a type of corruption that is frequently seen by Peruvians, such as taking bribes in exchange for public contracts, I make sure that subjects take the stimulus seriously and respond in a meaningful way. Also, by presenting the hypothetical legislative candidate as a former mayor, I make sure that the public works provision treatment is realistic. Moreover, to rule out the possibility that corruption is overlooked due to lack of credibility of information presented in the experimental vignette, I hold the source of the information constant and attribute the accusation to an international anticorruption organization, as a way to guard against less trustworthy national sources (Botero et al., 2015; Weitz-Shapiro and Winters, 2017).¹²

In terms of comparability, this design most approximates Winters and Weitz-Shapiro’s (2013) survey experiment in Brazil, in which they modify candidate competence while holding corruption type constant as bribes in exchange for public contracts. Nevertheless, this study differs from their study in some important regards. First, this study explicitly tests a nonadditive model of corruption voting, in which a candidate’s competence trait influences the way voters punish corruption. While their experiment focuses on the important question of whether the benefits of a competent administration can offset the cost of corruption, this study’s main focus of attention is rather how the effect of corruption on vote varies across candidate competence levels. Moreover, this study disentangles the effect of public works provision from that of voters’ expectations of corruption in government by manipulating the perceived level of prevalence of corruption in the district where the hypothetical candidate is running for office.

Finally, the present experiment is different from other studies of corruption in that it investigates accountability for a legislative candidate, a case of accountability that has been relatively understudied in the recent boom of experimental studies of corruption. Having a subnational authority switching to a legislative office instead of retaining the same office also serves a way to reduce the overreporting of rejection of a corrupt candidate.¹³ Seeking reelection is so disreputable in the Peruvian context that only 17% of district mayors, 10% of provincial mayors, 16% of governors, and 20% of legislators were reelected in the last regional and national elections (Aragón and Incio, 2014). Furthermore, in addition to carrying a distinct electoral disadvantage, reelection for subnational authorities in Peru has recently become an unrealistic political ambition, as term limits were enacted in 2015.

Personalistic politics and party system fragmentation also makes Peru an interesting case for the study of electoral accountability for corruption. In elections for a seat in the national legislature, citizens can vote for up to 2 individual candidates to a seat in the single-chamber National Congress of 130 members. The Peruvian territory is divided into 25 electoral districts with magnitudes ranging from 2 to 36 seats per district. The proportional representation electoral system with open lists translates into several dozens of candidates running for a seat in the legislature in each district (up to 200 candidates in large magnitude districts like Lima). These candidates often run under party labels that convey little meaning in terms of ideological positioning or policy programs. While some traditional political parties have survived in Peru, mass partisanship is one of the lowest in the region, with only around 15%–20% of citizens identifying with any of the numerous political parties that so often come and go from one election to the other.

Results

The main outcome of interest is the respondent’s willingness to cast a vote in support of the hypothetical candidate. I measured this outcome variable using a 1–7 scale of the likelihood that a respondent will opt to vote for the candidate, where 1 is very unlikely and 7 very likely.

Table 1 lists the mean values of support under the “corrupt” and “honest” conditions, the difference in means, and the corresponding standard errors. It shows that when a hypothetical corrupt candidate is contrasted with an honest one, support drops from 3.984 to 2.805, and this difference is statistically significant (p < 0.01). That is, a corrupt record decreases candidate support by 1.178 points on the 1–7 scale.¹⁴ To measure the standardized effect size, I divide the average treatment effect of corruption (ATE) by the average of the two groups’ standard deviations to reveal that the corruption treatment caused a decrease of 0.63 standard deviations. This strong negative effect of corruption suggests that when presented with a hypothetical situation, citizens are willing to take notice and reject a corrupt official. It confirms the intuition that alleged corrupt behavior has a considerable effect on candidate support, a finding that is consistent with the effect of audits found in large-scale field experimental studies (Chong et al., 2015; Ferraz and Finan, 2008; Olken, 2007).

Table 1.

The Effect of Corruption on Likelihood of Support (1–7).

	Combined	Corrupt	Honest	ATE	p-value
Support	3.392	2.805	3.984	1.179	0.00
Standard error	(0.11)	(0.07)	(0.08)

ATE: average treatment effect.

Do public works help candidates in cushioning the electoral impact of a corruption accusation, or do they harm their electoral bids even further? In support of Hypothesis 1a, I find that voters evaluate corruption of competent politicians in a less negative manner from that of incompetent ones. Figure 2 displays the heterogeneous impact of corruption on electoral support for a hypothetical candidate in the two-candidate competence conditions. The penalty for corruption is 0.378 (p < 0.1) points greater for those politicians who fail to deliver public works, as the conditional average treatment effect of corruption goes from −0.996 (p < 0.01) in the competent condition to −1.374 (p < 0.01) in the incompetent condition. Even though the difference of corruption effects under the high and low competence in public works conditions is only significant at the 90% confidence level, this finding provides some evidence in favor of the idea that voters would be more lenient toward corrupt politicians who successfully deliver tangible benefits to their constituencies.¹⁵

Figure 2.

Likelihood of Support by Corruption and Candidate Competence in Public Works Provision.

A substantive interpretation of the significant reduction in punishment for corruption suggests that voters evaluate corruption with a different mind-set when the alleged corrupt candidate is a competent politician. This finding runs against Hypothesis 1b that public works could increase the electoral costs of corruption by signaling greater government involvement in corruption. In contrast to the study in Spain and Sweden (Esaiasson and Muñoz, 2014), in which researchers found that corruption is more costly for competent politicians than for incompetent ones, competence in public works provision mitigates rather than exacerbates the influence of corruption on vote in this Peruvian sample.

Turning our attention to the second hypothesis, are citizens more lenient with an official’s wrongdoing when corruption is normalized? To explore the “they are all corrupt” hypothesis that corruption is penalized less when it is considered a widespread phenomenon, I examine whether the corruption effect is conditional on the level of corruption prevalence. Against Hypothesis 2a, I do not find evidence of this relationship. Figure 3 shows that the electoral punishment is slightly smaller, though not statistically significant, for corrupt candidates under the high prevalence of corruption treatment condition (−1.139 with p < 0.01) than under the low prevalence of corruption condition (−1.221 with p < 0.01). Although electoral penalties are on average 1.139 points lower for candidates who run in a low corruption environment than they are for candidates in a high corruption environment, I cannot reject the possibility that this difference is not statistically different from zero. This finding suggests that, although voters do not overlook corruption when corruption is widespread, they do not apply a corruption penalty of a different magnitude in a context of low prevalence of corruption.

Figure 3.

Likelihood of Support by Corruption Prevalence and Corruption.

To better understand this null result, I explore whether it is possible that the perception of corruption as a widespread phenomenon motivates voters to penalize all candidates, be they corrupt or honest. Would voters take cues from the environment and hold all leaders accountable for corruption observed in society, including an honest candidate who was not accused of any wrongdoing? As expected, I find that upstanding candidates suffer when corruption is widespread. In Table 2, the significant average treatment effect of widespread corruption suggests that all candidates are negatively affected when corruption is perceived to be high, and the conditional average treatment effects indicate that the honest candidates are particularly disfavored by perceptions of widespread corruption. This observation is in line with what Chong et al. (2015) found in Mexico about how corruption information decreased not only incumbent party support but also challenger party support and voter turnout. It also speaks to a broader literature on the demobilizing effects of corruption (Bowler and Karp, 2004; Carreras and Vera, 2018; Chang, 2005; Kostadinova, 2009; Sundström and Stockemer, 2015).

Table 2.

The Effect of Widespread Corruption on Electoral Support (0–100).

	Combined	Honest	Corrupt
Limited corruption	3.50	4.12	2.90
	(0.78)	(0.10)	(0.10)
Prevalent corruption	3.28	3.85	2.71
	(0.79)	(0.11)	(0.11)
Difference	−0.22**	−0.27*	−0.19

Standard errors in parentheses.

p < 0.10, **p < 0.05, *** p < 0.01.

I have presented evidence in favor of Hypothesis 1a suggesting that the cost of corruption is mitigated by the competence of a candidate in delivering public works, but it is possible that the trade-off effect is only present in highly corrupt environments. If the candidate were instead running in a context where corruption is rare, then the shielding effect of competence in delivering economic benefits might turn ineffective. To uncover where the trading effect comes from, I calculate the conditional effect of corruption on candidate competence at different levels of corruption prevalence. Contrary to my expectation, the trading is not present in the high prevalence of corruption group. Table 3 shows the coefficient estimates of electoral support for the respondents in the subgroups of high and low prevalence separately. While competence does mitigate the effect of corruption in the full sample, it does not seem to matter in this way for the subsample of respondents in the high prevalence of corruption treatment. Hence, perceptions of widespread corruption do not appear to be driving the competence-corruption trading strategy.

Table 3.

Trading by Prevalence of Corruption.

Variables	Combined	Full sample		Subsample
		Prevalent corruption	Limited corruption	Prevalent corruption	Limited corruption
Corruption	−1.374***	−1.310***	−1.437***	−1.169***	−1.408***
	(0.148)	(0.214)	(0.204)	(0.330)	(0.246)
Competence	0.423**	0.427**	0.426**	0.867**	0.669**
	(0.148)	(0.212)	(0.206)	(0.334)	(0.249)
Corruption × Competence	0.378*	0.335	0.414	0.204	0.219
	(0.209)	(0.301)	(0.289)	(0.477)	(0.357)
Intercept	3.774***	3.635***	3.912***	3.566***	3.797***
	(0.104)	(0.151)	(0.144)	(0.243)	(0.170)
N	1271	638	633	223	430

Standard errors in parentheses.

p < 0.10, **p < 0.05, *** p < 0.01.

Finally, to take into account the possibility that our contextual corruption treatment was unsuccessful, in an additional analysis I excluded the cases of a mismatch between the treatment and underlying levels of institutional trust. That is, I restricted the analysis to only the most likely cases for a successful manipulation. For the limited treatment condition, I considered only those respondents living in a region of high trust in institutions, and for the prevalence of corruption condition, I included just the cases of low trust in institutions.¹⁶ While this decision might interfere with the randomization, the new indicator of widespread corruption is not fully correlated with other contextual variables in the same way as an entirely observational indicator would. As expected, the conclusions are similar using both a full sample and a restricted sample of most likely cases; the mitigating effect of competence is not limited to highly corrupt environments.

Discussion and Conclusion

In this article, I considered different ways of the political context in which information is disseminated matters for political accountability for corruption. I identified and tested one particular form of trade-off between corruption and economic performance that distinctly embodies the puzzle of voters condoning corrupt politicians: precisely how would candidate competence in public works provision influence electoral support for a corrupt politician? I found that the magnitude of the effect of corruption is conditional upon the candidate’s perceived ability to provide public works in Peru but not in the same direction that a prior study of corruption and public works provision was found (Esaiasson and Muñoz, 2014). In this Peruvian sample, rather than exacerbating the costs of corruption, public works provision mitigates the negative effects of corruption on vote. This outcome is in line with what the “rouba mas faz” literature suggests about the exchange of economic gains for corruption, yet it remains contrary to what the experimental literature examining public works provision had found so far.

Adding further nuance to the discussion, this finding indicates that punishment for corruption is not necessarily negated by candidate competence but that citizens apply a smaller penalty for corruption if a politician has the reputation of an efficient public manager. Therefore, instead of a classical trade-off mentality where voters prefer a corrupt but competent candidate to an honest but incompetent one (Winters and Weitz-Shapiro, 2013), the evidence suggests that voters evaluate corruption of competent politicians in a qualitatively different manner from that of incompetent ones. Far from being fully acceptant of corruption, therefore, voters resist it by taking cues from competence traits to decide how to penalize a corrupt leader. Competent politicians are subject to more lenient standards but are not fully exonerated from an electoral penalty.

This study also relates to a central debate on the role of social norms in the persistence of corruption. Building upon existing evidence, I argued that when voters expect corrupt behavior to be the norm among the political elite, they do not update candidate evaluations after a single accusation of corruption as they might if corruption were atypical.¹⁷ Such a saturation effect is particularly concerning in the face of the increasing efficacy of oversight institutions and news media in scrutinizing and publicizing corruption around the world. Against expectations, however, I found that citizens do not overlook corruption when they view it as inevitable; they instead punish corrupt incumbents just as severely in both high and low corruption environments. This trend is in part due to the fact that voters blame all candidates for a highly corrupt environment, and this responsibility might especially hurt the honest candidates, setting an already low reference point.

Further exploring the role of the context of widespread corruption in electoral accountability for corruption, I also found that the competence-corruption exchange is not particular only to highly corrupt environments. This finding runs against cross-national studies suggesting that voters condone corruption in candidates who create positive economic conditions in highly corrupt countries but do not overlook it in less corrupt countries (Klašnja and Tucker, 2013). Testing this conditionality directly by randomizing corruption across varying levels of corruption prevalence, I found no effect. Furthermore, the replication of this null finding with a subsample of most likely cases suggests that it cannot be fully explained by a possibly unsuccessful treatment. Nevertheless, future studies should try other innovative methods to assess how voters’ expectations matter in political accountability for corruption.

Overall, these results are particularly interesting form a comparative perspective because Peru—a low-middle income country with a long history of electoral impunity—is an unlikely case in which to find that politicians cannot get away with taking bribes, even when corruption is perceived to be the norm rather than the exception. Although the present study also suggests that voters do apply smaller penalties to corruption when the candidate is presented as an efficient public works provider, the evidence that voters are not entirely forgiving sheds light on the micro-foundations of accountability for corruption. Citizens are less permissive and increasingly willing to voice their discontent with corrupt officials.

Nevertheless, these findings should be taken with caution. Before overextrapolating from a single experiment, other studies could explore in greater depth the nuances of corruption voting in high and low corruption environments. One interesting avenue for future research would be to compare these survey experimental results to a behavioral benchmark. Whereas some argue that survey experiments would overestimate corruption effects in comparison to field experimental measures using secret ballots (Boas et al., 2017), others have found that stated preferences in survey experiments match behavioral benchmarks rather well (Hainmueller et al., 2015). Note also that I do not mean to say that corruption effects will always be mitigated by public works provision. First, it is possible that if electoral campaigns were more party-centered, the personal reputation of a candidate who “gets things done” might not serve at all as a partial shield against corruption accusations. In real-world elections, policy issues might become more salient, especially in party-centered elections, and other factors such as media or campaign strategies could also play an important role. Second, the effect of public works provision might be particularly strong in Peru where both subnational authorities as well as national representatives are expected to provide public works to their constituencies, directly or indirectly. Nevertheless, I anticipate these results to be relevant for any developing country where political representatives are seen as agents of constituents’ demands vis-à-vis the central government.

Finally, the results of this study have implications for the challenges developing countries face in breaking the vicious cycle of corruption (Ashworth et al., 2013). While it is generally thought that social norms reinforce the persistence of corruption, these results suggest that when voters are exposed to credible information about corrupt behavior, even in the case of competent candidates, they will punish corruption. This implication calls into question the conventional image of citizens’ high tolerance of corruption, and it points to the need for additional research that develops and tests theories that will elucidate the circumstances in which exposed corruption information can in fact help eradicating electoral impunity. More studies about corruption are urgently needed in contemporary Latin America, where abundant information about government officials’ malfeasance inundates the news and where growing awareness appears to be turning cynicism into resistance. Finally, while electoral accountability is only one of the many mechanisms that can help prevent corruption in government, the changing citizen attitudes toward corruption can play a central role in the development of new and more effective institutional checks on corruption.

Supplemental Material

PSX868210_Supplemental_material – Supplemental material for Accepting or Resisting? Citizen Responses to Corruption Across Varying Levels of Competence and Corruption Prevalence

Supplemental material, PSX868210_Supplemental_material for Accepting or Resisting? Citizen Responses to Corruption Across Varying Levels of Competence and Corruption Prevalence by Sofia B Vera in Political Studies

Footnotes

Funding

The author disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Support for this research was provided by the Center for Latin American Studies and the Center for International Studies at the University of Pittsburgh.

ORCID iD

Sofia B Vera

Supplemental Material

Additional supplemental material may be found with the online version of this article.

Table A1. Summary Statistics.

Table A2. Balance Tests.

Table A3. Corruption Effect on Likelihood of Electoral Support.

Table A4. Corruption Effect on Electoral Support (Ordered Logistic Models).

Table A5. Predicted Values of Support for Competent and Incompetent Candidates, by Limited/Prevalent Condition.

Figure A1. Intra-National Variation of Perceptions of Corruption in Public Office in Peru.

Figure A2. Distribution of Likelihood of Electoral Support by Experimental Groups.

Notes

Author Biography

Sofia B Vera (svera@reed.edu) is an assistant professor of political science at Reed College. Her research interests include political behavior, representation, corruption, political institutions, and democracy. Replication datasets for this study are available upon request.

References

Anduiza

Gallego

Muñoz

(2013) Turning a Blind Eye: Experimental Evidence of Partisan Bias in Attitudes toward Corruption. Comparative Political Studies 46 (12): 1664–1692.

Aragón

Incio

(2014) La reelección de autoridades regionales y municipales en el perú, 2006-2014. Argumentos 8 (5): 16–30.

Arias

Larreguy

Marshall

et al. (2016) Priors Rule: When Do Malfeasance Revelations Help and Hurt Incumbent Parties. Unpublished manuscript.

Ashworth

Bueno

Mesquita

Friedenberg

(2013) Accountability Traps. Available at: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.739.7474&rep=rep1&type=pdf

Avenburg

(2017) Corruption, Accountability and Citizen Participation in Protests in Latin America. Revista Debates 11 (3): 11–28.

Banerjee

Green

et al. (2010) Can Voters be Primed to Choose Better Legislators? Experimental Evidence from Rural India. Working Paper, Presented and the Political Economics Seminar, Stanford University.

Banerjee

Pande

(2009) Parochial Politics: Ethnic Preferences and Politician Corruption. Working Paper, Kennedy School of Government, Harvard University, Cambridge, MA, August.

Barnes

Beaulieu

Saxton

(2017) Sex and Corruption: How Sexism Shapes Voters’ Responses to Scandal. Politics, Groups, and Identities 1–19.

Bauhr

(2017) Need or Greed? Conditions for Collective Action against Corruption. Governance 30 (4): 561–581.

10.

Bauhr

Charron

(2018) Insider or Outsider? Grand Corruption and Electoral Accountability. Comparative Political Studies 51 (4): 415–446.

11.

Boas

Hidalgo

Melo

(2017) Norms versus Action: Voting against Malfeasance in Brazil. Paper Presented at the Latin American Studies Association Conference, Lima, Peru, 22 April.

12.

Bobonis

Fuertes

LRC

Schwabe

(2010) Does Exposing Corrupt Politicians Reduce Corruption? Unpublished paper, University of Toronto.

13.

Botero

Cornejo

Gamboa

et al. (2015) Says Who? An Experiment on Allegations of Corruption and Credibility of Sources. Political Research Quarterly 68 (3): 493–504.

14.

Botero

Cornejo

Gamboa

et al. (2017) Are All Types of Wrongdoing Created Equal in the Eyes of Voters?

15.

Bowler

Karp

(2004) Politicians, Scandals, and Trust in Government. Political Behavior 26 (3): 271–287.

16.

Cameron

Chaudhuri

Erkal

et al. (2009) Propensities to Engage in and Punish Corrupt Behavior: Experimental Evidence from Australia, India, Indonesia and Singapore. Journal of Public Economics 93 (7–8): 843–851.

17.

Carreras

Vera

(2018) Do Corrupt Politicians Mobilize or Demobilize Voters? A Vignette Experiment in Colombia. Latin American Politics and Society 60 (3): 77–95.

18.

Chang

ECC

(2005) Electoral Incentives for Political Corruption Under Open-List Proportional Representation. The Journal of Politics 67 (3): 716–730.

19.

Chang

ECC

Golden

(2007) Electoral Systems, District Magnitude and Corruption. British Journal of Political Science 37 (1): 115–137.

20.

Chong

De La O

Karlan

et al. (2015) Does Corruption Information Inspire the Fight or Quash the Hope? A Field Experiment in Mexico on Voter Turnout, Choice, and Party Identification. The Journal of Politics 77 (1): 55–71.

21.

Corbacho

Gingerich

Oliveros

et al. (2016) Corruption as a Self-Fulfilling Prophecy: Evidence from a Survey Experiment in Costa Rica. American Journal of Political Science 60: 1077–1092.

22.

De Sousa

Moriconi

(2013) Why Voters Do Not Throw the Rascals Out?—A Conceptual Framework for Analysing Electoral Punishment of Corruption. Crime, Law and Social Change 60 (5): 471–502.

23.

De Vries

Solaz

(2017) The Electoral Consequences of Corruption. Annual Review of Political Science 20: 391–408.

24.

Druckman

Green

Kuklinski

et al. (2011) Cambridge Handbook of Experimental Political Science. New York: Cambridge University Press.

25.

Esaiasson

Muñoz

(2014) Roba Pero Hace? An Experimental Test of the Competence-Corruption Tradeoff Hypothesis in Spain and Sweden. Working Paper Series, Quality of Government Institute (QOG), Gothenburg, 2 February.

26.

Fernández-Vázquez

Barberá

Rivero

(2016) Rooting Out Corruption or Rooting For Corruption? The Heterogeneous Electoral Consequences of Scandals. Political Science Research and Methods 4 (2): 379–397.

27.

Ferraz

Finan

(2008) Exposing Corrupt Politicians: The Effects of Brazil’s Publicly Released Audits on Electoral Outcomes. The Quarterly Journal of Economics 123 (2): 703–745.

28.

Gottlieb

(2016) Greater Expectations: A Field Experiment to Improve Accountability in Mali. American Journal of Political Science 60 (1): 143–157.

29.

Hainmueller

Hangartner

Yamamoto

(2015) Validating Vignette and Conjoint Survey Experiments against Real-World Behavior. Proceedings of the National Academy of Sciences of the United States of America 112 (8): 2395–2400.

30.

Hainmueller

Hopkins

Yamamoto

(2013) Causal Inference in Conjoint Analysis: Understanding Multidimensional Choices Via Stated Preference Experiments. Political Analysis 22 (1): 1–30.

31.

Kenny

(2009) Measuring Corruption in Infrastructure: Evidence from Transition and Developing Countries. The Journal of Development Studies 45 (3): 314–332.

32.

Kitschelt

(2000) Linkages between Citizens and Politicians in Democratic Polities. Comparative Political Studies 33 (6–7): 845–879.

33.

Klašnja

Lupu

Tucker

(2017) When Do Voters Sanction Corrupt Politicians? Unpublished manuscript.

34.

Klašnja

Titiunik

(2017) The Incumbency Curse: Weak Parties, Term Limits, and Unfulfilled Accountability. American Political Science Review 111 (1): 129–148.

35.

Klašnja

Tucker

(2013) The Economy, Corruption, and the Vote: Evidence from Experiments in Sweden and Moldova. Electoral Studies 32 (3): 536–543.

36.

Klašnja

Lupu

Tucker

(2017) When Do Voters Sanction Corrupt Politicians? Paper Presented at 2017 Annual Meeting of the European Political Science Association, Milan, 21 December.

37.

Konstantinidis

Xezonakis

(2013) Sources of Tolerance towards Corrupted Politicians in Greece: The Role of Trade Offs and Individual Benefits. Crime, Law and Social Change 60 (5): 549–563.

38.

Kostadinova

(2009) Abstain or Rebel: Corruption Perceptions and Voting in East European Elections. Politics & Policy 37 (4): 691–714.

39.

Kunicová

Rose-Ackerman

(2005) Electoral Rules and Constitutional Structures as Constraints on Corruption. British Journal of Political Science 35 (4): 573–606.

40.

Levistsky

(2013) Peru the Challenge of a Democracy without Parties. In: Domínguez

Shifter

(eds) Constructing Democratic Governance in Latin America, 4th edn. Baltimore, MD: Johns Hopkins University Press, pp.282–315.

41.

Locatelli

Mariani

Sainati

et al. (2017) Corruption in Public Projects and Megaprojects: There Is an Elephant in the Room! International Journal of Project Management 35 (3): 252–268.

42.

Manzetti

Wilson

(2007) Why Do Corrupt Governments Maintain Public Support? Comparative Political Studies 40 (8): 949–970.

43.

Olken

(2007) Monitoring Corruption: Evidence from a Field Experiment in Indonesia. Journal of Political Economy 115 (2): 200–249.

44.

Pavão

(2018) Corruption as the Only Option: The Limits to Electoral Accountability. The Journal of Politics 80 (3): 996–1010.

45.

Polk

Rovny

Bakker

et al. (2017) Explaining the Salience of Anti-Elitism and Reducing Political Corruption for Political Parties in Europe with the 2014 Chapel Hill Expert Survey Data. Research & Politics 4 (1): 1–9.

46.

Quiroz

(2008) Corrupt Circles: A History of Unbound Graft in Peru. Washington, DC: Woodrow Wilson Center Press.

47.

Roberts

(2013) Market Reform, Programmatic (De)alignment, and Party System Stability in Latin America. Comparative Political Studies 46 (11): 1422–1452.

48.

Roh

(2017) The Incumbency Disadvantage in South Korean National Assembly Elections: Evidence from a Regression Discontinuity Approach. Electoral Studies 46 (Suppl. C): 112–122.

49.

Rosas

Manzetti

(2015) Reassessing the Trade-Off Hypothesis: How Misery Drives the Corruption Effect on Presidential Approval. Electoral Studies 39: 26–38.

50.

Rundquist

Strom

Peters

(1977) Corrupt Politicians and Their Electoral Support: Some Experimental Observations. American Political Science Review 71 (3): 954–963.

51.

Schwindt-Bayer

Tavits

(2016) Clarity of Responsibility, Accountability, and Corruption. Cambridge: Cambridge University Press.

52.

Sundström

Stockemer

(2015) Regional Variation in Voter Turnout in Europe: The Impact of Corruption Perceptions. Electoral Studies 40: 158–169.

53.

Tavits

(2007) Clarity of Responsibility and Corruption. American Journal of Political Science 51 (1): 218–229.

54.

Uppal

(2009) The Disadvantaged Incumbents: Estimating Incumbency Effects in Indian State Legislatures. Public Choice 138 (1–2): 9–27.

55.

Vergara

Watanabe

(2016) Peru Since Fujimori. Journal of Democracy 27 (3): 148–157.

56.

Weitz-Shapiro

Winters

(2014) Discerning Corruption: Credible Accusations and the Punishment of Politicians in Brazil. Paper Presented at the APSA 2014 Annual Meeting Paper, Washington, DC, 28–31 August.

57.

Weitz-Shapiro

Winters

(2017) Can Citizens Discern? Information Credibility, Political Sophistication, and the Punishment of Corruption in Brazil. The Journal of Politics 79 (1): 60–74.

58.

Weschle

(2016) Punishing Personal and Electoral Corruption: Experimental Evidence from India. Research & Politics 3 (2): 1–6.

59.

Winters

Weitz-Shapiro

(2013) Lacking Information or Condoning Corruption: When Do Voters Support Corrupt Politicians? Comparative Politics 45 (4): 418–436.

60.

Zechmeister

Zizumbo-Colunga

(2013) The Varying Political Toll of Concerns About Corruption in Good Versus Bad Economic Times. Comparative Political Studies 46 (10): 1190–1218.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.25 MB