Sage Journals: Discover world-class research

Abstract

A growing literature examines the effect of corruption on political behavior. However, little attention has been paid so far to the fact that politicians engage in it for various reasons and with different welfare consequences. In this article, I argue that voters judge corrupt politicians differently depending on what the money is used for. I show results from a survey experiment in India in which respondents heard about a politician who accepted money for a political favor. One treatment group was told that the politician used the money to personally enrich himself (personal corruption), while the other group was informed that he used it to buy votes (electoral corruption). Respondents who received the vote buying treatment were clearly and consistently less likely to agree with a series of potential punishments. This suggests that the overall welfare consequences of corrupt exchanges are an important factor when voters decide how to judge offending politicians.

Many countries in the past years have seen protest movements expressing their anger at the influence of money in politics. This has been accompanied by a surge in research examining how corruption affects political behavior (cf. Klašnja et al., 2016). Studies show that factors such as information, the economy, and partisanship influence the degree to which voters punish corrupt politicians. But lawmakers who engage in corruption do so for various reasons. Take the example of India, where politicians who become rich while holding office are not hard to find. Among the most high-profile cases are the former telecommunications minister A. Raja, who is accused of accepting millions of dollars in bribes, and the Chief Minster of Tamil Nadu, Jayaram Jayalalitha, who was convicted for illegally amassing “disproportionate assets” in excess of $10 million over five years in office.¹

But money from corrupt transactions is also used in other ways: “Parties and candidates tend to use their term of office to accumulate war chests for future elections and for nursing their constituencies. There is evidence that they raise these resources through any means available, including through corrupt means such as kickbacks for regulatory and allocative favors while in office.” (Gowda and Sridharan, 2012: 236). In Indian elections, candidates often try to buy votes by distributing cash, liquor, and small gifts (Kapur and Vaishnav, 2011; Krishna, 2007; Wilkinson, 2007). This has certain welfare consequences: an article in Bloomberg View calls Indian general elections a “quasi-Keynesian boost (…) rich with multiplier effects from boardroom to tea shop.”²

Do voters judge politicians who engage in corruption differently depending on what they use the money for? In this article, I present evidence that this is indeed the case. I show results from a survey experiment conducted on a representative sample of voters in Delhi, India. Respondents were asked about a politician who accepted money for a political favor. One treatment group was told that the politician used this money to personally enrich himself (personal corruption), while the other was informed that he used it to buy votes (electoral corruption). Respondents were then asked how much they supported different punishments the politician could face. Those who received the vote buying treatment were clearly and consistently less likely to agree with a series of potential punishments. The welfare consequences of corrupt exchanges thus are an important factor when voters decide how to evaluate offending politicians. This can help explain the central puzzle of the literature, namely that corrupt incumbents are often treated with leniency by voters (Golden, 2010; Golden and Mahdavi, 2015). I also show that the differences between the two treatments are larger among wealthier respondents, even though they are less likely to benefit from vote buying efforts. This suggests that the effect is not necessarily driven by narrow self-interest, but instead by a sociotropic evaluation of the overall welfare consequences of personal and electoral corruption among parts of the electorate.

This article contributes to the literature on the effect of corruption on political behavior, which so far has mainly focused on how environmental factors condition voters’ responses to corrupt politicians. For example, studies demonstrate the impact of information (Chong et al., 2015; Ferraz and Finan, 2008; Winters and Weitz-Shapiro, 2013; Chang et al., 2010), the state of the economy (Klašnja and Tucker, 2013; McCann and Domínguez, 1998; Zechmeister and Zizumbo-Colunga, 2013), political partisanship (Anderson and Tverdova, 2003; Anduiza et al., 2013; Eggers, 2014), clarity of responsibility (Tavits, 2007), and the level at which corruption is experienced (Gingerich, 2009; Klašnja et al., 2016). I complement the literature by drawing attention to the fact that the welfare consequences of corrupt exchanges matter as well.³ The article therefore is also part of a recently emerging literature that disaggregates corruption and special interest money (e.g. Bussell, 2013; Gingerich, 2014; Kaufmann and Vicente, 2011; Nyblade and Reed, 2008).

The paper relates most closely to two recent observational studies. Manzetti and Wilson (2007) find that support for governments perceived as corrupt among World Values Survey respondents is negatively related to the strength of government institutions. This is interpreted as evidence that voters in countries with strong patron–client relationships are more likely to support corrupt politicians, since they expect to receive tangible benefits. However, countries with low institutional strength differ from those with high strength in many other ways than their level of clientelism. More directly, Fernández-Vázquez et al. (2016) examine the reelection probabilities of Spanish mayors involved in corruption scandals. They show that only representatives whose corrupt dealings had no welfare enhancing consequences (e.g. embezzlement) saw a decrease in their reelection probabilities. If there were positive side benefits for voters (e.g. jobs through licensing illegal construction projects), there was no punishment. One worry with such an observational study is that politicians may select the type of corruption they engage in strategically, for example based on characteristics of the electorate.⁴ I contribute to this line of inquiry by providing direct experimental evidence of the micro-level mechanisms behind these findings, net of any potential confounders.

The article also speaks to contributions showing that voters tend to tolerate corruption when the offending politician delivers good results otherwise (Klašnja and Tucker, 2013; Muñoz et al., 2012; Rundquist et al., 1977).⁵ In contrast to these “implicit exchange” accounts, I provide evidence for a more explicit exchange that involves the usage of money obtained from a corrupt transaction itself. Finally, my findings are relevant for the literature on clientelism, and in particular for the debate about its welfare consequences. Clientelistic competition is not considered a normative ideal and has many negative effects (e.g. Magaloni, 2006; Stokes, 2005). However, in countries where the state provides few public goods, these payments are one of the only tangible benefits that especially poor citizens get from the political system (Auyero, 1999; Gonzalez Ocantos et al., 2014). I show evidence that voters do indeed recognize that clientelism has certain welfare-enhancing effects.

Survey experiment design

In this article, I focus on India, a country which currently ranks 85th out of 174 countries in Transparency International’s Corruption Perception Index and 130th out of 189 in the World Bank’s ease of doing business ranking.⁶ It is widely acknowledged that political corruption is very common (Bussell, 2012; Gowda and Sridharan, 2012; Kapur and Vaishnav, 2011). Furthermore, as the examples above make clear, it happens for several reasons. Many politicians use their tenure in office to enrich themselves (Fisman et al., 2014). But election campaigns are increasingly costly, with clientelistic expenditures making up a large part of the tab. Candidates need to invest considerable resources to be serious contenders (Gowda and Sridharan, 2012; Kapur and Vaishnav, 2011). India is thus an ideal setting to investigate the impact of corrupt money for personal enrichment versus vote buying.

The survey experiment for this study was conducted as part of a larger survey in the National Capital Territory of Delhi in January 2014. Respondents were selected from the voter roll in a multi-step process. First, 10 of 70 assembly constituencies in Delhi were chosen randomly. Second, five polling stations from each constituency were sampled.⁷ Finally, for each polling station, 20 voters were selected from the official electoral roll published by the Chief Electoral Officer of Delhi.⁸ They were assigned to one of two treatments in alternating order.

Personal Enrichment Treatment: Imagine a politician who received money from a company for a political favor. He used this money to personally enrich himself. What do you think the consequences should be?

Vote Buying Treatment: Imagine a politician who received money from a company for a political favor. He used this money to buy votes in an election. What do you think the consequences should be?

Both prompts are formulated in a neutral manner. In particular, the vote buying one does not stress who the recipients are or what benefits they receive. Instead, it simply refers to vote buying, which is illegal under Indian law, just like personal enrichment in exchange for a political favor. Note that I use a fictional rather than a real politician. While this comes with the downside of greater abstraction, previous research shows that pre-existing attitudes such as partisanship affect how much punishment voters exert for the same corruption offense. This is likely also true for ethnicity or, in the Indian context, caste. To abstract from these specifics, experimental studies of corruption typically ask respondents to evaluate fictional politicians, and I follow this convention.

Also note that the experiment does not have a control condition asking about a corrupt politician without giving any information about what the money was used for. The experiment thus cannot tell us whether the personal enrichment and vote buying treatments differ from a generic corruption allegation. But as the examples mentioned above indicate, corruption cases are rarely generic. They are usually discovered based on irregularities resulting from how the illegal money is used, so there are reports about politicians who increase their personal assets while in office, or who call in favors to engage in vote buying. In addition to not corresponding to many real-world situations, a generic corruption allegation would also introduce the amount of information a respondent receives as a second condition into the experiment. Because the goal is to tease out the effect of personal versus electoral corruption, I focus on comparing two conditions that hold constant the amount of information given.

After receiving the treatment statements, participants were presented with three punishments.

A politician who took money for political favors to [buy votes/enrich himself] should have to resign from his position.

A politician who took money for political favors to [buy votes/enrich himself] should be banned from contesting future elections.

A politician who took money for political favors to [buy votes/enrich himself] should be sentenced to time in jail.

For each statement, respondents were asked whether they agreed completely, agreed, neither agreed nor disagreed, disagreed, or disagreed completely. In addition to this experimental treatment, the survey asked about a number of political issues and included a standard battery of demographic questions. Interviews were conducted face-to-face in Hindi by trained interviewers of a Delhi-based polling company. If a sampled voter was unavailable, a replacement in the same locality and of the same gender and age group was selected. In total, 993 interviews were completed. The sample is balanced for most variables, with two small exceptions.⁹ To ensure that they do not drive the results, I account for them in various ways below.

Finally, it is important to note the context the survey was conducted in. In December 2013, the month before the survey, an anti-corruption party called the Aam Aadmi Party successfully competed in state elections for the Delhi assembly. The party gained almost 30 percent of the votes. It subsequently formed a minority government, which lasted from December 2013 to February 2014. This means that the survey experiment was conducted while an anti-corruption party was in power. Corruption therefore probably had a higher than average salience among its respondents. This makes it likely that they condemn corrupt practices more than would be the case in another place or at another time. But since the focus here is not on how much voters disapprove of corrupt politicians, but instead on whether their disapproval differs depending on what the money was used for, this should not limit our ability to draw inference.

Analysis

As an initial step, I treat the respondents’ answers to the three questions as continuous going from one (agree completely) to five (disagree completely). The first thing to note is that tolerance for politicians who accept special interest money for political favors is low, no matter how it is used. Most respondents agree or agree completely that a politician who engages in such behavior should be punished. The means in the entire sample are 1.43 (the politician should resign), 1.54 (the politician should be banned from contesting future elections), and 1.48 (the politician should be jailed). This was expected given the location and timing of the survey during the administration of an anti-corruption party.

Table 1 shows the means of the two treatment groups for the three response variables. Despite the high overall disapproval of corruption, the means are consistently smaller for respondents who were given the personal enrichment treatment compared to those who received the vote buying treatment (a lower number means more support for punishment). On the scale from one to five, respondents who received the personal enrichment treatment are on average between 0.124 (jail) and 0.247 (resign) points more supportive of punitive measures. These differences are significant at conventional levels, both when using uncorrected p-values as well as when accounting for multiple hypothesis testing using Benjamini–Hochberg corrections. Put the other way around, people are more tolerant of politicians accepting money in exchange for political favors when it is distributed to the population in an attempt to buy their support.

Table 1.

Differences in means for the three dependent variables between the two treatment groups.

Variable	Vote buying	Enrichment	Difference	Std. Err.	p-value	B-H p-value
Resign	1.555	1.308	0.247	0.058	0.000	0.000
Ban	1.613	1.460	0.153	0.061	0.006	0.016
Jail	1.541	1.417	0.124	0.060	0.025	0.046

B-H: Benjamini-Hochberg.

Of course, it is not clear that the response variables can be treated as continuous. Figure 1 shows the share of respondents who agree completely with the punishment statements for the personal enrichment treatment (gray) and the vote buying treatment (white). Again, it is apparent that there is widespread disapproval of the politician’s behavior. However, the share of respondents agreeing completely that he should resign, be banned, and be jailed is consistently larger for those who received the personal corruption treatment. These differences are statistically significant at conventional levels (see Table 2 in the Online Appendix).

Figure 1.

Share of respondents who agree completely with the punishment statements for the personal enrichment treatment (gray) and the vote buying treatment (white).

Because there are slight imbalances between the two treatment groups, I also estimate a series of regressions that control for demographic variables. Table 2 shows the coefficient of the vote buying treatment from Ordinary Least Squares (OLS) specifications with controls, constituency fixed effects, and standard errors clustered by polling station. The coefficients can be interpreted as the differences in means between the two treatments, controlling for demographic and constituency effects. Their magnitudes are actually larger than when looking at the raw differences, ranging between 0.178 (jail) and 0.285 (resign). In the Online Appendix, I present a series of additional model specifications as robustness checks. I also account for the imbalances in the treatment assignment groups by pre-processing the data using matching. In all cases, and for each dependent variable, the coefficient of the vote buying treatment is positive and significant at conventional levels.

Table 2.

Coefficient of the vote buying treatment for the three dependent variables, based on OLS specifications that control for demographic covariates, with assembly fixed effects and robust standard errors clustered by polling station. See Online Appendix for full table of results.

Variable	Coefficient	Sdt. Err.	p-value	B-H p-value
Resign	0.285	0.063	0.000	0.000
Ban	0.201	0.062	0.002	0.006
Jail	0.178	0.060	0.004	0.007

The fact that voters are more lenient with vote buying politicians begs the question what the reason behind this finding is. Is it driven by the egocentric self-interest of voters who usually are among the beneficiaries of clientelistic politics? Or is it because voters think about the sociotropic welfare consequences for society as a whole? Table 3 shows the effect of the two treatments separately for respondents in the lower and upper income halves. Because vote buying efforts primarily target the poor (e.g. Stokes, 2005; Weitz-Shapiro, 2012), respondents in the upper income half are less likely to benefit personally from it. Nevertheless, they are consistently more lenient with the vote buying candidate. Respondents in the lower income half, however, are only significantly less punitive for the vote buying politicians when it comes to the mildest form of punishment (resignation). For the other two punishments, the differences between the treatments are smaller and not significant. This suggests that the effect is not necessarily driven by narrow self-interest. Instead, it mainly comes from wealthier (and more educated) respondents differentiating the overall societal welfare consequences of personal and electoral corruption.¹⁰

Table 3.

Coefficient of the vote buying treatment for the three dependent variables separately for each income half, based on OLS specifications that control for demographic covariates, with assembly fixed effects and robust standard errors clustered by polling station. See Online Appendix for full table of results.

Variable	Income half	Coefficient	Std. Err.	p-value	B-H p-value
Resign	Lower	0.240	0.084	0.006	0.022
	Upper	0.331	0.074	0.000	0.000
Ban	Lower	0.113	0.061	0.070	0.206
	Upper	0.297	0.096	0.003	0.020
Jail	Lower	0.069	0.068	0.320	0.784
	Upper	0.297	0.098	0.004	0.020

Conclusions

One of the main advances in the study of corruption in recent years has been the disaggregation of the catch-all concept. Scholars now distinguish between bureaucratic and political corruption, petty corruption and grand corruption, personal and electoral corruption, and so on. Not only are these conceptually distinct (e.g. Bussell, 2013; Gingerich, 2014; Warren, 2004), they also have different welfare consequences. The survey experiment presented here contributes to this line of inquiry. It shows that voters are less willing to punish politicians engaging in a corrupt exchange if its proceeds are then “shared” with the wider electorate. This can help explain the central puzzle that motivates the literature, namely why voters frequently reelect incumbents they know to be corrupt (Golden, 2010; Golden and Mahdavi, 2015).

In future research, it will be necessary to replicate the survey experiment in different contexts to determine its empirical scope. Are voters generally more tolerant of vote buying, or are there situations where they are not? How does the magnitude of the effect differ, especially between different income levels? More broadly, (illegally) providing favors in exchange for money used to buy votes or for personal enrichment are only two ways in which special interest money enters politics. Others include (legal) campaign contributions, sponsored retreats and other travel, or lucrative post-government employment. What are their welfare consequences, and how do voters evaluate politicians engaged in them? Answering these questions will further advance the growing literature on different types of corruption as well as the study of its effects on political behavior. Ultimately, this will help us understand when voters punish corrupt politicians, and when they give them a free pass instead.

Footnotes

For their helpful comments and suggestions,I am grateful to Lorena Becerra,Pablo Beramendi,Pablo Fernández-Vázquez,Daniel Gingerich,Karen Remmer,Michael Ward,Erik Wibbels,as well as the editors and anonymous reviewers at Research and Politics . I thank Dhananjai Joshi and his team at Cicero Associates for their professional work in conducting the survey.

Declaration of conflicting interests

The author declared no potential conflicts of interest with respect to the research,authorship,and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research,authorship,and/or publication of this article: The project was supported financially by the Department of Political Science at Duke University.

Supplementary material

The online appendix is available at:

Carnegie Corporation of New York Grant

The open access article processing charge (APC) for this article was waived due to a grant awarded to Research & Politics from Carnegie Corporation of New York under its ‘Bridging the Gap’ initiative.

References

Anderson

Tverdova

(2003) Corruption, political allegiances, and attitudes toward government in contemporary democracies. American Journal of Political Science 47(1): 91–109.

Anduiza

Gallego

Muñoz

(2013) Turning a blind eye: Experimental evidence of partisan bias in attitudes toward corruption. Comparative Political Studies 46(12): 1664–1692.

Auyero

(1999) From the client’s point(s) of view: How poor people perceive and evaluate political clientelism. Theory and Society 28(2): 297–334.

Bussell

(2012) Corruption and Reform in India. Public Services in the Digital Age. Cambridge: Cambridge University Press.

Bussell

(2013) Varieties of corruption: The organization of rent-seeking in India. Unpublished Manuscript.

Chang

ECC

Golden

Hill

(2010) Legislative malfeasance and political accountability. World Politics 62(2): 177–220.

Chong

de la

O A

Karlan

Wantchekon

(2015) Does corruption information inspire the fight or quash the hope? A field experiment in mexico on voter turnout, choice, and party identification. Journal of Politics 77(1): 55–71.

Doherty

Dowling

Miller

(2011) Are financial or moral scandals worse? It depends. PS: Political Science & Politics 44(4): 749–757.

Eggers

(2014) Partisanship and electoral accountability: Evidence from the UK expenses scandal. Quarterly Journal of Political Science 9(4): 441–472.

10.

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.

11.

Ferraz

Finan

(2008) Exposing corrupt politicians: The effects of Brazil’s publicly released audits on electoral outcomes. Quarterly Journal of Economics 123(2): 703–745.

12.

Fisman

Schulz

Vig

(2014) The private returns to public office. Journal of Political Economy 122(4): 806–862.

13.

Gingerich

(2009) Corruption and political decay: Evidence from Bolivia. Quarterly Journal of Political Science 4(1): 1–34.

14.

Gingerich

(2014) Yesterday’s heroes, today’s villains: Ideology, corruption, and democratic performance. Journal of Theoretical Politics 26(2): 249–282.

15.

Golden

(2010) Puzzles of political corruption in modern advanced democracies. In: Magara

(ed) Democracy and Accountability: Globalized Political Responsibility. Tokyo: Fukosha.

16.

Golden

Mahdavi

(2015) The institutional components of political corruption. In: Gandhi

Ruiz-Rufino

(eds) Routledge Handbook of Comparative Political Institutions. New York: Routledge Press.

17.

Gonzalez Ocantos

Kiwiet de Jonge

Nickerson

(2014) The conditionality of vote-buying norms: Experimental evidence from Latin America. American Journal of Political Science 58(1): 197–211.

18.

Gowda

MVR

Sridharan

(2012) Reforming India’s party financing and election expenditure laws. Election Law Journal 11(2): 226–240.

19.

Kapur

Vaishnav

(2011) Quid pro quo: Builders, politicians, and election finance in India. Unpublished Manuscript.

20.

Kaufmann

Vicente

(2011) Legal corruption. Economics & Politics 23(2): 195–219.

21.

Klašnja

Tucker

(2013) The economy, corruption, and the vote: Evidence from experiments in Sweden and Moldova. Electoral Studies 32: 536–543.

22.

Klašnja

Tucker

Deegan-Krause

(2016) Pocketbook vs. sociotropic corruption voting. British Journal of Political Science 46(1): 67–94.

23.

Krishna

(2007) Politics in the middle: Mediating relationships between the citizens and the state in rural North India. In: Kitschelt

Wilkinson

(eds) Patrons, Clients, and Policies. Patterns of Democratic Accountability and Political Competition. Cambridge: Cambridge University Press.

24.

Magaloni

(2006) Voting for Autocracy. New York: Cambridge University Press.

25.

Manzetti

Wilson

(2007) Why do corrupt governments maintain public support? Comparative Political Studies 40(8): 949–970.

26.

McCann

Domínguez

(1998) Mexicans react to electoral fraud and political corruption: An assessment of public opinion and voting behavior. Electoral Studies 17(4): 483–503.

27.

Muñoz

Anduiza

Gallego

(2012) Why do voters forgive corrupt politicians? Cynicism, noise and implicit exchange. Unpublished Manuscript.

28.

Nyblade

Reed

(2008) Who cheats? Who loots? Political competition and corruption in Japan, 1947–1993. American Journal of Political Science 52(4): 926–941.

29.

Rundquist

Strom

Peters

(1977) Corrupt politicians and their electoral support: Some experimental observations. American Political Science Review 71: 954–963.

30.

Stokes

(2005) Perverse accountability: A formal model of machine politics with evidence from Argentina. American Political Science Review 99(3): 315–325.

31.

Tavits

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

32.

Warren

(2004) What does corruption mean in a democracy? American Journal of Political Science 48(2): 328–343.

33.

Weitz-Shapiro

(2012) What wins votes: Why some politicians opt out of clientelism. American Journal of Political Science 56(3): 568–583.

34.

Wilkinson

(2007) Explaining changing patterns of party-voter linkages in India. In: Kitschelt

Wilkinson

(eds) Patrons, Clients, and Policies. Patterns of Democratic Accountability and Political Competition. Cambridge: Cambridge University Press.

35.

Winters

Weitz-Shapiro

(2013) Lacking information or condoning corruption: When do voters support corrupt politicians? Comparative Politics 45(4): 418–436.

36.

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.

Punishing personal and electoral corruption: Experimental evidence from India

Abstract

Survey experiment design

Analysis

Conclusions

Footnotes

Declaration of conflicting interests

Funding

Supplementary material

Carnegie Corporation of New York Grant

References