The Electoral Consequences of Corruption

Catherine E. De Vries and Hector Solaz

Moreover, corruption is said to erode political trust and undermine political legitimacy in a
variety of institutional settings (Della Porta 2000, Seligson 2002, Andersen & Tverdova 2003,
Chang & Chu 2006).

For instance, majoritarian electoral systems are ´ associated with lower levels of corruption than
proportional ones, because they foster alternation in government and promote political
competition (Persson et al. 2003; see also Rudolph & Daubler ¨ 2016). The evidence regarding
proportional representation suggests that systems using closed lists are associated with more
corruption than those employing open lists (Carey & Shugart 1995, Kunicova & Rose-
Ackerman 2005). Finally, district magnitude might also play an important ´ role, especially
within proportional systems. Under open-list proportional systems, for example, corruption
increases with district magnitude (Chang & Golden 2007).

A core assumption underlying the research on the effects of electoral rules on corruption is that
voters punish corrupt politicians if they have the opportunity to do so. Yet, the empirical
evidence on the electoral punishment of corruption is mixed. The reelection of corrupt
politicians is not merely a trait of developing nations characterized by weak political and
economic institutions but is also found in established democracies such as Italy, Japan, and the
United States (e.g., Rundquist et al. 1977, Reed 1996, Chang et al. 2010). This raises the
question of why voters often fail to punish politicians for corrupt activities. At the contextual
level, researchers suggest that the presence of ´ strong economic growth (e.g., Klasnja & Tucker
2013, Zechmeister & Zizumbo-Colunga 2013) ˇ or a lack of institutional clarity (e.g., Tavits
2007, Schwindt-Bayer & Tavits 2016) might weaken the electoral punishment of politicians
who are known to be corrupt. Most research to date suggests that voters often do not hold
politicians accountable for corruption and thus deals with the breakdown of one of these steps,
although not explicitly.

People’s corruption perceptions may also vary extensively within one country. Some studies
have found that the wealthy and educated perceive more corruption; others find that the wealthy
and educated perceive less corruption (Davis et al. 2004, Redlawsk & McCann 2005, Tverdova
2011). Maeda & Ziegfeld (2015) show that these conflicting findings are resolved when
economic development is taken into consideration. Whereas in richer countries the poor and
uneducated tend to perceive higher levels of corruption, in poorer countries they perceive
corruption to be lower. The reported within-country heterogeneity may lead to a situation where
some voters in a specific election punish corruption while others tolerate it, and at the aggregate
level these competing responses might cancel out.

Voters punish corrupt politicians if they have the opportunity to do so. This idea is based
on the classical interpretation of the retrospective voting model as a means to either sanction
low-performing politicians or select high-performing ones (e.g., Key 1966, Fearon 1999). What
is more, knowing that voters disapprove of corruption, candidates have an incentive to actively
mislead the public by accusing competitors of corruption, even if this information is false.
Voters might become aware of this and perceive most corruption information not to be credible.
Furthermore, parties can provide explanations about corruption allegations or force individual
members to resign and take the blame.

Some studies demonstrate that corrupt activities take a considerable electoral toll on incumbent
politicians (see, e.g., Fackler & Lin 1995, Ferraz & Finan 2008, Krause & Mendez 2009,
Winters & Weitz-Shapiro 2013, Klasnja 2016). Krause & M ˇ endez (2009, p. 179), for
example, suggest ´ that “corruption in public office is effectively punished by voters.”

Other work suggests that the electoral retribution of corrupt behaviour in office is rather mild
(for an excellent overview, see Golden 2010). For example, studies on Italy (Chang et al. 2010),
the United Kingdom (Vivyan et al. 2012), Spain (Rivero & Fernandez-Vázquez 2011,
Fernández- ´ Vazquez et al. 2016), Japan (Reed 1996), and the United States (Rundquist et al.
1977, Peters ´ & Welch 1980) suggest that politicians implicated in corruption are often
reelected. What is more, recent work demonstrates the effect of specific individual or contextual
factors—such as information asymmetries (e.g., Chang et al. 2010, Botero et al. 2015),
partisanship (e.g., Anduiza et al. 2013, Munoz et al. 2016), or the state of the economy (Kla ˜
snja & Tucker 2013, Zechmeister ˇ & Zizumbo-Colunga 2013)—to explain the conditions that
lead to a lack of electoral punishment of corruption. The idea that voters punish a politician they
know to be corrupt is thus becoming increasingly contested.

Even if information is sufficient and of high quality, and blame is attributed correctly, voters
may not punish corruption. The third stage involves behavioral responses. Voters must consider
a large array of factors when deciding to cast their ballot, and corruption may not be a salient
issue in a specific election campaign (Chang et al. 2010, Klasnja et al. 2016); even if voters find
ˇ corruption important, they might be willing to trade it off against some other benefit.

Some studies suggest that the lack of electoral punishment of corruption is mostly a function of
the quantity of information available to voters (e.g., Ferraz & Finan 2008, Chang et al. 2010,
Costas-Perez et al. 2012). Information dissemination could be limited by a lack of media
reporting, ´ political control of the media, high levels of illiteracy, or a lack of prosecution of
corruption cases, for example. In a study of legislators in Italy’s lower ´ house between 1948
and 1994 and of candidates of the two largest parties (Christian Democracy and the Italian
Communist Party), Chang et al. (2010) show that only in the 1992–1994 legislature did corrupt
legislators or candidates face a serious electoral penalty. The authors suggest that this structural
break in the response to corruption by Italian voters was due to the media coverage that grew
out of the Clean Hands operation in the early 1990s.

Blame attribution is made difficult by the fact that people do not process information
objectively. Psychologists have demonstrated that attributions are affected by group-serving
biases (Taylor & Doria 1981). People tend to make internal attributions (to their own in-group)
for positive events or outcomes and make external attributions (to an out-group) for negative
events or outcomes. Retrospective voting studies, largely inspired by work in social psychology,
suggest that voters have strong group-serving biases when it comes to responsibility judgments
(Conover et al. 1987, Bartels 2002, Rudolph 2003, Evans & Andersen 2006, Evans & Pickup
2010).

Erroneous blame attribution may be the result not only of group-serving biases, but also of
institutional complexity. For example, direct experience with corruption might often involve
petty bureaucracy, such as being asked for a bribe by a policeman, and people may not always
be able to attribute this behaviour to the elected public officials who enable it (Abramo 2008,
Klasnja et al. ˇ 2016).

When voters are sufficiently informed about corruption and assign causal responsibility
correctly, they have several behavioural options: switching, abstaining, or sticking to their
previous vote. The viability of alternatives is a core factor that conditions voters’ ability to
switch party choice. Only when other parties perceived as credible and clean alternatives exist
can voters punish corruption by switching. Schleiter & Voznaya (2016) argue that party system
competitiveness plays a critical role in conditioning the electoral punishment of corruption.
When a viable alternative is lacking, voters might react to corruption by abstaining. The
literature on the relation between corruption and turnout is relatively small but provides a rather
consistent result: Corruption is negatively associated with turnout.

Voters might be also willing to turn a blind eye to corruption based on an implicit trade. It may
well be the case that poor performance in one area, in this case corruption, is compensated for
performance in some other area, such as economic growth (Klasnja & Tucker 2013,
Zechmeister ˇ & Zizumbo-Colunga 2013). Moreover, voters might care about other attributes of
candidates, such as ideological proximity (Rundquist et al. 1977), partisanship or some other
shared-group considerations (Solaz et al. 2017). By presenting experimental evidence from
Moldova, Klasnja & Tucker (2013) suggest that in high-corruption countries, ˇ corruption is
punished only in dire economic conditions.

Voters may be willing to overlook corruption if they receive representational benefits.

Rundquist et al. (1977) argue that voters might prefer a corrupt candidate who represents their
policy and issue positions to a clean candidate who does not.
Not only partisanship or ideology but in-group considerations more generally may lower the
electoral punishment of corruption. Banerjee & Pande (2007) present a theoretical model to
suggest that as a society becomes more ethnically polarized, and citizens vote following ethnic
identity rather than any other marker, corrupt activities of candidates of the same ethnic group
are less likely to be punished electorally. This is because voters “feel an instinctive pull towards
their coethnics” (Banerjee & Pande 2007, p. 5).

Solaz et al. (2017) provide empirical evidence for this type of in-group intuition. By utilizing
the fact that a large-scale corruption scandal involving the ruling party fell in the fieldwork
period of the European Social Survey, they are able to demonstrate that in-group loyalty
conditioned the electoral punishment of Spanish voters. They show that priming group identities
of both voters and candidates reduces the electoral punishment of corruption even when clean
alternatives exist and voters are fully informed about and receive no benefits from corruption.
The evidence presented by Solaz et al (2017) is important as it suggests that even when people
know their own leaders to be corrupt and accurately assign blame, they may still fail to sanction
because they place greater weight on other benefits they will receive from a candidate, namely
in-group status, than they do on that individual’s malfeasance.

If political polarization occurs among ethnic groups, corruption is less likely to be punished as
voters “feel an instinctive pull towards their coethnics” (Banerjee & Pande 2007, p. 5). The by-
product of this might be a selection effect. Parties alter their candidate choices in line with
ethnic polarization as the probability of winning an election increases with ethnic similarity
rather than quality, and thus the quality threshold for candidates is lowered. Following this
intuition, Solaz & De Vries (2017) present laboratory experimental evidence to show that a lack
of electoral punishment due to in-group loyalties indeed increases corruption.

Conventional wisdom suggests that democratic elections through which citizens can
periodically sanction or reward politicians have a constraining effect on corruption, because
voters are expected to punish corrupt activities. Yet, we often observe empirical deviations from
this expectation as voters keep reelecting politicians who steal from them. We have suggested
that punishing politicians for corrupt activities at the ballot box is indeed more difficult than
often assumed.
Populism and Corruption (Transparency International Anti-Corruption
Helpdesk Answer)

Niklas Kossow

Populism is not a new phenomenon. It has played a role in politics for generations and is often
associated with the rise of authoritarian leaders (De La Torre, Wejnert and Woods 2014). In
many of these countries, corruption and populism feed one another. On the one hand, populist
leaders use corruption to rally support for their political agenda. In this context, corruption
serves as a justification for populist politics. On the other hand, a populist rhetoric is then used
by the leaders to mask corrupt acts perpetrated by them or their collaborators, thus weakening
the effectiveness of anti-corruption policies. In many cases, populism also uses corruption as a
pretext to dismantle democratic institutions and propagate authoritarian policies (Kaltwasser
and Mudde 2012; Curini 2017).

Prominently among these is the Dutch political scientist Cas Mudde (2004). He defines
populism as a political ideology dividing society into two groups: the people on one side, and
the elite on the other. Populists claim to represent the will of the “pure people” against the
“corrupt elites” (Mudde and Kaltwasser 2017) and base political legitimacy on this
representation. He defines populism as the negation of pluralism, since populist leaders will
claim that it is they alone who represent the will of the people, as opposed to the existing elites.
By doing so, populist leaders can actively undermine democracy as, in their view, it is only
them who can legitimately represent the people. This negates one of the key premises of
democracy, that majorities can change and that interests of political minorities need to be
protected. Populism thus stands in direct opposition to pluralism.
Populism uses several sentiments in this context. Nationalism plays in an important part, as it is
used to delineate what Mudde (2004) refers to as the “pure people”. Populists thus often toy
with racism and xenophobia, suggesting they are needed to defend the people from outsiders
and that current elites are looking to undermine the purity of the people.

The focus on the political establishment in the capital can also be seen in the rhetoric used by
Philippine President Rodrigo Duterte who, as a former mayor in the southern Philippines,
campaigned on a promise to end the political domination of Manila, the country’s capital
(Balisacan 2017). In office, Duterte ran a campaign to crack down on several corrupt officials
and fired many of them. Yet, the success of this campaign remains questionable, and some of
the fired officials were re-hired, thus suggesting that the campaign was less serious and more
driven by the appearance of fighting corruption (Calonzo 2018).

A notable case is Viktor Orbán, who has ruled Hungary as prime minister since 2010. When he
came to power, the fight against the corrupt system represented by the socialist-liberal
governments that ruled between 2002 and 2010 was a policy priority. anley and Sikk (2013)
also find that high or rising corruption is a significant factor in explaining the success of “anti-
establishment” parties in Eastern Europe. Orbán took power after a series of governments had
benefitted from widespread corruption. Yet, by 2011, only one year after Orbán assumed the
office of prime minister, several observers noted a situation in Hungary they described as state
capture by elites surrounding Orbán and his party (Mong and Jancsics 2016). This does not only
include Hungary but the Czech Republic and Slovenia, which showed similar patterns in
elections in recent years, where corruption as an issue helped to get votes for antiestablishment
reform parties (Hanley and Sikk 2016).

The populist message of a corrupt elite responsible for a system that harms the people falls then
on fertile ground when corruption is perceived to be a problem in the country. This makes it
easier to sell the message that many elected politicians do indeed work against the best interests
of the people. What is questionable, however, is how effective populist leaders are in actually
fighting corruption.

Viktor Orbán and his populist rhetoric and politics did not only fail to fight systemic corruption
in Hungary but changed the logic of corruption, resulting in state capture through a narrow elite.
The slow dismantling of democratic institutions is a key part of the Orbán and Fidesz agenda. It
does not only serve to secure their power but also facilitates corruption, as safeguards are
removed and the state can be remodelled to serve the interests of a narrow elite.

Duterte’s government record, so far, is internationally known for its human rights violations in
his ruthless campaign to fight crimes and drug trafficking: extra-judicial killings resulted in over
12,000 victims since 2016. Already during his electoral campaign, it became known that he had
several bank account which he failed to declare in his asset declarations as a presidential
candidate (Santos Jr. 2016). After several officials, including the minister of the interior, were
fired on corruption charges, at least a quarter of them were re-hired shortly after (Calonzo
2018).

Populism and corruption are inherently interlinked. Corruption and the lack of good governance
are fuelling mistrust in public institutions and the grievances that populist can use for their
electoral campaigns. It supports the dichotomy of the “true people” standing against the
“corrupt elites” that is inherent in populism as a political ideology and in populist rhetoric.
Corruption

Attempts to define corruption more generally by international organizations, such as the World
Bank (1997), focus on the misuse or the abuse of public office for private gains (see also Rose-
Ackerman 1999, p. 91). Although this definition leaves considerable room for interpretation (for
example, what exactly constitutes misuse or private gain?), it has become widely accepted
within the literature (Rose-Ackerman 1999, Sandholtz & Koetzle 2000, Treisman 2000,
Lambsdorff 2002, Kunicova & Rose-Ackerman 2005, Golden 2010, Klasnja et al. 2016, to
name ˇ just a few). Empirical manifestations of corruption may thus include fraud,
misappropriation of public funds, or the acceptance of bribes.
Never Forget the First Time: The Persistent Effect of Corruption and the
Rise of Populism in Italy

Arnstein Aassvea , Gianmarco Danieleb and Marco Le Mogliea

Following the financial crisis in 2007, and the ensuing economic recession in Europe, combined
with austerity in the countries hit hardest by the crisis, unemployment rose to unprecedented
levels (Algan et al., 2017). Moreover, the economic recession and its austerity measures were
accompanied by labour market reforms, implying weaker protection for many European
workers and an increasing degree of economic insecurity (Guiso et al., 2017). Consequently, the
onset of austerity would quite naturally be interpreted as institutional failure (Guiso et al.,
2018), and therefore bring about a worsening of the public’s perception of politicians’ abilities,
both in terms managing public finances (Daniele et al., 2018) and their loyalty (Di Tella and
Rotemberg, 2018). In a similar vein, Armingeon and Guthmann (2014) argue that the economic
crisis made voters realize that their own institutions were no longer effectively protecting them,
which must have led to a decline in trust (Armingeon and Ceka, 2014).

An extensive literature argues that corruption is detrimental for the legitimacy of representative
democracy, through its negative effect on attitudes towards institutions (Ares and Hernández,
2017; Solé-Ollé and Sorribas-Navarro, 2017; Morris and Klesner, 2010; Bowler and Karp,
2004; Anderson and Tverdova, 2003; Della Porta and Vannucci, 1997), electoral outcomes (De
Vries and Solaz, 2017; Cobb and Taylor, 2015; Chong et al., 2015; Nannicini et al., 2013;
Stockemer et al., 2013; Costas-Pérez et al., 2012; Hirano and Snyder Jr, 2012; Ferraz and Finan,
2008) and quality of politicians (Cavalcanti et al., 2018; Bernheim and Kartik, 2014). However,
these studies only demonstrate short run effects, or, less frequently, medium term effects.
Whether corruption has long term effects is largely unknown.
The idea that youngsters tend to be more impressionable than elders has strong support in
psychology, where the phenomenon is referred to as the “sensitive age hypothesis” (Lau and
Redlawsk, 2008; Borghans et al., 2008; Sears and Funk, 1999; Sears and Valentino, 1997;
Krosnick and Alwin, 1989). Moreover, among young individuals, we focus on first time voters,
which are likely to be the most sensitive to failures of the institutional and political system, as in
the case of corruption. By permanently affecting individuals’ attitudes, the effect of youth
exposure to corruption might eventually extend to individuals’ current voting preferences.

Immediately after the Italian 2018 national election, we collected a follow-up survey on the
Trustlab sample for Italy, which reached a good take up rate (around 60% of the original
sample). We find that first time voters at the time of the “Clean Hands” scandal are more likely
to vote for populist parties, especially those at the right side of the political spectrum. The
probability of voting for a populist party at the 2018 elections is 9% higher for those being first
time voters when the Clean Hands scandal erupted. Therefore, youth exposure to corruption
seems to affect also current political preferences, as measured by current voting choice.

At the beginning of the 90s’ the “Clean Hands” investigation revealed the biggest corruption
scandal in Italian modern history, consisting in a vast and established corruption scheme, in
which public procurement were systematically assigned in exchange for illegal contributions to
political parties (Newell, 2000). About 23% of the Italian parliamentarians were charged with
corruption or other related crimes and national politicians were charged in 19 out of 20 regions
(Figure 1, panel (a) shows the increase in charges compared to previous electoral terms). The
two incumbent parties, the Christian Democrats (DC) and the Italian Socialist Party (PSI) were
particularly exposed and hard hit (i.e. around 75% of the MPs were involved in the corruption
scheme, see Figure A.3). The scandal lead in fact to the dissolution of both parties. It was also a
scandal that was extensively covered by all the main media outlets at the national and at the
local level, as shown in Figure 1. Figure 1, panel (b) shows the number of front pages of the
most read Italian newspaper (Corriere della Sera) devoted to corruption from 1990 to 1997.
During the peak of the scandal in 1993, almost 90% of all front pages covered the “Clean
Hands” scandal.

The “sensitive age hypothesis” argues that the propensity of beliefs and attitudes to change,
differs according to age, highlighting the potential long lasting impact of specific events
experienced during childhood or young adulthood.3 Several studies have validated this theory in
different contexts, showing that patterns of beliefs may be distinctive across generations due to
cohort-specific differences in economic, social and institutional environment (Roth and
Wohlfart, 2018; Alison et al., 2018; Giuliano and Spilimbergo, 2014; Dinas, 2013; Madestam
and Yanagizawa-Drott, 2012; Schuman and Corning, 2012; Erikson and Stoker, 2011; Osborne
et al., 2011; Firebaugh and Chen, 1995; Russell et al., 1992). Secondly, the “first time voter
hypothesis” makes a similar argument. Individuals participating in their first election, receives
an informational political shock, making them more exposed to important political events
occurring at that time, and in particular in the pre-electoral period (Ohme et al., 2017; Zeglovits
and Aichholzer, 2014; Bhatti and Hansen, 2014; Wagner et al., 2012).4 Indeed, early studies
show that first-time voters are more likely to learn and be influenced by campaign-related
information than voters who have already participated in election in the past (O’Keefe and Liu,
1980; Colwell Quarles, 1979; Winick, 1978). The combination of these two ideas leads to our
expectation that the “Clean Hands” scandal might permanently affect first-time voters at the
1994 election by depressing their support for liberal institutions.
The treatment group for our analysis is the cohort of those individuals who were first time
voters during the “Clean Hands” corruption scandal. Importantly for our estimation, two
national elections took place in this period, one right at the beginning of “Clean Hands” scandal,
when its consequences were still inconceivable (on 5th April 1992), and another at the peak of
the scandal on 27th March 1994. However, the impact of the scandal was negligible before the
1992 elections, which took place at the beginning of April. This pattern is clearly visible
looking at the media coverage of the scandal in Figure 1. This trend is also well documented by
Giglioli (1996), who reports that corruption coverage was extremely low in Italian newspapers
in February and March 1992, implicitly assigning a very limited relevance to these events. We
therefore exploit this window to test whether the scandal affects the political beliefs of first-time
voters at the 1994 national election.

Here we consider as first-time voters at the 1994 elections only those born in 1975 or 1976 and
we construct all the other groups consequently.7 In turn, the reference category includes first-
time voters at the election before the scandal, i.e. individuals born in the period 1970-1974 who
went voting for the first time in 1992. This means that in this specification we are including
individuals born in 1974 within the control group although some of them were actually first
time voters at the 1994 elections, while we are still considering as treated those born in 1976,
who instead became in part first-time voters only at the 1996 elections. In other words, for each
election, we are including as first-time voters all individuals born in that specific electoral year.

This choice is motivated by two considerations. First, the combination of the sensitive age
hypothesis and the first-time voter hypothesis naturally points towards a general decreasing size
of the impact of first-time voting with respect to the age at which the individuals vote for the
first time (Bhatti and Hansen, 2014; Bhatti et al., 2012). Second, individuals born between April
and September 19768 , who were not eligible at the 1994 elections, represent a cohort that could
be influenced by the “Clean Hands" scandal. Indeed, they might have been exposed to the 1994
electoral campaign through several channels, such as peer-effects at school (Ajilore and
Alberda, 2017; Campos et al., 2016)9 . Conversely, individuals born between April and
September 1974 might have been already exposed to the 1992 electoral campaign through the
same channels, so partially narrowing their differential exposure to the scandal as first-time
voters at the 1994 elections. In line with the relevance of peer-effects, survey data show that
first time voters at the 1994 elections had a very high propensity of discussing politics with their
peers.
Overall, these findings strongly suggest that “Clean Hands” represents a breaking point for the
generation of first-time voters at the time of the scandal, permanently depressing institutional
trust. Our main findings are confirmed for trust in the Parliament and in the National
Government, while we do not find an effect on trust towards Civil Servants. Importantly, we
find a negative effect on institutional integrity, a proxy of institutional corruption, closely linked
to our theory. Conversely, as a Placebo test, we do not find any effect on trust towards other
institutions not immediately related to the “Clean Hands” scandal, such as the media, the
financial institutions and other individuals (social trust).

The empirical analysis provided in Online Appendix A.5 rules out the effect of two other
important events taking place in Italy during this period and overlapping the "Clean Hands"
investigation: several terrorist attacks perpetrated by the mafia and a period of a severe
economic crisis. In addition to that, the 1994 national elections saw both the collapse of the
parties (e.g. DC), which ruled Italy up to the “Clean Hands” scandal and the entry of Silvio
Berlusconi in the political arena. One could argue that our findings might be linked to these
events. The disruptive and sudden reshape of the political party system and the presence within
the political scenario of an “ambiguous” individual as Berlusconi, in terms of corruption
attitude, might then have led to lower institutional trust. While we do not neglect the importance
of these hypothesis, we consider them part of our argument, since the collapse of the DC and the
entry of Berlusconi in politics were essentially driven by the “Clean Hands” scandal.

Our hypothesis concerns the long-term consequences of corruption. If there are long terms
effects, it would indeed be intuitive that there are both short run and medium run effects for this
cohort as well. In other words, it would be implausible if the long run effect is not preceded by
short and medium run effects. We find that right after “Clean Hands”, young people are more
likely to retain dictatorship to be sometimes better than democracy (i.e. Panel (a)). Moreover,
the results from the 2001 edition (i.e. Panel (b)) show that young individuals are less likely to
trust the national parliament.

We find an interesting geographical pattern, in which first-time voters are attracted by the main
populist player in their area: first-time voters from the South voted more for Five Star
Movement, while those from the North voted more Lega.
We find here a substantial scarring effect for the younger individuals, i.e. those going through
important formative years, a finding that is consistent with the “sensitive age hypothesis”. In
particular, young first-time voters exposed to a corruption scandal react through exerting lower
trust - a change that appears permanent. The fact that corruption can have a permanent impact
on trust is an important finding, which enriches our perspective on how state regimes and
institutions, as well as their failures, affect beliefs and attitudes (e.g. (Becker et al., 2016; Nunn
and Wantchekon, 2011; Grosjean and Senik, 2011; Alesina and Fuchs-Schündeln, 2007)). Our
study provides convincing evidence that memories of past corruption scandals among
individuals, eventually reactivated by the recent economic crisis, may have partially contributed
to the rise of populism in Italy.

Looking Beyond the Incumbent: The Effects of Exposing Corruption on

Electoral Outcomes

Alberto Chong, Ana L. De La O, Dean Karlan and Leonard Wantchekon

Information about corruption may not improve political accountability if voters respond to it by
withdrawing from the political process. We therefore need to further investigate the effect of
information beyond the political fate of the corrupt incumbents. Theoretically, information
about corruption could mobilize voters, who would otherwise abstain, to vote the corrupt
incumbent out of power. Less informed voters may also abstain in order to delegate their vote to
more informed voters (Feddersen and Pesendorfer 1996). On the other hand, information about
corruption can lead to abstentions if voters become disenchanted with their government, and
with the political system in general.3 If the exposure of rampant corruption leads voters to
believe that voting will not benefit them (either because they lose trust in governments, or in
their ability to respond to their constituents’ needs), then they will not bother to vote since the
costs of casting a ballot would be bigger than the benefits (Aldrich 1993).

This article provides unique experimental evidence that information about rampant corruption
decreases voter turnout, incumbent and challenger support, as well as erodes identification with
the corrupt incumbent’s party. The campaign distributed flyers door-to-door in selected voting
precincts with information about the mayor’s responsibility for the provision of public lighting,
safe water, sewage, and local roads. In one group of voting precincts, the flyer also included
information about the percentage of resources mayors spent in a corrupt manner (i.e. spending
where some form of irregularity was identified such as over-invoicing, fake receipts, diverting
resources, fraud, etc.). In another group, the flyer included information on the total amount of
resources available to mayors to invest in public services and the percentage actually spent by
the end of the prior year. In a third group, the flyer included information about the percentage of
total resources that mayors spent in poor areas. The control group received no information.
Using electoral data at the precinct level, as well as a post-intervention survey, we explore the
electoral effects of information about corruption.

Voters need information in order to discern representative from unrepresentative governments

(Manin, Przeworski, and Stokes 1999). In line with this argument, a growing research program
has found that improving information availability (via increased media access) promotes
government responsiveness (Besley and Burgess 2002), contains opportunistic behavior
(Besley, Pande and Rao 2005), prevents widespread theft of public resources (Adser´a et al.
2003, Khemani 2007, Reinikka and Svensson 2005) and improves the performance of
representatives in parliaments (Humphreys and Weinstein 2007; Snyder and Stromberg 2010,
Banerjee et al 2010). The questions that remain relate to how exposure of information about
corruption influences electoral turnout, and challengers’ support.

The evidence on the exposure of corrupt incumbents on challengers’ support is even more scant
than the evidence related to voter turnout. The few studies on this find that while perceived
corruption is not a significant predictor of voting for an opposition party, nonvoting hurts
opposition parties more than the incumbent. Furthermore, measures of perceptions and
experiences of corruption capture only indirectly the information available to voters about the
misuse of public resources by incumbents, which presumably is one of the most relevant pieces
of information that voters need to keep their governments accountable. A final challenge to the
study of the electoral consequences of exposure of political corruption is how to establish a
causal relation. Simultaneity between corruption exposure and voting is a prevalent concern
when using survey data since recall of corruption can be caused by and be itself a cause of
reported turnout.

The legacy of six decades of fiscal centralization may have left behind a variety of
misconceptions among voters. Among them, two of the most pervasive are that few resources
flow to local governments, and that state governors have the discretion to manipulate the
amount and timing of the funding that does trickle down. Even if decentralization has changed
the amount and allocation criteria of federal transfers, local governments commonly blame a
higher level of government for the lack of service delivery. As a result, political responsibility is
diluted, making it difficult for voters to hold their mayors accountable for the state of public
services in their municipality. Mexico seems to be stuck in what Khemani (2007) describes as a
“cycle of low performance and low expectations” in which voters do not expect politicians to be
able to improve service delivery.

Although ASF (Federal Auditor’s Office) reports contain reliable information about local
government’s performance and most of their content is newsworthy, local media do not pick up
this information as often as one would expect, and national media typically only covers
instances of misappropriation of public resources by federal agencies. By the time local
elections occur (approximately three months after the release of the reports for the elections in
this study), the ASF’s massive and detailed reports were no longer on journalists’ radar. Thus,
corruption at the local level is not exposed even if the information is available and free.

The field experiment took place in twelve municipalities that held elections for municipal
governments in 2009In all municipalities, the themes that were most prevalent over the course
of the campaigning were related to issues such as the economy, public safety, and the H1NI
virus. In addition, campaigns run by the nationally incumbent National Action Party (Partido
Acción Nacional, PAN) called for political continuity, while those of the opposition
Institutional Revolutionary Party (Partido Revolucionario Institucional, PRI) sought to convey a
commitment to keeping campaign promises, meeting people’s needs, and recovering the time
lost during PAN and Party of the Democratic Revolution (Partido de la Revolución
Democrática, PRD) administrations. PRI campaign messages were vague. For instance, the
slogan used throughout the country was “Mexico First, You First”.

While corruption was part of the campaigns in some instances, accusations of misuse of public
resources were often not supported by evidence from the ASF reports. Rather, accusations often
consisted of speculations about incumbent parties covertly lending unauthorized support to their
candidate. In collaboration with Innovations for Poverty Action, a nonprofit organization, we
coordinated the distribution of the flyers through two local firms with expertise in such flyer
distribution campaigns. The distribution of the flyers took place approximately one week before
local elections.

With respect to the information revealed by the audit reports and included in the flyers, we were
surprised to learn that mayors do not spend all the money they receive from the federal transfer.
Among the experimental municipalities, mayors spend on average 56 percent of the FISM
transfer (with a standard deviation of 23). By regulation, mayors are supposed to use the FISM
transfer to improve service delivery in poor areas of their municipality. The audit reports show
that among the municipalities in this experiment 83 percent of FISM was indeed allocated to
poor areas (with a standard deviation of 15). Finally, with respect to corruption, the audits show
that on average mayors spent 35 percent of FISM in corrupt manners (with a standard deviation
of 25).

Previous work has consistently find that the effect of exposing corruption on incumbents
depends on the severity of the malfeasance. For example, Chang and Golden (2004) find that
minor crimes had different effect compared to major crimes on legislators reelection rates.
Ferraz and Finan (2008) find that the effect of the audit is increasing in the number of violations
in the audit report. To take this into account, we split our sample into quartiles of corruption
exposed. Then, we compute the mean turnout, mean incumbent and mean challengers’ vote
shares by treatment and control for each quartile.

Why did information about rampant corruption depressed turnout, and challengers’ votes? As
explained in more detail before, one possibility is that voters disengage from politics when they
learn about corruption. Another possibility is that voters who identify with the party of the
corrupt incumbent prefer to abstain than to vote for a party other than their own. The results so
far show that abstentions from partisans of the incumbent party cannot explain by themselves
the effects because challengers’ votes also decrease when corruption is exposed. Furthermore,
voters who learn about rampant corruption are 40 percent less likely to identify with the party of
the corrupt incumbent compared to voters in the control group (significant at the one percent
level). Information about excessive corruption has no effect on the probability that respondents
on the treatment group identify with a challenger party, but it increases the probability that
respondents on the treatment group do not identify with a party by 30 percent (significant at the
one percent level). We conclude from these results that the effect of exposing rampant
corruption is primordially working through an effect of voter’s disengagement, rather than the
party identification argument.

The main contribution of this article it to provide experimental evidence that information about
political corruption influences voters’ electoral decisions beyond their evaluations of the
incumbent. We show that exposing rampant corruption leads to incumbents’ vote loses, but it
also leads to a decrease in electoral turnout, and a decrease in challengers’ votes. Furthermore,
exposure of corruption weakens partisan identification with the corrupt incumbent’s party and
increases share of voters who do not identify with a political party. Thus, under some
circumstances, information about corruption disengages voters from the political process. Our
results speak directly to one of the most pervasive themes in democratic theory. Voters need to
know what their governments do in order to use elections as instruments of control. Yet, voters
also need a more comprehensive institutional set up that allows them to use the information in
ways that promote accountability. John Dunn succinctly explains: “To promote representation,
democracy requires a regime of free information” (in Przeworski, Stokes, and Manin 1999 25).
Such regime involves the flow of information, and it also involves the construction of effective
institutions (i.e. independent courts, and independent accounting offices) that reinforce the
legitimacy and effectiveness of the political system.