research-article2017
CPSXXX10.1177/0010414017710258Comparative Political StudiesBauhr and Charron
Article
Comparative Political Studies
2018, Vol. 51(4) 415–446
Insider or Outsider? © The Author(s) 2017
Reprints and permissions:
Grand Corruption and sagepub.com/journalsPermissions.nav
DOI: 10.1177/0010414017710258
https://doi.org/10.1177/0010414017710258
Electoral Accountability journals.sagepub.com/home/cps
Abstract
While democratic accountability is widely expected to reduce corruption,
citizens to a surprisingly large extent opt to forgo their right to protest
and voice complaints, and refrain from using their electoral right to punish
corrupt politicians. This article examines how grand corruption and elite
collusion influence electoral accountability, in particular citizens’ willingness
to punish corrupt incumbents. Using new regional-level data across 21
European countries, we provide clear empirical evidence that the level
of societal grand corruption in which a voter finds herself systematically
affects how she responds to a political corruption scandal. Grand corruption
increases loyalty to corrupt politicians, demobilizes the citizenry, and crafts
a deep divide between insiders, or potential beneficiaries of the system, and
outsiders, left on the sidelines of the distribution of benefits. This explains why
outsiders fail to channel their discontent into effective electoral punishment,
and thereby how corruption undermines democratic accountability.
Keywords
corruption and patronage, elections, public opinion, voting behavior,
accountability, quality of democracy
1Quality
of Government (QoG) Institute, Department of Political Science, University of
Gothenburg, Sweden
Corresponding Author:
Monika Bauhr, Goteborgs Universitet, Sprängkullsgatan 19, Goteborg 405 30, Sweden.
Email: monika.bauhr@gu.se
416 Comparative Political Studies 51(4)
Introduction
While international organizations, policy experts, and nongovernmental
organizations promote democratic accountability as a critical component to
curb corruption, citizens to a surprisingly large extent opt to forgo their right
to protest and voice complaints, and refrain from using their electoral right to
punish corrupt politicians. The election and reelection of corrupt politicians
pose a major challenge to democratic theorists as it allows democracies to
underprovide public goods, such as health care, economic development, or
environmental protection, which in turn threatens to undermine the legiti-
macy of the democratic system in the eyes of the electorate (Norris, 2011).
This study examines why politicians involved in corruption continue to be
reelected. In doing so, we distinguish between two types of corruption—one,
political corruption scandals among elected officials, and, two, what we refer
to as high-level, “grand corruption” (GC). We explore how the latter influ-
ences voter behavior in response to the former type of corruption, and suggest
that pervasive GC corrodes electoral accountability and explains why institu-
tions rife with venality remain unchecked. In other words, the extent to which
a citizen punishes corrupt politicians is highly contingent upon the context in
which she finds herself. A polity dominated by egregious GC among elites is
organized around informal extended networks of acquaintances, and the per-
sonalized interchange of concrete rewards and punishment, rather than
around abstract impartial principles involving people who may not even be
part of your network (Hale, 2015, p. 20; Karklins, 2005; Ledeneva, 2013).
Such a systemic level of corruption permeates polities and “dictates every
political and economic pursuit” (Hale, 2015), and thereby shakes the very
foundations of liberal democracy, as it builds on “personal bonds where
detached judgment should be” (Piliavsky, 2014, p. 2) and particularism
(Mungiu-Pippidi, 2006) in lieu of impartiality (Rothstein & Teorell, 2008).
We broadly define GC as collusion among the highest levels of govern-
ment that involves major public sector projects, procurement, and large
financial benefits among high-level public and private elites (Rose-Ackerman,
1999, p. 27), transactions which are rarely directly observed by the general
public. We suggest that the level of GC varies considerably in scale and
severity between countries, and, at times, within countries from region to
region and risks corroding electoral accountability through three parallel but
interrelated electoral effects. In environments where GC is systemic, patron-
age-like exchanges are common and enhance loyalty to corrupt politicians, as
defectors may be deprived of spoils and face the moral costs involved of
breaching loyalty obligations. A polity with high or even systemic levels of
GC also produces resignation and voter abstention, as the electorate is more
Bauhr and Charron 417
likely to experience alienation and cynicism about the potential of the politi-
cal system to contain corruption. In addition, polities dominated by egregious
GC craft deep divide between citizens, particularly between the insiders of
collusive networks of exchange and spoils, or the real or potential beneficia-
ries of GC, and outsiders, those excluded from such collusion. In particular,
insiders remain loyal to corrupt politicians, irrespective of the politicians
being involved in a corruption scandal, whereas outsiders are more likely to
abstain from voting altogether. This contributes to explaining the failure of
outsiders to canalize their grievances into effective electoral punishment in
places where such behavior is most needed—in democracies with systemic
corruption.
Our analysis contributes to the growing literature on how corruption
affects voting behavior in democracies. We make a theoretical contribution
by developing a comprehensive multilevel theory, explaining why and how
GC corrodes electoral accountability and maintains corrupt politicians in
power. This contributes toward explaining why both virtuous and vicious
cycles of corruption are maintained and perpetuated. We model individual
characteristics of voters and macrolevel factors to interact to explain electoral
responses to a political corruption scandal. At the individual level, our
insider–outsider framework distinguishes between respondents, and at the
macrolevel we focus on variation in the level of grand (elite-level) corruption
across polities. In addition, as opposed to only focusing on isolated responses
to corruption, such as whether a voter would vote for a corrupt party or
whether they would abstain from voting altogether, we analyze three types of
electoral responses to corruption using a development of Hirschman’s (1970)
framework of Exit, Voice, and Loyalty. How citizens express discontent is of
central importance to the functioning of democracy, and mechanisms of
accountability operate very differently depending on the type of response that
citizens perceive to be most viable.
We also make an empirical contribution. We test our theory with a large
original survey that asks respondents directly how they would respond elec-
torally if their preferred party was involved in a corruption scandal, and com-
bine these survey data with a newly developed, hard measure of GC among
185 European regions in 21 countries. Our findings provide clear empirical
evidence that pervasive GC creates an environment where corrupt incum-
bents are more likely to be reelected, undermining the effectiveness of elec-
tions in curbing political corruption. In such contexts, voters are less inclined
to hold a corrupt incumbent accountable by switching to another party.
Furthermore, insiders are more likely to be loyal to their preferred party,
whereas outsiders become more inclined to abstain from voting. Thus, we
provide systematic evidence for a growing sentiment in the literature that has
418 Comparative Political Studies 51(4)
mostly been anecdotal—that elections can perpetuate (or even increase) cor-
ruption in places where elite collusion is pervasive (Bauhr & Grimes, 2014;
Mungiu-Pippidi, 2006; Persson, Rothstein, & Teorell, 2013). Conversely, we
find that, in environments where GC is less prevalent, accountability in the
form of voting for an alternative party becomes much more likely, and this is
particularly so among insiders. In other words, we find that insiders perpetu-
ate the status quo and defend the prevailing system in low as well as high
corrupt contexts: In “cleaner” societies, insiders defend the low corrupt equi-
librium, in contexts where GC is pervasive insiders loyally maintain corrupt
politicians in power.
The article proceeds as follows: First, we provide a brief overview of the
literature on electoral responses to corruption, and argue for the importance of
better integrating knowledge about how the functioning and nature of GC
shape loyalty structures and cynicism, and how these effects vary systemati-
cally across different groups of citizens. We then develop a theory about how
GC influences citizens’ responses to corruption scandals, focusing particularly
on loyalty, demobilization, and the divisiveness of GC. Third, we discuss our
data and measurements: We test our hypotheses using novel survey data for
more than 85,000 respondents in 24 European countries along with new data
on high-level administrative corruption in 185 regions. The data uses approxi-
mately 1.45 million awarded public procurement contracts and measure in
particular restrictions on open competition for government contracts to benefit
well-connected companies (Fazekas & Kocsis, 2015). The fourth section pres-
ents our results, and the fifth section concludes the article.
suggests that candidates are punished to only a very limited extent for being
implicated in corruption scandals.1
An emerging literature seeks explanations for the shortcomings of elec-
toral and societal accountability and for why it fails to contain corruption.
This body of research typically assumes that political corruption leads to
some level of indignation, but that this indignation is for various reasons not
canalized into electoral punishment (see, for instance, de Sousa & Moriconi,
2013). At least three explanations seem to be particularly prominent in this
literature. Explanations related to trade-off suggest that the reason why cor-
rupt politicians survive in office is that voters simply do not prioritize having
a noncorrupt political representative. Other factors, such as the politician’s
ability to attract investment, build successful coalitions, generate economic
growth, or make otherwise popular decisions, are often valued more highly
by most voters (see, that is, Konstantinidis & Xezonakis, 2013; Zechmeister
& Zizumbo-Colunga, 2013). Others find limited support for the existence of
trade-offs, and instead suggest that the reason why corrupt officials can main-
tain political support is that voters lack information about wrongdoings
(Winters & Weitz-Shapiro, 2013, see also Chang et al., 2010; Costas-Pérez,
Solé-Ollé, & Sorribas-Navarro, 2012). Furthermore, various institutional
explanations, such as press freedoms, electoral and party systems, the delega-
tion of power and the level of independence of the judiciary (Charron &
Bågenholm, 2016; Ferraz & Finan, 2008; Gerring & Thacker, 2004; Kunicová
& Rose-Ackerman, 2005), may contribute toward explaining reelection, not
least because they can influence the effectiveness of protests and citizen
mobilization. In short, the reason why citizens do not voice complaints is that
they lack information, lack institutional opportunities to do so, or simply do
not prioritize anticorruption efforts.
While all these explanations are plausible, we seek to take a step back in
this article and ask a more fundamental question: What determines citizens’
institutional opportunities to punish corrupt politicians, information about
corruption, and the extent to which they prioritize politicians running on anti-
corruption mandates? We suggest that the importance of these explanations
varies by context, and is in turn partly determined by the level of GC, and in
particular the extent to which such corruption pervades a given society. While
previous studies have distinguished between the effects of experiences of
corruption/bribery and perceptions of societal-level corruption on voting
behavior (Klašnja, Tucker, & Deegan-Krause, 2016) or compared the effects
of different forms of corruption on voting behavior and civic engagement
(Bauhr, 2016; Fernández-Vázquez, Barberá, & Rivero, 2016), our study is
unique in that we model the level of societal corruption as a macrolevel vari-
able using an objective measure, as opposed to the perceptions of individual
420 Comparative Political Studies 51(4)
future elections” (Folke, Hirano, & Snyder, 2011, p. 568; Golden, 2003). In
settings with high levels of GC, less formal rules (as opposed to impartial
merit) tend to dictate employment, and winners in the marketplace conse-
quently tend to be privileged insiders (Rose-Ackerman, 1999). GC increases
the importance of informal exchanges and produces strong loyalty ties among
both real and potential “clients” in a GC system, which reduces the extent to
which citizens elect to switch parties. These loyalty codes extend far beyond
the presence of direct vote buying, and may be present even in contexts where
elections are considered “free and fair,” violence and fraud are relatively
uncommon and voters are generally neither directly incentivized nor coerced
to vote for a corrupt incumbent.
Political change in contexts dominated by egregious GC often comes at a
great risk as it threatens potential and future job, career and even subsistence
opportunities for many citizens and their extended families. Thus, high levels
of GC often increase the stakes of defection and voice. Long-term exposure
to business, public career, and recruitment structures built on GC, where pub-
lic contracts and positions are distributed on the basis of personal contacts
rather than merit, fundamentally shapes loyalty structures in society as well
as perceptions of “How things get done” (Hale, 2015; Ledeneva, 2013). This
forms our first hypothesis.
Our theory thus posits that GC both decreases voice and increases exit, but
that these effects are strongly heterogeneous between different segments of
the population and depend on the extent to which GC corruption permeates
the system. GC divides societies between insiders and outsiders and, as the
prevalence of GC increases, insiders become increasingly loyal to corrupt
incumbents and less likely to punish corrupt politicians.
Question 1: What political party would you vote for if the national parlia-
mentary election were today?
426 Comparative Political Studies 51(4)
In all, 20.7% of all respondents answered “1,” whereas 33.7% and 38.6%
responded “2” and “3,” respectively. About 6.9% responded “Don’t know.”
The third response most closely represents our idea of exit, as the respondent
is clearly choosing to abstain from the democratic process. Thus, we create a
dummy variable that equals “1” if an individual selected this response and
“0” if otherwise. The second response, “vote for an alternative party not
involved in the corruption scandal,” most closely captures our idea of
accountability, or voice, as the respondent intends to vote, but elects to punish
the corrupt party (despite it being their first preference) by voting for an alter-
native. A dummy variable is thus created with this response as well, where an
individual receives a “1” if he or she gave the second answer and “0” if oth-
erwise. Loyalty is coded “1” if the respondent answered “still vote for the
preferred party” and “0” if otherwise. Figure 1 gives the proportion of each
response by country.
There are of course shortcomings with broad, hypothetical questions such
as these. First, the magnitude of the scandal is not mentioned and, even if
Bauhr and Charron 427
Our key independent variables are as follows: First, while there is no clear
consensus in the field as to how this concept should be measured, to avoid
possible endogeneity with a perceptions measure, we capture GC with an
objective measure with the aid of data from Fazekas and Kocsis (2015). This
measure uses public procurement data to assess the risk of high-level corrup-
tion and favoritism, which is the best measure available given that our con-
cept of high-level corruption among elites is of concern here. The data, for
EU28 between 2009 and 2013, contain information on individual public pro-
curement tenders that has to do with regulated administrative procedures in
which public bodies purchase goods and services, including contract value,
the number of bidders, the deadline for submitting bids, and the assessment
criteria used. The measures tap into deliberate restrictions on open competi-
tion for government contracts to benefit well-connected companies, which
we argue is the best available comparative metric of our concept of GC. To
avoid distortion from less competitive markets such as defense, the indicators
are defined only when at least 10 contracts were awarded in the period 2009-
2013, indicating a sufficient demand for bids from at least two operating
companies. In all, the dataset is based on approximately 1.45 million awarded
contracts. The simplest indication of corruption risk is when only one bid was
submitted on a tender in an otherwise competitive market. Hence, the per-
centage of single-bidder contracts awarded in all the awarded contracts is the
most straightforward measure we use to capture our concept and, we argue,
this indicates a climate very favorable for the type of GC corruption we envi-
sion in our theory. As this—like most other measures of corruption—is not a
direct measure of corruption, one clear potential pitfall with this measure is
that it is not actually capturing GC, but something else, such as lack of com-
petitive markets. We find however that correlations between this measure and
alternative measures based on citizen or expert corruption perceptions and
experiences are quite significant at both the country and regional levels for
this sample, which gives us some assurance of validity.7 The data are also
available at the regional level for NUTS 1 and NUTS 2 levels. Country-level
averages are reported in Online Appendix Table A2.
Second, there is also a lack of consensus on how one should operationalize
an insider. In political economy, particularly labor market studies, an insider
is someone with full-time (or willing part-time) employment, whereas out-
siders are those who are outside of the labor market, those who are unem-
ployed (Rueda, 2005). However, this distinction has been used much less so
in studies of voting behavior (Lindvall & Rueda, 2014). Most studies in polit-
ical science and development economics about how corruption affects elec-
toral behavior focus on the related concept of clients (a type of insider),
whereby voters are considered clients if they claim to directly rely on a
Bauhr and Charron 429
powerful patron for help when in need (Stokes et al., 2013). In a recent study
on corruption tolerance in African countries, Chang and Kerr (2016) distin-
guish two types of insiders—“network insiders” (similar to the aforemen-
tioned concept of client) and “identity insiders,” citizens who either share
partisan or ethnic ties with the sitting government. Most of these past studies
however focus on Latin American or African countries, whereas studies of
Europe using the insider–outsider framework focus on one’s position in the
labor market.
Our conception of an insider is similar to Kitschelt and Wilkinson’s (2007,
p. 10) idea of “contingent direct exchange,” in that we attempt to capture
whether a respondent potentially has “something to lose” politically and eco-
nomically by a political change resulting from an election. We therefore
attempt to merge the operationalization of the concept insider from both past
corruption and labor market studies, and thus include two components of
insiderness—economic and political. On the economic side, we follow the
operationalization of most political economy scholars of the labor market
(Rueda, 2005)—insiders are those with full-time employment. In several
robustness checks, we go one step further and include only full-time public
sector workers as a new administration can make political appointments, par-
ticularly within the public sector, making this type of employment potentially
especially relevant in this context. In addition, we add a political element.
Using a similar operationalization for the political element to Chang and Kerr
(2016) whereby we code an individual as an insider if he or she is a partisan
supporter of a sitting party in government.8
Each respondent is provided with a country-specific list of relevant politi-
cal parties (party lists included all sitting parliamentary parties and poten-
tially relevant parties outside parliament, based on country-expert
assessments). If a respondent answered that they would support a party in the
sitting government (at the time of the survey), then they would be coded “1”
and “0” if they said they would support any other party,9 and thus the compo-
nent of political insiderness is nonideological and has only to do with the
current power of one’s party. The binary measures are combined to equal “1”
for respondents with full-time employment who support the sitting party in
power and “0” if otherwise.
As stated in the hypothesis, we expect that insiders will be more inclined
to express loyalty as an electoral response as the level of contextual GC
increases, meaning that their personal gain through political ties would out-
weigh their desire for a cleaner political party. Conversely, the extent to
which insiders are also expected to opt for voice decreases in contexts with
higher levels of GC. We also hypothesize that GC affects insiders differently
than outsiders in terms of electoral outcomes. Therefore, we test the cross-
level interaction term—“insider×GC.”
430 Comparative Political Studies 51(4)
Several control variables are included at both the individual and country
levels based primarily on empirical research of election studies and voter
turnout (Geys, 2006; Smets & van Ham, 2013). First, we account for gender,
income, age, education, population of one’s residence, one’s self position on
the left–right scale, and general satisfaction with the economy. In addition,
we control for whether the respondent has paid a bribe in the last 12 months
for any public service. Thus, bribe corruption is measured at the microlevel,
whereas GC corruption is measured at the macrolevel (regional and country
level). The second-level fixed effects are population density (logged), ethnic
heterogeneity, recent past levels of voter turnout (average of the past two
election cycles), and the number of parties in the system, which Charron and
Bågenholm (2016) find to be relevant to corruption voting.
Due to our dependent variables being binary, we test the effects of corrup-
tion on voice and exit with logit estimation. Given that GC varies systemati-
cally within and across countries, we test for the effects with both logit
controlling for the country-level fixed effects described above as well as with
hierarchical logit, allowing for random country intercepts. Structuring the
data hierarchically has the added advantage of allowing us to test the cross-
level hypothesis more directly—that is, the insiders are more prone to loyalty
as the context of GC becomes greater.
Estimation
We begin by employing a hierarchical design, whereby the cross-level inter-
action effects are tested directly for the three outcomes separately. We find
that country-level differences in the dependent variable are present, and thus
country-level effects must be taken into account to avoid bias estimates.10
Simply accounting for regional and/or country dummy variables for individ-
ual-level variation in an ordinary-least-squares (OLS) model can lead to
problems because the error terms of the lowest level unit (individuals) within
the same group will still be correlated. Thus, we elect to explain individual
levels of voting in a hierarchical logit model with two levels, “i” and “j,” to
represent individual and country levels, respectively. The basic model used
here is as follows:
voting ij = α j + β2 Insiderij + β2 GC j +
β3 Insider × GCij + ϕnij +
ωn j + µ 0 j + e0ij ,
and ωn represent individual- and country (or regional)-level fixed effects con-
trols, respectively, µ0j shows the random-level intercepts for the macrolevel,
and e0ij is the error term.11
As the number of second-level observations (countries) is only 21, the
sample thus somewhat violates the so-called “30/30 rule” (Maas & Hox,
2005); we elected to remedy this in two ways: First, we estimate several of
the main models reported as individuals nested in regions (NUTS 1 and
NUTS 2 depending on the country).12 This has several advantages, as we see
it. One, the second level of the estimation becomes 185 (as opposed to 21),
giving us more than a sufficient sample size for hierarchical modeling, greater
degrees of freedom, and greater accuracy in the results as compared with the
limited country-level estimation on the second level (Maas & Hox, 2005).
Two, although the country context is salient, there is sufficient spatial varia-
tion in corruption, loyalty, and voting outcomes to warrant accounting for
regional differences,13 thus giving us a more precise picture of the landscape,
in particular in countries such as Italy, Spain, Belgium, and Germany that
have considerable regional variation. However, we do run models where the
country is used at the second level as well as standard logit with country fixed
effects. We also check for the effects of outliers in each model, and rerun the
analyses removing one country at a time in subsequent models (found in the
online appendix).
Results
Table 1 elucidates the results for each of the three outcomes. For the sake of
space, we report only the effects of GC and essential model statistics. For
each outcome, a baseline model was run (no controls—Models 1, 3, and 5)
along with a second model with full controls (Models 2, 4, and 6). We report
odds ratios to make the results more interpretable. We see that while GC does
not appear to have a significant effect on the probability of choosing loyalty,
it has an independent and significant effect in the full models for voice and
exit. For example, all things being equal, a 1-unit increase in the GC measure
is expected to increase the odds of exit by 13% and decrease the likelihood of
voice by 8%.
Overall, these results are more or less in line with our first two hypothe-
ses—that all things being equal, GC undermines electoral accountability by
decreasing the likelihood of voice and increasing the likelihood of exit,
although the effects for the latter are more robust to the inclusion of controls.
For example, as can be observed in Figure 2, for roughly the “cleanest” 20%
of the regions in the sample, the model predicts that the most likely electoral
432
Table 1. Electoral Responses to Corruption: Hierarchical Estimation Test of H1 and H2.
Fixed effects 1 2 3 4 5 6
Level 1 (individual)
Intercept 0.21 (.000) 0.04 (.000) 0.36 (.000) 0.07 (.000) 0.97 (.76) 3.28 (.000)
Level 2 (region)
Grand corruption 0.94 (.08) 1.01 (.65) 0.85 (.000) 0.92 (.04) 1.25 (.000) 1.13 (.000)
Random effects
Random intercept (SE) 0.45 (0.03) 0.37 (0.02) 0.46 (0.03) 0.42 (0.02) 0.51 (0.03) 0.38 (0.02)
Wald test Pr (χ2) .000 .000 .000 .000 .000 .000
ICC .06 .04 .07 .05 .07 .04
Log likelihood (It. 0) −36,635.6 −33,650.4 −46,215.3 −44,148.2 −47,862.4 −44,859.9
Log likelihood (final) −36,566.9 −33,538.7 −46,147.8 −44,020.7 −47,807.9 −44,729.1
LR test .000 .000 .000 .000 .000 .000
Ind. obs (regions) 74,338 (185) 74,338 (185) 74,338 (185) 71,477 (181) 74,338 (185) 71,477 (181)
Odds ratios reported with p values in parentheses from hierarchical logit estimation. Models 1, 3, and 5 show the baseline interaction terms
without controls, whereas 2, 4, and 6 include full controls and Level 1 and Level 2. Level 1 controls are gender, age, income, education, population
of residence, satisfaction with current economy, unemployment, left–right self-placement, support for government party (insider), and corruption
experience (bribe). Level 2 controls include ethnic heterogeneity, population density (logged), past turnout levels, and the effective number of
parties. ICC stands for “interclass correlation,” and ranges from 0 to 1. ICC = intraclass correlation; LR = likelihood ratio.
Bauhr and Charron 433
response is voice, whereas for more corrupt regions exit becomes the option
of the plurality of respondents, demonstrating evidence for Hypotheses 1
and 2.
In Table 2, we test Hypotheses 3 and 4 by including a cross-level interac-
tion between GC and whether an individual respondent is an insider or not.
Our anticipation is that being an insider will have an impact on how someone
views a corruption scandal in his or her party, and that this decision is also
conditional on the level of GC in the region where they reside. In Models 1,
3, and 5, we produce baseline estimates with only the main variable and the
interaction, whereas in 2, 4, and 6 all controls are included. Due to the insider
variable being binary, “1” for insiders and “0” for outsiders, the GC coeffi-
cient can be interpreted as the effect of GC among outsiders. Again, odds
ratios are reported.
With regard to loyalty, we find a significant interaction effect, implying a
divergence of behavior between insiders and outsiders, as GC increases. At
almost all levels of GC, insiders are more inclined to show loyalty than
434
Table 2. Electoral Responses to Corruption: Tests of H3.
Fixed effects 1 2 3 4 5 6
Level 1 (individual)
Insider 3.60 (.000) 3.16 (.000) 0.90 (.03) 0.81 (.000) 0.54 (.000) 0.65 (.000)
Intercept 0.15 (.000) 0.04 (.01) 0.37 (.000) 0.07 (.000) 1.09 (.31) 3.55 (.000)
Level 2 (region)
Grand corruption (CG) 0.89 (.003) 0.92 (.02) 0.87 (.000) 0.94 (.11) 1.25 (.000) 1.13 (.000)
Insider × GC 1.23 (.000) 1.23 (.000) 0.94 (.002) 0.93 (.000) 0.94 (.004) 0.96 (.04)
Random effects
Random intercept (SE) 0.42 (0.02) 0.38 (0.02) 0.46 (0.03) 0.42 (0.02) 0.49 (0.03) 0.38 (0.02)
Wald model test Pr (χ2) .000 .000 .000 .000 .000 .000
ICC .05 .04 .06 .05 .07 .04
Log likelihood −35,639.2 −33,532.8 −46,671.1 −44,013.5 −47,486.9 −44,735.4
LR test .000 .000 .000 .000 .000 .000
Ind. obs (regions) 74,338 (185) 71,477 (181) 74,338 (185) 71,477 (181) 74,338 (185) 71,477 (181)
Odds ratios reported with p values in parentheses from hierarchical logit estimation. Models 1, 3, and 5 show the baseline interaction terms
without controls, whereas 2, 4, and 6 include full controls and Level 1 and Level 2. Level 1 controls are gender, age, income, education, population
of residence, satisfaction with current economy, unemployment, left–right self-placement, corruption experience (bribe). Level 2 controls include
ethnic heterogeneity, population density (logged), past turnout levels, and the effective number of parties. ICC stands for “interclass correlation”
and ranges from 0 to 1. ICC = intraclass correlation; LR = likelihood ratio.
Bauhr and Charron 435
corrupt contexts, they will choose to hold those who engage in it accountable
in contexts where corruption is already comparatively low.
With regard to the third outcome, exit, we find that both being an insider
and the level of GC have strong independent effects. The odds of exit among
outsiders increase by 13% for every unit increase of GC, and there is even a
slight interaction effect between these two factors (the odds of outsiders exit-
ing compared with insiders increase slightly as a function of greater GC). Yet
in all cases, outsiders are considerably more likely to exit than insiders, as
seen in Figure 3, which corroborates Hypothesis 4.
Table 3 highlights the findings of the effects of the control variables. The
results corroborate much of the existing literature on turnout and party loy-
alty. On average, given a corruption scandal in one’s preferred political party,
females are less prone to loyalty and more prone to voice and exit. People
with higher socioeconomic status (education and income) show a greater pro-
pensity for voting—either expressing loyalty or voice—than for abstention.
Participants older than 60 are more loyal and less likely to switch parties,
while the effects of age on exit are strongly negative, with younger voters
being the most likely to abstain. Residents of larger urban areas show more
loyalty than rural residents, while the latter are more prone to voice and exit.
On a left–right spectrum (5-point scale), we observed clear “u-shaped” (and
“inverted u-shaped” patterns), with center voters least likely to show loyalty
and most likely to either voice or exit relative to voters on the ideological
fringes. The effects of perceptions on the economy work according to our
Bauhr and Charron 437
estimate these two sequences in Table A9 via sequential logit models. The
main findings are remarkably consistent and robust to these alternatives.
Conclusion
While democratic accountability is perhaps not only the most desirable but
also the most indispensable means to reduce political corruption, democra-
cies clearly vary in their ability to contain corruption. This study suggests that
GC corrodes electoral accountability and explains why institutions rife with
venality remain unchecked. In particular, we suggest that egregious GC is
severely detrimental to the electoral punishment of corrupt politicians as it
canalizes grievances into resignation, alienation, and exit rather than indigna-
tion and voice, enhances loyalty to corrupt politicians, and crafts deep soci-
etal divides between insiders, or beneficiaries of the dominant system, and
outsiders, those left on the sidelines of the spoils of the dominant network.
All these effects—loyalty, demobilization, and divisiveness—we suggest
maintain corrupt politicians in power.
Using a large-N multilevel comparative design that allows us to directly
model cross-level interactions between individual and societal factors, a new
objective measure of GC using approximately 1,45 million procurement con-
tracts and surveys at the regional level for 185 European regions in 21 coun-
tries, we provide several novel and noteworthy findings: First, we provide
clear empirical evidence that GC and elite collusion create an environment
where corrupt incumbents are more likely to be reelected, even if politicians’
involvement in corruption scandals is widely known. In contexts where GC
is high, citizens respond to corruption scandals either by simply staying loyal
to the corrupt politician (loyalty) or by abstaining from voting altogether
(exit). Conversely, in contexts where elite-level GC is low, citizens are more
likely to switch parties in response to a corruption scandal (voice).
Interestingly, we find no evidence for a systematic effect of personal experi-
ence with petty corruption on either voice or loyalty, and only a relatively
small effect on exit, suggesting that the contextual effects of elite-level cor-
ruption are more powerful in affecting voting behavior than personal experi-
ences in a comparative European context.16
Second, we develop an insider–outsider framework and observe important
differences not only in the way in which insiders and outsiders respond to a
corruption scandal but also that these responses are highly contingent upon
the context in which voters find themselves and in particular on the level of
GC in their polity. Insiders in polities where GC reaches systemic levels typi-
cally stay loyal to corrupt politicians, whereas outsiders in such contexts are
instead more likely to refrain from voting and exit the electoral system
Bauhr and Charron 439
Funding
The authors disclosed receipt of the following financial support for the research,
authorship, and/or publication of this article: This project has received funding from
the European Research Council (ERC) under the European Union’s Horizon 2020
research and innovation programme (Grant Agreement No. 339571) called “PERDEM,”
the Swedish foundation for humanities and social sciences and the Seventh Framework
Programme for Research and Development of the European Union (Project Number
290529) called “ANTICORRP” (http://anticorrp.eu/).
Notes
1. See Fernández-Vázquez, Barberá, and Rivero (2016) for an excellent overview
of the findings of these studies.
2. See Fernández-Vázquez, Barberá, and Rivero (2016); Funk (1996); Welch and
Hibbing (1997).
3. However, clientelist parties may also buy voter abstention as a strategy to win
elections (Cox & Kousser, 1981), indicating that the mobilization effect of party
machines may not be easily predictable.
Bauhr and Charron 441
References
Anduiza, E., Gallego, A., & Muñoz, J. (2013). Turning a blind eye: Experimental
evidence of partisan bias in attitudes toward corruption. Comparative Political
Studies, 46, 1664-1692. doi:10.1177/0010414013489081
Auyero, J. (2001). Poor people’s politics: Peronist survival networks and the legacy
of Evita. Durham: Duke University Press.
Bågenholm, A. (2013). Throwing the rascals out? The electoral effects of corruption
allegations and corruption scandals in Europe 1981–2011. Crime, Law and Social
Change, 60, 595-609.
Basinger, S. J. (2013). Scandals and Congressional elections in the post-Watergate
era. Political Research Quarterly, 66, 385-398.
Bauhr, M. (2016). Need or greed? Conditions for collective action against corruption.
Governance. Advance online publication. doi:10.1111/gove.12232
Bauhr, M., & Grimes, M. (2014). Indignation or resignation: The implications of transpar-
ency for societal accountability. Governance, 27, 291-320. doi:10.1111/gove.12033
Birch, S. (2010). Perceptions of electoral fairness and voter turnout. Comparative
Political Studies, 43, 1601-1622.
Chang, E. C. C., Golden, M. A., & Hill, S. J. (2010). Legislative malfeasance and
political accountability. World Politics, 62, 177-220. doi:10.1017/S004388
7110000031
Chang, E. C., & Kerr, N. N. (2017). An insider–outsider theory of popular tolerance
for corrupt politicians. Governance, 30(1), 67-84.
Charron, N., & Bågenholm, A. (2016). Ideology, party systems and corruption
voting in European democracies. Electoral Studies, 41, 35-49. doi:10.1016/j.
electstud.2015.11.022
Charron, N., Dahlström, C., Lapuente, V., & Fazekas, M. (2017). Careers, connec-
tions, and corruption risks: Investigating the impact of bureaucratic meritocracy
on public procurement processes. The Journal of Politics, 79(1), 89-104.
Charron, N., Dijkstra, L., & Lapuente, V. (2014). Regional governance matters:
Quality of government within European Union member states. Regional Studies,
48, 68-90. doi:10.1080/00343404.2013.770141
Charron, N., Dijkstra, L., & Lapuente, V. (2015). Mapping the regional divide in
Europe: A measure for assessing quality of government in 206 European regions.
Social Indicators Research, 122, 315-346.
Chong, A., De La O, A. L., Karlan, D., & Wantchekon, L. (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, 55-71.
doi:10.1086/678766
Cornell, A., & Grimes, M. (2015). Institutions as Incentives for Civic Action:
Bureaucratic Structures, Civil Society, and Disruptive Protests. The Journal of
Politics, 77, 664-678.
Costas-Pérez, E., Solé-Ollé, A., & Sorribas-Navarro, P. (2012). Corruption scandals,
voter information, and accountability. European Journal of Political Economy,
28, 469-484. doi:10.1016/j.ejpoleco.2012.05.007
Bauhr and Charron 443
Cox, G. W., & Kousser, J. M. (1981). Turnout and rural corruption: New York as a test
case. American Journal of Political Science, 25, 646-663. doi:10.2307/2110757
Crisp, B. F., Olivella, S., Potter, J. D., & Mishler, W. (2014). Elections as instruments
for punishing bad representatives and selecting good ones. Electoral Studies, 34,
1-15. doi:10.1016/j.electstud.2013.08.017
Davis, C. L., Camp, R. A., & Coleman, K. M. (2004). The influence of party systems
on citizens’ perceptions of corruption and electoral response in Latin America.
Comparative Political Studies, 37, 677-703. doi:10.1177/0010414004265879
De Mesquita, B. B., Morrow, J. D., Siverson, R. M., & Smith, A. (2002). Political
institutions, policy choice and the survival of leaders. British Journal of Political
Science, 32, 559-590.
Ecker, A., Glinitzer, K., & Meyer, T. M. (2016). Corruption performance voting and
the electoral context. European Political Science Review. Retrieved from /core/
journals/european-political-science-review/article/corruption-performance-vot-
ing-and-the-electoral-context/655C33877217827725D626BAB83CEAD4
Eggers, A. C., & Fisher, A. C. (2011). Electoral accountability and the UK parlia-
mentary expenses scandal: Did voters punish corrupt MPs? (SSRN Scholarly
Paper No. ID 1931868). Rochester, NY: Social Science Research Network.
Retrieved from https://papers.ssrn.com/abstract=1931868
Fazekas, M., & Kocsis, G. (2015). Uncovering high-level corruption: Cross-national
corruption proxies using government contracting data (SSRN Scholarly Paper
No. ID 2711932). Rochester, NY: Social Science Research Network. Retrieved
from https://papers.ssrn.com/abstract=2711932
Fernández-Vázquez, P., Barberá, P., & Rivero, G. (2016). Rooting out corruption or
rooting for corruption? The heterogeneous electoral consequences of scandals.
Political Science Research and Methods, 4, 379-397. doi:10.1017/psrm.2015.8
Ferraz, C., & Finan, F. (2008). Exposing corrupt politicians: The effects of Brazil’s
publicly released audits on electoral outcomes. The Quarterly Journal of
Economics, 123, 703-745. doi:10.1162/qjec.2008.123.2.703
Fukuyama, F. (2015). Why is democracy performing so poorly? Journal of
Democracy, 26(1), 11-20. Retrieved from https://doi.org/10.1353/jod.2015.0017
Folke, O., Hirano, S., & Snyder, J. M. J. (2011). Patronage and elections in U.S. States.
American Political Science Review, 105, 567-585. doi:10.1017/S0003055411000256
Funk, C. L. (1996). The impact of scandal on candidate evaluations: An experimen-
tal test of the role of candidate traits. Political Behavior, 18, 1-24. doi:10.1007/
BF01498658
Gerring, J., & Thacker, S. C. (2004). Political institutions and corruption: The role of
unitarism and parliamentarism. British Journal of Political Science, 34, 295-330.
doi:10.1017/S0007123404000067
Geys, B. (2006). Explaining voter turnout: A review of aggregate-level research.
Electoral Studies, 25, 637-663.
Gingerich, D. W. (2014). Yesterday’s heroes, today’s villains: Ideology, corrup-
tion, and democratic performance. Journal of Theoretical Politics, 26, 249-282.
doi:10.1177/0951629813495120
444 Comparative Political Studies 51(4)
Golden, M. (2003). Electoral connections: The effects of the personal vote on politi-
cal patronage, bureaucracy and legislation in postwar Italy. British Journal of
Political Science, 33, 189-212. doi:10.1017/S0007123403000085
Hainmueller, J., Hangartner, D., & Yamamoto, T. (2015). Validating vignette and
conjoint survey experiments against real-world behavior. Proceedings of the
National Academy of Sciences, 112, 2395-2400.
Hale, H. E. (2015). Patronal politics: Eurasian regime dynamics in a comparative
perspective. New York, NY: Cambridge University Press.
Hirschman, A. O. (1970). Exit, voice, and loyalty: Responses to decline in firms,
organizations, and states (Vol. 25). Cambridge, MA: Harvard University Press.
Jain, A. K. (2001). Corruption: A review. Journal of Economic Surveys, 15, 71-121.
Johnston, M. (1998). What can be done about entrenched corruption? In B. Pleskovic
& J. Stiglitz (Eds.), Annual World Bank conference on development economics,
1997 (pp. 69-90). Washington, DC: World Bank.
Johnston, R. J. (2006). Is hypothetical bias universal? Validating contingent valu-
ation responses using a binding public referendum. Journal of Environmental
Economics and Management, 52, 469-481. doi:10.1016/j.jeem.2005.12.003
Karklins, R. (2005). The system made me do it: Corruption in post-communist societ-
ies. Armonk, NY: M.E. Sharpe.
Keefer, P. (2007). Clientelism, credibility, and the policy choices of young democra-
cies. American Journal of Political Science, 51(4), 804-821.
King, G., Murray, C. J., Salomon, J. A., & Tandon, A. (2004). Enhancing the validity
and cross-cultural comparability of measurement in survey research. American
Political Science Review, 98, 191-207.
Kitschelt, H., & Wilkinson, S. I. (2007). Patrons clients policies: Patterns of demo-
cratic accountability and political competition. Cambridge, UK: Cambridge
University Press.
Klašnja, M., Tucker, J. A., & Deegan-Krause, K. (2016). Pocketbook vs. Sociotropic
corruption voting. British Journal of Political Science, 46, 67-94. doi:10.1017/
S0007123414000088
Konstantinidis, I., & Xezonakis, G. (2013). Sources of tolerance towards corrupted
politicians in Greece: The role of trade offs and individual benefits. Crime, Law
and Social Change, 60, 549-563. doi:10.1007/s10611-013-9478-2
Kunicová, J., & Rose-Ackerman, S. (2005). Electoral rules and constitutional struc-
tures as constraints on corruption. British Journal of Political Science, 35,
573-606. doi:10.1017/S0007123405000311
Ledeneva, A. V. (2013). Can Russia modernise? Sistema, power networks and infor-
mal governance. Cambridge, NY: Cambridge University Press.
Lindvall, J., & Rueda, D. (2014). The insider–outsider dilemma. British Journal of
Political Science, 44, 460-475.
Maas, C. J. M., & Hox, J. J. (2005). Sufficient sample sizes for multilevel modeling.
Methodology: European Journal of Research Methods for the Behavioral and
Social Sciences, 1(3), 86-92. Retrieved from http://dx.doi.org.ezproxy.ub.gu.se/
Manzetti, L., & Wilson, C. J. (2007). Why do corrupt governments maintain pub-
lic support? Comparative Political Studies, 40, 949-970. doi:10.1177/001
0414005285759
Bauhr and Charron 445
Sousa, L., & de Moriconi, M. (2013). Why voters do not throw the rascals out? A
conceptual framework for analysing electoral punishment of corruption. Crime,
Law and Social Change, 60, 471-502. doi:10.1007/s10611-013-9483-5
Stockemer, D., LaMontagne, B., & Scruggs, L. (2013). Bribes and ballots: The impact
of corruption on voter turnout in democracies. International Political Science
Review, 34, 74-90. doi:10.1177/0192512111419824
Stokes, S. C., Dunning, T., Nazareno, M., & Brusco, V. (2013). Brokers, voters,
and clientelism: The puzzle of distributive politics. Cambridge, UK: Cambridge
University Press.
Sundström, A., & Stockemer, D. (2015). Regional variation in voter turnout in
Europe: The impact of corruption perceptions. Electoral Studies, 40, 158-169.
doi:10.1016/j.electstud.2015.08.006
Sung, H.-E. (2004). Democracy and political corruption: A cross-national
comparison. Crime, Law and Social Change, 41, 179-193. doi:10.1023/
B:CRIS.0000016225.75792.02
Treisman, D. (2007). What have we learned about the causes of corruption from ten
years of cross-national empirical research? Annual Review of Political Science,
10, 211-244. doi:10.1146/annurev.polisci.10.081205.095418
Vossler, C. A., & Kerkvliet, J. (2003). A criterion validity test of the contingent valua-
tion method: Comparing hypothetical and actual voting behavior for a public ref-
erendum. Journal of Environmental Economics and Management, 45, 631-649.
Wantchekon, L. (2003). Clientelism and voting behavior: Evidence from a field
experiment in Benin. World Politics, 55, 399-422.
Welch, S., & Hibbing, J. R. (1997). The effects of charges of corruption on voting
behavior in Congressional elections, 1982-1990. The Journal of Politics, 59, 226-
239. doi:10.2307/2998224
Winters, M. S., & Weitz-Shapiro, R. (2013). Lacking information or condoning cor-
ruption: When do voters support corrupt politicians? Comparative Politics, 45,
418-436. doi:10.5129/001041513X13815259182857
Zechmeister, E. J., & Zizumbo-Colunga, D. (2013). The varying political toll of con-
cerns about corruption in good versus bad economic times. Comparative Political
Studies, 46, 1190-1218. doi:10.1177/0010414012472468
Author Biographies
Monika Bauhr is an Associate Professor at the Quality of Government Institute at the
department of political science, University of Gothenburg. She is the Principal inves-
tigator of the ANTICORRP (Anticorruption Policies Revisited: Global Trends and
European Responses to the Challenge of Corruption), a large-scale multidisciplinary
research program, funded by the European Commission. Her research is mainly con-
cerned with the link between democracy and corruption and the impact of corruption
on international redistribution and foreign aid.
Nicholas Charron is an Associate Professor in political science at Gothenburg
University in Sweden. His primary research interests are in political institutions, cor-
ruption, regional studies, and the European Union.