Corruption, Party Leaders, and Candidate Selection: Evidence from Italy

• Published date: 01 May 2020
• Author: Raffaele Asquer,Miriam A. Golden,Brian T. Hamel
• Date: 01 May 2020
• DOI: http://doi.org/10.1111/lsq.12259

291

LEGISLATIVE STUDIES QUARTERLY, 45, 2, May 2020

DOI: 10.1111/lsq.12259

RAFFAELE ASQUER

Independent Researcher

MIRIAM A. GOLDEN

European University Institute

BRIAN T. HAMEL

University of California, Los Angeles

Corruption, Party Leaders, and

Candidate Selection: Evidence from

Italy

Research indicates that voters are not particularly effective at removing

corrupt politicians from office, in part because voters make decisions on the basis

of many competing factors. Party leaders are much more single-minded than vot-

ers and will choose to deselect implicated legislators if it means maintaining a

positive party reputation and improving the odds of winning a legislative major-

ity. We examine renominations to Italy’s legislature in two periods marked by

corruption. We compare these renomination patterns with those from the prior

legislature, when corruption lacked political salience. Our analysis shows that in-

cumbent renominations are negatively associated with the number of press men-

tions that link the incumbent to corruption—but only when corruption is salient

to the public. Our study highlights the importance of party leaders in forcing

malfeasant legislators out of office—and reducing corruption—and redirects at-

tention from voters to political elites as a critical channel in enforcing democratic

accountability.

The theory of democratic accountability draws on the

idea that the fundamental role of voters is to threaten elected

politicians with loss of office if they do not perform adequately

(Ferejohn 1986; Przeworski, Stokes, and Manin 1999). The the-

ory is motivated by a bottom-up perspective that sees the evolu-

tion of democratic practices unfolding through the increasingly

effective work of an informed, educated, and vigilant citizenry.

Emerging from this framework is a large literature on whether

voters punish politicians implicated in scandals, especially those

involving political corruption. Summarizing, these studies report

two key findings. First, in survey experimental settings, voters

© 2019 Washington University in St. Louis

292 Raffaele Asquer, Miriam A. Golden, and Brian T. Hamel

overwhelmingly sanction corrupt politicians (Boas, Hidalgo, and

Melo 2019; Klašnja and Tucker 2013; Weitz-Shapiro and Winters

2013, 2017; Winters and Weitz-Shapiro 2016). This suggests a gen-

eral norm against corruption: citizens see corruption as wrong and

view those engaged in it as unfit for office. But studies of actual

voting behavior and electoral returns show that though malfeas-

ant politicians often receive fewer votes than those not accused of

corruption, the former still usually win reelection (Anduiza, Aina,

and Muñoz 2013; Basinger 2013; Boas, Hidalgo, and Melo 2019;

Carlson and Reed 2018; Chang, Golden, and Hill 2010; Chong

et al. 2015; Hamel and Miller 2019; Klašnja, Tucker, and Deegan-

Krause 2016; Peters and Welch 1980; Welch and Hibbing 1997).

For example, Basinger (2013) shows that 73% of scandal-tainted

legislators in the US House won their next primary election, and

of those, 81% won the next general election.

Voters, in short, are not particularly effective at removing

corrupt politicians from office. New work (e.g., Boas, Hidalgo,

and Melo, 2019) has sought to explain why this is—that is, why

corruption in and of itself is not enough to trigger effective politi-

cal sanctioning. One plausible interpretation is that voters’ goals

are multidimensional, and they make electoral choices on the basis

of many factors. Partisanship, racial and ethnic identity, and per-

sonal loyalties loom large in some contexts; likewise, clientelistic

electoral strategies allow some voters to benefit materially from

corruption in ways that overshadow their own negative ethical

evaluations of corruption itself (Anduiza, Aina, and Muñoz 2013;

Boas, Hidalgo, and Melo 2019; Chang and Kerr 2017; Klašnja and

Tucker 2013; Pereira and Barros 2014). The political context also

shapes voters’ decision-making. Voters are limited in the choices

available to them (Muñoz, Anduiza, and Gallego 2016; Pavão

2018). For example, in uncompetitive or one-party-dominant

political contexts, the absence of a viable alternative candidate

restricts the capacity of voters to sanction a corrupt incumbent

(Vera and Beatriz 2017). Or corruption may be so widespread that

all alternative candidates are no less corrupt (Pavão 2018). In these

ways, malfeasance becomes a weak trigger of political decision-

making relative to other, competing factors.

We see political party leaders as much more single-minded

than voters, as well as capable of coordinating party strategy across

constituencies. First, party leaders are maximizing a utility func-

tion that empirically maps onto a simple goal: increasing the likeli-

hood that the party wins or maintains a legislative majority. This

293Corruption, Party Leaders, and Candidate Selection

goal requires, second, that party leaders coordinate their strategy

across constituencies and present to the national electorate a slate

of candidates that collectively promotes a positive party reputa-

tion. Cross-constituency coordination thus operates through the

mechanism of candidate selection and deselection. Voters, as we

have argued above, have utility functions that map empirically

onto multiple goals, and there are often trade-offs across these

goals such that voting to clean up corruption may not always be

the principal goal that drives electoral choice. Voters might over-

look corruption in favor of other dimensions of candidate choice.

At the same time, accusations of corruption against a candidate

in one constituency may spill over onto voters elsewhere, such

that even if the accused were to win office again, candidates affili-

ated with the same party in other constituencies whose electoral

margins are tighter may be negatively affected. The latter consid-

erations may cause party leaders to withdraw renomination from

candidates accused of corruption.

Party leaders are well suited to police politicians because

doing so is compatible with their primary objective: to win elec-

tions. To win elections, parties need a strong party brand on both

policy and valence dimensions. Valence reflects the nonpolicy rep-

utation of the party. If the party is marred in scandal or unable to

legislative effectively and pass basic legislation, its valence value is

diminished. We argue that party leaders will decline to renominate

corrupt politicians from their own party ranks out of concern that

the national party may suffer, as voters come to perceive a weaker

party brand and choose to punish even those honest candidates

running under the same party label. The decisions of party lead-

ers are thus born out of electoral anticipation, but they primarily

reflect the assessments of an electorally motivated elite that co-

ordinates party strategy across the entire country. Unlike voters,

party leaders cannot easily overlook corruption as they seek to put

forth a nationally viable slate of candidates without compromising

the party brand. Our argument then is that party leaders anticipate

the potential behavior of voters across constituencies and decline

to renominate those who may make winning a legislative majority

more difficult.

Nonetheless, party leaders will not always act to clean up

corruption in their own ranks. Previous work shows that politi-

cal elites are more responsive to issues that are salient to the pub-

lic than they are to other issues (e.g., Hutchings 1998; Kollman

1998; Kuklinski and McCrone 1980; Miller and Stokes 1963). Our