ABSTRACT
While corruption of different types has been shown to lower popular political trust in democratic regimes, evidence from non-democracies remains inconsistent. In some post-Soviet countries, for instance, widespread bribery and nepotism in the government co-exist with enduring popularity of top political leadership. Drawing on an unusually nuanced dataset from Russia (N = 2,350), we show that, in general, encounters with corruption in the public sector are associated with citizens’ lower trust of their government. At the same time, we theorize two caveats that attenuate this relationship, contributing to inconsistent findings in previous studies. First, we find that the negative association between corruption and political trust is significantly weaker when such corruption is beneficial to ordinary people. Second, citizens tend to “penalize” local rather than central government officials, which, we argue, is a result of top leaders’ ability to manipulate public discourse around corruption.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1. Notably, actual electoral behavior of democratic citizens is less sensitive to corruption scandals than their attitudes (Boas, Daniel Hidalgo, and André Melo Citation2019; Sundström and Stockemer Citation2015, etc.) – possibly, because corruption simultaneously mobilizes action against corrupt politicians and discourages it by reducing the faith in the democratic process (Kostadinova Citation2009).
2. While this large body of scholarship frames political trust as a function of citizen evaluation of how widespread corruption is in the government as well as citizens’ own experiences with corruption, it is important to note that another body of research makes the opposite claim – that trust, especially social trust or generalized interpersonal trust, decreases corruption (Bjørnskov Citation2010; Graeff and Tinggaard Svendsen Citation2013; Kubbe Citation2014; Uslaner Citation2004, Citation2008). In his summary of relevant research, You (Citation2018, 482) concludes that “to date, there is still disagreement about the causal direction” in the well-documented relationship between different kinds of trust and corruption. Although, in this article, we build on a strong foundation of work that discusses the impact that corruption perceptions and behaviors have on political trust, we cannot definitively rule out the possibility of reverse causation and endogeneity.
3. The type of corruption discussed in this study is commonly known as bytovaya korruptsiya in Russian. Typically, it involves relatively small scale monetary or in-kind remuneration between ordinary citizens and public employees in positions that require direct interactions with clients. While, theoretically, such bribes can be paid on behalf of small businesses or entrepreneurs, public sector corruption is analytically and practically distinct from business corruption, which entails large-scale payments and high-stakes transactions between organizations and higher-level officials, often with political power or with protection from state actors with such power. Our measurement and conceptualization of public sector corruption is consistent with that of Transparency International, which estimates that in 48 countries around the world, upward of 25% of the entire population report engaging in transactions of this kind at least once a year (3. https://www.transparency.org/en/gcb/global/global-corruption-barometer-2017).
4. In this study, we follow the institutional framework, which links political trust to citizens’ evaluation of government’s performance (see Wang Citation2016 for an overview).
5. These expectations are broadly consistent with cultural theories of political trust that postulate that citizens’ evaluations of governmental performance are conditioned by national values and norms. Focusing on Asian societies, for instance, Deyong and Yang (Citation2014) explain high levels of political trust with traditional deference to authority; similarly, others attribute high levels of political trust in the Soviet Union to the legacy of submissive popular attitudes in Tsarist Russia (Jowitt Citation1998; Keenan Citation1986).
6. This argument is consistent with the work of political scientists who maintain that citizen support for authoritarian leadership represents a reaction to unfavorable structural conditions rather than distinct expectations of their leaders (Castelnovo, Popper, and Koren Citation2017; Kakkar and Sivanathan Citation2017; Emma et al. Citation2017). Others also find that citizens support autocratic leaders in the name of specific causes – such as national security, ethno-racial order, or restrictions on abortion, and do so despite, rather than because of, their core socio-political attitudes (see Harms et al. Citation2018 for an overview).
7. List experiments have received substantial attention from methodologists (Blair and Imai Citation2012; Bryn, Imai, and Shapiro Citation2016) and have been employed to study sensitive matters in both democratic and authoritarian settings (e.g. Kuklinski, Cobb, and Gilens Citation1997; Redlawsk, Tolbert, and Franko Citation2010). In such experiments, respondents are randomly divided into two groups. They receive a card with a list of statements and are instructed to indicate only how many of the statements they agree with, but not which ones. The treatment group’s list of statements is identical to that of the control group except it includes one additional statement concerning the sensitive matter. If respondents are placed into the groups randomly, the difference in the average number of statements by the control and treatment groups will reflect the number of those in the treatment group agreeing with the sensitive item. The sensitive item in our list experiment asked about whether the respondent had given a bribe or a gift or favor to a bureaucrat in one of four different spheres. Our direct questions queried separately about the same two possible actions for each of those bureaucratic spheres.
8. While we use the term “bribery” to denote our variable in the most straightforward way, our actual measure of public sector corruption includes monetary exchanges as well as exchanges of gifts and favors, covering a spectrum of remuneration types common in public sector corruption markets.
9. Because we are interested in the potential impact of experiences with corruption on political trust, we include respondents who had no institutional interaction in the last two years and therefore no experiences with corruption. However, the pattern of results is very similar when we exclude those with no institutional interaction.
10. Another benefit of the behavioral approach in our surveys is that reported encounters with corruption had occurred in two years preceding the survey, whereas the measure of trust assesses respondents’ attitudes at the time of the survey.
11. For more information on the Global Corruption Barometer, please see https://www.transparency.org/en/gcb?gclid=Cj0KCQiAxoiQBhCRARIsAPsvo-wO6O44DKd5fmydnvq_pYbp7Ct0qYQ0vdywDIvx5TECvJWDOaUMpYMaAhUdEALw_wcBm.
12. Because corruption experience was the topic of only one of three major modules in our survey (in addition to modules on citizens’ personal networks and their political attitudes and behaviors), we were limited in the number of sectors the respondents were queried about.
13. For more information on the European Quality of Government Index, please see https://www.gu.se/en/quality-government/qog-data/data-downloads/european-quality-of-government-index.
15. More than decade elapsed between the two surveys, and their different methodologies, likely, explain the difference.
16. Russian men are more likely than women to report being extorted and getting better service in exchange for corruption, older Russians report more experience with extortion, and higher income Russians are more likely to report receiving better service. These sociodemographic differences in experiencing desirable and undesirable corruption underscore the necessity of controlling for these attributes.
17. Ordered logistic regression yields the same pattern of substantive findings but does not allow for comparison of the magnitudes of coefficients in assessing our hypotheses (Mood Citation2010).
18. The differences in coefficients across the models for trust in central and local leaders were assessed using likelihood ratio tests from seemingly unrelated regression.