Pathways to Corruption: Institutional Context and Citizen Participation in Bureaucratic Corruption

Pathways to Corruption: Institutional Context and Citizen Participation in Bureaucratic Corruption Abstract Though bureaucratic corruption is widespread, social scientists have yet to develop a comprehensive model predicting ordinary people’s engagement in corrupt exchanges with street-level bureaucrats. Our article fills this gap by specifying an individual-level causal model of bureaucratic corruption centered around three theoretically derived predictors: beliefs about acceptability of corruption, its perceived riskiness, and its utility to the offender. In doing so, we develop a theory of how institutional stability affects rates and causal pathways to bureaucratic corruption. Using the data from nationally representative surveys in Russia (N = 2,000) and Ukraine (N = 1,535), we test a path-diagramed structural equation model that accounts for endogeneity and the relationships among the theoretically derived predictors of corruption. The tests of our model in institutionally more stable Russia and less stable Ukraine, combined with OLS regression models on split samples from conflict areas within each country, show that when institutions have greater stability, (1) citizens are less likely to partake in bureaucratic corruption, and (2) the effects of theoretically derived predictors on corruption behavior are stronger. We are the first to incorporate into a theory of bureaucratic corruption a feedback loop whereby engagement in corruption reinforces perceptions of its spread, which in turn contributes to the belief that bribery is acceptable. Our findings explain the variation in corruption behavior among citizens of high-corruption societies, extend the institutional theory of social control, testify to the importance of contextualizing theories of crime, and reveal the need to tailor anti-corruption policies to specific institutional environments. Bureaucratic corruption1 refers to the abuse of bureaucratic office for personal gain, whereby the employees of street-level organizations receive unsanctioned compensation for performing their job-related duties or extending additional, extralegal benefits to their clients. Such corruption is difficult to study: it is hidden, stigmatized, and often illegal. Thus, despite the negative economic consequences of corruption,2 we know little about its causes. While existing studies identify country-level predictors of corruption (e.g., Pellegrini 2008), put forth experimental tests of its select determinants (e.g., Graeff et al. 2014), or describe it using anecdotal data (e.g., Osipian 2009), to date, no study has advanced a comprehensive model of bureaucratic corruption at the level of the individual. We fill this gap by developing and testing a causal model that predicts citizens’ participation in bribery exchanges with street-level bureaucrats in high-corruption societies.3 We identify major theoretical predictors of corruption based on criminological theory, experimental work, and case studies. We then specify and show support for a structural equation model of bureaucratic corruption that considers the context of exchanges and stipulates causal relationships among determinants. In doing so, we develop a theory of how institutions influence bureaucratic corruption. We argue that institutional context is particularly relevant to low-level bribery because it often involves cooperation with virtual strangers. Inasmuch as participants lack prior knowledge of their exchange partners, they are particularly likely to draw on institutional rules and norms when deciding whether to participate in bureaucratic corruption. In contrast to other studies, however, we go beyond the analysis of how specific institutions shape people’s corruption conduct. We show that the stability of the institutional context, in and of itself, impacts actors’ causal pathways to this crime. Using new survey data from Russia and Ukraine, we conduct cross-national and within-country statistical tests of the determinants of bureaucratic corruption. As two high-corruption societies4 with markedly different institutional contexts, Russia and Ukraine provide valuable cases for our study. Previous research shows that bureaucratic corruption is commonplace in a variety of organizations in both countries, including hospitals and clinics (Rivkin-Fish 2005), universities and secondary schools (Osipian 2009), and police control posts (Gerber and Mendelson 2008). At the same time, Russia’s institutions are more stable than their Ukrainian counterparts. Both countries also have internal variation in institutional stability, associated with regionally concentrated military conflicts. The estimation of our model reveals that when institutions are more stable, (1) citizens are less likely to partake in bureaucratic corruption, and (2) the predictors of corrupt behavior, such as beliefs and utility assessments, have stronger effects. These findings have important implications for sociological theories of crime and for anti-corruption policy: they explain the variation in corrupt behavior among citizens of high-corruption societies, extend the social control theory of crime, and testify to the importance of contextualizing the theories of law-breaking. Main Predictors of Citizen Engagement in Bureaucratic Corruption The relevant literature suggests three predictors of corrupt behavior: beliefs about the acceptability of corruption, perceived riskiness of corrupt behavior, and its utility to the offender. First, sociological theories of crime often feature beliefs and attitudes as prominent predictors. According to Sutherland’s (1947) differential association theory, for instance, people commit crime because they are exposed to more definitions that are favorable to the violation of law than viewpoints that are unfavorable to crime. Social learning theories (Akers 2011) further suggest that acquired cultural schema translate into concrete behaviors, which are then sustained through positive reinforcement. Empirical studies confirm that individuals are more likely to break the law when exposed to crime-favorable definitions (Apel and Paternoster 2009; Jackall 1988). Correspondingly, popular attitudes are central to many explanations of corruption. For instance, formal theorists maintain that past experiences form beliefs about the economic environment, which in turn contribute to a culture of corruption (Sah 1988). Case studies of the former Soviet Union also suggest that normative systems accepting of corruption formed during the Soviet era, when the planned economy was characterized by an overbearing bureaucratic apparatus and shortages of consumer goods. In this context, relationship-based transactions known as blat emerged as widely accepted means for procuring goods and services that offered ordinary citizens a personalized, flexible, and effective alternative to rigid and politicized institutions. Blat was framed in terms of mutual assistance and support in the face of adversity; as a community-reliant and community-fostering mode of sustenance rather than a subversion of the public good by individuals (Ledeneva 1998; Rehn and Taalas 2004). In the economic dislocations of the post-communist era, popular justifications for bureaucratic corruption persisted (Karklins 2005; Polese 2008). Surveys confirm that citizens are more likely to engage in bureaucratic corruption when they do not define it as wrong (Tavits 2010). We therefore expect that people’s beliefs affect their corruption behavior: Hypothesis 1. The more acceptable citizens think that bribery and gift-giving are, the more likely they are to engage in bureaucratic corruption. Our second predictor of bureaucratic corruption, perceived risk of punishment, comes from the work of criminologists. Crime deterrence strategies are usually based on manipulating the scale, certainty, and speed of punishment (Cohen and Felson 1979; Kleck et al. 2005). Scholars of white-collar crime and corruption also acknowledge the impact of “criminal opportunity” on prevalence of law-breaking (Benson and Simpson 2015). Along these lines, Sajó (2003) links post-Soviet bribery to ineffective and politicized policing in the region and Karklins (2002) attributes corruption to the lack of oversight and transparency. Our second hypothesis reflects this school of thought in criminological research: Hypothesis 2. The lower citizens’ perceived risks of detection and punishment, the more likely they are to engage in bureaucratic corruption. The third major predictor of corruption stems from citizens’ rational calculus. When trying to obtain desired services, citizens weigh the costs and benefits of different strategies. Previous research shows that decisions to bribe a bureaucrat reflect citizens’ evaluations of how much they need a specific service, whether there are other ways to get it, and what a bureaucrat expects in return (McMann 2014; Polese 2008). We assess the effect of these determinants on citizens’ behavior in two corollary hypotheses. The first corollary hypothesis considers the role of extortion. Studies suggest that citizens of high-corruption societies often experience requests for bribes5 from employees of service-provision organizations (Grødeland, Koshechkina, and Miller 1998; Karklins 2002). Faced with such requests, clients may conclude that bribery is the only way to receive the desired service; they may also feel too uncomfortable or intimidated to refuse. In fact, Tavits (2010) finds that officials’ requests of bribes are the strongest predictor of having paid a bribe. Thus, we expect that such requests—whether explicit or deduced from hints, reticence, or delays in service—shape citizens’ behavior: Hypothesis 3a. When officials request bribes, citizens are more likely to engage in bureaucratic corruption. Yet, existing studies reveal that requests from officials cannot explain all bureaucratic corruption. Polese (2008) argues that Ukrainians often choose to engage in under-the-table transactions with bureaucrats even when bribes are not explicitly requested. Karklins (2002) writes that post-communist citizens often initiate corrupt exchanges with traffic cops, Rivkin-Fish (2005) documents voluntary bribery payments to doctors, and Werner (2000) mentions that bribes are offered in exchange for forged military documents. In Werner’s (2000, 11) words, bribery “can be useful for ‘greasing the machine’s wheels’ or speeding things up.” To capture the value that corrupt transactions hold for citizens distinct from officials’ requests, we develop a notion of utility of bureaucratic corruption and incorporate it into our second corollary hypothesis. This notion stems from another classic theory in sociological criminology, Anomie Theory (Merton 1968), whereby crime offers a way to achieve socially defined goals in the absence of legitimate avenues of achievement. Crime can facilitate economic redistribution (Gambetta 1996), perform employment functions (Deuchar 2009), and provide welfare (Wacquant 2008). Some also argue that bureaucratic corruption is functional in underdeveloped societies because it allows ineffective organizations to deliver services to citizens (Becquart-Leclercq 1989; Nye 1967). Case studies reveal that such corruption generates private markets, allowing citizens to satisfy their consumption goals in the absence of viable legal alternatives (McMann 2014; Smith 2010; Yang 1994). Economic rationality asserts that behavior emerges from a cost-benefit calculation, which means that the value of the service received in exchange for a bribe is evaluated vis-à-vis its expense. Existing research, however, focuses primarily on the benefits of illicit exchanges, without reference to their costs. Our study is the first to propose that an appropriate predictor of corrupt behavior is its utility, understood as citizens’ perceived need to resort to corruption relative to their financial capabilities. Hypothesis 3b. As the utility of the transaction increases, the likelihood of engaging in bureaucratic corruption also increases. Hypothesized Relationships among Key Concepts A theoretically sound model of bureaucratic corruption must account for interactions among its key predictors and acknowledge the cyclicality of the corruption process (Andvig and Moene 1990). In this section, we develop additional hypotheses about inter-relations among people’s ideas about acceptability, utility, riskiness, and perceptions of corruption in society. To date, individual-level studies of bureaucratic corruption have not fully specified nor employed the statistical models that accommodate the interactions among its predictors. The few existing large-N analyses use regression models that assume independence among predictor variables and do not account for feedback loops (Miller, Grødeland, and Koshechkina 2001; Tavits 2010). Our theory and analyses represent the first-ever specification and test of non-recursive structural equation models of bureaucratic corruption. Our first hypothesized interaction emerges from the recognition that attitudes shape people’s cost-and-benefit calculations in regards to rule-breaking (Vaughan 1999). Ashforth and Anand (2003) show that beliefs that corruption is normal go hand-in-hand with claims that corruption is necessary to achieve goals. The authors also argue that “once corruption is normative, it may accrue symbolic rewards, such as status and self-esteem, in addition to the utilitarian rewards” (2003, 13). Beliefs about acceptability of corruption may also reduce its perceived costs. Since breaking social norms may be difficult and uncomfortable (Posner 1997), the costs of bribery may be lower for people who believe that corruption is normal. Our next hypothesis reflects the expectation of a positive relationship between perceived acceptability and utility of corruption: Hypothesis 4a. The more acceptable citizens think that bribery and gift-giving are, the higher utility they attribute to bureaucratic corruption. Moreover, scholars argue that corruption is a collective action problem: the behavior of others impacts citizens’ own decisions to partake in corruption (Mungiu-Pippidi 2013; Persson, Rothstein, and Teorell 2013). We build on these arguments by specifying the pathways whereby perceptions lead to bribery behavior. We expect perceptions of corruption in society at large to influence individual bribery behavior in part by how people assess its risks. When people perceive that others are giving and accepting bribes with impunity, they are more likely to conclude that they would not be punished should they also do so (Čábelková and Hanousek 2004). Formal theorists exploring the concept of multiple equilibria in the incidence of corruption also demonstrate that as perceived incidence of corruption in society increases, the probability of getting caught in a corrupt transaction decreases (Andvig and Moene 1990). The idea that people gauge the riskiness of a behavior from observing others is, in fact, central to general deterrence theory, widely accepted in criminology (Homel 2012; Sunshine and Tyler 2003). Hypothesis 4b. The more corruption citizens perceive in society, the lower their perceived risks of detection and punishment. Acknowledging the cyclicality of the corruption process, we expect that perceptions of corruption in society are not exogenous, but, rather, depend on citizens’ own experiences. Reisinger, Zaloznaya, and Claypool (2017) report that when citizens encounter corruption in an interaction with a low-level bureaucrat, they are likely to infer the corruptibility of other officials. Our next hypothesis, therefore, accounts for a feedback loop that links corrupt behavior and its determinants to perceptions of corruption in society. This expectation raises the need to test our theory with a non-recursive structural equation model that allows for causation to flow in more than one direction. Hypothesis 4c. The more citizens engage in bribery and gift-giving, the more corruption they perceive in society. Institutional Stability and Bureaucratic Corruption Criminologists have long recognized that context shapes crime. For instance, neighborhood characteristics, such as “poverty, residential mobility, ethnic heterogeneity, and weak social networks,” have been shown to increase street crime (Kubrin and Weitzer 2003, 374; Rosenfeld, Baumer, and Messner 2001). In this article, we focus on institutions as a salient contextual determinant of causal pathways to bureaucratic corruption. We define institutions as systems of rules or rule-like prescriptions that exercise normative and instrumental pressure toward compliance and that are shared and embedded in existing social structures (North 1990; Ostrom and Basurto 2011). Social control theory of crime underscores the importance of institutional context (Hirschi 1969; Kornhauser 1978). According to this theory, when bonds that link individuals to other people and institutions weaken, individuals are more likely to break the law. Criminologists have demonstrated empirically that attachment to institutions—for instance, through work and marriage—inhibits the likelihood of criminal behavior among adults (Triplett, Gainey, and Sun 2003; Wiatrowski, Griswold, and Roberts 1981). We believe that institutional context is especially pertinent to bureaucratic corruption because such corruption rests on cooperation between social actors who have limited prior experience with each other. Since clients and bureaucrats usually do not have regularized relationships outside service-provision organizations, they are unlikely to share trust and reciprocity. Socio-legal scholars argue that such social distance increases actors’ reliance on institutionalized norms in their interactions with each other (Merry 1993; Ellickson 1986). Social exchange theory (SET) also suggests that people are likely to deduce the rules of exchange from the “normative definition of the situation” in contexts of uncertainty and risk, when wrong choices are especially costly (Emerson 1976, 351). In corrupt transactions, the risk of non-reciprocity is high because the mechanisms for ensuring that the other party delivers on the bargain, such as courts, are unavailable, and because one of the parties might report the exchange to authorities, triggering sanctions that range from stigma to criminal punishment. To mitigate these risks, social actors are particularly likely to invoke institutionalized norms. While others have studied how particular institutions, such as corruption-conducive norms (Köbis et al. 2015) or anti-corruption laws (Peisakhin and Pinto 2010), affect the prevalence of corruption, we argue that the stability of institutions, in and of itself, affects the rates of bureaucratic corruption and shapes how individuals come to participate in this crime. We conceptualize the notion of institutional stability as an ideal type, which anchors one end of what we see as a continuous variable. Such stability is associated with absence of frequent, erratic, or inconsistent social change. Importantly, when institutions are stable, actors are able to know what the behavioral norms are in a particular situation at a specific point in time. When norms are “knowable,” actors are able to predict, with a degree of confidence, the likely consequences of different courses of action. By contrast, when institutions are unstable, rules change frequently, rapidly, and erratically, making it harder for citizens to anticipate what will be considered right and wrong at any point in time. The knowability of “the rules of the game” does not always depend on their correspondence to formal laws and regulations. Sometimes, behavioral norms do reflect codified rules, which may make their knowability more straightforward. However, when “rules of the game” do diverge from formal regulations, institutional stability enables actors to infer behavioral expectations with a degree of certainty. Actors make such inferences from their own past experiences or from past experiences of others, discovered through conversations, hearsay, and observation. The less stable the institutional environment, the lower the predictive power of past experiences for the future behavior. It is important to note that institutional stability does not imply the immutability of behavioral norms, as institutions evolve and change even in the most stable of contexts. For example, law and society scholars show that law is constantly “in action”—evolving through contestation by social actors (Shamir 2004), interaction with other laws and customs (Merry 1988), and appropriation by organizations (Edelman 1992). When institutions are stable, however, this ongoing evolution does not undermine the knowability of rules as citizens develop reliable expectations about the pace and the direction of institutional change. In less stable settings, where social change is more abrupt and irregular, actors have difficulty predicting just how quickly, and in what way, the rules of the game may change. Also, in stable institutional settings, the rules that govern behavior have knowable limits of pertinence relative to other, co-extant rules. In any environment, a variety of rule-like prescriptions governs social action: for instance, different laws, religious postulates, and family norms all regulate one’s conduct at a family celebration of a religious holiday. Some norms might even conflict with one another; yet, with greater institutional stability, people better understand how far each rule extends. The degree of institutional stability has important implications for citizen behavior in bureaucracies. When it is high, citizens are confident about what needs to be done to obtain services, and what distinguishes appropriate from inappropriate conduct in bureaucratic organizations. In stable contexts, citizens have reliable expectations regarding the behavior of their exchange partners; for example, the probability that a bribe is expected and the likelihood of punishment in case of detection. In contrast, when institutional stability is low, the rules are variant and ambiguous, and controls on organizational behavior function unpredictably. Formal theorists link institutional stability to political stability and to fewer corrupt bureaucrats: the “higher the probability of a regime shift…the higher is the incidence of corruption” (Andvig and Moene 1990, 68). Increases in political uncertainty, which are associated with lower probabilities of getting caught for corrupt transactions, cause bureaucrats to become more prone to corruption (71). Additionally, weak destabilized governments are less able to prevent the establishment of independent corruption rackets (Bardhan 1997, 1325). We therefore hypothesize that the overall prevalence of corruption is lower in stable environments because the probabilities of getting caught are lower and because ordinary citizens have fewer incentives to resort to illicit strategies. Also, in contexts with institutional stability, law enforcement agencies and accountability systems prosecute law-breaking more consistently. Hypothesis 5. The more stable the institutional context, the lower the levels of citizen engagement in bureaucratic corruption. Even though overall levels of corrupt behavior are lower when institutions are more stable, we expect stronger causal effects from bribery predictors in stable environs. Stability allows citizens to anticipate how organizations will function and, therefore, rely more fully on beliefs, utility calculations, and risk assessments in making corruption-related decisions. In contrast, in less stable systems, the uncertainty about “how things work” makes it difficult to assess the risks and utility of rule-breaking. Hypothesis 6. The magnitude of the effects that acceptability beliefs, assessments of risk, and utility calculations have on bureaucratic corruption is greater when institutions are more stable. Figure 1 offers a graphic representation of our theoretically derived model of corruption behavior.6 Figure 1. View largeDownload slide Theoretical model of corruption behavior with utility Figure 1. View largeDownload slide Theoretical model of corruption behavior with utility Variation in Institutional Stability and Selection of Cases In comparing Russia and Ukraine, we explore how bureaucratic corruption is influenced by institutional stability at both the national and subnational levels. As indicated above, institutional stability is tied to political stability. One type of political instability that is particularly relevant to our cases is regime change,7 which is often accompanied by erratic and partial reforms, coexistence of multiple regulatory systems, and resource-deficiency, all of which undermine the ability of actors to anticipate the “rules of the game” (Spicer, McDermott, and Kogut 2000). In contrast, when a polity is more stable, “members of society restrict themselves to the behavior patterns that fall within the limits imposed by…expectations” (Ake 1975, 273). Russia, in the post-Soviet era, has been more stable, politically and institutionally, than Ukraine, which has undergone multiple regime changes. Russia has had essentially the same government between 2000 and 2017 (Geddes, Wright, and Frantz 2014). After President Putin’s ascent to power at the turn of the century, he retained a heavily controlled and steady system of governance, avoided risky overhauls, and contained destabilizing reforms through state capture, intimidation of the opposition, and extensive ideological propaganda. This political stability enhances ordinary Russians’ knowledge of the behavioral norms, their boundaries, and the likely pace and direction of their change (Forrat 2015; Gel’man 2015). Although higher than in Ukraine, Russia’s institutional stability is far from absolute. The knowability of the “rules of the game” in Russia is partially undermined by the “duality” of its institutions (Hendley 2017). Russia’s “formal constitutional order” coexists with “a second world of informal relations, factional conflict, and para-constitutional political practices” (Sakwa 2010, 185). This duality, however, is in itself knowable: failures of law, most often, happen in cases that involve political or economic elites and other influential actors (Gel’man 2004; Ledeneva 2008). Ordinary Russians, in contrast, can usually predict what is expected of them in street-level bureaucracies. Unlike Russia, Ukraine has oscillated between distinct forms of political governance since acquiring independence from the USSR. It experienced a pro-democratic “Orange Revolution” in 2004, which brought Viktor Yushchenko to presidency, but turned back to more autocratic governance under Viktor Yanukovych (Kuzio 2015). In 2013, a new wave of popular protests brought to power a pro-Western administration of Petro Poroshenko. This volatility of Ukraine’s government reduced its effectiveness in monitoring bureaucracies and allowed for more deviations from the agreed-upon patterns of behavior. Bardhan (1997, 1325) argues that weak central government creates a kind of economic warlordism, where different ministries, agencies, and levels of local government all set their own norms of exchange. Given this institutional instability, Ukrainians have a harder time predicting—and agreeing on their predictions of—the rules of behavior in bureaucracies. Scholars have described the institutional landscape of Ukraine as ambivalent (Riabchuk 2007), its population as having a divided mentality (Fimyar 2010), and its culture as “an unstable mixture of…ideologies” (Zlobina 2007, 96). In sum, despite their geographic and cultural proximity and centuries of intertwined history, since the breakup of the Soviet Union, Russia and Ukraine have pursued different political trajectories with distinct consequences for their institutional contexts. The two countries, therefore, offer good cases for a “paired comparison” of the effects of institutional stability on bribery behavior. Paired comparison provides the analytical leverage of similarity and contrast and “allows analysts to use differences in institutional form as a variable to demonstrate the sources of intrasystemic behaviors” (Tarrow 2010). No matter how well paired, however, no two case studies can produce fully generalizable conclusions. To achieve additional analytical power, we complement our cross-country comparison with within-country analyses. Both Russia and Ukraine are marked by significant regional heterogeneity. On the subnational level, each country has regions that are afflicted with military conflict that results in institutional instability. Military conflicts increase out-migration, create chaos, and detract resources from public infrastructure—all of which upset institutional stability (Bellows and Miguel 2006). Four distinct conflicts have destabilized select Russian and Ukrainian regions. First, fueled by ethnic and territorial disputes, frozen conflicts have developed in Russia’s North Caucasus region, most notably in Chechnya (Hesli 2007). Second, confrontations with Russia’s neighbors, such as the Republic of Georgia, have contributed to endemic instability in the adjacent regions (Cornell and Starr 2009). Third, Russia’s annexation of the Crimean Peninsula in March 2014 affected regions in both Russia and Ukraine. The prolonged military presence and the emergency state in the peninsula have undermined the institutional stability in neighboring areas (Birnbaum and Demirjian 2015). Fourth, regions in both countries suffer from the military conflict with Russia-backed separatists in Eastern Ukraine. In our analysis, we link these regionally concentrated conflicts and the associated subnational variation in institutional instability to rates of and pathways to corrupt behavior. By comparing more and less institutionally stable regions within Russia and Ukraine, as well as comparing the two countries with each other, we increase the number of comparison cases, and raise the generalizability of our findings. Data and Measures Our data come from representative national surveys conducted in Russia and Ukraine in June–July 2015. Face-to-face interviews were conducted with 2,000 respondents in Russia and 1,535 in Ukraine (see appendix A for more detail on the surveys). To measure bureaucratic corruption, we started with a query as to whether the respondent had contact with any of the following officials in the past 12 months: Officials of the judicial system (judges, clerks, justice department officials, lawyers, prosecutors) Doctors, nurses, medical workers, hospital administrators Inspectors (in health, construction, food quality, sanitary control, and licensing) Officials issuing certificates and permits (marriage, death, birth certificates; construction permits) Officials in housing and communal services Officials who issue governmental tenders Politicians at the local level (mayors, local executive heads) Professors, instructors in higher educational institutions Police officers Tax officials Teachers, school administrators Traffic police Other (respondent names the official) We then asked to which of these officials our respondents had given a bribe, a present, or done a favor in exchange for a service. In Russia, 83 percent of respondents had contact with at least one of these officials, and 18 percent of those who had contact paid a bribe (gave a gift or did a favor). In Ukraine, 77 percent had contact with at least one official, and 29 percent of those paid a bribe, made a gift, or did a favor.8 Each individual’s Corruption Behavior is the ratio of the number of bribes paid (as well as gifts given or favors made) to the number of contacts with officials. The ratio ranges from 1 (every contact resulted in a bribe) to 0.1 (no bribes),9 with a mean score of 0.37 for Russia (on average, over a third of contacts in Russia result in corrupt behavior) and 0.51 for Ukraine (on average, half of all contacts with officials in Ukraine result in corruption). Appendix B contains descriptive information on our measures. To measure beliefs, we asked whether it is acceptable to do any of the following to receive a faster or higher-quality service from officials: To give money To give a gift To do a favor Responses are coded from 1, never acceptable, to 3, always acceptable. We average responses to each behavior to create a scale of Belief in Acceptability of Bribery. In Ukraine, the average score on this variable is 1.6; in Russia, it is 1.7. Each person was also asked about the Spread of Corruption throughout the country. Specifically, respondents were asked whether over the past three years the level of corruption had increased a lot, increased a little, stayed the same, decreased a little, or decreased a lot. We coded responses from 1 to 5, with higher scores associated with increased corruption. The average score on this array is 3.3 for Russia and 3.9 for Ukraine. To measure risk of detection and punishment, we asked a series of questions on how likely the respondent thought people were to get into legal trouble for offering bribes or gifts to employees of Hospitals and clinics, Traffic police, Registration and tender offices, Universities, and Secondary schools. Responses range from 1, very likely, to 4, very unlikely. We average evaluations of these five bureaucracies to create a summary scale of each respondent’s perception of the Risk of Punishment if one engages in bureaucratic corruption. In Russia, the average score on this measure is 2.9; in Ukraine, it is 3.3. To measure the utility of bureaucratic corruption, we combine perceptions of the need to pay bribes with an evaluation of personal ability to finance them. An individual’s perception of the need to engage in bureaucratic corruption is determined by responses to a question about why ordinary people pay bribes to officials. The following options were offered: Because it is the easiest or fastest way to achieve the results Because bureaucrats demand bribes from people Because it is normal; everybody around them does it Because people feel grateful to the official Because they feel sorry for the official (they want to help him/her) We considered answers 1 or 2 as indicators of perceived need for corrupt conduct. These responses were aggregated depending on whether these answers were selected as the most important, the second most important, or the least important reason. We then measured respondents’ capacities to pay bribes based on their financial statuses. Thus, Utility of Bribery measures perception of need weighted by financial capacity. Our measure of utility ranges from 0.0 to 7.5, with averages of 3.5 in Russia and 2.4 in Ukraine. To indicate regionally concentrated institutional instability, we created a dummy variable for whether or not the respondent resides in or near a region marked by any of the military conflicts described above. Nineteen percent of the Russia sample and 39 percent of the Ukraine sample reside in a Conflict Area. To measure officials’ requests of bribes, a battery of questions asked whether each official with whom the respondent had contact had asked for a bribe, a present, or a favor in exchange for services. Official Requested a Bribe for each individual is the ratio of the number of times an official asked for a bribe to the number of contacts with officials he or she had. The average scores in Russia and Ukraine are 0.38 and 0.49, respectively, indicating a greater likelihood in Ukraine than in Russia of being asked by an official for a bribe. Also included in the analysis are demographic predictors that have been previously linked to differences in behavior: age, gender, education, and rural residence (Miller, Grødeland, and Koshechkina 2001; O’Brien and Wegren 2002; Tavits 2010; Verba and Nie 1972). Our structural equation model links demographics to beliefs about acceptability and to utility, as a vast literature evidences differences in attitudes based upon these characteristics (Hesli et al. 2001). Analyses and Results We start by testing the impact of institutional stability on engagement in bureaucratic corruption (H5). We argue that Russia will have lower levels of corrupt behavior than Ukraine because Russia has more institutional stability. To test, we calculate the average likelihood of engagement in bureaucratic corruption (given contact with an official) in Russia and Ukraine. The first columns of table 1 (and the corresponding t-test value for comparing the two means) reveal that engagement in bureaucratic corruption occurs at a higher rate in Ukraine (0.505) than in Russia (0.371). Importantly, in both countries, living in or adjacent to a conflict area significantly increases one’s likelihood of engaging in corruption (last four columns of table 1). Since conflict areas have more unstable institutions, this finding provides preliminary support for hypothesis 5 that in more stable institutional contexts levels of bribery behavior are lower. Table 1. Difference of Means Test for Corruption Behavior by Institutional Context All of Ukraine All of Russia Russia Ukraine Conflict areas Non-conflict areas Conflict areas Non-conflict areas Average level of corrupt behavior (means) 0.505 0.371 0.415 0.361 0.522 0.495 Number of cases 1,180 1,665 316 1,349 455 725 T value for comparing the two means 16.5123 4.6923 1.9686 All of Ukraine All of Russia Russia Ukraine Conflict areas Non-conflict areas Conflict areas Non-conflict areas Average level of corrupt behavior (means) 0.505 0.371 0.415 0.361 0.522 0.495 Number of cases 1,180 1,665 316 1,349 455 725 T value for comparing the two means 16.5123 4.6923 1.9686 Table 1. Difference of Means Test for Corruption Behavior by Institutional Context All of Ukraine All of Russia Russia Ukraine Conflict areas Non-conflict areas Conflict areas Non-conflict areas Average level of corrupt behavior (means) 0.505 0.371 0.415 0.361 0.522 0.495 Number of cases 1,180 1,665 316 1,349 455 725 T value for comparing the two means 16.5123 4.6923 1.9686 All of Ukraine All of Russia Russia Ukraine Conflict areas Non-conflict areas Conflict areas Non-conflict areas Average level of corrupt behavior (means) 0.505 0.371 0.415 0.361 0.522 0.495 Number of cases 1,180 1,665 316 1,349 455 725 T value for comparing the two means 16.5123 4.6923 1.9686 Next, we evaluate the proposition that the magnitudes of the effects of acceptability beliefs, assessments of risk, and utility calculations on bureaucratic corruption are greater when institutions are more stable (H6). To test this proposition, we employ ordinary least squares (OLS) regression on split samples in Russia and Ukraine (see table 2). By conducting our analyses separately in Russia and Ukraine, we see how national context affects predictors of bureaucratic corruption. By analyzing independently, and comparing, conflict and non-conflict areas in each country, we also assess the effect of regional instability on corruption determinants.10 We present two models in this table; the first (model 1) includes Utility as a predictor, and the second (model 2) contains Official Requests Bribe. Table 2. OLS Regression of Corruption Behavior upon Theoretical Predictors Dependent variable: Corruption behavior Model 1 (with utility) Model 2 (with official requested bribe) Russia Ukraine Russia Ukraine Predictor variables Non-conflict area Conflict area Non-conflict area Conflict zone area Non-conflict area Conflict area Non-conflict area Conflict area Belief in acceptability 0.045*** 0.071*** 0.040*** 0.002 0.025*** 0.028* 0.038*** 0.019 (0.009) (0.024) (0.015) (0.021) (0.006) (0.015) (0.011) (0.014) Punishment risk low 0.029*** 0.051*** 0.044*** 0.007 0.007 0.016 0.002 −0.008 (0.007) (0.018) (0.013) (0.016) (0.005) (0.012) (0.010) (0.011) Utility 0.008*** 0.001 −0.0001 0.016 (0.003) (0.008) (0.007) (0.010) Spread of corruption 0.012** 0.034*** 0.033*** 0.008 −0.002 0.007 0.020*** 0.004 (0.005) (0.012) (0.008) (0.011) (0.003) (0.007) (0.006) (0.007) Rural −0.026** −0.055** 0.041** 0.094*** −0.002 −0.028* 0.026** 0.063*** (0.012) (0.028) (0.018) (0.033) (0.008) (0.016) (0.013) (0.022) Age 0.001*** 0.0003 0.001 0.001 0.0004** −0.001* 0.0003 −0.0001 (0.0003) (0.001) (0.001) (0.001) (0.0002) (0.001) (0.0004) (0.0005) Female −0.027*** 0.006 0.012 −0.006 −0.003 0.004 0.025** 0.013 (0.009) (0.025) (0.018) (0.025) (0.006) (0.015) (0.013) (0.017) Education −0.011*** −0.0003 −0.0002 −0.005 −0.003 0.003 0.001 0.001 (0.003) (0.007) (0.005) (0.006) (0.002) (0.004) (0.004) (0.004) Official requested bribe 0.672*** 0.813*** 0.732*** 0.730*** (0.017) (0.037) (0.031) (0.033) Constant 0.180*** 0.020 0.112 0.404*** 0.047* −0.0004 −0.036 0.102 (0.037) (0.109) (0.078) (0.093) (0.025) (0.066) (0.057) (0.064) Observations 1307 312 715 435 1347 316 723 455 R2 0.074 0.105 0.054 0.036 0.558 0.648 0.475 0.535 Adjusted R2 0.068 0.081 0.043 0.018 0.556 0.639 0.470 0.527 Residual std. error (df) 0.167 0.214 0.220 0.239 0.115 0.134 0.164 0.165 (1298) (303) (706) (426) (1338) (307) (714) (446) F statistic (df) 12.971*** 4.445*** 5.010*** 1.979** 211.501*** 70.566*** 80.880*** 64.169*** (8; 1298) (8; 303) (8; 706) (8; 426) (8; 1338) (8; 307) (8; 714) (8; 446) Dependent variable: Corruption behavior Model 1 (with utility) Model 2 (with official requested bribe) Russia Ukraine Russia Ukraine Predictor variables Non-conflict area Conflict area Non-conflict area Conflict zone area Non-conflict area Conflict area Non-conflict area Conflict area Belief in acceptability 0.045*** 0.071*** 0.040*** 0.002 0.025*** 0.028* 0.038*** 0.019 (0.009) (0.024) (0.015) (0.021) (0.006) (0.015) (0.011) (0.014) Punishment risk low 0.029*** 0.051*** 0.044*** 0.007 0.007 0.016 0.002 −0.008 (0.007) (0.018) (0.013) (0.016) (0.005) (0.012) (0.010) (0.011) Utility 0.008*** 0.001 −0.0001 0.016 (0.003) (0.008) (0.007) (0.010) Spread of corruption 0.012** 0.034*** 0.033*** 0.008 −0.002 0.007 0.020*** 0.004 (0.005) (0.012) (0.008) (0.011) (0.003) (0.007) (0.006) (0.007) Rural −0.026** −0.055** 0.041** 0.094*** −0.002 −0.028* 0.026** 0.063*** (0.012) (0.028) (0.018) (0.033) (0.008) (0.016) (0.013) (0.022) Age 0.001*** 0.0003 0.001 0.001 0.0004** −0.001* 0.0003 −0.0001 (0.0003) (0.001) (0.001) (0.001) (0.0002) (0.001) (0.0004) (0.0005) Female −0.027*** 0.006 0.012 −0.006 −0.003 0.004 0.025** 0.013 (0.009) (0.025) (0.018) (0.025) (0.006) (0.015) (0.013) (0.017) Education −0.011*** −0.0003 −0.0002 −0.005 −0.003 0.003 0.001 0.001 (0.003) (0.007) (0.005) (0.006) (0.002) (0.004) (0.004) (0.004) Official requested bribe 0.672*** 0.813*** 0.732*** 0.730*** (0.017) (0.037) (0.031) (0.033) Constant 0.180*** 0.020 0.112 0.404*** 0.047* −0.0004 −0.036 0.102 (0.037) (0.109) (0.078) (0.093) (0.025) (0.066) (0.057) (0.064) Observations 1307 312 715 435 1347 316 723 455 R2 0.074 0.105 0.054 0.036 0.558 0.648 0.475 0.535 Adjusted R2 0.068 0.081 0.043 0.018 0.556 0.639 0.470 0.527 Residual std. error (df) 0.167 0.214 0.220 0.239 0.115 0.134 0.164 0.165 (1298) (303) (706) (426) (1338) (307) (714) (446) F statistic (df) 12.971*** 4.445*** 5.010*** 1.979** 211.501*** 70.566*** 80.880*** 64.169*** (8; 1298) (8; 303) (8; 706) (8; 426) (8; 1338) (8; 307) (8; 714) (8; 446) Note: Cells show estimated unstandardized coefficients with standard errors in parentheses. *p < 0.1 **p < 0.05 ***p < 0.01 Table 2. OLS Regression of Corruption Behavior upon Theoretical Predictors Dependent variable: Corruption behavior Model 1 (with utility) Model 2 (with official requested bribe) Russia Ukraine Russia Ukraine Predictor variables Non-conflict area Conflict area Non-conflict area Conflict zone area Non-conflict area Conflict area Non-conflict area Conflict area Belief in acceptability 0.045*** 0.071*** 0.040*** 0.002 0.025*** 0.028* 0.038*** 0.019 (0.009) (0.024) (0.015) (0.021) (0.006) (0.015) (0.011) (0.014) Punishment risk low 0.029*** 0.051*** 0.044*** 0.007 0.007 0.016 0.002 −0.008 (0.007) (0.018) (0.013) (0.016) (0.005) (0.012) (0.010) (0.011) Utility 0.008*** 0.001 −0.0001 0.016 (0.003) (0.008) (0.007) (0.010) Spread of corruption 0.012** 0.034*** 0.033*** 0.008 −0.002 0.007 0.020*** 0.004 (0.005) (0.012) (0.008) (0.011) (0.003) (0.007) (0.006) (0.007) Rural −0.026** −0.055** 0.041** 0.094*** −0.002 −0.028* 0.026** 0.063*** (0.012) (0.028) (0.018) (0.033) (0.008) (0.016) (0.013) (0.022) Age 0.001*** 0.0003 0.001 0.001 0.0004** −0.001* 0.0003 −0.0001 (0.0003) (0.001) (0.001) (0.001) (0.0002) (0.001) (0.0004) (0.0005) Female −0.027*** 0.006 0.012 −0.006 −0.003 0.004 0.025** 0.013 (0.009) (0.025) (0.018) (0.025) (0.006) (0.015) (0.013) (0.017) Education −0.011*** −0.0003 −0.0002 −0.005 −0.003 0.003 0.001 0.001 (0.003) (0.007) (0.005) (0.006) (0.002) (0.004) (0.004) (0.004) Official requested bribe 0.672*** 0.813*** 0.732*** 0.730*** (0.017) (0.037) (0.031) (0.033) Constant 0.180*** 0.020 0.112 0.404*** 0.047* −0.0004 −0.036 0.102 (0.037) (0.109) (0.078) (0.093) (0.025) (0.066) (0.057) (0.064) Observations 1307 312 715 435 1347 316 723 455 R2 0.074 0.105 0.054 0.036 0.558 0.648 0.475 0.535 Adjusted R2 0.068 0.081 0.043 0.018 0.556 0.639 0.470 0.527 Residual std. error (df) 0.167 0.214 0.220 0.239 0.115 0.134 0.164 0.165 (1298) (303) (706) (426) (1338) (307) (714) (446) F statistic (df) 12.971*** 4.445*** 5.010*** 1.979** 211.501*** 70.566*** 80.880*** 64.169*** (8; 1298) (8; 303) (8; 706) (8; 426) (8; 1338) (8; 307) (8; 714) (8; 446) Dependent variable: Corruption behavior Model 1 (with utility) Model 2 (with official requested bribe) Russia Ukraine Russia Ukraine Predictor variables Non-conflict area Conflict area Non-conflict area Conflict zone area Non-conflict area Conflict area Non-conflict area Conflict area Belief in acceptability 0.045*** 0.071*** 0.040*** 0.002 0.025*** 0.028* 0.038*** 0.019 (0.009) (0.024) (0.015) (0.021) (0.006) (0.015) (0.011) (0.014) Punishment risk low 0.029*** 0.051*** 0.044*** 0.007 0.007 0.016 0.002 −0.008 (0.007) (0.018) (0.013) (0.016) (0.005) (0.012) (0.010) (0.011) Utility 0.008*** 0.001 −0.0001 0.016 (0.003) (0.008) (0.007) (0.010) Spread of corruption 0.012** 0.034*** 0.033*** 0.008 −0.002 0.007 0.020*** 0.004 (0.005) (0.012) (0.008) (0.011) (0.003) (0.007) (0.006) (0.007) Rural −0.026** −0.055** 0.041** 0.094*** −0.002 −0.028* 0.026** 0.063*** (0.012) (0.028) (0.018) (0.033) (0.008) (0.016) (0.013) (0.022) Age 0.001*** 0.0003 0.001 0.001 0.0004** −0.001* 0.0003 −0.0001 (0.0003) (0.001) (0.001) (0.001) (0.0002) (0.001) (0.0004) (0.0005) Female −0.027*** 0.006 0.012 −0.006 −0.003 0.004 0.025** 0.013 (0.009) (0.025) (0.018) (0.025) (0.006) (0.015) (0.013) (0.017) Education −0.011*** −0.0003 −0.0002 −0.005 −0.003 0.003 0.001 0.001 (0.003) (0.007) (0.005) (0.006) (0.002) (0.004) (0.004) (0.004) Official requested bribe 0.672*** 0.813*** 0.732*** 0.730*** (0.017) (0.037) (0.031) (0.033) Constant 0.180*** 0.020 0.112 0.404*** 0.047* −0.0004 −0.036 0.102 (0.037) (0.109) (0.078) (0.093) (0.025) (0.066) (0.057) (0.064) Observations 1307 312 715 435 1347 316 723 455 R2 0.074 0.105 0.054 0.036 0.558 0.648 0.475 0.535 Adjusted R2 0.068 0.081 0.043 0.018 0.556 0.639 0.470 0.527 Residual std. error (df) 0.167 0.214 0.220 0.239 0.115 0.134 0.164 0.165 (1298) (303) (706) (426) (1338) (307) (714) (446) F statistic (df) 12.971*** 4.445*** 5.010*** 1.979** 211.501*** 70.566*** 80.880*** 64.169*** (8; 1298) (8; 303) (8; 706) (8; 426) (8; 1338) (8; 307) (8; 714) (8; 446) Note: Cells show estimated unstandardized coefficients with standard errors in parentheses. *p < 0.1 **p < 0.05 ***p < 0.01 The estimation result for model 1 of table 2 reveals that when institutions are more stable, the three main predictors identified in hypotheses 1a, 2, and 3 (Acceptability, Risk, and Utility) influence bureaucratic bribery in the hypothesized fashion. In non-conflict regions of institutionally stable Russia, all of these predictors are significantly associated with corruption behavior, while in conflict regions of Russia, and less institutionally stable Ukraine, Utility does not hold predictive significance. The magnitudes of the estimated coefficients relative to their standard errors for Belief in Acceptability and Risk of Punishment Low are greater in the more stable environments in comparison to those that are less stable.11 These findings lend preliminary support to hypothesis 6. When Official Requested a Bribe is added to the model and Utility is excluded (model 2 of table 2), officials’ demands for bribes emerge as, by far, the best predictor of bureaucratic corruption. These results demonstrate that once a provider demands a bribe, the client often has little option but to pay. According to the R2 statistic, we explain 65 percent of variance in Corruption Behavior in conflict areas of Russia when we account for whether or not an Official Requested a Bribe. Nevertheless, some variance remains to be explained by Belief in Acceptability in both conflict and non-conflict regions of Russia and in non-conflict regions of Ukraine. Although table 2 reveals the importance of hypothesized predictors as well as differences in the magnitudes of their effects in more and less stable institutional contexts, the OLS model provides only a rudimentary tool for assessing our hypotheses. Such a model assumes that predictor variables are uncorrelated with one another and that each is exogenous from the dependent variable (Weisberg 2005). Yet, these assumptions are often untrue in reality. Indeed, we have hypothesized that the more people engage in corruption, the more widespread they perceive corruption to be (H4c). Thus, we expect that variation in the dependent variable actually causes change in perceived spread of corruption, which the OLS model treats as exogenous. A correctly specified model must allow for this endogeneity. Additionally, we expect interactions among some predictors of corruption behavior, while OLS assumes their independence. Moreover, our statistical model needs to account for the fact that demographics influence corruption directly as well as through its perceived acceptability and utility. To accommodate these theoretical expectations, we fit the model represented in figure 1 using structural equation modeling (SEM). SEM estimates simultaneously a number of direct and indirect paths to corruption behavior. For example, our theory states that beliefs about acceptability affect behavior both directly (H1) and indirectly through utility (H4a). The OLS modeling is able to capture the acceptability-to-behavior path, but it cannot simultaneously estimate the acceptability-utility-behavior path. The multi-path patterns modeled by SEM better represent the complexity of the social world and have greater ecological validity. SEM is also appropriate for our analyses because our theory expects endogenous relationships: we not only model corruption behavior as influencing the perceived spread of corruption but also allow these perceptions to affect beliefs about acceptability and risks of corruption (see figure 1). The results from testing the SEM model are presented in tables 3 and 4.12 Table 3 includes Utility as a predictor, while in table 4, Utility is replaced with Official Requested a Bribe. Both tables involve multiple equations estimated simultaneously. In equation 1 of table 3, Utility, Acceptability, Risk, and demographics are the direct predictors of variation in Corruption Behavior (equation 1 resembles the single equation in table 2). The results of this more sophisticated estimation method (equation 1 of table 3) reveal that in both institutionally stable Russia and less stable Ukraine, these three predictors do significantly influence the likelihood of corruption behavior. Table 3. Structural Equation Model of Corruption Behavior with Utility Included Equation 1: Corruption behavior Russia (unstandardized) Ukraine (unstandardized) Russia (standardized) Ukraine (standardized) Belief in acceptability 0.410** 0.027* 1.201 0.066 (0.135) (0.015) Risk of punishment low 0.021* −0.229** 0.078 −0.680 (0.011) (0.088) Utility 0.017** 0.011* 0.151 0.058 (0.009) (0.006) Rural 0.004 −0.008 0.010 −0.016 (0.020) (0.014) Age 0.003*** 0.000 0.245 0.030 (0.001) (0.000) Female −0.003 0.017 −0.007 0.034 (0.015) (0.012) Education −0.023*** 0.003 −0.230 0.023 (0.006) (0.003) Equation 2: Belief in acceptability Spread of corruption −3.040*** 0.048** −5.561 0.086 (0.780) (0.016) Conflict area 0.359** −0.058 0.262 −0.049 (0.142) (0.036) Rural −0.303** −0.142*** −0.236 −0.115 (0.123) (0.038) Age −0.002 −0.003** −0.065 −0.092 (0.002) (0.001) Female −0.162** −0.054 −0.147 −0.045 (0.078) (0.035) Education 0.001 0.000 0.002 0.001 (0.022) (0.010) Equation 3: Risk of punishment low Spread of corruption 0.095*** 4.864** 0.135 7.260 (0.017) (2.246) Conflict area 0.132** −1.718** 0.075 −1.210 (0.043) (0.758) Rural −0.212*** −0.502 −0.130 −0.337 (0.040) (0.339) Equation 4: Utility Belief in acceptability 0.309** 0.213** 0.104 0.097 (0.156) (0.062) Conflict area 0.240** −0.172** 0.059 −0.067 (0.102) (0.075) Rural −0.463*** −0.134* −0.122 −0.050 (0.095) (0.081) Age −0.005* −0.010*** −0.047 −0.145 (0.003) (0.002) Female −0.290*** −0.272*** −0.089 −0.103 (0.080) (0.074) Education 0.100*** 0.104*** 0.115 0.154 (0.022) (0.020) Equation 5: Spread of corruption Corruption behavior 1.397** 6.797** 0.261 1.532 (0.698) (2.457) Model statistics Number of observations 1619 1150 Estimator ML ML Chi-square function test 12.405 16.295 Degrees of freedom 9 9 P-value (chi-square) 0.191 0.061 CFI 0.991 0.968 SRMR 0.012 0.017 PNFI 0.25 0.241 RMSEA 0.015 0.027 90 Percent CI of RMSEA [0.000, 0.034] [0.000, 0.047] P-value RMSEA ≤0.05 1.000 0.974 Equation 1: Corruption behavior Russia (unstandardized) Ukraine (unstandardized) Russia (standardized) Ukraine (standardized) Belief in acceptability 0.410** 0.027* 1.201 0.066 (0.135) (0.015) Risk of punishment low 0.021* −0.229** 0.078 −0.680 (0.011) (0.088) Utility 0.017** 0.011* 0.151 0.058 (0.009) (0.006) Rural 0.004 −0.008 0.010 −0.016 (0.020) (0.014) Age 0.003*** 0.000 0.245 0.030 (0.001) (0.000) Female −0.003 0.017 −0.007 0.034 (0.015) (0.012) Education −0.023*** 0.003 −0.230 0.023 (0.006) (0.003) Equation 2: Belief in acceptability Spread of corruption −3.040*** 0.048** −5.561 0.086 (0.780) (0.016) Conflict area 0.359** −0.058 0.262 −0.049 (0.142) (0.036) Rural −0.303** −0.142*** −0.236 −0.115 (0.123) (0.038) Age −0.002 −0.003** −0.065 −0.092 (0.002) (0.001) Female −0.162** −0.054 −0.147 −0.045 (0.078) (0.035) Education 0.001 0.000 0.002 0.001 (0.022) (0.010) Equation 3: Risk of punishment low Spread of corruption 0.095*** 4.864** 0.135 7.260 (0.017) (2.246) Conflict area 0.132** −1.718** 0.075 −1.210 (0.043) (0.758) Rural −0.212*** −0.502 −0.130 −0.337 (0.040) (0.339) Equation 4: Utility Belief in acceptability 0.309** 0.213** 0.104 0.097 (0.156) (0.062) Conflict area 0.240** −0.172** 0.059 −0.067 (0.102) (0.075) Rural −0.463*** −0.134* −0.122 −0.050 (0.095) (0.081) Age −0.005* −0.010*** −0.047 −0.145 (0.003) (0.002) Female −0.290*** −0.272*** −0.089 −0.103 (0.080) (0.074) Education 0.100*** 0.104*** 0.115 0.154 (0.022) (0.020) Equation 5: Spread of corruption Corruption behavior 1.397** 6.797** 0.261 1.532 (0.698) (2.457) Model statistics Number of observations 1619 1150 Estimator ML ML Chi-square function test 12.405 16.295 Degrees of freedom 9 9 P-value (chi-square) 0.191 0.061 CFI 0.991 0.968 SRMR 0.012 0.017 PNFI 0.25 0.241 RMSEA 0.015 0.027 90 Percent CI of RMSEA [0.000, 0.034] [0.000, 0.047] P-value RMSEA ≤0.05 1.000 0.974 Note: Standard errors are in parentheses. *p < 0.1 **p < 0.05 ***p < 0.001. Five post-estimation tests are applied to examine the goodness of fit of the model, chi-square, Comparative Fit Index (CFI), Standardized Root Mean Square Residual (SRMR), Parsimonious Normed Fit Index (PNFI), and Root Mean Square Error of Approximation (RMSEA). Chi-square p-value: A good model fit would provide an insignificant result at a 0.05 threshold (Barrett 2007). RMSEA: A value of RMSEA <0.06 is recognized as indicative of good fit (Hu and Bentler 1999). CFI: A value of CFI ≥0.95 is presently recognized as indicative of good fit (Hu and Bentler 1999). SRMR: Values for the SRMR range from zero to 1.0 with well-fitting models obtaining values less than 0.05 (Diamantopoulos and Siguaw 2000). Table 3. Structural Equation Model of Corruption Behavior with Utility Included Equation 1: Corruption behavior Russia (unstandardized) Ukraine (unstandardized) Russia (standardized) Ukraine (standardized) Belief in acceptability 0.410** 0.027* 1.201 0.066 (0.135) (0.015) Risk of punishment low 0.021* −0.229** 0.078 −0.680 (0.011) (0.088) Utility 0.017** 0.011* 0.151 0.058 (0.009) (0.006) Rural 0.004 −0.008 0.010 −0.016 (0.020) (0.014) Age 0.003*** 0.000 0.245 0.030 (0.001) (0.000) Female −0.003 0.017 −0.007 0.034 (0.015) (0.012) Education −0.023*** 0.003 −0.230 0.023 (0.006) (0.003) Equation 2: Belief in acceptability Spread of corruption −3.040*** 0.048** −5.561 0.086 (0.780) (0.016) Conflict area 0.359** −0.058 0.262 −0.049 (0.142) (0.036) Rural −0.303** −0.142*** −0.236 −0.115 (0.123) (0.038) Age −0.002 −0.003** −0.065 −0.092 (0.002) (0.001) Female −0.162** −0.054 −0.147 −0.045 (0.078) (0.035) Education 0.001 0.000 0.002 0.001 (0.022) (0.010) Equation 3: Risk of punishment low Spread of corruption 0.095*** 4.864** 0.135 7.260 (0.017) (2.246) Conflict area 0.132** −1.718** 0.075 −1.210 (0.043) (0.758) Rural −0.212*** −0.502 −0.130 −0.337 (0.040) (0.339) Equation 4: Utility Belief in acceptability 0.309** 0.213** 0.104 0.097 (0.156) (0.062) Conflict area 0.240** −0.172** 0.059 −0.067 (0.102) (0.075) Rural −0.463*** −0.134* −0.122 −0.050 (0.095) (0.081) Age −0.005* −0.010*** −0.047 −0.145 (0.003) (0.002) Female −0.290*** −0.272*** −0.089 −0.103 (0.080) (0.074) Education 0.100*** 0.104*** 0.115 0.154 (0.022) (0.020) Equation 5: Spread of corruption Corruption behavior 1.397** 6.797** 0.261 1.532 (0.698) (2.457) Model statistics Number of observations 1619 1150 Estimator ML ML Chi-square function test 12.405 16.295 Degrees of freedom 9 9 P-value (chi-square) 0.191 0.061 CFI 0.991 0.968 SRMR 0.012 0.017 PNFI 0.25 0.241 RMSEA 0.015 0.027 90 Percent CI of RMSEA [0.000, 0.034] [0.000, 0.047] P-value RMSEA ≤0.05 1.000 0.974 Equation 1: Corruption behavior Russia (unstandardized) Ukraine (unstandardized) Russia (standardized) Ukraine (standardized) Belief in acceptability 0.410** 0.027* 1.201 0.066 (0.135) (0.015) Risk of punishment low 0.021* −0.229** 0.078 −0.680 (0.011) (0.088) Utility 0.017** 0.011* 0.151 0.058 (0.009) (0.006) Rural 0.004 −0.008 0.010 −0.016 (0.020) (0.014) Age 0.003*** 0.000 0.245 0.030 (0.001) (0.000) Female −0.003 0.017 −0.007 0.034 (0.015) (0.012) Education −0.023*** 0.003 −0.230 0.023 (0.006) (0.003) Equation 2: Belief in acceptability Spread of corruption −3.040*** 0.048** −5.561 0.086 (0.780) (0.016) Conflict area 0.359** −0.058 0.262 −0.049 (0.142) (0.036) Rural −0.303** −0.142*** −0.236 −0.115 (0.123) (0.038) Age −0.002 −0.003** −0.065 −0.092 (0.002) (0.001) Female −0.162** −0.054 −0.147 −0.045 (0.078) (0.035) Education 0.001 0.000 0.002 0.001 (0.022) (0.010) Equation 3: Risk of punishment low Spread of corruption 0.095*** 4.864** 0.135 7.260 (0.017) (2.246) Conflict area 0.132** −1.718** 0.075 −1.210 (0.043) (0.758) Rural −0.212*** −0.502 −0.130 −0.337 (0.040) (0.339) Equation 4: Utility Belief in acceptability 0.309** 0.213** 0.104 0.097 (0.156) (0.062) Conflict area 0.240** −0.172** 0.059 −0.067 (0.102) (0.075) Rural −0.463*** −0.134* −0.122 −0.050 (0.095) (0.081) Age −0.005* −0.010*** −0.047 −0.145 (0.003) (0.002) Female −0.290*** −0.272*** −0.089 −0.103 (0.080) (0.074) Education 0.100*** 0.104*** 0.115 0.154 (0.022) (0.020) Equation 5: Spread of corruption Corruption behavior 1.397** 6.797** 0.261 1.532 (0.698) (2.457) Model statistics Number of observations 1619 1150 Estimator ML ML Chi-square function test 12.405 16.295 Degrees of freedom 9 9 P-value (chi-square) 0.191 0.061 CFI 0.991 0.968 SRMR 0.012 0.017 PNFI 0.25 0.241 RMSEA 0.015 0.027 90 Percent CI of RMSEA [0.000, 0.034] [0.000, 0.047] P-value RMSEA ≤0.05 1.000 0.974 Note: Standard errors are in parentheses. *p < 0.1 **p < 0.05 ***p < 0.001. Five post-estimation tests are applied to examine the goodness of fit of the model, chi-square, Comparative Fit Index (CFI), Standardized Root Mean Square Residual (SRMR), Parsimonious Normed Fit Index (PNFI), and Root Mean Square Error of Approximation (RMSEA). Chi-square p-value: A good model fit would provide an insignificant result at a 0.05 threshold (Barrett 2007). RMSEA: A value of RMSEA <0.06 is recognized as indicative of good fit (Hu and Bentler 1999). CFI: A value of CFI ≥0.95 is presently recognized as indicative of good fit (Hu and Bentler 1999). SRMR: Values for the SRMR range from zero to 1.0 with well-fitting models obtaining values less than 0.05 (Diamantopoulos and Siguaw 2000). Table 4. Structural Equation Model of Corruption Behavior with Official Requested a Bribe Included Equation 1: Corruption behavior Russia (unstandardized) Ukraine (unstandardized) Russia (standardized) Ukraine (standardized) Official requested bribe 0.707*** 0.729*** 0.735 0.688 (0.016) (0.026) Belief in acceptability 0.031*** 0.031*** 0.091 0.077 (0.007) (0.009) Risk of punishment low 0.010** 0.026 0.040 0.077 (0.004) (0.037) Rural −0.003 0.037** −0.006 0.074 (0.007) (0.011) Age 0.000 0.000 0.019 0.010 (0.000) (0.000) Female −0.000 0.019* −0.001 0.039 (0.006) (0.010) Education −0.002 0.001 −0.017 0.007 (0.002) (0.003) Equation 2: Belief in acceptability Spread of corruption 0.384** 0.047** 0.703 0.084 (0.125) (0.020) Conflict area 0.111** −0.056 0.081 −0.048 (0.033) (0.035) Rural −0.084** −0.134*** −0.065 −0.109 (0.031) (0.037) Age −0.005*** −0.003** −0.161 −0.086 (0.001) (0.001) Female −0.038 −0.055 −0.034 −0.046 (0.026) (0.035) Education 0.034*** 0.001 0.114 0.002 (0.007) (0.009) Equation 3: Risk of punishment low Spread of corruption 0.065** −0.880 0.093 −1.319 (0.027) (1.282) Conflict area 0.136** −0.192*** 0.077 −0.136 (0.043) (0.053) Rural −0.220*** −0.111** −0.134 −0.075 (0.040) (0.046) Equation 4: Spread of corruption Official requested bribe 0.694*** 0.412** 0.134 0.088 (0.126) (0.135) Equation 5: Official requested bribe Conflict area 0.025** 0.056*** 0.051 0.125 (0.012) (0.013) Rural −0.043*** 0.016 −0.095 0.035 (0.011) (0.014) Age 0.001** 0.001** 0.083 0.085 (0.000) (0.000) Female −0.039*** −0.024* −0.099 −0.052 (0.009) (0.013) Model statistics Number of observations 1663 1178 Estimator ML ML Chi-square function test 48.314 29.591 Degrees of freedom 10 10 P-value (chi-square) 0.000 0.001 CFI 0.978 0.979 SRMR 0.02 0.022 PNFI 0.278 0.277 RMSEA 0.048 0.041 90 percent CI [0.035, 0.062] [0.024, 0.058] P-value RMSEA ≤0.05 0.568 0.794 Equation 1: Corruption behavior Russia (unstandardized) Ukraine (unstandardized) Russia (standardized) Ukraine (standardized) Official requested bribe 0.707*** 0.729*** 0.735 0.688 (0.016) (0.026) Belief in acceptability 0.031*** 0.031*** 0.091 0.077 (0.007) (0.009) Risk of punishment low 0.010** 0.026 0.040 0.077 (0.004) (0.037) Rural −0.003 0.037** −0.006 0.074 (0.007) (0.011) Age 0.000 0.000 0.019 0.010 (0.000) (0.000) Female −0.000 0.019* −0.001 0.039 (0.006) (0.010) Education −0.002 0.001 −0.017 0.007 (0.002) (0.003) Equation 2: Belief in acceptability Spread of corruption 0.384** 0.047** 0.703 0.084 (0.125) (0.020) Conflict area 0.111** −0.056 0.081 −0.048 (0.033) (0.035) Rural −0.084** −0.134*** −0.065 −0.109 (0.031) (0.037) Age −0.005*** −0.003** −0.161 −0.086 (0.001) (0.001) Female −0.038 −0.055 −0.034 −0.046 (0.026) (0.035) Education 0.034*** 0.001 0.114 0.002 (0.007) (0.009) Equation 3: Risk of punishment low Spread of corruption 0.065** −0.880 0.093 −1.319 (0.027) (1.282) Conflict area 0.136** −0.192*** 0.077 −0.136 (0.043) (0.053) Rural −0.220*** −0.111** −0.134 −0.075 (0.040) (0.046) Equation 4: Spread of corruption Official requested bribe 0.694*** 0.412** 0.134 0.088 (0.126) (0.135) Equation 5: Official requested bribe Conflict area 0.025** 0.056*** 0.051 0.125 (0.012) (0.013) Rural −0.043*** 0.016 −0.095 0.035 (0.011) (0.014) Age 0.001** 0.001** 0.083 0.085 (0.000) (0.000) Female −0.039*** −0.024* −0.099 −0.052 (0.009) (0.013) Model statistics Number of observations 1663 1178 Estimator ML ML Chi-square function test 48.314 29.591 Degrees of freedom 10 10 P-value (chi-square) 0.000 0.001 CFI 0.978 0.979 SRMR 0.02 0.022 PNFI 0.278 0.277 RMSEA 0.048 0.041 90 percent CI [0.035, 0.062] [0.024, 0.058] P-value RMSEA ≤0.05 0.568 0.794 Note: See note to table 3. Table 4. Structural Equation Model of Corruption Behavior with Official Requested a Bribe Included Equation 1: Corruption behavior Russia (unstandardized) Ukraine (unstandardized) Russia (standardized) Ukraine (standardized) Official requested bribe 0.707*** 0.729*** 0.735 0.688 (0.016) (0.026) Belief in acceptability 0.031*** 0.031*** 0.091 0.077 (0.007) (0.009) Risk of punishment low 0.010** 0.026 0.040 0.077 (0.004) (0.037) Rural −0.003 0.037** −0.006 0.074 (0.007) (0.011) Age 0.000 0.000 0.019 0.010 (0.000) (0.000) Female −0.000 0.019* −0.001 0.039 (0.006) (0.010) Education −0.002 0.001 −0.017 0.007 (0.002) (0.003) Equation 2: Belief in acceptability Spread of corruption 0.384** 0.047** 0.703 0.084 (0.125) (0.020) Conflict area 0.111** −0.056 0.081 −0.048 (0.033) (0.035) Rural −0.084** −0.134*** −0.065 −0.109 (0.031) (0.037) Age −0.005*** −0.003** −0.161 −0.086 (0.001) (0.001) Female −0.038 −0.055 −0.034 −0.046 (0.026) (0.035) Education 0.034*** 0.001 0.114 0.002 (0.007) (0.009) Equation 3: Risk of punishment low Spread of corruption 0.065** −0.880 0.093 −1.319 (0.027) (1.282) Conflict area 0.136** −0.192*** 0.077 −0.136 (0.043) (0.053) Rural −0.220*** −0.111** −0.134 −0.075 (0.040) (0.046) Equation 4: Spread of corruption Official requested bribe 0.694*** 0.412** 0.134 0.088 (0.126) (0.135) Equation 5: Official requested bribe Conflict area 0.025** 0.056*** 0.051 0.125 (0.012) (0.013) Rural −0.043*** 0.016 −0.095 0.035 (0.011) (0.014) Age 0.001** 0.001** 0.083 0.085 (0.000) (0.000) Female −0.039*** −0.024* −0.099 −0.052 (0.009) (0.013) Model statistics Number of observations 1663 1178 Estimator ML ML Chi-square function test 48.314 29.591 Degrees of freedom 10 10 P-value (chi-square) 0.000 0.001 CFI 0.978 0.979 SRMR 0.02 0.022 PNFI 0.278 0.277 RMSEA 0.048 0.041 90 percent CI [0.035, 0.062] [0.024, 0.058] P-value RMSEA ≤0.05 0.568 0.794 Equation 1: Corruption behavior Russia (unstandardized) Ukraine (unstandardized) Russia (standardized) Ukraine (standardized) Official requested bribe 0.707*** 0.729*** 0.735 0.688 (0.016) (0.026) Belief in acceptability 0.031*** 0.031*** 0.091 0.077 (0.007) (0.009) Risk of punishment low 0.010** 0.026 0.040 0.077 (0.004) (0.037) Rural −0.003 0.037** −0.006 0.074 (0.007) (0.011) Age 0.000 0.000 0.019 0.010 (0.000) (0.000) Female −0.000 0.019* −0.001 0.039 (0.006) (0.010) Education −0.002 0.001 −0.017 0.007 (0.002) (0.003) Equation 2: Belief in acceptability Spread of corruption 0.384** 0.047** 0.703 0.084 (0.125) (0.020) Conflict area 0.111** −0.056 0.081 −0.048 (0.033) (0.035) Rural −0.084** −0.134*** −0.065 −0.109 (0.031) (0.037) Age −0.005*** −0.003** −0.161 −0.086 (0.001) (0.001) Female −0.038 −0.055 −0.034 −0.046 (0.026) (0.035) Education 0.034*** 0.001 0.114 0.002 (0.007) (0.009) Equation 3: Risk of punishment low Spread of corruption 0.065** −0.880 0.093 −1.319 (0.027) (1.282) Conflict area 0.136** −0.192*** 0.077 −0.136 (0.043) (0.053) Rural −0.220*** −0.111** −0.134 −0.075 (0.040) (0.046) Equation 4: Spread of corruption Official requested bribe 0.694*** 0.412** 0.134 0.088 (0.126) (0.135) Equation 5: Official requested bribe Conflict area 0.025** 0.056*** 0.051 0.125 (0.012) (0.013) Rural −0.043*** 0.016 −0.095 0.035 (0.011) (0.014) Age 0.001** 0.001** 0.083 0.085 (0.000) (0.000) Female −0.039*** −0.024* −0.099 −0.052 (0.009) (0.013) Model statistics Number of observations 1663 1178 Estimator ML ML Chi-square function test 48.314 29.591 Degrees of freedom 10 10 P-value (chi-square) 0.000 0.001 CFI 0.978 0.979 SRMR 0.02 0.022 PNFI 0.278 0.277 RMSEA 0.048 0.041 90 percent CI [0.035, 0.062] [0.024, 0.058] P-value RMSEA ≤0.05 0.568 0.794 Note: See note to table 3. To assess differences in the magnitudes of effects, we report standardized coefficients in tables 3 and 4. We confirm H6 for two of the three predictors: the magnitudes of the effects that acceptability beliefs and utility have on bureaucratic corruption are greater in the more stable institutions of Russia.13 For risk of punishment, the effect on corruption behavior is positive in Russia and negative in Ukraine.14 Equation 1 of table 3 also describes direct effects of demographic characteristics on the likelihood of engaging in bureaucratic corruption. The most significant finding is that in institutionally unstable Ukraine, none of the demographics significantly influence the likelihood of corrupt behavior. In stable Russia, however, older people are more likely to engage in bureaucratic corruption, while higher education is associated with lower chances of such engagement. Next, equation 2 in table 3, predicting Belief in Acceptability of Bribery, allows individual demographic characteristics, conflict area, and spread of corruption to influence corrupt behavior indirectly through beliefs about its acceptability (see also figure 1). We see that in Ukraine, the more corruption citizens perceive in society, the more acceptable they think that bribery is, while in Russia perceptions that corruption is increasing decrease beliefs in bribery’s acceptability. Another difference is that in Russia, with its more stable institutions, acceptability is significantly higher in conflict areas, while in less stable Ukraine, acceptability is unrelated to residence in a conflict region. Equation 3 in table 3, in turn, reveals that in both countries, risk of punishment is seen as lower when corruption is perceived to be increasing in society. This confirms hypothesis 4b. Importantly, this equation reveals that national institutional context matters for the relationship between proximity to conflict and risk of punishment. In the country with greater institutional stability, residence in a conflict area increases the perceived risk of detection and punishment. In contrast, in less stable Ukraine, residence in a conflict area decreases the perceived risk of detection and punishment. Next, equation 4 of table 3 for predicting utility supports H4a: beliefs about acceptability of bribery are significantly related to calculations of its utility. Again, this finding confirms the import of modeling the predictors of corrupt behavior in a way that accommodates the interactions among them. Equation 4 again underscores the importance of institutional context for bureaucratic corruption. In institutionally stable Russia, the utility of bureaucratic corruption is greater in conflict areas. In less stable Ukraine, by contrast, additional increases in instability (associated with residing in or near a conflict zone) reduce utility. The relevance of demographic characteristics to utility is, however, the same in Russia and Ukraine: utility is higher in non-rural settings, decreases with age, increases with education, and is lower for females. The last equation in table 3, equation 5, predicts perceptions about the spread of corruption with actual corrupt behavior. This equation incorporates our hypothesized feedback loop, whereby perceptions of corruption affect behavior through their effects on perceived acceptability of corruption (equation 2), while engagement in corruption in turn increases the likelihood of perceiving corruption as increasing in society (equation 5). Note that hypothesis 4c is confirmed in both Russia and Ukraine. In table 4, we test the direct effect of Official Requested a Bribe on bribery behavior in a separate SEM (see figure 2 for graphic representations). Since officials’ requests for bribes account for more than 50 percent of the variance in corruption behavior (from table 2), limited variance is left to be explained by other predictors. Nevertheless, in both stable and unstable institutional contexts, beliefs about the acceptability of bribery still significantly and independently influence the likelihood of corrupt behavior. In addition, in stable Russia, an individual is less likely to engage in bureaucratic corruption, even if asked for a bribe by an official, when perceived risk of punishment is high (equation 1 of table 4). Equation 2 of table 4 suggests that when officials’ requests for bribes are included in the SEM model, beliefs about the acceptability of bribery increase as perceptions of corruption in society rise. In addition, beliefs about the acceptability of bribery are consistently lower while perceived risk of punishment is higher in rural areas. Figure 2. View largeDownload slide Theoretical model of corruption behavior with extortion (official requested bribe) Figure 2. View largeDownload slide Theoretical model of corruption behavior with extortion (official requested bribe) Equation 5 in table 4 also shows where and from whom requests for bribes are most likely to occur: in both Russia and Ukraine, they occur more frequently in conflict areas and when the client is older and male. Discussion Our study reveals the micro-foundations of bureaucratic corruption. Based on sociological theory and case studies of high-corruption societies, we identify the processes behind corrupt behavior among the clients of service-provision organizations. Our statistical tests confirm that the beliefs about acceptability of corruption and the perceived risk of punishment affect the likelihood of citizens’ bribery behavior, whether or not an official has requested a bribe. Utility calculations affect the incidence of bureaucratic corruption distinct from explicit requests for a bribe. To specify and test the effects of these determinants on bribery behavior, our study pioneered a structural equation model, which accommodates multiple direct and indirect paths to the outcome, as well as causal dependencies among endogenous and exogenous variables. We demonstrate empirically that causes of bureaucratic corruption, while conceptually distinct from one another, are not independent in their operation. We are also the first to incorporate into a theory of bureaucratic corruption a feedback loop whereby engagement in corruption reinforces perceptions of its spread, which in turn contributes to the belief that bribery is acceptable. Although the cyclical nature of bureaucratic corruption has been demonstrated by formal theorists (e.g., Andvig and Moene 1990), prior to our analyses, no one had tested such cyclicality empirically. In addition, we offer a new measure: the utility of bribery, which is an improvement on existing measures, as it weighs perceived benefits of bribery against respondents’ financial means. Our main theoretical contribution, however, consists in revealing the effects of institutional stability on the scale and causes of bureaucratic corruption. First, our results reveal significant variance in rates of bureaucratic corruption in Ukraine and Russia despite their comparable scores on Transparency International’s Corruption Perception Index (TICPI). While TICPI (2015) reflects perceptions of the spread of political as well as bureaucratic corruption, our focus specifically on citizens’ reported engagement in corrupt exchanges with street-level bureaucrats reveals notable differences between and within the two countries. Thus, our analyses show that citizens resort to under-the-table transactions with bureaucrats more rarely in relatively stable Russia than they do in destabilized Ukraine; within each country, less stable conflict areas also have higher rates of corrupt engagement than other regions. This finding confirms that institutional instability, in and of itself, has criminogenic effects. Second, we demonstrate that institutional stability affects the strength of theoretically established determinants of corruption: not only do these predictors have greater explanatory power in stable Russia than in unstable Ukraine, but within each country, their impact on behavior diminishes in areas destabilized by military conflict. In other words, beliefs and utility calculations are more closely tied to bribery behavior when people deal with predictable bureaucracies. When institutions are destabilized, beliefs may be in flux and the utility of corruption is difficult to calculate, which diminishes their impact on people’s behavior. Institutions may be destabilized by a number of socio-political processes. Paired comparison of Russia and Ukraine provides preliminary evidence that frequent regime change at the national level, which undermines the stability of institutions, impacts citizens’ pathways to rule-breaking. Within countries, citizens’ pathways to corruption are systematically different in regions beset by military conflict. We see contextual effects of residence in a rural area: the intimacy of rural communities, with their heightened social reciprocity, makes corruption less accepted and its risks higher. Together, these findings at national, regional, and community levels reveal that stability of institutions influences crime. Our findings extend the institutional theory of social control of crime. Social control theorists have shown the importance of belonging to specific organizations, networks, family structures, and associations. We, in turn, have theorized the importance of living in a society ordered by stable institutions. We hypothesized that stable institutions allow rules to be more knowable, thereby reducing crimes, such as corruption. Our empirical results indicate that order begets order: when people are confident about how the social world operates, they are less likely to engage in illegality. Conclusion Every year, the international community spends billions of dollars on fighting corruption in non-Western societies (Sampson 2010). Based on their low rankings on TICPI and other indicators of corruption, academics and policymakers routinely refer to these countries as “pervasively” and “endemically” corrupt (Lovell, Ledeneva, and Rogachevskii 2000; Varese 2000). Such designation implies unvarying levels of corruption and generates unwarranted defeatism regarding the prospects of fighting corruption in these societies. In contrast, our study underscores and explains internal variation in high-corruption societies. Not only do our surveys reveal that in such societies many people regularly abstain from corruption, our model shows why some are more likely to resort to illicit transactions than others. Building on these findings, social scientists can avoid the trap of relegating certain countries to the lower ranks of what has, essentially, emerged as a global moral order. Our findings also underscore the importance of tailoring anti-corruption policy to specific national and subnational contexts. Most policy recommendations that international donors extend to high-corruption societies fall back on strategies that were conceptualized and tested in the Global North (Larmour 2007). The assumption behind such policy transfer is that determinants of corrupt behavior are stable across different contexts. While most critiques of policy adoption focus on differences in governance structures (Minogue 2004), our study underscores another important reason why policies successful elsewhere may fail in a different setting: citizens’ causal pathways to corruption depend on institutional stability. For instance, our data suggest that when institutions are destabilized, anti-corruption strategies that focus on reducing instrumental incentives for bribery may be less effective. In such contexts, more positive outcomes might emerge from policies targeting cultural predictors of corruption—that is, educational and outreach campaigns that focus on social harms of corruption and that raise the stigma associated with engagement. In contrast, when a system is stable but still marked by corruption, anti-corruption strategies that reduce the utility of corruption may have better results than policies focused on educating the citizenry. Ultimately, our findings also have implications for the sociology of knowledge, as they imply that existing theories of corruption (and, likely, other crimes) are contingent on institutional stability. We therefore call into question the generalizability of studies conducted in Western societies and invite more research in contexts with lower institutional stability. Notes 1 In the literature, such corruption also goes under the names of petty, administrative, or low-level corruption (see Karklins [2005] for examples). 2 Corruption distorts the allocation of resources, discourages investment, and impairs economic growth (Dabla-Norris 2000). 3 We define high-corruption countries broadly as countries that cluster in the lower 30th percentile of Transparency International’s corruption rankings (Transparency International 2016). 4 Since 2012, Transparency International (2016) has ranked both Ukraine and Russia in the 20th percentile of the most corrupt countries in the world. 5 In the relevant literature, the word extortion is used to refer to bureaucrats’ requests for bribes (Karklins 2005; Tavits 2010). This word, however, generally implies force or threat, and our survey questions focus on whether an official asked for a bribe, not on whether the request for a bribe was accompanied by a veiled threat or delays and rudeness in service. Therefore, we use the term “official’s request for bribe” instead of “extortion.” 6 To save space, we present a single representation, but we expect that the magnitude of the effects in this model are greater in contexts where institutions are more stable. 7 According to Sanders (1981), regime change encompasses changes in regime norms, types of party system, and/or military-civilian status. 8 Some readers may worry whether respondents are willing to divulge to the interviewer their engagement in bureaucratic corruption. We employed multiple strategies for making respondents feel comfortable in answering our queries honestly (see appendix A, “Safeguards against Social Desirability Bias”). We also used a list experiment to assess the accuracy of responses to questions about respondents’ engagement in corruption. Other researchers have also documented a lack of social desirability bias associated with discussing bureaucratic corruption in post-Soviet countries (McMann 2014; Petrov and Temple 2004). 9 The formula results in a non-zero number representing zero bribes paid because we added a constant to the numerator and denominator of the ratio to avoid dividing by zero. 10 As an alternative to this split-sample approach, we also use interaction terms to test the hypothesis that the relationship between main theoretical predictors and corruption behavior is different in conflict and non-conflict areas (table 2D of appendix D). The results in table 2D reflect those in table 2, and provide confirmation that the effects of Acceptability and Risk are significantly different in conflict and non-conflict zones. 11 Standardized coefficients are calculated by dividing the unstandardized estimated coefficients reported in table 2 by their standard errors (reported in parentheses). The larger standardized coefficient associated with Acceptability and Risk in non-conflict areas as compared with conflict areas indicates that the effects of these variables on corruption behavior are greater (relative to other predictors) in the non-conflict areas. For example, in Russia, the effect of Acceptability (relative to other predictors in the equation) is lower in less stable conflict areas. 12 These relations are simultaneously estimated with covariant residuals among perceived spread of corruption, corruption behavior, risk of punishment, and beliefs about acceptability, as shown in dashed lines in figure 1. The SEM models were estimated by the “sem” function of the lavaan package (0.5–23.1097) on R 3.2.5 (see appendix C). 13 We report standardized coefficients for illustrative purposes only, and understand the perils of comparing these across independent samples. 14 In the OLS regressions (model 1 of table 2), where the effects of the predictor variables affect corruption behavior independently from one another, we see that in conflict areas of Ukraine, Risk is unrelated to Corruption Behavior. In SEM, however, we have a predictive equation (equation 3) embedded into the analysis that allows residence in a conflict area to push risk of punishment higher in conflict areas of Ukraine. This has the effect of creating a negative relationship between low risk of punishment and corruption behavior in Ukraine. Conflict areas of Ukraine are in large part occupied by Russian troops (a foreign occupying army), thus risk of punishment can be very real in these regions. About the Authors Marina Zaloznaya is an Assistant Professor of Sociology at the University of Iowa. Her work is based on ethnographic, survey, and comparative-historical analyses of informal economies in non-democratic political systems, in Eastern Europe and China. Zaloznaya’s publications include a recent book, The Politics of Bureaucratic Corruption in Post-Transitional Eastern Europe (Cambridge University Press, 2017), and articles in a range of sociology and interdisciplinary journals. Vicki Hesli Claypool is Professor Emeritus of the Political Science Department at the University of Iowa. Her research career focused primarily on studying voting behavior and public opinion in Russia and Ukraine using mass surveys based on person-to-person interviews. Most of her recent publications have appeared in PS: Political Science & Politics and address topics associated with success in the academic profession, including job satisfaction, faculty rank attainment, research productivity, and academic salaries. William M. Reisinger is Professor of Political Science at the University of Iowa. He specializes in authoritarianism and democracy in the former communist states, especially Russia. He has written or edited eight books, including The Regional Roots of Russia’s Political Regime (University of Michigan Press, 2017), coauthored with Bryon J. Moraski, as well as numerous articles and book chapters. Supplementary Material Supplementary material is available at Social Forces online. References Ake , Claude . 1975 . “ A Definition of Political Stability .” Comparative Politics 7 ( 2 ): 271 – 83 . Google Scholar CrossRef Search ADS Akers , Ronald . 2011 . Social Learning and Social Structure: A General Theory of Crime and Deviance . New Brunswick, NJ : Transaction Publishers . Andvig , Jens , and Karl Ove Moene . 1990 . “ How Corruption May Corrupt .” Journal of Economic Behavior & Organization 13 ( 1 ): 63 – 76 . Google Scholar CrossRef Search ADS Apel , Robert , and Raymond Paternoster . 2009 . “Understanding ‘Criminogenic’ Corporate Culture: What White-Collar Crime Researchers Can Learn from Studies of the Adolescent Employment–Crime Relationship.” In The Criminology of White-Collar Crime , edited by Sally Simpson and David Weisburd , 15 – 33 . New York : Springer . Google Scholar CrossRef Search ADS Ashforth , Blake , and Vikas Anand . 2003 . “ The Normalization of Corruption in Organizations .” Research in Organizational Behavior 25 : 1 – 52 . Google Scholar CrossRef Search ADS Bardhan , Pranab . 1997 . “ Corruption and Development: A Review of Issues .” Journal of Economic Literature 35 ( 3 ): 1320 – 46 . Barrett , Paul . 2007 . “ Structural Equation Modelling: Adjudging Model Fit .” Personality and Individual Differences 42 ( 5 ): 815 – 24 . Google Scholar CrossRef Search ADS Becquart-Leclercq , Jeanne . 1989 . “Paradoxes of Political Corruption: A French View.” In Political Corruption: A Handbook , edited by Arnold Heidenheimer , Victor LeVine , and Michael Johnston , 19 – 36 . New Brunswick, NJ : Transaction Publishers . Bellows , John , and Edward Miguel . 2006 . “ War and Institutions: New Evidence from Sierra Leone .” American Economic Review 96 ( 2 ): 394 – 99 . Google Scholar CrossRef Search ADS Benson , Michael , and Sally Simpson . 2015 . Understanding White-Collar Crime: An Opportunity Perspective . New York : Taylor & Francis . Birnbaum , Michael , and Karoun Demirjian . 2015 . “Fighting Spreads in Eastern Ukraine after Rebel Rout.” Washington Post, February 20. Čábelková , Inna , and Jan Hanousek . 2004 . “ The Power of Negative Thinking: Corruption, Perception and Willingness to Bribe in Ukraine .” Applied Economics 36 ( 4 ): 383 – 97 . Google Scholar CrossRef Search ADS Cohen , Lawrence , and Marcus Felson . 1979 . “ Social Change and Crime Rate Trends: A Routine Activity Approach .” American Sociological Review 44 ( 4 ): 588 – 608 . Google Scholar CrossRef Search ADS Cornell , Svante , and Frederick Starr , eds. 2009 . The Guns of August: Russia’s War in Georgia . Armonk, NY : M. E. Sharpe . Dabla-Norris , Era . 2000 . A Game-Theoretic Analysis of Corruption in Bureaucracies . Washington, DC : International Monetary Fund . Deuchar , Ross . 2009 . Gangs, Marginalised Youth and Social Capital . Staffordshire, UK : Trentham Books . Diamantopoulos , Adamantios and Judy A. Siguaw . 2000 . Introducing LISREL: A Guide for the Uninitiated. London : Sage. Edelman , Lauren . 1992 . “ Legal Ambiguity and Symbolic Structures: Organizational Mediation of Civil Rights Law .” American Journal of Sociology 97 ( 6 ): 1531 – 76 . Google Scholar CrossRef Search ADS Ellickson , Robert . 1986 . “ Of Coase and Cattle: Dispute Resolution among Neighbors in Shasta County .” Stanford Law Review 38 : 623 – 87 . Google Scholar CrossRef Search ADS Emerson , Richard . 1976 . “ Social Exchange Theory .” Annual Review of Sociology 2 : 335 – 62 . Google Scholar CrossRef Search ADS Fimyar , Olena . 2010 . “ Policy Why(s): Policy Rationalities and the Changing Logic of Educational Reform in Postcommunist Ukraine .” International Perspectives on Education and Society 14 : 61 – 91 . Google Scholar CrossRef Search ADS Forrat , Natalia . 2015 . “The Political Economy of Russian Higher Education: Why Does Putin Support Research Universities?” Post-Soviet Affairs 32 ( 4 ): 1 – 39 . Gambetta , Diego . 1996 . The Sicilian Mafia: The Business of Private Protection . Cambridge, MA : Harvard University Press . Geddes , Barbara , Joseph Wright , and Erica Frantz . 2014 . “ Autocratic Breakdown and Regime Transitions: A New Data Set .” Perspectives on Politics 12 ( 2 ): 313 – 31 . Google Scholar CrossRef Search ADS Gel’man, Vladimir . 2004 . “ The Unrule of Law in the Making: The Politics of Informal Institution Building in Russia .” Europe-Asia Studies 56 ( 7 ): 1021 – 40 . Google Scholar CrossRef Search ADS Gel’man , Vladimir . 2015 . Authoritarian Russia: Analyzing Post-Soviet Regime Changes . Pittsburgh, PA : University of Pittsburgh Press . Google Scholar CrossRef Search ADS Gerber , Theodore , and Sarah Mendelson . 2008 . “ Public Experiences of Police Violence and Corruption in Contemporary Russia: A Case of Predatory Policing? ” Law & Society Review 42 ( 1 ): 1 – 44 . Google Scholar CrossRef Search ADS Graeff , Peter , Sebastian Sattler , Guido Mehlkop , and Carsten Sauer . 2014 . “ Incentives and Inhibitors of Abusing Academic Positions: Analysing University Students’ Decisions About Bribing Academic Staff .” European Sociological Review 30 ( 2 ): 230 – 41 . Google Scholar CrossRef Search ADS Grødeland , Åse , Tatyana Koshechkina , and William Miller . 1998 . “ ‘Foolish to Give and Yet More Foolish Not to Take’: In-Depth Interviews with Post-Communist Citizens on Their Everyday Use of Bribes and Contacts .” Europe-Asia Studies 50 ( 4 ): 651 – 77 . Google Scholar CrossRef Search ADS Hendley , Kathryn . 2017 . Everyday Law in Russia . Ithaca, NY : Cornell University Press . Hesli , Vicki . 2007 . Governments and Politics in Russia and the Post-Soviet Region . Boston : Houghton Mifflin . Hesli , Vicki L. , Ha-Lyong Jung , William M. Reisinger and Arthur H. Miller . 2001 . “ The Gender Divide in Russian Politics: Attitudinal and Behavioral Considerations .” Women & Politics 22 ( 2 ): 41 – 80 . Google Scholar CrossRef Search ADS Hirschi , Travis . 1969 . “A Control Theory of Delinquency.” In Criminology Theory: Selected Classic Readings , edited by Frank Williams and Marylin McShane , 289 – 305 . Cincinnati, OH : Anderson Publishing . Homel , Ross . 2012 . Policing and Punishing the Drinking Driver: A Study of General and Specific Deterrence . New York : Springer . Hu , Li‐tze , and Peter M. Bentler . 1999 . “ Cutoff Criteria for Fit Indexes in Covariance Structure Analysis: Conventional Criteria Versus New Alternatives .” Structural Equation Modeling: A Multidisciplinary Journal 6 ( 1 ): 1 – 5 . Google Scholar CrossRef Search ADS Jackall , Robert . 1988 . “ Moral Mazes: The World of Corporate Managers .” International Journal of Politics, Culture, and Society 1 ( 4 ): 598 – 614 . Google Scholar CrossRef Search ADS Karklins , Rasma . 2002 . “ Typology of Post-Communist Corruption .” Problems of Post-Communism 49 ( 4 ): 22 – 32 . Google Scholar CrossRef Search ADS ——— . 2005 . The System Made Me Do It: Corruption in Post-Communist Societies . Armonk, NY : M. E. Sharpe . Kleck , Gary , Brion Sever , Spencer Li , and Marc Gertz . 2005 . “ The Missing Link in General Deterrence Research .” Criminology 43 ( 3 ): 623 – 60 . Google Scholar CrossRef Search ADS Köbis , Nils , Jan-Willem van Prooijen , Francesca Righetti , and Paul Van Lange . 2015 . “ ‘Who Doesn’t?’—The Impact of Descriptive Norms on Corruption .” PloS One 10 ( 6 ): 1 – 14 . Google Scholar CrossRef Search ADS Kornhauser , Ruth . 1978 . Social Sources of Delinquency: An Appraisal of Analytic Models . Chicago : University of Chicago Press . Kubrin , Charis , and Ronald Weitzer . 2003 . “ New Directions in Social Disorganization Theory .” Journal of Research in Crime and Delinquency 40 ( 4 ): 374 – 402 . Google Scholar CrossRef Search ADS Kuzio , Taras . 2015 . “Ukraine: Leaving the Crossroads.” In Central and East European Politics: From Communism to Democracy , edited by Sharon Wolchik and Jane Curry , 481 – 512 . Lanham, MD : Rowman & Littlefield . Larmour , Peter . 2007 . “ International Action against Corruption in the Pacific Islands: Policy Transfer, Coercion and Effectiveness .” Asian Journal of Political Science 15 ( 1 ): 1 – 16 . Google Scholar CrossRef Search ADS Ledeneva , Alena . 1998 . Russia’s Economy of Favors: Blat, Networking and Informal Exchange . New York : Cambridge University Press . Ledeneva , Alena . 2008 . “ Telephone Justice in Russia .” Post-Soviet Affairs 24 ( 4 ): 324 – 50 . Google Scholar CrossRef Search ADS Lovell , Stephen , Alena Ledeneva , and A. Rogachevskii , eds. 2000 . Bribery and Blat in Russia: Negotiating Reciprocity from the Middle Ages to the 1990s . New York : St. Martin’s Press . McMann , Kelly . 2014 . Corruption as a Last Resort: Adapting to the Market in Central Asia . Ithaca, NY : Cornell University Press . Merry , Sally Engle . 1988 . “ Legal Pluralism .” Law & Society Review 22 ( 5 ): 869 – 96 . Google Scholar CrossRef Search ADS ——— . 1993 . “ Mending Walls and Building Fences: Constructing the Private Neighborhood .” Journal of Legal Pluralism and Unofficial Law 25 ( 33 ): 71 – 90 . Google Scholar CrossRef Search ADS Merton , Robert . 1968 . Social Theory and Social Structure . New York : Simon and Schuster . Miller , William , Ase Grødeland , and Tatyana Koshechkina . 2001 . A Culture of Corruption? Coping with Government in Post-Communist Europe . Budapest : Central European University Press . Minogue , Martin . 2004 . “Public Management and Regulatory Governance: Problems of Policy Transfer to Developing Countries.” In Leading Issues in Competition, Regulation, and Development , edited by Paul Cook , Colin Kirkpatrick , Martin Minogue , and David Parker , 165 – 82 . Cheltenham, UK : Edward Elgar Publishing . Mungiu-Pippidi , Alina . 2013 . “ Controlling Corruption through Collective Action .” Journal of Democracy 24 ( 1 ): 101 – 15 . Google Scholar CrossRef Search ADS North , Douglass . 1990 . Institutions, Institutional Change and Economic Performance . Cambridge : Cambridge University Press . Google Scholar CrossRef Search ADS Nye , Joseph . 1967 . “ Corruption and Political Development: A Cost-Benefit Analysis .” American Political Science Review 61 ( 2 ): 417 – 27 . Google Scholar CrossRef Search ADS O’Brien , David , and Stephen Wegren . 2002 . Rural Reform in Post-Soviet Russia . Washington, DC : Woodrow Wilson Center Press . Osipian , Ararat . 2009 . “ Corruption Hierarchies in Higher Education in the Former Soviet Bloc .” International Journal of Educational Development 29 ( 3 ): 321 – 30 . Google Scholar CrossRef Search ADS Ostrom , Elinor , and Xavier Basurto . 2011 . “ Crafting Analytical Tools to Study Institutional Change .” Journal of Institutional Economics 7 ( 03 ): 317 – 43 . Google Scholar CrossRef Search ADS Peisakhin , Leonid , and Paul Pinto . 2010 . “ Is Transparency an Effective Anti-Corruption Strategy? Evidence from a Field Experiment in India .” Regulation & Governance 4 ( 3 ): 261 – 80 . Google Scholar CrossRef Search ADS Pellegrini , Lorenzo . 2008 . “ Causes of Corruption: A Survey of Cross-Country Analyses and Extended Results .” Economics of Governance 9 ( 3 ): 245 – 63 . Google Scholar CrossRef Search ADS Persson , Anna , Bo Rothstein , and Jan Teorell . 2013 . “ Why Anticorruption Reforms Fail—Systemic Corruption as a Collective Action Problem .” Governance 26 ( 3 ): 449 – 71 . Google Scholar CrossRef Search ADS Petrov , Georgy , and Paul Temple . 2004 . “ Corruption in Higher Education .” Higher Education Management and Policy 16 ( 1 ): 83 – 99 . Google Scholar CrossRef Search ADS Polese , Abel . 2008 . “ ‘If I Receive It, It Is a Gift; If I Demand It, Then It Is a Bribe’: On the Local Meaning of Economic Transactions in Post-Soviet Ukraine .” Anthropology in Action 15 ( 3 ): 47 – 60 . Google Scholar CrossRef Search ADS Posner , Richard . 1997 . “ Social Norms and the Law: An Economic Approach .” American Economic Review 87 ( 2 ): 365 – 69 . Rehn , Alf , and Saara Taalas . 2004 . “ ‘Znakomstva I Svyazi’ (Acquaintances and Connections)—Blat, the Soviet Union, and Mundane Entrepreneurship .” Entrepreneurship & Regional Development 16 ( 3 ): 235 – 50 . Google Scholar CrossRef Search ADS Reisinger , William , Marina Zaloznaya , and Vicki Hesli Claypool . 2017 . “ Everyday Corruption and Regime Support in Russia and Ukraine .” Post-Soviet Affairs 33 ( 4 ): 255 – 75. Google Scholar CrossRef Search ADS Riabchuk , Mykola . 2007 . “Ambivalence or Ambiguity? Why Ukraine Is Trapped between East and West.” In Ukraine, the EU and Russia , edited by Stephen Velychenko , 70 – 88 . New York : Pallgrave Macmillan . Google Scholar CrossRef Search ADS Rivkin-Fish , Michele . 2005 . “Bribes, Gifts and Unofficial Payments: Rethinking Corruption in Post-Soviet Russian Health Care.” In Corruption: Anthropological Perspectives , edited by Dieter Haller and Cris Shore , 47 – 64 . Ann Arbor, MI : Pluto . Rosenfeld , Richard , Eric Baumer , and Steven Messner . 2001 . “ Social Capital and Homicide .” Social Forces 80 ( 1 ): 283 – 310 . Google Scholar CrossRef Search ADS Sah , Raaj . 1988 . “Persistence and Pervasiveness of Corruption: New Perspectives.” Yale Economic Growth Center Discussion Paper 560, accessed online at http://www.raajsah.com/uploads/4/3/4/9/43495347/wp1988_persistence-and-pervasiveness-of-corruption.pdf, October 26, 2017. Sajó , András . 2003 . “ From Corruption to Extortion: Conceptualization of Post-Communist Corruption .” Crime, Law and Social Change 40 ( 2 ): 171 – 94 . Google Scholar CrossRef Search ADS Sakwa , Richard . 2010 . “ The Dual State in Russia .” Post-Soviet Affairs 26 ( 3 ): 185 – 206 . Google Scholar CrossRef Search ADS Sampson , Steven . 2010 . “ The Anti-Corruption Industry: From Movement to Institution .” Global Crime 11 ( 2 ): 261 – 78 . Google Scholar CrossRef Search ADS Sanders , David . 1981 . Patterns of Political Instability . New York : St. Martin’s Press . Google Scholar CrossRef Search ADS Shamir , Ronen . 2004 . “ Between Self-Regulation and the Alien Tort Claims Act: On the Contested Concept of Corporate Social Responsibility .” Law & Society Review 38 ( 4 ): 635 – 64 . Google Scholar CrossRef Search ADS Smith , Daniel . 2010 . A Culture of Corruption: Everyday Deception and Popular Discontent in Nigeria . Princeton, NJ : Princeton University Press . Spicer , Andrew , Gerald McDermott , and Bruce Kogut . 2000 . “ Entrepreneurship and Privatization in Central Europe: The Tenuous Balance between Destruction and Creation .” Academy of Management Review 25 ( 3 ): 630 – 49 . Google Scholar CrossRef Search ADS Sunshine , Jason , and Tom Tyler . 2003 . “ The Role of Procedural Justice and Legitimacy in Shaping Public Support for Policing .” Law & Society Review 37 ( 3 ): 513 – 48 . Google Scholar CrossRef Search ADS Sutherland , Edwin H. 1947 . Criminology . Philadelphia : Lippincott . Tarrow , Sidney . 2010 . “ The Strategy of Paired Comparison: Toward a Theory of Practice .” Comparative Political Studies 43 ( 2 ): 230 – 59 . Google Scholar CrossRef Search ADS Tavits , Margit . 2010 . “ Why Do People Engage in Corruption? The Case of Estonia .” Social Forces 88 ( 3 ): 1257 – 79 . Google Scholar CrossRef Search ADS Transparency International . 2015 . Corruption Perceptions Index 2015, accessed online at https://www.transparency.org/cpi2015#downloads, September 2017. ——— . 2016 . Corruption Perception Index 2016, accessed online at https://www.transparency.org/news/feature/corruption_perceptions_index_2016, June 2017. Triplett , Ruth , Randy Gainey , and Ivan Sun . 2003 . “ Institutional Strength, Social Control and Neighborhood Crime Rates .” Theoretical Criminology 7 ( 4 ): 439 – 67 . Google Scholar CrossRef Search ADS Varese , Frederico . 2000 . “Pervasive Corruption.” In Economic Crime in Russia , edited by Alena Ledeneva and Marina Kurkchiyan , 99 – 111 . London : Kluwer Law International . Vaughan , Diane . 1999 . “ The Dark Side of Organizations: Mistake, Misconduct, and Disaster .” Annual Review of Sociology 25 : 271 – 305 . Google Scholar CrossRef Search ADS Verba , Sydney , and Norman Nie . 1972 . Participation in America: Political Participation and Social Equality . New York : Harper and Row . Wacquant , Loïc . 2008 . Urban Outcasts: A Comparative Sociology of Advanced Marginality . Cambridge : Polity . Weisberg , Sanford . 2005 . Applied Linear Regression . Vol. 528 . Hoboken, NJ : John Wiley & Sons . Google Scholar CrossRef Search ADS Werner , Cynthia . 2000 . “ Gifts, Bribes, and Development in Post Soviet Kazakstan .” Human Organization 59 ( 1 ): 11 – 22 . Google Scholar CrossRef Search ADS Wiatrowski , Michael , David Griswold , and Mary Roberts . 1981 . “ Social Control Theory and Delinquency .” American Sociological Review 46 ( 5 ): 525 – 41 . Google Scholar CrossRef Search ADS Yang , Mayfair . 1994 . Gifts, Favors, and Banquets: The Art of Social Relationships in China . Ithaca, NY : Cornell University Press . Zlobina , Tamara . 2007 . “Cultural Markers of Ukrainian Public Space. Mixture and Instability: The City of Lviv Case.” European Humanities University Center for Advanced Studies and Education, accessed online at http://old.ehu.lt/uploads/files/periodicalissue/docs/Digest_2_2007 %20small_51ee90a2e2c7a.pdf#page=85, October 26, 2017. Author notes This article is based upon work supported by the US Army Research Laboratory and the US Army Research Office under grant number W911NF-14-1-0541. We are grateful for the contributions of Jungmin Song, the assistance of Yue Hu and Jenny Juehring, and helpful comments of Michael Sauder, Karen Heimer, and participants of Theory Workshop at the University of Iowa Sociology Department. © The Author(s) 2018. Published by Oxford University Press on behalf of the University of North Carolina at Chapel Hill. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Social Forces Oxford University Press

Pathways to Corruption: Institutional Context and Citizen Participation in Bureaucratic Corruption

Loading next page...
 
/lp/ou_press/pathways-to-corruption-institutional-context-and-citizen-participation-wQM9qqWqGM
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the University of North Carolina at Chapel Hill. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.
ISSN
0037-7732
eISSN
1534-7605
D.O.I.
10.1093/sf/soy007
Publisher site
See Article on Publisher Site

Abstract

Abstract Though bureaucratic corruption is widespread, social scientists have yet to develop a comprehensive model predicting ordinary people’s engagement in corrupt exchanges with street-level bureaucrats. Our article fills this gap by specifying an individual-level causal model of bureaucratic corruption centered around three theoretically derived predictors: beliefs about acceptability of corruption, its perceived riskiness, and its utility to the offender. In doing so, we develop a theory of how institutional stability affects rates and causal pathways to bureaucratic corruption. Using the data from nationally representative surveys in Russia (N = 2,000) and Ukraine (N = 1,535), we test a path-diagramed structural equation model that accounts for endogeneity and the relationships among the theoretically derived predictors of corruption. The tests of our model in institutionally more stable Russia and less stable Ukraine, combined with OLS regression models on split samples from conflict areas within each country, show that when institutions have greater stability, (1) citizens are less likely to partake in bureaucratic corruption, and (2) the effects of theoretically derived predictors on corruption behavior are stronger. We are the first to incorporate into a theory of bureaucratic corruption a feedback loop whereby engagement in corruption reinforces perceptions of its spread, which in turn contributes to the belief that bribery is acceptable. Our findings explain the variation in corruption behavior among citizens of high-corruption societies, extend the institutional theory of social control, testify to the importance of contextualizing theories of crime, and reveal the need to tailor anti-corruption policies to specific institutional environments. Bureaucratic corruption1 refers to the abuse of bureaucratic office for personal gain, whereby the employees of street-level organizations receive unsanctioned compensation for performing their job-related duties or extending additional, extralegal benefits to their clients. Such corruption is difficult to study: it is hidden, stigmatized, and often illegal. Thus, despite the negative economic consequences of corruption,2 we know little about its causes. While existing studies identify country-level predictors of corruption (e.g., Pellegrini 2008), put forth experimental tests of its select determinants (e.g., Graeff et al. 2014), or describe it using anecdotal data (e.g., Osipian 2009), to date, no study has advanced a comprehensive model of bureaucratic corruption at the level of the individual. We fill this gap by developing and testing a causal model that predicts citizens’ participation in bribery exchanges with street-level bureaucrats in high-corruption societies.3 We identify major theoretical predictors of corruption based on criminological theory, experimental work, and case studies. We then specify and show support for a structural equation model of bureaucratic corruption that considers the context of exchanges and stipulates causal relationships among determinants. In doing so, we develop a theory of how institutions influence bureaucratic corruption. We argue that institutional context is particularly relevant to low-level bribery because it often involves cooperation with virtual strangers. Inasmuch as participants lack prior knowledge of their exchange partners, they are particularly likely to draw on institutional rules and norms when deciding whether to participate in bureaucratic corruption. In contrast to other studies, however, we go beyond the analysis of how specific institutions shape people’s corruption conduct. We show that the stability of the institutional context, in and of itself, impacts actors’ causal pathways to this crime. Using new survey data from Russia and Ukraine, we conduct cross-national and within-country statistical tests of the determinants of bureaucratic corruption. As two high-corruption societies4 with markedly different institutional contexts, Russia and Ukraine provide valuable cases for our study. Previous research shows that bureaucratic corruption is commonplace in a variety of organizations in both countries, including hospitals and clinics (Rivkin-Fish 2005), universities and secondary schools (Osipian 2009), and police control posts (Gerber and Mendelson 2008). At the same time, Russia’s institutions are more stable than their Ukrainian counterparts. Both countries also have internal variation in institutional stability, associated with regionally concentrated military conflicts. The estimation of our model reveals that when institutions are more stable, (1) citizens are less likely to partake in bureaucratic corruption, and (2) the predictors of corrupt behavior, such as beliefs and utility assessments, have stronger effects. These findings have important implications for sociological theories of crime and for anti-corruption policy: they explain the variation in corrupt behavior among citizens of high-corruption societies, extend the social control theory of crime, and testify to the importance of contextualizing the theories of law-breaking. Main Predictors of Citizen Engagement in Bureaucratic Corruption The relevant literature suggests three predictors of corrupt behavior: beliefs about the acceptability of corruption, perceived riskiness of corrupt behavior, and its utility to the offender. First, sociological theories of crime often feature beliefs and attitudes as prominent predictors. According to Sutherland’s (1947) differential association theory, for instance, people commit crime because they are exposed to more definitions that are favorable to the violation of law than viewpoints that are unfavorable to crime. Social learning theories (Akers 2011) further suggest that acquired cultural schema translate into concrete behaviors, which are then sustained through positive reinforcement. Empirical studies confirm that individuals are more likely to break the law when exposed to crime-favorable definitions (Apel and Paternoster 2009; Jackall 1988). Correspondingly, popular attitudes are central to many explanations of corruption. For instance, formal theorists maintain that past experiences form beliefs about the economic environment, which in turn contribute to a culture of corruption (Sah 1988). Case studies of the former Soviet Union also suggest that normative systems accepting of corruption formed during the Soviet era, when the planned economy was characterized by an overbearing bureaucratic apparatus and shortages of consumer goods. In this context, relationship-based transactions known as blat emerged as widely accepted means for procuring goods and services that offered ordinary citizens a personalized, flexible, and effective alternative to rigid and politicized institutions. Blat was framed in terms of mutual assistance and support in the face of adversity; as a community-reliant and community-fostering mode of sustenance rather than a subversion of the public good by individuals (Ledeneva 1998; Rehn and Taalas 2004). In the economic dislocations of the post-communist era, popular justifications for bureaucratic corruption persisted (Karklins 2005; Polese 2008). Surveys confirm that citizens are more likely to engage in bureaucratic corruption when they do not define it as wrong (Tavits 2010). We therefore expect that people’s beliefs affect their corruption behavior: Hypothesis 1. The more acceptable citizens think that bribery and gift-giving are, the more likely they are to engage in bureaucratic corruption. Our second predictor of bureaucratic corruption, perceived risk of punishment, comes from the work of criminologists. Crime deterrence strategies are usually based on manipulating the scale, certainty, and speed of punishment (Cohen and Felson 1979; Kleck et al. 2005). Scholars of white-collar crime and corruption also acknowledge the impact of “criminal opportunity” on prevalence of law-breaking (Benson and Simpson 2015). Along these lines, Sajó (2003) links post-Soviet bribery to ineffective and politicized policing in the region and Karklins (2002) attributes corruption to the lack of oversight and transparency. Our second hypothesis reflects this school of thought in criminological research: Hypothesis 2. The lower citizens’ perceived risks of detection and punishment, the more likely they are to engage in bureaucratic corruption. The third major predictor of corruption stems from citizens’ rational calculus. When trying to obtain desired services, citizens weigh the costs and benefits of different strategies. Previous research shows that decisions to bribe a bureaucrat reflect citizens’ evaluations of how much they need a specific service, whether there are other ways to get it, and what a bureaucrat expects in return (McMann 2014; Polese 2008). We assess the effect of these determinants on citizens’ behavior in two corollary hypotheses. The first corollary hypothesis considers the role of extortion. Studies suggest that citizens of high-corruption societies often experience requests for bribes5 from employees of service-provision organizations (Grødeland, Koshechkina, and Miller 1998; Karklins 2002). Faced with such requests, clients may conclude that bribery is the only way to receive the desired service; they may also feel too uncomfortable or intimidated to refuse. In fact, Tavits (2010) finds that officials’ requests of bribes are the strongest predictor of having paid a bribe. Thus, we expect that such requests—whether explicit or deduced from hints, reticence, or delays in service—shape citizens’ behavior: Hypothesis 3a. When officials request bribes, citizens are more likely to engage in bureaucratic corruption. Yet, existing studies reveal that requests from officials cannot explain all bureaucratic corruption. Polese (2008) argues that Ukrainians often choose to engage in under-the-table transactions with bureaucrats even when bribes are not explicitly requested. Karklins (2002) writes that post-communist citizens often initiate corrupt exchanges with traffic cops, Rivkin-Fish (2005) documents voluntary bribery payments to doctors, and Werner (2000) mentions that bribes are offered in exchange for forged military documents. In Werner’s (2000, 11) words, bribery “can be useful for ‘greasing the machine’s wheels’ or speeding things up.” To capture the value that corrupt transactions hold for citizens distinct from officials’ requests, we develop a notion of utility of bureaucratic corruption and incorporate it into our second corollary hypothesis. This notion stems from another classic theory in sociological criminology, Anomie Theory (Merton 1968), whereby crime offers a way to achieve socially defined goals in the absence of legitimate avenues of achievement. Crime can facilitate economic redistribution (Gambetta 1996), perform employment functions (Deuchar 2009), and provide welfare (Wacquant 2008). Some also argue that bureaucratic corruption is functional in underdeveloped societies because it allows ineffective organizations to deliver services to citizens (Becquart-Leclercq 1989; Nye 1967). Case studies reveal that such corruption generates private markets, allowing citizens to satisfy their consumption goals in the absence of viable legal alternatives (McMann 2014; Smith 2010; Yang 1994). Economic rationality asserts that behavior emerges from a cost-benefit calculation, which means that the value of the service received in exchange for a bribe is evaluated vis-à-vis its expense. Existing research, however, focuses primarily on the benefits of illicit exchanges, without reference to their costs. Our study is the first to propose that an appropriate predictor of corrupt behavior is its utility, understood as citizens’ perceived need to resort to corruption relative to their financial capabilities. Hypothesis 3b. As the utility of the transaction increases, the likelihood of engaging in bureaucratic corruption also increases. Hypothesized Relationships among Key Concepts A theoretically sound model of bureaucratic corruption must account for interactions among its key predictors and acknowledge the cyclicality of the corruption process (Andvig and Moene 1990). In this section, we develop additional hypotheses about inter-relations among people’s ideas about acceptability, utility, riskiness, and perceptions of corruption in society. To date, individual-level studies of bureaucratic corruption have not fully specified nor employed the statistical models that accommodate the interactions among its predictors. The few existing large-N analyses use regression models that assume independence among predictor variables and do not account for feedback loops (Miller, Grødeland, and Koshechkina 2001; Tavits 2010). Our theory and analyses represent the first-ever specification and test of non-recursive structural equation models of bureaucratic corruption. Our first hypothesized interaction emerges from the recognition that attitudes shape people’s cost-and-benefit calculations in regards to rule-breaking (Vaughan 1999). Ashforth and Anand (2003) show that beliefs that corruption is normal go hand-in-hand with claims that corruption is necessary to achieve goals. The authors also argue that “once corruption is normative, it may accrue symbolic rewards, such as status and self-esteem, in addition to the utilitarian rewards” (2003, 13). Beliefs about acceptability of corruption may also reduce its perceived costs. Since breaking social norms may be difficult and uncomfortable (Posner 1997), the costs of bribery may be lower for people who believe that corruption is normal. Our next hypothesis reflects the expectation of a positive relationship between perceived acceptability and utility of corruption: Hypothesis 4a. The more acceptable citizens think that bribery and gift-giving are, the higher utility they attribute to bureaucratic corruption. Moreover, scholars argue that corruption is a collective action problem: the behavior of others impacts citizens’ own decisions to partake in corruption (Mungiu-Pippidi 2013; Persson, Rothstein, and Teorell 2013). We build on these arguments by specifying the pathways whereby perceptions lead to bribery behavior. We expect perceptions of corruption in society at large to influence individual bribery behavior in part by how people assess its risks. When people perceive that others are giving and accepting bribes with impunity, they are more likely to conclude that they would not be punished should they also do so (Čábelková and Hanousek 2004). Formal theorists exploring the concept of multiple equilibria in the incidence of corruption also demonstrate that as perceived incidence of corruption in society increases, the probability of getting caught in a corrupt transaction decreases (Andvig and Moene 1990). The idea that people gauge the riskiness of a behavior from observing others is, in fact, central to general deterrence theory, widely accepted in criminology (Homel 2012; Sunshine and Tyler 2003). Hypothesis 4b. The more corruption citizens perceive in society, the lower their perceived risks of detection and punishment. Acknowledging the cyclicality of the corruption process, we expect that perceptions of corruption in society are not exogenous, but, rather, depend on citizens’ own experiences. Reisinger, Zaloznaya, and Claypool (2017) report that when citizens encounter corruption in an interaction with a low-level bureaucrat, they are likely to infer the corruptibility of other officials. Our next hypothesis, therefore, accounts for a feedback loop that links corrupt behavior and its determinants to perceptions of corruption in society. This expectation raises the need to test our theory with a non-recursive structural equation model that allows for causation to flow in more than one direction. Hypothesis 4c. The more citizens engage in bribery and gift-giving, the more corruption they perceive in society. Institutional Stability and Bureaucratic Corruption Criminologists have long recognized that context shapes crime. For instance, neighborhood characteristics, such as “poverty, residential mobility, ethnic heterogeneity, and weak social networks,” have been shown to increase street crime (Kubrin and Weitzer 2003, 374; Rosenfeld, Baumer, and Messner 2001). In this article, we focus on institutions as a salient contextual determinant of causal pathways to bureaucratic corruption. We define institutions as systems of rules or rule-like prescriptions that exercise normative and instrumental pressure toward compliance and that are shared and embedded in existing social structures (North 1990; Ostrom and Basurto 2011). Social control theory of crime underscores the importance of institutional context (Hirschi 1969; Kornhauser 1978). According to this theory, when bonds that link individuals to other people and institutions weaken, individuals are more likely to break the law. Criminologists have demonstrated empirically that attachment to institutions—for instance, through work and marriage—inhibits the likelihood of criminal behavior among adults (Triplett, Gainey, and Sun 2003; Wiatrowski, Griswold, and Roberts 1981). We believe that institutional context is especially pertinent to bureaucratic corruption because such corruption rests on cooperation between social actors who have limited prior experience with each other. Since clients and bureaucrats usually do not have regularized relationships outside service-provision organizations, they are unlikely to share trust and reciprocity. Socio-legal scholars argue that such social distance increases actors’ reliance on institutionalized norms in their interactions with each other (Merry 1993; Ellickson 1986). Social exchange theory (SET) also suggests that people are likely to deduce the rules of exchange from the “normative definition of the situation” in contexts of uncertainty and risk, when wrong choices are especially costly (Emerson 1976, 351). In corrupt transactions, the risk of non-reciprocity is high because the mechanisms for ensuring that the other party delivers on the bargain, such as courts, are unavailable, and because one of the parties might report the exchange to authorities, triggering sanctions that range from stigma to criminal punishment. To mitigate these risks, social actors are particularly likely to invoke institutionalized norms. While others have studied how particular institutions, such as corruption-conducive norms (Köbis et al. 2015) or anti-corruption laws (Peisakhin and Pinto 2010), affect the prevalence of corruption, we argue that the stability of institutions, in and of itself, affects the rates of bureaucratic corruption and shapes how individuals come to participate in this crime. We conceptualize the notion of institutional stability as an ideal type, which anchors one end of what we see as a continuous variable. Such stability is associated with absence of frequent, erratic, or inconsistent social change. Importantly, when institutions are stable, actors are able to know what the behavioral norms are in a particular situation at a specific point in time. When norms are “knowable,” actors are able to predict, with a degree of confidence, the likely consequences of different courses of action. By contrast, when institutions are unstable, rules change frequently, rapidly, and erratically, making it harder for citizens to anticipate what will be considered right and wrong at any point in time. The knowability of “the rules of the game” does not always depend on their correspondence to formal laws and regulations. Sometimes, behavioral norms do reflect codified rules, which may make their knowability more straightforward. However, when “rules of the game” do diverge from formal regulations, institutional stability enables actors to infer behavioral expectations with a degree of certainty. Actors make such inferences from their own past experiences or from past experiences of others, discovered through conversations, hearsay, and observation. The less stable the institutional environment, the lower the predictive power of past experiences for the future behavior. It is important to note that institutional stability does not imply the immutability of behavioral norms, as institutions evolve and change even in the most stable of contexts. For example, law and society scholars show that law is constantly “in action”—evolving through contestation by social actors (Shamir 2004), interaction with other laws and customs (Merry 1988), and appropriation by organizations (Edelman 1992). When institutions are stable, however, this ongoing evolution does not undermine the knowability of rules as citizens develop reliable expectations about the pace and the direction of institutional change. In less stable settings, where social change is more abrupt and irregular, actors have difficulty predicting just how quickly, and in what way, the rules of the game may change. Also, in stable institutional settings, the rules that govern behavior have knowable limits of pertinence relative to other, co-extant rules. In any environment, a variety of rule-like prescriptions governs social action: for instance, different laws, religious postulates, and family norms all regulate one’s conduct at a family celebration of a religious holiday. Some norms might even conflict with one another; yet, with greater institutional stability, people better understand how far each rule extends. The degree of institutional stability has important implications for citizen behavior in bureaucracies. When it is high, citizens are confident about what needs to be done to obtain services, and what distinguishes appropriate from inappropriate conduct in bureaucratic organizations. In stable contexts, citizens have reliable expectations regarding the behavior of their exchange partners; for example, the probability that a bribe is expected and the likelihood of punishment in case of detection. In contrast, when institutional stability is low, the rules are variant and ambiguous, and controls on organizational behavior function unpredictably. Formal theorists link institutional stability to political stability and to fewer corrupt bureaucrats: the “higher the probability of a regime shift…the higher is the incidence of corruption” (Andvig and Moene 1990, 68). Increases in political uncertainty, which are associated with lower probabilities of getting caught for corrupt transactions, cause bureaucrats to become more prone to corruption (71). Additionally, weak destabilized governments are less able to prevent the establishment of independent corruption rackets (Bardhan 1997, 1325). We therefore hypothesize that the overall prevalence of corruption is lower in stable environments because the probabilities of getting caught are lower and because ordinary citizens have fewer incentives to resort to illicit strategies. Also, in contexts with institutional stability, law enforcement agencies and accountability systems prosecute law-breaking more consistently. Hypothesis 5. The more stable the institutional context, the lower the levels of citizen engagement in bureaucratic corruption. Even though overall levels of corrupt behavior are lower when institutions are more stable, we expect stronger causal effects from bribery predictors in stable environs. Stability allows citizens to anticipate how organizations will function and, therefore, rely more fully on beliefs, utility calculations, and risk assessments in making corruption-related decisions. In contrast, in less stable systems, the uncertainty about “how things work” makes it difficult to assess the risks and utility of rule-breaking. Hypothesis 6. The magnitude of the effects that acceptability beliefs, assessments of risk, and utility calculations have on bureaucratic corruption is greater when institutions are more stable. Figure 1 offers a graphic representation of our theoretically derived model of corruption behavior.6 Figure 1. View largeDownload slide Theoretical model of corruption behavior with utility Figure 1. View largeDownload slide Theoretical model of corruption behavior with utility Variation in Institutional Stability and Selection of Cases In comparing Russia and Ukraine, we explore how bureaucratic corruption is influenced by institutional stability at both the national and subnational levels. As indicated above, institutional stability is tied to political stability. One type of political instability that is particularly relevant to our cases is regime change,7 which is often accompanied by erratic and partial reforms, coexistence of multiple regulatory systems, and resource-deficiency, all of which undermine the ability of actors to anticipate the “rules of the game” (Spicer, McDermott, and Kogut 2000). In contrast, when a polity is more stable, “members of society restrict themselves to the behavior patterns that fall within the limits imposed by…expectations” (Ake 1975, 273). Russia, in the post-Soviet era, has been more stable, politically and institutionally, than Ukraine, which has undergone multiple regime changes. Russia has had essentially the same government between 2000 and 2017 (Geddes, Wright, and Frantz 2014). After President Putin’s ascent to power at the turn of the century, he retained a heavily controlled and steady system of governance, avoided risky overhauls, and contained destabilizing reforms through state capture, intimidation of the opposition, and extensive ideological propaganda. This political stability enhances ordinary Russians’ knowledge of the behavioral norms, their boundaries, and the likely pace and direction of their change (Forrat 2015; Gel’man 2015). Although higher than in Ukraine, Russia’s institutional stability is far from absolute. The knowability of the “rules of the game” in Russia is partially undermined by the “duality” of its institutions (Hendley 2017). Russia’s “formal constitutional order” coexists with “a second world of informal relations, factional conflict, and para-constitutional political practices” (Sakwa 2010, 185). This duality, however, is in itself knowable: failures of law, most often, happen in cases that involve political or economic elites and other influential actors (Gel’man 2004; Ledeneva 2008). Ordinary Russians, in contrast, can usually predict what is expected of them in street-level bureaucracies. Unlike Russia, Ukraine has oscillated between distinct forms of political governance since acquiring independence from the USSR. It experienced a pro-democratic “Orange Revolution” in 2004, which brought Viktor Yushchenko to presidency, but turned back to more autocratic governance under Viktor Yanukovych (Kuzio 2015). In 2013, a new wave of popular protests brought to power a pro-Western administration of Petro Poroshenko. This volatility of Ukraine’s government reduced its effectiveness in monitoring bureaucracies and allowed for more deviations from the agreed-upon patterns of behavior. Bardhan (1997, 1325) argues that weak central government creates a kind of economic warlordism, where different ministries, agencies, and levels of local government all set their own norms of exchange. Given this institutional instability, Ukrainians have a harder time predicting—and agreeing on their predictions of—the rules of behavior in bureaucracies. Scholars have described the institutional landscape of Ukraine as ambivalent (Riabchuk 2007), its population as having a divided mentality (Fimyar 2010), and its culture as “an unstable mixture of…ideologies” (Zlobina 2007, 96). In sum, despite their geographic and cultural proximity and centuries of intertwined history, since the breakup of the Soviet Union, Russia and Ukraine have pursued different political trajectories with distinct consequences for their institutional contexts. The two countries, therefore, offer good cases for a “paired comparison” of the effects of institutional stability on bribery behavior. Paired comparison provides the analytical leverage of similarity and contrast and “allows analysts to use differences in institutional form as a variable to demonstrate the sources of intrasystemic behaviors” (Tarrow 2010). No matter how well paired, however, no two case studies can produce fully generalizable conclusions. To achieve additional analytical power, we complement our cross-country comparison with within-country analyses. Both Russia and Ukraine are marked by significant regional heterogeneity. On the subnational level, each country has regions that are afflicted with military conflict that results in institutional instability. Military conflicts increase out-migration, create chaos, and detract resources from public infrastructure—all of which upset institutional stability (Bellows and Miguel 2006). Four distinct conflicts have destabilized select Russian and Ukrainian regions. First, fueled by ethnic and territorial disputes, frozen conflicts have developed in Russia’s North Caucasus region, most notably in Chechnya (Hesli 2007). Second, confrontations with Russia’s neighbors, such as the Republic of Georgia, have contributed to endemic instability in the adjacent regions (Cornell and Starr 2009). Third, Russia’s annexation of the Crimean Peninsula in March 2014 affected regions in both Russia and Ukraine. The prolonged military presence and the emergency state in the peninsula have undermined the institutional stability in neighboring areas (Birnbaum and Demirjian 2015). Fourth, regions in both countries suffer from the military conflict with Russia-backed separatists in Eastern Ukraine. In our analysis, we link these regionally concentrated conflicts and the associated subnational variation in institutional instability to rates of and pathways to corrupt behavior. By comparing more and less institutionally stable regions within Russia and Ukraine, as well as comparing the two countries with each other, we increase the number of comparison cases, and raise the generalizability of our findings. Data and Measures Our data come from representative national surveys conducted in Russia and Ukraine in June–July 2015. Face-to-face interviews were conducted with 2,000 respondents in Russia and 1,535 in Ukraine (see appendix A for more detail on the surveys). To measure bureaucratic corruption, we started with a query as to whether the respondent had contact with any of the following officials in the past 12 months: Officials of the judicial system (judges, clerks, justice department officials, lawyers, prosecutors) Doctors, nurses, medical workers, hospital administrators Inspectors (in health, construction, food quality, sanitary control, and licensing) Officials issuing certificates and permits (marriage, death, birth certificates; construction permits) Officials in housing and communal services Officials who issue governmental tenders Politicians at the local level (mayors, local executive heads) Professors, instructors in higher educational institutions Police officers Tax officials Teachers, school administrators Traffic police Other (respondent names the official) We then asked to which of these officials our respondents had given a bribe, a present, or done a favor in exchange for a service. In Russia, 83 percent of respondents had contact with at least one of these officials, and 18 percent of those who had contact paid a bribe (gave a gift or did a favor). In Ukraine, 77 percent had contact with at least one official, and 29 percent of those paid a bribe, made a gift, or did a favor.8 Each individual’s Corruption Behavior is the ratio of the number of bribes paid (as well as gifts given or favors made) to the number of contacts with officials. The ratio ranges from 1 (every contact resulted in a bribe) to 0.1 (no bribes),9 with a mean score of 0.37 for Russia (on average, over a third of contacts in Russia result in corrupt behavior) and 0.51 for Ukraine (on average, half of all contacts with officials in Ukraine result in corruption). Appendix B contains descriptive information on our measures. To measure beliefs, we asked whether it is acceptable to do any of the following to receive a faster or higher-quality service from officials: To give money To give a gift To do a favor Responses are coded from 1, never acceptable, to 3, always acceptable. We average responses to each behavior to create a scale of Belief in Acceptability of Bribery. In Ukraine, the average score on this variable is 1.6; in Russia, it is 1.7. Each person was also asked about the Spread of Corruption throughout the country. Specifically, respondents were asked whether over the past three years the level of corruption had increased a lot, increased a little, stayed the same, decreased a little, or decreased a lot. We coded responses from 1 to 5, with higher scores associated with increased corruption. The average score on this array is 3.3 for Russia and 3.9 for Ukraine. To measure risk of detection and punishment, we asked a series of questions on how likely the respondent thought people were to get into legal trouble for offering bribes or gifts to employees of Hospitals and clinics, Traffic police, Registration and tender offices, Universities, and Secondary schools. Responses range from 1, very likely, to 4, very unlikely. We average evaluations of these five bureaucracies to create a summary scale of each respondent’s perception of the Risk of Punishment if one engages in bureaucratic corruption. In Russia, the average score on this measure is 2.9; in Ukraine, it is 3.3. To measure the utility of bureaucratic corruption, we combine perceptions of the need to pay bribes with an evaluation of personal ability to finance them. An individual’s perception of the need to engage in bureaucratic corruption is determined by responses to a question about why ordinary people pay bribes to officials. The following options were offered: Because it is the easiest or fastest way to achieve the results Because bureaucrats demand bribes from people Because it is normal; everybody around them does it Because people feel grateful to the official Because they feel sorry for the official (they want to help him/her) We considered answers 1 or 2 as indicators of perceived need for corrupt conduct. These responses were aggregated depending on whether these answers were selected as the most important, the second most important, or the least important reason. We then measured respondents’ capacities to pay bribes based on their financial statuses. Thus, Utility of Bribery measures perception of need weighted by financial capacity. Our measure of utility ranges from 0.0 to 7.5, with averages of 3.5 in Russia and 2.4 in Ukraine. To indicate regionally concentrated institutional instability, we created a dummy variable for whether or not the respondent resides in or near a region marked by any of the military conflicts described above. Nineteen percent of the Russia sample and 39 percent of the Ukraine sample reside in a Conflict Area. To measure officials’ requests of bribes, a battery of questions asked whether each official with whom the respondent had contact had asked for a bribe, a present, or a favor in exchange for services. Official Requested a Bribe for each individual is the ratio of the number of times an official asked for a bribe to the number of contacts with officials he or she had. The average scores in Russia and Ukraine are 0.38 and 0.49, respectively, indicating a greater likelihood in Ukraine than in Russia of being asked by an official for a bribe. Also included in the analysis are demographic predictors that have been previously linked to differences in behavior: age, gender, education, and rural residence (Miller, Grødeland, and Koshechkina 2001; O’Brien and Wegren 2002; Tavits 2010; Verba and Nie 1972). Our structural equation model links demographics to beliefs about acceptability and to utility, as a vast literature evidences differences in attitudes based upon these characteristics (Hesli et al. 2001). Analyses and Results We start by testing the impact of institutional stability on engagement in bureaucratic corruption (H5). We argue that Russia will have lower levels of corrupt behavior than Ukraine because Russia has more institutional stability. To test, we calculate the average likelihood of engagement in bureaucratic corruption (given contact with an official) in Russia and Ukraine. The first columns of table 1 (and the corresponding t-test value for comparing the two means) reveal that engagement in bureaucratic corruption occurs at a higher rate in Ukraine (0.505) than in Russia (0.371). Importantly, in both countries, living in or adjacent to a conflict area significantly increases one’s likelihood of engaging in corruption (last four columns of table 1). Since conflict areas have more unstable institutions, this finding provides preliminary support for hypothesis 5 that in more stable institutional contexts levels of bribery behavior are lower. Table 1. Difference of Means Test for Corruption Behavior by Institutional Context All of Ukraine All of Russia Russia Ukraine Conflict areas Non-conflict areas Conflict areas Non-conflict areas Average level of corrupt behavior (means) 0.505 0.371 0.415 0.361 0.522 0.495 Number of cases 1,180 1,665 316 1,349 455 725 T value for comparing the two means 16.5123 4.6923 1.9686 All of Ukraine All of Russia Russia Ukraine Conflict areas Non-conflict areas Conflict areas Non-conflict areas Average level of corrupt behavior (means) 0.505 0.371 0.415 0.361 0.522 0.495 Number of cases 1,180 1,665 316 1,349 455 725 T value for comparing the two means 16.5123 4.6923 1.9686 Table 1. Difference of Means Test for Corruption Behavior by Institutional Context All of Ukraine All of Russia Russia Ukraine Conflict areas Non-conflict areas Conflict areas Non-conflict areas Average level of corrupt behavior (means) 0.505 0.371 0.415 0.361 0.522 0.495 Number of cases 1,180 1,665 316 1,349 455 725 T value for comparing the two means 16.5123 4.6923 1.9686 All of Ukraine All of Russia Russia Ukraine Conflict areas Non-conflict areas Conflict areas Non-conflict areas Average level of corrupt behavior (means) 0.505 0.371 0.415 0.361 0.522 0.495 Number of cases 1,180 1,665 316 1,349 455 725 T value for comparing the two means 16.5123 4.6923 1.9686 Next, we evaluate the proposition that the magnitudes of the effects of acceptability beliefs, assessments of risk, and utility calculations on bureaucratic corruption are greater when institutions are more stable (H6). To test this proposition, we employ ordinary least squares (OLS) regression on split samples in Russia and Ukraine (see table 2). By conducting our analyses separately in Russia and Ukraine, we see how national context affects predictors of bureaucratic corruption. By analyzing independently, and comparing, conflict and non-conflict areas in each country, we also assess the effect of regional instability on corruption determinants.10 We present two models in this table; the first (model 1) includes Utility as a predictor, and the second (model 2) contains Official Requests Bribe. Table 2. OLS Regression of Corruption Behavior upon Theoretical Predictors Dependent variable: Corruption behavior Model 1 (with utility) Model 2 (with official requested bribe) Russia Ukraine Russia Ukraine Predictor variables Non-conflict area Conflict area Non-conflict area Conflict zone area Non-conflict area Conflict area Non-conflict area Conflict area Belief in acceptability 0.045*** 0.071*** 0.040*** 0.002 0.025*** 0.028* 0.038*** 0.019 (0.009) (0.024) (0.015) (0.021) (0.006) (0.015) (0.011) (0.014) Punishment risk low 0.029*** 0.051*** 0.044*** 0.007 0.007 0.016 0.002 −0.008 (0.007) (0.018) (0.013) (0.016) (0.005) (0.012) (0.010) (0.011) Utility 0.008*** 0.001 −0.0001 0.016 (0.003) (0.008) (0.007) (0.010) Spread of corruption 0.012** 0.034*** 0.033*** 0.008 −0.002 0.007 0.020*** 0.004 (0.005) (0.012) (0.008) (0.011) (0.003) (0.007) (0.006) (0.007) Rural −0.026** −0.055** 0.041** 0.094*** −0.002 −0.028* 0.026** 0.063*** (0.012) (0.028) (0.018) (0.033) (0.008) (0.016) (0.013) (0.022) Age 0.001*** 0.0003 0.001 0.001 0.0004** −0.001* 0.0003 −0.0001 (0.0003) (0.001) (0.001) (0.001) (0.0002) (0.001) (0.0004) (0.0005) Female −0.027*** 0.006 0.012 −0.006 −0.003 0.004 0.025** 0.013 (0.009) (0.025) (0.018) (0.025) (0.006) (0.015) (0.013) (0.017) Education −0.011*** −0.0003 −0.0002 −0.005 −0.003 0.003 0.001 0.001 (0.003) (0.007) (0.005) (0.006) (0.002) (0.004) (0.004) (0.004) Official requested bribe 0.672*** 0.813*** 0.732*** 0.730*** (0.017) (0.037) (0.031) (0.033) Constant 0.180*** 0.020 0.112 0.404*** 0.047* −0.0004 −0.036 0.102 (0.037) (0.109) (0.078) (0.093) (0.025) (0.066) (0.057) (0.064) Observations 1307 312 715 435 1347 316 723 455 R2 0.074 0.105 0.054 0.036 0.558 0.648 0.475 0.535 Adjusted R2 0.068 0.081 0.043 0.018 0.556 0.639 0.470 0.527 Residual std. error (df) 0.167 0.214 0.220 0.239 0.115 0.134 0.164 0.165 (1298) (303) (706) (426) (1338) (307) (714) (446) F statistic (df) 12.971*** 4.445*** 5.010*** 1.979** 211.501*** 70.566*** 80.880*** 64.169*** (8; 1298) (8; 303) (8; 706) (8; 426) (8; 1338) (8; 307) (8; 714) (8; 446) Dependent variable: Corruption behavior Model 1 (with utility) Model 2 (with official requested bribe) Russia Ukraine Russia Ukraine Predictor variables Non-conflict area Conflict area Non-conflict area Conflict zone area Non-conflict area Conflict area Non-conflict area Conflict area Belief in acceptability 0.045*** 0.071*** 0.040*** 0.002 0.025*** 0.028* 0.038*** 0.019 (0.009) (0.024) (0.015) (0.021) (0.006) (0.015) (0.011) (0.014) Punishment risk low 0.029*** 0.051*** 0.044*** 0.007 0.007 0.016 0.002 −0.008 (0.007) (0.018) (0.013) (0.016) (0.005) (0.012) (0.010) (0.011) Utility 0.008*** 0.001 −0.0001 0.016 (0.003) (0.008) (0.007) (0.010) Spread of corruption 0.012** 0.034*** 0.033*** 0.008 −0.002 0.007 0.020*** 0.004 (0.005) (0.012) (0.008) (0.011) (0.003) (0.007) (0.006) (0.007) Rural −0.026** −0.055** 0.041** 0.094*** −0.002 −0.028* 0.026** 0.063*** (0.012) (0.028) (0.018) (0.033) (0.008) (0.016) (0.013) (0.022) Age 0.001*** 0.0003 0.001 0.001 0.0004** −0.001* 0.0003 −0.0001 (0.0003) (0.001) (0.001) (0.001) (0.0002) (0.001) (0.0004) (0.0005) Female −0.027*** 0.006 0.012 −0.006 −0.003 0.004 0.025** 0.013 (0.009) (0.025) (0.018) (0.025) (0.006) (0.015) (0.013) (0.017) Education −0.011*** −0.0003 −0.0002 −0.005 −0.003 0.003 0.001 0.001 (0.003) (0.007) (0.005) (0.006) (0.002) (0.004) (0.004) (0.004) Official requested bribe 0.672*** 0.813*** 0.732*** 0.730*** (0.017) (0.037) (0.031) (0.033) Constant 0.180*** 0.020 0.112 0.404*** 0.047* −0.0004 −0.036 0.102 (0.037) (0.109) (0.078) (0.093) (0.025) (0.066) (0.057) (0.064) Observations 1307 312 715 435 1347 316 723 455 R2 0.074 0.105 0.054 0.036 0.558 0.648 0.475 0.535 Adjusted R2 0.068 0.081 0.043 0.018 0.556 0.639 0.470 0.527 Residual std. error (df) 0.167 0.214 0.220 0.239 0.115 0.134 0.164 0.165 (1298) (303) (706) (426) (1338) (307) (714) (446) F statistic (df) 12.971*** 4.445*** 5.010*** 1.979** 211.501*** 70.566*** 80.880*** 64.169*** (8; 1298) (8; 303) (8; 706) (8; 426) (8; 1338) (8; 307) (8; 714) (8; 446) Note: Cells show estimated unstandardized coefficients with standard errors in parentheses. *p < 0.1 **p < 0.05 ***p < 0.01 Table 2. OLS Regression of Corruption Behavior upon Theoretical Predictors Dependent variable: Corruption behavior Model 1 (with utility) Model 2 (with official requested bribe) Russia Ukraine Russia Ukraine Predictor variables Non-conflict area Conflict area Non-conflict area Conflict zone area Non-conflict area Conflict area Non-conflict area Conflict area Belief in acceptability 0.045*** 0.071*** 0.040*** 0.002 0.025*** 0.028* 0.038*** 0.019 (0.009) (0.024) (0.015) (0.021) (0.006) (0.015) (0.011) (0.014) Punishment risk low 0.029*** 0.051*** 0.044*** 0.007 0.007 0.016 0.002 −0.008 (0.007) (0.018) (0.013) (0.016) (0.005) (0.012) (0.010) (0.011) Utility 0.008*** 0.001 −0.0001 0.016 (0.003) (0.008) (0.007) (0.010) Spread of corruption 0.012** 0.034*** 0.033*** 0.008 −0.002 0.007 0.020*** 0.004 (0.005) (0.012) (0.008) (0.011) (0.003) (0.007) (0.006) (0.007) Rural −0.026** −0.055** 0.041** 0.094*** −0.002 −0.028* 0.026** 0.063*** (0.012) (0.028) (0.018) (0.033) (0.008) (0.016) (0.013) (0.022) Age 0.001*** 0.0003 0.001 0.001 0.0004** −0.001* 0.0003 −0.0001 (0.0003) (0.001) (0.001) (0.001) (0.0002) (0.001) (0.0004) (0.0005) Female −0.027*** 0.006 0.012 −0.006 −0.003 0.004 0.025** 0.013 (0.009) (0.025) (0.018) (0.025) (0.006) (0.015) (0.013) (0.017) Education −0.011*** −0.0003 −0.0002 −0.005 −0.003 0.003 0.001 0.001 (0.003) (0.007) (0.005) (0.006) (0.002) (0.004) (0.004) (0.004) Official requested bribe 0.672*** 0.813*** 0.732*** 0.730*** (0.017) (0.037) (0.031) (0.033) Constant 0.180*** 0.020 0.112 0.404*** 0.047* −0.0004 −0.036 0.102 (0.037) (0.109) (0.078) (0.093) (0.025) (0.066) (0.057) (0.064) Observations 1307 312 715 435 1347 316 723 455 R2 0.074 0.105 0.054 0.036 0.558 0.648 0.475 0.535 Adjusted R2 0.068 0.081 0.043 0.018 0.556 0.639 0.470 0.527 Residual std. error (df) 0.167 0.214 0.220 0.239 0.115 0.134 0.164 0.165 (1298) (303) (706) (426) (1338) (307) (714) (446) F statistic (df) 12.971*** 4.445*** 5.010*** 1.979** 211.501*** 70.566*** 80.880*** 64.169*** (8; 1298) (8; 303) (8; 706) (8; 426) (8; 1338) (8; 307) (8; 714) (8; 446) Dependent variable: Corruption behavior Model 1 (with utility) Model 2 (with official requested bribe) Russia Ukraine Russia Ukraine Predictor variables Non-conflict area Conflict area Non-conflict area Conflict zone area Non-conflict area Conflict area Non-conflict area Conflict area Belief in acceptability 0.045*** 0.071*** 0.040*** 0.002 0.025*** 0.028* 0.038*** 0.019 (0.009) (0.024) (0.015) (0.021) (0.006) (0.015) (0.011) (0.014) Punishment risk low 0.029*** 0.051*** 0.044*** 0.007 0.007 0.016 0.002 −0.008 (0.007) (0.018) (0.013) (0.016) (0.005) (0.012) (0.010) (0.011) Utility 0.008*** 0.001 −0.0001 0.016 (0.003) (0.008) (0.007) (0.010) Spread of corruption 0.012** 0.034*** 0.033*** 0.008 −0.002 0.007 0.020*** 0.004 (0.005) (0.012) (0.008) (0.011) (0.003) (0.007) (0.006) (0.007) Rural −0.026** −0.055** 0.041** 0.094*** −0.002 −0.028* 0.026** 0.063*** (0.012) (0.028) (0.018) (0.033) (0.008) (0.016) (0.013) (0.022) Age 0.001*** 0.0003 0.001 0.001 0.0004** −0.001* 0.0003 −0.0001 (0.0003) (0.001) (0.001) (0.001) (0.0002) (0.001) (0.0004) (0.0005) Female −0.027*** 0.006 0.012 −0.006 −0.003 0.004 0.025** 0.013 (0.009) (0.025) (0.018) (0.025) (0.006) (0.015) (0.013) (0.017) Education −0.011*** −0.0003 −0.0002 −0.005 −0.003 0.003 0.001 0.001 (0.003) (0.007) (0.005) (0.006) (0.002) (0.004) (0.004) (0.004) Official requested bribe 0.672*** 0.813*** 0.732*** 0.730*** (0.017) (0.037) (0.031) (0.033) Constant 0.180*** 0.020 0.112 0.404*** 0.047* −0.0004 −0.036 0.102 (0.037) (0.109) (0.078) (0.093) (0.025) (0.066) (0.057) (0.064) Observations 1307 312 715 435 1347 316 723 455 R2 0.074 0.105 0.054 0.036 0.558 0.648 0.475 0.535 Adjusted R2 0.068 0.081 0.043 0.018 0.556 0.639 0.470 0.527 Residual std. error (df) 0.167 0.214 0.220 0.239 0.115 0.134 0.164 0.165 (1298) (303) (706) (426) (1338) (307) (714) (446) F statistic (df) 12.971*** 4.445*** 5.010*** 1.979** 211.501*** 70.566*** 80.880*** 64.169*** (8; 1298) (8; 303) (8; 706) (8; 426) (8; 1338) (8; 307) (8; 714) (8; 446) Note: Cells show estimated unstandardized coefficients with standard errors in parentheses. *p < 0.1 **p < 0.05 ***p < 0.01 The estimation result for model 1 of table 2 reveals that when institutions are more stable, the three main predictors identified in hypotheses 1a, 2, and 3 (Acceptability, Risk, and Utility) influence bureaucratic bribery in the hypothesized fashion. In non-conflict regions of institutionally stable Russia, all of these predictors are significantly associated with corruption behavior, while in conflict regions of Russia, and less institutionally stable Ukraine, Utility does not hold predictive significance. The magnitudes of the estimated coefficients relative to their standard errors for Belief in Acceptability and Risk of Punishment Low are greater in the more stable environments in comparison to those that are less stable.11 These findings lend preliminary support to hypothesis 6. When Official Requested a Bribe is added to the model and Utility is excluded (model 2 of table 2), officials’ demands for bribes emerge as, by far, the best predictor of bureaucratic corruption. These results demonstrate that once a provider demands a bribe, the client often has little option but to pay. According to the R2 statistic, we explain 65 percent of variance in Corruption Behavior in conflict areas of Russia when we account for whether or not an Official Requested a Bribe. Nevertheless, some variance remains to be explained by Belief in Acceptability in both conflict and non-conflict regions of Russia and in non-conflict regions of Ukraine. Although table 2 reveals the importance of hypothesized predictors as well as differences in the magnitudes of their effects in more and less stable institutional contexts, the OLS model provides only a rudimentary tool for assessing our hypotheses. Such a model assumes that predictor variables are uncorrelated with one another and that each is exogenous from the dependent variable (Weisberg 2005). Yet, these assumptions are often untrue in reality. Indeed, we have hypothesized that the more people engage in corruption, the more widespread they perceive corruption to be (H4c). Thus, we expect that variation in the dependent variable actually causes change in perceived spread of corruption, which the OLS model treats as exogenous. A correctly specified model must allow for this endogeneity. Additionally, we expect interactions among some predictors of corruption behavior, while OLS assumes their independence. Moreover, our statistical model needs to account for the fact that demographics influence corruption directly as well as through its perceived acceptability and utility. To accommodate these theoretical expectations, we fit the model represented in figure 1 using structural equation modeling (SEM). SEM estimates simultaneously a number of direct and indirect paths to corruption behavior. For example, our theory states that beliefs about acceptability affect behavior both directly (H1) and indirectly through utility (H4a). The OLS modeling is able to capture the acceptability-to-behavior path, but it cannot simultaneously estimate the acceptability-utility-behavior path. The multi-path patterns modeled by SEM better represent the complexity of the social world and have greater ecological validity. SEM is also appropriate for our analyses because our theory expects endogenous relationships: we not only model corruption behavior as influencing the perceived spread of corruption but also allow these perceptions to affect beliefs about acceptability and risks of corruption (see figure 1). The results from testing the SEM model are presented in tables 3 and 4.12 Table 3 includes Utility as a predictor, while in table 4, Utility is replaced with Official Requested a Bribe. Both tables involve multiple equations estimated simultaneously. In equation 1 of table 3, Utility, Acceptability, Risk, and demographics are the direct predictors of variation in Corruption Behavior (equation 1 resembles the single equation in table 2). The results of this more sophisticated estimation method (equation 1 of table 3) reveal that in both institutionally stable Russia and less stable Ukraine, these three predictors do significantly influence the likelihood of corruption behavior. Table 3. Structural Equation Model of Corruption Behavior with Utility Included Equation 1: Corruption behavior Russia (unstandardized) Ukraine (unstandardized) Russia (standardized) Ukraine (standardized) Belief in acceptability 0.410** 0.027* 1.201 0.066 (0.135) (0.015) Risk of punishment low 0.021* −0.229** 0.078 −0.680 (0.011) (0.088) Utility 0.017** 0.011* 0.151 0.058 (0.009) (0.006) Rural 0.004 −0.008 0.010 −0.016 (0.020) (0.014) Age 0.003*** 0.000 0.245 0.030 (0.001) (0.000) Female −0.003 0.017 −0.007 0.034 (0.015) (0.012) Education −0.023*** 0.003 −0.230 0.023 (0.006) (0.003) Equation 2: Belief in acceptability Spread of corruption −3.040*** 0.048** −5.561 0.086 (0.780) (0.016) Conflict area 0.359** −0.058 0.262 −0.049 (0.142) (0.036) Rural −0.303** −0.142*** −0.236 −0.115 (0.123) (0.038) Age −0.002 −0.003** −0.065 −0.092 (0.002) (0.001) Female −0.162** −0.054 −0.147 −0.045 (0.078) (0.035) Education 0.001 0.000 0.002 0.001 (0.022) (0.010) Equation 3: Risk of punishment low Spread of corruption 0.095*** 4.864** 0.135 7.260 (0.017) (2.246) Conflict area 0.132** −1.718** 0.075 −1.210 (0.043) (0.758) Rural −0.212*** −0.502 −0.130 −0.337 (0.040) (0.339) Equation 4: Utility Belief in acceptability 0.309** 0.213** 0.104 0.097 (0.156) (0.062) Conflict area 0.240** −0.172** 0.059 −0.067 (0.102) (0.075) Rural −0.463*** −0.134* −0.122 −0.050 (0.095) (0.081) Age −0.005* −0.010*** −0.047 −0.145 (0.003) (0.002) Female −0.290*** −0.272*** −0.089 −0.103 (0.080) (0.074) Education 0.100*** 0.104*** 0.115 0.154 (0.022) (0.020) Equation 5: Spread of corruption Corruption behavior 1.397** 6.797** 0.261 1.532 (0.698) (2.457) Model statistics Number of observations 1619 1150 Estimator ML ML Chi-square function test 12.405 16.295 Degrees of freedom 9 9 P-value (chi-square) 0.191 0.061 CFI 0.991 0.968 SRMR 0.012 0.017 PNFI 0.25 0.241 RMSEA 0.015 0.027 90 Percent CI of RMSEA [0.000, 0.034] [0.000, 0.047] P-value RMSEA ≤0.05 1.000 0.974 Equation 1: Corruption behavior Russia (unstandardized) Ukraine (unstandardized) Russia (standardized) Ukraine (standardized) Belief in acceptability 0.410** 0.027* 1.201 0.066 (0.135) (0.015) Risk of punishment low 0.021* −0.229** 0.078 −0.680 (0.011) (0.088) Utility 0.017** 0.011* 0.151 0.058 (0.009) (0.006) Rural 0.004 −0.008 0.010 −0.016 (0.020) (0.014) Age 0.003*** 0.000 0.245 0.030 (0.001) (0.000) Female −0.003 0.017 −0.007 0.034 (0.015) (0.012) Education −0.023*** 0.003 −0.230 0.023 (0.006) (0.003) Equation 2: Belief in acceptability Spread of corruption −3.040*** 0.048** −5.561 0.086 (0.780) (0.016) Conflict area 0.359** −0.058 0.262 −0.049 (0.142) (0.036) Rural −0.303** −0.142*** −0.236 −0.115 (0.123) (0.038) Age −0.002 −0.003** −0.065 −0.092 (0.002) (0.001) Female −0.162** −0.054 −0.147 −0.045 (0.078) (0.035) Education 0.001 0.000 0.002 0.001 (0.022) (0.010) Equation 3: Risk of punishment low Spread of corruption 0.095*** 4.864** 0.135 7.260 (0.017) (2.246) Conflict area 0.132** −1.718** 0.075 −1.210 (0.043) (0.758) Rural −0.212*** −0.502 −0.130 −0.337 (0.040) (0.339) Equation 4: Utility Belief in acceptability 0.309** 0.213** 0.104 0.097 (0.156) (0.062) Conflict area 0.240** −0.172** 0.059 −0.067 (0.102) (0.075) Rural −0.463*** −0.134* −0.122 −0.050 (0.095) (0.081) Age −0.005* −0.010*** −0.047 −0.145 (0.003) (0.002) Female −0.290*** −0.272*** −0.089 −0.103 (0.080) (0.074) Education 0.100*** 0.104*** 0.115 0.154 (0.022) (0.020) Equation 5: Spread of corruption Corruption behavior 1.397** 6.797** 0.261 1.532 (0.698) (2.457) Model statistics Number of observations 1619 1150 Estimator ML ML Chi-square function test 12.405 16.295 Degrees of freedom 9 9 P-value (chi-square) 0.191 0.061 CFI 0.991 0.968 SRMR 0.012 0.017 PNFI 0.25 0.241 RMSEA 0.015 0.027 90 Percent CI of RMSEA [0.000, 0.034] [0.000, 0.047] P-value RMSEA ≤0.05 1.000 0.974 Note: Standard errors are in parentheses. *p < 0.1 **p < 0.05 ***p < 0.001. Five post-estimation tests are applied to examine the goodness of fit of the model, chi-square, Comparative Fit Index (CFI), Standardized Root Mean Square Residual (SRMR), Parsimonious Normed Fit Index (PNFI), and Root Mean Square Error of Approximation (RMSEA). Chi-square p-value: A good model fit would provide an insignificant result at a 0.05 threshold (Barrett 2007). RMSEA: A value of RMSEA <0.06 is recognized as indicative of good fit (Hu and Bentler 1999). CFI: A value of CFI ≥0.95 is presently recognized as indicative of good fit (Hu and Bentler 1999). SRMR: Values for the SRMR range from zero to 1.0 with well-fitting models obtaining values less than 0.05 (Diamantopoulos and Siguaw 2000). Table 3. Structural Equation Model of Corruption Behavior with Utility Included Equation 1: Corruption behavior Russia (unstandardized) Ukraine (unstandardized) Russia (standardized) Ukraine (standardized) Belief in acceptability 0.410** 0.027* 1.201 0.066 (0.135) (0.015) Risk of punishment low 0.021* −0.229** 0.078 −0.680 (0.011) (0.088) Utility 0.017** 0.011* 0.151 0.058 (0.009) (0.006) Rural 0.004 −0.008 0.010 −0.016 (0.020) (0.014) Age 0.003*** 0.000 0.245 0.030 (0.001) (0.000) Female −0.003 0.017 −0.007 0.034 (0.015) (0.012) Education −0.023*** 0.003 −0.230 0.023 (0.006) (0.003) Equation 2: Belief in acceptability Spread of corruption −3.040*** 0.048** −5.561 0.086 (0.780) (0.016) Conflict area 0.359** −0.058 0.262 −0.049 (0.142) (0.036) Rural −0.303** −0.142*** −0.236 −0.115 (0.123) (0.038) Age −0.002 −0.003** −0.065 −0.092 (0.002) (0.001) Female −0.162** −0.054 −0.147 −0.045 (0.078) (0.035) Education 0.001 0.000 0.002 0.001 (0.022) (0.010) Equation 3: Risk of punishment low Spread of corruption 0.095*** 4.864** 0.135 7.260 (0.017) (2.246) Conflict area 0.132** −1.718** 0.075 −1.210 (0.043) (0.758) Rural −0.212*** −0.502 −0.130 −0.337 (0.040) (0.339) Equation 4: Utility Belief in acceptability 0.309** 0.213** 0.104 0.097 (0.156) (0.062) Conflict area 0.240** −0.172** 0.059 −0.067 (0.102) (0.075) Rural −0.463*** −0.134* −0.122 −0.050 (0.095) (0.081) Age −0.005* −0.010*** −0.047 −0.145 (0.003) (0.002) Female −0.290*** −0.272*** −0.089 −0.103 (0.080) (0.074) Education 0.100*** 0.104*** 0.115 0.154 (0.022) (0.020) Equation 5: Spread of corruption Corruption behavior 1.397** 6.797** 0.261 1.532 (0.698) (2.457) Model statistics Number of observations 1619 1150 Estimator ML ML Chi-square function test 12.405 16.295 Degrees of freedom 9 9 P-value (chi-square) 0.191 0.061 CFI 0.991 0.968 SRMR 0.012 0.017 PNFI 0.25 0.241 RMSEA 0.015 0.027 90 Percent CI of RMSEA [0.000, 0.034] [0.000, 0.047] P-value RMSEA ≤0.05 1.000 0.974 Equation 1: Corruption behavior Russia (unstandardized) Ukraine (unstandardized) Russia (standardized) Ukraine (standardized) Belief in acceptability 0.410** 0.027* 1.201 0.066 (0.135) (0.015) Risk of punishment low 0.021* −0.229** 0.078 −0.680 (0.011) (0.088) Utility 0.017** 0.011* 0.151 0.058 (0.009) (0.006) Rural 0.004 −0.008 0.010 −0.016 (0.020) (0.014) Age 0.003*** 0.000 0.245 0.030 (0.001) (0.000) Female −0.003 0.017 −0.007 0.034 (0.015) (0.012) Education −0.023*** 0.003 −0.230 0.023 (0.006) (0.003) Equation 2: Belief in acceptability Spread of corruption −3.040*** 0.048** −5.561 0.086 (0.780) (0.016) Conflict area 0.359** −0.058 0.262 −0.049 (0.142) (0.036) Rural −0.303** −0.142*** −0.236 −0.115 (0.123) (0.038) Age −0.002 −0.003** −0.065 −0.092 (0.002) (0.001) Female −0.162** −0.054 −0.147 −0.045 (0.078) (0.035) Education 0.001 0.000 0.002 0.001 (0.022) (0.010) Equation 3: Risk of punishment low Spread of corruption 0.095*** 4.864** 0.135 7.260 (0.017) (2.246) Conflict area 0.132** −1.718** 0.075 −1.210 (0.043) (0.758) Rural −0.212*** −0.502 −0.130 −0.337 (0.040) (0.339) Equation 4: Utility Belief in acceptability 0.309** 0.213** 0.104 0.097 (0.156) (0.062) Conflict area 0.240** −0.172** 0.059 −0.067 (0.102) (0.075) Rural −0.463*** −0.134* −0.122 −0.050 (0.095) (0.081) Age −0.005* −0.010*** −0.047 −0.145 (0.003) (0.002) Female −0.290*** −0.272*** −0.089 −0.103 (0.080) (0.074) Education 0.100*** 0.104*** 0.115 0.154 (0.022) (0.020) Equation 5: Spread of corruption Corruption behavior 1.397** 6.797** 0.261 1.532 (0.698) (2.457) Model statistics Number of observations 1619 1150 Estimator ML ML Chi-square function test 12.405 16.295 Degrees of freedom 9 9 P-value (chi-square) 0.191 0.061 CFI 0.991 0.968 SRMR 0.012 0.017 PNFI 0.25 0.241 RMSEA 0.015 0.027 90 Percent CI of RMSEA [0.000, 0.034] [0.000, 0.047] P-value RMSEA ≤0.05 1.000 0.974 Note: Standard errors are in parentheses. *p < 0.1 **p < 0.05 ***p < 0.001. Five post-estimation tests are applied to examine the goodness of fit of the model, chi-square, Comparative Fit Index (CFI), Standardized Root Mean Square Residual (SRMR), Parsimonious Normed Fit Index (PNFI), and Root Mean Square Error of Approximation (RMSEA). Chi-square p-value: A good model fit would provide an insignificant result at a 0.05 threshold (Barrett 2007). RMSEA: A value of RMSEA <0.06 is recognized as indicative of good fit (Hu and Bentler 1999). CFI: A value of CFI ≥0.95 is presently recognized as indicative of good fit (Hu and Bentler 1999). SRMR: Values for the SRMR range from zero to 1.0 with well-fitting models obtaining values less than 0.05 (Diamantopoulos and Siguaw 2000). Table 4. Structural Equation Model of Corruption Behavior with Official Requested a Bribe Included Equation 1: Corruption behavior Russia (unstandardized) Ukraine (unstandardized) Russia (standardized) Ukraine (standardized) Official requested bribe 0.707*** 0.729*** 0.735 0.688 (0.016) (0.026) Belief in acceptability 0.031*** 0.031*** 0.091 0.077 (0.007) (0.009) Risk of punishment low 0.010** 0.026 0.040 0.077 (0.004) (0.037) Rural −0.003 0.037** −0.006 0.074 (0.007) (0.011) Age 0.000 0.000 0.019 0.010 (0.000) (0.000) Female −0.000 0.019* −0.001 0.039 (0.006) (0.010) Education −0.002 0.001 −0.017 0.007 (0.002) (0.003) Equation 2: Belief in acceptability Spread of corruption 0.384** 0.047** 0.703 0.084 (0.125) (0.020) Conflict area 0.111** −0.056 0.081 −0.048 (0.033) (0.035) Rural −0.084** −0.134*** −0.065 −0.109 (0.031) (0.037) Age −0.005*** −0.003** −0.161 −0.086 (0.001) (0.001) Female −0.038 −0.055 −0.034 −0.046 (0.026) (0.035) Education 0.034*** 0.001 0.114 0.002 (0.007) (0.009) Equation 3: Risk of punishment low Spread of corruption 0.065** −0.880 0.093 −1.319 (0.027) (1.282) Conflict area 0.136** −0.192*** 0.077 −0.136 (0.043) (0.053) Rural −0.220*** −0.111** −0.134 −0.075 (0.040) (0.046) Equation 4: Spread of corruption Official requested bribe 0.694*** 0.412** 0.134 0.088 (0.126) (0.135) Equation 5: Official requested bribe Conflict area 0.025** 0.056*** 0.051 0.125 (0.012) (0.013) Rural −0.043*** 0.016 −0.095 0.035 (0.011) (0.014) Age 0.001** 0.001** 0.083 0.085 (0.000) (0.000) Female −0.039*** −0.024* −0.099 −0.052 (0.009) (0.013) Model statistics Number of observations 1663 1178 Estimator ML ML Chi-square function test 48.314 29.591 Degrees of freedom 10 10 P-value (chi-square) 0.000 0.001 CFI 0.978 0.979 SRMR 0.02 0.022 PNFI 0.278 0.277 RMSEA 0.048 0.041 90 percent CI [0.035, 0.062] [0.024, 0.058] P-value RMSEA ≤0.05 0.568 0.794 Equation 1: Corruption behavior Russia (unstandardized) Ukraine (unstandardized) Russia (standardized) Ukraine (standardized) Official requested bribe 0.707*** 0.729*** 0.735 0.688 (0.016) (0.026) Belief in acceptability 0.031*** 0.031*** 0.091 0.077 (0.007) (0.009) Risk of punishment low 0.010** 0.026 0.040 0.077 (0.004) (0.037) Rural −0.003 0.037** −0.006 0.074 (0.007) (0.011) Age 0.000 0.000 0.019 0.010 (0.000) (0.000) Female −0.000 0.019* −0.001 0.039 (0.006) (0.010) Education −0.002 0.001 −0.017 0.007 (0.002) (0.003) Equation 2: Belief in acceptability Spread of corruption 0.384** 0.047** 0.703 0.084 (0.125) (0.020) Conflict area 0.111** −0.056 0.081 −0.048 (0.033) (0.035) Rural −0.084** −0.134*** −0.065 −0.109 (0.031) (0.037) Age −0.005*** −0.003** −0.161 −0.086 (0.001) (0.001) Female −0.038 −0.055 −0.034 −0.046 (0.026) (0.035) Education 0.034*** 0.001 0.114 0.002 (0.007) (0.009) Equation 3: Risk of punishment low Spread of corruption 0.065** −0.880 0.093 −1.319 (0.027) (1.282) Conflict area 0.136** −0.192*** 0.077 −0.136 (0.043) (0.053) Rural −0.220*** −0.111** −0.134 −0.075 (0.040) (0.046) Equation 4: Spread of corruption Official requested bribe 0.694*** 0.412** 0.134 0.088 (0.126) (0.135) Equation 5: Official requested bribe Conflict area 0.025** 0.056*** 0.051 0.125 (0.012) (0.013) Rural −0.043*** 0.016 −0.095 0.035 (0.011) (0.014) Age 0.001** 0.001** 0.083 0.085 (0.000) (0.000) Female −0.039*** −0.024* −0.099 −0.052 (0.009) (0.013) Model statistics Number of observations 1663 1178 Estimator ML ML Chi-square function test 48.314 29.591 Degrees of freedom 10 10 P-value (chi-square) 0.000 0.001 CFI 0.978 0.979 SRMR 0.02 0.022 PNFI 0.278 0.277 RMSEA 0.048 0.041 90 percent CI [0.035, 0.062] [0.024, 0.058] P-value RMSEA ≤0.05 0.568 0.794 Note: See note to table 3. Table 4. Structural Equation Model of Corruption Behavior with Official Requested a Bribe Included Equation 1: Corruption behavior Russia (unstandardized) Ukraine (unstandardized) Russia (standardized) Ukraine (standardized) Official requested bribe 0.707*** 0.729*** 0.735 0.688 (0.016) (0.026) Belief in acceptability 0.031*** 0.031*** 0.091 0.077 (0.007) (0.009) Risk of punishment low 0.010** 0.026 0.040 0.077 (0.004) (0.037) Rural −0.003 0.037** −0.006 0.074 (0.007) (0.011) Age 0.000 0.000 0.019 0.010 (0.000) (0.000) Female −0.000 0.019* −0.001 0.039 (0.006) (0.010) Education −0.002 0.001 −0.017 0.007 (0.002) (0.003) Equation 2: Belief in acceptability Spread of corruption 0.384** 0.047** 0.703 0.084 (0.125) (0.020) Conflict area 0.111** −0.056 0.081 −0.048 (0.033) (0.035) Rural −0.084** −0.134*** −0.065 −0.109 (0.031) (0.037) Age −0.005*** −0.003** −0.161 −0.086 (0.001) (0.001) Female −0.038 −0.055 −0.034 −0.046 (0.026) (0.035) Education 0.034*** 0.001 0.114 0.002 (0.007) (0.009) Equation 3: Risk of punishment low Spread of corruption 0.065** −0.880 0.093 −1.319 (0.027) (1.282) Conflict area 0.136** −0.192*** 0.077 −0.136 (0.043) (0.053) Rural −0.220*** −0.111** −0.134 −0.075 (0.040) (0.046) Equation 4: Spread of corruption Official requested bribe 0.694*** 0.412** 0.134 0.088 (0.126) (0.135) Equation 5: Official requested bribe Conflict area 0.025** 0.056*** 0.051 0.125 (0.012) (0.013) Rural −0.043*** 0.016 −0.095 0.035 (0.011) (0.014) Age 0.001** 0.001** 0.083 0.085 (0.000) (0.000) Female −0.039*** −0.024* −0.099 −0.052 (0.009) (0.013) Model statistics Number of observations 1663 1178 Estimator ML ML Chi-square function test 48.314 29.591 Degrees of freedom 10 10 P-value (chi-square) 0.000 0.001 CFI 0.978 0.979 SRMR 0.02 0.022 PNFI 0.278 0.277 RMSEA 0.048 0.041 90 percent CI [0.035, 0.062] [0.024, 0.058] P-value RMSEA ≤0.05 0.568 0.794 Equation 1: Corruption behavior Russia (unstandardized) Ukraine (unstandardized) Russia (standardized) Ukraine (standardized) Official requested bribe 0.707*** 0.729*** 0.735 0.688 (0.016) (0.026) Belief in acceptability 0.031*** 0.031*** 0.091 0.077 (0.007) (0.009) Risk of punishment low 0.010** 0.026 0.040 0.077 (0.004) (0.037) Rural −0.003 0.037** −0.006 0.074 (0.007) (0.011) Age 0.000 0.000 0.019 0.010 (0.000) (0.000) Female −0.000 0.019* −0.001 0.039 (0.006) (0.010) Education −0.002 0.001 −0.017 0.007 (0.002) (0.003) Equation 2: Belief in acceptability Spread of corruption 0.384** 0.047** 0.703 0.084 (0.125) (0.020) Conflict area 0.111** −0.056 0.081 −0.048 (0.033) (0.035) Rural −0.084** −0.134*** −0.065 −0.109 (0.031) (0.037) Age −0.005*** −0.003** −0.161 −0.086 (0.001) (0.001) Female −0.038 −0.055 −0.034 −0.046 (0.026) (0.035) Education 0.034*** 0.001 0.114 0.002 (0.007) (0.009) Equation 3: Risk of punishment low Spread of corruption 0.065** −0.880 0.093 −1.319 (0.027) (1.282) Conflict area 0.136** −0.192*** 0.077 −0.136 (0.043) (0.053) Rural −0.220*** −0.111** −0.134 −0.075 (0.040) (0.046) Equation 4: Spread of corruption Official requested bribe 0.694*** 0.412** 0.134 0.088 (0.126) (0.135) Equation 5: Official requested bribe Conflict area 0.025** 0.056*** 0.051 0.125 (0.012) (0.013) Rural −0.043*** 0.016 −0.095 0.035 (0.011) (0.014) Age 0.001** 0.001** 0.083 0.085 (0.000) (0.000) Female −0.039*** −0.024* −0.099 −0.052 (0.009) (0.013) Model statistics Number of observations 1663 1178 Estimator ML ML Chi-square function test 48.314 29.591 Degrees of freedom 10 10 P-value (chi-square) 0.000 0.001 CFI 0.978 0.979 SRMR 0.02 0.022 PNFI 0.278 0.277 RMSEA 0.048 0.041 90 percent CI [0.035, 0.062] [0.024, 0.058] P-value RMSEA ≤0.05 0.568 0.794 Note: See note to table 3. To assess differences in the magnitudes of effects, we report standardized coefficients in tables 3 and 4. We confirm H6 for two of the three predictors: the magnitudes of the effects that acceptability beliefs and utility have on bureaucratic corruption are greater in the more stable institutions of Russia.13 For risk of punishment, the effect on corruption behavior is positive in Russia and negative in Ukraine.14 Equation 1 of table 3 also describes direct effects of demographic characteristics on the likelihood of engaging in bureaucratic corruption. The most significant finding is that in institutionally unstable Ukraine, none of the demographics significantly influence the likelihood of corrupt behavior. In stable Russia, however, older people are more likely to engage in bureaucratic corruption, while higher education is associated with lower chances of such engagement. Next, equation 2 in table 3, predicting Belief in Acceptability of Bribery, allows individual demographic characteristics, conflict area, and spread of corruption to influence corrupt behavior indirectly through beliefs about its acceptability (see also figure 1). We see that in Ukraine, the more corruption citizens perceive in society, the more acceptable they think that bribery is, while in Russia perceptions that corruption is increasing decrease beliefs in bribery’s acceptability. Another difference is that in Russia, with its more stable institutions, acceptability is significantly higher in conflict areas, while in less stable Ukraine, acceptability is unrelated to residence in a conflict region. Equation 3 in table 3, in turn, reveals that in both countries, risk of punishment is seen as lower when corruption is perceived to be increasing in society. This confirms hypothesis 4b. Importantly, this equation reveals that national institutional context matters for the relationship between proximity to conflict and risk of punishment. In the country with greater institutional stability, residence in a conflict area increases the perceived risk of detection and punishment. In contrast, in less stable Ukraine, residence in a conflict area decreases the perceived risk of detection and punishment. Next, equation 4 of table 3 for predicting utility supports H4a: beliefs about acceptability of bribery are significantly related to calculations of its utility. Again, this finding confirms the import of modeling the predictors of corrupt behavior in a way that accommodates the interactions among them. Equation 4 again underscores the importance of institutional context for bureaucratic corruption. In institutionally stable Russia, the utility of bureaucratic corruption is greater in conflict areas. In less stable Ukraine, by contrast, additional increases in instability (associated with residing in or near a conflict zone) reduce utility. The relevance of demographic characteristics to utility is, however, the same in Russia and Ukraine: utility is higher in non-rural settings, decreases with age, increases with education, and is lower for females. The last equation in table 3, equation 5, predicts perceptions about the spread of corruption with actual corrupt behavior. This equation incorporates our hypothesized feedback loop, whereby perceptions of corruption affect behavior through their effects on perceived acceptability of corruption (equation 2), while engagement in corruption in turn increases the likelihood of perceiving corruption as increasing in society (equation 5). Note that hypothesis 4c is confirmed in both Russia and Ukraine. In table 4, we test the direct effect of Official Requested a Bribe on bribery behavior in a separate SEM (see figure 2 for graphic representations). Since officials’ requests for bribes account for more than 50 percent of the variance in corruption behavior (from table 2), limited variance is left to be explained by other predictors. Nevertheless, in both stable and unstable institutional contexts, beliefs about the acceptability of bribery still significantly and independently influence the likelihood of corrupt behavior. In addition, in stable Russia, an individual is less likely to engage in bureaucratic corruption, even if asked for a bribe by an official, when perceived risk of punishment is high (equation 1 of table 4). Equation 2 of table 4 suggests that when officials’ requests for bribes are included in the SEM model, beliefs about the acceptability of bribery increase as perceptions of corruption in society rise. In addition, beliefs about the acceptability of bribery are consistently lower while perceived risk of punishment is higher in rural areas. Figure 2. View largeDownload slide Theoretical model of corruption behavior with extortion (official requested bribe) Figure 2. View largeDownload slide Theoretical model of corruption behavior with extortion (official requested bribe) Equation 5 in table 4 also shows where and from whom requests for bribes are most likely to occur: in both Russia and Ukraine, they occur more frequently in conflict areas and when the client is older and male. Discussion Our study reveals the micro-foundations of bureaucratic corruption. Based on sociological theory and case studies of high-corruption societies, we identify the processes behind corrupt behavior among the clients of service-provision organizations. Our statistical tests confirm that the beliefs about acceptability of corruption and the perceived risk of punishment affect the likelihood of citizens’ bribery behavior, whether or not an official has requested a bribe. Utility calculations affect the incidence of bureaucratic corruption distinct from explicit requests for a bribe. To specify and test the effects of these determinants on bribery behavior, our study pioneered a structural equation model, which accommodates multiple direct and indirect paths to the outcome, as well as causal dependencies among endogenous and exogenous variables. We demonstrate empirically that causes of bureaucratic corruption, while conceptually distinct from one another, are not independent in their operation. We are also the first to incorporate into a theory of bureaucratic corruption a feedback loop whereby engagement in corruption reinforces perceptions of its spread, which in turn contributes to the belief that bribery is acceptable. Although the cyclical nature of bureaucratic corruption has been demonstrated by formal theorists (e.g., Andvig and Moene 1990), prior to our analyses, no one had tested such cyclicality empirically. In addition, we offer a new measure: the utility of bribery, which is an improvement on existing measures, as it weighs perceived benefits of bribery against respondents’ financial means. Our main theoretical contribution, however, consists in revealing the effects of institutional stability on the scale and causes of bureaucratic corruption. First, our results reveal significant variance in rates of bureaucratic corruption in Ukraine and Russia despite their comparable scores on Transparency International’s Corruption Perception Index (TICPI). While TICPI (2015) reflects perceptions of the spread of political as well as bureaucratic corruption, our focus specifically on citizens’ reported engagement in corrupt exchanges with street-level bureaucrats reveals notable differences between and within the two countries. Thus, our analyses show that citizens resort to under-the-table transactions with bureaucrats more rarely in relatively stable Russia than they do in destabilized Ukraine; within each country, less stable conflict areas also have higher rates of corrupt engagement than other regions. This finding confirms that institutional instability, in and of itself, has criminogenic effects. Second, we demonstrate that institutional stability affects the strength of theoretically established determinants of corruption: not only do these predictors have greater explanatory power in stable Russia than in unstable Ukraine, but within each country, their impact on behavior diminishes in areas destabilized by military conflict. In other words, beliefs and utility calculations are more closely tied to bribery behavior when people deal with predictable bureaucracies. When institutions are destabilized, beliefs may be in flux and the utility of corruption is difficult to calculate, which diminishes their impact on people’s behavior. Institutions may be destabilized by a number of socio-political processes. Paired comparison of Russia and Ukraine provides preliminary evidence that frequent regime change at the national level, which undermines the stability of institutions, impacts citizens’ pathways to rule-breaking. Within countries, citizens’ pathways to corruption are systematically different in regions beset by military conflict. We see contextual effects of residence in a rural area: the intimacy of rural communities, with their heightened social reciprocity, makes corruption less accepted and its risks higher. Together, these findings at national, regional, and community levels reveal that stability of institutions influences crime. Our findings extend the institutional theory of social control of crime. Social control theorists have shown the importance of belonging to specific organizations, networks, family structures, and associations. We, in turn, have theorized the importance of living in a society ordered by stable institutions. We hypothesized that stable institutions allow rules to be more knowable, thereby reducing crimes, such as corruption. Our empirical results indicate that order begets order: when people are confident about how the social world operates, they are less likely to engage in illegality. Conclusion Every year, the international community spends billions of dollars on fighting corruption in non-Western societies (Sampson 2010). Based on their low rankings on TICPI and other indicators of corruption, academics and policymakers routinely refer to these countries as “pervasively” and “endemically” corrupt (Lovell, Ledeneva, and Rogachevskii 2000; Varese 2000). Such designation implies unvarying levels of corruption and generates unwarranted defeatism regarding the prospects of fighting corruption in these societies. In contrast, our study underscores and explains internal variation in high-corruption societies. Not only do our surveys reveal that in such societies many people regularly abstain from corruption, our model shows why some are more likely to resort to illicit transactions than others. Building on these findings, social scientists can avoid the trap of relegating certain countries to the lower ranks of what has, essentially, emerged as a global moral order. Our findings also underscore the importance of tailoring anti-corruption policy to specific national and subnational contexts. Most policy recommendations that international donors extend to high-corruption societies fall back on strategies that were conceptualized and tested in the Global North (Larmour 2007). The assumption behind such policy transfer is that determinants of corrupt behavior are stable across different contexts. While most critiques of policy adoption focus on differences in governance structures (Minogue 2004), our study underscores another important reason why policies successful elsewhere may fail in a different setting: citizens’ causal pathways to corruption depend on institutional stability. For instance, our data suggest that when institutions are destabilized, anti-corruption strategies that focus on reducing instrumental incentives for bribery may be less effective. In such contexts, more positive outcomes might emerge from policies targeting cultural predictors of corruption—that is, educational and outreach campaigns that focus on social harms of corruption and that raise the stigma associated with engagement. In contrast, when a system is stable but still marked by corruption, anti-corruption strategies that reduce the utility of corruption may have better results than policies focused on educating the citizenry. Ultimately, our findings also have implications for the sociology of knowledge, as they imply that existing theories of corruption (and, likely, other crimes) are contingent on institutional stability. We therefore call into question the generalizability of studies conducted in Western societies and invite more research in contexts with lower institutional stability. Notes 1 In the literature, such corruption also goes under the names of petty, administrative, or low-level corruption (see Karklins [2005] for examples). 2 Corruption distorts the allocation of resources, discourages investment, and impairs economic growth (Dabla-Norris 2000). 3 We define high-corruption countries broadly as countries that cluster in the lower 30th percentile of Transparency International’s corruption rankings (Transparency International 2016). 4 Since 2012, Transparency International (2016) has ranked both Ukraine and Russia in the 20th percentile of the most corrupt countries in the world. 5 In the relevant literature, the word extortion is used to refer to bureaucrats’ requests for bribes (Karklins 2005; Tavits 2010). This word, however, generally implies force or threat, and our survey questions focus on whether an official asked for a bribe, not on whether the request for a bribe was accompanied by a veiled threat or delays and rudeness in service. Therefore, we use the term “official’s request for bribe” instead of “extortion.” 6 To save space, we present a single representation, but we expect that the magnitude of the effects in this model are greater in contexts where institutions are more stable. 7 According to Sanders (1981), regime change encompasses changes in regime norms, types of party system, and/or military-civilian status. 8 Some readers may worry whether respondents are willing to divulge to the interviewer their engagement in bureaucratic corruption. We employed multiple strategies for making respondents feel comfortable in answering our queries honestly (see appendix A, “Safeguards against Social Desirability Bias”). We also used a list experiment to assess the accuracy of responses to questions about respondents’ engagement in corruption. Other researchers have also documented a lack of social desirability bias associated with discussing bureaucratic corruption in post-Soviet countries (McMann 2014; Petrov and Temple 2004). 9 The formula results in a non-zero number representing zero bribes paid because we added a constant to the numerator and denominator of the ratio to avoid dividing by zero. 10 As an alternative to this split-sample approach, we also use interaction terms to test the hypothesis that the relationship between main theoretical predictors and corruption behavior is different in conflict and non-conflict areas (table 2D of appendix D). The results in table 2D reflect those in table 2, and provide confirmation that the effects of Acceptability and Risk are significantly different in conflict and non-conflict zones. 11 Standardized coefficients are calculated by dividing the unstandardized estimated coefficients reported in table 2 by their standard errors (reported in parentheses). The larger standardized coefficient associated with Acceptability and Risk in non-conflict areas as compared with conflict areas indicates that the effects of these variables on corruption behavior are greater (relative to other predictors) in the non-conflict areas. For example, in Russia, the effect of Acceptability (relative to other predictors in the equation) is lower in less stable conflict areas. 12 These relations are simultaneously estimated with covariant residuals among perceived spread of corruption, corruption behavior, risk of punishment, and beliefs about acceptability, as shown in dashed lines in figure 1. The SEM models were estimated by the “sem” function of the lavaan package (0.5–23.1097) on R 3.2.5 (see appendix C). 13 We report standardized coefficients for illustrative purposes only, and understand the perils of comparing these across independent samples. 14 In the OLS regressions (model 1 of table 2), where the effects of the predictor variables affect corruption behavior independently from one another, we see that in conflict areas of Ukraine, Risk is unrelated to Corruption Behavior. In SEM, however, we have a predictive equation (equation 3) embedded into the analysis that allows residence in a conflict area to push risk of punishment higher in conflict areas of Ukraine. This has the effect of creating a negative relationship between low risk of punishment and corruption behavior in Ukraine. Conflict areas of Ukraine are in large part occupied by Russian troops (a foreign occupying army), thus risk of punishment can be very real in these regions. About the Authors Marina Zaloznaya is an Assistant Professor of Sociology at the University of Iowa. Her work is based on ethnographic, survey, and comparative-historical analyses of informal economies in non-democratic political systems, in Eastern Europe and China. Zaloznaya’s publications include a recent book, The Politics of Bureaucratic Corruption in Post-Transitional Eastern Europe (Cambridge University Press, 2017), and articles in a range of sociology and interdisciplinary journals. Vicki Hesli Claypool is Professor Emeritus of the Political Science Department at the University of Iowa. Her research career focused primarily on studying voting behavior and public opinion in Russia and Ukraine using mass surveys based on person-to-person interviews. Most of her recent publications have appeared in PS: Political Science & Politics and address topics associated with success in the academic profession, including job satisfaction, faculty rank attainment, research productivity, and academic salaries. William M. Reisinger is Professor of Political Science at the University of Iowa. He specializes in authoritarianism and democracy in the former communist states, especially Russia. He has written or edited eight books, including The Regional Roots of Russia’s Political Regime (University of Michigan Press, 2017), coauthored with Bryon J. Moraski, as well as numerous articles and book chapters. Supplementary Material Supplementary material is available at Social Forces online. References Ake , Claude . 1975 . “ A Definition of Political Stability .” Comparative Politics 7 ( 2 ): 271 – 83 . Google Scholar CrossRef Search ADS Akers , Ronald . 2011 . Social Learning and Social Structure: A General Theory of Crime and Deviance . New Brunswick, NJ : Transaction Publishers . Andvig , Jens , and Karl Ove Moene . 1990 . “ How Corruption May Corrupt .” Journal of Economic Behavior & Organization 13 ( 1 ): 63 – 76 . Google Scholar CrossRef Search ADS Apel , Robert , and Raymond Paternoster . 2009 . “Understanding ‘Criminogenic’ Corporate Culture: What White-Collar Crime Researchers Can Learn from Studies of the Adolescent Employment–Crime Relationship.” In The Criminology of White-Collar Crime , edited by Sally Simpson and David Weisburd , 15 – 33 . New York : Springer . Google Scholar CrossRef Search ADS Ashforth , Blake , and Vikas Anand . 2003 . “ The Normalization of Corruption in Organizations .” Research in Organizational Behavior 25 : 1 – 52 . Google Scholar CrossRef Search ADS Bardhan , Pranab . 1997 . “ Corruption and Development: A Review of Issues .” Journal of Economic Literature 35 ( 3 ): 1320 – 46 . Barrett , Paul . 2007 . “ Structural Equation Modelling: Adjudging Model Fit .” Personality and Individual Differences 42 ( 5 ): 815 – 24 . Google Scholar CrossRef Search ADS Becquart-Leclercq , Jeanne . 1989 . “Paradoxes of Political Corruption: A French View.” In Political Corruption: A Handbook , edited by Arnold Heidenheimer , Victor LeVine , and Michael Johnston , 19 – 36 . New Brunswick, NJ : Transaction Publishers . Bellows , John , and Edward Miguel . 2006 . “ War and Institutions: New Evidence from Sierra Leone .” American Economic Review 96 ( 2 ): 394 – 99 . Google Scholar CrossRef Search ADS Benson , Michael , and Sally Simpson . 2015 . Understanding White-Collar Crime: An Opportunity Perspective . New York : Taylor & Francis . Birnbaum , Michael , and Karoun Demirjian . 2015 . “Fighting Spreads in Eastern Ukraine after Rebel Rout.” Washington Post, February 20. Čábelková , Inna , and Jan Hanousek . 2004 . “ The Power of Negative Thinking: Corruption, Perception and Willingness to Bribe in Ukraine .” Applied Economics 36 ( 4 ): 383 – 97 . Google Scholar CrossRef Search ADS Cohen , Lawrence , and Marcus Felson . 1979 . “ Social Change and Crime Rate Trends: A Routine Activity Approach .” American Sociological Review 44 ( 4 ): 588 – 608 . Google Scholar CrossRef Search ADS Cornell , Svante , and Frederick Starr , eds. 2009 . The Guns of August: Russia’s War in Georgia . Armonk, NY : M. E. Sharpe . Dabla-Norris , Era . 2000 . A Game-Theoretic Analysis of Corruption in Bureaucracies . Washington, DC : International Monetary Fund . Deuchar , Ross . 2009 . Gangs, Marginalised Youth and Social Capital . Staffordshire, UK : Trentham Books . Diamantopoulos , Adamantios and Judy A. Siguaw . 2000 . Introducing LISREL: A Guide for the Uninitiated. London : Sage. Edelman , Lauren . 1992 . “ Legal Ambiguity and Symbolic Structures: Organizational Mediation of Civil Rights Law .” American Journal of Sociology 97 ( 6 ): 1531 – 76 . Google Scholar CrossRef Search ADS Ellickson , Robert . 1986 . “ Of Coase and Cattle: Dispute Resolution among Neighbors in Shasta County .” Stanford Law Review 38 : 623 – 87 . Google Scholar CrossRef Search ADS Emerson , Richard . 1976 . “ Social Exchange Theory .” Annual Review of Sociology 2 : 335 – 62 . Google Scholar CrossRef Search ADS Fimyar , Olena . 2010 . “ Policy Why(s): Policy Rationalities and the Changing Logic of Educational Reform in Postcommunist Ukraine .” International Perspectives on Education and Society 14 : 61 – 91 . Google Scholar CrossRef Search ADS Forrat , Natalia . 2015 . “The Political Economy of Russian Higher Education: Why Does Putin Support Research Universities?” Post-Soviet Affairs 32 ( 4 ): 1 – 39 . Gambetta , Diego . 1996 . The Sicilian Mafia: The Business of Private Protection . Cambridge, MA : Harvard University Press . Geddes , Barbara , Joseph Wright , and Erica Frantz . 2014 . “ Autocratic Breakdown and Regime Transitions: A New Data Set .” Perspectives on Politics 12 ( 2 ): 313 – 31 . Google Scholar CrossRef Search ADS Gel’man, Vladimir . 2004 . “ The Unrule of Law in the Making: The Politics of Informal Institution Building in Russia .” Europe-Asia Studies 56 ( 7 ): 1021 – 40 . Google Scholar CrossRef Search ADS Gel’man , Vladimir . 2015 . Authoritarian Russia: Analyzing Post-Soviet Regime Changes . Pittsburgh, PA : University of Pittsburgh Press . Google Scholar CrossRef Search ADS Gerber , Theodore , and Sarah Mendelson . 2008 . “ Public Experiences of Police Violence and Corruption in Contemporary Russia: A Case of Predatory Policing? ” Law & Society Review 42 ( 1 ): 1 – 44 . Google Scholar CrossRef Search ADS Graeff , Peter , Sebastian Sattler , Guido Mehlkop , and Carsten Sauer . 2014 . “ Incentives and Inhibitors of Abusing Academic Positions: Analysing University Students’ Decisions About Bribing Academic Staff .” European Sociological Review 30 ( 2 ): 230 – 41 . Google Scholar CrossRef Search ADS Grødeland , Åse , Tatyana Koshechkina , and William Miller . 1998 . “ ‘Foolish to Give and Yet More Foolish Not to Take’: In-Depth Interviews with Post-Communist Citizens on Their Everyday Use of Bribes and Contacts .” Europe-Asia Studies 50 ( 4 ): 651 – 77 . Google Scholar CrossRef Search ADS Hendley , Kathryn . 2017 . Everyday Law in Russia . Ithaca, NY : Cornell University Press . Hesli , Vicki . 2007 . Governments and Politics in Russia and the Post-Soviet Region . Boston : Houghton Mifflin . Hesli , Vicki L. , Ha-Lyong Jung , William M. Reisinger and Arthur H. Miller . 2001 . “ The Gender Divide in Russian Politics: Attitudinal and Behavioral Considerations .” Women & Politics 22 ( 2 ): 41 – 80 . Google Scholar CrossRef Search ADS Hirschi , Travis . 1969 . “A Control Theory of Delinquency.” In Criminology Theory: Selected Classic Readings , edited by Frank Williams and Marylin McShane , 289 – 305 . Cincinnati, OH : Anderson Publishing . Homel , Ross . 2012 . Policing and Punishing the Drinking Driver: A Study of General and Specific Deterrence . New York : Springer . Hu , Li‐tze , and Peter M. Bentler . 1999 . “ Cutoff Criteria for Fit Indexes in Covariance Structure Analysis: Conventional Criteria Versus New Alternatives .” Structural Equation Modeling: A Multidisciplinary Journal 6 ( 1 ): 1 – 5 . Google Scholar CrossRef Search ADS Jackall , Robert . 1988 . “ Moral Mazes: The World of Corporate Managers .” International Journal of Politics, Culture, and Society 1 ( 4 ): 598 – 614 . Google Scholar CrossRef Search ADS Karklins , Rasma . 2002 . “ Typology of Post-Communist Corruption .” Problems of Post-Communism 49 ( 4 ): 22 – 32 . Google Scholar CrossRef Search ADS ——— . 2005 . The System Made Me Do It: Corruption in Post-Communist Societies . Armonk, NY : M. E. Sharpe . Kleck , Gary , Brion Sever , Spencer Li , and Marc Gertz . 2005 . “ The Missing Link in General Deterrence Research .” Criminology 43 ( 3 ): 623 – 60 . Google Scholar CrossRef Search ADS Köbis , Nils , Jan-Willem van Prooijen , Francesca Righetti , and Paul Van Lange . 2015 . “ ‘Who Doesn’t?’—The Impact of Descriptive Norms on Corruption .” PloS One 10 ( 6 ): 1 – 14 . Google Scholar CrossRef Search ADS Kornhauser , Ruth . 1978 . Social Sources of Delinquency: An Appraisal of Analytic Models . Chicago : University of Chicago Press . Kubrin , Charis , and Ronald Weitzer . 2003 . “ New Directions in Social Disorganization Theory .” Journal of Research in Crime and Delinquency 40 ( 4 ): 374 – 402 . Google Scholar CrossRef Search ADS Kuzio , Taras . 2015 . “Ukraine: Leaving the Crossroads.” In Central and East European Politics: From Communism to Democracy , edited by Sharon Wolchik and Jane Curry , 481 – 512 . Lanham, MD : Rowman & Littlefield . Larmour , Peter . 2007 . “ International Action against Corruption in the Pacific Islands: Policy Transfer, Coercion and Effectiveness .” Asian Journal of Political Science 15 ( 1 ): 1 – 16 . Google Scholar CrossRef Search ADS Ledeneva , Alena . 1998 . Russia’s Economy of Favors: Blat, Networking and Informal Exchange . New York : Cambridge University Press . Ledeneva , Alena . 2008 . “ Telephone Justice in Russia .” Post-Soviet Affairs 24 ( 4 ): 324 – 50 . Google Scholar CrossRef Search ADS Lovell , Stephen , Alena Ledeneva , and A. Rogachevskii , eds. 2000 . Bribery and Blat in Russia: Negotiating Reciprocity from the Middle Ages to the 1990s . New York : St. Martin’s Press . McMann , Kelly . 2014 . Corruption as a Last Resort: Adapting to the Market in Central Asia . Ithaca, NY : Cornell University Press . Merry , Sally Engle . 1988 . “ Legal Pluralism .” Law & Society Review 22 ( 5 ): 869 – 96 . Google Scholar CrossRef Search ADS ——— . 1993 . “ Mending Walls and Building Fences: Constructing the Private Neighborhood .” Journal of Legal Pluralism and Unofficial Law 25 ( 33 ): 71 – 90 . Google Scholar CrossRef Search ADS Merton , Robert . 1968 . Social Theory and Social Structure . New York : Simon and Schuster . Miller , William , Ase Grødeland , and Tatyana Koshechkina . 2001 . A Culture of Corruption? Coping with Government in Post-Communist Europe . Budapest : Central European University Press . Minogue , Martin . 2004 . “Public Management and Regulatory Governance: Problems of Policy Transfer to Developing Countries.” In Leading Issues in Competition, Regulation, and Development , edited by Paul Cook , Colin Kirkpatrick , Martin Minogue , and David Parker , 165 – 82 . Cheltenham, UK : Edward Elgar Publishing . Mungiu-Pippidi , Alina . 2013 . “ Controlling Corruption through Collective Action .” Journal of Democracy 24 ( 1 ): 101 – 15 . Google Scholar CrossRef Search ADS North , Douglass . 1990 . Institutions, Institutional Change and Economic Performance . Cambridge : Cambridge University Press . Google Scholar CrossRef Search ADS Nye , Joseph . 1967 . “ Corruption and Political Development: A Cost-Benefit Analysis .” American Political Science Review 61 ( 2 ): 417 – 27 . Google Scholar CrossRef Search ADS O’Brien , David , and Stephen Wegren . 2002 . Rural Reform in Post-Soviet Russia . Washington, DC : Woodrow Wilson Center Press . Osipian , Ararat . 2009 . “ Corruption Hierarchies in Higher Education in the Former Soviet Bloc .” International Journal of Educational Development 29 ( 3 ): 321 – 30 . Google Scholar CrossRef Search ADS Ostrom , Elinor , and Xavier Basurto . 2011 . “ Crafting Analytical Tools to Study Institutional Change .” Journal of Institutional Economics 7 ( 03 ): 317 – 43 . Google Scholar CrossRef Search ADS Peisakhin , Leonid , and Paul Pinto . 2010 . “ Is Transparency an Effective Anti-Corruption Strategy? Evidence from a Field Experiment in India .” Regulation & Governance 4 ( 3 ): 261 – 80 . Google Scholar CrossRef Search ADS Pellegrini , Lorenzo . 2008 . “ Causes of Corruption: A Survey of Cross-Country Analyses and Extended Results .” Economics of Governance 9 ( 3 ): 245 – 63 . Google Scholar CrossRef Search ADS Persson , Anna , Bo Rothstein , and Jan Teorell . 2013 . “ Why Anticorruption Reforms Fail—Systemic Corruption as a Collective Action Problem .” Governance 26 ( 3 ): 449 – 71 . Google Scholar CrossRef Search ADS Petrov , Georgy , and Paul Temple . 2004 . “ Corruption in Higher Education .” Higher Education Management and Policy 16 ( 1 ): 83 – 99 . Google Scholar CrossRef Search ADS Polese , Abel . 2008 . “ ‘If I Receive It, It Is a Gift; If I Demand It, Then It Is a Bribe’: On the Local Meaning of Economic Transactions in Post-Soviet Ukraine .” Anthropology in Action 15 ( 3 ): 47 – 60 . Google Scholar CrossRef Search ADS Posner , Richard . 1997 . “ Social Norms and the Law: An Economic Approach .” American Economic Review 87 ( 2 ): 365 – 69 . Rehn , Alf , and Saara Taalas . 2004 . “ ‘Znakomstva I Svyazi’ (Acquaintances and Connections)—Blat, the Soviet Union, and Mundane Entrepreneurship .” Entrepreneurship & Regional Development 16 ( 3 ): 235 – 50 . Google Scholar CrossRef Search ADS Reisinger , William , Marina Zaloznaya , and Vicki Hesli Claypool . 2017 . “ Everyday Corruption and Regime Support in Russia and Ukraine .” Post-Soviet Affairs 33 ( 4 ): 255 – 75. Google Scholar CrossRef Search ADS Riabchuk , Mykola . 2007 . “Ambivalence or Ambiguity? Why Ukraine Is Trapped between East and West.” In Ukraine, the EU and Russia , edited by Stephen Velychenko , 70 – 88 . New York : Pallgrave Macmillan . Google Scholar CrossRef Search ADS Rivkin-Fish , Michele . 2005 . “Bribes, Gifts and Unofficial Payments: Rethinking Corruption in Post-Soviet Russian Health Care.” In Corruption: Anthropological Perspectives , edited by Dieter Haller and Cris Shore , 47 – 64 . Ann Arbor, MI : Pluto . Rosenfeld , Richard , Eric Baumer , and Steven Messner . 2001 . “ Social Capital and Homicide .” Social Forces 80 ( 1 ): 283 – 310 . Google Scholar CrossRef Search ADS Sah , Raaj . 1988 . “Persistence and Pervasiveness of Corruption: New Perspectives.” Yale Economic Growth Center Discussion Paper 560, accessed online at http://www.raajsah.com/uploads/4/3/4/9/43495347/wp1988_persistence-and-pervasiveness-of-corruption.pdf, October 26, 2017. Sajó , András . 2003 . “ From Corruption to Extortion: Conceptualization of Post-Communist Corruption .” Crime, Law and Social Change 40 ( 2 ): 171 – 94 . Google Scholar CrossRef Search ADS Sakwa , Richard . 2010 . “ The Dual State in Russia .” Post-Soviet Affairs 26 ( 3 ): 185 – 206 . Google Scholar CrossRef Search ADS Sampson , Steven . 2010 . “ The Anti-Corruption Industry: From Movement to Institution .” Global Crime 11 ( 2 ): 261 – 78 . Google Scholar CrossRef Search ADS Sanders , David . 1981 . Patterns of Political Instability . New York : St. Martin’s Press . Google Scholar CrossRef Search ADS Shamir , Ronen . 2004 . “ Between Self-Regulation and the Alien Tort Claims Act: On the Contested Concept of Corporate Social Responsibility .” Law & Society Review 38 ( 4 ): 635 – 64 . Google Scholar CrossRef Search ADS Smith , Daniel . 2010 . A Culture of Corruption: Everyday Deception and Popular Discontent in Nigeria . Princeton, NJ : Princeton University Press . Spicer , Andrew , Gerald McDermott , and Bruce Kogut . 2000 . “ Entrepreneurship and Privatization in Central Europe: The Tenuous Balance between Destruction and Creation .” Academy of Management Review 25 ( 3 ): 630 – 49 . Google Scholar CrossRef Search ADS Sunshine , Jason , and Tom Tyler . 2003 . “ The Role of Procedural Justice and Legitimacy in Shaping Public Support for Policing .” Law & Society Review 37 ( 3 ): 513 – 48 . Google Scholar CrossRef Search ADS Sutherland , Edwin H. 1947 . Criminology . Philadelphia : Lippincott . Tarrow , Sidney . 2010 . “ The Strategy of Paired Comparison: Toward a Theory of Practice .” Comparative Political Studies 43 ( 2 ): 230 – 59 . Google Scholar CrossRef Search ADS Tavits , Margit . 2010 . “ Why Do People Engage in Corruption? The Case of Estonia .” Social Forces 88 ( 3 ): 1257 – 79 . Google Scholar CrossRef Search ADS Transparency International . 2015 . Corruption Perceptions Index 2015, accessed online at https://www.transparency.org/cpi2015#downloads, September 2017. ——— . 2016 . Corruption Perception Index 2016, accessed online at https://www.transparency.org/news/feature/corruption_perceptions_index_2016, June 2017. Triplett , Ruth , Randy Gainey , and Ivan Sun . 2003 . “ Institutional Strength, Social Control and Neighborhood Crime Rates .” Theoretical Criminology 7 ( 4 ): 439 – 67 . Google Scholar CrossRef Search ADS Varese , Frederico . 2000 . “Pervasive Corruption.” In Economic Crime in Russia , edited by Alena Ledeneva and Marina Kurkchiyan , 99 – 111 . London : Kluwer Law International . Vaughan , Diane . 1999 . “ The Dark Side of Organizations: Mistake, Misconduct, and Disaster .” Annual Review of Sociology 25 : 271 – 305 . Google Scholar CrossRef Search ADS Verba , Sydney , and Norman Nie . 1972 . Participation in America: Political Participation and Social Equality . New York : Harper and Row . Wacquant , Loïc . 2008 . Urban Outcasts: A Comparative Sociology of Advanced Marginality . Cambridge : Polity . Weisberg , Sanford . 2005 . Applied Linear Regression . Vol. 528 . Hoboken, NJ : John Wiley & Sons . Google Scholar CrossRef Search ADS Werner , Cynthia . 2000 . “ Gifts, Bribes, and Development in Post Soviet Kazakstan .” Human Organization 59 ( 1 ): 11 – 22 . Google Scholar CrossRef Search ADS Wiatrowski , Michael , David Griswold , and Mary Roberts . 1981 . “ Social Control Theory and Delinquency .” American Sociological Review 46 ( 5 ): 525 – 41 . Google Scholar CrossRef Search ADS Yang , Mayfair . 1994 . Gifts, Favors, and Banquets: The Art of Social Relationships in China . Ithaca, NY : Cornell University Press . Zlobina , Tamara . 2007 . “Cultural Markers of Ukrainian Public Space. Mixture and Instability: The City of Lviv Case.” European Humanities University Center for Advanced Studies and Education, accessed online at http://old.ehu.lt/uploads/files/periodicalissue/docs/Digest_2_2007 %20small_51ee90a2e2c7a.pdf#page=85, October 26, 2017. Author notes This article is based upon work supported by the US Army Research Laboratory and the US Army Research Office under grant number W911NF-14-1-0541. We are grateful for the contributions of Jungmin Song, the assistance of Yue Hu and Jenny Juehring, and helpful comments of Michael Sauder, Karen Heimer, and participants of Theory Workshop at the University of Iowa Sociology Department. © The Author(s) 2018. Published by Oxford University Press on behalf of the University of North Carolina at Chapel Hill. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Social ForcesOxford University Press

Published: Feb 16, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off