Corruption has been associated with lower economic growth and increased poverty (Fisman and Svensson 2007; Olken and Pande 2012; Mo 2001; Gupta et al. 2002; Méon and Sekkat 2005; Rose-Ackerman and Palifka 2016). Ordinary citizens in developing countries tell pollsters that corruption degrades their quality of life and is one of their biggest complaints (IPSOS/MORI 2016; World Economic Forum 2016). Several scholars have argued that corruption harms poor citizens in a country more than rich ones, increasing inequality (Olken 2006; Reinikka and Svensson 2004; Gupta et al. 1998).

For all of these reasons, in the past two decades, anti-corruption reform has become a central part of the global development agenda, and hundreds of studies of corruption have been conducted across the fields of economics, political science, sociology, and anthropology. Despite this, there is a general consensus that corruption remains resistant to efforts to tame it. There have been some striking successes, as we will document below, but studies routinely conclude that corruption remains no less prevalent today than it was 20 years ago, when the global effort to combat corruption began (Mungiu-Pippidi 2011; De Sousa et al. 2012; DFID 2015).

We argue that a meta-analysis of the extensive scholarly literature on corruption can point the way to a better understanding of corruption as well as new strategies to control it. The scholarly literature has so far developed in separate disciplines, each of which has produced important insights, but each of which also has some crucial limitations. Bringing them together into an overarching research agenda can be a powerful way forward. A few other efforts to categorize the scholarship exist (e.g. Jancsics 2014), but ours is the first to confront the existing approaches to corruption systematically with the considerable ethnographic research that has been conducted over the last several decades, and through this to suggest advances in both practical strategies to control corruption, and the intellectual understanding of corruption.

This project began as a literature review on corruption for the United States Agency for International Development (USAID). Our research team began by identifying approximately 600 articles, book chapters, working papers, and dissertations on corruption from the fields of anthropology, economics, history, political science, and sociology. We focused on bureaucratic corruption, in which everyday citizens are forced to pay bribes for government services, rather than grand corruption, in which business officials have close ties with government elites, with money flowing to the government elites and favors flowing to the business officials. Grand corruption and bureaucratic corruption are sociologically distinct phenomena. Grand corruption involves many fewer people and much larger amounts exchanged and can be corrosive to the legitimacy of the state; on the other hand, some scholarship suggests that bureaucratic corruption may be worse for economic development than grand corruption, as bureaucratic corruption distorts decision-making, increases unpredictability, and undermines transparency throughout the economy. For example, the rapid development of the East Asian countries was not hindered by grand corruption, but each of these countries did take efforts to stamp out bureaucratic corruption (Kang 2002; Prasad and Nickow 2016). While some of the arguments we make here may apply to grand corruption (such as the blurred distinction between gift and bribe), and grand corruption may also undermine the capacity of the state, in the limited space of this article we focus only on bureaucratic corruption.

Our research team skimmed the identified works and eventually found 260 studies that we deemed worthy of inclusion in the literature review, based on the criteria of whether the work focused primarily on bureaucratic corruption, was primarily empirical, offered insights into combating corruption, and was sufficiently of high quality as judged by the members of the research team. For quantitative research and experimental research the members of the research team with quantitative training made an expert judgment on aspects of the study such as the causal identification strategy, the sample size, the validity of measures, and the statistical significance of the findings. For qualitative research, the members of the research team with expertise in qualitative methods made an expert judgment on the depth of the empirical research (for example, the range of primary and secondary sources consulted, the duration of the fieldwork, the quality and number of interviews), the soundness of the research design (for example, the proper use of techniques of comparative or historical analysis), and the logic of the argumentation. A full list of the 260 articles is given in the appendix.

The team members read these studies and then turned this research into two articles. A companion article to this one (Gans-Morse et al. 2018) assesses the empirical evidence on seven different anti-corruption strategies: (1) rewards and penalties, (2) monitoring, (3) restructuring bureaucracies, (4) screening and recruiting, (5) anti-corruption agencies, (6) educational campaigns, and (7) international agreements. The companion article examines studies that employ field experiments, laboratory experiments, quantitative analyses, ethnographic research, and historical research, assessing the effectiveness of these seven strategies using these various methods. This present article, on the other hand, takes a broader approach, attempting to put theories and insights from different disciplines into conversation, and confronting them against an increasing body of ethnographic evidence in order to develop a new research agenda for understanding, and new strategies for combating, corruption. Thus, while the goal of the companion article was to synthesize the existing anti-corruption research to draw broad policy lessons, the present paper attempts to advance theory and suggest new approaches and a new research agenda.

Part I of this article identifies three broad strategies that have been proposed for combating corruption: the individual-level strategy of rewarding actors for avoiding corruption, and punishing them for engaging in corruption; the strategy of attempting to address corruption in a “big bang,” effecting rapid change at once in a whole society; and organizational-level strategies that attempt to address corruption one organization at a time. We identify both the promise and the problems inherent in each strategy: the reward/punish approach has shown evidence of success, but it can only be successful to the extent that those doing the rewarding and punishing, or monitoring the data, are not themselves corrupt; the big bang strategy fits well with the historical experience of several countries that overcame corruption, but it requires waiting for exceptional moments of crisis; and the organizational-level strategies have been shown to work in some state agencies, but it is not clear whether they can diffuse beyond these isolated examples.

Each of these strategies has emerged in isolation from an ethnographic literature on corruption that over the last two decades has become extensive, but has never before been synthesized into an overarching framework. Part II therefore turns to a thorough analysis of the ethnographic literature on corruption, distilling it into three main insights: corruption persists because diminished state capacity forces ordinary people to engage in it to meet everyday needs, because the distinction between corrupt and non-corrupt acts is not clear, and because there can be counteracting moral pressures, such as the pressure to take care of ethnic networks and kin groups. We derive from these observations three lessons of what successful corruption reform must do: it must meet the challenge of state capacity (the resource challenge), the challenge of clear definitions (the definitional challenge), and the challenge of actions that are moral according to alternative criteria (the alternative moralities challenge).

Part III puts the findings of part I and part II together, investigating what each of the three strategies to control corruption can learn from the ethnographic literature, and developing a research agenda for how to move forward in the study of corruption and the practice of corruption reform. We argue that the ethnographic evidence shows how each strategy needs to be reformulated: rewards and punishments should be embedded in strategies to cultivate commitment among employees, and experimental methods should build in ethnographic components to ensure the reliability of the measurements; research on “big bangs” needs to shift toward examining whether smaller-scale crises can open space for reform; and the organizational strategy needs systematic analysis of the causes of sub-national variation in corruption, as well as analysis of how isolated organizations that are corruption-free interact with the corrupt environment, to understand both how such organizations maintain themselves, and how they may diffuse. An overarching lesson from the ethnographic scholarship is that corruption is often the result of people trying to meet legitimate needs, and corruption reform must therefore proceed by offering alternative ways to meet those needs.

Part I: Three Approaches to Fighting Corruption

The dominant approach to corruption conceives of corruption as a “principal-agent problem” (Rose-Ackerman 1978; Klitgaard 1988), that is, one in which the agent (the bureaucrat) is not behaving in the interests of the principal (the public). The standard solutions to principal-agent problems are to monitor the agent and reward her for honesty or punish her for corruption. Global anti-corruption strategies are still dominated by a principal-agent model. But while there have been some instances of success with this model, there are also many failures (Fjeldstad and Isaksen 2008; Persson et al. 2013; Charron 2011; Enweremadu 2012; Gilbert and Sharman 2016).

An emerging consensus is that the reward/punish approach is necessary, but not sufficient on its own, because in many developing countries corruption is not a problem of individual deviance from the system; rather, corruption is the system, and nearly everyone is corrupt. Thus, programs to reward good behavior or punish bad behavior eventually fail because those doing the rewarding and punishing are themselves corruptible, and programs of monitoring are only as corruption-free as the monitors themselves (Sundström 2017; Persson et al. 2013; Fjeldstad 2003; Foltz and Opoku-Agyemang 2015; Hira and Shiao 2016; Osburg 2018). Individual-level rewards and punishments may be necessary for controlling corruption, but they do not seem to be sufficient.

A recent example of this dynamic is the growing consensus on the virtues of e-governance, that is, electronic technologies of surveillance. Several studies find e-governance to be effective in reducing corruption (Banerjee et al. 2014; Elbahnasawy 2014; Lewis-Faupel et al. 2016; Muralidharan et al. 2016; Olken 2007; Pathak et al. 2009). But when anthropologist Jeffrey Witsoe (2016) conducted an ethnographic examination of one of these electronic governance programs in north India, he discovered that the staff of the program were keeping two different sets of accounts—“offline” accounts and “online” accounts. That is, those charged with the monitoring of the surveillance technologies might themselves be unreliable.

Because of the problems that have emerged with the principal-agent approach, a second set of scholars has developed a critique and an alternative approach. These scholars make the point discussed above that corruption is a problem of the system, a collective action problem in which there are no “principals”: “in a context in which corruption is the expected behavior, monitoring devices and punishment regimes should be largely ineffective since there will simply be no actors that have an incentive to hold corrupt officials accountable” (Persson et al. 2013, 457; Mungiu-Pippidi 2013). These scholars often point out that structural features make corruption the most rational means of capital accumulation in particular contexts and suggest that we should not conceive of corruption as dysfunction or a defect from the correct order, but rather, as an alternative order (North et al. 2009; Khan 2005; Gray 2015; Uberti 2016). This alternative conceptualization, while not as dominant as the principal-agent conceptualization, has received support from a wide range of scholars from several disciplines (Mungiu-Pippidi 2015; Persson et al. 2013; Heilman and Ndumbaro 2002; Blundo et al. 2006; Anand 2015; Collier 2000).

As we will explore below, this understanding of corruption is certainly more empirically accurate when weighed against the ethnographic evidence. But this approach has important limitations as well, because it does not offer a realistic alternative to combating corruption. Even if corruption is an alternative order, its costs, particularly to the most vulnerable, are by now clear, and these scholars are not suggesting that we should abandon efforts to combat corruption. Rather, they suggest that because corruption is a systemic problem, it can only be resolved through a “big bang” of reform of the entire system in a short period of time (Rothstein 2011). But in instances where a big bang approach has worked, it has been rendered possible by historical circumstances such as defeat in war. For example, Teorell and Rothstein (2015) emphasize that dramatic change was possible in Sweden only because the loss of a war to Russia in the early nineteenth century led to a sense of urgency that overcame collective action problems. Popa (2015) similarly argues that war raised the costs of corruption, and therefore raised interest in controlling corruption, in nineteenth century Britain. Such “big bangs” have indeed been implemented in recent years in a few countries, including Georgia, Romania, and Italy, and have seen some success (Schueth 2012; Mungiu-Pippidi 2015; Acconcia and Cantabene 2008). But each of these cases depended on a historically specific constellation of forces and cannot underpin a general strategy for fighting corruption. It is perhaps for this reason that the principal-agent model has been hard to dislodge as a practical anti-corruption strategy. The big bang approach also neglects cases where anti-corruption reform was successful at sub-national levels. For example, the USA overcame corruption through several episodes that do not qualify as a “big bang” (Kernell and McDonald 1999; Skowronek 1982; Bodenhorn 2007; Parrillo 2013).

Recently, a third approach to corruption, one that does not need to wait for the arrival of moments of crisis during which an entire country can change, has emerged. Scholars have pointed out that even within countries known for widespread corruption, certain organizations are relatively free of corruption, and have remained so for decades. McDonnell notes that “the variation in governance within states is as large as the variation between states” (McDonnell 2012, 97–98; McDonnell 2017). This nascent literature has not yet produced a consensus as to the causes of variation in sub-national rates of corruption. Suggestions for the factors that affect corruption range from members’ exposure to other, non-corrupt organizations (McDonnell 2012) to members’ professional ties to networks outside the organization (Mistree 2015) to the demands of powerful benefactors for non-corrupt behavior (Johnson 2009) to the particular features of the market the organization is involved in (Zaloznaya 2017) and the organization’s autonomy (Mistree 2015, Roll 2014, Zaloznaya 2017).

Despite the lack of consensus on exactly how successful organizations become successful, the abilities of organizations to resist the corrupt social practices of the surrounding context are by now well-demonstrated within this literature and form the foundation for an approach that suggests the “big bang” can be attempted within particular organizations. Over time, as more and more organizations are created that are free of corruption—“islands of integrity”—the entire bureaucratic culture of a country could be changed.

This approach also has limitations. As McDonnell (2017) notes, the main reality of these islands of integrity is that they are constantly managing the corruption of the surrounding culture and it is not clear how many such islands can survive over the long term given the pressures of the wider context. A second issue is that one factor that makes existing islands of integrity cleave to their integrity is precisely the knowledge that they are different and special; a reputation for integrity will be worth less when many other organizations also can demonstrate integrity and such a reputation may therefore be harder to preserve. Moreover, as we discuss below, avoiding corruption takes resources, including the resources to pay employees well, limiting how many such organizations can be developed.

Having identified these three broad approaches to corruption reform, we now turn to a deeper discussion of the ethnographic literature on corruption, so that we can ask what the studies on the ground can tell us about how to move forward with these three approaches.

Part II: Why Corruption Persists

Despite a stereotype that some cultures are more accepting of corruption than others, systematic research suggests that corruption is condemned even in countries where it is widespread. Rothstein and Torsello note that “Surveys carried out in regions throughout India and in sub-Saharan Africa show that people in these societies take a very clear stand against corruption and understand the problem in the same manner as it is understood, for example, by organizations such as the World Bank and Transparency International” (Rothstein and Torsello 2014, 276; see also Miller et al. 2001). Surveys around the world routinely find corruption to be one of the issues that most frustrates ordinary citizens in developing countries (IPSOS/MORI 2016; World Economic Forum 2016).

But if corruption is universally condemned, then why does it exist? Upon closer investigation, three factors are at play that undermine anti-corruption reforms. One of the remarkable features of the empirical scholarship is how much agreement there is on these three factors across countries and over time, and therefore we arrange our review of the literature thematically rather than geographically or chronologically.

Corruption as Necessary, Corruption as Skill

First, even if corruption is universally condemned, in some contexts corruption is necessary for everyday survival, and given this, the ability to successfully execute a corrupt act is seen as a skill and a source of pride even while corruption is condemned in the abstract.

Das (2015) argues that in South Asia mundane everyday acts of corruption are what allow individuals to procure electricity and water and to send their children to school. Anand (2015) makes a similar argument: negotiating illegal water connections with local politicians, municipal water engineers, and community members, through bribes and promises of political support, is an everyday tactic for people in Mumbai to keep water flowing to their communities. Managing leakages allows marginalized citizens to access water without having to riot for it. Endres (2014) shows how local state border officials decide not to enforce the law by making an exception (through bribes) to small-scale traders crossing the border in Vietnam. State officials recognized that if the law was implemented an entire class of individuals would be put out of business, especially the small-scale petty traders along the Vietnam–China border, who are already a marginalized group. Hanlon (2004) makes a similar argument for Mozambique, Anders (2010) for Malawi, and McMann (2014) for Central Asia. Uberti (2016) argues that in developing countries, patronage can help to stabilize political situations that have been destabilized by colonialism and extractive industrialization. In Paraguay, the patronage system was perceived by ordinary Paraguayans as the normal channel for accessing jobs, services, and benefits (Hetherington 2018). Osburg (2018) shows how the ritualized entertaining practices cultivated by entrepreneurs in China to forge ties with state officials have allowed individuals of humble origin to make their way into the system.

The need to participate in corrupt exchanges is exacerbated by the weakness of the state in some contexts. Given the consistent incapacity of the state to provide adequate services, a range of informal practices is created to circumvent the official rules. Ledeneva (2009) shows through a historical study of blat (the use of personal networks for obtaining goods and services and for circumventing formal procedures) from the Soviet period to contemporary Russia that corruption was functional in granting individuals access to goods and services in shortage (also Ledeneva 1998). In Russia, Reeves (2013) shows how bribe payments by migrant workers to officers are seen by many Kyrgyz workers as “socially acceptable” as these officers are one of the few class of state employees whose salaries are actually lower than that of migrants who procure factory work. Also in Russia, Rivkin-Fish (2005) shows how unofficially paying a doctor was perceived as a bribe in the past but now is perceived as more appropriate and ethical than adhering to official institutional regulations, since the formal system is not able to provide for users’ needs nor to properly compensate doctors for their work; to unofficially pay a doctor thus becomes a positive act that represents respect for the doctor’s effort and expertise (particularly as the market-price system was widely misunderstood). Polese (2008) argues that bribes in Ukraine are not a choice but a necessity as public officials are not able to survive with their wages; indeed, to fight these transactions is counterproductive as they allow doctors and teachers to survive in Ukraine and not flee the country in search of better living conditions. Chew (1990) examines a case of falling salaries in Uganda and concludes “while corruption cannot be condoned, one should understand that dismally low pay can impel government employees into corruption through need rather than greed” (1010). McMullan’s (1961) ethnographic research in West Africa argues that corruption often emerges as a result of the divergence between “a literate government and an illiterate society”. He argues that illiterate people are more vulnerable to corrupt civil servants as they may not be certain what the rules are, whether they did in fact pay the fees that were required of them, and in general are in no position to resist paying bribes from literate professional people asking for them. A few authors emphasize that attacking corruption can undermine the effectiveness of a bureaucracy in both developing and developed countries (Mathur 2012, Anechiarico and Jacobs 1994, Shore 2005).

Given these structural impediments that make corruption necessary, the ability to successfully participate in corruption becomes seen as a kind of skill and is sometimes publicly or privately celebrated by actors as a way to display their connections and to reaffirm their social identity. In Bucharest, Romania, instead of feeling shame, former property owners talked proudly about their ability to engage efficiently with bribes as this ability displayed both their social capital (connections with judges) as well as their economic capital (the capacity to pay bribes) and reaffirmed their identity as owners (Zerilli 2005). In rural western Uttar Pradesh, rich farmers, despite publicly participating in protests against corruption, celebrated in private the corrupt local practices that guarantee their privileges in the sugar cane market (Jeffrey 2002).

The celebration of corruption as a display of power is not unique to elites. Many works in the ethnographic tradition show that the poor accept corruption because they see their own ability to engage in it as a form of agency. In Nigeria, corruption is perceived as a way to get rich (Smith 2010). In rural western Uttar Pradesh, Jeffrey (2002) argues that when the poor were able to purchase influence within cane societies it was celebrated as an ability to “stand on their own feet.” Ruud (2000) describes how in West Bengal the process of negotiating with the bureaucracy by using network ties is perceived as a social skill which some people are better at exercising than others, and which is therefore seen as a source of pride. De Sardan (1999) emphasizes how petty corruption is such a part of everyday experience when dealing with bureaucracy in contemporary African states that it becomes a part of the social landscape, a know-how, that is considered indispensable for survival. Witsoe (2011) argues that the continuous support of the poor for lower caste leaders who are widely known to be corrupt in Bihar, India, reflects the popular acceptance of corruption as a means to caste empowerment. In this context, corruption becomes a means to appropriate the state by the poor and as such a means of leveling inequalities in Indian society. People supported politicians not only despite perceptions that they were corrupt but also precisely because they were perceived as corrupt and therefore capable of using their positions to benefit their supporters. In a similar vein, during his fieldwork with lower levels of the bureaucracy in a small village in North India, Gupta (1995) links individuals’ frustration with corruption to the fact that not only are they exploited, but they also lack social capital required to negotiate the services. In a related analysis conducted in a very different context, de Vries (2007) argues that the corrupt but popular rule of the politician Lopez in the city of El Grullo, Mexico, was rooted in his ability to talk with pride about his ability to transgress the rules. Lopez and his clients’ spectacle of enjoyment and their proud talk about their ability to arrange things outside the law was a form of performing the cultural representation of power.

The ability to play the game is perceived, then, not as a source of embarrassment, but rather as a source of pride and a measure of competence. Blundo et al. (2006) note that refusing to engage in corruption in these contexts can even be perceived as a lack of propriety or a break with normal solidarity, particularly given how corruption permeates state institutions to which individuals are closely related in their daily lives, such as hospitals, day care, education, and military registration. For all these problems, individuals have to interact with state agents, and informal practices of circumventing official procedures are so widespread that they have become social norms.

This complex of attitudes can serve to undermine anti-corruption reform. Where corruption is necessary for survival, widespread condemnation of corrupt practices is not enough if individuals are not given other ways to ensure their survival and engage with the state.

Corruption vs. Gift-Giving, Public vs. Private

Universal condemnation of corruption also coexists with ambiguity over which exchanges qualify as corrupt; some are seen as parts of gift-giving rituals, and there is often no clear distinction between gift and bribe, or public and private. Several authors call attention to how the cultural competence of some actors enables them to manage these perceptions by making their acts appear to be something other than an outright bribe.

Smart and Hsu (2007) argue that perception of corruption in China is intrinsically related to political power and its abuse. Despite the rhetorical distinction between corruption or bribe and guanxi (the use of social networks to accomplish tasks), the behaviors are the same: exchange of gifts for instrumental purposes. However, bribery is seen to violate the basic foundation of guanxi, warm personal relations, by making instrumental gain its sole purpose. The difference between reasonable guanxi practice and an act of corruption hinges on managing perceptions. The style and specifics of the interaction will affect how it will be perceived by those involved; therefore, it is of utmost importance that actors are willing to tactfully behave as though the transactions were primarily about human sentiment or the collective good (see also Sedlenieks 2004). Avenarius and Zhao (2012) highlight how the combination of traditional convictions about gift giving with more modern values, such as meritocracy, varies from urban to rural citizens and from older to younger Chinese. Caplan (1971) argues that exchanges made in kind are seen as gifts while exchanges made in cash are seen as bribes. Werner (2000) argues that the distinction between gift and bribe in Kazakhstan is difficult and at times impossible to make, and it takes high levels of cultural competency to do so. Individuals’ views on the morality of bribery depend on multiple elements: the personality and generosity of the official, their regular salary, how the bribe compares relationally to others in the same career position, and whether or not the bribe was presented voluntarily.

Jeffrey (2002) discerned from long-term fieldwork in Uttar Pradesh state in northern India several key findings regarding social perception of corruption. First, the perception of whether a monetary gift is a bribe or not is strongly tied to class. If the bribe is initiated by the rich toward the poor, it is understood as a gift; if the roles are reversed, and the poor are the initiators, it is understood as a bribe. Second, the poor are much less likely to build the necessary social relationships with government officials and civil servants in order for their monetary gifts to be understood as gifts and not bribes. Jusionyte (2015) discusses the ability of civil servants to function at the border of what is legal and illegal, and legitimate and illegitimate. Hasty (2005) shows the extremes that anti-corruption investigators go to in such contexts to resist the cycle of gift-giving, refusing even soft drinks for fear that this might be, or might be seen to be, a bribe (See also Rose-Ackerman and Palifka 2016, 235–242; Philp 2008).

Another concern is that condemnation of corruption is undermined when there is no clear distinction between public and private. Rothstein and Torsello’s (2014) work on Human Relations Area Files data suggests that even though corruption is uniformly condemned, what counts as public and private varies in different cultures. In general, a higher number of occurrences of bribery instances is found in societies where there is no clear-cut distinction between private and public goods. In the west, avoidance of corruption is rooted in a carefully established distinction between processes of social and political exchange, and also between political and economic exchange, but other social and cultural contexts throughout the world have a quite different evolution of public vs. private (Heidenheimer 1996).

Munoz’s (2014) long-term ethnographic fieldwork with civil servants in Cameroon reveals that the emergence of corruption and the anti-corruption apparatus are new policy challenges for how civil service reforms are negotiated, as rewarding individuals and organizations with contracts was an established mechanism of awarding favors and redistributing wealth—blurring the boundaries of public and private. Gupta’s (2012) analysis of corruption in northern India shows how bureaucrats collapse their role of public and private servants. Those blurred boundaries instead of being an anomaly are part of the normal and routine conditions in which states operate through local officials.

Schatzberg (2001) examines newspapers, government communiqués, and literary sources to analyze discourses around political life from eight African countries, and finds that the Western discourse of legitimacy of the state based on a sharp distinction between the state and civil society is unable to capture how many African citizens understand the state. Instead, Schatzberg finds a familial understanding of state offices and the redistribution of public goods; and if the nation is a family guided by a paternal leader, as father-chiefs political leaders are expected to eat, and to eat well during prosperous periods.

Kinship-Based Moralities and Other Moralities

Even though corruption is generally seen as immoral, in practice other dimensions of morality can pressure civil servants into corrupt behavior, particularly the morality of taking care of one’s kin or ethnic networks. Ruud (2000), Smith (2010), and Bukuluki (2013) all make reference to the existence of a social pressure to provide for relatives and friends as one of the reasons why bending the rules for private gain might be more acceptable in contexts as distinct as India, Nigeria, and Uganda. In a similar vein, Smith (2010) also notes in Nigeria the social pressure to redistribute to kin as a strategy of survival of the poor. Not all forms of bending the rules are equal. The morality of certain practices that could be classified as corrupt depends on the motivations for such practices. Some forms of corruption are undertaken based on positive values (kinship, reciprocity) whereas others are negatively labeled. Ekeh (1975) argues that African societies have two public realms, the "primordial public," identified with primordial groupings and kinship, and the "civic public," associated with civic structures and state institutions. The individual sees sustaining the primordial public as a moral obligation, but the civic public is amoral: individuals seek to gain from the civic public but have no moral urge to give back to the civic public.Footnote 1

Wihantoro et al. (2015) study an attempt at reform of the Indonesian tax administration. Because these reforms aimed to increase “impersonality”, they conflicted with the organization’s practice of treating employees as one big family and helping with colleagues’ financial and family matters. Reforms geared toward employee evaluation by supervisors, and systems of employee praise to encourage workplace competition such as Employee of the Month programs, proved particularly difficult to enforce due to their incompatibility with values of harmony, respect, and hierarchy. Refusals of gifts from clients were often taken as insults. Managers were reluctant to praise productive employees and were excessively compassionate to poor performers. In this and other ways, the authors argue that Western bureaucratic models were inappropriate.

Von Holdt (2010) draws on participant observation in public hospitals and health departments to argue that nationalism and the struggle against apartheid and racism undermine bureaucratization efforts in South Africa. The project of nationalism leads to a set of practices that directly undermine bureaucratic functioning; for example, the wish to promote the hitherto oppressed can lead to unskilled actors in bureaucracies, rigid hierarchy combined with this absence of meritocracy can lead managers to care more about pleasing higher-ups (e.g. in controlling costs) than solving the problems of the rank and file, and lack of respect for authority as a legacy of apartheid struggles can lead to corruption. By studying cases of property restitution in Romania, Zerilli (2005) shows how attitudes toward corrupt behavior shift according to personal experience, circumstance, and collective histories. Farmers seeking restoration of property rights, while blaming corruption, justified their use of corruption (bribing judges) as a way to repair injustices perpetrated by the old regime.

In sum, civil servants and ordinary people engage in everyday corrupt practices in order to navigate a broken system and avoid state neglect, because the distinction between a corrupt act and a non-corrupt act is unclear, or in response to pressure from ethnic and/or kin groups or as a way to address prior oppression. These practices undermine anti-corruption reforms.

This survey of the empirical scholarship reveals that anti-corruption reform must meet three challenges. First, corruption persists because people need to engage in corruption to meet their needs. This is the resource challenge. Second, corruption persists because there is uncertainty over what constitutes a gift, and what constitutes a bribe, as well as confusion over what is private and what is public. This is the definitional challenge. And third, corruption persists because there can be pressure to behave in ways that are considered moral according to alternative criteria, such as taking care of one’s kin, or standing up to legacies of racism and oppression. This is the alternative moralities challenge.

Part III: Addressing the Challenges

Each of the three strategies holds promise in addressing the three challenges, and each has limitations. The principal-agent approach conceptualizes corruption as individual deviance driven by greed and therefore does not explicitly aim to address the problem of resources or of underdeveloped state capacity that drive people to engage in corrupt activities. However, although current incarnations of the principal-agent approach focus on monitoring and punishing corrupt behavior and rewarding non-corrupt behavior, it is a simple extension of the approach to take into account the utilitarian reasons why people need to behave in corrupt ways and to address these needs, for example by raising salaries to levels where civil servants do not feel the need to participate in bribes. In practice, attempts such as raising wages are part of any systematic attempt to control corruption.

The principal-agent models may also be able to address the challenge of definition if they make the conditions for punishment and reward clear: by establishing exactly what behaviors will be rewarded and what behaviors will be punished, clear definitions may be promulgated of what constitutes corruption. This is again an extension of the principal-agent model rather than a pure version of it, however, because in this formulation, the punishment and reward are having an effect not because they change the costs and benefits, but because any regime of punishment and reward communicates information to a group.

Where the principal-agent models are weakest is in addressing the question of alternative moralities. In some contexts, it may seem more moral to sacrifice one’s individual-level reward, or submit to the individual-level punishment, in order to provide for one’s kin or ethnic group. For example, in the South African context that von Holdt studies, individual-level rewards and punishments would have been seen as part of a regime that favored whites, inherited from the apartheid era, and it is easy to understand why actors would develop strategies of resistance that seemed to them entirely moral, and morally preferable. Indeed, one study of the water and sanitation sector in South Asia finds that successful corruption reform drew on techniques of monitoring and reducing discretion, but embedded these in reforms that “sometimes unintentionally, built new commitment and pride among the concerned staff…Indeed, in those instances in which a…reform was implemented with no concomitant driver that changed the way employees felt about their jobs, the result was resentment and eventually sabotage of the reform itself” (Davis 2004, 67). Schulze and Frank (2003) suggest the scholarship on corruption needs a clearer understanding of intrinsic motivation (see also Armentier and Boly 2011).

As to the second set of approaches, the ethnographic scholarship helps us understand exactly what the “big bang” does—and how it might be replicated at lower levels. First, a crisis caused by an event such as a defeat in war can alter the willingness of groups to redistribute resources, so that it becomes possible to meet the resource challenge. For example, in eighteenth and nineteenth century Britain, corruption was how civil servants survived; anti-corruption reforms succeeded because they introduced salaries  (Harling and Mandler 1993; see also Sundell 2014). Although punishment was a big part of a recent effort against corruption in Georgia, reform was also enabled by donors who “provided ample salaries to 11,000 civil servants in 2004…Support from international donors enabled raising civil service salaries to levels that obviated the need to engage in petty corruption in order to ensure livelihoods” (Schueth 2012, 138; see also Quah 2007).

The big bang model addresses the problem of definitions by creating a new, explicit definitional moment. The reform of irregular emoluments in Britain gives an example. When reformers tried to argue for providing politicians with regular salaries, a definitional struggle ensued, with those opposed to it arguing that regular salaries would be a “violation of the principle of gratuitous public service” and would attract “the very worst class that a country can be governed by—the caucus-fed professional politicians. Log-rolling and corruption are the inevitable corollary” (quoted in Seaward 2010, 45). But within a generation reformers had succeeded in arguing that professional salaries were necessary to allow those other than the wealthy to serve in public office (Seaward 2010). Similarly, in Sweden, the idea of a public office as a personal possession was replaced with a meritocratic definition rapidly, between 1860 and 1875 (Rothstein 2011, 243).

These explicit efforts at redefinition aggregate to offer participants a way to resist alternative moralities, although more research on this issue is needed. In countries that experience successful anti-corruption reform, “wherever individuals looked after the massive reforms had taken place, they realized that there was ... a new ‘game in town’” (Persson et al. 2013, 465). For example, in Hong Kong, the Independent Commission Against Corruption used formal and informal mechanisms to change public attitudes in a generation, using television, concerts, and sporting events to foster resistance to corruption such that “by the 1980s young people in Hong Kong took a stricter view of corruption than did their elders—a contrast found in few other societies” (Johnston 1998, 97).

Thus, the “big bang” scholarship can be brought together with the three challenges to ask whether smaller-scale crises can have some of the same results in addressing corruption. And unpacking the mechanisms through which the big bang affects corruption can allow anti-corruption reformers to be prepared for moments when large-scale crisis opens up the possibility for large-scale change.

The organizational strategy holds promise on all three challenges. Organizations can address the problem of resources because even in poor countries, certain organizations are able to generate the resources necessary to allow their employees to maintain a stable livelihood without resorting to bribery. For example, Johnson writes of the Directorate of Economic and Financial Cooperation (DCEF) in the Ministry of Finance and the Economy (MEF) of Senegal that it “benefited from relatively stable financial support from the Senegalese government” (2009, 161). An important limitation of this approach is that only some organizations are likely to have the resources to pay their employees well. One suggestion is therefore to start by attacking corruption in the tax agencies. When the tax agencies are corrupt, tax revenues are lower than they would otherwise be, and it is not possible to generate the kinds of state revenues that would allow a wider resolution of the resource problem.

Organizations can address the problem of definition because they are small enough to be able to impose particular definitions on their members. Consider the interviewees whom Zaloznaya asks about bribery in both corrupt and non-corrupt Ukrainian universities. One interviewee from a corrupt university tells her: “I would give you a 95% guarantee that it’s not bribes that matter here… but, rather, good relationships, long evenings over a rich dinner table, holidays – student’s day today, teacher’s day tomorrow… some drinks, some snacks, maybe occasional present but mostly niceness and brotherhood. A lot of the times it’s these relationships that decide everything” (Zaloznaya 2012, 207). This “niceness and brotherhood” are not defined by the respondent as corruption, even though this is a context in which academic merit rather than relationships should “decide everything”. On the other hand, an interviewee from a non-corrupt university says: “I have never accepted anything from students… For me, the work with students and the positive energy they give me is that salary supplement that the government does not give me” (184). It is specific organizations that transfer the definition of whether an act is “niceness and brotherhood” or a bribe that saps “positive energy”. Zaloznaya notes that informal mechanisms such as gossip and observation of others are the mechanisms of transmission of the local norms.

Finally, Mistree (2015) gives a good picture of why organizations are the relevant unit for addressing the challenge of alternative moralities. He investigates the Indian Institutes of Technology, a set of universities in India that, unlike other educational institutions in the country, are remarkably free of bribery and corruption. When Mistree asked his interviewees why bribery was uncommon at the IITs, two of the interviewees “turned the question around, rhetorically asking whether such deplorable behavior would be expected or tolerated at an American university. If it does not happen at American universities, why should it happen at the IITs?” (206). These Indian interviewees did not identify with the surrounding context of corruption in India, but with the relatively corruption-free context of American universities. Mistree documents the strength of this culture, and the mechanisms that sustain it, including allocation of resources based on merit, which draws even those who have more lucrative options in developed countries (207–8). This culture is particularly strong because professors generally live on campus, and socialize with each other rather than with the surrounding community. In the face of all these organizational-level incentives toward meritocracy, defecting to the benefit of one’s family or kin groups—a practice widespread throughout the surrounding Indian context—becomes unthinkable.

Conclusion: A Program for Research

This tour through the empirical literature on corruption gives suggestions for both a research agenda and more empirically informed strategies for fighting corruption.

We first argued that three broad strategies of corruption reform can be identified: individual-level reward/punish strategies, macro strategies of effecting change in an entire society during a “big bang,” and organizational-level strategies that address individual organizations. While each of these is promising in certain ways, each also has its limits. To improve these strategies, we next turned to a review of the ethnographic literature on corruption and distilled from it three points: corruption exists because people need to engage in it to fulfill their everyday needs, and that in this situation the ability to successfully execute a corrupt exchange becomes seen as a kind of skill; corruption exists because the distinctions between gift and bribe, and private and public, are not clear; and corruption exists because alternative moralities to distribution by merit exist, such as distribution according to kin groups.

This confrontation with the ethnographic literature allows us to argue that the research on individual-level rewards and punishments can move forward in two ways. First, these studies, including recent studies on e-governance, need to build in ethnographic components. We should not base policy on experimental methods without a more detailed understanding of the reality on the ground, including the reliability of the measurements. Development organizations have begun to understand the need to integrate experimental methods with qualitative ethnographic analysis, and corruption research should follow that lead. Second, while reward/punish approaches can address resources and definitions, their Achilles’ heel is the question of alternative moralities. The best way forward is to imitate those reforms that embed rewards and punishments in strategies to cultivate commitment and pride among employees (Davis 2004).

Research on the “big bang” needs to focus on how it might be possible to bring about the conditions necessary for corruption reform on many fronts without war or economic crisis. Historical research may help in this, particularly in showing how smaller crises can open space for reform. For example, Bodenhorn (2007) shows how the combination of several contingent events led to corruption reform in nineteenth century New York. The Panic of 1837, while not as significant as some of the events that have been highlighted by the scholars of “big bangs”, nevertheless kicked off a process in which small entrepreneurs along with other groups came to identify their interests with anti-corruption reform, and members of the media kept up the call for reform. Research is needed on how events that are less encompassing than “big bangs” can nevertheless lead to systemic change, as well as on the specific mechanisms through which the big bang brings about change.

Finally, the organizational strategy offers several ways to move forward on research: first, we need to systematically analyze the causes of sub-national variation in corruption. Comparative historical studies as well as comparative ethnography are necessary to tease out the reasons for variation among organizations. Second, this research needs to examine carefully how islands of integrity interact with the corrupt field around them, and the strategies that allow some of them to succeed in maintaining their absence of corruption, and might allow some of their practices to diffuse to other organizations.

This analysis has also given us an important overarching warning about what not to do: anti-corruption reforms should not aim to punish the person giving the bribe. Although our stereotype of a bribe-giver is of the rich driver of a BMW paying off a traffic policeman, or a business owner bribing fire inspectors, in fact the ethnographic literature shows that bribe-giving is often an attempt by citizens to meet their legitimate needs, including in cases where they are legally entitled to the good or service but cannot get it without a bribe. Moreover, the ethnographic evidence is clear that the poor are the ones most likely to get caught by anti-corruption reforms. Das (2015) argues that the poor are actually in positions of greatest risk in the political agendas of anti-corruption in India, as their attempts to secure the elements needed for survival can be represented as corruption. Werner (2000) argues that anti-corruption laws that target gift-giving are likely to implicate the wrong individuals—those in more desperate or marginal situations who resort to bribing public officials—rather than targeting officials who routinely benefit from bribes, as the latter typically possess the resources and connections to escape identification by national anti-corruption legislation (see also McMullan 1961; De Sousa et al. 2012). The poor are much less likely to have the resources and connections that allow them to portray their monetary gifts as gifts and not bribes. Jancsics (2013) finds clear class differences in the need to bribe officials, with middle-class and upper-class respondents often commenting that they did not need to bribe low-level government agents: they can use their high-level contacts to gain what lower-class respondents need to gain through bribery. Furthermore, anti-corruption legislation may be used as a political tool by public officials against their enemies. For all these reasons, Torsello (2012) argues that policy makers should be wary of corruption narratives as “truth telling” or “revealing” moments by individuals or civic groups, and consider the political goals of these narratives (see also Kajsiu 2013). Given these observations, it is worth taking seriously Basu’s (2011) suggestion that the act of giving a bribe in circumstances where it is necessary to do so to get something the giver is entitled to should be declared legal, as this creates a divergence between the interests of the bribe-giver and the bribe-taker, giving the bribe-giver reason to report the bribe-taker. But the legitimate needs of the bribe-taker must also be kept in mind, and corruption reform can only be successful if the salaries of government bureaucrats are at levels commensurate with their needs.

Our research suggests that while all three anti-corruption approaches are valuable tools in the arsenal against bureaucratic corruption, all three approaches need more systematic research targeted in the ways we have identified here to develop viable methods of corruption reform.