« Does Power Corrupt? The Effect of Holding Local Political Office in Zambia on Generalized Trust and Altruistic Reciprocity Daniel P. Enemark,* ...»
Does Power Corrupt?
The Effect of Holding Local Political Office in Zambia
on Generalized Trust and Altruistic Reciprocity
Daniel P. Enemark,* Clark C. Gibson,†
Mathew D. McCubbins,* and Brigitte Zimmerman†
DRAFT: May 2012
Abstract: Scholars universally acknowledge government corruption to be widespread in Africa,
yet there is no such agreement on whether office-holding tends to corrupt politicians or corrupt
politicians tend to obtain office. Building on the corruption and behavioral economics literatures, we identify behavioral choices that we argue are linked to corruption. We conduct controlled experiments with over 140 local politicians in Zambia, sampling those who won or lost the last election by a margin of 10% or less, and use a regression discontinuity design to identify the local average treatment effect of holding office on the behavior of politicians. We find that holding office does have a significant effect on the strategic behavior on politicians. Specifically, at the electoral margin, holding office leads politicians to display higher levels of altruistic reciprocity. Possibly because of our low statistical power, we were not able to identify a significant effect of holding office on politicians’ level of generalized trust, although we did observe insignificant differences in the expected direction.
* University of Southern California † University of California, San Diego
1. Introduction Does power corrupt, or do the corrupt seek power? Political observers have explored this question for millennia: Plato held deep concerns about the corrupt pursuing power, but expressed confidence that true philosophers would not be seduced by the temptations of office. Socrates warns Glaucon in the Republic that office-holding must be devoid of lavish rewards, or those who are “hungering after their own private advantage … will be fighting about office,” and will produce civil disputes that undermine the state (Plato 380 BC/1888). Sir John Dalberg-Acton uttered perhaps the most famous warning about the perils of institutional authority in his response to Pope Pius IX’s pronouncement of papal infallibility: “All power tends to corrupt,” he cautions, “and absolute power corrupts absolutely” (Dalberg-Acton 1949).
The ubiquity and consequences of corruption have made it central to political discussions by citizens and scholars alike. It has been blamed for a long list of ills, including the misuse of public funds, the inefficient provision of public goods, the lack of government legitimacy, stunted economic growth, and citizen disenfranchisement. A deep and wide scholarly literature has sought to explain its origins and endurance, pointing to possible causes such as human nature, institutions, enculturation, and even genotype.
But the apparent simplicity of the question – are people or are institutions responsible for corruption -- belies the difficulty of answering it rigorously. Current studies of corruption offer a variety of definitions that range from narrow concepts like criminal bribery to the broader view that corruption includes any action that exploits public office for private gain. Measures of corruption rely on surveys of perceptions of corruption and objective data on policy outputs, both of which suffer from selection bias and measurement error. To explain whether individuals or office generates corruption, research requires observations from the counterfactual world in which incumbents make choices without holding office.
We explore the determinants of corruption with a novel research design employing behavioral games played by politicians. We begin by developing a theory of the proclivity for corruption at the individual level, based on the widely accepted assumption that corruption is motivated by an individual’s preference for private over public welfare. We assert that prosocial behavioral traits, specifically generalized trust and altruistic reciprocity, are associated with lower levels of corruption. We ground this assertion in the corruption and behavioral economics literatures and support it with in-depth interviews in the field. Behavioral game theory provides a means by which to measure these traits at the individual level in an unbiased way.
To measure prosocial traits, we recruited subjects from a pool of district councilors in Zambia, an African country with an average level of corruption for the continent, and estimate politicians’ preference for private over public welfare, and therefore their proclivity for corruption. We believe this is the first effort to recruit actual politicians for behavioral games, as well as the first study to use such games to identify whether those seeking power have equal proclivity for corruption absent the treatment of office-holding. Using a regression-discontinuity design, we sampled Zambian councilors who competed in the last local government election and won or lost by no more than 10% of the vote. We expect such individuals to share similar underlying traits before taking office, an assumption we support econometrically in our analysis.
The combination of the “as if” random assignment to the treatment of holding office and our outcome variable, derived from highly incentivized tasks in the laboratory, allows us to determine whether the prosocial traits of generalized trust and altruistic reciprocity differ across winning and losing politicians. We find that that holding office leads politicians to display higher levels of altruistic reciprocity. This is particularly compelling given the relatively low statistical power of our test. We find no significant effect of holding office on politicians’ level of generalized trust, although we do observe insignificant differences in the expected direction.
2. Corruption Literature Scholars, journalists, and policy practitioners have documented widespread government corruption and its disastrous downstream effects in all nations, but a great deal of current research narrows its focus to developing countries (Shleifer and Vishny 1993; Coolidge and Rose-Ackerman 1997; Hoff and Stiglitz 2003). Media reports about bribery and embezzlement among ministers, party members, bureaucrats, and political supporters are commonplace.
Government officials engage in a vast web of clientelism, a softer form of corruption, by rerouting jobs and funds away from the intended recipients to supporters and dependents. Over time, a culture of corruption promotes self-interested behavior and punishes prosocial behaviors such as egalitarianism and meritocratic hiring (Rose-Ackerman 1999; 2001; 2004).
While nearly as many definitions of corruption exist as there are scholars in the field, most agree that the motivation for corruption is a politician’s preference for private gain over public welfare. Shleifer and Vishny (1993) define corruption as the use of public office for private gain. Others, such as Bardhan (1997) and Aidt (2003), suggest that corruption induces agency loss, since the private gain of the office holder comes at some public cost. Some narrow this definition of corruption to require illegality (Jain 2001) or include only the giving and taking of bribes (Gray and Kaufmann 1998). The constant, however, remains that idea that the politician chooses personal over public gain.
A variety of theories also address the systemic determinants of corruption. Beyond the widely established negative correlation between economic development and corruption (Treisman 2007), the main debate explores whether cultural or institutional phenomena cause corruption. Institutionalists argue that the incentives generated by electoral rules, systems of government, and bureaucratic institutions may all affect corruption by altering the incentives that motivate politicians and the monitoring mechanisms that constrain them (Rose-Ackerman 1999;
Reinikka and Svensson 2003; Treisman 2007; Andvig and Moene 1990; Chang and Golden 2004; Kunikova and Rose-Ackerman 2005; Acemoglu et al. 2001). Warner (2007) argues that decentralized government institutions, the context of our research, are particularly vulnerable to fostering corruption. African countries often lack precisely those institutions that scholars find important in restraining corruption. Without such mechanisms to check them, politicians turn to corrupt tactics to retain political power and enrich themselves (Aidt 2003; Coolidge and RoseAckerman 1997). Although some believe that democratic institutions in Africa will reduce corruption (e.g. Lindberg 2006), most analysts are far less sanguine.
Scholars also advance cultural theories to explain corrupt behavior (Hauk and Saez-Marti 2002; Treisman 2007; Johnston 2005). In this view, corruption is the result of a long history of social conventions, enforced and reinforced by social sanctions and internalized feelings of guilt or shame. For example, Fisman and Miguel (2007) find that U.N. diplomats from highcorruption countries accumulate more unpaid parking tickets, concluding that norms are an important determinant of corruption. De Sardhan (1999) argues that corruption in Africa is socially embedded in the negotiation mechanisms and authority structure of the continent’s cultures and is therefore immutable. However, Bayart (1993) explicitly attempts to combat this assertion by arguing that Africa’s poverty, not its embedded culture, makes it prone to corruption and clientelism.
3. Behavioral Economics and Political Behavior Whether cultural or institutional, explanations for corrupt behavior all assume that a politician’s preferences trump what is best for society. We believe that behavioral economics provides us with an ideal tool for measuring the underlying traits that capture the tradeoff between private and public welfare. The use of economic games in the laboratory has led to major advances in our understanding of human decision-making (Camerer 2003). One of the key areas of exploration has been the study of an individual’s “value orientation,” that is, a comparative weight placed on their own welfare compared to the welfare of others (early examples in psychology include Kelley and Stahelski I970; Kuhlman and Wimberley 1976; the first economists to adopt the idea of value orientations explicitly were Offerman et al. 1996).
Researchers find that individuals tend to make choices quite consistent with their orientations.
The dominant value orientations observed in behavioral experiments are “individualistic” (selfish, profit-maximizing) and “cooperative” (jointly maximizing private and public welfare) (e.g. Liebrand 1984; Van Lange 1999; Offerman et al. 1996).
One way to summarize the differences between individualistic and cooperative types would be to say that all individuals put weight X on private welfare and weight Y on public welfare; for individualists Y = 0, while for cooperators Y is significantly greater. In our experiments, we consider prosocial choices as those consistent with a “cooperative” value orientation. We associate prosocial behavior with a lower proclivity for corruption, and selfish choices, those consistent with the “individualistic” value orientation, with a higher proclivity for corruption.
If the institutional arena drives corruption, then holding office would changes underlying traits as well as surface-level behaviors. This assertion is well grounded in the literature that links traits to corrupt behaviors. La Porta et al. (1997), Knack and Keefer (1997), and Rothstein (2011) found a negative association between corruption and generalized trust as measured in behavioral games. In this work, generalized trust implies trust without regard to the characteristics of the individual, and particularized refers to trust of known individuals in multi-round interactions.