Three essays on corruption

#Abstract
In most of the studies the corruption is treated as an illicit contract between a venal agent, who breaks the trust of her principal by pursuing her private interests, and the client who looks for maximization of his utility. All the essays that make part of this thesis, shift the focus from a corruption as an individual dyadic interaction to a corruption as a part of group dynamics. I start with a global level, considering if corrupt governments tend to have more aggressive foreign policy towards their neighbours (Chapter 1). The next part inquiries into a link between macro- and micro-levels: whether an institution of collective sanctions for a crime of one changes the incentive structure of a wrongdoer and his team mates (Chapter 2). The last part covers a micro-to-micro component, focusing on mechanisms of conditionality that push people towards bribery when they observe others doing so (Chapter 3).

Are corrupt countries more aggressive on international arena than non-corrupt ones? That is the question I look into in the first chapter. It is an empirically established fact that corrupt governments spend relatively more on military expense items. But if new guns are purchased as a part of military procurement contract lobbied by a light-fingered politician, will these guns be necessarily used? Or, in other words, do the overblown military expenses have real consequences? In order to control for overall level of a country’s foreign activity, a new indicator of aggression was used: a share of conflictual events in total number of events generated by a government toward foreign actors. To build such an indicator I employed two large datasets of events which became just recently available to scholars (ICEWS and GDELT). Analysis of the panel data for the more than 140 countries over the time span of 20 years (1995-2014) confirms that indeed there is a strong positive correlation between a share of aggressive actions in foreign activity of a country and an overall level of the corruption, controlling for a battery of confounders. This relation has passed the series of checks for robustness.

Punishing the whole team for a bribery or any other crime committed by one of the team members looks like an attractive solution: a furtive contract between bribe-giver and -taker is hard to observe and to detect by an external authority, but can be easily controlled and curbed by the peers of a bribe-taker. In the second paper, I test if the sanctions applied to the entire group for a misconduct of one of its members can be an efficient prevention measure. Using a modified public good game with a threshold, I show that overall collective sanctions show unimposing results. Moreover, when subjects are able to punish their peers, the level of cooperation is even lower in the regime of collective sanctions than under individual sanctions. But there is difference across gender: under collective sanctions the level of compliance among males is substantially higher than among females. This difference is mostly explained by the perception of unfairness of the collective sanctions regime. Another important difference between two regimes is in reaction on previous experience of punishment: under collective sanctions a person got caught complies in the future at least in the short run. A ‘wrongdoer’ who was punished individually decreases his level of compliance in the next period.

As with many other decisions driven by normative expectations, the decision to give or not to give a bribe heavily depends on what kind of behavior a person observes among ones of his kind. The third paper studies the phenomenon of the conditional corruption. There I present the results of the online experiment that should increase our understanding of the sociological micromechanisms behind this conditionality. I suggest that one of the reasons why people are so much affected by observing the decisions of other is an in-group competition: client expects the decisions of a previous client might affect the attitudes of a bureaucrat towards him/her. The results of the experiment demonstrate that the effect of conditionality in an in-group is stronger than in an out-group. However, a stronger willingness for bribe-giving in an in-group does not affect individual beliefs about the incidence of corruption in general.