| Economic analysis offers useful insight into the incentives for corruption. In this thesis we examine corruption within the principal-agent framework with delegation where the government is the principal, the supervisor is a bureaucrat, and the agent is a citizen or a firm. We focus, in particular, on examining incentive structures that may encourage a: government bureaucrat to misuse his authority for personal gain. The first chapter examines the welfare implications of one such misuse, namely that of extortion. When a bureaucrat is authorized to penalize an agent for violating some government regulation, he may misuse his authority in two ways. First, he may accept a bribe from an agent who has violated the regulation, in exchange for dropping the penalty. However, more importantly, he may threaten to penalize an agent who has complied with the regulation, unless this agent offers him a bribe. This second form of corruption, which we identify as extortionary corruption, has not been examined extensively in the previous literature on corruption. In this chapter we examine the conditions under which extortionary corruption and extortion may arise and study some of its welfare implications.; The second chapter of this thesis considers a model in which a planner hires a supervisor to enforce an agent's compliance with regulation. Other papers consider the possibility of corruption in this context, after the supervisor has expended effort to observe and document non-compliance, which if reported would result in the agent being fined with a certain probability. The agent may choose to bribe the supervisor to not report his findings. This paper extends their model in two ways, to treat heterogeneous supervisors who differ in their ability and to allow the agent to choose to bribe the supervisor either before or after he has expended effort. We say that the former type of bribe entails preemptive collusion while the later entails ex-post collusion. This paper shows that there is a critical link between the ability of the bureaucrat and whether or not preemptive collusion will occur. In fact, when bureaucratic ability is high preemptive collusion is more likely, but when bureaucratic ability is low ex-post collusion is more likely to occur. We also show that there is a non-monotonic relationship between the magnitude of the bribe exchanged and the welfare distortion that results from corruption.; The third chapter extends the one-period static model presented in chapter two to a two-period dynamic model of corruption. Bureaucrats once again differ in their ability to monitor the firm. However, in contrast to chapter two, their ability is unknown. Players in this game are Bayesian learners who have rational priors about the bureaucrat's ability and over time may learn his true ability. Learning is sometimes prevented when the bureaucrat colludes with the firm preemptively. We find that the possibility of learning introduces an additional dimension to the conflict of interest between the principal and collusive supervisors. In the two period model we find that corruption sometimes prevents learning. We also find that the firm sometimes pays the bureaucrat a higher bribe in the first period in order to prevent the bureaucrat from gaining experience and learning his true level of ability. Additionally, we find that the size of the bribe never increases over time. |
