Abstract
This dissertation consists of three chapters that, theoretically and experimentally, address the effectiveness of anti-corruption mechanisms. In the first chapter, I analyze the effects of monitoring on an agent's incentives in a two-period principal-agent model in which the agent decides on his effort and corruptibility.
The agent's type and strategy are unknown to the principal. I compare incentive-compatible wages under three different scenarios: 1) the principal does not monitor and only observes output; 2) the principal monitors the agent's effort choice; and 3) the principal monitors the agent's corruptibility.
I find that monitoring of effort improves the sorting of types but it might also give the agent more incentive to be corrupt. Monitoring of corruption does not improve the sorting of types but it negatively affects the agent's incentive to be corrupt.
In the second and in the third chapter I analyze experimentally how promising as anti-corruption measures leniency policies really are. Buccirossi and Spagnolo (2006) had conjectured, based on theoretical work, that ill-designed legal environments might, in fact, produce results that contradict the intentions of the designers of the leniency policies.
And, indeed, I demonstrate, for the first time as far as I know, that real-world subjects understand and use ill-designed legal environments to enforce occasional corrupt transactions.