Making Collusion Hard: Asymmetric Information as a Counter-Corruption Measure

Abstract

We study the problem of a principal who relies on the reports of a monitor to provide incentives to an agent. We allow for collusion, so that the agent and monitor can side-contract on what report to send. We show that the principal can benefit from creating endogenous asymmetric information between the agent and the monitor, thereby making side-contracting more difficult. Specifically, it may be optimal to randomize the incentives given to the monitor, and let the magnitude of her incentives serve as her private information vis a vis the agent.Plausible numerical computations in simple environments suggest that the potential efficiency gains from random incentives can be large. However, in general, the optimality of random incentives will depend on patterns of pre-existing asymmetric information: it is not always effective to add new sources of asymmetric information. We solve for both the Bayesian and max-min optimal policies, as well as provide an experiment-ready framework for prior-free policy evaluation. We show that even though monitors' reports do not provide a reliable measure of actual corruption, it is possible to evaluate local policy changes using only unverified report data.