Cost-Effective Control of Corruption in Public Offices

Abstract

We study the problem of controlling bribery at minimum incentives cost in a four-layer, principal-supervisor-agent-client, model of public office. The model includes a second check of corruption by an incorruptible detector such as an outside inspector, the media or any whistle-blower. We consider two occasions for collusion, ex-ante (before corruption occurs) and ex-post (upon detection), and we endogenize bribes and the supervisor's monitoring effort. We show that motivating the supervisor through a scheme that allows for ex-post collusion (side payments upon detection of corruption for covering up) can be cost-effective. But the scheme may implement only an intermediate range of corruption targets. Low corruption levels cannot be implemented because the ex-post side payment is too small to motivate the supervisor for an effective monitoring of the agent. High corruption targets cannot be implemented because the ex-ante collusion-proofness constraint will be violated. Preventing both types collusion may require asymmetric treatment of the two offenders (the client and the bribe-taking agent), with a relatively low penalty on the bribe-taker. The analysis provides several insights into the roles of incorruptible detectors in controlling corruption. Such detectors i) make implementation of zero corruption possible if they are sufficiently effective, ii) reduce implementation costs by making the internal corruptible supervisor accountable, iii) limit the sizes of side payments within the hierarchy, and iv) allow for the possibility of a bounty hunter arrangement that economizes on reward payments to the supervisor, hence total expected implementation costs. The paper also provides potential explanations for internal cover up of corruption and adverse selection of (corruptible) supervisors.