Relational Contracts With and Between Agents

Firms often use both objective/verifiable and subjective/non-verifiable performance measures to provide employees with effort incentives. We study a principal/multi-agent model in which a verifiable team-based performance measure and nonverifiable individual performance measures (one for each agent) are available for contracting. A problem with tying rewards to non-verifiable measures is that the principal has incentives to understate the realization of those measures in order to reduce compensation. We compare two mechanisms for overcoming this credibility problem: bonus pools and reputation. While reputation is fostered by repeated interactions (a low discount rate), repeated interactions create opportunities for agent-agent collusion under bonus pools. These opportunities for collusion can be exacerbated by the team performance measure, to the point that it can be optimal to make the size of the bonus pool independent of the realization of the team measure. In general, strong task interdependencies improve the effectiveness of reputationbased contracting and reduce the effectiveness of bonus pool arrangements.