Multi-Agent Moral Hazard and Expectation-Based Reference-Dependent Preferences

抄録

This paper studies a multi-agent moral hazard model in which the agents have expectation-based reference-dependent preferences developed by Köszegi and Rabin (2006, 2007). The agents' utilities depend not only on the realized outcomes but also on the comparisons of them with the reference outcomes. Due to loss aversion, the agents dislike wage uncertainty, and reducing it by partially compensating their failure may be beneficial for a principal. We show that when the agents are loss-averse, the optimal contract is based on team incentives (joint performance evaluation or relative performance evaluation) in which the agents partly share their wage uncertainty. Our results provide a new insight that team incentives serve as a loss-sharing device among agents.