Ambiguity in Dynamic Contracts

Abstract

I study a dynamic principal agent model in which the effort cost is unknown to the principal. The principal is ambiguity averse, and designs a contract which is robust to the worst case effort cost process. Ambiguity divides the contract into two regions. After sufficiently high performance, the agent reaches the over-compensation region, where he receives excessive benefits compared to the contract without ambiguity, while after low performance, he enters the under-compensation region. Ambiguity also causes a disconnect between the current effort cost and the strength of incentives. That is, even when the agent is under-compensated, his incentives are as strong as in the over-compensation region, since the principal fears the agent might shirk otherwise. Under ambiguity, the agent's true effort cost does not need to equal the worst-case. I analyze the agent's incentives for this case, and show that the possibility of firing is detrimental to the agent's incentives. I study several extensions concerning the timing structure and the nature of the principal's ambiguity aversion.