Moral Hazard, Dynamic Incentives, and Ambiguous Perceptions

Abstract

This paper considers dynamic moral hazard settings, in which the consequences of the agent's actions is not precisely understood. In a new continuous-time principal-agent model with drift ambiguity, the agent's unobservable action translates to drift set for the diffusion processes that describe evolution of output. The agent and the principal has symmetric imprecise information about the technology, and both seek robust performance from a contract in relation to their respective worst-case scenarios. We show that the optimal long-term contract aligns the party's pessimistic expectations and broadly features flattening of the high-powered incentives. Methodologically, we provide a tractable way to formulate and characterize optimal long-run contracts with drift ambiguity. Substantively, our results provide some insights into the formal link between robustness and simplicity of dynamic contracts, in particular that presence of ambiguity flattens high-powered incentives.

Keywords: Dynamic Moral Hazard, Compensation, Ambiguity, Continuous-Time Methods

JEL Classification: D81, D82, D86