Ambiguity sharing and the lack of relative performance evaluation

Abstract

We extend the Holmstrom and Milgrom (Econometrica 55:303–328, 1987) model by introducing model uncertainty to study robust long-term contracting and focus on relative performance evaluation. Concerns regarding model misspecification induce a tradeoff between incentives and ambiguity sharing, which increases the pay-performance sensitivity. When compensation contracts can be written on some additional signal, such as industry average performance, we find that an ambiguity-averse principal increases the agent’s exposure to the common shock by reducing the use of relative performance evaluation. Thus, optimal contracting involves effectively paying for luck and our model provides a theoretical explanation for the well-documented lack of relative performance evaluation in CEO compensation from the perspective of model uncertainty.