Correlation and relative performance evaluation

This paper reexamines the issue of relative versus joint incentive schemes in a mul-tiagent moral hazard framework. It allows to contrast the inference dimension and the insurance property of relative performance evaluation. Importantly, the widespread idea that the principal should use all the more competitive schemes that the equilib-rium outcomes are more correlated is shown not to be robust. When correlation varies with the efforts chosen, more equilibrium correlation can make joint performance eval-uation more likely to be optimal, because a pair of good performances can become a relatively better signal that both agents work hard than a pair of asymmetric perfor-mances. With risk-averse agents, that informational effect has to be traded off against the agents ’ insurance concerns. As a result, the optimal incentive scheme is some-times mixed, which can be interpreted in a firm context as the use of aggregate profit sharing in combination with selective firing or promotion