Correlation and relative performance evaluation

Abstract

This paper reexamines the issue of relative versus joint incentive schemes in a multi-agent moral-hazard framework. The model allows a full analysis of the information and dependence structure. An important result is that the widespread notion that greater correlation in outcomes calls for more competition is not robust. First, when the dependence structure is effort-sensitive, the optimal incentive scheme in general mixes elements of relative evaluation and joint evaluation. Second, under limited liability, higher equilibrium correlation tends to make joint performance evaluation more desirable. Examples are provided regarding incentives in firms, finance and innovation.