Relative Performance Evaluation between Multitask Agents

Abstract

We investigate the moral hazard problem in which the principal delegates multiple tasks to two agents. She imperfectly monitors the action choices by observing the public signals that are correlated through the macro shock and that satisfy conditional independence. When the number of tasks is sufficiently high, relative performance evaluation functions effectively for unique implementation, where the desirable action choices are supported by an approximate Nash equilibrium, and any approximate Nash equilibrium virtually induces the first-best allocation. Thus, this is an extremely effective method through which the principal divides the workers into two groups and makes them compete with each other.