September 30, 2011, 13:45–15:00
Toulouse
Room MS 003
Decision Mathematics Seminar
Abstract
We study a repeated principal-agent model with subjective evaluations. We make minimal assumptions on the joint distribution of the principal's output and the agent's private signal. We show that there exists a class of perfect Bayesian equilibrium which approaches efficiency as the discount factor increases to1. In each equilibrium in this class, the principal evaluates the agent every T periods. The principal pays a bonus and asks the agent to work for T more periods if the evaluation is positive. The agent is fired if the evaluationis negative. The model shows that the inefficiency due to subjective evaluations vanishes as the parties become more patient.