Identification and Estimation of Incentive Problems : Adverse Selection

The adverse selection model is a principal-agent model defined by the objective functionof the principal, the agents’ utility function and the distribution of agents’ types. We provethat the nonparametric identification of this model requires the knowledge of at least one ofthe three functions. We also show that some exogenous changes in the objective function ofthe principal are sufficient to obtain partial or full nonparametric identification of the model.A nonparametric estimation procedure based on these results is proposed. We apply thismethod to test if firms provide the right incentives to their workers. Using contract databetween the French National Institute of Statistics and its interviewers, we test and rejectthe contracts’ optimality. We estimate, however, the loss of using linear contracts insteadof optimal ones to be less than 10%, which may explain why these simple contracts are sopopular.