Identifying A Screening Model with Multidimensional Private Information

In this paper I study the nonparametric identification of screening (price discrimination) models when consumers have multidimensional private information about their taste for product characteristics. In particular, I consider the model developed by Rochet and Chone (1998) and determine conditions to identify the cost function, the joint density of taste and the utility functions, from individual level data (on demand and prices). When the utility function is nonlinear, exogenous binary cost shifter is sufficient for identification. Moreover, I show that if there are some consumer covariates and the utility is nonlinear the model is over identified. I also characterize all testable restrictions of the model on the data.