Identifying Dynamic Discrete Choice Models: an Application to School-Leaving in France.

In this paper, the authors analyze the non paramatric identification of dynamic discrete choice without consumption smoothing using individual data on discrete choices. They posit that preferences are not restricted and that agents have private information about heir tastes at each period that no other agents, including econometricians can observe. Agents however are supposed to coordinate their expectations of future random tastes on a (common knowledge) distribution function.