Econometric Methods for Public Policies Evaluation

This document presents an overview of econometric tools recently developed for empirical ex post policy evaluation. The emphasis is laid on selection problems (how to disentangle the real effect of a policy from the personal features of the affected people), and on the practical issues of a policy evaluation (access to relevant data, identification hypotheses, interpretation of the results) illustrated by examples provided by recent empirical articles. It presents first the Rubin framework for causal inference, and details then the four main empirical designs for evaluation: selection on observables, differences in differences, instrumental variables and regression on discontinuity.