INTRODUCTION

The papers by Costas Meghir and Fablen Postel-Vinay and Jean-Marc Robin provide useful discussions of developments in two important areas in labor economics and public finance. There is little overlap between the two papers, and so I consider them seperately.

Meghir assesses the relative strengths and weaknesses of policy evaluation based on straightforward experimental designs and on structural models. He makes a strong case that the two approaches are complementary, particularly when dynamics and general equilibrium effects are likely to be important. Below I begin with a short summary of his most important points. I then discuss the potential to utilize more of a continuum of models between a simple experimental or quasi-experimental analysis on the one hand and a dynamic structural model on the other, even in complicated dynamic settings where reduced form analysis is difficult. In particular, I suggest using simulation to derive the reduced form implied by the structural model and then treating the implied reduced form as a baseline reduced form around which to expand.

Postel-Vinay and Robin provide a valuable overview of developments in search/matching models and their application using matched employer/employee data. I briefly supplement their coverage of the literature and their research agenda.

DYNAMIC MODELS AND POLICY EVALUATION

Meghir addresses three broad issues of methodology. The first is the role of analysis conducted in a simple “treatment effects” framework in the presence of experimental variation (or plausibly exogenous non-experimental variation) versus analysis based on a structural model.