A Critical Survey of Empirical Methods for Evaluating Active Labor Market Policies

This paper considers different methods for solving the evaluation problem. I highlight the role of heterogeneity in program impacts in defining evaluation parameters of interest and in interpreting estimated program impacts. I discuss the strengths and weaknesses of social experiments and conclude that they require careful implementation and interpretation. I review and critique two popular non-experimental evaluation methods: difference-in-differences and propensity score matching. I find that the former relies on assumptions at odds with the empirical data and that the latter is not a magical solution to all evaluation problems. Finally, I argue for the importance of paying attention to data quality and general equilibrium effects.