56

This paper surveys recent developments in the literature on microeconometric evaluation methods. The canonical statistical model is the Rubin causal model, the assumptions of which are outlined in the paper. The model provides a suitable framework for defining and examining the identifiability conditions of parameters of interest, e.g., the effect of participation in an active employment policy program. Our overview puts special emphasis on matching estimators in the situation where the selectivity depends only on observables, and on parametric and semiparametric estimators when the selectivity depends on both observable and unobservable characteristics. This is illustrated using results from studies we have undertaken to evaluate several active employment policy measures and programs in France.