Abstract
This paper aims to estimate the effect of an institutional training program on participants' chances of finding a job, using a rich dataset which comes from the official records of the Employment Service of Slovenia and taking into account the potential bias due to the existence of unobserved confounding factors. To deal with these selection biases, three methods are implemented in a comparative perspective: (1) instrumental variable (IV) regression; (2) Heckman's two-stage approach and (3) propensity score matching. This paper underlines important divergences between the results of parametric and non-parametric estimators. Some of the results, however, show the impact of the institutional training program on participants' chances of finding a job, especially in the short run. In the long run, however, the results are not so obvious.