This paper extends the traditional focus of active labor market policy evaluation from a static comparison of participation in a program versus nonparticipation (or participation in another program) to the evaluation of the effects of program sequences, i.e. multiple participation or timing of such programs. We use a dynamic evaluation framework that explicitly allows for dynamic selection into different stages of such sequences based on past intermediate outcomes to analyse multiple programs, the timing of programs, or the order of program. Such a demanding analysis is made possible by having available exceptionally comprehensive data of the Austrian workforce population. Our findings suggest that (i) active job search programs are more effective after a qualification program compared to the reverse order, that (ii) multiple participation in qualification measures dominates single participation, and that (iii) the effectiveness of specific labor market programs deteriorates the later they start.
Not available in German. This paper advocates the use of a nonparametric bounds analysis to check the robustness of the results of applied evaluation studies to the problem of a lack of common support. The typical responses by researchers of either ignoring it, or obtaining estimates only for the subpopulation within the common support, can both be misleading: Ignoring the problem may result in biases because the comparison group may not be comparable. Deleting observations at best yields an estimator that is consistent for the common support. When treatment effects are heterogeneous inside and outside the common support this estimator does is inconsistent. Furthermore, useful information is ignored, because the response of the treated to the treatment can be estimated even outside the common support. This information can be used to derive bounds with width depending on the configuration of the data. The application to an evaluation study of Swiss active labour market policies shows that the relevance of the bounds for changing the interpretation of the results depends very much on the particular data configuration. Download Discussion Paper: (pdf, 264 kb)
