Partial identification of wage effects of training programs

In an evaluation of a job-training program, the influence of the program on the in-dividual wages is important, because it reflects the program effect on human capital. Esti-mating these effects is complicated because we observe wages only for employed individuals, and employment is itself an outcome of the program. Only usually implausible assumptions allow identifying these treatment effects. Therefore, we suggest weaker and more credible assumptions that bound various average and quantile effects. For these bounds, consistent, nonparametric estimators are proposed. In a reevaluation of a German training program, we find that a considerable improvement of the long-run potential wages of its participants.