How effective are language trainings for immigrants? Evaluating active labor market policies by applying propensity score matching

Abstract: Integration programs are important and at the same time controversial parts of an immigration policy. This paper contributes to this debate by evaluating the effectiveness of German language trainings for immigrants with respect to their labor market performance. To allow for possible sample selection bias, the propensity score matching method is applied. This method matches treated and control persons with similar characteristics such that the assignment to each group can be considered as random placement and treatment effects can be worked out by comparing the outcomes of the corresponding matching partners. The empirical analysis relies on a new administrative dataset containing extensive employment information. Contrary to expectations about the effectiveness of a policy program, the results show significantly negative effects of language trainings, both on the employment status and on earnings. There is some evidence that the effects are more pronounced in Eastern Germany and for female participants. In accordance to previous studies, the negative effects can partly be explained by locking-in effects, which indicate that the job search intensity is usually lower during and after the participation in a program. Data limitations and arising econometric problems in the matching procedure might also cause the effects to be negative.