Many Dropouts? Never Mind! – Employment Prospects of Dropouts from Training Programs

Abstract

The employment effects of training programs for the unemployed are frequently studied, but little is known about the number, the characteristics, and the labor market prospects of those who drop out of these programs. The aim of this paper is to contribute to filling this gap. Dropouts are identified in German administrative data. The medium- and long-term effect of dropout is estimated using a bivariate dynamic random effects probit model which takes into account state dependence and selection on unobservables. The model is estimated with Markov Chain Monte Carlo (MCMC) methods.