Is labour market training a curse for the unemployed?
Evidence from a social experiment

In this paper, we investigate the impact of classroom training programs on subsequent labor market outcomes in Denmark. In 1994 a social experiment was conducted, where unemployed applicants for labor market training were randomized into treatment and control groups. The experimental data are contaminated by the presence of no-shows and cross-overs, which implies that the traditional experimental estimator is biased. We interpret our experiment within an economic model of agents maximizing outcomes facing different cost regimes. We present two sets of results interpretable within this model, one set being effects of training among people induced to switching participation status because of exogenous changes in regimes and another set being average treatment effects within the two regimes. To obtain the second set of results we implement matching estimators using extensive survey and register data. We find - surprisingly - that classroom training significantly increases individual unemployment rates and decreases employability. We discuss possible reasons for this finding and some related policy issues.