Evaluating continuous training programmes by using the generalized propensity score

Summary.  The paper assesses the heterogeneity of treatment effects arising from variation in the duration of training. We use German administrative data that have the extraordinary feature that the amount of treatment varies continuously from 10 days to 395 days (i.e. 13 months). This feature allows us to estimate a continuous dose–response function that relates each value of the dose, i.e. days of training, to the individual post‐treatment probability of employment (the response). The dose–response function is estimated after adjusting for covariate imbalance by using the generalized propensity score, which is a recently developed method for covariate adjustment under continuous treatment regimes. Our data have the advantage that we can consider both the actual and the planned durations of training as treatment variables: if only actual durations are observed, treatment effect estimates may be biased because of endogenous exits. Our results indicate an increasing dose–response function for treatments of up to 120 days, which then flattens out, i.e. longer training programmes do not seem to add an additional treatment effect.