In this paper, we study sensitive of propensity score methods to the specifications through Monte Carlo experiments. We find that the estimated treatment effects on the treated are not sensitive to the specifications if the unconfoundedness assumption holds. When the unconfoundedness assumption...

We analyze the causal effect of training duration on employment outcomes of unemployed workers. Observed training durations might be endogenous. We use planned duration as an instrument for actual duration. LATE estimates indicate that an increase in duration has a positive impact for short...