We study the effectiveness of nonexperimental strategies in adjusting for comparison group differences when using data from several programs, each implemented at a different location, to compare their effect if implemented at alternative locations. First, we adjust for individual characteristics differences simultaneously across all groups using unconfoundedness-based and conditional difference-in-difference methods for multiple treatments. Second, we adjust for differences in local economic conditions and stress their role after program participation. Our results show that it is critical to have sufficient overlap across locations in both dimensions and illustrate the difficulty of adjusting for local economic conditions that differ greatly across locations.

We semiparametrically estimate average causal effects of different lengths of exposure to academic and vocational instruction in the Job Corps (JC) under the assumption that selection into different lengths is based on a rich set of observed covariates and time-invariant factors. We find that the estimated effects on future earnings increase in the length of exposure and that the marginal effects of additional instruction decrease with length of exposure. We also document differences in the estimated effects across demographic groups, which are particularly large between males and females. Finally, our results suggest an important lock-in effect in JC training.