Comparing the Results of Youth Training Programs in Latin America and the Caribbean
The evidence on the effectiveness of youth training programs in Latin American and the Caribbean (LAC) tends to be encouraging regarding the quality of employment of beneficiaries (positive impacts are observed regarding the access to formal employment), although there is significant heterogeneity across countries and by gender of the beneficiaries. It is not clear how easily one can generalize from the results of an impact evaluation in a particular country. We address the underlying heterogeneity in the characteristics of the beneficiaries of youth training programs in LAC by relying on the individual-level data used in the experimental impact evaluations of three of these programs. We show that we can identify, and characterize, the individuals satisfying a common support condition, i.e. those who are similar across programs. We use non-experimental multiple treatment estimators to eliminate differences across programs, which work better for men than for women. For men satisfying common support (i.e. comparable), who have worse initial conditions than non-comparable individuals, the positive treatment effects on formality disappear for some programs. The results highlight the importance of treatment effect heterogeneity, which may have implications for focalization and program design. They also make explicit the limits to the external validity of each of the experiments, and how difficult is the interpretation of the results from meta-analysis studies. By stressing for which types of individuals it is possible to make comparisons across programs, our study points towards a more nuanced comparison of impact evaluation studies of youth training programs.