This paper aims at evaluating the effect of the amount of older workers exits (aged 50 or more) on the entries of youngsters at a local labour market level, during years 1985 - 2002. If we can observe some effect of the exits on the entries, it will shed light on the substitution between older workers and youngsters. Moreover, since in our model the causal agent cannot be specified a – priori, we don’t know what causes what. Hence, we are actually looking for a correlation between these two quantities. Systematic differences in background characteristics, between local markets with different levels of the older workers exits, can bias the effect estimation on the entries of youngsters. In order to adjust for this, we apply propensity score methods as extended and generalized in a setting with a continuous treatment by Hirano and Imbens (2004). Our results show a positive and significant correlation between exits of older workers and entries of youngsters._x000d_ _x000d_ Keywords: Synthetic firms, Evaluation, Non-experimental methods, Continuous treatment, Matching, Generalized propensity score, Dose-response function._x000d_ _x000d_ JEL codes: C13, C49, J68.
