his paper presents estimates of the impact of programs for unemployed workers on the performance of program beneficiaries in Mexico. We emphasize the significance of applying methodologies capable of avoiding statistical bias attributable to unobserved variables when measuring the impact on earnings and allowing to us properly estimate unemployment duration and work status after exiting from unemployment.

Using a technique that combines the matching method with the double difference method (pre and post-intervention differences in differences between the treatment and control group) we measure the impact of SICAT on unemployed males with previous working experience who participated in the 2002, 2003 and 2004 programs. Outcomes using this method do not support the hypothesis that SICAT has a positive impact on the earnings per hour of program participants three months after finishing the training provided by the program. Furthermore, when the impact is statistically significant, such as in years prior to 2004, the effect was the opposite to the one that was expected, as if those participating in SICAT had been stigmatized in the labor market to the extent that it turns out to be counterproductive.