Learning But Not Earning? The Value of Job Corps Training for Hispanics

The National Job Corps Study (NJCS) was a four-year longitudinal social experiment in which over 15,000 Job Corps eligible applicants were randomized into treatment and control groups. Using experimental estimators, Job Corps was found to have positive impacts in the weekly earnings of white and black youths 48 months after randomization, but not for Hispanic youths, a puzzling outcome that eluded explanation in the NJCS. This study considers explanations for why Job Corps does not increase the earnings of Hispanics in the NJCS. First, we show that the randomization in the NJCS did not create comparable treatment and control groups for Hispanics, possibly because Hispanics represent 20% of the entire randomized sample and are more concentrated in metropolitan areas. We then apply alternative estimators that adjust for systematic differences in observable and time-invariant characteristics of the Hispanic subsample, but still find statistically insignificant effects of Job Corps 48 months after randomization. Finally, we estimate the "net treatment effect" controlling for post-treatment experience to advance an explanation for why Job Corps fails to benefit Hispanics 48 months after randomization: non-treated Hispanics earn a significant amount of labor market experience during the study compared to treated Hispanics (and non-treated blacks and whites). This higher level of experience translates into higher earnings that Hispanic treated individuals are not able to overcome by the end of the study, despite having higher earnings growth in the 48-month period compared to whites and blacks (either treated or non-treated)