Effects of Heterogeneous Length of Stay in a Residential Job Training Program for at-Risk Youth

Introduction: Job Corps is the nation’s largest vocationally focused education and training program for at-risk youth. Evaluating the impact of Job Corps has been of great interest to policymakers and economists because other job training programs for at-risk youth have not been found to be effective. Previous studies examining the Job Corps effects have improved our understanding of the cost benefit aspects of Job Corps. However, the literature lacks an explanation for how the program works. This study investigates the likely heterogeneous effects in terms of length of stay. Because the participants enroll for different lengths of time, we hypothesize that heterogeneous duration of stay leads to heterogeneous program effects. We then investigate which training classes were more useful than others for attainment of desirable youth outcomes.

Methods: This study executes a secondary data analysis of the National Job Corps Study. Our analytic sample includes 11,313 youth aged 16 to 24 at the time of the baseline interview. Youth outcomes are employment and earnings, welfare dependency, and criminal behavior, measured at 48-month follow-up. Intervention variables are five discrete lengths of stay. Our analysis includes twenty class variables that constitute academic and job training program. These variables are also used as controls in statistical models. We use a regression tree technique to separate the length of stay into a number of different parts. We next construct multivariate generalized linear models to understand the heterogeneous effects of length of stay in the context of other features of the program.

Results: Results indicate that length of stay plays a key role in determining the program effect on youth outcomes. Optimal length of stay for labor market outcomes is 6-8 months, and 8-14 months for criminal behavior. Less than 2 months of stay has deleterious effects on all youth outcomes. As to class effects, world of work and performance panels classes are beneficial for increasing total hours worked and earnings per week. Clerical and parenting skills classes help reducing criminal activities. Surprisingly, reading, math, and GED classes do not play key roles in enhancing youth outcomes.

Conclusions: This study brings new empirical evidence to the debate on the Job Corps effects in youth development. The study establishes that there is indeed a net association between length of stay and youth outcomes. Exploration of exact program content was intrinsic to a better understanding of the program impact, and the results suggest the ways in which future program can be improved.