Government-sponsored education and training programs have the goal to enhance participants' skills so as to become more employable, productive and dependable citizens and thus alleviate poverty and decrease public dependence. While most of the literature evaluating training programs concentrates on estimating their total average treatment effect, these programs offer a variety of services to participants. Estimating the effect of these components is of importance for the design and the evaluation of labor market programs. In this paper, we employ a recent nonparametric approach to estimate bounds on the "mechanism average treatment effect" to evaluate the causal effect of attaining a high school diploma, General Education Development or vocational certificate within a training program for disadvantaged youth 16-24 (Job Corps) relative to other services pffered, on two labor outcomes: employment probability and weekly earnings. We provide these estimates for different demographic groups by race, ethnicity, gender, and two age-risk groups (youth and young adults). Our analysis depicts a positive impact of a degree attainment within the training program on employment probability and weekly earnings for the majority of its participants which in general accounts for 55 - 63 percent of the effect of the program. The heterogeneity of the key demographic subgroups is documented in the relative importance of a degree attainment and of the other services provided in Job Corps.