Long-Term Impacts of Youth Mentorship Program

Bell (2025) analyzes the mentorship program run by the national non-profit Big Brothers Big Sisters of America (BBBSA). The program is targeted at youth (mainly aged between 10 and 14) who have faced adversity. In addition, the participants in the study were majority male and majority belonging to a racial minority group. Youth are matched with a volunteer mentor who has been screened through criminal background checks and character references. Mentor and volunteer pairs are matched on specific characteristics such as activities of interest and shared life experiences. These relationships are designed to last for at least a year and expected to continue for longer if they are beneficial. Mentor-mentee pairs commit to meeting once a week for a social or cultural activity such as playing games, seeing a movie, going on a bike ride, or eating at a restaurant.

In 1991, BBBSA implemented a randomized control trial to study the effects of their mentoring program. Given that not all participants who are randomly assigned to treatment actually end up matched to a mentor, the paper uses this research design to measure the intent-to-treat effect of the mentorship program. The paper also estimates a treatment on the treated effect using a larger observational sample of applicants to a single BBBSA location (Boston) between 1991 and 2010. These applicants are then linked to the tax records to provide information on long-run outcomes.

The paper estimates the effect of mentoring on economic outcomes such as wages, non-employment, unemployment and EITC benefits, as well as on behavioral outcomes such as marriage, teen birth, and incarceration. The analysis using only the RCT finds positive and significant effects on behavioral outcomes like college-going, and a negative but imprecise effect on earnings between ages 25 and 30. However, within the observational sample, the paper estimates a statistically significant effect of 20% on earnings. The paper combines the RCT and observational estimates for a more conservative estimate of a 15% effect on earnings. This estimate is the main result used in the authors’ calculation of an MVPF for the BBBSA youth mentorship program.