Association for Public Policy Analysis & Management

In the United States, the gap in college completion rates between students in low- versus high-income families is larger for students born on the cusp of the millennial generation (i.e., the early 1980s) than for those born decades earlier (Bailey & Dynarski, 2011). Recent policy discussions have centered on the genesis and evolution of such gaps, as well as on ways to improve outcomes for low-income students across the educational pipeline. In this paper, we study the impacts of a generous, multifaceted aid program for low-income students on measures of college success: earned credits, GPA, and graduation.

In the fall of 2003, the University of North Carolina at Chapel Hill announced a new aid program for low-income, college-bound students. Beginning with students entering UNC-Chapel Hill in the fall of 2004, this aid program (called  the “Carolina Covenant”) covered all the financial costs of college attendance (without loans) and provided awardees with additional supports (e.g., peer and faculty mentoring, career and professional development opportunities, and social events).

Covenant scholars are identified after the admission process using a straightforward, transparent family income cutoff. We exploit this cutoff in the context of a regression discontinuity (RD) design to estimate impacts on our set of postsecondary outcomes. As a complementary strategy, we employ data on students entering UNC-Chapel Hill in the fall of 2003 in the context of a difference-in-differences design to estimate effects. In the RD design, the counterfactual consists of students whose family income placed them just above the Covenant threshold (in each year of the program’s existence). Aside from the aid award, we assume these students are just like those who barely met the eligibility criteria to receive the scholarship (in terms of measurable characteristics like financial need, as well as unobservable characteristics). In the difference-in-differences design, students who would have been eligible for the Covenant in 2003 (before it existed) comprise the counterfactual.

Preliminary results are consistent across these two research designs and suggest that eligibility to receive the Carolina Covenant had little impact on the likelihood of completing college in four years for initial cohorts of scholars, but a positive, statistically and practically significant effect for more recent cohorts of scholars (i.e., 2007 and later). These results are consistent with changes over time in components of the aid package. We discuss this evolution and also explore impacts of the program on credit accumulation and academic performance in college.

References

Bailey, M., & Dynarski, S. (2011). “Inequality in Postsecondary Attainment.” In Greg Duncan

and Richard Murnane (Eds.), Whither Opportunity: Rising Inequality, Schools, and Children's Life Chances, pp. 117-132. New York, NY: Russell Sage Foundation.