Association for Public Policy Analysis & Management

Concerns about the rising tuition in higher education, and continued inequities in outcomes, have led to national interest in making community college free of charge for a substantial share of the population though "promise scholarships" and other means. Initially promoted by President Obama, the idea is now being taken up by both current Democratic presidential candidates and many governors across the country.

What effect would the various forms of free community college have on student outcomes? This is a difficult question to answer, but we can make some progress toward an answer through randomized control trials (RCTs). In this study, we summarize evidence from an RCT of a college promise made to all first-time 9th graders in half of the 36 Milwaukee Public Schools (MPS) high schools in 2011, using a cluster-paired design. If students obtained a 2.5 GPA, attended class 90 percent of the time, graduated from high school on time, and met certain other conditions, they received up to $12,000 for college, to be used in any 2- or 4-year public colleges in the State of Wisconsin (as well as some private institutions). The scholarship is enough to pay the direct costs (tuition and fees) for a 2-year degree.

Leveraging the random assignment, I examine effects on both high school and college outcomes. Preliminary results on high school outcomes suggest no average treatment effects (ATEs) on standard academic outcomes (GPA, attendance, and graduation), but positive effects on FAFSA completion, the number of colleges applied to, participation in college access programs, and perceived college affordability. As predicted, the ATEs vary over time in a U-shaped pattern, with the largest effects arising near the announcement and at the very end of high school. The effects are also largest for students who were near the performance thresholds at baseline. There are signs of negative effects away from the thresholds, explaining the overall null ATE. Estimation of effects on college entry are in progress.

These results have implications for the design of financial aid (e.g., the use of performance thresholds) and the cost-effectiveness of increased aid and free community college. In particular, the variation in results near the thresholds raises questions about the equity and efficiency of performance requirements.