States have an economic interest in developing and retaining human capital, but face challenges from underinvestment in education and the outmigration of skilled workers. This paper evaluates the long-run effects of a prominent policy response to these challenges: state merit aid programs that subsidize in-state college attendance. Using national survey data on college enrollment and U.S. Census data, I exploit differences in program adoption timing to compare changes in outcomes between states and across cohorts in a staggered difference-in-differences design. Merit aid shifts students from out-of-state to in-state institutions, which are relatively less selective on average. There is little evidence that these programs increase degree attainment or long-run in-state residence. For women, program exposure reduces post-college employment and earnings, while there is limited evidence of effects for men. Welfare analysis shows net losses for program recipients and adopting states, with additional negative fiscal spillovers to other jurisdictions.
with Jing Liu and David Blazar, revise and resubmit at Education Finance and Policy
This study evaluates the impact of expanding high school Computer Science (CS) course offerings on students’ college major choices and early-career labor market outcomes. Using longitudinal data from Maryland, we exploit variation from the staggered expansion of CS course offerings across schools. We find that taking a high school CS course increases students’ likelihood of declaring a CS major by 8 percentage points and receiving a CS bachelor’s degree by 5 percentage points. Access to CS coursework also raises students’ total quarters of employment by 8 percent over ages 24-27. Notably, students who are female, socioeconomically disadvantaged, Black, or lower-achieving experience similar or larger gains in CS degree attainment and labor market outcomes; however, their take-up rates are much lower than their peers. Our findings suggest that closing this participation gap requires policy innovations beyond course expansion to actively promote CS participation among underrepresented groups.
Media: Fordham, Axios, Fortune, Code.org, EdSurge
Research Briefs: Brookings, MLDS Center
with Nolan Pope and George Zuo
While schools devote immense resources to developing students’ skills, little is known about the relative long-run returns to specific subjects and skill dimensions across the K-12 curriculum. Using linked administrative data from Maryland, we estimate how K-12 test scores across subjects, subscores, and grades translate to postsecondary and labor market outcomes. While math and English both strongly predict high school and college completion, the math earnings premium is four times larger. Controlling for educational attainment and college major choices only modestly attenuates this gap. Patterns in subscores indicate that earnings associations are concentrated in analytical and applied reasoning rather than content knowledge. Finally, math and ELA are partially interchangeable for high school graduation and early postsecondary attainment, but become increasingly complementary for BA attainment and earnings.
Research Briefs: Urban Institute, MLDS Center
with Thomas Dee, Elizabeth Huffaker, Maranna Yoder, and George Zuo
Draft Available Upon Request
Access to advanced coursework plays a central role in shaping students’ academic trajectories, yet placement often depends on discretionary decisions by schools and families. We study one of the earliest known district adopters of automatic enrollment, where most schools initially used one of two achievement cutoffs—either the 50th or 75th percentile—to automatically enroll rising middle school students in advanced math. Using regression discontinuity designs, we find that automatic enrollment significantly expanded access to advanced math while producing durable gains in progression and course completion. At the lower cutoff, the policy more than doubled initial advanced math enrollment and eventual AP math passing on an intent-to-treat basis, with substantially larger implied effects among students whose placement changed. At the higher cutoff, effects were similar through early high school but attenuated thereafter.
with Thomas Dee, Elizabeth Huffaker, Maranna Yoder, and George Zuo
Policies that use test score thresholds to automatically place students into advanced math coursework have become increasingly common in recent years, yet there is limited evidence on whether these policies improve student achievement or how their effects vary across schools. To address this gap, we study Texas’s statewide Middle School Advanced Mathematics Program, which requires districts to automatically enroll sixth-grade students in an advanced math course if they score at or above the 60th percentile on their fifth-grade standardized test beginning in the 2024-25 school year. We leverage the discontinuity created by this test score threshold to examine how effects on student achievement differ across student subgroups and how policy implementation varies across districts. These findings provide early statewide evidence on who benefits from automatic enrollment and how effects vary across school contexts, with implications for policies designed to expand access to advanced math coursework.
with Jordy Berne, Trey Miller, and George Zuo
Community college students have low rates of persistence and completion, particularly those who enroll part-time while balancing other responsibilities in addition to school. One policy that aims to address these low persistence rates is the Alamo Colleges District Summer Momentum Plan (SMP), which automatically provides eligible students with up to six credit hours of free summer tuition. Using a synthetic difference-in-differences design that compares changes in outcomes at Alamo Colleges with those of a synthetic control constructed from other Texas community college districts, we evaluate the effects of SMP on continued enrollment, degree completion, and early-career earnings. These findings shed light on whether reducing both financial and administrative barriers improves outcomes for community college students, the vast majority of whom enroll part-time. This study assesses the cost-effectiveness of SMP relative to other interventions and informs the design of scalable approaches to improving college persistence.
The Landscape of Auto-Enrollment Policies for K-12 Advanced Math Coursework, 2026 (with George Zuo, Thomas Dee, Elizabeth Huffaker, and Maranna Yoder), RAND Research Report
Draft Available Upon Request