Working Papers
Distance to Schools and Equal Access in School Choice Systems [paper]
Revised and Resubmitted, Journal of Public Economics
This paper studies the extent to which school choice systems reduce disparities in access to high-achieving peers and effective schools, focusing on the role of geography. Using data from Boston Public Schools, I estimate a model of school demand and use it to construct counterfactual assignments that isolate the effects of residential location and assignment rules. I find that racial gaps in access persist under school choice and are largely driven by differences in distance to schools. For Black families, geography plays a central role: if they faced the same distribution of distances as White families, they would be assigned to more effective schools and to schools with higher-achieving peers, in some cases reversing observed gaps. For Hispanic families, geography explains differences in access to high-achieving peers but not to effective schools, where outcomes are similar to or more favorable than those of White families. In contrast, assignment rules such as proximity priorities and restricted choice sets have limited effects on these disparities. The results highlight the importance of travel costs in shaping parental demand and show that, even in choice-based systems with broad access and transportation provision, geography can remain a key source of inequity.
Match Effects and the Gains from Alternative Job Assignments: Evidence from a Teacher Labor Market, with Elton Mykerezi, Aaron Sojourner, and Aradhya Sood [paper]
Updated February 2026, Under review
This paper studies the relative importance of teacher match effects and teacher general effectiveness in the production of student learning and quantifies the learning gains attainable from alternative teacher assignments. Because general effectiveness and match effects are typically confounded in observed value-added measures, we develop and estimate a framework that separately identifies each, allowing match quality to vary along both observable student characteristics and unobservable teacher-school dimensions. Using more than a decade of administrative data from a large urban school district, we address endogenous teacher sorting by exploiting quasi-random variation in assignments induced by differences in driving time between teachers and schools. We find that match effects exhibit substantial dispersion, comparable in magnitude to general teacher effectiveness, and that observed sorting patterns are negatively correlated with match quality. Counterfactual assignments that are acceptable to teachers under existing compensation schemes indicate that alternative assignment of teachers while holding student composition fixed can raise average test scores by approximately 0.13 standard deviations, with significant gains driven by unobservable match effects.
Selected Work in Progress
Disentangling School and Peer Effects: A Non-linear Approach with School Lotteries, with Minseon Park and Suk Joon Son
Draft coming up soon
This paper leverages data from a centralized school choice system to separately estimate the contributions of school and peer effects to student learning. A central challenge in this setting is that students are not only sorted across schools, but are also exposed to different—and potentially endogenous—peer environments. We show that school admission lotteries generate quasi-random variation that can be exploited to separately identify these effects, and we implement this strategy within a flexible model of peer effects that allows student outcomes to depend on the full distribution of peer achievement rather than only on summary measures such as mean peer quality. To make this high-dimensional problem tractable, we use functional principal component analysis to summarize peer composition, capturing the primary dimensions of variation in peer distributions while preserving the flexibility needed to model rich peer effects.
Targeting and Efficiency in the Allocation of Childcare Subsidies, with Minseon Park, Suk Joon Son, and Esperanza Johnson
This project studies the trade-off between efficiency and redistribution in waitlist-based allocation systems applied to Massachusetts’ childcare subsidy program. The program allocates assistance through two mechanisms—vouchers that allow parental choice of providers and direct assignments to specific centers—creating natural variation in the degree of choice families face. Restricting choice may reduce match quality but can improve targeting if families with better outside options self-select out, while also encouraging provider expansion in lower-income areas by stabilizing funding. Using administrative data on all applicants and providers from 2015 to 2019, we quantify how variation in the degree of choice shapes both allocative efficiency and redistribution toward more disadvantaged families.