Research

This paper assesses the implications of heterogeneity in teacher effects for student achievement and classroom segregation. Leveraging administrative data from North Carolina public schools, I estimate a teacher value-added model that incorporates heterogeneity in teacher effects on test scores across five student demographics: English language learner (ELL) status, economic disadvantage, disability, gender, and race. There is substantial variation in teacher effects, particularly for ELL students and students with disabilities. Using these estimates, I identify student-teacher reassignment policies that maximize achievement under varying constraints. Reassignment yields notable gains in average test scores (up to 0.04σ in both math and reading). A weak correlation between comparative advantage for teaching ELL students and a teacher's absolute quality allows ELL students to benefit from large match effects without sacrificing the overall quality of their assigned teachers. As a result, reassignment significantly narrows the achievement gap between ELL and English-proficient students, with modest reductions across the other dimensions. However, these improvements are accompanied by large increases in classroom segregation, indicating a trade-off between raising achievement and maintaining classroom diversity.

Abstract

In a large urban setting with school choice, we examine whether transportation poses a barrier to attending high-rated high schools. By analyzing public transportation travel times, we find significant racial disparities in student travel times to the nearest high-rated school. Black students experience the longest travel times and are the least likely to rank a high-rated school as their top choice. We estimate a discrete choice model using ranked school choice data to estimate parental preferences for schools and their distaste to distance. Next, we simulate new express bus routes to high-rated schools from low socioeconomic status (SES) neighborhoods that are poorly connected to the public transportation network. Our findings suggest that with reduced travel times, up to 25% of students would switch to a high-rated school as their first choice. This indicates that transportation time significantly influences school choice decisions, and targeted policies, such as express bus routes, could encourage families to select higher-rated schools.

Abstract

Value-added models typically attribute student test score gains to a student’s teachers or their schools, but teachers and schools may both play a role in determining student outcomes. In this paper, we first provide empirical evidence that accounting for school quality is important for correctly estimating teacher quality (and vice versa). We then develop a new statistical approach to jointly estimate teacher and school value-added. In doing so, we find that schools differ in how well they educate students, even after accounting for varying levels of teacher quality at the schools. This implies that ignoring such school effects biases conventional measures of teacher value-added, and we show empirically that adjusting for measures of school quality leads to meaningful changes in who is considered an effective or ineffective teacher.

Abstract

Many school systems implement choice-based assignment to schools while incorporating mechanisms into the matching process to promote diversity. However, differences in preferences and residential patterns across student groups can limit these efforts. This paper examines Oakland Unified School District's (OUSD) school assignment system, where higher-income families' preference for neighborhood schools contributes to income segregation. We simulate alternative assignment rules that increase low-income students' access to high-demand schools. Our findings suggest that small adjustments favoring low-income applicants can substantially reduce income segregation. However, we observe that denying students their top-choice school leads some to exit the district—a cost for OUSD, as it results in decreased diversity and funding. Our analysis shows this exit effect is similar across income groups, though destinations differ: low-income students often transfer to charter schools, while high-income students shift to private institutions.