Do Charter Schools Contribute to Rising Within-School Racial Segregation?

Crema (2025) investigates whether charter school openings increase racial segregation within traditional public school classrooms. She analyzes North Carolina administrative data covering 97 charter openings from 1997–2015, along with classroom-level records showing the racial composition of students in grades 1–5. She finds that segregation in nearby public schools rises as soon as a charter opening is announced, before any students transfer. Classroom segregation increases by roughly 6–15 percent, meaning about 8 more non-white students in an average school would need to switch classrooms to achieve racial balance.

Why This Article Was Selected for The Policy Scientist

Crema’s (2025) article examines a broad policy question: how charter school entry alters the racial composition of classrooms within nearby public schools, a topic linked to systemwide patterns of separation that affect access to teachers, peers, and academic opportunities. The issue is timely as many states continue to expand charter sectors while reassessing the structural sources of segregation. The article contributes by identifying within-school sorting responses that major prior studies, focused mainly on cross-school shifts, could not detect. The dataset is unusually granular, and the event-study design offers credible causal evidence. Future work could be strengthened through randomized or quasi-experimental designs.

Full Citation and Link to Article

Crema, A. (2025). The segregative effects of charter schools. American Economic Journal: Economic Policy. Advance online publication. https://www.aeaweb.org/articles?id=10.1257/pol.20230539

Central Research Question

This study asks whether charter school openings alter the racial composition of classrooms inside nearby traditional public schools (TPS), and whether these within-school responses meaningfully contribute to the broader segregative effects attributed to school choice. The author focuses on a mechanism that has received little empirical attention: changes in the assignment of white and non-white students to classrooms within the same school and grade. The central question is therefore twofold: first, do TPS increase classroom-level racial segregation in response to nearby charter openings; second, does accounting for such within-school shifts meaningfully change the field’s understanding of charter-induced segregation more generally.

Previous Literature

Research on charter schools and segregation has largely concentrated on between-school sorting—how student transfers, residential choices, and school choice policies affect the racial composition of individual campuses. Foundational work by Bifulco, Ladd, and others documents modest but consistent increases in cross-school segregation stemming from charter entry, with heterogeneity by geography, curriculum, and student demographics. More recent national analyses, such as Monarrez et al. (2022), identify small aggregate effects that become sizable in localities with rapid charter growth. These studies emphasize parental school selection and residential mobility as the primary channels through which segregation intensifies.

A parallel literature examines within-school segregation, noting that 30–50 percent of total segregation may occur inside buildings rather than across them. Sociologists and education economists have shown that classroom assignments deviate systematically from random allocation, often producing racially unbalanced learning environments tied to tracking, teacher requests, and various informal administrative practices. Yet this line of work has rarely been integrated with school choice research. The competitive pressures exerted by charter schools have been shown to influence staffing, spending, and test-score dynamics in TPS, but prior studies have not focused on how classroom formation practices themselves may constitute a strategic response. The present paper connects these strands by arguing that competition may reshape the internal organization of TPS, not only their student pools, and by supplying empirical evidence consistent with this mechanism.

Data

The analysis combines administrative data from the North Carolina Education Research Data Center (NCERDC) with detailed records of classroom assignments across grades 1–5 from 1994 to 2017. These classroom files include student-level race indicators, course identifiers, and section assignments, which allow the author to compute the dissimilarity index—a widely used measure of racial unevenness—at the school-year level. The sample includes 628 TPS located within 10 miles of at least one charter opening, of which 390 experienced a charter introduction within a 5-mile radius during the 1997–2015 period.

The dataset includes 97 elementary charter openings in two waves: an initial expansion following the 1996 North Carolina Charter School Act and a second wave after the state lifted its charter cap in 2011. The author exploits both the timing of opening announcements and specific grade-expansion commitments within charter applications, which provide variation in competitive pressure before students can physically transfer. The dataset is unusually detailed for U.S. administrative sources, especially given that most states do not track classroom assignment with comparable consistency. These features enable precise measurement of within-school segregation and the timing of TPS responses.

Methods

The empirical strategy is an event-study difference-in-differences design that compares TPS located within five miles of a new charter to schools slightly farther away but still within local educational markets. Following best practices for staggered treatment timing, the author adopts the Cengiz et al. (2019) stacked cohort approach and includes saturated fixed effects to avoid bias from heterogeneous treatment effects. The primary outcome is the school-by-year dissimilarity index, calculated across classrooms within each grade and averaged to produce an annual school-level measure.

To distinguish strategic behavior from compositional changes in TPS enrollment, the author uses two identification strategies that isolate periods when competitive information is available but cross-school student movement cannot yet occur. First, the analysis restricts to openings announced at least one year before operation, allowing estimation of segregation effects before the charter enrolls students. Second, leveraging charters that commit to adding specific grades in later years, the author tests whether segregation rises within those grades before the expansions become operational. These tests allow cleaner attribution of segregation changes to TPS agency rather than mechanical shifts in racial composition originating from charters’ enrollment.

The author also constructs district-level segregation measures with and without classroom information to assess whether incorporating within-school changes materially alters estimated charter effects on overall segregation. Additional analyses consider other within-school margins, such as tracking by achievement, access to gifted programs, and teacher assignment, though data limitations require a narrower interpretation for these results.

Findings/Size Effects

Across specifications, TPS located near a new charter increase classroom-level racial segregation. This effect occurs even before charter schools begin enrolling students, which strongly suggests that TPS administrators modify classroom assignments as an adaptive response to anticipated competition rather than in reaction to actual enrollment losses. Using pre-announced openings, the author finds a roughly 6 percent increase in classroom segregation at event time zero. When exploiting promised future grade expansions, the estimated increase rises to approximately 15 percent. These effect sizes imply that, at an average school with roughly 300 students, an additional eight non-white students would need to reassign classrooms to reach racial balance after charter entry.

The segregative response is strongest in schools where white students are not the overwhelming majority and in cases where opening charters serve disproportionately white student bodies. Curricular specialization—particularly Montessori, classical education, and college-preparatory models—predicts larger TPS responses, consistent with the idea that charters appealing to advantaged families pose a more salient competitive threat. The analysis also reveals that TPS achieve a higher fraction of feasible segregation following entry, reinforcing that the observed effects are not mechanical artifacts of changing student bodies.

When extended to district-level segregation measures, the results indicate that within-school changes contribute modestly but meaningfully to total segregation, although they do not overturn the consensus that charters increase segregation only moderately on average. The author finds little evidence that TPS reassign students to more effective teachers or smaller classrooms, but some suggestive evidence that gifted program access expands disproportionately for white students. Test-score heterogeneity increases, with gains concentrated among students above their school’s median performance.

Conclusion

This paper provides the first clear evidence that charter school expansion influences segregation not only between schools but also within them, via altered classroom assignments in TPS facing competitive pressure. By leveraging announcement timing and grade-expansion commitments, the author isolates behavioral responses unrelated to student transfers, strengthening the causal interpretation. The findings suggest that internal organizational practices within TPS play a previously underappreciated role in the segregative dynamics surrounding school choice.

The study deepens the literature by identifying mechanisms embedded in day-to-day administrative decisions rather than parental school selection alone. While the within-school channel does not dramatically alter the overall assessment of charter-induced segregation, it adds a meaningful layer to understanding how education systems respond to competition. The North Carolina context—characterized by rapid charter growth and detailed administrative data—supports strong internal validity, though generalizability may depend on the structure of charter sectors and administrative discretion in other states. The paper highlights the need for further research on how competitive pressures shape internal practices within public schools and how these changes interact with broader patterns of educational inequality.