Zoned Out: The Long-Term Consequences of School Choice for Wealth Segregation

Georgy Artemov, Kentaro Tomoeda · 2025

Read Zoned Out: The Long-Term Consequences of School Choice for Wealth Segregation by Georgy Artemov, Kentaro Tomoeda (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-09967.txt.

Introduction

Georgy Artemov and Kentaro Tomoeda study how school choice systems affect wealth segregation over the long run, once families can also choose where to live. Families buy homes in zones that set admission priority, face shocks to their school preferences, and take part in one of three mechanisms: neighborhood assignment, deferred acceptance, or top trading cycles. Housing and schooling choices feed back into each other in their model.

Neighborhood segregation rises in that same order across the three mechanisms, and prices in zones for oversubscribed schools rise likewise. Choice mechanisms do reduce segregation inside schools relative to neighborhoods, but usually not enough to overturn the ranking created by residential sorting. Two desegregation policies further show how short-term and long-term effects can point in opposite directions.

Catalog record

CREATORGeorgy Artemov, Kentaro Tomoeda
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE125 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2511-09967

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Zoned Out: The Long-Term Consequences of School Choice for Wealth Segregation Georgy Artemov, Kentaro Tomoeda Abstract We study how school choice mechanisms shape wealth segregation in the long term by endogenizing residential choice. Families buy houses in school zones that determine admission priority, experience shocks to school preferences, and participate in one of three mechanisms: neighborhood assignment (N), Deferred Acceptance (DA), or Top Trading Cycles (TTC). Neighborhood segregation increases from N to DA to TTC. DA and TTC reduce school-level segregation relative to neighborhoods but typically not enough to reverse this ranking, and housing prices in oversubscribed zones rise in the same order. Two desegregation policies further illustrate how short- and long-term perspectives can differ. (JEL C78, D47, D63, I21, I28, R31) Keywords: wealth segregation, wealth sorting, school choice, housing, school admission priorities, Deferred Acceptance, Top Trading Cycles, market design A substantial premium for housing in zones of desirable schools is well documented in the empirical literature. 1 1 1 Many studies estimate a premium of 2–4% in housing prices per standard deviation increase in test scores (see 8 for a seminal contribution; 31 for a review; 28 for a recent study exploiting policy variation; and 19 for a study employing an alternative methodology). 26 finds an even higher premium of about 7% for family housing, where school considerations are most salient, though the aggregate effect is consistent with prior estimates. This premium reflects priority access to local school seats. The value of this access depends on the school assignment mechanism—particularly on the extent to which seats are available to out-of-zone applicants—and can influence residential location decisions. Yet most theoretical analyses of school assignment mechanisms treat residential locations—or priorities they confer—as fixed ( 1 , see, e.g., a review by) . Treating residential locations as fixed is appropriate for short-term analysis, but does not account for assignment mechanisms’ impact on long-term residential wealth segregation (or sorting). To provide a long-term analysis, we endogenize families’ location choices in a two-stage model: families first purchase a house in one of the school zones and then participate in school assignment. In-zone housing gives residents priority at their local school, which we assume to be the sole reason families purchase in-zone housing. In our model, families’ preferences over schools are heterogeneous—there is no uniformly best school—but there exists an undersubscribed school that is second best for most families. This assumption makes the matching problem tractable while preserving the key differences in school assignments under the mechanisms we study: Neighborhood Assignment (N), Deferred Acceptance (DA), and Top Trading Cycles (TTC). Assignment mechanisms matter in our model because families experience a school-preference shock after purchasing a house. The shock reflects the fact that, in practice, families choose when and where to buy housing based on factors outside our model, and may find it too costly to relocate once preferences change. Under the flexible mechanisms, DA and TTC, families who no longer wish to attend the in-zone school may be assigned elsewhere, freeing local seats for out-of-zone enrollments (Section 2.1 explains how DA and TTC operate in our model). 2 2 2 Heterogeneity of school preferences—the importance students place on different school characteristics—is well documented in the education policy literature ( 20 ; 24 , see, e.g.,) . If instead there were a universally preferred school, our framework could readily accommodate it (see Section 4 ). Our model can also incorporate, at the cost of extra notation, families who do not experience a preference shock and instead purchase houses with complete information (see Section 4 ). Finally, while preferences for neighborhood amenities can also drive out-of-zone enrollment, we abstract from this channel to focus on uncertainty about school preferences. Out-of-zone enrollments help DA and TTC reduce school-level segregation relative to N in the short term—when neighborhood segregation is held constant—by allowing students from poorer out-of-zone neighborhoods to access schools in richer ones. In the long term, neighborhoods become segregated because families with different wealth levels resolve the tradeoff between the higher cost of in-zone housing and the expected benefit of a better school assignment by making different housing choices. Obtaining priority at an oversubscribed school improves the expected assignment quality under any mechanism with local priority. DA and TTC further improve the expected quality by providing out-of-zone assignment options, which, as we explain next, lead to different location choices across mechanisms: N produces the least, and TTC the most, segregated neighborhoods (Theorem 1 ). As in the short term, DA and TTC continue to reduce school-level segregation relative to neighborhood-level one; yet, this reduction is usually insufficient to offset the increase in neighborhood segregation that arises under these mechanisms, resulting in more segregated schools than N (Theorem 2 ). At the time of housing decisions, families differ only in wealth and their “signals:” their expected values for schools. Because disutility from housing expenditure is lower for wealthier families, two families with identical signals but differing wealth may make different housing decisions. As wealthier families choose in-zone housing more often than poorer families, neighborhoods are segregated under any mechanism that grants in-zone priority. Yet, the level of segregation varies among the three mechanisms. As DA and TTC provide an out-of-zone assignment option, signals play less of a role in housing decisions, compared to N. Consequently, under DA and TTC, wealthier families with weaker signals may outbid those poorer families with strong signals who buy in-zone housing under N. Thus, at the population level, as the importance of signals declines, wealth becomes the primary determinant of housing competition, increasing neighborhood-level segregation under flexible mechanisms compared to N. Between the two flexible mechanisms, TTC results in higher segregation because residents of oversubscribed schools’ zones have a higher probability of obtaining an out-of-zone assignment under TTC than under DA. This probability is identical for all out-of-zone applicants under DA, implying no residential advantage. Under TTC, an out-of-zone assignment requires finding a partner willing to exchange seats, and such partners always exist for students residing in the zones of oversubscribed schools. Hence, out-of-zone assignments are more accessible under TTC, leading to greater segregation. The argument connecting neighborhood segregation to the probability of obtaining an out-of-zone assignment ignores the housing price changes induced by the mechanisms, but we show that the price-driven effect only strengthens our argument. We also show that, under general conditions, housing prices increase from N to DA to TTC (Theorem 3 ). This finding may be of independent interest, as prices capture externalities present in the school choice problem and are a persistent worry among residents of popular school zones in public debates. Neighborhood segregation does not directly translate to segregation at the school level. Relative to N, two offsetting forces operate under DA: it increases residential segregation, as discussed above, but decreases segregation at each school relative to its own neighborhood. This decrease arises because residents of oversubscribed school zones are, on average, wealthier. When they forgo seats at their local school, these seats are filled by poorer students. Theorem 2 quantifies these forces and provides conditions under which one dominates the other. The shape of the signal distribution plays an important role: under a natural parameter restriction, we identify the shape that minimizes the forces driving segregation

Read the complete text