How Do School Closures Affect Students’ Educational and Labor Market Outcomes?

Kim (2026) examines how public school closures affect students’ academic progress and later economic outcomes. He uses linked Texas administrative records covering K–12 schooling, college attendance, employment, and earnings, combined with difference-in-differences designs comparing students in closed schools with matched controls. School closures reduce math and reading scores by about 0.03 standard deviations and increase disciplinary days by 23 percent. By age 26, closures lower high school graduation by 1.8 percentage points and college enrollment by 1.4 points. At ages 25–27, employment falls 1.0 percentage point and annual earnings decline by about $700, or 3.5 percent.

Why This Article Was Selected for The Policy Scientist

This article addresses an increasingly important policy problem as demographic change and declining enrollment force more school districts to consider consolidation and closure. School closures matter well beyond short-term test performance because disruption during childhood may affect educational attainment, employment, and earnings many years later. Kim’s contribution is especially timely because enrollment pressures are likely to make closures increasingly common. The Texas administrative data are unusually strong, linking K–12 records to college and workforce outcomes over many years. His matched difference-in-differences designs provide a credible causal inference strategy, substantially strengthening the analysis over conventional multivariate regression. The statewide setting also improves generalizability beyond studies of individual districts. The article is forthcoming in the American Economic Journal: Economic Policy, a leading peer-reviewed journal in applied economics and public policy.

Full Citation and Link to Article

Kim, J. (2026). The long shadow of school closures: Impacts on students’ educational and labor market outcomes. American Economic Journal: Economic Policy, forthcoming. https://www.aeaweb.org/articles?id=10.1257%2Fpol.20240815

Central Research Question

The study asks whether public school closures have causal effects on students that extend beyond temporary disruptions in test performance. The central question is whether being displaced by a school closure changes students’ academic achievement, attendance, disciplinary outcomes, high school completion, college enrollment, employment, and earnings later in life. This question is important because prior research often finds that test-score losses following closures diminish within several years, which could suggest that closures impose only short-lived costs. The study tests whether that interpretation is incomplete by examining outcomes into students’ mid-20s. It also asks whether the effects vary across students, particularly by grade level, economic disadvantage, race and ethnicity, and the quality of the schools students attend after displacement. The broader objective is to identify whether school closure should be understood primarily as a temporary academic interruption or as a more consequential disruption with lasting educational and labor-market effects.

Previous Literature

The study builds on several strands of prior research. One concerns the direct effects of school closures. Earlier studies generally emphasize short-run academic outcomes, particularly standardized test scores. Brummet (2014), Engberg et al. (2012), Larsen (2020), Steinberg and MacDonald (2019), Beuchert et al. (2018), and Torre and Gwynne (2009) are among the key studies cited. Across this literature, school closures often produce an initial decline in academic achievement, although some studies find that test scores recover after several years. This recovery has contributed to the view that the effects of closure may be temporary.

The study also draws on research showing that childhood educational experiences can have consequences that are not fully captured by contemporaneous test scores. Chetty, Friedman, and Rockoff (2014) demonstrate that teacher quality can affect students’ later earnings, while Chetty et al. (2011) link classroom experiences to adult outcomes. Heckman, Pinto, and Savelyev (2013) likewise show that childhood interventions can have persistent effects on later socioeconomic outcomes. Jackson, Johnson, and Persico (2016) provide related evidence that school inputs can shape long-term educational and economic trajectories.

A third relevant literature concerns student mobility. Hanushek, Kain, and Rivkin (2004), Rockoff and Lockwood (2010), Schwerdt and West (2013), and Schwartz, Stiefel, and Cordes (2017) examine how moving between schools affects achievement. These studies generally suggest that school transitions can be disruptive. School closures provide a distinct setting because students are forced to change schools even when their families do not move residences. The study therefore contributes to both the school-closure and student-mobility literatures by examining outcomes well beyond test scores.

Data

The analysis uses longitudinal administrative records from Texas. The data link individual students across three major state systems: the Texas Education Agency, the Texas Higher Education Coordinating Board, and the Texas Workforce Commission. This linkage allows students to be followed from K–12 schooling into postsecondary education and then into the labor market.

The K–12 records begin in the 1994–1995 academic year and contain information on standardized math and reading scores, attendance, disciplinary actions, high school graduation, student demographic characteristics, and school characteristics. Postsecondary records identify enrollment in Texas colleges and universities and bachelor’s degree attainment. Workforce records provide quarterly employment and earnings information for workers covered by the state unemployment insurance system.

The analysis identifies 323 public school closures in Texas between 1998 and 2015 for the short-run analysis. The author requires that a school appear on the Texas Education Agency’s closure list, disappear from the administrative data, and not be replaced by a substantially overlapping school at the same address. Most closures were associated with enrollment declines, financial pressures, aging facilities, or district restructuring rather than persistent academic underperformance.

The short-run samples contain approximately 31,600 students for test-score analyses, 57,300 for disciplinary outcomes, and 69,200 for attendance. The long-run analysis uses 130 closed schools and includes approximately 42,400 students in grades 2 through 12.

Methods

The study uses two difference-in-differences research designs intended to estimate causal effects. For short-run outcomes, the analysis exploits within-student changes over time. Students enrolled in schools that eventually close are compared with students in matched control schools before and after closure. Control schools are selected using nearest-neighbor matching based on school type, location, racial and ethnic composition, and measures of economic disadvantage.

The short-run models include individual fixed effects and matched-group-by-time fixed effects. This design controls for stable student characteristics and for common changes affecting similar schools over time. Event-study specifications are also used to test whether treated and comparison students followed similar trends before closure. The absence of substantial differential pre-trends provides support for the parallel-trends assumption required for a causal interpretation.

For long-run outcomes, the author cannot compare the same students before and after closure because outcomes such as college attendance and adult earnings are only observed later. Instead, the study compares younger cohorts who experienced closure with older cohorts from the same schools who had already progressed beyond the affected school before it closed. These differences are then compared with analogous cohort differences in matched control schools.

The long-run models include school fixed effects, cohort-by-match-group fixed effects, student demographic controls, and pre-closure academic and attendance measures. Event-study specifications are again used to assess preexisting differences across cohorts. The combination of matching, fixed effects, cohort comparisons, and pre-trend testing provides a comparatively strong quasi-experimental identification strategy.

Findings/Size Effects

School closures produce immediate academic and behavioral disruptions. Math scores fall by approximately 0.030 standard deviations and reading scores by about 0.034 standard deviations. These test-score effects generally dissipate within three years. Attendance and disciplinary effects are more persistent. Absences increase by approximately 0.05 days, or 0.7 percent relative to the pre-closure mean. Disciplinary action increases by approximately 0.49 days during the first two years, representing a 23 percent increase. After three to four years, the increase in disciplinary days reaches approximately 0.78 days.

The long-term results are substantially more consequential. By age 26, experiencing a school closure reduces the probability of graduating from high school by 1.8 percentage points, equivalent to a 2.7 percent decline relative to the comparison mean. College enrollment falls by 1.4 percentage points, or 2.8 percent. The estimated effect on bachelor’s degree completion is not statistically significant. College quality, measured using expected earnings associated with the institution attended, declines by approximately $191, or 0.9 percent.

Labor-market effects persist into the mid-20s. Employment at ages 25 through 27 declines by 1.0 percentage point, or approximately 1.9 percent. Average annual earnings fall by roughly $700, representing a 3.5 percent reduction. Only about one-quarter of this earnings decline can be explained by lower educational attainment or lower expected earnings associated with college quality. This indicates that the labor-market consequences extend beyond changes in formal education alone.

The effects are heterogeneous. Students who are in higher grades when their schools close experience larger and more persistent negative effects than younger students. Economically disadvantaged students also experience larger adverse effects on several outcomes. Students transferring to lower-performing schools show more persistent test-score declines and larger earnings losses, while students moving to higher-performing schools sometimes experience greater disciplinary disruption. Students who move with fewer peers or travel farther to their new schools also tend to experience larger negative effects.

Conclusion

The findings indicate that school closures have consequences that are not adequately summarized by short-term test-score changes. Although average math and reading scores recover within several years, students exposed to closures experience persistent increases in disciplinary problems and measurable reductions in high school graduation, college enrollment, employment, and earnings.

The results therefore extend the existing school-closure literature by demonstrating that apparent recovery in test scores does not imply the absence of long-run effects. The study also broadens the literature on student mobility by showing that forced school changes, even without residential relocation, can have lasting consequences.

The evidence suggests that disruption itself is an important mechanism. Students frequently move to schools that appear comparable or even stronger on conventional academic measures, yet they still experience negative outcomes. Changes in peer environments, separation from classmates, adaptation to new institutional expectations, longer travel distances, and behavioral adjustment may therefore contribute to the effects.

The study estimates that the present discounted value of lifetime earnings falls by approximately $31,000 per displaced student. Applied to roughly 250,000 students affected by school closures nationally in a typical year, this corresponds to an estimated aggregate annual cost of approximately $7.8 billion. The study does not attempt to estimate offsetting fiscal or organizational benefits from closures, so this figure should be interpreted as a measure of costs borne by displaced students rather than as a complete cost-benefit assessment.