Abstract Total public school enrollment has decreased in recent years and is forecast to continue decreasing for the foreseeable future. Experts and educators are concerned about how enrollment declines will change resources available for students who remain in district public schools. Using data on all public school districts from 1998 to 2019, we estimate how resources change as district enrollments increase and how resources change as enrollments decline. Our preferred empirical approach operationalizes all explanatory variables as spline variables, which allows for different covariates for growing and declining districts. This flexible approach allows for the possibility that school funding systems, and therefore fiscal and resource changes, differ between districts with growing and declining student enrollment. We find generally that districts with declining student enrollment have experienced larger increases in per-pupil funding and resources compared to districts with growing enrollments, as enrollment declines were associated with larger increases in local, federal, and, to a lesser extent, state revenues per pupil. Our descriptive results also suggest there may be heterogeneity across districts with respect to relationships between enrollment and fiscal and resource changes, which should be a subject of future research.
Abstract Connecticut expanded interdistrict magnet schools (IMS) intending to reduce racial and socioeconomic segregation across districts, yet the potential unintended effects on student composition in nearby schools remains unclear. Leveraging the staggered rollout of IMS openings, this study finds that IMS openings reduce enrollment by about 5 percent in nearby private K8 and traditional public high schools (TPHS). In private K8, the share of White students declines by 3.5 percent, reducing White exposure for remaining Black and Hispanic students, with effects driven by Catholic schools. In TPHS, the share of students eligible for free or reduced-price lunch increases by 18 percent, suggesting that higher-income students are more likely to exit, with larger effects in a region under court-ordered desegregation. These findings suggest that policies designed to reduce segregation across districts may have unintended socioeconomic and racial compositional consequences for the students left behind in nearby schools.
Abstract Using the detailed college-level data from the College Scorecard on students’ post-college earnings from the near universe of four-year colleges, we assess the usefulness of going beyond comparing colleges based only on median earnings and analyze the descriptive relationship between college selectivity and earnings outcomes and how this relationship may differ by student gender and field of study. We show that considering the full distribution of earnings outcomes can greatly improve our understanding of college-level earnings dynamics. Using different points in the distribution other than the median would markedly change how colleges rank on these metrics, largely due to widely overlapping earnings distributions across colleges. On the link between college selectivity and earnings, we find that there is virtually no correlation among less-selective colleges but a strong relationship among more selective colleges. Earnings are significantly higher at colleges with lower acceptance rates, especially at the high end of the earnings distribution, and the gender earnings gap is also larger at these colleges. We also show that the relationship between selectivity and earnings varies dramatically by field of study, with some fields showing a large return to college selectivity while others revealing none at all.
Abstract We combine data from the NLSY-97 and O*NET to compare the educational attainment, wages, and occupational characteristics of similar (propensity weighted) two- and four-year starters in their thirties. Consistent with prior research, we find that two-year starters are over 20 percentage points less likely to earn a Bachelor’s degree than four-year peers. However, we find mixed results on wages. We generally identify under 10% differences in two- and four-year starters’ wages, but magnitudes and significance are somewhat sensitive to weighting specification in ways that suggest wage penalties of two-year attendance may be greater for students who more closely resemble typical four-year (rather than two-year) starters. Across specifications, we find that differences in key occupational characteristics are more modest than differences in educational attainment. Two-year starters are about 6 percentage points less likely to work in a job that typically requires a Bachelor’s and hold jobs with similar math, though lower social, skill requirements as four-year peers. Finally, we find that two-year starters more often work in the trades and applied sciences, whereas four-year peers more often work in professions such as teaching. Results highlight a need for more understanding of how college level affects life course trajectories beyond educational attainment.
Abstract A large and growing body of research demonstrates the benefits of assigning students to teachers who share their racial or ethnic backgrounds. These benefits range from improved test scores and attendance to reduced suspension rates and higher graduation and college enrollment rates. Yet much of this evidence has been limited to the general student population. Our recent studies extend this literature by focusing on two student groups often overlooked in research and policy: students with disabilities and English learners. Taken together, the evidence underscores the need for educational researchers and policy makers to consider not just broad teacher diversity efforts, but also the targeted importance of race-matching—and in some cases language-matching—for students facing additional systemic barriers.
Abstract Previous research shows that Achieve Atlanta's place-based scholarship and associated services meaningfully improve college persistence and completion. In this follow-up study that uses similar methods but additional and more detailed data, we examine whether scholarship recipients exhibit different student loan portfolios, course-taking patterns, or academic performance. Using regression analyses, we find that students in their first semester of college who receive the Achieve Atlanta scholarship and associated services are less likely to take out loans, borrow lower amounts, earn more credit hours, and attain higher GPAs. Additionally, we find no evidence that the place-based scholarship crowds out institutional aid. These effects are important in their own right and shed light on potential mechanisms that may drive earlier findings on college success.
Abstract The use of four-day school weeks (4dsw) in the United States has expanded rapidly over the past two decades. Previous work examines the impact of 4dsw on student outcomes, but less research to date examines its effect on teachers even though schools in some locales have adopted 4dsw to improve recruitment and retention. This paper uses a difference-in-differences design to estimate the effect of 4dsw adoption in Oregon, a state with early and widespread 4dsw use, on teacher retention in both the short and long term. We find that adopting a four-day week increased turnover among teachers by approximately 2 percentage points overall. These increases are driven by schools that adopted prior to the COVID-19 pandemic and continue to impact schools in the longer term. The findings suggest that policymakers interested in implementing 4dsw for improved teacher retention should exercise caution.
Abstract Against the background of ongoing governmental discussions on the introduction of tuition fees for public higher education, this paper examines the impact of tuition fees on university graduation rates among high-school completers. We analyze the introduction of fees in several German states in 2006-07 (around EUR 1,000 annually), using administrative microdata on all university graduates and a difference-in-differences approach. We observe a reduction of about 2.0 percentage points in university graduation rates among eligible school completers. This reduction in graduation is considerably smaller than the reduction in enrollment of 3.4 percentage points, suggesting that a substantial share of students deterred from enrolling by tuition fees would not have completed their degree even in the absence of fees. This highlights the importance of studying not only enrollment but also graduation, especially in contexts with high dropout rates.
Abstract Using data from the Vietnam Household Living Standards Surveys from 2014, 2016, and 2018, and the Vietnam Population and Housing Census from 2019, I apply a differences-in-differences framework to assess the impacts of Vietnam's 1991 universal primary education policy on women's education, marriage, fertility, and employment. Results from the reduced form estimates suggest that the policy increased the probability of having at least primary education by 3.10 percentage points (3.69 percent) to 16 percentage points (17.02 percent) and the probability of having at least secondary education by 6.30 percentage points (8.51 percent) to 10.50 percentage points (14.19 percent). There is a 6.5 percentage point (12.75 percent) increase in women's probability of being literate in the post-policy period. The policy decreased the number of child deaths. However, there is no evidence of the effects of the policy on women's marriage, the probability of having a job, and the probability of having at least one child. The instrumental variables estimates on the effects of education on fertility and child mortality suggest that having more education may decrease women's fertility and child mortality. The instrumental variables estimates of the effects of education on wages further indicate that having more education may increase women's annual wages.
Abstract Community college graduation rates remain low, despite the existence of interventions shown to improve student outcomes. Adoption of these interventions has been limited, in part due to concerns about financial sustainability. This study estimates the extent to which such interventions “pay for themselves” from the college's perspective—that is, whether the tuition revenue and state funding they generate offset their implementation costs. Using cost and impact estimates from nineteen rigorously evaluated interventions, we simulate implementation at 857 community colleges across forty-one states and systems. On average, these interventions cost around $2,100 per student and generate $200 of revenue per student, leaving the college with a gap of around $1,900 per student to cover. Under more optimistic assumptions, these interventions generate $500 of revenue per student, leaving a gap of $1,600 per student. There is, however, variation across interventions, with one low-cost intervention recouping almost all its costs on average. There is also variation across colleges, driven primarily by differences in state funding models. These findings highlight the need to identify interventions that are both effective and financially sustainable—and to expand public and philanthropic investment if proven strategies are to be adopted at scale.
Abstract Many states incentivize college students to major in fields aligned with specific, often “in-demand” industries. While their goal is often to raise students’ labor market outcomes, little is known about whether matching one's degree with an industry of work improves employment and earnings. We leverage a novel education–industry crosswalk applied to student and worker panel data covering over 295,000 graduates to estimate an education–industry match premium that leverages within-person variation in earnings. We document which fields have the most and least education–industry matching, how match premia vary across fields, and how match premia evolve over time. We show that workers in industries “matched” with their college degree field experience an average earnings premium of about 5–10 percent, though these premia vary by gender, field, level of degree, and how long a worker has been working in a matched industry. These findings shed new light on how education, occupation, and industry explain wages and offer important insights for future work and individual investments in higher education.
While numerous studies examine the effects of school competition on student performance, little research directly addresses a key critique of competition: its potential to negatively affect parental engagement and voice. We draw on Hirschman's theory of voice to argue that voucher-based school competition increases opportunities for exit but may crowd out voice. To assess the causal effects of competition on parental voice, we employ a robust two-way fixed effects difference-in-differences framework, comparing municipalities in Sweden that introduced competition with those that did not. Our findings indicate that school complaints decline following the introduction of competition. This decrease in voice is driven by neither a decrease in problems in school nor by changes in teaching staff quality or attrition. This suggests that the decrease in complaints is driven not by an increase in school quality but rather by a change in how much parents voice their concerns. These results offer new insights into the unintended consequences of school competition reforms, highlighting how marketbased policies may shape school quality beyond their intended effects.
Grappling with challenges stemming from the COVID-19 pandemic, colleges and universities faced the grim realities of serious budget shortfalls. The federal government responded with three waves of the Higher Education Emergency Relief Fund (HEERF I, HEERF II, and HEERF III), which provided nearly $5 billion for all eligible Minority-Serving Institutions (MSI). This brief provides a descriptive analysis of the MSI provision of each HEERF by examining allocations across different MSIs, which reveal considerable distribution differences when comparing between two-year and four-year institutions, across designations, and per student at different MSIs. While shaped by the unique circumstances of the pandemic, this research facilitates a greater discussion of MSI funding policy among researchers, advocates, and policymakers by offering considerations for future equitable appropriations. The brief argues that a more nuanced approach to federal appropriations, which moves beyond the current and contentious proportional funding practice, is necessary. Furthermore, the brief contends that a more equitable approach to funding MSIs would require, at a minimum, for each MSI designation be funded independently of one another and, in doing so, account for their unique and respective histories, missions, needs, and characteristics.
U.S. colleges face persistent inequities—not only in who enrolls, but in who ultimately graduates. While institutions have expanded efforts to diversify incoming classes, students from historically underserved backgrounds, including those who are first-generation and economically disadvantaged, continue to face several barriers which affect their rate of degree completion. This paper evaluates the causal impact of the a bridge program called BSP, an eight-week pre-collegiate summer initiative at a large public land grant university designed to support such students in their transition to college. Leveraging a fuzzy regression discontinuity design (FRDD) within a local randomization inference framework, I find that BSP substantially improves first-year completion and second-year persistence. While participants receive a $2,500 annual scholarship, comparative analyses of similar financial aid programs suggest that the bundled nature of BSP—including structured summer programming, academic mentoring, and peer support—is central to its effectiveness. These findings underscore the importance of holistic interventions that address both financial and non-financial barriers to student success.
The effect of remote learning on student performance has been a frequent topic of research and discussion in the aftermath of the COVID-19 pandemic, yet little is known about the impact of remote instruction on the performance of teachers. This study documents how relative effectiveness of teachers changed when moving from in-person to remote instruction and analyzes the characteristics of teachers associated with greater relative effectiveness during remote instruction. Using matched student/teacher-level data from three large metro-Atlanta school districts, we estimate teacher value-added models to measure the association between teacher characteristics and a teacher's relative contribution to test score growth before and during the period of virtual instruction in the 2020-21 school year. We find evidence of increased variation in overall teacher effectiveness during remote instruction, driven largely by changes in the relative performance of early elementary (K-2) and middle school teachers. Veteran teachers appear relatively more effective in virtual instruction than their less-experienced peers, with lessexperienced teachers performing relatively worse regardless of in-person teaching ability. Finally, we find that the very best in-person teachers are more likely to experience large declines in relative effectiveness when shifting to remote instruction compared to a baseline period with in-person instruction
Abstract In 2015, the City University of New York (CUNY) launched a new program—Accelerate, Complete, and Engage (ACE)—aimed at improving college graduation rates. A randomized-control evaluation of the program found a nearly 12 percentage point increase in graduation five years after college entry. Using this impact estimate and national data on earnings by gender, age, and degree status, we estimate incremental expected long-run benefits and costs for participants, as well as intergenerational effects for the children of participants, relative to “business as usual” for the control group. Our main estimate indicates net social benefits of $48,000 over a lifetime per participant from greater earnings and labor force attachment, improvements in health, and savings in public transfers. A major contribution of our analysis is the estimation of second-generational benefits. Including intergenerational benefits for children who grow up in newly higher-earning families more than triples this estimate, to $167,000 in net social benefits per participant. These results are sensitive to assumptions about whether the impact on graduation after five years persists indefinitely, or whether the control group eventually catches up. Still, net social benefits are strongly positive even under our most conservative assumptions.
Critical race theory (CRT) and claims of political indoctrination in K-12 classrooms are at the forefront of the ongoing culture wars surrounding public education. Despite a wave of legislative action targeting CRT-related instruction, little systematic evidence documents the extent to which critics' claims have merit. Using a nationally representative survey of American high school students, we examine the frequency with which students are exposed to CRT-related content and political bias in their classrooms. We further examine how students' experiences differ by political preference, race/ethnicity, and local political context, finding that classroom content likely mirrors community preferences. Our findings suggest that although some divisive topics are discussed, their frequency and context do not support the claims of the most vocal critics.
Teacher hiring plays an important role in determining the composition of the teacher workforce and a small but growing body of research has sought to better understand the degree to which various types of applicant information predict teacher outcomes. We add to this literature by examining the relationship between structured ratings of applicants collected from their professional references and teacher retention, an outcome that has implications for student achievement and the expenditure of school and district resources. We find that teachers who scored one standard deviation above average in their reference ratings were 2.6 percentage points more likely to remain at the schools that hired them. When we include rater-fixed effects in the model (so comparisons are between teacher applicants within rater), we find reference ratings are substantially more predictive of retention (including district- and state-level retention), suggesting that attention to issues of interrater reliability could make applicant ratings more useful. The results of this study support the idea that surveying professional references to get assessments of teacher applicants is a promising and low-cost approach to assessing job candidates that can complement other screening practices.
Community colleges are designed to serve their local communities by responding to the workforce demands of their local economies. Despite its ability to decrease the funding gap between two- and four-year institutions, many states provide no local funding to their community colleges. Among states that provide local funding to community colleges, we know very little about the specific sources of local funding that may be associated with educational inequities as suggested by K-12 literature. We developed a novel dataset regarding local funding sources for all community colleges across five years and used it to describe the sources of local funding to community colleges and their relationship to the colleges' institutional characteristics. Results showed that community colleges were predominantly funded by local property taxes but were also funded by other local sources, including consumption and income taxes. We also found geographic variations in where different local funding sources were collected. Our study provides a more nuanced understanding of local funding to community colleges and may aid policy makers in developing funding strategies to support community colleges and the students they serve.
