Using data from the Current Population Survey’s Education Supplement, we examine the countercyclicality of post-secondary education (PSE) enrollment in the U.S. between 1977 and 2023. The unemployment–enrollment relationship is weak prior to 2000 and when Covid-19 years are included, but it is much stronger in 2000-2019. Focusing on these two decades, we show that higher unemployment increases enrollment at both 2- and 4-year colleges. We also show that persistence (continued enrollment among prior PSE participants) accounts for a larger share of the countercyclicality than matriculation (new enrollments). As for matriculation, eighteen-year-olds experience a particularly pronounced increase in the probability of starting a 2-year college degree when unemployment is high. We also find that men in their 20s are more likely to start a 4-year college degree when economic conditions deteriorate.
Compensatory education programs seek to reduce educational inequality, yet evidence on their effectiveness remains mixed. I estimate the impact of a program in Chile that targets additional resources to schools with a share of disadvantaged students above four different thresholds (15, 30, 45 and 60%). Effect estimates are consistently statistically indistinguishable from zero at each cutoff, for both academic and non-academic outcomes. I link the absence of detectable effects to the fact that schools at the cutoffs do not meaningfully increase their spending. These results suggest that policies that provide modest, formula-based marginal revenue increases may prove ineffective if they are not accompanied by strong implementation guidance on how to spend.
Career and Technical Education (CTE) has expanded with the intention to meet a changing economy and broaden pathways to higher education and training. Using longitudinal administrative data from Massachusetts, we generate causal evidence on the impact of high-quality CTE expansion in comprehensive high schools. Leveraging the staggered rollout of these programs, we find that access to a new CTE program induces 5.4 percent of prior non-participants to enroll. On average, while CTE exposure decreases four-year college enrollment, initial earnings one-year post-graduation increase by 8.8 percent among those with in-state employment. However, this early wage premium diminishes by age 23. We find suggestive evidence that these patterns are driven largely by male students and those with access to less college-aligned programs. Conversely, more college-aligned programs induce a 4.8 percent increase in quarters of employment through age 23 while experiencing a more muted decline in four-year college enrollment and completion.
This study investigates how distance learning during COVID-19 affected educational outcomes of lower secondary students in Sweden. In contrast to most previously studied contexts characterized by prolonged and fully remote school closures, Sweden represents a school system with a high level of digital preparedness, in which distance learning was implemented for a limited period, often combined with in-person instruction, and accompanied by exemptions for students with special needs. Using a difference-in-differences design with entropy balancing, we compare pre-pandemic and pandemic-affected cohorts. Our findings suggest that students in schools that adopted remote instruction performed similarly on standardized tests to those in schools that maintained in-person teaching throughout the pandemic. Moreover, progression to upper secondary school was not negatively affected. In some cases, we even find positive effects of remote instruction.
This paper examines how linguistic similarity between students’ mother tongue (L1) and the language of instruction (LOI) affects early literacy skills in Cameroon, a multilingual country with over 270 living languages. Using data from the 2019 Early Grade Reading Assessment (EGRA) and lexicostatistical methods, I construct two continuous measures of phonetic similarity between 15 major local languages and the two official LOIs, French and English. Identification relies on a two-stage least squares strategy that exploits geographic variation in long-run linguistic diffusion. The results show that students whose L1 is phonetically closer to the LOI perform better in listening comprehension, partly offsetting the learning loss associated with linguistic mismatch. The effects are larger within the Francophone education subsystem, in rural areas, among students who receive parental support with homework, and among younger learners, with estimates declining with age. A complementary provincial analysis shows that the predicted gap between low- and high-similarity scenarios varies across provinces and is larger in several low-performing areas. These findings underscore the importance of linguistic alignment in language-in-education policy for improving early learning and promoting equity in multilingual settings.
Asthma is a leading chronic condition among school-aged children and a major driver of absenteeism. Since attendance is linked to academic success, graduation, and earnings, interventions reducing asthma-related absences may improve educational trajectories. We evaluate a program that placed asthma case managers in New York City public schools to engage directly with students and families. Using novel linked health-education data and a triple-differences design exploiting staggered rollout and asthma targeting, we find significant gains in school-based asthma management: absenteeism among children with asthma fell by 5%, narrowing the asthma/no-asthma gap by 20%, and active asthma prescriptions rose by nearly 10%. These results show that school-embedded supports can improve children’s health while also boosting educational outcomes.
While all children in Sweden are entitled to free, universal preschool from age 3, enrollment rates for children with an immigrant background – and especially for those in newly-arrived families – remain well below the 95 percent national average. At the same time, studies suggest that these children have particularly high returns from preschool attendance. Thus, policies that increase preschool enrollment among children with an immigrant background have potential to improve long-term educational and labor market outcomes and narrow gaps relative to native-born children. We evaluate a reform in 2023 that required municipalities to offer reserved preschool slots to 3–5-year-olds in newly-arrived immigrant families without the parents needing to apply – so-called automatic offers. Using a difference-in-differences specification, we compare next-year enrollment rates for currently non-enrolled children in newly-arrived families and other families and find that the policy substantially increased preschool enrollment. Our results suggest that administrative hurdles prevent some immigrant families from enrolling their children and that simplifying the application process is an effective policy tool for increasing preschool enrollment in this group.
This study provides pre-registered, experimental evidence on the use of non-generative artificial intelligence (AI) chatbots to support students in large-enrollment undergraduate courses. We find the chatbot messaging increased students’ final grades and engagement with academic supports, such as tutoring. Treatment effects were generally consistent across student demographics, with the exception of treated women in a Microeconomics course, who earned final grades that were seven percentage points higher than women in the control group. This study provides evidence that integrating AI-enabled outreach and communication to students in their college courses can enhance student engagement, learning, and course performance.
Despite widespread concern about teacher shortages, there is little evidence on when teaching positions are posted and how many people apply for them. Using detailed job-posting and application data on nearly 57,000 applicants from 19 traditional school districts and 24 charter organizations, we provide descriptive evidence on the timing of job postings, the supply of applicants, and the characteristics of applicants, emphasizing how each varies across organizations and between the district and charter sectors. We find that late hiring is less prevalent than prior single-district studies suggest—about 11% of openings are posted late—but varies widely across organizations. Applicant supply differs widely across organizations and by sector: districts average 3.3 applicants per opening, versus 6.1 for charters. We also find sectoral differences in applicant pool composition; compared to charters, district applicant pools have higher percentages of White applicants and applicants with prior teaching experience.
Using data from the Current Population Surveys Education Supplement for 1977-2023, we explore two important, yet understudied facets of the pattern of countercyclical post-secondary education (PSE) enrollment in the U.S. First, we show that economic downturns are associated with higher enrollment probabilities at both 2- and 4-year colleges among young men, but only at 2-year institutions among young women. Second, we show that the overall increase in enrollment propensities during downturns is primarily driven by persistence (i.e., changes in enrollment among individuals with prior PSE participation), rather than matriculation (i.e., new enrollments). However, higher unemployment rates increase matriculation probabilities at 2-year colleges among 18-year-old men and women, and at 4-year colleges among individuals in their early 20s. Our findings improve our understanding of the dimensions along which aggregate economic fluctuations induce changes in human capital acquisition.
Technological advancements in automation are transforming job tasks and skill demands, raising concerns about the adaptability of the workforce. While reskilling and upskilling are widely recommended for workers in high-risk occupations, empirical research consistently finds an “automation training gap”: workers in occupations classified by experts as highly susceptible to automation participate in less training than workers in lower-risk occupations. Existing studies have primarily focused on employer-side explanations, but the role of workers’ own perceptions has received less attention. Using representative survey data for the Dutch workforce matched to occupational OECD automation risk indicators, we examine whether self-assessed job automation risk is associated with training participation and helps explaining the automation training gap. Our findings show that workers who perceive high automation risk have a 7 percentage-point higher probability of participating in training. Moreover, the negative association between expert-assessed automation risk and training is significantly attenuated when workers perceive their own job to be at risk. We also find that lower rates of self-initiated training among workers who do not perceive automation risk contribute to the overall gap. These results suggest that worker perceptions play a critical role in human capital investment decisions under automation risk. Policies aimed at increasing awareness of task-based technological vulnerability may enhance training participation and help mitigate skill mismatches in the evolving labor market.
We study a low-cost, teacher-led program for 4th–5th graders in Brazilian public schools that embeds financial, fiscal, and civic concepts in short, comics-based stories paired with structured class activities. We evaluate a school-level randomized trial in Brazil’s capital Brasília covering 106 schools (≈13,600 students). Outcomes – knowledge and self-reported habits/beliefs – were measured with an age-appropriate instrument administered in person by an independent survey firm under exam-like conditions. Intent-to-treat estimates on a complete-case analysis sample show gains of 0.10 SD in financial knowledge, 0.06 SD in fiscal knowledge, and 0.06 SD in civic knowledge, all statistically significant and robust to alternative treatments of missing responses. We also find significant improvements in declared financial habits and beliefs and in civic beliefs. Beyond knowledge and beliefs, students also report greater school interest, more family interaction around money and public services, and higher future-self expectations. Placebo outcomes not targeted by the program, such as mathematics and monitoring of common goods at school, show no robust impacts. Monitoring data indicate that higher teacher engagement with the materials is associated with better student outcomes. Overall, the results indicate that culturally salient narrative media, delivered during regular class time, can measurably improve targeted economic and civic competencies at scale.
This paper examines earnings differences between first-generation and continuing-generation college graduates across 24 OECD countries using data from the OECD Survey of Adult Skills (PIAAC). In most countries analyzed, first-generation graduates earn significantly less than their peers from college-educated families, with an average gap across all countries of approximately 7.5%. We investigate potential mechanisms behind this result and find that first-generation graduates are less likely to pursue postgraduate education, more likely to hold professional degrees, more likely to work in smaller firms and lower paying occupation, and tend to have lower cognitive skills. These findings highlight the need for policy interventions to enhance educational mobility and promote equality of opportunity.
This study examines the impact of private school attendance on segregation and student achievement in compulsory school in Denmark. We show that increased private school attendance is driven by students from high socio-economic groups. Leveraging variation across municipalities, grade and calendar years and instrumental variables based on private school openings, we find that higher private school enrollment is associated with higher segregation of disadvantaged children. From event study models of the private school openings and a mover design that controls for student parental background, peer parental background, past achievement and non-cognitive scores, we find small achievement effects of private school attendance.
Retention of effective teachers has significant implications for student outcomes. Thus, un-derstanding factors that influence teacher retention is a central concern in education research. I use a performance pay scheme to study the relationship between salary and teacher decisions to exit or become an assistant principal along the performance distribution of teachers. Applying a regression discontinuity design, I find that teachers whose salary can increase across the cutoff are 4.2 percentage points less likely to exit the district public school system when they score just above the cutoff, even in the absence of dismissal threats. I find no statistically significant effect for teachers with no financial incentive around the cutoff. Exploring heterogeneity in teacher responses across the performance distribution yields suggestive evidence that high-performing teachers are less responsive to salary. While low-performing teachers, who are subject to additional performance requirements to advance in the salary schedule, are more likely to exit when they face no financial incentive and score just above the cutoff. I find suggestive evidence that performance burdens can be as effective as dismissal threats in incentivizing attrition, but only among low-performing teachers with competitive outside options.
Using administrative data from North Carolina, I study the decision to take the SAT in the presence of universal ACT testing. I find that low-income students are less likely than their peers to take the SAT in addition to the state-mandated ACT, and that they improve upon their initial ACT scores by less when doing so. Taken together, these disparities decrease low-income students’ rankings in the test score distribution when evaluating students on their maximum score rather than their initial ACT score. Income gaps in SAT taking are partially driven by differential responses to the ability signal sent by a student’s initial ACT score.
Tuition-free college programs are gaining momentum as policymakers address rising college costs and workforce readiness. Despite their growing adoption, limited research examines how workforce-focused eligibility criteria impact student outcomes beyond enrollment. This pre-registered study employs two within-study quasi-experimental designs — difference-in-discontinuity and difference-in-differences — to estimate the causal impact of Virginia’s Get a Skill, Get a Job, Get Ahead (G3) initiative on financial aid and academic outcomes for community college students. Launched as a pandemic recovery effort, G3 aimed to reverse enrollment declines and address labor shortages by leveraging simplified ‘free college’ messaging and offering last-dollar scholarships and additional advising support for students in high-demand workforce programs. We find that G3 significantly increased FAFSA completion, total financial aid, and grant aid, with gains concentrated among middle-income students. The program also reduced student borrowing, consistent with crowd-out by grant aid. Certificate completion rose by 5.1 to 6.6 percentage points, and higher enrollment translated these gains into a net increase in the number of students earning certificates in targeted fields. These findings suggest workforce-targeted tuition-free programs can expand financial aid access, increase the supply of certified workers in priority fields, and better align higher education with workforce demands.
We examine the Great Recession's long-term impact on federal student loan borrowing and repayment, using detailed federal administrative records and leveraging comparisons across labor markets with varying unemployment severity. A one percentage point increase in recession unemployment caused 7% higher outstanding debt and 6% more defaults. By 2019, the recession accounted for 19%-32% of undergraduate debt increases and 10%-25% of default increases across institutional sectors. Borrowers enrolled at the onset of the recession experienced the largest effects on debt accumulation and repayment distress. For this group, recession-induced borrowing is linked with sustained declines in program completion, suggesting reduced financial welfare.
Shortly after the new president of Colombia took office in 2018, the Minister of Education announced the ending of the Ser Pilo Paga scholarship program. The scholarship, awarded on both the basis of need and merit, provided awardees with funding covering both tuition costs and a living stipend at any high-quality university in the country. It was replaced by a new scholarship, Generacion E - Excelencia, which was formally identical in terms of benefits and requirements. Exploiting the need-based eligibility threshold using a regression discontinuity design, we find that this policy transition reduced test scores for eligible students. An analysis of how this transition affected the socioeconomic composition of the highest performers shows that the representation of eligible students decreased. These findings underscore the importance of program visibility and trust in sustaining the behavioral effects of financial aid.
This study examines whether increasing young individuals’ access to higher education by creating and expanding higher education institutions affects their survival or mental health, or those of their parents. Our quasi-experimental analysis leverages changes in access to university resulting from the geographical expansion of the Finnish university system in the 1960s and 1970s. The results suggest that greater access to university for 19-year-olds reduces their probability of mental health-related hospitalization and drug use while also generating positive spillovers on their mothers’ longevity. However, we do not find strong effects on individuals’ early mortality, fathers’ old-age survival or parental mental health.
