Introduction

Air pollution exposure is an immense global challenge, with 99 per cent of the world’s population living in locations that exceed the World Health Organisation’s (WHO) recommended levels for ambient particulate matter (PM2.5). This research paper explores how sustained exposure to air pollution over a school year affects the rate of human capital accumulation. Unlike previous literature that focused on single-day exam performance, this study examines the cumulative impact on the rate at which students actually learn. The outcome data come from the College Board and comprise nearly 19 million exam scores from 10 million U.S. high school students attending approximately 25,000 schools between 2007 and 2016.

Mechanisms and policy context

PM2.5 particles are small enough to enter the lungs and bloodstream, potentially affecting neural function through oxidative stress and low-grade inflammation. The report identifies three primary channels through which pollution affects learning: acute cognitive interference affecting attention and processing speed; respiratory and physical health issues, such as asthma, that drive absenteeism; and disruption to school operations, where schools cancel outdoor activities or close entirely during heavy pollution events. In the United States, air pollution is regulated by the Environmental Protection Agency (EPA), which tightened the annual PM2.5 standard from 12 µg/m3 to 9 µg/m3 in February 2024. This revision was upheld by the D.C. Circuit in June 2026.

Data and descriptive statistics

The analysis combines nationwide student test scores with high-resolution daily PM2.5 concentrations and wildfire smoke data. Researchers categorised exposure based on the number of days in the year prior to an exam falling into three ranges: non-polluted (below 8 µg/m3), moderately polluted (8–12 µg/m3), and heavily polluted (at or above 12 µg/m3). The average student in the sample experiences 189.1 polluted days per year, with 102.7 days involving moderate pollution levels. Exposure is unequally distributed, with Black and Hispanic students, and those in socio-economically disadvantaged schools, experiencing more polluted days on average.

Empirical strategy

To identify the causal effect of pollution, the researchers employed a within-student design that compares the same student’s performance across multiple test attempts preceded by differing air quality levels. To isolate exogenous variation and address potential measurement errors, the study used wildfire smoke from distant fires (at least 10 kilometres from the school) as an instrumental variable for local PM2.5. This strategy purges bias from local economic conditions that might simultaneously influence air quality and educational achievement. The model includes student fixed effects to control for time-invariant factors like ability and household background.

Results

The study finds that pollution exposure meaningfully slows the rate of learning. A year of observed pollution exposure reduces annual learning by approximately 0.055 standard deviations, equivalent to 19 per cent of typical annual score growth. This magnitude is comparable to the learning loss associated with 5 to 10 additional days of student absence. The damage is driven almost entirely by moderate pollution days (8–12 µg/m3) and by exposure during the school year rather than the summer. This concentration during instructional time suggests that instructional disruption and cognitive interference are the primary drivers of learning loss.

Magnitudes

The harm from air pollution is not evenly distributed. The impact is approximately three times larger in socio-economically disadvantaged schools and among Black and Hispanic students compared to White and Asian students. This gap is driven by differential sensitivity rather than just differential exposure; the same amount of pollution causes more damage to students with fewer community or family-based resources. The research suggests that the total learning cost of pollution accounts for approximately 15 per cent of the achievement gap between Black/Hispanic and White/Asian students, and 13 per cent of the gap between high- and low-challenge schools.

Conclusion

The report concludes that ambient air pollution reduces academic performance and widens existing achievement gaps. The findings indicate that meaningful social costs accrue at pollution levels previously treated as safe by regulators. Reducing PM2.5 concentrations back toward historical levels could impose significant achievement costs, while further improvements in air quality could support student achievement. Future research is suggested to further distinguish between specific mechanisms, such as teacher absenteeism versus reduced student attention, to better inform mitigation strategies like improved HVAC systems and air filtration.