Air Pollution and Learning

• In Event: The effects of environmental risks on children

Abstract

We estimate the causal impact of air pollution on student learning using longitudinal PSAT data from over 16 million U.S. high school students who took the exam multiple times between 2007 and 2016. Our within-student design, combined with an instrumental variables strategy exploiting variation in wildfire smoke from distant fires, isolates plausibly exogenous changes in local PM2.5 concentrations. IV estimates are substantially larger than OLS, consistent with attenuation bias from measurement error. Exposure to moderately polluted days (8-12 μg/m3) during the school year significantly reduces academic performance; scaling estimated effects by observed exposure implies an annual learning loss of approximately 0.055 standard deviations, or roughly 19% of typical yearly score growth. Effects are concentrated during the academic year rather than summer months, pointing toward instructional disruption and short-run cognitive interference rather than longer-run health channels. We document meaningful heterogeneity: students attending high-challenge schools and those with lower baseline achievement experience substantially larger adverse effects. Our results demonstrate harmful impacts at PM2.5 concentrations as low as 8 μg/m3, below current and recently proposed EPA standards, with important implications for environmental regulation and educational equity. + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +

Author

• Joshua Goodman, Boston University
  Presenting Author