This article documents social inequality in exposure to air pollution in French urban areas over 2012–2021, using daily concentrations of PM2.5, PM10, NO2 and O3 at the neighbourhood level. For each neighbourhood and year I count the days on which at least one pollutant exceeds its WHO guideline, and I record which pollutants exceed together on each of those days. Comparing neighbourhoods of the same urban area and year, the richest 30% experience 9.7 fewer polluted days per year than the poorest 30%, a gap that steepens with city size. This gap nets two opposing forces. Every combination of pollutants that includes NO2 is more frequent in poorer neighbourhoods, by 41% of its level, and every combination without NO2 is more frequent in richer ones, by 9%. Richer and poorer neighbourhoods thus experience comparable numbers of fine-particle days, but in poorer neighbourhoods those days come with NO2. This is an inequality in what people breathe, and it is invisible at coarser spatial scales, where averaging pollution removes almost all of the gradient. Over the decade exposure fell by about 40% at every point of the income distribution, which left the gap in days unchanged and the relative gap wider.

Who bears the educational cost of air pollution? Using administrative data on five million French students and wind-driven variation in PM2.5, we show that exam day pollution shocks lower exam scores by 4–7% of a standard deviation among disadvantaged students—equivalent to 10–25% of the socio-economic achievement gap—while leaving advantaged students unaffected. We trace this unequal vulnerability to socially stratified differences in cognitive resilience, defined as the capacity to perform under high-stakes conditions, for which we develop a novel individual-level measure. Pollution shocks therefore widen pre-existing educational inequalities. Yet vulnerability is not immutable: high value-added schools mitigate the effects of pollution among low-resilience students

This paper estimates the causal effect of chronic exposure to industrial pollution during middle school on performance in the end-of-middle-school national examination (brevet) and on high-school track choice. We link administrative records for nearly 1.8 million French students from three cohorts (2010-2012) to the universe of industrial facilities reporting emissions of thirteen major pollutants and construct each school cohort's cumulative exposure over the four years of middle school. We identify its causal effect using a wind-based instrument that exploits within-school, cohort-to-cohort variation in realized wind direction. A one standard deviation (SD) increase in realized downwind exposure lowers brevet scores by 0.069 SD, reduces the brevet pass rate by 1.8 percentage points (pp), and decreases entry into the academic high-school track by 2.5 pp. The effects are driven primarily by emissions from metallurgy and energy plants, especially particulates (PM10), reactive gases, and neurotoxic metals, and are highly concentrated geographically and among disadvantaged students