³ÉÈË´óƬ

I examine the impact of prenatal suspended particulate pollution on educational outcomes, using ambient total suspended particulates (TSPs) as a measure of particulate exposure and standardized test scores of exposed individuals as a measure of educational achievement. I focus on individuals born between 1979 and 1985 to exploit the shock of the industrial recession of the early 1980s. This variation helps separate the causal e ects of pollution reduction from general time trends. Considering the 7-year time period as a whole yields statistically insigni cant results, but focusing on the 3-year period around the recession (1981-1983) yields negative and statistically significant results, suggesting that the relationship is subtle enough to require large-scale changes to be detectable. My ndings suggest a standard deviation decrease in the mean pollution level in a student's year of birth is associated with 1.87% of a standard deviation increase in test scores in high school. I also employ an instrumental variables strategy, using changes in relative manufacturing employment driven by the recession as an instrument for TSP levels. Instrumental variables results are approximately 3.7 times the size of the OLS results, suggesting the potential presence of measurement error in ambient pollution. Results are robust to the inclusion of school xed e ects, year of birth and year of test xed e ects, and various demographic and economic covariates. I also investigate the potential bias sources of migration and selection into motherhood, and show these are unlikely to explain my results.