Search
Browse By Day
Browse By Time
Browse By Person
Browse By Policy Area
Browse By Session Type
Browse By Keyword
Browse Artificial Intelligence Presentations
Program Calendar
Sign In
Search Tips
Air pollution is one of the leading global mortality risk factors. A large literature shows that both acute and chronic exposure to air pollution harm health, productivity, and cognitive performance. Yet an important and understudied question remains: do individuals adapt to repeated exposure? This question matters for policy because standard welfare calculations typically assume that the marginal damage of pollution is constant and does not depend on prior exposure history. This paper studies whether repeated exposure reduces the short-run cognitive costs of air pollution. Specifically, we ask whether students from chronically polluted environments are less sensitive to acute pollution shocks during high-stakes exams than students from cleaner environments, and whether this sensitivity changes over time as individuals move to and remain in a highly polluted city.
To answer these questions, we use a rich dataset of more than 75,000 student-examination records from a selective university in Beijing spanning 2008 to 2016. We match these records to hourly PM2.5 readings from the U.S. Embassy monitor, located near campus, to measure pollution exposure at the time of exams. To capture students’ prior pollution experience, we use satellite-based PM2.5 estimates from the China High Air Pollutants (CHAP) dataset to classify the air quality of their home counties over the three years of high school before college entry.
Our empirical analysis begins with an OLS model that estimates the effect of exam-time PM2.5 on test performance. To identify adaptation, we then implement a triple-differences design that allows pollution sensitivity to vary with both students’ hometown pollution background and their duration of residence in Beijing. In addition, we use a semiparametric procedure to flexibly estimate the dose-response relationship between pollution and academic performance.
We find that a one-standard-deviation increase in exam-time PM2.5 reduces test scores by 2.1% of a standard deviation on average. However, this effect is substantially smaller for students from more polluted hometowns (0.7% SD) than for students from cleaner-air areas (2.5% SD). Moreover, among students from cleaner hometowns, sensitivity declines with time spent in Beijing: the performance penalty falls from 6.4% of a standard deviation in the freshman year to statistically insignificant levels by the junior and senior years. This study focuses on a period with low public awareness of PM2.5, which suggests that these patterns are more consistent with physiological adaptation than with changes in avoidance behavior.
Overall, the findings provide large-scale empirical evidence of dynamic adaptation to air pollution in a real-world setting. They imply that conventional benefit-cost analyses may overstate pollution damages for chronically exposed populations. At the same time, adaptation to acute shocks should not be interpreted as the absence of harm, since cumulative health and cognitive costs may still persist.