The Pollution-Productivity Curve: Non-Linear Effects and Adaptation in High-Pollution Environments

Over 2.8 billion people are chronic ally exposed to hazardous levels offine particulate matter air pollution. This paper provides novel evidence that workers in such settings partially adapt to chronic exposure but that adaptation does not offset cumulative harm. We estimate the effect of PM2.5 on labor productivity using individual-level performance data from 14 years of professional cricket in India (2008–2022), paired with a machine learning data product providing daily PM 2.5 estimates at 10 km resolution. Leveraging variation in day-to-day exposure generated by league scheduling rules, we find that a 10 μgm−3 increase in same-day PM 2.5 (half a standard deviation) reduces bowler performance by about 1 percent relative to batters, consistent with bowlers’ heightened exposure via higher respiration rates. Effects are non-linear, with the largest marginal damages above approximately 50 μgm−3—levels common in developing countries but uncommon in causal studies. Using variation in chronic exposure from player assignment to teams according to salary cap rules, we find that acute shocks harm those with the highest past exposure approximately 40 percent less than those with median exposure histories, indicating adaptation over both 30-day and career-spanning horizons. Nevertheless, chronic exposure degrades performance by more than adaptation offsets, except under extremely rare pollution conditions. These findings underscore the importance of regulating these cond moment of the pollution distribution: non-linearity implies that marginal damages are largest when pollution is in the uppertail, while partial adaptation implies that spikes above mean levels amplify marginal damages.