Reproducibility package for Air Pollution And Student Learning In Pakistan

Ambient air pollution is a major health risk, but its implications for human capital formation are not well understood, particularly in low- and middle-income countries where exposure is high. This paper estimates the causal effect of sustained exposure to fine particulate matter (PM2.5) on primary-school learning in rural Punjab, Pakistan. We combine longitudinal student achievement data from the Learning and Educational Achievements in Punjab Schools (LEAPS) project with high-resolution satellite-based PM2.5 estimates. To address endogeneity of pollution to local economic activity, we exploit plausibly exogenous variation generated by open burning of agricultural residue interacted with wind direction, constructing school-year measures of upwind fire exposure as instruments for local PM2.5. Our instrumental variables estimates show that higher school-year PM2.5 significantly slows learning, with larger effects in mathematics than in language subjects. Results are consistent with impacts operating during instructional months: pollution exposure during the school year predicts lower achievement, while exposure during vacation months has little explanatory power. We find statistically imprecise heterogeneity by age and gender, opposite to patterns found in comparable settings. We organize the analysis around a simple framework in which learning depends on attendance, cognition, effort, and instructional quality, each a function of exposure to poor air quality. Pollution increases student absences, but this attendance channel explains only a small portion of the total effect, implying that most of the learning loss is non-attendance-related and likely operates through the framework’s other channels—direct effects on cognition, reduced effort, and degraded instructional quality.