Working Papers

Air Pollution and Economic Activity: Evidence from Foot Traffic Patterns (Job Market Paper)
I investigate how air pollution affects economic activity. Using over 600 million phone-location-based foot traffic data points from SafeGraph, I conduct a large-scale analysis to examine the causal effect of air pollution on activity patterns across the US. Using changes in local wind direction as an instrumental variable (IV) for air pollution, I characterize the dynamic response to pollution exposure. I find that PM2.5 reduces economic activity in both the short and medium run, with effects persisting for up to two weeks before gradually recovering. The reductions are widespread across different economic sectors, with recreational activities experiencing the largest decline. The effect is more pronounced in higher-income counties and areas with larger shares of children, suggesting greater awareness among wealthier or more vulnerable populations.

Go with the Wind: Polluters’ Strategic Response to Wind Directions
I investigate the strategic behavior of polluters in the US. Under the Clean Air Act, air quality monitor data determine whether an area meets environmental standards. Because wind carries pollution, monitors detect more pollution from upwind sources and less from downwind sources. This disparity incentivizes polluters to emit less when they are upwind of a monitor, and more when they are downwind. I identify such strategic behavior among US coal-fired power plants, finding that a one-standard-deviation increase in favorable wind direction—when wind blows pollutants away from monitors—leads to a 0.8% (172 lbs) and 0.4% (54 lbs) increase in sulfur dioxide ($SO_2$) and nitrogen dioxide ($NO_x$) emissions, respectively. At the same time, fuel input remains constant while emission rates rise, suggesting that power plants temporarily turn off pollution control equipment. Additionally, the increase is more pronounced when power plants are in a non-attainment county, located in the same state as the monitor, or surrounded by fewer nearby polluters. These findings suggest that polluters facing stricter regulatory pressure are more likely to respond strategically when conditions are favorable.

Works in Progress

Natural Disasters and Occupational Mobility: Evidence from the 1927 Mississippi Flood (joint with Xiaohan Wang)
We investigate how a major natural disaster affect intergenerational occupational mobility in the agricultural sector. Linking individuals across the 1920, 1930, and 1940 Censuses, we show that children residing in counties affected by the 1927 Mississippi Flood were significantly less likely to become farmers as adults. This effect is mainly driven by children from non-farming families, who were 9% less likely to work in agriculture and 10% more likely to transition into the service sector. In the longer run, these individuals achieved socioeconomic status scores about 3% higher than comparable children in non-flooded counties. Our findings suggest that natural disasters can accelerate structural change by shifting new labor market entrants out of agriculture and into higher-status occupations.

Social Networks and Household Protective Investments: Evidence from Wildfire Smoke (joint with Sanjukta Mitra)

Deforestation and Air Pollution in the Indonesian Rainforest (joint with Francesco Cenerini)

Which robust method better? - Simulation Studies (joing with EunYi Chung and Yoko Yamasaki)