The Response of Online Consumer Spending to Air Pollution : Evidence from a Quasi-experiment in China

This paper estimates how online consumer spending responds to changes in air pollution. We address the endogeneity of air pollution in affecting household consumption by exploiting plausibly exogenous variations in air quality caused by China's Huai River heating policy. We use a comprehensive and proprietary panel dataset obtained from China's largest e-commerce company consisting of complete transaction information for more than 4,369 active online merchants between January 2017 and September 2019. At the Seller-BuyerCity-Month level, our difference-in-differences estimations show that air pollution caused by the heating season increases consumption from sellers of health-related products and necessities by 7.3% and 8.0%, respectively, whereas consumption from sellers of non-essential products drops by 3.5%. Moreover, the two-stage least square regression discontinuity design at the BuyerCity-Year level yields substantially larger causal effects. Using online keyword searches related to "air pollution" and "mood", we show that avoidance behavior and negative mood jointly provide potential mechanisms