COVID-19, City Lockdown, and Air Pollution: Evidence from China

Guojun He; Yuhang Pan; Takanao Tanaka

We employ two sets of difference-in-differences (DiD) models to quantify the impact of a city's lockdown on air pollution in different groups of cities. The DiD model is a wellestablished econometric technique and has been widely applied in understanding the impacts of government policies (Angrist and Pischke, 2008) . In our DiD specification, the city-level air pollution is as a function of the city-specific lock-down policy, city-specific fixed effects that capture city-level time-invariant determinants of air pollution (e.g., geographical conditions, baseline income), month by week fixed effects that capture shocks common to all cities in a given week (e.g., nationwide holiday policy, macroeconomic conditions) and weather conditions (see Appendix Methods). Adopting this method, we can estimate the difference in air pollution levels between the treatment group (locked-down cities) and the control group (non-locked-down cities) before and after city lockdowns in 2020. In addition, because cities without formal lockdown policies might also have been affected by the disease preventive measures (e.g., all cities extended the Spring Festival holiday, required social distancing, and urged people to stay at home), we also apply the DiD model to compare air pollution levels in the control (no-lockdown) cities before and after China's Spring Festival relative to the previous year. 3 Combining these two sets of DiD results, we can evaluate the total effects of city lockdown on air quality.