Effects of Weather on COVID-19 and an Instrumental Variable Strategy to Address Endogeneity

Author: 
Alden Ming Yang Tan
Adviser(s): 
Prof. Yuichi Kitamura
Abstract: 

Amidst the COVID-19 pandemic, there has been an intense and wide-ranging effort by academics, scientists and policymakers all around the world to understand what affects COVID-19 transmission. A plethora of factors are being examined, ranging from social distancing measures and mask mandates, to demographic and socioeconomic variables. One particular variable of interest is the weather. It has often been suggested that weather variables such as temperature would affect both the viability of the virus itself, as well as human behavior, which would both have an impact on COVID-19 transmission. I investigate the causal effect of weather on COVID-19 cases and deaths, with particular focus on the United States, using county-level, daily data from January to December 2020. I identify a potential endogeneity problem with regressions in the existing literature and propose a novel instrumental variable (IV) strategy to address it. Specifically, when regressions include stay-home or lockdown orders as a control variable, this variable may be endogenous as policymakers may anticipate future COVID-19 shocks when deciding whether or not to implement stay-home orders. The proposed IV strategy shows promising results in correcting for the endogeneity problem in the regressions of both COVID-19 cases and deaths. Using the IV strategy, I find that both temperature and dewpoint have a statistically significant and negative effect on COVID-19 cases and deaths. My results suggest that a 1 degree Fahrenheit rise in temperature decreases COVID-19 cases by 1.84%, and decreases deaths by 0.239%. In addition, a 1 degree Fahrenheit rise in dewpoint decreases cases by 1.37% and deaths by 0.281%. My results also provide evidence that both cases and deaths decrease a few weeks after stay-home orders are implemented. These results are robust to changes in the model specification, and have instructive implications for public health policymaking in the US.

Term: 
Spring 2022