Weather, pollution and Covid-19 spread : a time series and Wavelet reassessment

Faced with the global pandemic of Covid-19, we need to better understand the links between meteorological factors, air quality and the virus. In the vein of a recent empirical literature, we reassess the impact of weather factors like temperatures, humidity and air quality indicators on Covid-19 daily cases in China both for Wuhan and Beijing. Using a consistent number of observations (104), we compute, for the first time, correlations but also Granger causality and above all, a spectral analysis using Wavelet methods. Our results go further previous studies and reveal the complexity of the studied relationships when both time and frequency domains are taken into account. Wavelet analysis enables us to go further usual correlations analysis. Though negative humidity impact on Covid-19 cases was expected to be relatively clear regarding previous literature based on correlations, we do not find evidence of such a result. The controversial effect of warmer temperatures on the Covid-19, often difficult to identify or sometimes identified as surprisingly positive, can negatively emerge via Wavelet analysis for some periods only. This result is however clear-cut for the Hubei Province but for the Beijing one. Finally, our results reveal a bi-directional causality between air quality and the number of infected people. Short-run causality from Covid-19 to air quality (better induced air quality) via lockdown policies disappear in a medium-run and turns to become a significant causal link from induced air quality improvement to Covid-19 daily cases (reduction of infected people).