Rising temperature levels during spring and summer are often argued to enable lifting of strict containment measures even in the absence of herd immunity. Despite broad scholarly interest in the relationship between weather and coronavirus spread, previous studies come to very mixed results. To contribute to this puzzle, the paper examines the impact of weather on the COVID-19 pandemic using a unique granular dataset of over 1.2 million daily observations covering over 3700 counties in nine countries for all seasons of 2020. Our results show that temperature and wind speed have a robust negative effect on virus spread after controlling for a range of potential confounding factors. These effects, however, are substantially larger during mealtimes, as well as in periods of high mobility and low containment, suggesting an important role for social behaviour.

The effectiveness of containment measures has been shown to depend on both epidemiological and sociological mechanisms, most notably compliance with national lockdown rules. Yet, there has been growing discontent with social distancing rules during national lockdowns across several countries, particularly among certain demographic and socio-economic groups. Using a highly granular dataset on compliance of over 105,000 individuals between March and May 2020 in the United Kingdom (UK), we find that compliance with lockdown policies was initially high in the overall population during the earlier phase of the pandemic, but that compliance fell substantially over time, especially among specific segments of society. Warmer temperatures increased the non-compliance of individuals who are male, divorced, part-time employed, and/or parent of more than two children. Thus, while epidemiologically the virus spread was naturally more limited during the warmer period of 2020, sociologically the higher temperature led to lower individual-level compliance with public health measures. As long as new strains emerge, governments may therefore be required to complement vaccination campaigns with targeted and time limited restrictions. Since non-complying individuals at the beginning of the pandemic share certain characteristics with vaccination sceptics, understanding their compliance behaviour will remain essential for future policymaking.

Early non-pharmaceutical interventions (NPI) significantly reduced the death toll of the COVID-19 pandemic. Yet, there are vast differences in how quickly governments implemented NPIs. In this paper, we analyze the role of public attention, measured as the share of daily Google searches in a country related to COVID-19, in the timing of the NPI responses. We first show that public attention depends strongly on whether there are cases in own country. We then show that countries with high levels of public attention are more likely to implement non-pharmaceutical interventions, even after controlling for the number of cases and deaths. Finally, we show that the extent to which a government responds to public attention is highly dependent on the country's institutional quality. The positive effect of public attention on policy implementation is driven entirely by countries with good institutions.