Emotional states and the Covid-19 pandemic: evolution, national differences, and their relationship with emergency policies

We propose to study the effect of Covid-19 on the emotional states of people in OECD and other countries by applying content analysis to internet searches recorded via Google Trends since 31 December 2019. Our aim is to understand how emotional states evolved as people progressively came to know the first confirmed cases of Covid-19 and deaths, and then experienced governments’ introduction of social distancing, lockdowns and programs to support firms and employment. In particular, we will focus on emotions concerning employment and economic considerations, but also particular attention to the rise of emotions related to mental health and wellbeing. We will analyse the relationship between timing of government responses to the pandemic in various countries and individuals’ Google searches related to various emotional states. We will employ a variant of the normal content analysis as used by Savage and Torgler (2013) to compare frequency and count analysis of Google search terms. The data are classified as ‘open microphone’, meaning that every search term is captured without knowing the number people involved or the intended focus of the search. This means that it is difficult to compare changes in word counts over time as it is unclear how many people are using the service. However, only comparing the frequency of certain terms appearing does not give a clear indication of the volume of searches. As such both frequency and count analysis are used since neither one alone provides a complete by itself. Furthermore, it is important to understand the distribution of the ‘normal’ daily search traffic, so we are able compare any variations that are triggered by a specific event. Other explanatory variables will include country-specific counts of Covid-19 cases and death rates at a given date, to measure the stock of information about the pandemic available to the population and variables measuring country proximity to the known hotspots. Additional country characteristics, such as population density, demographic structure, living conditions, and cultural values will be used as control variables in the analysis.