Abstract
We use daily happiness scores (Gross National Happiness (GNH)) to illustrate how happiness changed throughout 2020 in ten countries across Europe and the Southern hemisphere. More frequently and regularly available than survey data, the GNH reveals how happiness sharply declined at the onset of the pandemic and lockdown, quickly recovered, and then trended downward throughout much of the year in Europe. GNH is derived by applying sentiment and emotion analysis-based on Natural Language Processing using machine learning algorithms-to Twitter posts (tweets). Using a similar approach, we generate another 11 variables: eight emotions and three new context-specific variables, in particular: trust in national institutions, sadness in relation to loneliness, and fear concerning the economy. Given the novelty of the dataset, we use multiple methods to assess validity. We also assess the correlates of GNH. The results indicate that GNH is negatively correlated with new COVID-19 cases, containment policies, and disgust and positively correlated with staying at home, surprise, and generalised trust. Altogether the analyses indicate tools based on Big Data, such as the GNH, offer relevant data that often fill information gaps and can valuably supplement traditional tools. In this case, the GNH results suggest that both the severity of the pandemic and containment policies negatively correlated with happiness.