BACKGROUND
Since the spread of the SARS-CoV-2 virus and the lockdown measures went hand-in-hand, it is difficult to distinguish how public opinion reacted to the lockdown measures from the reactions to COVID-19.
OBJECTIVE
We analyze the causal effect of COVID-19 lockdown policies on sentiment and uncertainty using the Italian lockdown in February 2020 as a quasi-experiment. Communities inside and just outside the lockdown area were equally confronted with COVID-19 at the time of the implementation of the policy, offering a form of random allocation of the lockdown. The two areas had also balanced socioeconomic and demographic characteristics before the lockdown, indicating that the definition of the boundaries of the area under strict lockdown approximates a randomized experiment. This allows to identify the causal impact of lockdowns on public emotions, disentangling the changed due to the policy itself, from the changes induced by the spread of the novel virus.
METHODS
We employ Twitter data, natural language models (N = 24,261), and a difference-in-differences approach to compare sentiment changes within (n=1,567) and outside (n=22,694) the lockdown areas before and after the beginning of the lockdown. Tweets are classified into four categories—economics, health, politics, and lockdown policy—to analyze the corresponding emotional responses.
RESULTS
We find that the lockdown had no significant effect on economic uncertainty (b=0.005, SE=0.007, t(125)=0.70, P =.48) or economic negative sentiment (b=-0.011, SE=0.0089, t(125)=-1.32, P =.19), but increased uncertainty about health (b=0.036, SE=0.0065, t(125)=5.55, P<.001) and the lockdown policy (b=0.026, SE=0.006, t(125)=4.47, P<.001) and negative sentiment towards politics (b=0.025, SE=0.011, t(125)=2.33, P =.02), suggesting that lockdowns have wide externalities beyond health.
CONCLUSIONS
Our results emphasize the need for authorities to use these findings to improve future policies and communication efforts to mitigate uncertainty and social panic.