Psychological predictors of socioeconomic resilience amidst the COVID-19 pandemic : evidence from machine learning

Abstract

What predicts cross-country differences in the recovery of socioeconomic activity from the COVID-19 pandemic? To answer this question, we examined how quickly countries’ socioeconomic activity bounced back to normalcy from disruptions caused by the COVID-19 pandemic based on residents’ attitudes, values, and beliefs as measured in the World Values Survey. We trained nine preregistered machine learning models to predict the rate at which various socioeconomic metrics (e.g., public transportation occupancy, cinema attendance) recovered from their COVID-19 lows based on the World Values Survey. All models had high predictive accuracy when presented with out-of-sample data (rs ≥ .83). Feature importance analyses identified five psychological predictors that most strongly predicted socioeconomic recovery from COVID-19: religiosity, liberal social attitudes, the value of independence, obedience to authority, and the Protestant work ethic. Although past research has established the role of religiosity, liberalism, and independence in predicting resilience, it has not yet considered obedience to authority or the Protestant work ethic. Thus, the current research suggests new directions for future work on resilience that may not be apparent from either a deductive or an inductive approach.