Social Interactions in Pandemics: Fear, Altruism, and Reciprocity

In SIR models, infection rates are typically exogenous, whereas individuals adjust their behavior in reality. City-level data across the globe suggest that mobility falls in response to fear, proxied by Google searches. Incorporating experimentally validated measures of social preferences at the regional level, we find that stringency measures (conversely their lifting) matter less when individuals are more patient and altruistic, or exhibit less negative reciprocity. To account for these findings, we extend homogeneous and networked SIR models so as to endogenize agents' social-activity intensity. We derive the social planner's problem and draw implications on the optimality of targeted lockdown policies.