We develop and implement a heterogeneous-agents network-based empirical model to analyze alternative policies during a pandemic outbreak. We combine several data sources, including information on individuals mobility and encounters across metropolitan areas, information on health records for millions of individuals, and information on the possibility to be productive while working from home. This rich combination of data sources allows us to build a framework in which the severity of a disease outbreak varies across locations and industries, and across individuals who differ by age, occupation, and preexisting health conditions.
We use this framework to analyze the impact of different social distancing policies in the context of the COVID-19 outbreaks across US metropolitan areas. Our results highlight how outcomes vary across areas in relation to the underlying heterogeneity in population density, social network structures, population health, and employment characteristics. We find that policies by which individuals who can work from home continue to do so, or in which schools and ﬁrms alternate schedules across different groups of students and employees, can be effective in limiting the health and healthcare costs of the pandemic outbreak while also reducing employment losses.