• Last updated: Wed, Nov 8, 2023Status: Ongoing
• Adrian Adermon, Lisa Laun, Costanza Naguib, Martin Olsson, Jan Sauermann, and Anna Sjögren

Heterogeneous impacts of COVID-19 on incomes

An important question for policymakers is which groups experienced the largest income losses in the pandemic, and how well they were protected by the welfare system. Understanding this will allow potential holes in the safety net to be patched before another crisis. It is also useful to know which policies protected different groups, to better understand the distributional effects of changing these policies. In this project, we study the heterogeneous impacts of the COVID-19 pandemic on individual incomes for the full Swedish population. We apply state-of-the-art causal machine learning methods (specifically, the generalized random forest), which allow us to identify potentially narrow groups which were hit especially hard by the pandemic. These groups will be defined by complex interactions among characteristics such as age, gender, nationality, level of education, sector of employment, family structure. The approach is data-driven, which prevents ad hoc methodological decisions that could result in detection of heterogeneity that is not really present, and helps avoid selective reporting of findings. Since we have access to a wide range of income sources, including labor incomes; transfers from pre-pandemic benefit systems; and transfers from specific pandemic measures, we are able to study heterogeneity in pandemic impacts separately across these different income measures, as well as for their sum.