Regional employment during recessions and recoveries in Japan: A data-driven approach

Abstract

This study investigates the impact of economic shocks on regional employment in Japan using causal forest algorithms and compares the data-driven findings with estimates derived from standard ordinary least squares (OLS) regression. Our results reveal no significant difference in the magnitude of impact between small and large regions. Instead, regional differences in impact stem from variations in pre-crisis employment growth and the concentration of local employment. Specifically, regions with a high proportion of manufacturing jobs demonstrate resilience to recession-related job losses but recover slowly, while regions with pre-crisis employment growth exceeding 5% suffer steep declines during recessions but exhibit strong recoveries.