Working Papers

Asymmetric Migration Response and Its Implications in Big Shocks (July, 2025) New Draft!

Abstract: Standard migration models with a single migration elasticity imply that post-shock migration patterns mirror pre-shock behavior, regardless of the shock’s magnitude—a prediction inconsistent with observed data. To address this limitation, I develop a nested logit migration framework in which nests are defined by both geographic and location's characteristics. This model allows migration elasticities to vary across locations—a feature that enables post-shock migration patterns to depend on the size of the shock. Using U.S. commuting-zone-level migration flow data, I document substantial heterogeneity in migration elasticities. I validate the model using the case of Hurricane Katrina in 2005, showing through simulation that the model’s structure—featuring heterogeneous elasticities—can account for much of the persistently disproportionate migration responses observed over the decade following the disaster. Finally, the framework reveals that localized shocks generate more uneven spatial welfare losses compared to a single-elasticity model, with high-migration-elasticity destinations disproportionately burdened due to overflow migrants from the affected origin.