Coauthored with Kirill Borusyak and Rafael Dix-Carneiro
In this paper, we investigate regressions commonly used to assess the effects of local economic shocks on local populations. Using a simple model with frictional mobility, we find that these regressions are severely misspecified, particularly when migrant-connected regions face similar shocks. We propose a model-consistent empirical approach and present simulation evidence using real-world geography and migration patterns to show that the standard approach is likely to yield substantially biased results in practice.
Image source: Flickr