From Danny Yagan:
Over the Great Recession, the employment rate in some U.S. cities declined by more than twice the aggregate decline. To what extent did the ability to migrate insure workers against these idiosyncratic local shocks? I answer this question using geo-coded administrative panel data on the universe of U.S. males from years 2000-2011. I find that despite migration flows that were in principle large enough to provide full insurance, migration has provided only 7% insurance: the 2006 residents of the average local area have borne 93% of the area’s idiosyncratic Great Recession labor demand shock. I find similarly small degrees of insurance across specifications, demographic groups, and labor market outcomes. Insurance was three times greater over the 2001 recession, driven entirely by greater insurance for above-average earners. The relative failure of migratory insurance over the Great Recession is attributable to unusually small employment gains for those migrating from heavily-shocked areas to lightly-shocked areas, rather than to a decline in migration rates. The results imply large spatial adjustment frictions in the U.S. labor market and indicate that past location may be a powerful tag for directing social insurance.