Abstract
This paper asks whether Americans were jobless in 2014 because of where they were living in 2007. In the cross section, employment rates diverged across U.S. local areas 2007-2009 and—in contrast to history—have barely converged. This “great divergence” could reflect spatial differences in human capital, rather than causal location effects. I therefore use administrative data to compare two million workers with very similar pre-2007 human capital: those who in 2006 earned the same amount from the same retail firm, at establishments located in different local areas. I find that conditional on 2006 firm-x-wages fixed effects, living in 2007 in a below-median 2007-2009-fluctuation area caused those workers to have a 1.3%-lower 2014 employment rate. Hence, U.S. local labor markets are limitedly integrated: location has caused long-term joblessness and exacerbated within-skill inequality. The enduring impact is not explained by enduringly high unemployment, more layoffs, more disability enrollment, or reduced migration. Instead, the employment outcomes of cross-area movers are consistent with severe-fluctuation areas continuing to depress residents’ labor force participation. Impacts are correlated with housing busts but not manufacturing busts, possibly reconciling current experience with history. If recent trends continue, employment rates are estimated to remain diverged into the 2020s—adding up to over a relative lost decade for half the country. Employment models should allow market-wide shocks to cause persistent labor force exit, leaving employment depressed even after unemployment recovers.