Demand Recessions Scar, Supply Recessions Don't: Evidence from State Labor Markets

Abstract

Not all recessions are created equal. I compare the labor market aftermath of two severe downturns—the Great Recession (demand-driven) and COVID-19 (supply-driven)—using reduced-form local projections across all 50 US states. Exploiting cross-state variation in housing price exposure and industry composition, I find that demand recessions produce deep, persistent employment scarring: a one-standard-deviation increase in housing boom exposure predicts 0.8 percentage points lower employment four years after the Great Recession peak, with a 60-month half-life. COVID-exposed states recovered fully within 18 months. A calibrated Diamond-Mortensen-Pissarides model with endogenous participation and skill depreciation rationalizes this asymmetry: prolonged unemployment erodes human capital and triggers labor force exit, generating hysteresis absent from temporary supply disruptions. Skill depreciation accounts for 51% of demand-shock welfare losses.