We propose using sign restrictions to identify regional labor demand shocks in a panel VAR of US federal states. Observed migration responds significantly, but less persistently than the residual-based migration measure constructed by Blanchard and Katz (1992).