Abstract
Past researchers, examining the duration of joblessness, have failed to include a variable to account for the unemployed's migration decision. By employing a simultaneous structure with a censored dependent variable and an endogenous binary migration variable, the present study demonstrates that single stage models, which do not control for migration, produce selectivity bias in the estimated duration equation. The empirical model used in this study provides estimated coefficients that differ significantly from those of single stage models. For example, most past studies conclude that nonwhites suffer longer periods of unemployment than whites. The findings from the present study indicate that, after controlling for geographic mobility, there is no statistical difference in unemployment duration between whites and nonwhites. -Authors