Persistence in U.S. State Unemployment Rates

Abstract

Romero‐Ávila and Usabiaga (2007) find that many U.S. state unemployment rates are stationary, a result at odds with the traditional view that unemployment rates are path‐dependent and subject to shocks that have permanent effects. They base their results on multivariate unit root tests that provide for two breaks in mean. This note extends the analysis to directly examine whether the series were fractionally integrated. When no allowance is made for breaking means, the results suggest evidence in favor of hysteresis, an outcome that generally applies when one break in mean is considered. Allowing for two breaks demonstrates that the evidence in favor of the natural rate and the hysteresis hypotheses is temporally sensitive.