This paper examines the nature of asymmetry of U.S. state unemployment rates using the time reversibility test developed by Ramsey and Rothman (1996). These authors and others have found asymmetry in aggregate unemployment rates in this study we examine whether or not these results extend to state level unemployment series. Alaska, District of Columbia, Hawaii, Louisiana, Missouri, Montana, and Puerto Rico, exhibit changes in unemployment rates that are symmetric. California, Georgia, Kansas, and North Carolina, show evidence of asymmetry of the change in unemployment rates due to non-linearity in the model. Unemployment rate asymmetry documented in other states is attributable to non-Gaussian errors. Asymmetric patterns documented in most states are consistent with the fast-up and slow-down dynamics observed in aggregate unemployment data.
The Network Composition of Aggregate Unemployment
