42nd Congress of the European Regional Science Association: "From Industry to Advanced Services - Perspectives of European Metropolitan Regions", August 27th - 31st, 2002, Dortmund, Germany
This paper analyses the time series properties of the unemployment rates of the 50 US States, as well as the global rate of the USA. Our results, based on the use of ADF-type tests, show that the inclusion of some breaks is vital in order to reduce the persistence on these rates. Thus, we can reject the unit root null hypothesis versus a double mean-shifted stationary alternative for 46 States and for the US total rate. We also find that the behavior of this latter rate is not congruent with that of the States, implying the presence of some aggregation problems which have not been commonly accounted in the literature.