Unit Root Tests on the Level and Rate of Unemployment:
 Australia, 1960-1997

Abstract

The aim of this paper is to analyze the rate and level of Australian unemployment, and also their logarithms, in the period of 1960 to 1997, with special regard to the unit-root versus stationarity hypotheses. Theoretically, the level of unemployment, the rate of unemployment and the logarithm of the unemployment rate cannot be generated by random walk processes. In practice, however, the actual time series at hand might behave similarly to some final segments of unit-root processes, making it impossible to reject the unit root hypothesis at conventional levels of significance. As our analysis shows, the results of unit-root and stationarity tests depend on whether we consider unemployment, the unemployment rate or their logarithms. Moreover, they are very sensitive to the existence of structural changes.