Unemployment Hysteresis in Australian States and Territories: Evidence from Panel Data Unit Root Tests

This article tests for hysteresis by applying panel data unit root tests to quarterly unemployment rates for Australian states and territories between 1982:2 and 2002:1. Panel tests proposed by Levin and Lin (1992) using ordinary least squares and O'Connell (1998) using feasible generalised least squares (which assume that under the alternative hypothesis of stationarity, all labour markets revert to the natural rate at the same speed) provide evidence in support of the natural rate hypothesis. However, the panel test proposed by Im, Pesaran and Shin (1997), which does not assume that all cross-sectional units converge towards the equilibrium value at the same speed under the alternative and is therefore less restrictive than the other two panel tests, finds evidence of hysteresis. Given the advantages of the Im et al. (1997) test over the other two panel tests the results can be interpreted as being consistent with the existence of hysteresis in unemployment Copyright 2003 The University of Melbourne, Melbourne Institute of Applied Economic and Social Research.