Working Paper, Department of Economics, Queen Mary, University of London 495
An attractive feature of panel unit root tests is the ability to exploit coefficient homogeneity under the null hypothesis of a unit root for all series in order to obtain a more powerful test of the unit root hypothesis. However, under the alternative hypothesis of heterogeneous panel unit root tests of at least one series being stationary, the researcher is left with little idea of how to proceed. In other words if we reject the unit root hypothesis we do not know which series caused the rejection. We propose a method that enables the distinction of a set of series into a group of stationary and a group of nonstationary series. We discuss its theoretical properties and investigate its small sample performance in a Monte Carlo study.