Series-Specific Unit Root Tests with Panel Data.

A unit root testing procedure is presented that exploits the well-established power advantages of panel estimation while rectifying a deficiency in other panel unit root tests. This test (called SURADF) is based on seemingly unrelated regressions applied to Augmented Dickey-Fuller (ADF) tests for a unit root. In contrast to extant panel unit root tests, our test allows for determination of which members of the panel reject the null hypothesis of a unit root and which ones do not. The power of the test is investigated with Monte Carlo simulation and demonstrated with application to several panels of real exchange rates. We find that when the contemporaneous cross-correlations of the residuals are high, our procedure has substantially more power to reject a unit root than the single equation Dickey-Fuller test. Copyright 2002 by Blackwell Publishing Ltd