Abstract
This paper uses a recently suggested test for unit roots in panels of time series data (Maddala and Wu, Oxford Bulletin of Economics and Statistics, 61, 631-52, 1999) to consider the Purchasing Power Parity hypothesis. The major innovation of this test is that it allows both the testing of unit root null, using the ADF test, and the stationarity null, using the KPSS test. It is found that the results are inconsistent, suggesting that either alternative hypotheses to a unit root may need to be considered or that panel based testing in this particular context may be of limited value.