Abstract
—This paper employs recently developed techniques for testing hypotheses in cointegrated panels to test the strong version of purchasing power parity for a panel of post Bretton Woods data. We compare results using fully modié ed and dynamic OLS approaches, and strongly reject the hypothesis. We also introduce a new between-dimension dynamic OLS estimator and é nd that the between-dimension FMOLS and DOLS esti- mates of the long-run deviation from purchasing power parity are larger than the corresponding within-dimension estimates. Finally, we attempt to reconcile these rejections with the mixed é ndings that have been reported in panel unit root studies.