Abstract
The empirical validity of long-run purchasing power parity is investigated using multivariate cointegration techniques. Both bilateral and multilateral PPP is examined. The data set is monthly and covers almost 22 years (January 1970-August 1991) for four countries - Germany, Japan, the U.S., and Great Britain. While three cointegrating relations are detected among the set of nominal exchange rates and domestic price levels (or equivalently, there exists a reduced number of common stochastic trends) none of these satisfy the linear constraints implied by PPP. We conclude that neither bilateral nor multilateral PPP can be supported by the behavior of the data.