| Abstract: | We examine long-run PPP between Germany, Great Britain, Japan and the United States over the period 1930–1996 using multivariate cointegration techniques. Bilateral PPP between the four countries is examined in one system (as opposed to e.g. series of trivariate systems). In all of the statistical analysis, asymptotic tests are augmented by parametric bootstrap analogues, whereby we reduce, if not eliminate, the size distortion typically present in small-sample studies. The cointegration analysis provides support for the necessary conditions for PPP (i.e. cointegrating relations are found) but not for the sufficient conditions (i.e., the coefficients in the cointegrating relations are far from what PPP predicts). These results are at odds with results from other studies that also analyze long-horizon data sets.First version received: November 2000/Final version received: February 2003Comments by Stefan Norrbin, Lee Ohanian, Anders Vredin, seminar-participants at Sveriges Riksbank (the central bank of Sweden) and at the Econometric Society European Meeting in Santiago de Compostela are gratefully acknowledged. |
