Long-run purchasing power parity redux

Abstract

We test for long-run covariability between nominal exchange rate depreciation and inflation differentials to provide a new perspective on long-run Purchasing Power Parity (PPP). The method directly tests Cassel’s concept of relative PPP and is more robust to departures from exact unit roots in nominal exchange rates or relative prices than standard unit root and cointegration tests for PPP. The central result of the paper is that the 90 percent confidence interval for (1) the long-run correlation coefficient is above zero and (2) the long-run linear regression coefficient contains one and, therefore, long-run PPP cannot be rejected for 9 of the 16 countries. For six of the countries, adding a third criterion that the confidence interval for the long-run linear regression coefficient have relatively narrow bands provides strong evidence of long-run PPP. The evidence of long-run PPP is much stronger for high inflation/high depreciation countries than for low inflation/low depreciation countries.