On the Importance of Span of the Data in Univariate Estimation of the Persistence in Real Exchange Rates

This paper revisits the empirical evidence on real exchange rates' convergence to their purchasing power parity (PPP) levels. In their recent empirical study, Murray and Papell (2002) claim that the univariate approach provides no useful information on the size of the half-lives of real exchange rate deviations from PPP. However, we obtain finite confidence intervals for the half-life for a maximum of 8 out of 16 countries by applying the nonparametric grid bootstrap technique of Hansen (1999) to over a century of real exchange rates data for 16 developed countries relative to the US dollar. Our finding sharply contrasts to that of Murray and Papell (2002) with the post Bretton Woods real exchange rates. Our finding suggests that span of the data, not the estimation methods, matters more for obtaining useful information on long-run propositions such as PPP.