Is Real Exchange Rate Mean Reversion Caused By Arbitrage?

The presence of purchasing power parity is often attributed to the exploitation of arbitrage opportunities in goods markets. We examine this presumption for a 1960-1996 monthly panel of bilateral exchange rates and trade for the G7 countries. The data exhibit strong mean reversion. However, despite allowing for substantial latitude in specification, we find very limited support for a simple arbitrage view. The deviations of real exchange rates and trade from trend are virtually uncorrelated. Large trade deviations neither trigger nor accelerate mean reversion. Large real exchange rate deviations do not lead to systematic changes in trade. Constricting the sample to eighteen-month episodes of notable mean reversion - large persistent depreciations starting from overvalued levels - does not reveal any systematic relation either. The timing of these episodes does point, however, to an alternative explanation of mean reversion: the majority of episodes occur during periods of nominal exchange rate regime instability, pointing towards exchange rate policy or speculation as the immediate cause of mean reversion. Both may, of course, reflect expectations of trade responses, opening an indirect role for incipient arbitrage in explaining mean reversion.