Can nominal exchange rates be differenced to stationarity?

The objective of this paper is to examine the well known assumption of difference stationarity of nominal exchange rates with the method of Leybourne et al. (1996). We conclude that the assumption of difference stationarity is innocuous for quarterly exchange rates but not for monthly exchange rates. In fact, a randomized unit-root process is appropriate for monthly exchange rates. These findings are important for exchange rates modeling.