Abstract
The characterization of real exchange rate series as random in nature has been questioned in recent times by the application of new statistical tools. This study uses a new test proposed by Giraitis et al. (Journal of Econometrics, 112, pp. 265–9, Citation2003) and based on KPSS (Journal of Econometrics, 54, pp. 159–78, 1992) test. The rescaled variance (V/S) statistic is shown to have a simpler asymptotic distribution and achieve a better balance of size and power than Lo's (1991) and KPSS (1992) test. Application of the new test suggests that real exchange rate movements do not show evidence of long memory.
Notes
1Froot and Rogoff (Citation1995) and Rogoff (Citation1996) provide excellent discussions.