Systematic sampling and real exchange rates

Systematic Sampling and Real Exchange Rates. — Four major real bilateral exchange rates are shown to be well characterized by nonlinear stationary models over the recent float. Using Monte Carlo methods, this paper examines the effects of systematic sampling on the behaviour of real exchange rates and shows that: systematic sampling reduces significantly nonlinearity in real exchange rates and affects their lag structure; given a certain span, the frequency of the data set becomes crucial for detecting mean reversion in real exchange rates once the analysis is switched from a linear to a nonlinear model. Monte Carlo simulations also suggest that the parameter governing the speed of nonlinear mean reversion may be upward biased.