Weekly Univariate Nominal Exchange Rate Fluctuations

Abstract

The difficulties involved in explaining exchange rate movements during the post-1973 float with standard purchasing power parity, monetary, or portfolio balance models have become increasingly apparent. Meese and Rogoff (1983a, 1983b) systematically document the pervasive out-of-sample empirical failure of these models, and they find that a simple random walk model predicts the major rates during the floating period as well as (or better than) any of the alternative models.¹ These models (both structural and nonstructural) include a flexible price monetary model (Frenkel, 1976; Bilson, 1979), a sticky price monetary model (Dornbusch, 1976; Frankel, 1979), a sticky price monetary model with current account effects (Hooper and Morton, 1982), six univariate time series models, a vector autoregressive model, and the forward rate. The failure of the structural models is all the more striking in light of the fact that the Meese-Rogoff predictive comparisons use ex post realizations of exogenous variables.