Nonlinear error correction in spot and forward exchange rates

Abstract

Recent research has increasingly suggested that exchange rates may be characterised by non-linear behaviour which results from the existence of market frictions. This paper examines whether such non-linear behaviour is evident, not in rates themselves, but in the adjustment of rates back to some fundamental equilibrium. Thus, we examine a series of six spot and forward exchange rates to see whether a non-linear error-correction model, which exhibits asymmetric adjustment back to equilibrium either in terms of the size of the deviation from equilibrium or the sign of the deviation outperforms either a random walk model for rates or a linear error-correction model. Our in-sample results suggest that the non-linear models outperform both the linear models, with evidence of significant sign and size threshold effects. Out-of-sample forecasts lend further support for the non-linear models.