Forecasting exchange rates: Non-linear adjustment and time-varying equilibrium

Citation

Grossmann A & McMillan D (2010) Forecasting exchange rates: Non-linear adjustment and time-varying equilibrium. Journal of International Financial Markets, Institutions and Money, 20 (4), pp. 436-450. https://doi.org/10.1016/j.intfin.2010.06.004

Abstract
By linking two main strands of equilibrium exchange rate research, this paper models and forecasts exchange rate movements around a time-varying equilibrium using both linear and non-linear techniques. Our results support evidence of linear and non-linear (ESTR) stationary behaviour around a time-varying equilibrium, particularly when using a trade based price index. The latter results are largely robust across a break due to the Plaza Accord. Forecasts of both the equilibrium deviations and exchange rates themselves are largely supportive of the ESTR model over several alternatives. This is notably so across most measures with respect to the equilibrium deviations and over the sign based measures for the exchange rate forecasts. Overall, our results suggest that shortrun changes in exchange rates are forecastable when allowing for a time-varying equilibrium rate and using an appropriate price index. Such a result has important implications for researchers, policy-makers and goods and financial market participants. For example, policy-makers need to be cognisant of a changing equilibrium level and not necessarily conduct policy in such a manner as to restore a previous equilibrium. Similarly, those engaged in hedging need to be aware that equilibrium rates are time varying but, beneficially, movements around equilibrium appear predictable.

Keywords
Exchange rates; Forecasting; ESTR model; Time-varying equilibrium

Journal
Journal of International Financial Markets, Institutions and Money: Volume 20, Issue 4

People

Professor David McMillan

Professor in Finance, Accounting & Finance