Forecasting exchange rates: a robust regression approach

The least squares estimation method as well as other ordinary estimation method for regression models can be severely affected by a small number of outliers, thus providing poor out-of-sample forecasts. This paper suggests a robust regression approach,based on the S-estimation method, to construct forecasting models that are less sensitive to data contamination by outliers. A robust linear autoregressive (RAR) and a robust neural network (RNN) models are estimated to study the predictability of twoexchange rates at the 1-, 3- and 6-month horizon. We compare the predictive ability of the robust models to those of the random walk (RW), the standard linear autoregressive (AR) and neural networks (NN) models in terms of forecast accuracy and sign predictability measures. We find that robust models tend to improve the forecasting accuracy of the AR and of theNNat all time horizons, and even of the RWfor forecasts carried out at the 1-month horizon. Robust models are also shown to have significantmarket timing ability at all forecast horizons.