Bicorrelations and cross-bicorrelations as non-linearity tests and tools for exchange rate forecasting

This paper proposes and implements a new methodology for forecasting time s eries, based on bicorrelations and cross-bicorrelations. It is shown that t he forecasting technique arises as a natural extension of, and as a complem ent to, existing univariate and multivariate non-linearity tests. The formu lations are essentially modified autoregressive or vector autoregressive mo dels respectively, which can be estimated using ordinary least squares. The techniques are applied to a set of high-frequency exchange rate returns, a nd their out-of-sample forecasting performance is compared to that of other time series models. Copyright (C) 2001 John Wiley & Sons, Ltd.