Exchange Rate Modeling by Multivariate Nonlinear Cointegration Analysis Using Artiicial Neural Networks

In this paper we investigate the merits of artiicial neural networks in forecasting foreign exchange rates. From previous research it is known that it is hard to beat the random walk model using structural exchange rate models. In this paper we show that by using a suitable multivariate speciication a structural model can be derived that beats the random walk. By introducing a new method for multivariate nonlinear cointegration analysis, based on the linear method of Johansen (1988), we construct a neural network error correction model for the yen/dollar, pound/dollar and DM/dollar exchange rates that signiicantly outperforms both the random walk model and a linear vector error correction model.