International Journal of
Innovation, Management and Technology

Abstract

The need of exchange rate forecasting in order to prevent its disruptive fluctuations has encouraged the monetary policy makers and economists for many years to find a powerful method to predict it. The determinants of exchange rate make its behavior to be complex, volatile and non-linear. In most of the studies done by researchers for exchange rate prediction, linear models such as econometric models and non-linear models such as neural networks have been applied. The lack of studies on the application of dynamic networks is the most important motivation of this study. In this paper Neural Network Autoregressive with Exogenous Input (NNARX) as a dynamic non-linear neural network, Artificial Neural Network (ANN) as a static neural network, Generalized Autoregressive Conditional Heteroscedasticity (GARCH) as a non-linear econometric model and Autoregressive Integrated Moving Average (ARIMA) as a linear econometric model are applied to forecast the Singaporean Dollar over US Dollar (SGD/USD) exchange rate in three time horizons. Comparison of the performance of different models is measured by different criteria. Results reveal that among all models, NNARX outperformed other models and among nonlinear models, NNARX outperformed ANN and both outperformed the GARCH model.