COMPARATIVE FORECASTING OF MOROCCAN DIRHAM EXCHANGE RATES: ARMA VS. MARKOV SWITCHING AUTOREGRESSIVE (MS-AR) MODELS

Abstract

This paper has modelled and predicted the exchange rate between the Moroccan Dirham, the EURO and the US Dollar, for this reason two types of models have been used (the MS-AR and the ARMA model), the results of our study show that the MS-AR model outperformed the ARMA in modeling the two real exchange rates EUR/MAD and USD/MAD. The BIC information criterion by shows that the MS-AR model beats the ARMA model based on random walk, while the forecasts of both models show that the ARMA has more accurate forecasts.