TIME SERIES FORECAST MODELS FOR FOREIGN EXCHANGE MARKET IN A DEVELOPING ECONOMY

Abstract

Technicalities in foreign exchange forecasting have been of interest to investors and academia, particularly in a developing economy. Data of foreign exchange are time series in nature and several techniques have been developed to modeling and forecasting foreign exchange rates. In this study, Nigeria's foreign exchange rate against three world-leading currencies (US Dollars, Euro and Pounds Sterling) are modeled with ARIMA, Auto. arima, Box-Cox, random walk forecast, and Exponential Smoothing and subjected to comparative tests using  Diebold-Mariano criteria with a modern model time series model. The empirical analysis shows that that the modern model outperforms some of the other techniques in forecasting Nigeria exchange rates against world-leading currencies particularly when the forecast horizon is low.