Comparative Analysis of the Forecasting Ability of the GARCH-Type Models on the Returns Series of Nigeria Exchange

ABSTRACT

The aim of the work was to explore GARCH-type models using three different distributions (Normal, Student-t and the Generalized Error Distributions) with a view of comparing their forecasting ability of volatility on the returns of Nigeria exchange rate for period of January 2002 to December 2020 comprising of 228 observations. The following GARCH models were employed GARCH (1,1), GJR-GARCH and EGARCH models. The result revealed that the returns were characterized by significant volatility persistence and asymmetric effect irrespective of the distribution. Using the RMSE, MAE and the Theil’s inequality coefficient as the measure of predictive accuracy to appraise the forecast performance of all the models it was found that GJR-GARCH model produced better forecast regardless of the distribution adopted for the exchange rate returns series.