ARIMA and Symmetric GARCH-type Models in Forecasting Malaysia Gold Price

Abstract

Abstract Gold price modelling is crucial in gold price pattern determination since the information can be used for investors to enter and exit the market. The model selection is important and corresponds to the gold price movement characteristics. This study examines the forecasting performance of autoregressive integrated moving average (ARIMA) with symmetric generalised autoregressive conditional heteroscedastic (GARCH)-type models (standard GARCH, IGARCH and GARCH-M) under three types of innovations that are Gaussian, t and generalized error distributions to model gold price. The proposed models are employed to daily Malaysia gold price from year 2003 to 2014. The empirical results indicate that ARIMA(0,1,0) - standard GARCH(1,1) using t innovations is the most preferred ARIMA with symmetric GARCH-type model.