Modelling of volatile data has become the area of interest in financial tim series recently. Volatility refers to the phenomenon where the conditional variance of the time series varies over time. The objective of this study is to compare the modelling performance of Generalized Autoregressive Conditional Heteroscedasticity (GARCH) and hybrid ARIMA-GARCH model for the prices of crude oil. Eviews and Minitab software are used to analyze the data. The models investigated are GARCH and hybrid ARIMA-GARCH model. In parameter estimation, Maximum Likelihood Estimation (MLE) is the preferred technique for GARCH models while Ordinary Least Squares Estimation (OLS) and MLE will be used for hybrid ARIMA-GARCH models. The goodness of fit of the model is measured using Akaike’s Information Criterion (AIC). The diagnostic checking is conducted to validate the goodness of fit of the model using Jarque-Bera test, Serial Correlation test and Heteroskedasticity test. Forecasting accuracies for both models are assessed using Mean Absolute Error (MAE) and Mean Absolute Percentage Error (MAPE). The model which gives the lowest measure of error is considered to be the most appropriate model. Empirical results indicate that modelling using hybrid model has smaller AIC, MAE and MAPE values compared to GARCH model. It can be concluded that hybrid ARIMA-GARCH model is better in modelling crude oil prices data compared to GARCH model
展开▼
