Modeling crude oil price volatility in Nigeria: using GARCH (1,1), EGARCH (1,1), and GJR-GARCH (1,1) models

This study investigates the performance of various GARCH models for volatility forecasting, focusing on the GARCH (1,1), EGARCH (1,1), and GJR-GARCH (1,1) frameworks, each tested with normal and Student’s t-distributions. The models are evaluated using four information criteria: Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), Shibata Criterion, and Hannan-Quinn Criterion. The results reveal that the GARCH (1,1)-NORM model consistently outperforms other models in terms of all criteria, demonstrating the lowest values across AIC, BIC, Shibata, and Hannan-Quinn measures. In contrast, models using the student’s t-distribution (-STD) generally exhibit slightly higher information criteria values compared to their normal distribution counterparts (NORM). This suggests that the normal distribution provides a better fit for the data in this analysis. Additionally, while more complex models such as EGARCH and GJR-GARCH offer advanced features like capturing asymmetry and leverage effects, they do not significantly improve model performance over the simpler GARCH (1,1)-NORM model. The study recommends using the GARCH (1,1)-NORM model for its optimal balance of fit and simplicity while acknowledging that alternative models like ARIMA-GARCH, TARCH, and Stochastic Volatility models might be explored based on specific data characteristics and forecasting needs.