GARCH Modelling of Nigeria Stock Exchange Returns with Odd Generalized Exponential Laplace Distribution

Abstract

Communication in Physical Sciences, 2023, 9(4): 572-584

Authors: Job Obalowu and ^*Reuben Oluwabukunmi David

Received: 18 March 2023/Accepted 05 September 2023/

The modeling of the volatility of asset returns plays an important role in risk assessment and decision-making processes for both investors and financial institutions. In this paper, we have modeled the volatility in Nigeria Stock Exchange (NSE) returns using the Generalized Autoregressive Conditional Heteroskedasticity (GARCH) model and some of its variants with an Odd Generalized Exponential Laplace Distribution (OGELAD) due to its ability to capture the time-varying and nonlinear nature of financial time series. Fitting the different models indicates the new error distribution outperforms other error distributions for all volatility models. The majority of the parameters for all fitted models and error distributions are significant at 5%, 1%, and 0.1% level of significance. The diagnostic check of the fitted models shows they have been adequately specified. Furthermore, the forecasting performance of the fitted models shows that the new error distribution outperformed existing error distributions in out-sample forecasts. While GARCH (1,1) with an OGELAD is selected for fitting the volatility, the GJR-GARCH (1,1) model with an OGELAD is preferred for forecasting the volatility of NSE returns. Thus, GARCH models with a non-normal error distribution provide a robust distribution for modeling volatility