This study examines the dynamic volatility of the Nigerian Stock Exchange All-Share Index (NGSEINDEX) daily log returns from October 28, 2015, to October 28, 2025, to provide a statistically sound basis for risk assessment in this critical emerging market. The empirical methodology employed a grid search of Generalized Autoregressive Conditional Heteroskedasticity (GARCH) family models under four distributional assumptions. Pre-estimation diagnostics confirmed the series’ mean stationarity and the presence of strong conditional heteroskedasticity. The EGARCH(1,1) model with the Generalized Error Distribution (GED) was selected as the superior specification based on information criteria (AIC=3660.22, BIC=3687.37). The estimation confirmed highly significant volatility persistence (beta1=0.8732) resulting in a slow decay half-life of approximately 5.11 trading days, and pronounced leptokurtosis (nu=1.0100), validating the heavy-tailed GED choice. The model exhibited strong out-of-sample predictive power (QLIKE Loss of 0.4013). These robust findings offer critical implications for portfolio managers, emphasizing the necessity of employing dynamic risk models like Value-at-Risk (VaR) that explicitly account for the observed persistence and heavy-tailed risk structure in the NGSEINDEX.
SHREEVASTAVA et al. (Mon,) studied this question.