MODELLING THE VOLATILITY OF STOCK EXCHANGE MARKET CAPITALIZATION RETURNS IN NIGERIA USING GARCH MODELS
Abstract
Financial market volatility remains a significant concern for investors and policymakers, particularly in emerging economies, where market inefficiencies exacerbate risks. This study provided fresh insights into Nigeria's stock market volatility by comprehensively evaluating Generalized Autoregressive Conditional Heteroscedasticity (GARCH)-family models with alternative error distributions for market capitalization returns from 1990 to 2023. The analysis revealed striking findings. While standard GARCH models captured basic volatility clustering, only specifications incorporating heavy-tailed distributions adequately addressed the extreme fluctuations characteristic of this emerging market. The Threshold GARCH(1,1) model with Student-t innovations emerged as superior in modelling asymmetric volatility responses, with the EGARCH-Generalized Error Distribution (GED) specification showing infinite persistence - a remarkable finding suggesting shock impacts may never fully dissipate. Through rigorous comparison of Normal, Student-t and GED innovations, the study demonstrated that distributional assumptions significantly influenced volatility persistence estimates and forecast accuracy. The results challenged conventional modelling approaches by showing that even sophisticated GARCH variants leave some nonlinear dependencies unaccounted for, pointing to potential avenues for future methodological improvements. These findings carry important implications for risk management practices and regulatory policies in volatile emerging markets, particularly for portfolio managers seeking to mitigate downside risks in Nigeria's equity market. The study advances the empirical literature on volatility modelling while providing practical guidance for financial market participants operating in similar emerging market contexts.
Downloads
Published
Issue
Section
License
Copyright (c) 2025 Science World Journal

This work is licensed under a Creative Commons Attribution 4.0 International License.