Volatility Forecasting Using GARCH Models in Emerging Stock Markets: A Study of India
Abstract:
This study examines the volatility dynamics of the Indian stock market by forecasting the volatility of the BSE Sensex and NSE Nifty 50 indices using GARCH-type models, including GARCH(1,1), EGARCH, TGARCH, and FIGARCH. Daily closing price data from January 1, 2001 to March 30, 2025 were used to compute log re turns and estimate models through Maximum Likelihood Estimation. The results reveal strong evidence of vol atility clustering and high volatility persistence in both indices. Asymmetric volatility, where negative shocks exert greater impact than positive shocks of equal magnitude, is confirmed through EGARCH and TGARCH models. Comparative evaluation using AIC, BIC, RMSE, and MAE indicates that EGARCH and TGARCH outperform GARCH(1,1) and FIGARCH in both fit and forecasting accuracy. FIGARCH captures long-mem ory effects but does not provide superior forecasting performance for the Indian market. Structural break tests show no significant breaks over the study period, suggesting stable volatility behavior across major economic events. The findings carry practical implications for investors, risk managers, and policymakers, particularly in enhancing risk management, portfolio optimization, and regulatory oversight in emerging markets. The study highlights the importance of incorporating asymmetric models for more reliable volatility forecasting and recommends integrating macroeconomic factors and high-frequency data in future research.
