Modelling Nifty Volatility: Application of Hybrid GARCH Models

Abstract

In this study, the financial time series data on Nifty and India VIX have been used for estimating and forecasting volatility of returns on Nifty. Daily returns on Nifty are found to be negatively skewed and leptokurtic in nature. The daily return data are found to be stationary and show presence of ARCH effects. Volatility of daily return is found to be time-varying in nature. Leverage effects are found to be present and the same is reconfirmed by the shapes of the news impact curves. For evaluating the forecasting power of volatility models, an exogenous variable in the form of a lagged value of the India VIX, a proxy for implied volatility estimate, was included in the basic GARCH and GJR GARCH equations to obtain hybrid versions of these models. The experiment produced mixed results in terms of improvements in the predictive powers of the hybrid models. For the hybrid Basic-GARCH model, the predictive power improved while the predictive power weakened, though marginally, for the hybrid GJR GARCH model.