Volume 12, Number 1, June 2017
| Minimum Variance Hedge Performance of the Realized-Volatility-Based GARCH Model with Alternative Conditional Correlation Settings |
Abstract
We propose the augmented GARCH model (GARCH-RV) which extends the bivariate GARCH model by including the realized volatility (RV) as explanatory variable for variance equations. We comprehensively consider constant-, dynamic- and asymmetric generalized dynamic- conditional correlation settings to model the variance-covariance matrix for the GARCH-RV model. The proposed hedging models are employed to estimate minimum variance hedge ratios and evaluate their hedge performance for S&P 500 stock index futures under high and low volatility periods, where the high and low volatility periods can be clearly identified by using a Markov switching model with two-volatility regimes. Empirical results indicate that the hedging performance based on the bivariate GARCH models with RV estimators can be substantially improved in terms of risk reductions and economic value gains as compared with those without RV estimators.
Keywords: Hedge, GARCH-RV Model, Markov Switching Model, Economic Value
JEL Classification: C52, C53, G32