Improving Volatility Identification and Prediction of Realized Stochastic Volatility Model with Implied volatility
33 Pages Posted: 3 Jun 2019 Last revised: 15 Jul 2020
Date Written: August 4, 2019
Abstract
We examine a group of extended realized stochastic volatility (RSV) models with ex-ante volatility information added to the framework. The most advantageous specification is the one with implied volatility (IV) as an explanatory variable in the latent volatility process, which produces an estimated latent volatility almost identical to the realized volatility. Including IV generalizes the traditional AR(1) specification of the volatility process to a market-based process, which captures the dynamics of the daily realized volatility. The purposed RSV model with IV outperforms all benchmark models in out-of-sample return density prediction. The empirical finding is valid for both equity index and major individual stocks for a data sample with daily observations over twenty years.
Keywords: Stochastic volatility, realized volatility, implied volatility, MCMC, Bayesian forecasting
JEL Classification: C11, C15, C22, C52
Suggested Citation: Suggested Citation