Acknowledgments: This work was supported by the Hankuk University of Foreign Studies Research Fund of 2012. We are grateful to the Editor and two anonymous referees for many constructive comments and suggestions.
Empirical Comparison of Alternative Implied Volatility Measures of the Forecasting Performance of Future Volatility*
Article first published online: 9 FEB 2012
© 2012 Korean Securities Association
Asia-Pacific Journal of Financial Studies
Volume 41, Issue 1, pages 103–124, February 2012
How to Cite
Rhee, D. W., Byun, S. J. and Kim, S. (2012), Empirical Comparison of Alternative Implied Volatility Measures of the Forecasting Performance of Future Volatility. Asia-Pacific Journal of Financial Studies, 41: 103–124. doi: 10.1111/j.2041-6156.2011.01066.x
- Issue published online: 9 FEB 2012
- Article first published online: 9 FEB 2012
- Received 10 May 2011; Accepted 28 November 2011
- Stochastic volatility;
Implied volatility from the Black and Scholes (Journal of Political Economy 81, 1973, p. 637) model has been empirically analyzed for the forecasting performance of future volatility and is well known to be biased. Based on the belief that implied volatility from option prices can best estimate future volatility, this study identifies the best way to derive implied volatility to overcome the forecast bias associated with the Black–Scholes model. For this, the following three models are considered: Heston’s model, which best addresses the problems associated with the Black–Scholes model for pricing and hedging options; Britten-Jones and Neuberger’s model-free implied volatility (MFIV), which eliminates the model-oriented bias; and VKOSPI, the Korean version of the Chicago Board Options Exchange Market Volatility Index. This study conducts a comparative analysis of implied volatilities from the Black–Scholes model, Heston’s model, the MFIV, and VKOSPI for their abilities to forecast future volatility. The results of the empirical analysis of the KOSPI 200 options market indicate that Heston’s model can eliminate most of the bias associated with the Black–Scholes model, whereas the MFIV and VKOSPI do not show any improvement in terms of forecasting performance.