在预测未来波动率时,究竞是基于历史数据的时间序列模型还是基于期权价格的隐含波动率模型效率更高?本文对香港恒生指数期权市场所含信息的研究发现,在预测期限较短(一周)时,GARCH(1,1)模型所含信息较多,预测能力最强,但在预测较长期限(一个月)时,隐含波动率所含信息较多,预测能力较强。同时,期权市场交易越活跃,所反映的信息就越全面,隐含波动率的预测能力也就越强。
It is an interesting question that which is more efficient in forecasting the future volatilities, the time series models based on historical data or implied volatilities obtained directly from the option prices. The study based on Hang Seng Index (HSI) options suggests that when the forecast horizon is one week, the GARCH (1, 1) volatilities contains all information in implied volatilities, while the result is the opposite and implied volatilities are more efficient in the prediction of future volatilities when the horizon is one month. The larger the option trading volume, the more the information contained in implied volatilities.