Realized Volatility Of CSI300Stock Index Futures:an Empirical Study

In theoretical study and practical application, volatility of asset palys an important role in asset pricing and risk management. Thus, how to model and forecast the daily return volatility has become increasingly important. Meanwhile, with the development of computer technology, trading frequency has be-en significantly improved. And to get high-frequency data has become quite easy. Therefore, how to effectively incorporate intraday high-frequency data into the modeling of daily return volatility is supposed to be the main issue to many researchers and practitioners. Lots of literatures having the concern with modeling and forecasting the dynamic dependencies in financial market volatility have appeared over the past two decades.Until fairly recently, most of the empirical results in the literature were based on the use of daily, and coarser frequency data, coupled with formulations within the GARCH or stochastic volatility class of models. Meanwhile, somewhat of a paradigm shift has started to occur, which can incorporate high-frequency data and use it to model and forecast longer-run volatility problems. We do all of the above through the use of simple reduced-form time series models for non-parametric daily realized volatility measures basing on the summation of intraday squared returns. Furthermore, decomposing the total daily return variability into its continuous and discontinuous components is based on the bi-power variation measures which is developed by Barndorff-Nielsen and Shephard (2004,2006). Following the analysis of Andersen, we begin to decompose the total return variability du-ring the trading days into its continuous sample path variation and the variation due to jumps. And we test for structural changes of the CSI300future contract return volatility by using Bai and Perron(2003) method. Our empirical results with a380-day sample of high-frequency intraday CSI300futures contract returns confirm that earlier findings of the dynamic dependencies in the daily continuous sample path variability has been well described by an approximate long-memory HAR model, as originally proposed by Corsi (2004). At the same time, the dynamic dependencies in the non-parametrically identified significant jumps appear to be well described by an ACH model. And we found two structural changes in the realized volatility model. Lastly, we discuss how the resulting reduced form model structures for each one of the three components, which may be used in the construction of out-of-sample forecasts for the total return volatility.