基于微观结构噪音修正的波动率估计——以中国股市逐笔交易数据为样本  

Volatility Estimation Based on Noise Bias Correction of Microstructure——Based on the Tick-by-tick Data of the China Stock Markets

作　　者：梁崴[1] 王春峰[1] 房振明[1] 张蕊[1] 

机构地区：[1]天津大学管理学院金融工程研究中心,天津300072

出　　处：《系统工程》2009年第2期1-6,共6页Systems Engineering

基　　金：国家自然科学基金资助项目(70771076);国家杰出青年科学基金资助项目(70225002)

摘　　要：基于中国股市逐笔交易数据,有效利用全样本不规则采样数据,采用二尺度已实现波动率(TSRV)方法对我国股票市场分离噪音下的日波动率进行估计,并将TSRV方法拓展至日内高频时段,结合RV模型进行了对比研究。研究结果显示逐笔交易数据相对传统超高频数据包含更多的交易信息,噪音的存在严重的干扰了波动率估计。TSRV方法在中国市场条件日间和日内都具有较好的适用性,能有效剔出微观结构噪音对波动率估计的影响,提高波动率的估计精度和稳定性。实际应用中,TSRV方法对低频采样频率的选取具有很好的鲁棒性。Based on the irregularly-spaced tick-by-tick data of the China stock markets, we estimate daily volatility under the noise of the split stock markets. In order to correct the noise bias, the Two-scale Realized Volatility estimator is introduced in this paper. Then the higher frequency interval volatility is estimated with TSRV method. The research indicates that the tick-by-tick data contain more information than other ultra-high frequency data~ that the microstructure noise severely interferes with volatility estimation. The TSRV works on inter-day and intraday intervals in the context of China stock market. As a less variable and more accurate estimator, the TSRV method has robustness in the choice of low frequency sampling rate in practice.

关 键 词：二尺度已实现波动率 遂笔交易数据 微观结构噪音 子采样 

分 类 号：F830[经济管理—金融学]