摘要
在高频金融数据研究中,估计金融资产价格序列积分波动率时,往往需要考虑市场微观结构噪声与资产价格跳跃的影响。本文将市场微观结构噪声部分地表示成交易信息的参数函数,并结合资产收益序列的跳跃特征,提出资产收益的高斯混合模型。本文利用EM算法进行噪声参数估计的同时,识别资产价格的跳跃,进而提出一种新的积分波动率的估计量。本文提出的方法可以视为Li等(2016)的改进,并在模拟研究中,得到了比Li等(2016)更好的参数估计效果,且即使在跳跃幅度分布误设的情况下,也具有良好的识别跳跃的功能。在应用举例中,对比了本文方法与Lee和Mykland(2008)的跳跃发现方法,论证了本文的模型在识别跳跃方面的可靠性。
When estimating integrated volatility of a financial asset, the impacts of market microstructure noise and jumps should be taken into consideration in the research of high-frequency financial data. This paper proposes a Gaussian mixture model based on the market microstrncture noise partially expressed as a parametric function of trading data and the jump characteristics of the asset returns series. A new estimator of integrated volatility is put forward after the jumps of the assets prices are identified while EM algorithm is applied to estimate the parameters of noise. The model put forward in this paper could be regarded as an improvement of Li et 81. （2016）, with a better result in simulation study, and is able to perform well in detecting the jumps even when the distribution of jump range was set by mistake. At the end, in a practical example, In comparison with Lee and Mykland （2008）, the model has been justified in terms of its reliability in detecting the jumps.
出处
《统计研究》
CSSCI
北大核心
2017年第8期109-119,共11页
Statistical Research
基金
国家自然科学基金项目"金融资产配置中面板数据动态因子模型研究"(71271210)
国家自然科学基金项目"非对称随机波动建模及其在金融风险管理中的应用研究"(71471173)
教育部人文社会科学重点研究基地项目"金融风险测度与管理若干前沿问题研究"(14JJD910002)资助
中国人民大学2016年度拔尖创新人才培育资助计划成果
