金融时间序列的标度特性实证研究  

Empirical Research on the Scaling Properties of Financial Time Series

作　　者：苑莹[1] 庄新田[1] 

机构地区：[1]东北大学工商管理学院,辽宁沈阳110004

出　　处：《管理工程学报》2008年第2期85-89,共5页Journal of Industrial Engineering and Engineering Management

基　　金：国家自然科学基金资助项目(70371062)

摘　　要：分别从时间序列分析,简单分形分析及多重分形分析的角度,运用自相关、功率谱、盒维数以及多重分形谱方法对上证指数时间序列的标度特征进行了实证研究。首先,自相关函数和功率谱分析结果表明,上证指数时间序列在跨时间尺度的指数之间存在着相关性,表现为分形时间序列,说明市场未达到弱式有效;其次,盒维数的分析结果揭示了上证指数的聚类特征及标度变化,并且盒维数随临界值增加而递减的规律体现了多标度特征的存在;最后,多重分形谱的方法进一步确认了上海股票市场的多标度特性,并指出该特性是股票市场信息倍增级联过程的结果。Some statistical methods, including autocorrelation, power spectrum, functional box dimension and muhifractal analysis, were used on the daily Shanghai stock price index to examine the scale properties in the market. Firstly, the long-term memory was identified through the analysis of autocorrelation and power spectrum, respectively. It was found that the variation of indexes is statistically correlated, indicating that the stock market did not reached the soft efficiency. In addition, the clustering structure and scale properties of the time series were revealed by the analysis of functional box dimension, and the box dimension was shown to he a decreasing function of the threshold index level, implying muhifractal characteristics. Finally, the muhiscaling characteristics were analyzed by muhifractal scaling analysis. The origin of muhifractal properties in the Shanghai stock market can he interpreted in terms of the muhiplicative cascade process of stock market information.

关 键 词：资本市场复杂性 标度 盒维数 倍增级联过程 

分 类 号：F830.9[经济管理—金融学]