社科网首页|论坛|人文社区|客户端|官方微博|报刊投稿|邮箱 中国社会科学网
返回首页
基于多标度分形理论的金融资产收益非对称性测度方法研究

作者:王 鹏

2013-03-19 10:13:33 来源:数量经济技术经济研究

(西南财经大学金融学院)

 

【摘要】 基于多标度分形理论,提出了一种新的更适用于实际金融资产收益数据的非对称性测度方法——两阶段非对称性检验法(Two-step asymmetry testingTAT),并运用Monte Carlo模拟考察了其与传统的偏度系数检验法的非对称性判定结论差异。实证结果表明:总体来讲,本文提出的两阶段非对称性检验法在常用检验水平下取得了较偏度系数法更为准确的金融资产收益非对称性判定结论,且两阶段非对称性检验法较偏度系数法更适用于具有非独立、非正态特性数据的非对称性检验。

关键词  多标度分形理论  两阶段非对称性检验法  偏度系数检验法  Monte Carlo模拟

中图分类号  F224  F830                 文献标识码  A

 

Testing Asymmetry in Financial Asset Return Based on Multifractal Theory

Abstract: Asymmetry in financial asset returns is not only one factor should be considered in asset pricing and portfolio selection, but also related to risk measurement and derivatives pricing. In traditional study, the common approach to test asymmetry in asset return distributions is using the coefficient of skewness which is defined as the standardized third central moment. However, when using the coefficient of skewness to test asymmetry, the key to make the conclusion right and effective is that not only asset prices are independent of each other, but also the asset return should obey normal distribution. In this paper, a new asymmetry test based on multifractal theory, two-step asymmetry testing (TAT), is proposed. A Monte Carlo study showed that this test is competitive with coefficient of skewness test in common significance levels generally and that, for dependent and non-normal data, TAT testing work more properly.

Key words: Multifractal Theory; Two-step Asymmetry Testing; Coefficient of Skewness Test; Monte Carlo Simulation

 

 

无论是从理论上,还是从金融管理的实践上,对金融资产收益非对称性的测度都是具有重要意义的课题。这不仅是由于非对称性是投资组合选择过程中应该考虑的一个重要因子,而且还对金融风险识别与测度和衍生品定价的准确性等方面都有较大影响。因此,从20世纪90年代开始,对金融资产收益非对称特征进行检验和描述的相关研究就开始逐渐兴起,并在最近几年得到了越来越多的重视。

传统上,测度金融资产收益非对称性的方法是基于定义为随机变量分布标准化三阶中心矩(The standardized third central moment)的偏度系数(Coefficient of skewness)(Lau et al.1989Christofferson2003Korkie et al.2006)。然而,在运用偏度系数进行非对称性检验时,决定所得结论有效性的关键不仅在于资产价格变化之间的独立性,而且还取决于其对资产收益服从正态分布这一重要前提假定的满足(Kendall1969)。换句话说,当资产价格变化(收益率)不服从正态分布或相互间不独立时,基于传统偏度系数的非对称性判断结论是非常值得怀疑的。然而,众多严谨的实证研究早已表明,尽管金融资产收益水平值的相关性并不明显,但收益率的平方(或绝对值)之间却往往呈现出较强的正相关,这说明不同时点上的资产价格变化(收益率)并不是独立的(Engle1982Mcmillan and Speight2006;许启发,2006Poshakwale and Aquino2008)。除此之外,在常用的抽样频率上(如日收益率或周收益率),正态分布假设更是会被强烈拒绝(Linden2001Tolikas and Gettinby2009;黄德龙、杨晓光,2008;王天一、黄卓,2012)。



[1] 本文获得国家自然科学基金资助,编号:71101119