证券市场的噪音测度及其影响研究

发布时间：2018-02-25 05:13

本文关键词： 噪音交易噪音收益噪音水平噪音定价噪音价格行为　出处：《天津大学》2014年博士论文　论文类型：学位论文

【摘要】：随着行为金融学、连续金融学、非线性金融、证券市场微观结构、风险定价理论与实证等研究的深入，市场完美与完全的理想假设被不断放开与拓展，噪音研究也日益受到更大的重视。噪音的研究主要包含两方面，一是噪音的存在性及随之而来的如何识别或估计问题，二是，如果存在，它对市场的影响如何；一方面其存在破坏了市场有效性，因此，直观上认为其应越小越好；另一方面其又是市场存在的重要条件，如果没有噪音证券市场无利可图，市场也就无法长期存活。但由于其成份复杂，测度问题较难解决，因此影响研究也受其限制。因此，本文从噪音研究的两大问题出发，首先从交易、收益、水平三个层次探究噪音测度方法，然后从资产定价及价格行为角度探析了其对证券市场的影响。具体来说，本文做了如下工作并形成了相应的结论： 1）通过将噪音交易者引入交易树扩展经典的EKOP模型，本文重新构建了交易到达过程，然后，基于泊松到达理论推演交易到达过程的单期似然函数及多期联合似然函数，进而得到有效估计证券市场的噪音交易比例的模型。以HS300样本股为样本估计的估计结果表明，中国股票市场2011年1-7月平均噪音交易比例达0.2432，高于知情交易者比例。同时，参数结果分析表明期间噪音交易者情绪偏乐观，市场的信息效率不高。 2）通过将噪音因子引入O-U过程，本文构建了包含具有均值回复特征的噪音水平过程的三因子模型。然后，采用Kalman滤波方法并借助EM算法实现未知参数估计和噪音收益度量。最后，对上证综指（即SH000001）1991年1月4日至2012年2月24日期间的周噪音收益进行度量，结果表明，其噪音收益水平处于-23.00%至83.51%之间，存在右偏及尖峰特征，分析表明投资者理性程度及监管是影响噪音收益的重要因素。 3）从证券市场的非线性、确定性及混沌特征角度出发，本文将信号学中基于相空间重构理论下噪音水平的估计思想引入证券市场并构建了相应的估计模型。以20100104-20101214期间H300指数为样本本文检验了其高频资产价格时间序列数据的非线性、确定性及混沌特征，，在此基础上估计了其日噪音水平，估计结果表明H300指数期间噪音处于21.55-65.40%之间，且噪音水平存在右偏及扁平特征，与资产价格走势及市场信息之间可能存在复杂的关系。 4）在充分梳理资产定价理论与模型的基础上，本文借鉴MPT、CAPM、APT、LAPM、BAPM的思想及其结论，同时效仿Fama-French三因素模型的做法的基础上构建新的融入噪音的CAPM模型（NAPM）。并采用NAPM模型考察了中国股票市场的噪音定价情况。以HS300指数成分股为样本的实证结果表明，在中国股市中，噪音风险具有较强的正定价能力，且噪音水平越高的股票能得到更多的风险补偿。 5）最后，本文采用包含噪音与能够反映市场质量的价格行为指标的动态面板VAR模型去探究证券市场噪音与价格行为之间的真实动态关系。以HS300成份股为样本的实证研究表明，噪音与价格行为之间存在复杂的动态关系，噪音与非流动性及信息不对称成正相关，与波动率及交易量成反比，但在噪音形成初期关系可能不稳定。
[Abstract]:With the study of behavioral finance , continuous finance , non - linear finance , security market microstructure , risk pricing theory and demonstration , the study of noise is more and more attention . The research of noise mainly includes two aspects , one is the existence of noise and how to identify or estimate the problem . 1 ) The transaction arrival process is reconstructed by introducing the noise trader into the classical ekOP model of the transaction tree , and then , based on the arrival theory of Poisson arrival , the single - phase likelihood function and the multi - period joint likelihood function of the transaction arrival process are reconstructed , and then the model of effective estimation of the noise trade ratio of the security market is obtained . 2 ) By introducing the noise factor into the O - U process , this paper constructs a three - factor model containing the noise level process with the mean reversion feature . Then , using the Kalman filter method and using the EM algorithm to implement the unknown parameter estimation and the noise income measure , the results show that the noise income level lies between - 23.00 % and 83.51 % , and the right deviation and the peak characteristic exist . The analysis indicates that the investor ' s rational degree and regulation are the important factors that affect the noise income . 3 ) Based on the nonlinear , deterministic and chaotic characteristics of the security market , this paper introduces the idea of noise level under the theory of phase space reconstruction into the security market and constructs the corresponding estimation model . Based on this , the nonlinear , deterministic and chaotic characteristics of the high frequency asset price time series data are tested . The results show that the noise is between 21.55 and 65.40 % during the H300 index , and the noise level has right and flat characteristics , and there may be a complex relationship between the price trend of the asset and the market information . 4 ) Based on fully combing the theory and model of asset pricing , this paper uses the ideas and conclusions of MPT , CAPM , APT , LAPM , BAPM and constructs a new CAPM model ( NAPM ) based on the practice of Fama - French three - factor model . 5 ) Finally , a dynamic panel VAR model containing noise and price behavior index which can reflect the market quality is used to explore the real dynamic relationship between the noise and price behavior of the stock market . The empirical research on HS300 as a sample shows that there is a complex dynamic relationship between the noise and price behavior , and the noise and the non - liquidity and the information asymmetry are positively correlated , which is inversely proportional to the fluctuation rate and the transaction amount , but may not be stable at the early stage of noise formation .

【学位授予单位】：天津大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：F832.51

【参考文献】