【作者】 曲圣宁；

【导师】 张宗成；

【作者基本信息】 华中科技大学 ， 数量经济学， 2011， 博士

【摘要】 “噪音”这一概念最初是由包括Kyle在内的有效市场理论的支持者们从物理学界引入金融学界,并首次被界定为“信号”的对立概念。两者间都包含了信息：“信号”中包含的是反映了标的资产基本面或宏观经济政策等指标改变等真实信息；“噪音”中包含的是错误的信息或者无信息。但是,这些最初的研究并没有充分考虑“噪音”对标的风险资产交易价格的影响。随后,面对有效市场理论与现实金融市场大量的显著不符,新兴的行为金融学理论利用了这个概念建立起了噪音理论,认为由于噪音交易行为将噪音成分注入到交易价格中,因此该价格同时反映了“信息”和“噪音”。而基于噪音成分的交易导致市场套利有限,难以达到有效状态。根据噪音成分的形成机制和市场表现,本文将噪音的本质定义为：在金融市场中,一切引起投资者对标的资产现在或未来价值产生误判,并按照错误的定价进行交易的所有因素的总和。“噪音”不仅影响市场上的交易价格,也影响标的资产的交易数量。而且,基于“噪音”的交易包含于成交的每一笔交易之中。围绕着这一概念,本文分别从形成机制、具体表现、度量方法和影响因素检验等方面展开了较为深入的讨论。在所有的金融市场中,宏观经济政策、公司基本面改变、信息不对称程度和投资者情绪等因素都在影响着噪音交易的多少,和交易价格中噪音成分的大小,区别只是影响程度的不同而已。在中国股票市场上,沪、深两市总市值随经济总量的增加而上涨,但当前阶段中国股票市场上的投资者主要还是由中小投资者构成,虽然机构投资者比例有不断上升趋势。利用EGARCH模型,我们发现两市代表性股指收益率存在明显聚类效应。“利空”消息比“利好”消息会给市场收益率的条件方差产生更大的冲击,存在不对称效应。短期的“羊群效应”和“动量效应”以及长期的“反转效应”显著存在。在噪音成分的度量上,本文分别比较分析了以交易价差法和行为资产定价法为代表的直接度量方法和以Jensen’sα、R-S分形法以及Variance-Ratio模型为代表的间接度量方法。我们还对传统的Variance-Ratio模型进行了修正,根据金融市场的实际经验,加入了单位标的风险资产波动对组合或整个市场噪音成分波动的影响权重的考虑,建立了平均方差比(AVR)模型及其检验方法。利用R-S分析法以及AVR模型所计算得到的噪音成分指标,结合动态面板模型和系统GMM估计方法,本文比较和检验了不同持有期限两市噪音成分的大小以及影响因素对噪音成分的冲击。发现代表了上市公司基本面的各变量对噪音成分的解释力度较低,而且随着持有期限的加长,无论是影响力度还是系数的显著性都出现了下降的态势。宏观变量对噪音成分的解释力度次之,在长期持续显著。投资者情绪和不对称信息程度对噪音成分的影响最大,长期显著。最后,根据全文分析结果,我们对监管当局提出了一定的政策建议,期望能够促进中国证券市场健康发展。更多还原

【Abstract】 The concept "noise" was firstly introduced into the field of finance from physics by supporters of efficient market theory, including Kyle, and was defined as the opposite of the "signal". Both of these two things involve information, but "signal" contains true information which reflects the change of the foundation of the object or the fluctuation of the macroeconomics situation. On the contrary, "noise" covers wrong or non information. These original researchers did not give enough attention about the impact of the noise to the trading prices of the risky assets. Then, concerning the obvious unconformity between the efficient market theory and the market realities, new emerging behavior financial theory builds up noise trading theory and believes the investors’trading put the noise into the trading prices which reflect both the noise and the information, and then lead to the limitation of arbitrage and make the market hard to be efficient. According to the forming institution and expression of the noise in the market, we define the nature of the noise as the summation of all the factors that result to the investors’mispricing to the objective risky asset and its biased estimation of the future value. "Noise" affects not only the trading prices but also the trading volume in the market, and is involved in every individual trading. Based on this concept, this article makes research and examination from the angles of founding institution, representation, measuring methods, and impacting factors.In the financial markets of various countries, macroeconomics situation, foundation of the firms, information asymmetric, and the investors’emotion influence the volume of the noise trading and the noise component involved in the trading prices, and the difference is only about the extent.In both Shanghai and Shenzhen stock market, the total market value increased with the economic growth, but most part of the investors are middle and small ones, although the ratio of the institutional investors has a trend of increasing. By means of EGARCH model, we find the yield of the representative portfolio in two markets tend to be clustering. And "bad" information brings more strike to the conditional variance of the portfolio than "good" news, which means the impacts of these two kinds are dissymmetric. In the markets, herding and momentum effect in short run and reversal effect in long run are obvious.When measuring the noise, we introduce the direct methods, which include bid-ask spread and behavior asset pricing model, and indirect methods, which include Jensen’s a, R-S model and Variance-Ratio model. And we also make amendment about the traditional variance-ratio mode, and build up AVR model and its corresponding examining method.By use of the computed noise index from R-S model and AVR model, we analyze and examine the volume and the impact of some variables to the noise of different holding periods, through dynamic panel-data model and system GMM estimation method. We find the explanation from the variables representing the foundation of the firms to the noise is weak and decreasing with the time going, the explanation from the macroeconomic variables is better and obvious in the long run, and the investors’ emotion and the information asymmetry have the strongest explanation to the noise and also obvious in the long run.Finally, according to the conclusion of this paper, we bring forward some suggestion to those who are in charge of the operation of the China stock market to guarantee the healthy growth of the market.更多还原