【作者】 毛羽丰；

【导师】 门明；

【作者基本信息】 对外经济贸易大学 ， 金融学， 2014， 博士

【摘要】 反馈交易者是金融市场上存在的一类特殊的非理性交易者，他们不以证券的基础价值为投资决策的准绳，而是以过去的价格涨跌为交易准则，他们在市场上长期存在并与理性交易者进行博弈，影响资产价格的形成过程和市场的稳定性。我国的股票市场具有“新兴加转轨”的特点，一些独特的市场条件和制度因素使得我国股票市场的反馈交易特征、影响因素和作用机制都具有其独特性。本文主要分五个方面对此进行探讨：第一，本文厘清了反馈交易的基本概念和历史渊源；从行为金融学的角度，关注认知偏差和情绪偏差的影响，对反馈交易产生的心理根源进行剖析；在此基础上，进一步挖掘了投资实践中反馈交易的形成机制；并对反馈交易的相关理论进行介绍和推证，其中DSSW噪音交易理论和行为资产定价模型（BAPM）是反馈交易理论的先导和理论根源，而行为资本资产定价模型（BCAPM）和行为跨期资本资产定价模型（BICAPM）分别基于局部均衡和一般均衡的理念为反馈交易提供了可供检验的实证分析框架，是本文的理论基础。第二，本文在BCAPM框架下，使用EGARCH(1,1)-M模型对我国股市各行业板块之间反馈交易的差异性进行了探讨，把对反馈交易的研究延伸至了更加微观的行业层面。通过分析，本文证实了上海股票市场的反馈交易行为主要来自能源、医药、消费和电信四个行业的股票，而其余六个行业板块不存在显著的时变性反馈交易行为。波动率和市场涨跌因素对能源、医药、消费和电信行业反馈交易行为的影响有差异的，表现在：一方面，消费行业投资者的反馈交易行为对波动率的敏感度最高，其次为医药行业，电信和能源行业的这一敏感度略低；另一方面，能源和医药行业股票投资者的反馈交易行为受到市场涨跌因素的影响，具有显著的不对称性，在市场下跌期间，正反馈交易效应强于市场上涨期间；而消费和电信行业的反馈交易行为受市场涨跌因素的影响并不显著，因此不具有显著的不对称性。从正反馈交易效应作用的时间上看，对于存在时变性反馈交易行为的四个行业，正反馈交易者主导市场的现象均一致的出现于2006至2009年间，其余时期正反馈交易者对市场的主导作用并不显著。第三，本文运用BCAPM分析框架，建立GJR-GARCH(1,1)-M模型，使用几乎覆盖我国股市全部历史的数据，对存在一定程度分割的沪深A、B股和香港H股股票市场的反馈交易行为进行了实证分析，结果显示分割市场各部分之间的反馈交易性行为特征和影响因素具有差异性：（1）沪深A、B股市场存在显著的反馈交易者行为作用的特征，倾向于表现出新兴加转轨市场容易受正反馈交易者的不稳定影响的特点；H股、红筹股市场并不存在反馈交易者主导市场定价的显著证据，更多的体现出成熟股票市场的特征。（2）影响A、B股市场正反馈交易行为的因素存在差异， A股市场的正反馈交易更多取决于市场涨跌因素，一旦市场出现下跌，正反馈交易者就会对市场定价呈现显著影响，“杀跌”行为比“追涨”行为对市场的影响显著，使得市场呈现更大的不稳定性；而B股市场的正反馈交易更多取决于市场波动率因素，较高的波动率水平会有助于正反馈交易者发挥主导作用。（3）沪市和深市对反馈交易行为影响因素的敏感程度具有显著差异，深市的反馈交易行为对波动率和市场涨跌的敏感程度比沪市更高，也就是说，深市比沪市更容易出现正反馈交易者主导市场的现象。深市的投资者比沪市有更显著的追涨杀跌的行为。在关注A、B、H股市场反馈交易横向差异的同时，本文还使用引入虚拟变量修正的实证模型，分析了B股对内地个人投资者开放和QFII制度引入这两大改变市场分割状态和投资者结构的事件分别对沪深B股和A股市场的反馈交易行为在短期和长期均的影响，结论显示：（1）由于内地个人投资者的加入，B股市场总的来说更容易出现显著的正反馈交易效应，市场价格更加趋于不稳定。这种改变主要来自与波动率变化无关的收益率自相关性的减小，市场出现正反馈交易效应的门槛波动率水平较事件发生之前大为降低。可见内地散户投资者是组成正反馈交易者队伍的“生力军”，他们的非理性行为为市场增加了更大的不稳定因素。（2）QFII制度的引入在短期（2-4年）对A股市场的反馈交易行为模式没有影响；在长期来看，上证A股市场与波动率变化无关的收益率自相关水平显著降低，市场更容易出现显著的正反馈交易效应；而深证A股市场在QFII制度引入前后反馈交易行为的相关特征和波动率水平均无显著变化。可以认为A股市场对外国机构投资者开放并未使得A股市场变得更加不稳定。第四，运用与分析B股准入管制放开和QFII制度引入的影响相同的实证方法，评估股指期货推出对沪深股市反馈交易行为的影响，发现股指期货推出后，沪深A股市场的反馈交易行为特征发生了显著变化，正反馈交易行为在一定程度上得到了抑制。其中，变化最显著的是与波动率水平相关的那部分反馈交易行为，在股指期货推出后的2年至4年的时间里，沪深A股市场收益率自相关性对波动率的敏感程度相比之前显著下降，即使市场波动率升高，正反馈交易者对市场定价的影响也不再显著。与波动率变化无关的那部分收益率自相关性虽然只在2年的时间跨度上有显著增加，但也一定程度上增加了正反馈交易者发挥作用的门槛波动率水平，间接有助于抑制正反馈交易者使市场更加不稳定的影响。第五，使用BICAPM分析框架，以一般均衡的思维，建立二元GARCH(1,1)-BEKK-M模型，对我国外汇市场和股票市场的反馈交易行为进行联合实证分析，结果显示：除波动率水平和市场涨跌因素之外，外汇市场和股票市场间的联系是影响沪深A、B股市场反馈交易行为的重要因素，其影响程度超过波动率水平。在上证B股市场上，市场关联对该市场反馈交易的影响要小于其他市场，且方向与其他市场相反，表现为：在股市上涨、人民币汇率贬值或股市下跌、人民币升值时，上证B股市场容易呈现显著的正反馈交易效应，否则可能负反馈交易者在定价中发挥主导作用。而对于上证A股、深证A股和深证B股市场，则在股市上涨、人民币升值或股市下跌、人民币贬值时，上证A股、深证A股和深证B股市场会容易呈现显著的正反馈交易效应，否则多半负反馈交易者会在市场定价中发挥主要作用。更多还原

【Abstract】 Feedback traders are a special class of irrational traders in financial markets.Theymake investment decisions according to the last trading price change rather than thefundamentalvalue of securities.They coexist with the rational traders in the long runand interact with them, which influences the formation process of asset prices andmarket stability. Chinese stock market is subject to the characteristics of emergingmarkets and running in the transition period.There exists uniqueness for thecharacteristics, factors and mechanism of the feedback trading in Chinese stockmarket due to some special market conditions and arrangements. Wewill explore thisin five different parts:First, we clarify the basic concepts and historical roots of feedback trading.Also, weconduct behavioral financial analysis on the psychological basis of feedback trading,considering the impact of cognitive biases and emotional bias. Furthermore, wediscuss the generation mechanism of feedback trading in investment practice.Then weintroduce the relative theories of feedback trading,including DSSW noise tradingtheory and behavioral asset pricing model (BAPM), both of what are the preludes andsource of feedback trading theories, also including the behavioral capital assetspricing model (BCAPM) and the behavioral intertemporal capital asset pricing model(BICAPM), which are based on the concept of partial equilibrium and generalequilibrium and provide a testable empirical framework for feedback trading. Weemploy the BCAPM and the BICAPM as the theoretical basis of this dissertation.Second, with BCAPM framework and EGARCH (1,1)-M model, we discuss thefeedback trading behavior differences among various industry sectors of Chinesestock market and extend the study of the feedback trading to a more micro perspective.Through analysis, we confirmed that the feedback trading effects in Shanghai stockmarket mainly stem from the energy, health care, consumer staples and telecomservices sectors, while the other six industry sectors have no significant time-varyingfeedback trading effects. The volatility and market up-and-down have different impacton thefeedback trading of energy,health care, consumer staples and telecom servicessectors. On the one hand, feedback trading behaviors of the consumer staple sectorinvestors are most sensitive to the volatility, followed that of the health care sectorinvestors. In telecom service and energy sectors, the sensitivities of feedback tradingto volatility are slightly lower.On the other hand, the feedback trading behaviors fromenergy and health care sectors are significantly asymmetric.Positive feedback tradingeffect is much stronger during the market decline than the market increasing. It’s notthe case for the consumer staples and telecom service sectors. The timing of thepositive feedback trading effect is similar for the four sectors. The positive feedbacktraders dominate the market consistently during the period from2006to2009. Forother time within the sample period, it’s not the case. Third, we adopt the BCAPM analytical framework and establish a GJR-GARCH(1,1)-M model to examine the differences of feedback trading behaviors amongvarious segmented parts of Chinese stork market. Our empirical studies suggest that A-share, B-share and H-share markets exhibit different feedback trading behaviors:(1)There exists significant feedback trading behavior in A-share and B-share market butnone for H-share and red-chip market.(2) Volatility and market up-and-down havedifferent effects on the feedback trading behavior in A and B-share markets. ForA-share market, positive feedback trading behavior depends more on the marketup-and-down. Once the market declines, the positive feedback traders will dominatethe A-share market. For B-share market, positive feedback behavior depends more onthe volatility. Positive feedback traders will dominates B-share market during theperiod with high volatility.(3) The feedback trading behaviors in the Shanghai andShenzhen stock markets have significant differences in the sensitivity. Compared tothat in Shanghai stock market, feedback trading in Shenzhen stock market is moresensitive to both volatility and market ups and downs.In other word, Shenzhen stockmarket is more prone to positive feedback traders than Shanghai stock market.Concerned about the horizontal differences of feedback trading behaviors among A, B,and H-share markets, we also introduce an empirical model modified by dummyvariables toanalyze the policy effects of two events. One is the deregulation ofB-share market, the other is the launch of QFII. Both have changed the marketsegmentation and the structures of investors in the relevant parts of Chinese stockmarkets. We concluded that:(1) B-share market,rather than A market, has moresignificant positive feedback trading effects after the access of individual investorsfrom mainland.And B-share market price tends to be more volatile. This change isprimarily attributed to the decrease inreturn autocorrelation, which is irrelevant to thevolatility change. And the change of autocorrelation lifts the thresholdvolatility abovewhich positive feedback traders dominates the market. Therefore, individual investorsfrom mainland of China consist of the main force of the positive feedback traders inChinese B-share market.Their irrational investment behaviorsdestabilize the market.(2) In short term (2~4years), the launch of the QFII has no significant effect on thefeedback trading pattern of the A-share market; in the long run,it lowers theautocorrelations irrelevant to the conditional volatility in Shanghai A-share marketand makes the market more prone to significant positive feedback trading effect. As tothe Shenzhen A-share market, after the launch of the QFII,there is no significantchange on the feedback trading behavior and volatility levels. The launch of QFIIenables the foreign institutional investors to participate the A-share markets and thisdoesn’t destabilize the A-share market.Fourth, we analyze the effects of the introduction of stock index futures with the samemodel as we did in the analysis of the deregulation of B-share market and the launchof QFII. We find that after the introduction of the stock futures index, thecharacteristics of the feedback trading behaviors changes a lot in the A-sharemarket.The positive feedback trading has been alleviated to some extent, whichattributes to the changes in the sensitivity of feedback trading behavior to thevolatility levels. During the two or four years after the introduction of the stock index futures, the volatility sensitivity of the return autocorrelation decreased significantlycompared to the period before the event.Even if market volatility increases, thepositive feedback traders have no significant influence on the pricing of the A-sharemarket. The autocorrelations irrelevant to the volatility increase significant only in thetwo-year window. That increases the threshold volatility above with positive feedbacktraders dominate the market. In this indirect way, it limits the destabilizing influenceof positive feedback traders.Fifth, on the basis of BICAPM analytical framework and general equilibrium thoughts,we build an bivariate GARCH (1,1)-BEKK-M model and jointly examine thefeedback trading behaviorsin foreign exchange market and the stock markets. Ourfindings suggest: In addition to volatility and market ups and downs, covariancebetween the foreign exchange and stock markets is also an important factor which hassignificant impact on the feedback trading behaviors in the A and B-share stockmarkets.And its influence is larger than that of the volatility and the marketup-and-down. The covariance influences the Shanghai A-share market, ShenzhenA-share market and the Shenzhen B-share market more than the Shanghai B-sharemarket. The covariance works in the way as follows: When stock markets rise,Renminbi appreciates, or the stock markets decrease, Renminbi depreciates, positivefeedback traders will dominate in the Shanghai A-share market, Shenzhen A-sharemarket and the Shenzhen B-share market; When stock markets rise, Renminbidepreciates, or the stock markets decrease, Renminbi appreciates, positive feedbacktraders will dominate in the Shanghai B-share market.更多还原