【作者】 闫伟；

【导师】 杨春鹏；

【作者基本信息】 华南理工大学 ， 金融工程与经济发展， 2012， 博士

【摘要】 现代金融理论产生后得到了长足的发展，然而，自70年代末以来涌现出大量投资者异常行为及金融市场异象，从而使得现代金融理论的完备体系面临了金融事实的挑战。自此，行为金融学产生并发展开来。对于行为金融领域中的重要问题，即资产定价模型而言，现有文献主要从噪音交易者和投资者认知偏差两个角度进行了研究。然而，基于投资者情绪的研究与基于这两个研究角度的分析相比，具有更大的优势，既能较容易实证检验相关结论，也能得到心理行为实验与神经医学实验的证据支持。因此，从投资者情绪的角度进行行为资产定价模型的研究或许能够得到更多的有益结论。本文即基于投资者情绪理论研究行为资产定价问题，继而深入分析所构建的理论模型，文章主要研究内容有三：一、针对目前投资者情绪指数构建过程中面临的一些不足，本文提出了一套选取投资者情绪代理变量的优化程序。本文提出的优化程序对原始代理变量的选取做了改进，同时对不适合表征情绪的代理变量增加了剔除操作，从而使得构建的投资者情绪指数更加符合理论逻辑。随后，本文应用我国股市实际数据，以EGARCH模型验证了该优化情绪指数的有效性，从而为本文理论模型的研究奠定了坚实的基础。二、基于BSV模型与DSSW模型，本文相继构建了风险资产的情绪认知价格模型和情绪均衡价格模型，并数理推导了其价格解析表达式。首先，针对市场上仅存在风险资产、以及市场情绪投资者同质即均为单向情绪投资者的假设条件下，本文改进了BSV模型的不足之处，建立了单向情绪资产认知价格模型，并通过数理推导求解得到了其价格表达式。随后将模型拓展至市场情绪投资者异质即可分为正向情绪投资者与负向情绪投资者两类时的情况，构建并求解了双向情绪资产认知价格模型。其次，本文考虑了当市场上同时存在风险资产和无风险资产的情况下，投资者在受到情绪影响的条件下如何配置其资产并最终决定风险资产的价格，在改进了DSSW模型的基础上分别构建了单向情绪资产均衡价格及双向情绪资产均衡价格模型，并通过推导求得了两模型的解析表达式。三、针对文中两类理论模型求解得出的解析表达式，本文分析了实际投资策略、资产超额收益、资产情绪泡沫及负的期望收益等问题，并得到了丰富的有益结论。针对两类情绪认知价格模型，特别是单向情绪认知价格模型，设计了一套投资策略，利用本文提出的投资者情绪指数构造优化程序，构建了一类新的市场整体情绪指数，并以上证综指为对象进行了实际数据的验证，发现该投资策略相较于基准指数能够获取较高的超额收益，从而验证了本文所提出投资策略的有效性。随后参数灵敏度分析证明了初始参数的设置并不影响投资策略的盈利率，最终又验证了这一投资策略盈利的稳健性。对于双向情绪认知价格模型，本文给出了数值模拟，并以参数分析验证了众多行为金融实验的结论，即正向情绪投资者与负向情绪投资者权衡博弈时，正向情绪投资者数量的增加将引致市场平均情绪水平高涨，并最终导致资产价格升高。针对两类情绪均衡价格模型，本文分别分析了情绪对风险资产超额收益、泡沫及期望收益的影响。在单向情绪资产均衡价格模型方面，研究结论表明：情绪高涨时资产有正的期望收益，情绪低落时资产有负的期望收益；然而，这一结论对超额收益和泡沫而言并不成立，即当情绪低落时，风险资产确实会有正的超额收益及负的情绪泡沫，但当情绪高涨并大于某一临界值时，即情绪剧烈高涨时，资产具有负的超额收益及正的情绪泡沫，当情绪高涨小于该临界值时，即情绪温和高涨时，资产具有正的超额收益及负的情绪泡沫。这一数理推导观点是对现有研究结论的细化。在双向情绪资产均衡价格模型方面，研究结论表明：负向情绪投资者的悲观气氛强于正向投资者的乐观气氛时，风险资产有超额收益；正向情绪投资者的乐观气氛强于负向情绪投资者的悲观气氛但强度不是很大即小于某一临界值时，风险资产有超额收益；正向情绪投资者的乐观气氛远远强于负向情绪投资者的悲观气氛即大于该临界值时，风险资产没有超额收益。对泡沫的分析与对超额收益的分析均得到了相同的情绪临界值，这一结论的性质与前面的单向情绪资产均衡价格模型分析一致。最后分析了双向情绪水平的变化与投资者数量的变化分别对资产期望收益的影响，结果发现当两类情绪发生同向变化且低落时资产期望收益为负，这一情况意味着市场整体情绪水平的低落，与单向情绪资产价格模型的结论是一致的；当两类情绪发生异向变化时资产期望收益的正负取决于价格模型中其它参数的具体取值情况。而乐观情绪投资者数量的增加并不一定使得风险资产价格必然地上升，该类投资者的过度涌现可能造成资产价格的风险抑制效应增强，从而可能使其具有负的期望收益，这与实际市场中的股票价格行为一致。更多还原

【Abstract】 There were a great many of research achievements after the modern finance theoryestablished. However, lots of abnormalies of investors’ behavior and anomalies of financemarket avalanched since the end of1970s and challeged the mature frame of modern financetheory. Therefore, the behavioral finance appeared and developed. The asset pricing model,which is a kernel problem in behavioral finance, was investigated in two approaches whichare noise trader and investor biases belonged. Nevertheless, the researches based on investorsentiment compared with achievements based on noise trader and investor biases have moreadvantages which some conclusion could be validated empirically and could be supported bymany evidences from behavioral and physiological experiments. Consequently, morebeneficial conclusion may be obtained from the behavioral asset pricing model based oninvestor sentiment. This dissertation intends to investigate the asset pricing model and itsextended problems by using the investor sentiment theory. In principle, there are threecontents of this dissertation:Firstly, this dissertation proposed an optimized procedure for selecting the proxies ofinvestor sentiment, in order to ameliorate some shortcomings during constructing the investorsentiment index in currently literature.The optimized procedure in this dissertation extends the selection for the original proxies,meanwhile eliminates some proxies which are not suitable for describing the sentiment. Thefinal investor sentiment index constructed by this method is more logical. Subsequently, thevalidity for this optimized sentiment index is testified through EGARCH model by adoptingthe real data in Chinese stock market. This optimization research laid a solid foundation forthe theoretical models latterly.Secondly, this dissertation established two sentiment cognitive models and two sentimentequilibrium models for the risky asset based on BSV model and DSSW model, and obtainedthe analytical expressions for the asset price.On one hand, the shortages in BSV model are improved in this dissertation, and aunidirectional sentiment cognitive asset pricing model is proposed based on the consumptionthat there is only one risky asset in the market and the sentiment investors are homogeneous. The analytical expression is obtained subsequently. Furthermore, a bidirectional sentimentcognitive asset pricing model is established and deduced when the consumption is changedthat the sentiment investors are heterogeneous including call sentiment investoes and putsentiment investors. On the other hand, this dissertation investigates how to select portfoliofor investors and ultimately influence the price of risky asset when all of the investors areaffected by sentiment and there is risk-free asset in the market as well as risky asset. Aunidirectional sentiment equilibrium asset pricing model and a bidirectional sentimentequilibrium asset pricing model are constructed respectively, and their analytical expressionsof the price are acquired on the foundation of modifing DSSW model.Thirdly, the investment strategy, exccesive return, sentiment bubbles and negativeexpected return are studied according to the theoretical models proposed in the dissertation,and many profitable conclusions are obtained.An investment strategy is designed according to the sentiment cognitive price model,especially of the unidirectional sentiment cognitive price model. The efficience of thisinvestment strategy is validated by empirical test with the Shanghai composite index and anew market sentiment index constructed by the optimized procedure introduced in thisdissertation. This investment strategy is of strong earning power compared with the Shanghaicomposite index. The solidity of this investment strategy that the setting of initial paremetersin the theoretical model doesn’t affect the investment return is proved subsequently byparameters sensitive analysis. Moreover, the numerical simulation and parameter analysisaccording to the bidirectional sentiment cognitive price model verifies the conclusion of manybehavioral finance experiments. When the call sentiment investors gamble with the put, theincrease of their number will lead the average sentiment level to ascend, and ultimately causethe high price.This dissertation analyzes the affect from investor sentiment on the excessive return,bubble and expected return of risky asset according to the sentiment equilibrium price models.The conclusion of research on unidirectional sentiment asset equilibrium price model showsthat: There is positive expected return for risky asset when the sentiment is high, while isnegative when the sentiment is low. However, this conclusion doesn’t hold for excessivereturn and bubble. There do is positive excessive return and negative bubble when the sentiment is low, but not vise versa. The result shows that there is negative excessive returnand positive bubble when the sentiment is high and higher than a critical value, while there ispositive excessive return and negative bubble when the sentiment is high but lower than thiscritical value. This conclusion conduced logically is a detailed extension for the existingachievements. The conclusion of research on bidirectional sentiment asset equilibrium pricemodel shows that: There is excessive return when the pessimism level of put sentimentinvestor is stronger than the optimism level of call sentiment investor, but not vise versa.There is excessive return when the optimism level of call sentiment investor is stronger thanthe pessimism level of put sentiment investor but is lower than a critical value. However,there is no excessive return when the optimism level of call sentiment investor is muchstronger than the pessimism level of put sentiment investor meanwhile is higher than thiscritical value. The same critical value of sentiment is obtained in the research on bubble. Thisconclusion is consistent with the analysis of unidirectional sentiment asset equilibrium pricemodel. Lastly, the influence on expected return from bidirectional sentiment level andinvestor numbers is investigated. The result implies that: the expected return is negative whenboth of these two sentiment change toward the same direction. This situation means theaverage level of market sentiment is low, and this conclusion is consistent with theunidirectional sentiment asset equilibrium price model. The positive or negative expectedreturn when these two sentiment change toward different directions depends on the values ofother parameters. The increasing number of call sentiment investors does not necessarily raisethe asset price. The nimiety of these investors may enhance the risk of asset price, so theexpected return of risky asset may be negative. This conclusion is consistent with stock pricebehavior in actual stock market.更多还原