尾部风险、投资者情绪与横截面股票预期收益率——来自我国 A股市场的证据
摘
要
在传统金融理论中，如果资产附带高风险，那么这个资产是具有较高收益的，所以
许多学者基于证券收益率的风险进行分析，从而实现对资产定价的目的。从马科维茨的
均值方差理论到如今的 Fama-French因子理论，这些理论都在资产定价的研究上发挥着
重要作用。然而，传统金融理论的研究是以股票市场的收益率服从正态分布为前提条件
的，但是后来的学者对股票市场收益分布进行检验，发现股票市场存在“厚尾”现象。
早在 1988年，Rietz就说明了这个问题，并表示这种“厚尾”现象是不能够被忽视的，
之后他也发现这种尾部现象可以有效解释金融学中的“风险溢价之谜”。而后 Barro对
近 100年的美国宏观数据分析，发现一旦外界发生小概率事件，股票市场收益率会随着
发生强烈的变化。因为股票市场表现的这一特点，所以人们通过市场的小概率区间的收
益率来测度这种外界极端事件的冲击，后来人们称之为尾部风险。之后，不管是国内还
是国外，学术界对于尾部风险的研究越来越集中，并且都得出了相似的结论，即当尾部
风险越大时股票预期收益率也越高。虽然这些文献都有类似的结论，但没有解释尾部风
险为什么能够对资本市场定价。而张一等（2017）分析了投资者情绪与股票预期收益率
的问题，这为本文指明了研究方向。他们在理论上认为投资者情绪对于尾部风险的影响
是显著的，因为股市的尾部风险代理变量本身是一种极端收益率，含有股票收益率的信
息，而同时投资者情绪高低会引起市场波动，这说明两者具有很强的关联性。所以，本
文就大胆猜测投资者情绪能够解释尾部风险溢价的问题，并基于投资者情绪的角度对尾
部风险与股票预期收益率的问题进行研究。而在投资者情绪的问题上，本文发现国内外
绝大部分的研究都是基于市场层面，构建的指标也是市场投资者情绪，在个股层面上研
究投资者情绪的问题较少。因此，本文构建了个股投资者情绪，并从个股微观角度到市
场层面来全面研究投资者情绪与尾部风险溢价的问题。
为了能够更好的探究投资者情绪与尾部风险溢价的问题，本文先是寻找个股尾部风
险的代理变量，参考 Kelly和 Jiang的方法构建市场尾部风险，再通过个股收益率与市
场尾部风险的滚动回归结果得到因子载荷系数，并将该系数作为反映个股尾部风险的代
理变量。之后，本文选取较为常规的资产定价解释变量，通过变量排序法与皮尔森相关
性检验初步验证代理变量的合理性。随后本文参考 Fama-French方法对代理变量进行排
序分析验证其独立定价能力，并进一步通过 Fama-Macbeth横截面回归方法验证本文个
股尾部风险代理变量的实证效果。紧接着，在找到构建个股投资者情绪的指标后，本文
通过将这些指标进行主成分分析，并由此构建了个股投资者情绪，根据结果将个股投资
者情绪高低分成五组，并相应地对各组进行面板回归，初步分析在不同的个股投资者情
绪的资产组合下个股尾部风险溢价的效果。同时，本文为了更进一步地研究个股投资者
情绪对个股尾部风险溢价能力的影响效果，参考 Jeffrey的方法对尾部风险分解，分解出
I

摘要
被个股投资者情绪解释的个股尾部风险和其余部分，根据这两组数据再一次对股票预期
收益率进行 Fama-Macbeth横截面回归，以此来证明个股尾部风险溢价与个股投资者情
绪有关。最后，本文转向市场的角度，进一步分析这个问题，根据市场投资者情绪将市
场尾部风险分解，将分解后的含有市场情绪的市场尾部风险与个股收益率进行滚动回归，
得到的因子载荷系数作为含有市场情绪的个股尾部风险，并通过 Fama-French方法变量
排序以及 Fama-Macbeth横截面回归的方法来验证市场投资者情绪影响尾部风险溢价的
能力。同时，为了进一步分析市场投资者情绪能否影响市场尾部风险的预测能力，本文
也将分解前后的市场尾部风险进行了时间序列回归，对比两者实证结果，以此来说明本
文的结论。
经过上述的实证分析，本文根据这些结果归纳出四个主要结论。第一，依据个股收
益率与市场尾部风险滚动回归得到的风险载荷系数作为个股尾部风险的代理变量是合
理的，并且具有独立的定价能力。第二，尾部风险作为左端收益风险的代理变量，对股
票市场具有稳健的正向预测作用。不管是在个股层面上利用 Fama-Macbeth横截面检验，
还是在市场层面上通过时间序列回归检验，本文都得出了一致的结论。同时在本文增加
了市场主流的风险控制变量后，虽然尾部风险的参数在减少，但是显著性及其相关性没
有变化。第三，本文对组建的个股投资者情绪的代理变量进行了验证，并充分证明了数
据的合理性，且这些变量无论是对个股尾部风险还是市场尾部风险的分解，都起到了重
要作用。第四，本文证明了投资者情绪影响尾部风险溢价的问题。这种影响不论是基于
个股层面还是市场层面，都是成立的。本文根据投资者情绪的构建指标分解了尾部风险，
根据实证结果，发现被个股投资者情绪解释的那一部分个股尾部风险溢价能力显著高于
不被解释的部分，同时，被市场投资者情绪解释的那一部分尾部风险的预测能力显著高
于其余部分，这充分说明了投资者情绪在尾部风险与股票预期收益率的问题存在着重要
影响。
关键词：尾部风险；资产定价；个股投资者情绪；尾部风险溢价
II

尾部风险、投资者情绪与横截面股票预期收益率——来自我国 A股市场的证据
Abstract
Traditional financial theory discovers the characteristics of high risk and high return in
capital markets, so many scholars analyze the risk of securities yield based on it to achieve the
purpose of asset pricing. With the birth of Markowitz's theory of mean variance, and now
Fama-French's factor theory, these advances have advanced the innovation of asset pricing
theory. However, the study of traditional financial theory is premised on the stock market's
yield obeying the normal distribution, but after later scholars' research on the stock market, it
was found that there is a "thick tail" phenomenon in the stock market. As early as 1988, Rietz
explained the problem and said that this thick tail phenomenon could not be ignored, and he
later observed that such tail events could effectively explain the "risk premium mystery" in
finance. Then the scholar Barro analyzed the macro data of the United States for nearly 100
years and found that once a small probability event occurs in the outside world, the stock
market yield will change strongly with it. Because of this nature of the stock market, people
measure the impact of this extreme event of the outside world through the yield of the
market's small probability range, which was later called tail risk. Since then, academic
research on tail risk, both internally and externally, has become more and more concentrated,
and they have all come to a similar conclusion, that is, the higher the expected return on
stocks when the tail risk is greater. While the literature has similar conclusions, there is no
explanation as to why tail risk can be priced in capital markets. In his 2017 thesis, Zhang Yi et
al. analyzed the problems of investor sentiment and expected return on stocks, and gave the
research direction of this thesis an idea. They theoretically believe that the impact of investor
sentiment on tail risk is significant, because the tail risk proxy variable of the stock market
itself is an extreme yield, containing information on the stock yield, while investor sentiment
will cause market volatility, which shows that the two have a strong correlation. Therefore,
this article will boldly guess that investor sentiment can explain the problem of tail risk
premium, and explain the problem of tail risk and the expected return of stocks based on the
perspective of investor sentiment. On the issue of investor sentiment, we found that most of
the research at home and abroad is at the market level, and the indicators constructed are also
market investor sentiment, and there are fewer problems in studying investor sentiment at the
individual stock level. Therefore, this article will construct individual stock investor sentiment
and comprehensively study the problem of investor sentiment and tail risk premium from the
micro perspective of individual stocks to the market level.
In order to better explore the problem of investor sentiment and tail risk premium, this
thesis first looks for the proxy variable of individual stock tail risk, refers to Kelly and Jiang's
method to construct market tail risk, and then obtains the factor load coefficient through the
rolling regression results of individual stock yield and market tail risk, and uses this
coefficient as a proxy variable reflecting the tail risk of individual stocks. After that, the
III
。。。以下略