摘要
经典资本资产定价模型假设资产收益服从正态分布或者投资者效用函数是二次型,然而实际中资产收益的分布呈现尖峰厚尾特征,因而无法在均值-方差框架下分析风险与收益的关系。与资产收益正常变化相比,极端变化带来的尾风险会对市场风险和投资者的决策行为产生更为显著的影响,因此研究尾风险的度量方法和定价能力具有一定理论和实践意义。依据极值理论,基于Fama-French的3因子模型无法解释的收益残差构建尾风险度量模型。选取中国股票市场沪深A股作为研究对象,以历史滑动窗估计尾风险,投资组合分析和横截面回归分析的结果都表明尾风险可以稳定地负向预测股票收益,在控制规模、交易量、非流动性、短期反转、中期动量、特质波动率、特质偏度等股票特征后,尾风险的负向预测能力依然显著。引入融资融券交易,发现融资融券标的股票的尾风险显著低于非融资融券标的股票的尾风险,而且可以消除尾风险的负向溢价,因而中国资本市场尚未完全开放融资融券交易机制可能是产生尾风险负向溢价的主要原因。通过买多-卖空组合构建尾风险定价因子,发现尾风险因子比投资因子和盈利因子有更高的风险价格,而且并不是相对于Fama-French的5因子的冗余因子。与Fama-French的3因子模型和5因子模型相比,Fama-French的3因子或5因子基础上包含尾风险因子的定价模型对股票横截面收益有更好的解释能力,因而有更高的定价效率。2008年全球金融危机爆发后,对尾风险定价能力的忽略使已有资产定价模型受到广泛批评。因此,探究尾风险的度量方法以及分析尾风险的定价能力,不仅有助于完善和拓展资产定价理论,而且有助于降低投资者对股票价格的认知偏差,提升市场资源配置效率和维护市场稳定性。
The classical capital asset pricing model assumes that asset return obeys a normal distribution or the utility function of investors is the quadratic form, but the distribution of actual asset returns shows leptokurtic and thick tail features. Therefore, it is unreasonable to analyze the relationship between risk and return in the framework of mean-variance. Compared with the normal varies in asset returns, the tail risk brought by extreme varies in asset returns has more significant impacts on the market risk and investor decision-making behavior. Thus, in this paper we study the measurement and pricing ability of the tail risk. Based on the Extreme Value Theory, a tail risk measurement model is constructed from the return residuals that cannot be explained by the Fama-French three factor model. A shares of Shanghai and Shenzhen stock markets are selected as the research object, and tail risk is estimated using a roiling window. Portfolio analyses and cross-sectional regression analyses indicate that tail risk can persistently predict stock returns negatively. The predictability are still significant to controls for stock characteristics such as size, trading volume, illiquidity, short-term reversals, medium-term momentum, idiosyncratic volatility, idiosyncratic skewness, and so on. By introducing margin trading, it is found that the tail risk of margin trading stock is significantly lower than the tail risk of non margin trading stock, and margin trading can eliminate the negative premium of tail risk, thus the incomplete open margin trading mechanism in Chinese capital market may be the main reason for the negative premium of tail risk. By constructing the tail risk pricing factor by long-short portfolios, it is found that tail risk factor has higher risk price than profitability and investment factors, and is not a redundant factor relative to Fama-French five factors. Compared with Fama- French three factor model and Fama-French five factor model, the pricing models including FF3 factors or FF5 factors and additional tail risk factors have stronger explanatory capability on cross-sectional stock returns. After the outbreak of the global financial crisis in 2005, the neglect of the tail risk pricing ability has made the existing asset pricing model widely criticized. Therefore, analyzing the tail risk measurement methods and exploring the tail risk pricing ability in this paper not only help improve and expand the asset pricing theory, but also help reduce the cognitive biases of investors on the stock price, and improve the efficiency of market resource allocation and maintain market stabihty.
出处
《管理科学》
CSSCI
北大核心
2017年第6期65-78,共14页
Journal of Management Science
基金
教育部人文社会科学青年项目(16YJC790113)
山西省哲学社会科学研究项目(2017203)~~
关键词
尾风险
预期收益
定价效率
极值理论
融资融券
tail risk
expected return
pricing efficiency
extreme value theory
margin trading
