基于非参数方法的我国资本资产定价模型研究

发布时间：2018-03-26 15:26

本文选题：资产定价模型　切入点：非参数与半参数　出处：《南京财经大学》2014年硕士论文

【摘要】：自从深圳证券市场与上海证券市场成立以来,经过短短二十年的发展,中国资本市场取得了巨大的进步,成为全球第二资本市场。中国金融资本市场对于全球金融市场的影响作用越来越大,中国资本市场开放程度逐渐提高,因此需要建立能准确反映我国证券市场运行的资产定价模型,揭示资本市场风险与收益的关系。文章在借鉴国内外研究成果的基础上,结合我国股市的具体特点,利用非参数和半参数的方法对我国证券市场的资产定价进行研究。论文首先对抽样获得的每只股票均采用非参数光滑样条方法建立单因子资产定价模型并进行估计,分析这种模型能否解释我国资本市场的时变市场风险和非线性资产定价,基于该非参数资本市场特征线(cML)检验β系数的线性和稳定性。并将非参数单因子模型与线性CAPM模型的结果进行比较,找出两个模型拟合效果、β系数、α系数的差异。文章还以同样的思路和方法建立Fama-French三因子的线性模型与非参数、半参数模型,对我国股市进行研究,并将结果与Fama-French的线性三因子模型进行比较。通过实证建模分析,论文得出以下主要结论：1.非线性CAPM模型对于我国证券市场的解释力优于传统的线性CAPM模型,这不仅体现在似然比检验与模型拟合效果比较中,还体现在通过采用非参数方法绘制的特征曲线可以生动反映个股的α值都表现为随着市场收益的变化而变动,这也就证明了α值存在时变性,而传统的线性CAPM模型假定α值是固定的常数,不随市场态势转变而变化是错误的。2.尽管通过Wilcoxon-Mann-Whitney对大中小三个规模组的股票进行检验,不能认为通过线性方法与非参数方法求得的β系数存在差异,但是不论规模大小以及拒绝还是接受线性CAPM模型,非参数方法绘制的个股β值都表现为随着市场收益的变化而变动,线性CAPM模型暗含着β系数是不变常数的假设被否定,即证明了β系数同样具有时变性。3.我国证券市场存在规模效应,对规模效应的解释,可以从各个上市公司业绩和二级市场交易这两个不同的层面去理解。4.账面市值比效应仅仅在小规模(S)组中有所体现；而在中等规模(M)组与大规模(B)组中并没有这种效应。5.非参数三因子模型发现β值、规模因子sMB与账面市值比因子HML的系数同样存在时变性,线性三因子模型并不能准确的解释股票的超额收益,非参数模型的优越性再一次得到体现。6.为了选择最能准确刻画我国证券市场的资产定价模型,分别建立线性三因子模型、非参数三因子模型与各个半参数三因子模型,并对模型进行比较和假设检验,结论发现非参数三因子模型解释力最强。
[Abstract]:Since the establishment of the Shenzhen Securities Market and the Shanghai Securities Market, after just 20 years of development, China's capital market has made great progress. As the second capital market in the world, China's financial capital market is playing a more and more important role in the global financial market, and the opening up of China's capital market is gradually increasing. Therefore, it is necessary to establish an asset pricing model that can accurately reflect the operation of China's securities market, and to reveal the relationship between capital market risks and returns. Using non-parametric and semi-parametric methods to study the asset pricing in China's securities market. Firstly, the paper uses the non-parametric smooth spline method to establish a single-factor asset pricing model and estimates the asset pricing model. Whether this model can explain the time-varying market risk and nonlinear asset pricing in China's capital market is analyzed. Based on the nonparametric capital market characteristic line, we test the linearity and stability of 尾 coefficient, and compare the results of the nonparametric single factor model with the linear CAPM model. To find out the difference of fitting effect between the two models, 尾 coefficient and 伪 coefficient. The paper also establishes the linear model and non-parametric and semi-parametric model of Fama-French three-factor with the same thinking and method, and studies the stock market in our country. By comparing the results with Fama-French 's linear three-factor model, the paper draws the following main conclusions: 1. The nonlinear CAPM model is superior to the traditional linear CAPM model in explaining the stock market in China. This is not only reflected in the comparison between likelihood ratio test and model fitting effect, but also in that the characteristic curve drawn by non-parametric method can vividly reflect that the 伪 value of individual stock changes with the change of market income. This proves that the 伪 value is time-varying, while the traditional linear CAPM model assumes that the 伪 value is a fixed constant and does not change with the change of market situation. The 尾 coefficients obtained by the linear method and the nonparametric method cannot be considered to be different, but regardless of the size and the rejection or acceptance of the linear CAPM model, The 尾 values of individual stocks plotted by the nonparametric method change with the change of market returns. The assumption that the 尾 coefficient is constant is denied in the linear CAPM model. It is proved that the 尾 coefficient is also time-varying. 3. There is a scale effect in the stock market of our country, and the explanation of the scale effect is discussed. It can be understood from the two different levels of performance of listed companies and secondary market trading. The book market value ratio effect is only reflected in the small scale Steam; However, there is no such effect in the medium scale M) group and the large scale B group. The nonparametric three factor model shows 尾 value, and the coefficient of the scale factor sMB and the book market value ratio factor HML is also time-varying. Linear three-factor model can not accurately explain the excess return of stock, and the superiority of non-parametric model is reflected again. In order to select the asset pricing model which can best describe the stock market in China, the linear three-factor model is established separately. The non-parametric three-factor model is compared with each semi-parametric three-factor model. The conclusion is that the non-parametric three-factor model has the strongest explanatory power.
【学位授予单位】：南京财经大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F832.51;F224

【相似文献】