基于GARCH-M模型对中国股票市场风险溢价的研究

发布时间：2018-01-07 21:46

本文关键词：基于GARCH-M模型对中国股票市场风险溢价的研究　出处：《重庆大学》2013年硕士论文　论文类型：学位论文

【摘要】：风险溢价是金融经济学的一个核心概念，对它的有效度量直接影响着资产定价，投资分析和风险管理等金融市场活动。风险溢价的水平可以直观的反映金融市场带给投资者的风险回报，对投资者的投资决策具有指导意义。尤其是在中国，金融市场变化日新月异，各种各样的金融衍生品不断出现，金融市场风险变得越来越难以捉摸，研究风险溢价的水平可以帮助我们更好的了解市场的微观结构以及金融资产的流动性等问题，从宏观上把握市场的走势。众所周知，风险溢价的研究最早起源于资本资产定价理论，，认为风险承担者应该获得相应的风险回报。在对波动率风险溢价进行分析时，GARCH-M模型是常用的工具之一，但它作为参数模型，不可避免地给出一个具体的模型形式并对模型误差做出假设，当这些假设不成立时，统计推断便不再精确，甚至没有实际意义。针对这些缺点，人们将非参数技术引用到时间序列分析中，提出了半参数GARCH-M模型，参数部分可对模型的估计结果进行一定的解释，而非参数部分则弥补了参数模型的缺陷，降低了估计值的偏差。本文提出的半参数非对称GARCH-M模型继承了参数和非参数模型的优点，将波动率方程部分处理成传统的TGARCH过程，而均值方程处理成非参数函数的形式，能更好的拟合波动率与风险溢价之间的关系。文中首先分别采用局部多项式拟合和加权最小二乘法对均值函数和波动率参数进行了估计并证明了其大样本性质，包括渐近正态性和相合性。其次通过模拟实验验证了半参数模型的可行性，并与参数模型比较证明了其优良性。最后，我们将半参数TGARCH-M模型用于上证指数波动率风险溢价的实证研究。结果表明，在MSE和QLIKE准则下，半参数TGARCH-M模型的拟合优度明显高于参数模型。同时，我们利用半参数模型估计结果对波动率与超额收益率之间的关系进行了分析，说明风险溢价的曲线是非线性非单调的，所得结论符合中国股票市场的现状。
[Abstract]:Risk premium is a core concept of financial economics, and its effective measurement directly affects asset pricing. Investment analysis, risk management and other financial market activities. The level of risk premium can directly reflect the return on risk brought by the financial market, which has guiding significance for investors to make investment decisions, especially in China. Financial market changes with each passing day, a variety of financial derivatives continue to appear, financial market risks become more and more elusive. Studying the level of risk premium can help us better understand the microstructure of the market and the liquidity of financial assets, and grasp the trend of the market from the macro perspective. As we all know, the research of risk premium originated from the capital asset pricing theory. GARCH-M model is one of the commonly used tools, but as a parameter model, it inevitably gives a specific model form and makes assumptions about model errors, when these assumptions do not hold true. Statistical inference is no longer accurate, even without practical significance. In view of these shortcomings, non-parametric techniques are applied to time series analysis, and a semi-parametric GARCH-M model is proposed. The parameter part can explain the estimation result of the model to a certain extent, but the non-parametric part can make up the defect of the parameter model and reduce the deviation of the estimated value. The semi-parametric asymmetric GARCH-M model inherits the advantages of both parametric and non-parametric models, and the volatility equation is partially treated as a traditional TGARCH process. The mean equation is treated as a nonparametric function. Firstly, the local polynomial fitting and the weighted least square method are used to estimate the mean function and the volatility parameter, and the properties of large sample are proved. It includes asymptotic normality and consistency. Secondly, the feasibility of the semi-parametric model is verified by simulation experiments, and its superiority is proved by comparison with the parametric model. We apply the semi-parametric TGARCH-M model to the empirical study of the risk premium of Shanghai stock index volatility. The results show that under the MSE and QLIKE criteria. The goodness of fit of semi-parametric TGARCH-M model is obviously higher than that of parametric model. At the same time, we use semi-parametric model to estimate the relationship between volatility and excess return. It shows that the curve of risk premium is nonlinear and nonmonotone, and the conclusion is in line with the present situation of Chinese stock market.
【学位授予单位】：重庆大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：F832.51;F224;O212.1

【参考文献】