股票风险因子与经济增长的关联性研究

发布时间：2018-02-28 07:13

本文关键词： 股票风险因子经济增长 Fama-French三因素模型 Chen-Zhang三因素模型　出处：《南京财经大学》2014年硕士论文　论文类型：学位论文

【摘要】：股票市场一直被看做经济的晴雨表,股票市场与实体经济的发展息息相关,研究股市收益率与经济增长是一个备受关注的课题,但在我国股票市场与宏观经济走势之间的关系似乎并没有那么清晰。正是基于此激发了本文通过构造资产组合,分解股市收益率来研究各股票风险因子与我国经济增长之间的关系,是否具有“晴雨表”的作用。Fama-French三因素模型被广泛用来解释股市超额回报率,近些年的Chen-Zhang三因素模型也很好地解释了各国股市异象,本文选用上述两个模型中的风险因素做为股市收益率的资产组合,研究各资产组合收益率能否预测我国未来经济增长,且总结两个模型中的风险因素构造了一个五因素模型以增强解释我国宏观经济走势的能力。Fama-French三因素模型中的股票风险因素包括帐市比因素HML、规模因素SMB和市场风险溢价MRF,Chen-Zhang三因素模型中包括投资资本变化率INV、总资产收益率ROA和市场风险溢价MRF,五因素模型中的股票风险因素包括帐市比因素HML、规模因素SMB投资资本变化率INV、总资产收益率ROA和市场风险溢价MRF。本文对三种种资产组合中的风险因素HML、SMB、INV和ROA分别做了描述性统计,统计发现在三种资产组合中HML的收益率均值为正,SMB的收益率均值为负,INV的收益率均值为负,ROA的收益率均值为正,这为实证检验提供了基础。本文通过三种方法来检验了Fama-French三因素模型、Chen-Zhang三因素模型和五因素模型对我国未来经济增长的预测力,结果表明用分布滞后模型和基于滚动数据的方法能较好地解释我国未来经济增长,基于L-V构造模型预测效果不佳。其中在分布滞后模型和基于滚动数据的方法中风险因子HML和ROA的实证检验效果与预期一致,风险因子SMB和1NV的实证检验结果与预期相反。在分布滞后模型中五因素模型并不能很好地解释GDP增长率,各风险因子系数T值很低,通不过显著性检验。但在基于滚动数据的方法中五因素模型相对于Fama-French三因素模型和Chen-Zhang三因素模型有更好的解释力,这可能是因为滚动数据增加了样本期间。在基于L-V构造的模型中无论是Fama-French三因素模型还是Chen-Zhang三因素模型中的单变量回归还是多变量回归,各模型在整体上都不显著,通不过在10%的水平上F检验。
[Abstract]:The stock market has always been regarded as a barometer of the economy. The stock market is closely related to the development of the real economy. It is a subject of great concern to study the stock market yield and economic growth. However, the relationship between the stock market and macroeconomic trends in China does not seem to be so clear. Decomposition of stock market returns to study the relationship between stock risk factors and China's economic growth. Whether the "barometer" function. Fama-French three-factor model is widely used to explain the excess return of stock market. In recent years, the Chen-Zhang three-factor model also explains the anomalies of the stock market in various countries. In this paper, the risk factors of the above two models are chosen as the portfolio of stock market returns to study whether the return rate of each portfolio can predict the future economic growth of our country. This paper summarizes the risk factors in the two models and constructs a five-factor model to enhance the ability to explain the macroeconomic trend of our country. The stock risk factors in the three-factor model of Fama-French include book-market ratio factor, scale factor SMB and market wind. Risk premium MRF / Chen-Zhang three-factor model includes Inv, total return on assets (ROA) and market risk premium (MRF). The equity risk factors in the five-factor model include book-market ratio factor (HMLs), scale factor (SMB) investment capital change rate (Inv), total risk factor (ROA). The return on assets (ROA) and the market risk premium (MRF) are analyzed in this paper. The risk factors HMLSMBINV and ROA are analyzed respectively. The statistical results show that the average return of HML is positive, the average of return is negative, the average of return is negative, the average of return of HML is positive. This provides the basis for the empirical test. This paper tests the forecasting power of Fama-French three-factor model, Chen-Zhang three-factor model and five-factor model, to the future economic growth of our country through three methods. The results show that the distributed lag model and the method based on rolling data can better explain the future economic growth of China. The prediction effect of L-V based model is not good. In the distributed lag model and rolling data based method, the empirical test results of risk factors HML and ROA are consistent with expectations. The empirical results of risk factor SMB and 1NV are contrary to expectations. In the distributed lag model, the five-factor model can not explain the growth rate of GDP well, and the T value of each risk factor coefficient is very low. But in the rolling data based method, the five-factor model has a better explanation than the Fama-French three-factor model and the Chen-Zhang three-factor model. This may be due to the fact that the rolling data increases the sample duration. In the L-V-based model, whether it is univariate regression or multivariate regression in the Fama-French three-factor model or the Chen-Zhang three-factor model, each model is not significant as a whole. It does not pass the F test at the level of 10%.
【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;F124.1

【相似文献】