基于GARCH-M模型的中国创业板收益率的实证分析

An Empirical Analysis of Chinese GEM Yields Based on the GARCH M Model

对中国创业板指数收益率进行统计性分析和残差检验,并建立了AR模型,发现残差序列存在明显的ARCH效应,即过程的方差不仅随时间而变化,而且有时变化很激烈;按时间观察会有"波动丛集性"或"波动聚集性";从取值的分布来看有"高峰厚尾"的特征.因此,主要借助能够避免过度拟合并且更加简约、在处理金融数据易变性方面和非对称效应方面具有优势的广义ARCH模型(GARCH)消去ARCH效应、分析每日收益率,并利用GARCH-M模型对我国创业板的收益率与风险进行实证分析.最后利用信息冲击曲线来分析正负消息的冲击力差异,即"杠杆效应".

In this paper,first and foremost,by means of basic statistical analysis and residual test for China＇s GEM board yield,we can find that the ARCH effect of residual error sequence has been obvious after the AR model is set up,namely the variance of the process not only changes with time,but sometimes changes fiercely.There is＂volatility clusters＂according to the observation of time.In view of the distribution of values,there are the characteristics of＂peak thick-tail＂distribution.Therefore,the ARCH effect is removed and the daily yield analyzed by taking advantages of the generalized ARCH model（GARCH）which is more simple and helpful for avoiding excessive proposed merger and has advantages in dealing with financial data variability and asymmetric effect;the GARCH M model is also used to make an empirical analysis of the yield and risk of China＇s GEM index.In the end,this paper analyzes the difference of impact between the positive and negative news,namely＂leverage effect＂by information impact curve.