The Analysis And Comparison Of Stochastic Volatility Model Based On Bayesian Method

A lot of scholars study stochastic volatility Models because stochastic volatilityModels can describe the returns’ characteristics of sharp peak and heavy tail,volatility aggregation and long memory quite well. Bayesian method assumes that theparameters are random variables, and combines prior information and sampleinformation to get parameters’ inferences, which is more reasonable than traditionalstatistical method, so this article studys stochastic volatility models by Bayesianmethod. Domestic researches on stochastic volatility models are mainly limited to thestandard stochastic volatility model and some of its expansion models, but this articlestudys the constant elasticity of variance model and raises the heavy-tail constantelasticity of variance model, this article’s work are following:1. Obtaining the theoretical Bayesian inferences of the constant elasticity ofvariance model and heavy-tailed constant elasticity of variance model in this article,which are studied by few scolars, and this models are different from the standardstochastic volatility model and its expansion models which have been gotten theirtheoretical Bayesian inferences by some scholars.2. The empirical analysis of this article are based on sample datas of the CSI300Index closing prices from January5,2009to December31,2012with the Winbugssoftware. The empirical results show that the elasticity coefficients of the constantelasticity of variance model and the heavy-tailed constant elasticity of variance modelare bigger than1, and elasticity coefficient of the heavy-tailed constant elasticity ofvariance model is bigger, that is to say the the heavy-tailed constant elasticity ofvariance model shows the stronger level dependent than the constant elasticity ofvariance model. The heavy-tailed constant elasticity of variance model shows thestronger leverage effect than the leverage effect stochastic volatility model, theheavy-tailed constant elasticity of variance model can describe the returns’characteristic of sharp peak and heavy tail. This article compares the models whichare carred on this article by DIC criterion. The empirical results show that the model’sfit degree of constant elasticity of variance model is the highest, but this model is themost complex, increasing its DIC value, the heavy-tailed constant elasticity ofvariance model is better than the heavy-tailed stochastic volatility models in model’s fit degree and DIC value.