基于Griddy-Gibbs抽样的混合高斯AR-GJR-GARCH模型的贝叶斯估计
Bayesian estimation of the Gaussian mixture AR-GJR-GARCH model with Griddy-Gibbs sampler

综合考虑波动率的尖峰厚尾性、杠杆效应、自回归条件异方差性以及收益率的自回归性等特点，作者提出了混合高斯AR-GJR-GARCH模型，并用基于Griddy-Gibbs抽样的MCMC方法对模型的参数进行了贝叶斯估计, 以新东方的股票市场为例用Matlab和R软件对模型进行了实现与检验. 模型对波动率的各种特性都有一定的体现，并且估计方法的收敛速度较快、自相关性弱、算法复杂度低、稳定性良好.
Considering the characteristics of the volatility such as excess kurtosis and leverage effect, the authors propose a Gaussian mixture AR-GJR-GARCH model. The parameters of the model are estimated by using MCMC method based on Griddy-Gibbs sampler. The model is implemented and tested by Matlab and R software taking EDU stock market as an example. The method has a certain manifestation on the characteristics of the volatility and the method has the good convergence, the weak autocorrelation, the simple algorithm, and the nice stability