针对多变量随机波动模型难以刻画金融时间序列尖峰厚尾特征的问题,构建了贝叶斯多变量厚尾随机波动模型。通过模型的贝叶斯分析,选择参数先验分布,设计基于Gibbs抽样的MCMC算法,据此估计模型参数,解决多变量随机波动模型参数较多难以估计的问题;并利用沪深300股指期货与现货交易数据进行实证分析。研究结果表明:贝叶斯多变量厚尾随机波动模型能更准确地刻画金融市场的波动特征以及金融市场间的波动溢出效应。
To solve the problem that multivariate tailed characteristics of financial time series, this paper stochastic volatility model cannot describe heavy proposes a Bayesian heavy-tailed stochastic volatili ty model. Based on the analysis of model statistic structure and the selection of parameters prior, the pape constructs a Markov Chain Monte Carlo algorithm procedure with Gibbs sampler to estimate parameters avoiding the difficulty of parameter estimation. The suggested approach is applied to analyze the linkag effect between CSI 300 futures market and spot market. The results show that the proposed model de scribes not only volatility character of financial market more accurately, but also volatility spillover effec of the two financial markets.