论文部分内容阅读
提出了带有结构变点的条件分位数自回归模型,借助于不对称拉普拉斯分布,给出了基于贝叶斯推理和马尔科夫链蒙特卡罗模拟抽样的参数估计方法.通过(无、单、双)变点间接GARCH方程(IG)模型与对称绝对值方程(SAV)模型的比较以及风险值预测效果统计检验,发现单变点SAV模型是预测我国创业板指数市场风险的最优模型,且变点的估计均值紧邻首次限售解禁日期;证实了首次限售股解禁引起的原始股大量减持和频繁的高管离职是市场风险演化模式结构变化的主要成因,且变点后的市场风险平均水平以及滞后消息对风险的冲击强度均低于变点前的水平,因此必须加强对高管减持行为的监管,以保护中小投资者利益.
A conditional quantile autoregressive model with structural change points was proposed. Based on asymmetric Laplacian distribution, a parameter estimation method based on Bayesian inference and Markov chain Monte-Carlo sampling was given. (No, single, double) variable point indirect GARCH equation (IG) model and the symmetrical absolute value equation (SAV) model comparison and the risk value prediction effect statistical test found that single-point SAV model is to predict the market index of China’s GEM The optimal model and the change point of the estimated average value immediately after the date of the first sale of the lifting of the ban date; confirms the lifting of the ban for the first time a large number of shares of the original stock reduction and frequent turnover of senior management is the main cause of market risk evolution model structure changes, and change The average market risk after the point and the impact of the lagged news on the risk are lower than before the change point, so we must strengthen the supervision of senior management to reduce the interests of small and medium-sized investors.