变点CAViaR市场风险测量模型及创业板应用

The change point CAViaR model for measuring market risk with an application to Growth Enterprise Market

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.

This paper proposed a new conditional autoregressive value at risk by regression quantiles model with structural change points.Bayesian inference and Markov chain Monte Carlo simulation was applied to estimate parameters based on asymmetric Laplace distribution.Comparing the IG and SAV models both with no,one or two change points as well as statistically testing the forecasting effect of value at risk,it found that SAV model with one change point was the optimal one to forecast the market risk of China＇s Growth Enterprise Market,and the estimated change point in one-change-point SAV model was close to the first unlocking day of restricted shares.It verified that the dramatically reduction of original share and the senor executives＇ dismission drove by the first unlocking of large restricted shares were two main causes to result in structural change of risk evolution patterns,and the average VaRs and lagged returns＇ impact to the current VaR after the change point were both weaker than the ones before the change point.Therefore,it is necessary to strengthen the supervision of executives＇ reduced holding in order to protect small investors＇ interests.