论文部分内容阅读
针对已有的信用溢价模型只考虑了单一尺度的波动均值回复过程,从而导致的信息损失问题,提出了贝叶斯复合状态信用溢价模型,据此辨析不同尺度的信用溢价波动回复状态。利用不同剩余偿付期的中国企业债信用溢价指数序列,引入了基于混合正态分布的多步MCMC方法对复合状态模型进行贝叶斯分析,研究结果表明:不同剩余偿付期的债券具有不同的异方差水平,均值回复过程可以区分为长期和短期两种趋势,并且分别具有不同的尺度维度;长期回复过程显示了序列的整体波动趋势,短期回复过程更细致地刻画了极值点的影响;与传统模型的比较突出了复合状态模型在拟合效果上的优越性。
In view of the fact that the existing credit premium model considers only the single-scale volatility-averaged response process, the Bayesian composite credit premium model is proposed, and the credit repricing volatility recovery status at different scales is analyzed. Using the sequence of the credit surpluses of Chinese corporate bonds with different residual payoffs, a multi-step MCMC method based on mixed normal distribution is introduced to Bayesian analysis of the composite state model. The results show that bonds with different residual pay-off periods have different differences Variance level and mean recovery process can be divided into two kinds of long-term and short-term trends, and each has different scale dimensions. The long-term recovery process shows the trend of the whole fluctuation of the sequence, and the short-term recovery process describes the impact of the extreme point more elaborately. The comparison of the traditional model highlights the superiority of the composite state model in the fitting effect.