基于贝叶斯分层模型的可违约债券利率期限结构

Term Structure Model for Defaultable Bonds Based on Bayesian Hierarchical Model

本文构建可违约债券利率期限结构的贝叶斯分层模型,通过采用Dirichlet多层先验分布和联合估计思路,获得了不同信用级别下的收益曲线。利用交易所市场的企业债券交易数据对模型的有效性进行了实证检验。检验结果表明,基于Svensson函数的贝叶斯分层模型可以有效的克服小样本问题和异常值的影响,提高了模型拟合精度和样本预测绩效,而且能避免单曲线模型导出的收益率曲线交错的问题,说明贝叶斯分层模型可以有效的拟合交易所企业债券收益率曲线。模型丰富了中国企业债券收益曲线估计方法,对于企业债券的合理定价具有较强的参考价值。

This paper constructs a bayesian hierarchical model based on the Svensson Function. The yield curve based on different credit rates is gained by means of Dirichlet hierarchical prior and joint estimation approach. The empirical examination for proposed model is implemented using the data of exchange market in Chinese defaultable bond market. The empirical examination shows that the proposed model can overcome the question of a small sample, avoid the impact from outliers, and improve efficiently the fitted accuracy and predictive performance. The proposed model can enrich the way of estimating defaultable bonds, and the proposed model has a strong reference value for the reasonable pricing of corporate bonds.