中国债务抵押债券定价模型及实证分析

Pricing Model of Collateralized Debt Obligation for Chinese Bond Market and its Empirical Test

利用KMV模型估算出各债务人的违约概率,并用Copula函数分别估算出债务人之间的违约相关系数,模拟出各债务人的违约时点,在此基础上对债务抵押债券各系列进行定价;进一步应用Gaussian Copula和Student-t Copula对中国金融市场上发行的信贷抵押债券08招元一期产品各系列价差进行实证研究,结果发现两种Copula方法在估计优先A1级系列与B级系列时出现低估现象,而在估计优先A2级时出现高估;但一般都在5个基点的可接受误差范围之内。这表明该方法可以应用于信贷资产支持证券的定价,并可对高收益级的到期收益率进行估计。

The paper used KMV model to assess default probability of each obligor and their correlation coefficient between them with Copulas,simulating default time,and pricing every tranches of Collateralized debt obligation.Furthermore,this article used Gaussion Copula and t Copula to research a CLO-08 zhaoYuan,which discovered Gaussion Copula underestimated A1 and B tranches,but overestimated A2 tranch.The error was in 5 basic points,which was acceptable.Therefore this method could apply in pricing of CLO,and might be able to estimate the level due returns ratio of senior tranch.