两因子CIR模型对上交所利率期限结构的实证研究

以上交所债券价格隐含的利率期限结构数据作为分析对象，利用推广的卡尔曼滤波法，实证分析了连续时间的两因子CIR模型，发现估计出的两因子CIR模型能够反映实际观测到的利率期限结构的形状，模型下的利率期限结构与实际观测到的利率期限结构形状基本相同．但估计出的两因子CIR模型对利率期限结构的预测误差具有一定的序列相关性，说明估计出的两因子CIR模型没有充分反映债券回报率和利率期限结构的可预测性．实证表明估计出的CIR模型可以用于上交所债券的定价，但用于利率期限结构变化的预测会产生一定的系统偏差．

Modeling term-structure of yields in SSE with two-factor CIR model

With the monthly data of term structures in the Shanghai stock exchange(SSE) from January 1997 to April 2002,and making use of Kalman filter and maximum likelihood estimation approaches,continuous-time two-factor CIR model is estimated.Empirical study indicates that the two-factor CIR model can model the relative changes of the term structures very well.But it can not fully reflect the predictability of the bond excess returns.The predicting errors of the model for yield curves have obvious serial correlation.It is concluded that the estimated CIR model can be used to price bonds in the SSE,but it will cause some error when it is used to predict the change of yield curves.