利率期限结构的区制转移动态Nelson-Siegel模型  

A Markov Switching Dynamic Nelson-Siegel Model of Term Structure

作　　者：魏立佳[1] 蔡远飞 Wei Lijia;Cai Yuanfei

机构地区：[1]武汉大学经济与管理学院行为科学研究实验中心 [2]武汉大学经济与管理学院

出　　处：《金融学季刊》2018年第3期49-73,共25页Quarterly Journal of Finance

基　　金：国家自然科学基金项目(71403192);教育部社会科学研究项目(14YJC790130)的支持

摘　　要：为了刻画宏观经济周期对国债收益率的影响，本文在动态Nelson-Siegel（NS）模型基础上引入经济扩张、经济收缩两个不同的宏观经济状态，构建马尔科夫区制转移动态利率期限结构模型（MS-DNS模型）。根据卡尔曼滤波的极大似然估计结果，本文对中国国债收益率的动态潜在因子进行估计，并对未来的国债收益率进行向前预测。实证结果发现：MS-DNS利率期限结构模型取得了较好的拟合效果，而预测效果整体上较DNS模型有所改进。此外，MS-DNS模型体现了利率期限结构的非线性特征，区制转移与经济周期之间存在紧密的联系，可较为准确地捕获经济周期的阶段性变化。同时，美国国债收益率的稳健性研究也验证了MS-DNS模型的广泛适用性。To investigate how business cycle affects treasury yields, this paper considers two regimes dynamic model by assuming an economy is in either a boom or a recession. In this paper, we propose a Markov switching dynamic Nelson - Siegel term structure model（ MS-DNS model）. According to the maximum likelihood estimation of Kahnan Filtering, we estimate the dynamic latent factors of the China＇s treasury yields and conduct one- month-ahead forecasting on its treasury yields. The empirical results show that the MS-DNS model has a better goodness-in-fit and a better out- of-sample forecasting in yield curves than the classical dynamic Nelson-Siegel modek In addition, we also find that the MS-DNS model captures the nonlinear dynamics of term structure. Two regimes are closely related to the boom anti recession in the business cycle of China＇s economy. Meanwhile, the robustness analysis on US treasury yields also confirms our MS-DNS model is wildly applicable.

关 键 词：利率期限结构 区制转移 动态Nelson-Siegel模型 宏观经济周期 

分 类 号：F832.51[经济管理—金融学]