摘要
无套利Nelson-Siegel模型形式上具有Nelson-Siegel模型的简约性,本质上是满足无套利假设的仿射类动态模型。本文以Fama-Bliss方法获得的上交所国债利率期限结构为研究对象,利用卡尔曼滤波法方法实证分析了无套利Nelson-Siegel(AFNS)模型在中国国债市场的适用性。研究发现估计出的AFNS模型能够很好反映我国国债市场利率期限结构的动态特征,模型中的三个状态因子能更有效地描述中国国债收益率的水平、斜率和曲率因子的动态变化,而收益率调整项的非线性作用使得该模型能有效减少对中国国债长期收益率的拟合误差。和动态Nelson-Siegel模型的对比分析表明,相关因子假设的无套利Nelson-Siegel模型样本内拟合能力更佳,独立因子假设的无套利Nelson-Siegel模型则具有最优的样本外预测能力。
Arbitrage-free Nelson-Siegel model has the simplicity form of the Nelson-Siegel model, while in essence it is one of the affine dynamic term structure model. In this paper, the dynamic behavior of Chi- nese interest rate term structure is strudied using this model and it is found that its three state factors have the strong explanatory power on the level, slope and curvature factors of bond yield curve. Compared with dynamic Nelson-Siegel model, the Arbitrage-flee Nelson-Siegel model is shown to has the excellent out-of- sample forecasting ability.
出处
《中国管理科学》
CSSCI
北大核心
2012年第6期18-27,共10页
Chinese Journal of Management Science
基金
国家自然科学基金资助项目(71001069)
关键词
国债利率期限结构
状态因子
拟合
预测
interest rate term structure
state factor
fitting
forecast
