无套利Nelson-Siegel模型在中国国债市场的实证分析

发布时间：2018-02-05 19:57

本文关键词： 国债利率期限结构状态因子拟合预测　出处：《中国管理科学》2012年06期 　论文类型：期刊论文

【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。
[Abstract]:The Nelson-Siegel model with no arbitrage has the simplicity of Nelson-Siegel model in form. In essence, it is an affine dynamic model which satisfies the assumption of no arbitrage. In this paper, the term structure of bond interest rate obtained by Fama-Bliss method is taken as the object of study. An empirical Analysis of Non-arbitrage Nelson-Siegelelus AFNSs using Kalman filter method. The research shows that the estimated AFNS model can well reflect the dynamic characteristics of the term structure of interest rate in China's treasury bond market. The three state factors in the model can more effectively describe the dynamic changes of the level, slope and curvature factor of the Chinese government bond yield. The nonlinear effect of the yield adjustment makes the model can effectively reduce the fitting error of the long-term yield of Chinese government bonds. The comparison analysis with the dynamic Nelson-Siegel model shows that this model can effectively reduce the long-term yield of Chinese government bonds. The fitting ability of the Nelson-Siegel model without arbitrage is better than that of the correlation factor hypothesis. The independent factor-free Nelson-Siegel model has the best ability of extrasample prediction.
【作者单位】：中国金融期货交易所博士后科研工作站;复旦大学博士后流动站;上海理工大学管理学院;
【基金】：国家自然科学基金资助项目(71001069)
【分类号】：F224;F832.51

【参考文献】