中国银行间市场国债收益率曲线:基于静态插值模型的估计  

Estimating Yield Curves of China Interbank Treasury Security Market:An Interpolation Approach

作　　者：郭枫 倪婧钰 GUO Feng;NI Jing-yu

机构地区：[1]中央财经大学中国金融发展研究院

出　　处：《中央财经大学学报》2020年第4期75-90,共16页Journal of Central University of Finance & Economics

摘　　要：本文依据多指数衰减插值模型(Multiple Exponential Decay Interpolation Model),对中国银行间市场国债收益率曲线进行拟合。同时将多指数衰减模型与McCulloch三次样条模型和Nelson-Siegel-Svensson模型进行比较。实证结果表明,多指数衰减模型能更好适应中国银行间市场国债收益率曲线的形态特点,拟合优度更高,且兼具模型简约、计算负荷量小等优点。通过对比样本期间拟合价格残差的月加权均方根误差(wRMSE),多指数衰减模型在收益率曲线拟合方面具有相对优势,兼顾有效性、最优性和简约性,同时最大程度克服了过度拟合。此外,多指数衰减模型能够适应波动性较大的极端市场价格数据,在识别债券价格异象方面有一定优势。Based on the multiple exponential decay interpolation model,this paper studies the yield-curve fitting of the China's interbank Treasury security market,and compares the fitting performance of the new model to that of other prevailing interpolation models,such as McCulloch cubic-spline model and Nelson-Siegel-Svensson model.The empirical results suggest that the multiple exponential decay interpolation model is more adaptive to the time-varying shapes of Chinese risk-free yield curve.Not only does it outperform other competing models in terms of weighted rooted mean square error(wRMSE),but it also incurs least over-fitting.Moreover,it has great potential in fitting volatile data during extreme cyclical periods and identifying market price anomalies.

关 键 词：利率期限结构 收益率曲线拟合 插值 国债 残差分析 

分 类 号：F015[经济管理—政治经济学]