基于深度学习的国债利率期限结构的预测与分析  

Prediction and Analysis of the Term Structure of Government Bond Interest Rate Based on Deep Learning

作　　者：宋辉鹏 张金良[1] 

机构地区：[1]河南科技大学数学与统计学院,河南 洛阳

出　　处：《应用数学进展》2025年第2期448-462,共15页Advances in Applied Mathematics

基　　金：国家自然科学基金(51675161)资助项目。

摘　　要：利率期限结构及相关问题一直是金融学的研究热点。文章借助于动态Svensson模型、深度学习,构建了国债利率期限结构的深度学习预测模型。选取2012年1月~2024年5月银行间零息国债收益率数据,对其进行了实证分析,并与DS模型、加入宏观经济变量的DS模型、不加入宏观经济变量的深度学习模型的预测结果进行了对比,结果显示文章提出的深度学习预测模型效果显著提升,鲁棒性更好。The term structure of interest rates and related issues has always been a research hotspot in finance. With the help of the dynamic Svensson model and deep learning, this paper constructs a deep learning prediction model for the term structure of treasury bond interest rates. This paper selects the yield data of inter-bank zero-interest treasury bond bonds from January 2012 to April 2024 to conduct an empirical analysis. The results show that compared with the prediction results of the DS model, the DS model with macroeconomic variables, and the deep learning model without macroeconomic variables, the deep learning prediction model proposed in this paper has significantly improved its effectiveness and has better robustness.

关 键 词：利率期限结构 动态Svensson模型 深度学习 鲁棒性 

分 类 号：F83[经济管理—金融学]