Predicting Credit Bond Default with Deep Learning: Evidence from China

作     者：Ning Zhang Wenhe Li Haoxiang Chen Binshu Jia Pei Deng 

作者机构：School of InformationCentral University of Finance and EconomicsBeijing 100081China SDIC Taikang Trust Co.Ltd.Beijing 100034China 

出 版 物：《Journal of Social Computing》 (社会计算（英文）)

年 卷 期：2024年第5卷第1期

页      面：36-45页

核心收录：

学科分类：12[管理学] 02[经济学] 0202[经济学-应用经济学] 1201[管理学-管理科学与工程(可授管理学、工学学位)] 020204[经济学-金融学（含∶保险学）] 

基　　金：supported in part by the Emerging Interdisciplinary Project of Central University of Finance and Economics Beijing China 

主　　题：credit bond default prediction convolutional neural network imbalanced data processing generative adversarial network 

摘      要：China’s credit bond market has rapidly expanded in recent ***,since 2014,the number of credit bond defaults has been increasing rapidly,posing enormous potential risks to the stability of the financial *** study proposed a deep learning approach to predict credit bond defaults in the Chinese market.A convolutional neural network(CNN)was selected as the classification model and to reduce the extreme imbalance between defaulted and non-defaulted bonds,and a generative adversarial network(GAN)was used as the oversampling *** on 31 financial and 20 non-financial indicators,we collected Wind data on all credit bonds issued and matured or defaulted from 2014 to *** experimental results showed that our GAN+CNN approach had superior predictive performance with an area under the curve(AUC)of 0.9157 and precision of 0.8871 compared to previous research and other commonly used classification models-including the logistic regression,support vector machine,and fully connected neural network models-and oversampling techniques-including the synthetic minority oversampling technique(SMOTE)and Borderline SMOTE *** one-year predictions,indicators of solvency,capital structure,and fundamental properties of bonds are proved to be the most important indicators.