Planning Of Corporate Bond Default Risk Identification Program Incorporating ESG Information

In recent years,as China’s capital market is one of the most important way of direct financing,bond financing scale increased dramatically,has far more than the same period of direct financing,and in March 2014,"super November,debt" appear substantial default break after the "rigid redemption" China’s bond market,bond default events emerge in endlessly,future bond defaults possible development trend of the "normal".As a large number of default bonds,corporate bonds are representative to analyze the specific reasons for default.The purpose of this paper is to design a default risk identification scheme for corporate bonds and to provide an early warning system for bond investors.In observing default industry distribution of corporate bonds,found that high pollution,high energy consumption industries accounted for is higher,and with the state for this type of enterprises tend to be more strict supervision on the environmental impact of certainly will affect the enterprise management,therefore,this paper will reflect the corporate environmental responsibility,social responsibility,and internal management ability of ESG index and the influence of the traditional corporate debt default index,construct outside the company(macro),internal factor,ESG,bond related factors(financial)index system of two levels,a total of 14 variables,the qualitative analysis is the formation of a more perfect theoretical logic system;Second,under the condition of considering the data availability,this paper based on(t-1)data to predict,select the 35 only corporate debt default of 2014-2019 and 175 normal corporate bonds as building corporate bond default risk identification model of samples,aiming at the balance of the samples,this article uses ADASYN algorithm optimize the training set,improve the identification ability of model samples for breach of contract.In this paper,Catboost,single machine learning algorithm(SVM,BP)and traditional integrated learning algorithm(GBDT,Adaboost)were used to design the default risk identification scheme for corporate bonds.The rationality of Catboost as the main algorithm was verified by comparing the evaluation indexes of different schemes.Then,by comparing the results of three types of 10 models with or without ESG indexes on the test set,this paper determines the role of ESG information in identifying the default risk of corporate bonds.Finally,the paper analyzes the practical approach of the corporate bond default risk identification scheme based on Catboost algorithm.Based on the above analysis,this paper draws the following conclusions :(1)ESG information has a significant effect on the identification of default risk of corporate bonds;(2)the model based on Catboost algorithm has a better effect on the identification of default risk of corporate bonds compared with the model based on single machine learning algorithm and the model based on traditional integrated learning algorithm;(3)the corporate bond default risk identification scheme based on Catboost algorithm can help investors to screen investment targets and establish a bond default warning system.