Optimization Of Default Warning Model For Corporate Bond Integrating Negative Events

In recent years,more and more enterprises begin to use the issuance of corporate bonds for financing,the issuance of corporate bonds has increased year by year,the issuance scale has doubled and doubled.As an important part of the bond market,corporate bonds involve a wide range of enterprises and investors,and their influence is deep.The stable development of corporate bonds is an important part of maintaining a stable financial order and maintaining the general tone of progress in stability in China.At the same time,the default problem of corporate bonds cannot be ignored.To timely identify the default risk and make a judgment before the maturity of bonds will have certain positive significance for investors to reduce or avoid losses,bond issuers to timely self-examination and self-help,regulatory departments to resolutely strengthen supervision and guidance,and keep the bottom line of no systemic risk.When observing the information related to defaulted corporate bonds,it is found that the number of negative events before the date of default of corporate bonds is significantly larger by issuers,which may directly reflect the sub-health state of the company in advance.Therefore,based on the traditional corporate bond default warning index system,this paper integrates the negative event index to construct the index system of this paper,including 22 variables in four categories of macroeconomic situation,company-level information,bond characteristics and negative events,completing the first optimization of the traditional model in this paper.Secondly,this paper selects 53 defaulted corporate bonds and 167 normal corporate bonds issued by listed companies from 2014 to 2021,a total of 220 bonds,as the samples to build a default warning model of corporate bonds.In the corporate bond default warning model established based on Cat Boost algorithm,a set of optimal parameters are found to participate in the subsequent training and prediction of the model by using grid search and cross-validation methods,so as to complete the second step optimization of the model in this paper.Finally,by comparing the performance of the model in the evaluation system before and after the integration of negative event indicators and the tuning of parameters,the optimization effect of the early warning model is evaluated and the effects of the two optimizations are verified.Through the analysis of the importance of the index,the contribution of the negative event index to the prediction effect of the corporate bond default warning model in this paper and the role of other key indicators in the default warning is clarified.Through theoretical and empirical analysis,this paper draws three conclusions.First,the incorporation of negative events helps to improve the classification ability of corporate bond default warning model.Second,parameter optimization is beneficial to effectively improve the prediction effect of corporate bond default warning model;Third,the financial information of the issuing subject plays an important role in the early warning of corporate bond default.Finally,suggestions are given to investors,issuers and regulators,that is,investors can pay attention to negative events and consider risks comprehensively.Issuers should pay attention to negative events,prudently manage positive response;The regulatory layer should strictly comply with the requirements of event announcement and provide comprehensive information support for early warning.