Default Probability Prediction Of Credit Bonds Based On Logistic Elastic Net Model

Bond market is an important part of China’s financial market,and its financing scale far exceeds that of stock market.Therefore,whether the bond market can develop healthily is very important to a country’s financial system.As one of the direct financing channels of enterprises,the prosperity of bond market directly affects the expansion and development of enterprise financing channels.At the same time,the stability of the bond market is conducive to preventing the occurrence of financial systemic risks,and is also one of the channels of macroeconomic control,so it has become the focus of regulators.In recent years,bond market defaults occur frequently,while external rating failure,high rating bonds’ default are also common.Therefore,how to accurately quantify the credit risk of bonds and predict the probability of bond default is of great significance.This paper first elaborated the definition of credit debt and credit default,delimited the research scope.Then,the relevant theories of credit risk are sorted out.On this basis,the possible influencing factors of credit default are deeply analyzed,which lays a foundation for the empirical analysis in the following paper.In the empirical research part,all maturing credit bonds(excluding financial bonds and ABS)from 2014 to November 2021 are selected as samples,and the index database is established by using macro data and financial data of bond issuers.Finally,the final model variables are obtained through a series of variable screening methods,variable sorting and coding.Finally,by comparing the prediction results of traditional Logistic model and Logistic Elastic Net model,it is found that the prediction ability of the model is improved in both default group and non-default group after the penalty factor is added.At the same time,the main factors affecting bond default are obtained.