A Research On The Credit Risk Measurement Of China's Listed Companies

As China is further opening its capital market in recent years,the continuously expanded listed company has become a key important component of the whole national economy.Meanwhile,the opening of market economy and the deepening of market diversity,the expanding listed company is facing numerous problems,among which the credit risk issue is of great importance and difficulty that blocks listed companies' operation and especially the outbreak of financial crisis makes the problem exposed by credit risk get more and more prominent.The credit quality of listed companies directly determines the development of company itself,the stability of capital market and even the sound development of the country's whole economy.So,it is of great practical significance to construct a set of scientific methods of credit risk management,which contributes to the credit risk measurement of listed companies,to accurately identify and better control risks.In the researches for credit risk of existing listed companies,specialists and scholars have constructed a series of credit risk measurement models.On one hand,the model based on distance-to-default is capable of a better dynamic and responsive measurement of credit risk of listed companies.On the other hand,the model based on default probability is mainly capable of the measurement of credit risk of listed companies by means of analyzing static information.The researches show that both models have a good effect.However,there is few existing research that evaluates the credit risk of listed companies by analyzing information from both dynamic and static perspectives.Yet,most domestic researches are preferable to analyze information from the perspective of default probability,which only mirrors the static credit quality of listed companies.As a result,the researches could not accurately and comprehensively reflect credit quality of listed companies.For this reason,this thesis,taking the existing 128 listed companies of free exchange of A-share in China as the main objects of study,analyzes Logistic Model and KMV Model based on default probability and distance-to-default respectively in order to accurately and comprehensively reflect the credit risk of listed companies.In Logistic Model,the first part filters the financial indicators through normality test,T-test,nonparametric test,and multicollinearity test,and then includes the filtered indicators in the model construction;the second part analyzes the model test results 1-3 years previous to the credit crisis respectively,and then discusses the prominent indicators making companies fall into crises;the third part employs bipartition confusion matrix and the tests sample to examine the prediction effect of constructed model.In KMV Model,after determining relative parameters,the first part does empirical research on traditional KMV Model and makes analysis of it,and then improves traditional KMV Model for the purpose of comparison of models' measurement capabilities before and after the improvement;the second part further confirm an optimal DP through resetting the default point(DP)by discussing the DD's distinguishing ability of sample companies under the 10 set DPs and by using mean difference comparison and T test of every single sample,and then,applies the optimized DP to the improved KMV Model prediction through bipartition confusion matrix;the final part makes comprehensive comparisons of performances between improved KMV and traditional KMV Models,and of performances between improved KMV and Logistic Model through G-mean,F-measure of credit risk sample and Area Under the Curve(AUC)in order to select the optimal measurement model and applies it to the risk control practice of listed companies in the end.Empirical researches show:(1)the application of Logistic Model in China's listed credit risk measurement is effective and stable.Asset-liability ratio has significant distinguishing ability before the outbreak of credit crisis;in other words,credit risk of companies is correlative with profitability/earning capacity.(2)Selecting the optimal DP through the calculation of the formula DP=STD+LTD: when long-term liabilities coefficient is 1,the distance-to-default between 2 kinds of sample companies gets distinctive differences.At this time,the improved KMV Model based on the optimal DP is the most effective in the measurement of credit risk of listed companies.(3)The unanimous prediction result of the Logistic and improved KMV Models: The model 1 year previous to crisis is the most sensitive and accurate to the reflection of happening of risk crisis.The model 2 years previous to crisis takes the second.And those 3 years previous to crisis is the most insensitive,and gets highest wrong decision ratio.In other words,the closer to the happening time of credit risk crisis,the more effective prediction the model will get.(4)Compared to the traditional KMV Model,the improved KMV Model gets more effective measurement,which shows that the improvement is feasible.(5)In the comparison study of models reliability,the value of Both G-mean and F-measure of credit risk sample in the improved KMV Model are bigger than those in Logistic Model;it shows that the improved KMV Model has a stronger overall prediction capability of 2 kinds of sample and a stronger sample sorting capability than Logistic Model.The AUC value of improved KMV Model is bigger than that of Logistic Model;it shows that the prediction accuracy of improved KMV Model is higher than that of Logistic Model.To sum up,for the credit risk measurement of listed companies in China,the improved KMV Model based on distance-to-default is much more reliable than the Logistic Model based on default probability.