An Empirical Study On The Modification Of Default Points In KMV Model Based On Particle Swarm Optimization

Since the reform and opening up,China's financial market is expanding rapidly.At the same time,various types of risk events occur frequently.In general,financial risks include credit risk,market risk,operational risk and liquidity risk.Among them,the financial market and the business involved in credit risk are the most extensive,and it is the primary risk that needs to be managed and prevented.At present,there are four mature credit risk measurement models in the international market,namely,Credit Metrics model,Credit risk+ model,Credit Portfolio View model,and KMV model.The KMV model can effectively predict changes in the company's credit status before the company's credit risk or corporate bankruptcy,and it is mainly based on the company's asset value,debt status to predict the probability of default,as the basic data needed for the model is obtainable,we can use it in the China's financial market.However,the ownership structure of listed companies in China differs significantly from that of listed companies in Western countries.For example,the complex structure of stock capital and the excessive proportion of non-tradable shares all require the adjustment of parameters when using the KMV model.The KMV model has a key parameter that is the setting of the default point.In the classical model,it is set to the sum of 1 times the short-term debt and 0.5 times the long-term debt,but the financial market environment,the company's liability capacity,and the risk explosion point are all different from those in the West.It is not appropriate to use(1,0.5)coefficients.Therefore,this paper uses intelligent algorithm particle swarm optimization algorithm,and based on the data of 2996 companies in A-shares,after setting 5 initial particles and looking for 30 iterations,it finds the optimal short-term liability coefficient and optimal long-term liability coefficient in China.It is 3.79 and 1.64.This empirically sets the optimal default point of the KMV model under the Chinese market environment.Subsequently,the Mann-Whitney method and the ROC curve were used for testing and analysis.The test shows that the short-term liability coefficient and long-term liability coefficient(1,0.5)of the original KMV model cannot distinguish between ST shares and non-ST shares,and 3.79 and 1.64 groups.The coefficient can better distinguish between ST shares and non-ST shares,that is,the revised KMV model can better identify the credit risk of listed companies in China.