我国上市公司动态违约概率KMV模型改进

A Correction of Chinese Listed Companies' Dynamic Default Probability KMV Model

应用KMV模型检验中国上市公司的信用风险,需要对KMV模型做出修正并在此情况下检验模型的有效性。动态违约点的设定是对KMV模型修正的重要内容。本文通过设置动态违约点,对违约距离DD进行描述性统计分析并应用多种检验方法论证违约点参数设定,构建了大样本下违约距离和经验EDF的映射关系,在此基础上检验KMV模型的区分能力。结论表明,(1)均值差比较,均值t检验和秩和检验的结果在5年内具有一致性的结论,即当违约点在DP1时KMV模型的区分能力最强。(2)违约距离判别精度效果最好,而经验EDF的判别精度相比较低。

To apply KMV model to check the credit risk of Chinese listed companies,this paper makes a correction on the KMV model,whose effectiveness is tested.The set of dynamic default point is very important to the correction on the KMV model.By setting dynamic default point of the distance,making descriptive statistics on DD and using a variety of testing methods to demonstrate default point parameter,the mapping between DD and experience EDF under large samples is built;KMV model's ability of discrimination is tested based on what is mentioned above.Simulation results show that:(1)the result of the mean difference comparison,the mean t test and rank sum test in five years are sharing a consistent conclusion that:if the default point is DP1,the discriminating ability of KMV model is the strongest;(2)DD's discrimination precision is the best,while that of the experience EDF is lower.