中文摘要：

研究不确定性KMV信用风险测度问题,用差分进化算法（DE）来优化违约点系数,提出一种中国上市公司信用风险测度的不确定性DE-KMV模型.实证结果表明,常用的KMV模型往往低估了中国上市公司的风险值,而不确定性DE-KMV模型在面对中国上市公司各种风险情况下的违约系数值与实际风险很接近,模型通过分位数回归分析,其系数在置信区间内显著性更好.因此,相对于常用的KMV模型,化模型更据灵活性,能提高上市公司信用风险测度的准确性.

英文摘要：

This paper focuses on the measurement of uncertainty KMV credit risk. Using differential evolution algorithm to optimize the coefficient of default point, we propose a uncertainty DE-KMV model to measure the credit risk of public companies. Empirical results show that common KMV model often underestimate the value at risk of Chinese listed companies, but Uncertainty DE-KMV model can get coefficient values very close to the actual risks considering the various risks of of China＇s listed companies. This model passes the quantile regression test, and the coefficient is significantly better in the confidence interval. The evidence above can prove the superiority of the uncertainty DE-KMV model relative to common KMV models.