摘要
世界各国学者分别用不同的统计模型对信用风险进行全行业的实证研究。中国在此方面的研究尚处起步阶段。综合运用多元判别模型、Logistic模型、主成分模型,分不同行业对企业财务危机进行预警研究。比较分析了不同行业预警模型的判别准确率,不同预警技术的判别准确率,多年度预警的可行性,预警模型的稳定性,大类、中类行业预警的通用性等问题。商业银行可以使用这些模型进行信用风险度量和信贷风险预警。
Scholars of the world use different statistical models to conduct the empirical research on credit risk in entire industry respectively. The research in our country is still at the beginning stage. The paper makes use of the multiple discriminate analysis model, Logit model, principal component analysis model, to conduct the early-warning research of the corporations' financial distress by industry. The paper analyzes the discretion accuracy of the early - warning models in different industries, the precision of the different models, the possibility of early - warning in several years, the stability of the early - warning models, the replacement of the models in related industries. These models can be used for credit risk measurement and forecasting in commercial banks.
出处
《统计与信息论坛》
2007年第6期39-44,共6页
Journal of Statistics and Information
关键词
财务危机
多元判别分析
LOGISTIC回归
主成分分析
行业
financial distress
multiple discriminate analysis
Logit regression
principal component analysis
industry