摘要
正确估计中小企业信用风险对银行规避信用风险具有重要影响。然而由于中小企业信息资料不完善等原因,银行惜贷现象严重。本文利用德国某中小企业信用风险数据库的数据,采用三种生存模型(贝叶斯模型平均生存模型、传统生存模型和拔靴生存模型)来估计中小企业信用违约情况。结果表明,贝叶斯模型平均生存模型的变量选择确定性更强,结果准确性更高,拔靴生存模型次之;并非所有指标均会对企业信用违约产生重要影响,定量指标中企业盈利能力与总资产比率、产权比率、负债率指标影响重大,其中,企业盈利能力与总资产比率影响最大;而定性指标中仅专家对中小企业过去偿还历史的判断影响重大。银行需重点关注的年度为该中小企业贷款后第三年,常会出现不规则还款现象,其次为贷款后第二年。
The estimation of the credit risk of SMEs is important to avoid the credit risk.The dataset of a German SME credit risk database,with three survival models(Bayesian model averaging Survival model,classical Survival model and bootstrapped Survival model) is used to estimate the credit default of SMEs.The results show that the certainty of Bayesian model averaging Survival model is stronger,the result is more accurate,followed by bootstrapped Survival model.And not all indicators have major impact on whether the company will be credit default,the quantitative indicators of result ratio,equity ratio,liabilities ratio are of significant impact,in which the result ratio is the greatest impact ratio;and of qualitative indicators of SMEs only the variable summarizing the payment history of each SME is of great impact.The most important year that SMEs will default the loan is the third year after the loan,followed by the second year after the loan.
出处
《中国管理科学》
CSSCI
北大核心
2012年第S1期327-331,共5页
Chinese Journal of Management Science
基金
山东省自然科学基金资助项目(ZR2010GL011)
关键词
金融学
信用风险
贝叶斯模型平均生存模型
中小企业
后验包含概率
Financial
credit risk
Bayesian model averaging Survival model
small and medium-sized enterprises(SMEs)
posterior included probabilities