论文部分内容阅读
随着信贷规模的扩张,企业信用风险评估成为银行重点关注的问题。本文首先梳理企业信用风险评估文献,然后引入非财务指标,对现有指标体系进行完善;在此基础上,利用随机森林方法建立企业信用风险评估模型;并从指标类型和评估方法两个角度对所建模型进行评价。结果表明:盈利能力、偿债能力以及管理层激励对企业信用风险影响较大;改进后的指标体系能显著提高模型预测准确率;随机森林的预测性能优于CART决策树。
With the expansion of credit scale, corporate credit risk assessment has become a key issue for banks. This paper first sorts out the corporate credit risk assessment literature, and then introduces the non-financial indicators to improve the existing indicator system. On this basis, we use the random forest method to establish the enterprise credit risk assessment model; and from the point of view of the types of indicators and evaluation methods The model is evaluated. The results show that profitability, solvency and management incentives have a great impact on the credit risk of enterprises. The improved index system can significantly improve the accuracy of model prediction. The performance of random forest is better than that of CART decision tree.