基于LS-SVM的小微企业信用评估研究  

Small and Micro Businesses Credit Evaluation Research Based on LSSVM

作　　者：肖斌卿[1,2] 柏巍[1] 姚瑶 李心丹[1] 

机构地区：[1]南京大学工程管理学院,江苏南京210093 [2]纽约大学Stern商学院,纽约10012 [3]江苏中诚信信用管理有限公司评级部,江苏南京210019

出　　处：《审计与经济研究》2016年第6期102-111,共10页Journal of Audit & Economics

基　　金：国家自然科学基金(71271109;71671083;71203144);教育部人文社会科学研究青年项目(13YJC790174)

摘　　要：构建并优化小微企业信用评估技术已经成为商业银行开展小微业务必然选择。基于小微企业内在特征,设计以小微企业现金流信息为违约触发机制的小微企业信用评估指标体系,构建最小二乘支持向量机模型(LSSVM),运用某国有控股银行的小微企业贷款微观数据证实该模型能够相对提高预测精确度和稳定性。应用LSSVM构建的小微企业信用评估指标体系以及评估模型有助于提升银行对小微企业的了解程度和小微业务风险管理能力,减轻信息不对称,在一定程度能够化解供给型信贷配给导致的小微企业融资难问题。Based on the characteristics of small and micro businesses, we combine the theory with commercial bank practice to sort out and analyze the factors influencing the credit situation of small and micro enterprises. The index of small micro enterprise credit evaluation is constructed, which we add the factor of cash flow information. Based on a sample of actual borrowers from a large China commercial bank, the results reveal that LSSVM outperform the other techniques with the measure of ROC. In addition, this paper proves that cash flow index plays an important role in the credit evaluation of the small and micro enterprises through the empirical test. This study provides theoretical support and practical experience for China＇s commercial bank screening small and micro businesses credit evaluation index and building credit evaluation model.

关 键 词：小微企业 信用评估 企业信用评估 现金流信息 LS-SVM 小微金融 商业银行 信用配给 信用评分 

分 类 号：F275[经济管理—企业管理]