【作者】 王平；

【导师】 杨宝臣；

【作者基本信息】 天津大学 ， 工商管理硕士（专业学位）， 2020， 硕士

【副题名】以H城商行为例

【摘要】 自20世纪90年代以来,城市商业银行开始进行股份制改造,逐步发展壮大。自2008年开始,城市商业银行贷款规模逐年递增。随着贷款规模的增长,城市商业银行的信用风险管理能力存在短板。自2015年以来,以往城市商业银行粗放式的依赖专家经验判断信用风险的模式遇到了挑战,不良贷款率大幅攀升,尤其是中小企业贷款不良率也节节攀升。在这一背景下,对中小企业信用风险计量就成为城商行提升核心竞争力的关键。本文以H城市商业银行为例,在系统梳理总结信用风险计量、中小企业、城市商业银行的国内外相关研究基础之上,同时结合实际情况合理确定研究样本,构建信用风险计量指标体系,综合运用Fisher模型、Logit模型、SVM模型,对批发业、工业中小企业正常与违约状态进行识别。在此基础上,深入剖析批发业、工业企业的实际案例,实现了实践与理论的结合,并有针对性的提出了中小企业信用风险防范策略,具备重要的实践意义,主要研究结果如下:一是合理构建信用风险计量指标体系,能够为模型正确识别奠定重要基础。在财务指标具备重要性、可操作性、科学性的基础之上,优选了衡量企业偿债能力6项指标、盈利能力3项指标、营运能力3项指标等共计12项指标,构建了信用风险计量指标体系。该体系对于批发业、工业在整体样本描述性统计上,呈现出了均值相差较大、波动性较小的优势,为模型准确识别中小企业违约与正常状态奠定基础、具备较强的指导意义。二是批发业和工业中小企业影响模型识别的关键性财务指标存在显著性差异。模型结果显示,影响批发业中小企业识别结果的有六项关键性财务指标(流动比率、经营活动现金流量比率、资产负债率、利息支出保障倍数、毛利润率、存货周转率),影响工业中小企业识别结果的有七项关键性财务指标(流动比率、速动比率、资产负债率、利润获现率、存货周转率、应收账款周转率、固定资产周转率)。虽然两个行业之间存在着共同的三项关键性财务指标(流动比率、资产负债率、存货周转率)影响识别结果,但是该三项关键性财务指标影响方向存在互异性并且仍有其他不同的关键性财务指标影响存在显著性差异。三是批发业和工业中小企业采用Fisher、Logit、SVM模型比较识别效果,SVM模型的识别效果最佳。采用不同模型对不同行业的中小企业进行正常与违约情况识别,同一模型对某行业识别概率高、对另外一行业识别概率却呈现出相反的情况,仅仅使用一个模型对不同行业进行识别,可能无法达到最优识别。从整体而言SVM模型识别准确率要高于Fisher模型、Logit模型。四是通过对批发业和工业的具体中小企业案例分别予以实践,经过模型计量实证对比,信用风险计量模型当中支持向量机(SVM)模型判别准确率最高。通过实证案例也再次印证了支持向量机模型对于批发业、工业中小企业的正常与违约情况识别效果最佳。依据H城商行现有数据样本,支持向量机(SVM)模型是中小企业信用风险计量的最佳模型。五是基于实证研究,H城商行应加快风险管理转型。通过深入调查关键财务指标,为模型准确识别奠定数据基础。引入多模型识别信用风险,转变成以信用风险计量系统为核心的专业化信贷分析中心,实现授信客户精准识别。利用模型推动管理体系改革,将相关部门转变成为信用风险计量系统提供基础数据的服务部门,为模型精准识别风险保驾护航。更多还原

【Abstract】 Since the 1990 s,the urban commercial banks began to make transformation in the shareholding system,after that developed gradually.Since 2008,the proportion of loans has increased year by year for urban commercial banks,but the credit risk management ability still has weak point.Since 2015,the previous mode of judge credit risk which reliance on expert has encountered greet challenges.The non-performing loan rate has risen sharply,especially the SMEs(small and medium-sized enterprises).Therefore,the most import competitiveness for urban commercial banks is the ability of measuring the credit risk of SMEs.In the needs of practice,this thesis takes H Urban Commercial Bank as an example,according to systematically combing and summarizing the domestic and foreign researches on credit risk measurement,SMEs and urban commercial banks,and comprehensively using Fisher model,Logit model,and SVM model to identify the normal and default states of SMEs in wholesale and industry.This thesis applies the models to the actual cases of wholesale and industrial SMEs,and realizes the combination of practice and theory.At last this thesis proposes some prevention strategies for SMEs.According to research,there are few conclusions blow:First,the important basement of correct identification is to establish the credit risk measurement index system.The financial indicators must own the feature of importance,operability and scientific,12 indicators are optimized,including 6indicators to measure corporate solvency,3 indicators to measure profitability and 3indicators to measure operational capability,and a credit risk measurement index system is constructed.The system presents a large difference in mean value and volatility for the whole sample descriptive statistics of wholesale and industry Small advantages lay a foundation for the models to distinguish defaults from normal samples in SMEs.Second,the impact indicators of SMEs in different industries(wholesale and industrial)are different,and there are significant differences in main financial indicators.According to analysis result of the normal and default situations for SMEs in different industries using different model,there are six main financial indicators(current ratio,cash flow ratio of operating activities,asset-liability ratio,interest expense guarantee multiples,gross profit)which affect the identification results of SMEs in the wholesale industry,there are seven main indicators(current ratio,quick ratio,asset-liability ratio,profit realization rate,inventory turnover rate,receivables turnover rate,fixed Asset turnover ratio)which affect the identification results of industrial SMEs.Although there are three common main financial indicators(current ratio,asset-liability ratio,inventory turnover ratio)which affect the identification results between the two industries,but the three key financial indicators have different influence directions and there are still significant differences in the impact of other different key financial indicators.Third,using three different risk measurement models(Fisher,Logit,SVM)to identify the different style of the SMEs in different industries(wholesale and industrial),the SVM model recognizes best.This thesis is using different models to identify the normal and default situations of SMEs in different industries.The same model has a high probability of identifying one industry,but the opposite probability of identifying another industry.Using only one model to identify different industries may not to get the beast recognition,the SVM model has a better performance than the Fisher model and the Logit model.Fourth,through empirical comparison of model measurement on the cases of SMEs in the wholesale and industry,the SVM model has the best recognition.According to actual practices,the SVM model shows the best identity,and the SVM model is the best model for the measurement of the credit risk of SMEs.Based on the existing data of H urban commercial bank,the support vector machine(SVM)model is the best model for measuring the credit risk of SMEs.Fifth,according to the research,H urban commercial bank should take the transformation of risk management as quickly as possible.Through in-depth investigation of main financial indicators,lay the data foundation for accurate identification of the model.The introduction of multiple models to identify credit risk,H urban commercial bank has transformed into a specialized credit analysis center,at last achieve accurate identification of credit customers.Use the model to promote organizational structure reform,transform relevant departments into service departments that provide basic data for credit risk measurement system,and guarantee the accurate risk identification of the model.更多还原