基于多元有序Logit回归模型的商业银行信用风险评级实证研究

发布时间：2018-04-29 04:19

本文选题：经济下行期 + 大企业　；参考：《上海交通大学》2015年硕士论文

【摘要】：信用风险评级模型是《巴塞尔新资本协议》中内部评级法(IRB)的核心内容。银行利用信用风险评级模型对借款人的违约概率实施预测,并将结果运用到风险加权资产的计量和资本监管领域,以更好地抵御风险。自内部评级法在银行业实施以来,尽管世界上爆发过多次债务危机,但我国经济尚未发生过剧烈波动。从2013年开始,受国内国外多重因素的影响,中国经济的复杂程度逐步升高,一些区域出现了明显的风险征兆,某些企业(如钢贸)在银行的违约个案激增。在此情形下,商业银行使用的信用风险评级模型开始出现“超期服役”的现象。所谓“超期”是指商业银行对历史违约数据的模拟停留在3年甚至更早以前,未将经济下行周期内的企业违约风险特征纳入模型的研究和开发。为了改变这一现状,本文对2013年初到2014年末银行企业客户的信用风险新特征进行了重新模拟,对企业信用违约的主要财务因素进行了再挖掘。本文采取实证研究的方法,对2013年初到2014年末银行“正常类”、“违约”客户的财务指标进行因子分析,揭示出银行企业客户在经济下行周期内的信用风险特征,总结出与企业违约关系密切的4项财务因素:资本效率、现金获取能力、经营收益和偿债能力。最终在统计分析的基础上构建了基于多元有序Logit回归模型的信用风险评级模型。模型通过了拟合度检验,总体适配性良好。本文还通过商业银行实际案例,证明了模型具有较好的预测信度和效度。
[Abstract]:Credit risk rating model is the core content of IRB in Basel New Capital Accord. The bank uses the credit risk rating model to predict the borrower's default probability and applies the results to the measurement of risk-weighted assets and the field of capital supervision in order to resist risks better. Since the implementation of the internal rating law in the banking sector, although there have been many debt crises in the world, China's economy has not yet experienced violent fluctuations. Since 2013, China's economy has grown in complexity due to multiple factors at home and abroad, with clear signs of risk in some regions and a surge in bank defaults by some companies, such as steel trading. In this case, the credit risk rating model used by commercial banks began to appear the phenomenon of "extended service". The so-called "extended period" means that the simulation of historical default data of commercial banks stays at three years or even earlier, and the characteristics of enterprise default risk in the economic downward cycle are not included in the research and development of the model. In order to change this situation, this paper resimulates the new characteristics of credit risk of bank enterprise customers from the beginning of 2013 to the end of 2014, and excavates the main financial factors of enterprise credit default. This paper adopts the method of empirical research, carries on factor analysis to the financial index of bank "normal class" and "default" customer from the beginning of 2013 to the end of 2014, and reveals the characteristics of credit risk of bank enterprise customer in the down-cycle of economy. Four financial factors which are closely related to enterprise default are summarized: capital efficiency, cash acquisition ability, operating income and solvency. Finally, a credit risk rating model based on multivariate ordered Logit regression model is constructed on the basis of statistical analysis. The model has passed the test of fitting degree, and the overall adaptability is good. This paper also proves that the model has good predictive reliability and validity through practical cases of commercial banks.
【学位授予单位】：上海交通大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：F832.33

【相似文献】