【作者】 何锋；

【导师】 朱静；

【作者基本信息】 贵州大学 ， 金融（专业学位）， 2022， 硕士

【摘要】 在经济全球化的加速背景下,我国的商业银行都在实施国际化策略,通过发展个人信贷业务来提高自身的整体竞争力。同时,随着我国房地产业的快速发展和居民消费理念的变化,银行个人信贷业务得到了快速发展。但是我国尚未形成统一和公认的个人贷款违约风险评估体系,大多数银行仅依靠央行征信系统的信用记录和银行自身已有的客户记录来了解客户征信情况,网贷和民间贷款与银行信用信息不共享,银行为全面了解客户信用情况就只能额外展开尽职调查,大大增加人力、物力和时间成本。此外,自2019年底新冠肺炎疫情的出现以来,全球经济遭受重创,不少行业工人面临下岗和收入大幅下降,增加商业银行不良贷款率的风险。有鉴于此,本文在研读相关文献后以A银行GY分行为例,研究该银行个人贷款业务违约风险评估情况,通过对个人贷款业务违约评估现状进行分析,发现该分行存在风险评估模型预测精度差,评估流程问题多发,管理体制不完善等情况导致违约风险评估效果差。因此,A银行GY分行亟需完善个人贷款业务违约风险评估模型,提高风险管理人员风险意识,以及改进管理体制,设立奖惩机制。随后,本文针对现行逻辑回归模型精度低的问题研究并开发出泛化能力较好的梯度提升决策树模型,并分别采用逻辑回归和梯度提升决策树模型对A银行GY分行的个人贷款数据进行违约风险评估预测,预测结果显示梯度提升决策树模型预测精度更高,能精准预测出91%的违约客户,相较于之前的评估模型提高了12%。对违约贷款的区分度和泛化能力都更好,更适用于A银行GY分行个人贷款业务违约风险的评估。最后,文章进行了总结并提出构建合适的个人贷款违约风险评估体系、引进高金融素养人才、加大培训力度,完善个人贷款违约风险管理制度的建议。更多还原

【Abstract】 Under the background of the acceleration of economic globalization,China’s commercial banks are implementing an internationalization strategy to improve their overall competitiveness by developing personal credit business.At the same time,with the rapid development of China’s real estate industry and the changes in residents’ consumption concepts,the bank’s personal credit business has developed rapidly.However,China has not yet formed a unified and recognized personal default risk assessment system,most banks only rely on the credit records of the central bank credit system and the bank’s own existing customer records to understand the customer credit situation,online loans and private loans and bank credit information are not shared,banks in order to fully understand the customer credit situation can only carry out additional due diligence,greatly increasing manpower,material resources and time costs.In addition,since the emergence of the new crown pneumonia epidemic at the end of 2019,the global economy has been hit hard,and many workers in industries have faced layoffs and sharp declines in incomes,increasing the risk of non-performing loan ratios of commercial banks.In view of this,after reading the relevant literature,this paper takes the GY branch of Bank A as an example to study the default risk assessment of the bank’s personal loan business,and through the analysis of the current situation of the default assessment of the personal loan business,it is found that the branch has poor prediction accuracy of the risk assessment model,the management system is not perfect,etc.,resulting in poor default risk assessment effect,and many problems in the assessment process.In view of this,Bank A GY Branch urgently needs to improve the default risk assessment model of personal loan business,improve the risk awareness of risk managers,improve the management system,and set up a reward and punishment mechanism.Subsequently,this paper studies and develops a gradient elevated decision tree model with good generalization ability for the problem of low accuracy of the current logistic regression model,and uses logistic regression and gradient escalation decision tree model to conduct default risk assessment prediction on the personal loan data of Bank A GY Branch,and the prediction results show that the gradient enhancement decision tree model has higher prediction accuracy and can accurately predict 91% of defaulting customers,which is 12% higher than the previous evaluation model.The differentiation and generalization ability of defaulted loans are better,and it is more suitable for the assessment of default risk of personal loan business of Bank A GY Branch.Finally,the article summarizes and puts forward suggestions such as building a suitable personal default risk assessment system,improving incentive mechanisms and sharing credit information.更多还原