：以往在个人住房贷款的信贷风险评估过程中，银行业务人员依 据借款人本身的资信状况及考虑房产价值为基础，再依据业务人员本 身的经验加以分析来确定是否发放贷款。在此过程中，经常容易会因 为人为疏忽、判断错误、同业竞争及内部人员舞弊等因素造成贷款质 量低下，违约风险增加，如何降低外在的人为因素干扰，并建立一套 标准的信贷风险评估模型，用以降低借款人的违约风险，是银行进行 风险控制重要课题。 为能解决上述问题，本文以C农商行作为研究对象，抽取了389个 2007年至2016年发放的个人住宅抵押贷款案例作为研究样本，其中正 常还款的客户346个、违约客户43个，并选取借款人的个人属性、贷款 属性、房产属性、区域经济属性、银行往来关系属性等5个主要因素， 21个指标变量，构建了信贷风险评价指标体系，并使用K-S检验遴选出 了10个显著的解释变量，用以建立信贷风险评估模型，并使用Logistic 回归模型及判别分析模型进行实证研究，最后利用ROC曲线分析对两模 型的效力进行比较。 实证结果发现，使用Logistic回归模型有4个变量与违约风险显著 相关，收入还贷比、房龄、还款方式与违约风险呈现正相关，而业务来 往银行数量与违约风险呈现负相关。使用判别分析模型有5个变量与违 约风险显著相关，进入模型的变量比Logistic回归模型多一个风险因 子，即业务来往银行数量，其余相同；同时，收入还贷比的系数绝对值 是最大的，其次是业务来往银行数量、还款方式、房龄，与本行来往业 务量的系数为负且绝对值最小。ROC曲线分析结果表明，Logistic回归 模型和判别分析模型的预测准确性都在90%以上，但前者的预测能力要 比后者高。 II 因此，C农商行防控个人住房风险，可以通过控制关键的风险因素 降低信贷风险，如降低收入还贷比，限制高房龄房产贷款等；同时，构 建科学的风险评价体系，建立信贷客户信息数据库、完善风险评价指 标体系、建立信贷风险评价模型；另外，加强银行内部控制，如严格内 控制度，加强操作风险控制，健全培训体制等。 关键词：风险评价；个人住房贷款；农商行；Logistic回归模型； 判别分析模型 III STUDY ON PERSONAL MORTGAGED LOAN RISK EVALUATION OF C RURAL COMMERCIAL BANK Xin Zhang (MBA) Directed by Zhenhua Lei Abstract：In the past, in the personal credit risk assessment of housing loans, the bank staff based on the borrower's credit status and consider the value of real estate, based on the analysis of the business staff to determine whether to grant loans. In the process, it is often easy to reduce the quality of loans due to human negligence, misjudgment, peer competition and internal fraud, increase the default risk, reduce external interference and establish a set of standard credit risk assessment Model to reduce the default risk of the borrower is an important issue for bank risk control. In order to solve the above problems, this paper takes C-firm as the research object, and draws a sample of 389 cases of individual-home mortgages issued from 2007 to 2016 as the research sample, 346 of whom are regular repayments and 43 are default clients We choose five main factors and 21 index variables, including personal attributes, loan attributes, real estate attributes, regional economic attributes, and bank relationship attributes of the borrower. We construct a credit risk assessment index system and use KS test to select 10 significant Of the explanatory variables used to establish a credit risk assessment model, and the use of Logistic regression model and discriminant analysis model empirical research, and finally the use of ROC curve analysis of the effectiveness of the two models are compared. The empirical results show that there are four variables associated with the risk of default significantly related to the use of Logistic regression model, income repayment ratio, age, repayment method and the risk of IV default is positively correlated, while the number of business transactions and the risk of default is negatively correlated. There are five variables in the discriminant analysis model which are significantly correlated with the default risk. The variables entering the model have one more risk factor than the Logistic regression model, namely, the number of the banks in service and the rest are the same; meanwhile, the absolute value of the coefficient of the loan- The second is the number of business transactions between banks, repayment methods, building age, and the bank's business from the negative coefficient of the negative and the absolute minimum. The results of ROC curve analysis show that the predictive accuracy of Logistic regression model and discriminant analysis model are both above 90%, but the prediction ability of the former is higher than the latter. Therefore, C-firm can reduce credit risk by controlling key risk factors, such as reducing income-to-loan ratio and restricting housing loans for the elderly. At the same time, it should build a scientific risk evaluation system, establish credit customer information database, improve risk evaluation index system and establish credit risk. In addition, the internal control of banks should be strengthened, such as strict internal control system, operational risk control, and training system. Keyword:Risk Evaluation;Personal Mortgaged Loan; Rural Commercial Bank; Logistic Regression Model; Discriminant Analysis Model 目 录 摘要 .................................................... Ⅰ Abstract ................................................ Ⅲ 第1章 绪论 .............................................. 1 1.1 研究背景及意义 .................................... 1 1.1.1 研究背景 ..................................... 1 1.1.2 研究意义 ..................................... 2 1.2 文献综述 .......................................... 3 1.2.1 关于商业银行个人住房贷款违约风险相关研究 ..... 3 1.2.2 商业银行个人住房贷款风险评价指标 ............. 5 1.2.3 商业银行个人住房贷款风险评价方法 ............. 8 1.2.4 商业银行个人住房贷款风险防范策略 ............ 10 1.2.5 研究述评 .................................... 11 1.3 研究内容和方法 ................................... 11 1.3.1 研究内容 .................................... 11 1.3.2 评价方法 .................................... 13 第2章 相关理论 ......................................... 15 2.1 个人住房贷款及其风险概述 ......................... 15 2.2 风险评价方法 ..................................... 15 2.2.1 风险评价方法种类 ............................ 15 2.2.2 本文使用的风险评价方法 ...................... 16 第3章 C农商行个人住房贷款业务现状 ..................... 21 3.1 C农商行基本情况 ................................. 21 3.2 C农商行个人住房贷款业务发展现状 ................. 21 3.2.1 个人住房贷款余额 ............................ 21 3.2.2 个人住房贷款质量 ............................ 22 3.2.3 个人住房贷款区域分布 ........................ 23 3.2.4 个人住房贷款业务特征 ........................ 23 第4章 C农商行个人住房贷款风险评价指标体系构建 ......... 25 4.1 C农商行的风险识别 ............................... 25 4.1.1 农商行个人住房贷款的风险类型与特征 .......... 25 4.1.2 农商行个人住房贷款的风险成因分析 ............ 27 4.2 指标体系构建原则与方法 ........................... 30 4.2.1 构建原则 .................................... 30 4.2.2 构建方法 .................................... 32 4.3 C农商行个人住房贷款风险评估指标选取 ............. 32 4.4 C农商行个人住房贷款风险评价指标体系 ............. 33 4.4.1 风险评价指标体系 ............................ 33 4.4.2 风险评价指标解析 ............................ 34 4.4.3 风险评价指标说明 ............................ 36 第5章 C农商行个人住房贷款风险评价 ..................... 39 5.1 数据来源与描述统计 ............................... 39 5.1.1 数据来源 .................................... 39 5.1.2 描述性统计分析 .............................. 39 5.2 实证分析 ......................................... 44 5.2.1 K–S检验 ................................... 44 5.2.2 Logistic回归分析 ........................... 45 5.2.3 判别分析 .................................... 50 5.2.4 ROC曲线分析 ................................ 53 5.3 C农商行个人住房贷款风险评价结果分析 ............. 54 5.4 C农商行个人住房贷款风险评价的问题识别 ........... 55 5.4.1 未能有效运用关键风险因素控制风险 ............ 55 5.4.2 缺乏有效的量化风险评价体系 .................. 55 5.4.3 银行违约风险控制体制不完善 .................. 56 第6章 C农商行个人住房风险防控策略 ...