I 摘要 近年来，在党和国家的正确引导和政策扶持下，我国的“三农”问题日益缓解， 在扶贫攻坚的道路上，我们取得了巨大的成绩。农村信用社作为金融扶贫的主力 军，是农民金融服务的重要提供者。目前农村信用社针对农民的金融产品主要有 三种：小额农贷、再贷款、金融债。 小额农贷是最早开始的，也是累计发放贷款最多的产品。小额农贷主要是针 对农户发放5万元以下的信用贷款，其初衷就是为贫困农民提供资金帮助，帮助 他们摆脱困境，利用贷款重新开始生产、经营小企业或改善生活等。小额农贷的 申请手续比较简单，农户只需提供个人信息，如收入、住宿、婚姻状况、健康状 况、保险、职业等，信贷员再将收集到的农户信息录入信用评级系统，通过预设 的信用评级模型，得到一个信用评分，作为农户贷款的授信额度及贷款利率的评 价标准。 作为我国“贫中之贫”的三州地区（甘阿凉地区），由于自然条件恶劣、基础设 施差、信息闭塞、历史原因等客观因素影响，多年来都是扶贫攻坚道路上的重中 之重。以C农信为例，截至2019年6月，三州地区小额农贷借款总金额约为39.9 亿元，为当地农户提供了有效的金融扶持，一定程度上满足了三州农户资金需要， 缓解了“三农问题”。但是，根据2019年上半年数据显示，三州地区小额农贷的逾 期率远远高于全省平均水平，很多农户思想上就认为小额农贷是救济款，可还可 不还，在扶贫工作取得巨大成效的同时，也增加了银行的风险。 为降低C农信小额农贷的信贷风险，压降逾期率，本文通过研究甘阿凉三州 地区农户的小额农贷信用评级特征和逾期还款数据，采用SPSS工具构建四种算法 模型，将特征属性（如年龄、收入状况、贷款用途、货币性资产等）作为自变量， 对小额农贷的逾期率进行预测分析。一方面帮助C农信优化和改善现有的信用评 级模型；另一方面帮助C农信缩小贫困地区与全省的逾期率差距，降低信贷管理 风险，在扶持贫困人口脱贫的同时，减少经济损失。结果表明，本文构建的算法 模型对贫困地区小额农贷逾期情况具有较高的预测准确率，可以为后续信用评级 和信贷管理提供技术支撑。 关键词：信用评级，小额农贷，组合模型，贫困地区，信贷特征 ABSTRACT II ABSTRACT In recent years, China's three rural issues (agriculture, countryside and farmers) have been alleviated with the correct guidance and policy support of the Party and the State. We have made great achievements in poverty alleviation. Rural Credit Union is the main force of financial poverty alleviation that is the important provider of financial services for peasantry. Currently, there are three kinds of financial products for farmers — Rural Micro Credit, Reloan and Financial Debt. Rural Micro Credit is used to loaning for rural poverty population. Generally, the credit limit is mostly under 50000 Yuan. The targeted clients are most for poverty people who are living in countryside and their living environments are very severe and harsh. The original purpose of Rural Micro Credit is assisting for poverty group, and helping these people get out from tribulation, they can use the loaned money to restart production, run little business or improve their lives and etc. The application procedures of Rural Micro Credit are fairly easy. Borrowers only provide their personal information, such as income, accommodation, marital status, healthy, insurance, occupation and etc. Loan officers simply check these kinds of information or even not, and then type it into the Client Credit Grading System. Finally, borrowers will get a credit score which comes from system. The credit score help borrowers know how much money they could get from Rural Credit Union. This paper will focus on studying credit rating of Rural Micro Credit in poorer regions.The research data is from three minority areas — Ganzi, Aba and Liangshan, which is located in the eastern mountain areas of Sichuan province. These areas are knowned as one of the poorest areas in China. With C Rural Credit Union as an example, until June 2019, the total amount of Rural Micro Credit was about 3.99 billion yuan in three minority areas that is the strong financial supporting for local farmers. However, according to the data in the first half year of 2019, the overdue rate of Rural Micro Credit in three minority areas is far higher than the average level of the whole province. Many farmers think that Rural Micro Credit is relief fund, which no need to pay back. It certainly has increased the risk of banks. For declining the credit risk of Rural Micro Credit and reducing overdue rate, it builds the classification model by the data preprocessing, variable index, and analyzing ABSTRACT III key factors via SPSS in Chapter five. Overdue rate as a dependent variable, and rural households’ characteristics and features as independent variables such as insurance coverage, gambling and drug use records, family population, professional skills, health status and etc. It will improve the accuracy and focalization of credit rating result though training data model. In China, it is the key to content the financial demands of poverty rural households that helps to resolve three rural issues and promote the development of Inclusive Finance. Therefore, this paper based on the study of the credit rating of farmers in poverty-stricken areas that can help Rural Credit Union to improve the current credit model，and finally achieve the purpose of identifying risks and reducing the default repayment rate. Keywords: credit rating, Rural Micro Credit, hybrid model, poverty rural group, credit characteristics 目 录 IV 目 录 第一章 绪论 .................................................................................................................... 1 1.1 选题背景 ............................................................................................................ 1 1.2 研究目的和意义 ................................................................................................ 2 1.2.1 研究目的 ................................................................................................ 2 1.2.2 研究意义 ................................................................................................ 2 1.3 研究内容及论文框架 ........................................................................................ 3 1.3.1 研究内容 ................................................................................................ 3 1.3.2 论文框架 ................................................................................................ 3 1.4 研究思路与数据来源 ........................................................................................ 4 1.4.1 研究思路 ................................................................................................ 4 1.4.2 数据来源 ................................................................................................ 4 第二章 国内外研究进展及理论基础 ............................................................................ 5 2.1 国内外研究历程及发展现状 ............................................................................ 5 2.1.1 商业银行客户信用评级 ........................................................................ 5 2.1.2 个人信用评级的主要方法 .................................................................... 6 2.1.3 国外信用评级方法的发展 .................................................................... 7 2.2 农户信用评级的研究 ........................................................................................ 9 2.2.1 农户信用评级的发展 ............................................................................ 9 2.2.2 农户信用评级的方法 ............................................................................ 9 2.3 信贷机构与农户间信息不对称理论 .............................................................. 10 2.3.1 不完全市场竞争理论 .......................................................................... 10 2.3.2 信息不对称理论 ................................................................................... 11 2.3.3 交易费用理论 ....................................................................................... 11 2.3.4 普惠金融理论 ...................................................................................... 12 2.4 本周小结 .......................................................................................................... 13 第三章 C农信小额农贷信用评级体系的现状及问题 .....