【作者】 王九龙；

【导师】 谭静；

【作者基本信息】 四川农业大学 ， 农业经济管理， 2010， 硕士

【摘要】 消费信贷是指为了生活消费的需要,金融或商业机构以货物、货币等形式向有一定支付能力的消费者提供信用,消费者在将来的某个时期偿还的一种信贷行为和信用体系。我国消费信贷起步于20世纪80年代中期,但真正迅速发展却是从20世纪90年代末期开始,1997年底,全国个人消费信贷规模仅为172亿元,到2008年底,个人消费信贷增至3.7万亿元,比1997年增长217倍；个人消费贷款余额占人民币各项贷款总额的比例也由1997年的0.2%上升到2008年的12.28%。个人消费信贷规模的逐年增大,产生了对个人信用风险的控制需求。然而,受制于我国长期以来以企业信用制度建设为重点,缺乏专门的机构对个人信用记录数据收集等一系列因素的影响,我国商业银行信用风险管理存在着贷款发放时信息不对称现象严重、个人信用评估科学方法缺乏等问题,与快速增长的个人消费信贷不相适应。国外经验表明,个人信用评分模型作为一种有效的个人信用风险评估管理工具,在商业银行的消费信贷风险管理中起着非常重要的作用。在目前个人信用数据不完备、信用技术不完善、专业信用人员缺乏的条件下借鉴国外先进经验,充分利用商业银行现有信息系统中的数据和信息,建立一个具有一定预测能力、在信贷决策中具有一定参考价值、符合我国国情的个人信用风险评估模型是很有实践意义的。本文以H农村商业银行2008年3月26日至2009年10月13日审批通过的1356笔个人消费贷款客户信息为样本,尝试应用统计方法中生存分析、Logistic回归及判别分析等方法来构建个人信用风险评估模型,并通过模型间变量作用的差别比较找出影响个人消费信贷风险的主要因素,为H银行个人贷款风险管理提供相应的可行建议。本文共分五部分。首先,本文对选题的背景和意义作了简要陈述,同时鉴于本文研究的重点是通过生存分析、Logistic回归及判别分析等方法的应用来构建个人信用风险评估模型,所以本文在第一部分中重点对目前国际比较流行的信用评分模型建模方法、优劣势以及在我国信用评估中的应用等进行了详细阐述,为下阶段评分模型的建立提供参考和思路。然后,本文就消费信贷、信用风险、个人信用评估等基本概念、内涵及定位作了界定。第三部分,本文接着对文章中要用到的χ2检验、单变量生存函数、Cox回归、Logistic回归及判别分析等方法及建模过程作了详细研究和分析。最后,本文通过对H农村商业银行的实证分析,得出了在H农村商业银行现有信贷系统数据中,学历、贷款方式、年龄这三个变量对客户类型的识别具有重要意义,并认为对优良客户和淘汰客户的识别方式上,应尽可能采用两种以上的信用评分模型进行综合识别,即采用Logistic回归模型和Fisher判别分析,做到最大限度的提高信贷风险的识别能力。更多还原

【Abstract】 Consumer credit is a credit conduct and system that for the need of living, financial or commercial organizations provide consumers of the capacity to pay with some consumer credit in the forms of goods, money and other forms, and the consumers pay back a certain period in the future. Consumer Credit in China started in the mid 1980s of the 20th century, but its rapid development began in the late 1990s of the 20th century. By the end of 1997, national personal consumption is only 17.2 billion credit scale, and to the end of 2008,consumer credit had increased to 37 000 billion, an increase of 217 times over 1997;personal consumer loans accounted for RMB loans ratio 0.2% in 1997 rose to 12.28% in 2008.The scale of consumer credit increased year by year, resulting in the needs of the individual credit risk management. However, because of China has long been emphases on the enterprise credit system construction, the lack of specialized agencies to the credit history data collection and a series of other factors, there are many problems in the China’s commercial bank credit risk management, that are serious information asymmetry during the loans are provide and the lack of scientific methods of individuals credit evaluation and other issues, which is incompatible with the rapid growth of consumer credit. Overseas experience shows that personal credit score model as an effective management tool for individual credit risk assessment, has played a very important role in consumer credit risk management of the commercial banks.Under the current conditions of the imperfect personal credit data, imperfect credit technology and the lack of professional staff it has a very practical sense that making full use of existing data and information of information systems in commercial banks and establishing a personal credit risk assessment model which has certain predictability and some reference value in the credit decision-making in line with our national conditions.In this paper, with the sample of 1356 personal consumer loans document information approved by H Rural Commercial Bank from March 26,2008 to October 13,2009,the author has tried to build personal credit risk assessment model by the statistical methods of survival analysis, Logistic regression and discriminant analysis and find out the main factors affecting consumer credit risk through the model variables comparison in order to put forward practical recommendations about making the appropriate credit decision criteria. This article is divided into five parts.Firstly, it has the brief statement on the research background and significance. In view of the emphases on building individual credit risk assessment model through survival analysis, Logistic regression and discriminant analysis, the first part of the paper has the focus on the currently popular method of credit scoring modeling, the advantages and disadvantages as well as the application in our credit evaluation for considering the score model establishment of the next stage. Then, this paper has defined the basic concepts, content and location of consumer credit, credit risk, credit scoring and other concepts.In the PartⅢthis paper has the research and analysis in detail about the methods of the survival function of a single variable, Cox regression, Logistic regression and discriminant analysis and the modeling which are used in the article in the test. Finally, by the empirical analysis of H Rural Commercial Bank, the paper has showed that the three variables of education, loans modes and the age is important to the customer type recognition in the existing credit systems of H Rural Commercial Bank and proposed the solution.更多还原