基于DEA模型的信用卡客户分类方法及应用研究

发布时间：2018-03-18 07:23

本文选题：信用卡　切入点：客户分类　出处：《重庆交通大学》2015年硕士论文　论文类型：学位论文

【摘要】：近年来,随着信用卡在国内的普及,信用卡的客户资产管理问题成为各大银行和专家学者关注的重点领域。因此进行科学的客户资产管理是信用卡业务的重中之重,而识别出优质客户与不良客户对客户资产管理又尤为关键。该项工作不仅影响着业务的利润情况,更与银行的经营风险关系密切。目前国内银行仍没有科学统一的客户分类体系,鉴别价值客户仅通过资金存量或利润贡献等单一因素进行筛选。然而高收益往往伴随着高风险,如果客户风险无法度量并加以控制,即使高价值也变得毫无意义可言。由此看来,仅仅从客户收益角度来分类客户是不全面的,所以在客户分类中风险与收益须同时考虑,缺一不可。基于目前银行界对信用卡客户分类理论及实践的需要,本文提出的基于客户风险与收益二维角度的分类方法是具有理论和实践意义的。本文运用了先进的DEA模型替代实际工作中主观性较强、具有信息滞后性等缺点的信用评分表法对信用卡客户进行分类。大体研究思路如下:首先在详细介绍了DEA模型的思想及模型原理的基础上,分析了本文运用DEA模型的研究目的及可行性分析。然后分别确定了本文DEA模型中的两个输入变量(客户信用风险和流失风险)和两个输出变量(特约商户佣金收入和利费收入)。随后在第四章实证研究部分以建行XX分行为样本通过问卷调查的方式获取了相关数据。根据样本的客户基本信息运用logistic模型预测出违约概率来反映信用风险,邀请了银行内部评估专家根据问卷中的客户流失倾向信息预测出流失概率来反映流失风险。在实证部分的最后运行了DEAP2.1软件计算出每个客户的RAR分数(反映客户风险-收益的相对效率),依据RAR分数由高到低识别出了最优客户、优质客户和不良客户,并且依据软件输出结果可得到每个客户达到效率最大时的目标输入及输出值。最后在第五章笔者提出了本文客户分类方法在客户资产会计核算和信用卡营销两个领域的应用与建议。本文确认了信用卡客户两种风险类型并将其应用到DEA模型中,基于客户风险和收益二维角度对信用卡客户进行了有效的分类,经实证研究表明本文提出的客户分类方法在恰当性、有效性以及实用性方面均取得了满意的结果。本文提出的客户分类方法可以较为实际地运用到信用卡客户分类工作及客户资产管理中,希望通过此研究能够推动学者们不断改进和完善客户分类方法,为科学的客户资产管理工作奠定良好的基础。
[Abstract]:In recent years, with the popularity of credit card in China, the problem of customer asset management of credit card has become the focus of attention of various banks and experts, so carrying out scientific customer asset management is the most important part of credit card business. Identifying good and bad customers is particularly critical to customer asset management. This work not only affects the profitability of the business, At present, domestic banks still do not have a scientific and unified customer classification system, and the value customers are screened only by a single factor, such as capital stock or profit contribution. However, high returns are often accompanied by high risk. If customer risk cannot be measured and controlled, even high value becomes meaningless. Therefore, it is not comprehensive to classify customers solely from the perspective of customer income, so risk and profit should be considered simultaneously in customer classification. On the basis of the current needs of the banking sector on the theory and practice of credit card customer classification, The classification method proposed in this paper based on the two-dimensional angle of customer risk and profit is of theoretical and practical significance. In this paper, the advanced DEA model is used to replace the subjectivity in practical work. Credit rating table method, which has the disadvantage of information lag, classifies credit card customers. The main research ideas are as follows: firstly, the idea and principle of DEA model are introduced in detail. This paper analyzes the purpose and feasibility of using DEA model in this paper, and then determines two input variables (customer credit risk and loss risk) and two output variables (commission collection) in this DEA model. In chapter 4th, the empirical study takes the XX branch of CCB as a sample to obtain the relevant data through questionnaire survey. According to the basic information of the customer of the sample, the probability of default is predicted by using logistic model to reflect the credit risk. The bank internal evaluation experts were invited to predict the loss probability according to the customer turnover tendency information in the questionnaire to reflect the loss risk. At the end of the empirical section, the DEAP2.1 software was run to calculate the RAR score of each customer (reflecting the customer). The relative efficiency of risk-to-profit, according to the RAR score from high to low to identify the best customer, High quality customers and bad customers, And according to the result of software output, we can get the target input and output value when each customer reaches the maximum efficiency. Finally, in chapter 5th, the author puts forward the method of customer classification in customer asset accounting and credit card marketing. This paper identifies two types of risk for credit card customers and applies them to the DEA model. Based on the two-dimensional perspective of customer risk and income, this paper makes an effective classification of credit card customers. The empirical research shows that the proposed customer classification method is appropriate. Satisfactory results have been obtained in terms of effectiveness and practicability. The customer classification method proposed in this paper can be applied to credit card customer classification and customer asset management. It is hoped that this research can promote the scholars to improve and perfect the customer classification method and lay a good foundation for the scientific customer asset management.
【学位授予单位】：重庆交通大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：F832.2

【相似文献】