Research On Credit Card Customer Classification And Application Based On DEA Model

With the popularity of credit cards in our country recently, the Customer Asset Management of credit cards has become the focus of banks and experts in the field.Therefore, scientific management of customer assets is a top priority of the credit card business. Identifying high quality customers and bad customers is also particularly critical for the Customer Asset Management. This work not only affects the profits of the business, but also have a closer relationship with the bank’s risk management. China’s banks have no scientific and unified customer segmentation system so far.They use a single factor, such as capital stock or profit contribution to identify valuable customer However, high returns are often accompanied by high risk,if the customer risks can not be measured and controlled, even high value will become meaningless. In the view of this, only from the perspective of the customer profits to classify the customer is not complete, so the customers risks and profits should be taken into consideration together during the customer classification. Based on our banks’ needs of customer classification theory and practice of credit card customers, this paper proposes a new classification method that is based on a two-dimensional perspective of customer risk and profits, which has theoretical and practical significance.This paper classifies the credit cards’ customer using the advanced DEA model instead of the credit score table which has the shortage of subjectivity and information lag in actual work. The general idea of this paper is as follows:firstly, analyze the purpose and feasibility of applying DEA models in our paper,which is based on the detailed description of basic idea and model principle. And then determine the two input variables(customer credit risk and loss risk) and two output variables(merchants commission income and interest-fee income) respectively. Next, in the fourth chapter empirical research part, obtain the required data through questionnaires that take CCB XX branch as the survey sample. According to the basic information in questionnaires use the logistic model to predict the probability of default to reflect the credit risk, at the same time,invite the bank’s internal assessment experts to get the loss probability score and predict customer loss risk. In the last of empirical research part,run the DEAP2.1 software to calculate each customer’s RAR score(it reflects the customer risk-profits relative efficiency). Thus we can identify the optimal customer, quality customer and bad customer depending on theirs RAR scores from highest to lowest. Most importantly, we can get the information of each customer’s target input and output when each customer reaches the highest efficiency in accordance with the software output result. Finally, in the fifth chapter, the author proposes some advice on our customer classification method applying to customer assets accounting and credit card marketing.In this paper, we confirm two types of credit card customers’ risk and apply it to the DEA model. And we assort customers effectively based on a two-dimensional perspective of customer risk and profits. Empirical research shows that the proposed method has achieved satisfactory results in aspects of appropriateness, effectiveness and practicality. The researched method can be practically applied to the credit card customer classification work and customer asset management. The author hopes that scholars can be promoted to improve and perfect customer classification method through this research, and laying a good foundation for the scientific customer asset management.