|
With the rapid growth in Taiwan’s cash card market, banks need to investigate customers’ credit histories before issuing cash cards and approving credit limits. In order for banks to make more precise judgments and better decisions, an automatic and scientific evaluation model about customers’ credit rating should be integrated into a simplified rating process, so that applications for cash cards can be processed quickly and accurately to reduce loss from bad debt. This research focuses on building up an evaluation model comprising risk variables affecting cash cards. A Genetic Algorithm characterized with multi-point searching power is employed to analyze and resolve the issue regarding two-level weight parameters of risk variables of cash card. The subjects studied are customers from a specific bank involved in the cash card business. A total of 1398 subjects were sampled in this study, 1276 of them pay on time, and the others pay after the due date. Through analyzing empirical data, it’s expected to set up an evaluation model comprising two level risk variables affecting cash cards. The higher rating scores you make, the lower risk you have. As the research shows, the first level weight parameters of risk variables affecting cash cards ranked in order of highest to lowest risk are home ownership, education background, annual income, marital status, gender, credit balance, credit records and finally age. The second level variables of low risk are: females, those are earning an annual income between 300,000 to 500,000 NT dollars, holding a high school diploma, divorced or living with a live-in boyfriend or girlfriend, being a homeowner without loans, paying by credit cards or checks for over one year, under three years and without bad credit records, as well as maintaining a credit balance over 1,000,000 NT dollars. Another important finding is that those aged 46 to 55 have the lowest credit risk, and those 55 years and older second in low credit risk. By identifying potential risks and taking preventive measures, the goal to reduce risks for cash cards can be eventually achieved. In addition, appropriate management and oversight in the application process, as well as reviews after the cash cards have been issued and follow-up improvements are also critical steps to carry out such a goal. In conclusion, this research builds up a model combining the statistical method and information technology, offering solutions to enhance efficiency, reduce costs and above all optimize profits.
|