現金卡違約風險評估因素之研究-以T銀行為例

A Study on the Evaluation Factors of Cash Card Default Risk-An Example if the T Bank in Taiwan

廖元堂(Liao Yuan-Tang)

指導教授 ： 劉欽泉

亞洲大學/管理學院/經營管理學系碩士班/碩士(2004年)

摘要

本研究利用交叉分析與羅吉斯迴歸分析，就現金卡申請人信用額度核定書中之各項信用評等因素，探討各項因素與持卡人逾期違約之間的關係。
本研究發現信用卡持有張數、年收入、貸款餘額等變數均對正常卡／不良卡具有顯著性之差異，其中年收入越高、貸款餘額越多的持卡人其違約風險越低。信用卡張數越多的持卡人，其違約的可能性越高，顯示持卡人若信用卡張數越多張，越容易擴張信用，會成為現金卡違約的可能性越高。該銀行於信用額度核定書中之評等因素未將信用卡張數列入考量，值得該銀行注意。
本研究發現不良卡中的73.2％，核貸款額度介於1∼15萬元間，於該區間的族群其信用評等較低違約比率卻高於其他等級，建議該銀行對得分在70分以下的族群，額度的核定應特別審慎，以做好風險預警。
擔保與信用貸款金額越高得分越低的持卡人其違約風險越低，擔保貸款越多表示申請人的不動產越有價值，得分卻越低，因此貸款餘額越多其現金卡所能申請到的額度越少，申請人比較不可能為了現金卡小額度違約而影響到其他擔保貸款的信用，宜反向思考將擔保貸款與信用貸款餘額分開計分並將僅有擔保貸款餘額分數提高，如此較能反應申請人的資力狀況。
以核貸金額戶數隨機抽取樣本所作外部測試羅吉斯迴歸模式的預測效率，結果發現整體與樣本的外部測試均具有非常相近的預測效率。

關鍵字

現金卡 ； 現金卡信用額度核定書 ； 羅吉斯迴歸模式

並列摘要

The main aim of this Research is to explore each factor concerning credit evaluations for the credit line of applicants’ in approved letters as well as its relation with reneging behind time made by the person who holds the card, using cross analysis and logistic regression analysis.
According to the research, it can be found that variables such as the amount of credit cards a person has, his year incomes, balance of loan, possess apparent differences for normal cards and bad cards. Among them, a person holding cards who has higher year incomes and loan balances has lower risks of reneging. The one with more amounts of credit cards has higher possibility of reneging. This means that the more credit cards a person has, the higher possibility of cash card’s reneging he has due to easily expanding his credit. The fact is worth noticing for the bank that the amount of credit cards weren’t listed in the approved letter of credit line when this bank made its evaluation factors.
In the research, we can learn that the approved loan lines of 73.2％ of bad cards are the range from 10,000 to 150,000, among which the one is with a lower lever of personal credit evaluation but the rate of reneging is higher than other levels. We suggest that the bank should be particularly cautious when approving loan line for persons with the grade under 70 to ensure good early warning of risks.
A person holding cards with higher amount of guarantee and fiduciary loan and with lower evaluation has a lower reneging risk. The more guarantee loan he has, the more valuable his real estate is. But in this case, this kind of applicant has lower evaluation grade. Therefore, the more his loan balance is, the less the loan line can be applied by his cash card. The applicant won’t make harm on the credit of other guarantee loans for reneging small line of cash cards. In reverse, by separately computing the grade of guarantee loan and credit loan balance and raising the grade of loan balance with guarantee, the applicant’s money ability can be truly shown.
To make an external test of the forecast efficiency of logistic regression by taking random sampling of the account amount of approved loan line, its result is both the whole and external tests of the sample have extraordinarily close forecast efficiency.

並列關鍵字

Cash card ； approved letters of cash cards’ loan line ； logistic regression

參考文獻

14.張文生，「銀行建構信用卡信用風險即時預警系統之研究」，私立中原大學企業管理研究所碩士論文，2000年。

1.Cronan,T. P.,& Glorfeld, L. W.(1991) Production system development for expert systems using a recursive partitioning induction approach : An application to mortgage, Commercial and consumer lending, Decision science, 22, 812-845。

2.Duchessi, P. (1988). A knowledge-engineered system for commercial loan decision, Financial management, 17(3), 57-65

4.Klayman, J. (1988). Cue discovery in probalilistic environments: Uncertainty and experimentation, Learning memory and cognition, 14(2), 317-330

5.Noel, C. (1982), Credit Scoring System: A Critical Analysis, Journal of Marketing, 46,：82-91