| 摘要: | 在過去許多信用卡相關文獻探討上，大多使用客戶基本資料如性別、婚姻、教育、居住狀況、現在年齡、年收入、職業類別…等變數再配合聯合徵信中心資料協助發卡機構進行客戶評分審核，確實也發現到許多影響風險甚鉅的顯著因子，但在針對外在環境所影響到的區域性的變數並未列入評估系統，鑑於富風險性的客戶可能形成區域性群聚現象，因此針對區域性變數是否宜列入評分系統中，產生了此次研究的動機。
本次研究資料以國內某銀行已核發的信用卡客戶為研究對象，蒐集到2007年1月份至2009年1月份有動用的客戶並排除資料不全的部分一共100,394筆，其中逾期呆帳客戶24,087筆，與正常使用客戶76,307筆，透過LR模型研究方法將基本資料變數與再加上區域性變數相互比較，最後實證結果發現：選取3項區域性變數「各地區平均每戶人口數」、「各地區平均每人存款餘額」、「各地區人口密度」後，模型更加的完整，評估風險更具準確性，也因此確認區域性變數對於信用卡風險評估深具影響力。此一發現意味著銀行發行信用卡時，應更加注重地區性變數影響風險的層面並更加的要求客戶提供存款或財力證明，如此才能在利差優勢下維持授信品質降低營運風險並穩健的擴大市占率，避免風險控管失靈，遭金融監管機關強制停止發卡的命運。
According to past statistics on credit card related researches and measurement, information of Credit Information Center combines with client’s gender, marriage, education, living standard, age, income, occupation and so on; these results assist credit card agencies to measure customer’s credit. On face, some factors might cause risk. However this result does not include external environmental factors, rich-risk customers might be a cluster and live in the same region; therefore, regional variables will be discussed on following chapters.
The research materials is based on credit card owners from a domestic bank from 2007 to 2009; there are 100,394 records which only gathered from usage of credit cards and completed client profile , non-performing loans overdue customers are 24,087 records, and normal usage of credit card are 76,307 records. LR model calculated basic data and regional variables, and then added three more variables: "the average household population of the region", "regional average deposit balance" and "regional population density "; this combination was more accurate and therefore confirmed that regional variables is the most important element on credit card measurement system. This findings show that banks should request clients to provide deposit statement or financial resource and consider about regional variables for application of credit cards. Otherwise, it could make sure quality on issue credit cards and reduce operational risks, moreover, expend its market and avoid forbiddance of issue credit from government. |
