我國信用卡循環信用年利率近20%，在金融產業因銀行家數過多，持續過度競爭，各類放款利差皆大幅縮水下，挖掘出潛在可創造高利潤又不會倒帳的邊際持卡人，並發展適當的行銷策略，在微利市場中創造獲利，已成為極迫切的議題。 本研究以國內某發卡銀行信用卡客戶作為研究對象，從2006年12月31日以前發卡，正常使用、未發生逾期且最近一年有使用該銀行信用卡的客戶中隨機取2,340戶為實驗樣本，其中有使用循環信用功能與未使用者各半，利用邏輯斯迴歸(Logistic Regression; LR)統計方法，以個人基本資料以及持卡人在聯合徵信中心(JCIC)各銀行往來紀錄兩構面共十二個變數，建構發卡銀行使用循環信用持卡人特性的實證模式。 結果發現有八項變數與使用循環信用具顯著相關性。包括個人基本資料構面的住宅狀況、職業、年收入，各銀行往來紀錄構面的持有個案銀行信月卡月數、信用卡總應繳金額、在個案銀行以外是否使用循環信用功能、是否有無擔保貸款餘額，以及是否使用預借現金。最後，另隨機抽取598戶為測試樣本，以驗證模型整體之預測力。 The purpose of this study is to find out the characteristics of revolvers, based on the cases of the credit cards issued by a selected bank. Sample source is obtained from the normal accounts with holding at least one active credit card before December 31 in 2006.This study uses random sampling to collect the building data. Total effective sample is 2,340, which includes just half of revolvers and non-revolving samples, respectively. In addition, this paper uses logistic regression method and totally 12 variables in basic information scope of customer and in JCIC scope of cardholders to construct the empirical model. The findings of this study indicate that there are 8 variables showing significant relations with revolvers. The main prediction variables in basic information scope of customer are housing condition, occupation, and annual income group. The significant prediction variables in JCIC scope of cardholders are the time since first issue card, total amount of credit card liabilities in selected bank, and whether or not be a revolver in other banks, and whether or not using unsecured loan, cash advance in all banks. At last, this study uses another 598 random sampling data to test the correct classification power of the final logistic regression model.
