|
This study intends to use collected credit ratings for credit card users from a certain domestic bank, as well as information regarding approved customers by this credit issuing bank, as samples of evidence, to further analyze, verify and identify important factors that affect and influence customer willingness to pay back both owed interests and principals. In conjunction with cross-referencing past published literature, this study intends to extract key factors to be used as notable elements in influencing credit risks by any individual bank in the future. Then followed by applying Logistic Regression model to further establish credit risk appraisal model.
Source materials for this study are credit card user samplings from a domestic bank. There were a total of 25,613 samples taken, and listed with all possible credit-influencing variables. By using SPSS package software, this study proceeded Logistic Regression analysis, and identified most notable risk variables which would influence credit risk. Next, this study also observed existence of mutual and interactive relationships among all variables identified, which did exhibit striking influences on credit risks.
Out the six variables chosen, this study identified characteristic elements which affect credit risk from credit card user. The following characteristics are observed with notable standards of 1%：
1.Correlations between academic backgrounds with occurrences of payment overdue were unnoticeable.
2.Variables such as credit allowance amount, age, sex, marriage status and number of cards possessed, do reveal distinctive correlation with occurrences of payment overdue.
3.Variables of amount of credit allowance and age, do exhibit positive correlation. But sex, marriage status and number of cards possessed, demonstrate negative correlation.
Functions generated from Logistic Regression model after further analyzing and proving with evidence, can be further classified based on important factors： amount of allowed credit, number of cards possessed, age, sex and marital status. Then card-issuing bank can follow priorities of these important factors, through model of credit-rating risk appraisal for credit card, assign relevant weighting factors. After standardizing review models for rating credit risk, the bank can then objectively and promptly detect credit breach from credit card applicants, and rely these as future reviewing and approval principles.
|