MODELING THE DEFAULT RISK OF CARD LOANS CONSIDERING INDIVIDUAL BEHAVIOR CHARACTERISTICS BASED ON BANK ACCOUNT DEPOSIT AND WITHDRAWAL DATA

Abstract

In general, banks keep management costs for card loans down because the amounts are smaller than those of corporate loans. Therefore, they are not sufficiently managed after the money is lent. This study analyzes the deposit and withdrawal data of bank accounts used by card loan borrowers and proposes a default risk management technique. A previous study analyzed bank account data for screening applicants for card loans, and constructed a model to evaluate default risk. However, to the best of our knowledge, there are no studies that analyze bank account data to determine measures for managing the default risk while lending money with a card loan or construct a management model. In addition to the explanatory variables used in the screening model, we employ some variables observable after lending money that are related to individual behavior characteristics, such as number of borrowings and contract repayment rate, and analyze their relationships with default. We construct a logit model using these variables and examine the model using approximately 60 million observations. The result shows that the accuracy ratios exceed 50%, and the model is effective in practice. We also confirm the robustness of the results through cross-validation and an out-of-sample test.