A Study on the Effectiveness of Credit Scoring Using the Current and Modified Models of the Lao Credit Information Company

Abstract

This study aimed to compare the accuracy performance of credit scoring models between the current and modified versions, as well as to evaluate the risk performance of the new model in reducing risks associated with customer selection. The comparison involved the current credit scoring model (13 indicators) and an improved model (16 indicators, including additional indicators for utility bill payments) used by the Lao Credit Information Company (LCIC). The study utilized customer credit data submitted by commercial banks and financial institutions across Laos to the CIC database. Testing was conducted with datasets of 100, 500, and 1,000 customers, based on indicators such as gender, age, CIB repayment history, marital status, loan purpose, loan term, loan amount, outstanding debt, number of inquiries, overdue days, collateral type, collateral value, province, electricity bill, water bill, and telecommunication bill. The results revealed that the modified model outperformed the current model in both accuracy and risk identification. The new model more effectively predicted repayment ability and reduced the risk of selecting customers likely to default. The findings suggest that the modified model can enhance the credit services of financial institutions, mitigate lending risks, and contribute to the socio-economic development of the Lao PDR by strengthening financial stability.