Banks Face Rising Costs in Developing Alternative Credit Assessments

As financial authorities push for greater access to credit through alternative credit assessments, banks are facing increased costs. Each bank is responsible for developing and operating models that utilize non-financial data, raising concerns about the financial burden without clear standards or common infrastructure.

According to the financial sector on July 27, major banks are developing their own alternative credit assessment models to evaluate the repayment ability of customers with limited financial transaction histories.

Shinhan Bank has been using its 'Alternative Credit Assessment Model for Low-Income Individuals,' which incorporates automatic payment histories for living expenses and utility bills, since March for its low-income credit loan evaluations. KB Kookmin Bank is also applying financial and non-financial information, including payment histories for communication fees, in its 'KB First EASY Credit Loan' assessments. NH Nonghyup Financial Group plans to finalize its machine learning assessment strategy using alternative information by October.

Alternative credit assessments evaluate borrowers' repayment capabilities using non-financial data such as payment histories for communication fees, utilities, rent, and taxes, aiming to improve loan access for young adults and those new to the workforce who may struggle to establish creditworthiness with traditional financial data.

However, developing and operating these models incurs significant costs. While financial authorities emphasize the use of alternative credit assessments, the criteria and methods remain at the discretion of individual banks. Each institution must independently develop models and establish IT systems.

Banking sources estimate that developing new assessment models or modifying existing ones can cost tens of millions of won. To enhance accuracy, banks need to acquire, process, and refine data from various institutions. They must also integrate this data with existing loan assessment systems and validate the appropriateness of the models.

Acquiring data also incurs costs. To utilize information held by telecommunications, retail, and platform companies, banks must enter separate contracts and establish procedures for data protection and consent. Additionally, the lack of sufficient long-term data to verify the accuracy and risk of alternative credit assessments poses a challenge.

Each time a new evaluation criterion is added, the model must undergo validation and IT integration, extending the development timeline. Including external consulting, testing, and system stabilization can take over a year, according to banking sources.

The financial sector criticizes the current structure, where individual banks purchase similar data and repeatedly develop similar models, as inefficient. There is a need for at least minimal common standards regarding the scope of alternative information, data formats, and model validation methods.

A financial sector representative stated, “While authorities provide direction, they leave the detailed standards and infrastructure development to the banks, which must bear the development and operational costs. It is essential to establish common guidelines for data utilization and model validation while ensuring banks' autonomy.”