Banksalad is using MyData with Sogang University to analyse the relationship between spending behaviour and personal credit risk and develop an alternative credit scoring model based on the findings.

Banksalad said on Monday it published a paper titled "Spending Behaviour and Personal Credit Risk: MyData-Based" jointly with a research team led by emeritus professor Ju-ha Nam (남주하) of Sogang University.

The study used about 200,000 card payment records held by Banksalad and the company's in-house system for classifying spending categories. Based on actual transaction data, the researchers conducted an empirical analysis of the link between spending behaviour and personal credit risk.

The study found that users who spend consistently on medical and health items have a lower risk of default. The researchers analysed that steady health management could reduce the likelihood of a halt in economic activity due to illness and could act as a factor that eases liquidity burdens from sudden health deterioration.

By contrast, the risk of default tended to rise as the share of spending on telecoms bills, convenience stores, cafes and snacks increased. The researchers found that the emergence of new spending that differs from usual patterns, or shifts in existing spending patterns, is a meaningful variable for predicting credit risk.

Banksalad plans to develop the alternative credit scoring model, called the Banksalad Score, with HonestAI and Korea Credit Bureau (KCB), and to push for commercial use in the financial sector based on the findings.

According to the company, the model applies machine-learning algorithms to distinguish between creditworthy borrowers and high-risk borrowers and posted 60 percent in a K-S statistic evaluation. The K-S statistic is used as an indicator to measure the discriminative power of credit scoring models.

Banksalad has built 28 spending categories and 119 detailed spending items covering areas such as telecoms, transport, online shopping and travel and has used them for its asset management services. The same classification system was also applied to the study and the development of the alternative credit scoring model.

A Banksalad official said, "We confirmed the possibility that consumption data can be used as credit assessment information to identify individuals' financial characteristics." The official added, "We will expand data-based financial services to improve financial access for users who lack sufficient transaction history."