Researchers from BI Norwegian Business School and NHH Norwegian School of Economics have developed a new behavioral credit-risk model that integrates credit and debit transactions, significantly outperforming state-of-the-art machine learning methods in predicting credit card delinquency. The model, detailed in a study published in The Journal of Finance and Data Science, offers clearer insight into the behavioral drivers behind repayment problems by moving beyond traditional monthly aggregates.
"Credit data alone gives only a partial picture of a customer's financial situation," explained first author Håvard Huse. "By integrating debit transactions, we gain insight into payday spending, repayment behavior, and income patterns—factors that strongly influence whether someone is at risk of missing payments." The study, available via its DOI link, draws on detailed transaction data from a large Norwegian bank.
Traditional credit-risk models rely heavily on static monthly figures like balance and credit limit, which fail to reveal how customers manage finances day-to-day. The new hierarchical Bayesian behavioral model captures dynamics such as evolving repayment patterns and post-payday spending spikes. "By capturing behavioral dynamics... the new model explains both why delinquency occurs and who is likely to default," Huse shared. This approach allows the model to identify distinct behavioral segments with different "memory lengths," reflecting how past financial states influence current behavior.
Co-author Auke Hunneman noted that customers in financial distress are more influenced by earlier months' behavior, a dynamic the model captures better than standard machine-learning tools like XGBoost, GBM, neural networks, and stacked ensembles. Beyond superior accuracy, the model offers greater interpretability. "Banks not only need accurate predictions—they also need to understand which behavioral patterns drive risk," Hunneman added.
The practical implications are substantial. Using a three-month prediction horizon, early detection of at-risk cardholders could generate significant cost savings by enabling timely interventions. Co-author Sven A. Haugland stated, "For banks, this is more than an accuracy improvement—it is a way to proactively help customers avoid serious financial problems." The findings signal a shift in credit scoring from traditional static models toward richer behavioral analytics based on a complete view of customer transactions, as highlighted in the research published in The Journal of Finance and Data Science.
