Driving Financial Inclusion Through Advanced Scoring

This study is a research article that applies standard credit scoring methodologies to open banking data. It incorporates reject inference through multi-dimensional parcelling and advanced feature engineering for transaction data. The research addresses the unique opportunities and challenges presented by open banking data in credit risk modelling, particularly regarding rejected applicants. The study demonstrates the effectiveness of logistic regression combined with weights-of-evidence transformation, showing that this method can outperform traditional bureau credit scores for subprime populations. It discusses the importance of recalibrating score cutoffs to ensure fair assessments of newly inferred applicants while balancing risk and approval rates. The findings indicate that the proposed approach can lead to a reduction in bad rates, an increase in overall lending, and improved financial inclusion for underrepresented populations. By leveraging open banking data and refining reject inference techniques, the study outlines a pathway to more accurate and fair credit decisioning.