Graph neural networks improve corporate default prediction using trade data
DefaultGNN: A Dual-Perspective GNN Framework for Predicting Corporate Default from Buyer-Seller Transaction Networks
Machine LearningArtificial Intelligence
Summary
Predicting if a company will fail to pay its debts is important but hard when financial information is missing. The authors studied how companies buy and sell from each other using real tax invoice data over six years. They found that risks travel differently depending on the company’s role as buyer or seller and the size of transactions. Using this insight, they built a special kind of AI model that looks at both those views and predicts defaults better than existing methods. Their model also helps explain which trading partners increase a company’s risk, and works well alongside traditional credit scores.
What this means in practice
- •For credit analysts: Enhance approval rates by integrating transaction network risk insights into credit scoring for firms with limited financial data.
- •For financial institutions: Deploy dual-perspective graph models to detect default risk from buyer-seller data beyond conventional financial statements.
Authors
Junghoon Kim, Hyunsung Kim, Seungyoon Choi, KyoungYong Park, Jihun Lee, YongGu Ji, Chanyoung Park
Abstract
Corporate default prediction is a core problem in financial risk management, yet traditional credit models rely heavily on financial statements that are often sparse or unavailable for many firms. Corporate transaction networks offer a complementary view of real economic activity, but how risk propagates through buyer-seller relationships remains underexplored. We conduct a large-scale empirical study using real-world electronic tax-invoice data spanning six years that links transaction histories with default events, revealing that transaction-driven risk is both role-dependent (buyer or seller) and scale-dependent. Based on these findings, we construct multiplex buyer-view and seller-view transaction networks and propose DefaultGNN, a dual-perspective graph neural network-based framework for corporate default prediction. DefaultGNN integrates both views to model how risk flows through transactional relationships, achieving strong improvements over both attribute-based and graph-based baselines, especially for firms with limited intrinsic risk signals. We further provide interpretable network-based explanations by visualizing how distressed trading partners contribute to default risk. In collaboration with a licensed credit rating agency, we validate that DefaultGNN's predictions complement existing credit scoring models, improving approval rates by 7-11%p without increasing default risk among approved firms. The source code can be found at https://github.com/jhkim611/DefaultGNN