A hierarchical Bayesian model for payment delay prediction in supply chain financing
针对供应链融资中公司级信用评级与订单级交易不匹配的问题，提出分层贝叶斯模型，结合宏观信用与微观订单数据，提升支付延迟预测精度，并在航空供应链案例中验证其优于传统模型。
Supply Chain Financing (SCF) has gained importance as a critical tool for improving liquidity and working capital efficiency across supply chains, particularly by enabling suppliers to receive early payments and allowing buyers to extend payment terms without disrupting operations. However, current risk evaluation practices in SCF rely heavily on credit ratings at the company level, which often misalign with the order-level nature of SCF transactions. This misalignment can lead to inaccuracies in assessing payment risk. To address this issue, this paper introduces a Hierarchical Bayesian Model (HBM) that integrates macro-level credit ratings with micro-level order and delivery data to enhance the precision of payment delay predictions. The model not only identifies common behaviour patterns among suppliers, but also offers detailed supplier-specific risk assessments. Through a real-world case study in the aerospace supply chain sector, we demonstrate that HBM significantly outperforms non-hierarchical models in prediction accuracy and provides actionable insights, such as the positive correlation between delivery and payment delays, and the unusual finding that higher credit ratings may be associated with longer payment delays. Our findings suggest that combining company- and order-level data within an interpretable probabilistic framework enhances the explainability and precision of SCF risk assessment.