Cascading Credit Risk Assessment in Multiplex Supply Chain Networks

Credit risk identification in financial markets is critically important as global economies become more linked. This is particularly evident in multiplex networked supply-chain platforms. In these systems, risks can damage market stability. They also create information gaps and resource misallocations across different parts of the supply chain. Additionally, they may trigger cascading failures at key nodes. Most current methods treat these complex multiplex structures as single-layer networks. This simplification fails to capture their true dynamic behavior. Another issue is that nodes often operate across multiple chains, which makes risk mitigation harder. To solve these problems, we propose CIRAM, a cascading risk assessment method that uses neural networks. CIRAM includes contagion strength coefficients and a multi-transfer probability framework. These support a new multi-label propagation mechanism. Tests on supply chains of different scales show that CIRAM beats four baseline methods in precision, recall, and F1 score. CCS CONCEPTS Mathematics of computing ~Probability and statistics ~Probabilistic inference problems