A geometric-causal knowledge framework for quantifying and simulating counterfactual credit risk cascades under supply chain shocks

Conventional credit risk models often fail to capture the non-linear cascades triggered by supply chain shocks, treating systemic risk within a flat linear Euclidean framework. This study introduces a unified geometric-causal knowledge discovery framework that conceptualizes the supply chain ecosystem as a Riemannian manifold ( M , g ) . We derive a deterministic information-geometric metric tensor based on Shannon entropy, which warps the state-space to reflect multimodal information density. To validate this framework, we develop a rigorous data generating process (DGP) that simulates a systemic disruption in the Strait of Malacca, modeling the shock as a metric perturbation that propagates through the manifold's topology. By projecting the simulated states onto the manifold using kernel principal component analysis, we move beyond stochastic visualizations to a mathematically robust local isometry. Our framework yields three primary knowledge discoveries: Structural Fragility: The DGP reveals that upstream nodes act as high-curvature "risk sinks," where systemic volatility aggregates and intensifies. Early Warning: We introduce geodesic migration magnitude as a superior proxy for structural sensitivity, identifying distressed companies long before traditional financial ratios (e.g., Z-score) signal default. Causal Transition: Through tangent space gradient mapping, we prove a structural shift in risk drivers, where revenue growth emerges as the dominant geometric force following the maritime disruption. Finally, we operationalize these insights in a "policy sandbox" using parallel transport to simulate counterfactual interventions. Results demonstrate that our geometric-causal approach significantly outperforms baseline linear models, providing regulators with a proactive laboratory for enhancing supply chain resilience.