Artificial Intelligence for Coupled Financial and Logistics Risk Management in Supply Chains

Abstract

Supply chains move goods and money together, yet firms, lenders and logistics providers still manage the two flows with separate models, separate data and separate risk owners. A delayed vessel lengthens the cash conversion cycle, a stretched cash cycle raises a supplier’s default probability, and a supplier default removes physical capacity from the network. Artificial intelligence now forecasts each of these events well in isolation, but the coupling between them is rarely modelled. This paper reviews the literature at the intersection of financial risk, logistics and artificial intelligence, drawing on a structured corpus of peer-reviewed studies, regulatory frameworks and recent work up to early 2026. The review organises the evidence into five method families: credit and default prediction, disruption and freight market forecasting, network learning, language models for unstructured risk signals, and prescriptive decision models. It finds that each family is mature on its own side of the boundary and thin across it. On this basis the paper proposes the Coupled Flow Risk framework, a five-layer architecture of sensing, prediction, propagation, prescription and governance, with a formal coupling term that lets logistics stress enter credit risk and credit stress enter logistics capacity, and a decision rule that trades expected cost against conditional value-at-risk. The paper closes with a research agenda on calibration, causal validation, fair access to supply chain finance and model risk governance.