Scholarlink

<p>The global semiconductor supply chain faces unprecedented risks driven by geopolitical tensions, rare earth element (REE) concentration, and technological dependencies. This paper develops a comprehensive theoretical and empirical framework for supply chain risk management in the semiconductor industry by integrating the Functional Resonance Analysis Method (FRAM), the Analytical Hierarchy Process (AHP), and Explainable Artificial Intelligence (XAI). The framework is designed for decision-making under conditions of geopolitical fragmentation, where supply concentration, export control regimes, and industrial policy interventions, such as the United States CHIPS and Science Act and the European Critical Raw Materials Act, jointly shape the strategic space available to manufacturers, policymakers, and regulators. We introduce two theoretical results, the <i>Geopolitical Concentration Risk Propagation Theorem</i>, which formalizes how supply concentration amplifies systemic risk through the supply chain network, and the <i>Rare Earth Criticality Equilibrium Theorem</i>, which establishes conditions for stable multi-sourcing strategies under export control uncertainty. The framework is validated using real-world data from the U.S. Geological Survey (USGS), the International Energy Agency (IEA), and the Semiconductor Industry Association (SIA), covering the period 2020–2024. The empirical analysis is built on a curated dataset of 15 critical minerals across 47 countries, with clearly specified training and out-of-sample validation partitions, and risk-prediction accuracy metrics aligned with the supply chain risk management literature. The framework achieves a 34% improvement in risk prediction accuracy compared with traditional AHP-based approaches, and SHAP-based explanations ensure transparency of the resulting policy guidance, with LIME-SHAP agreement exceeding 92% across the test scenarios considered. The findings provide actionable insights for navigating the increasingly fragmented semiconductor landscape while embedding rare earth dependencies, low-carbon transition needs, and trade policy dynamics into a single, interpretable decision architecture.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

An explainable FRAM–AHP–XAI framework for semiconductor supply chain risk under rare earth dependencies

  • 期刊论文

  • 发表日期：

    2026年7月30日

查看全文
  • Massimiliano Ferrara,
  • Valeria Isgrò

摘要

The global semiconductor supply chain faces unprecedented risks driven by geopolitical tensions, rare earth element (REE) concentration, and technological dependencies. This paper develops a comprehensive theoretical and empirical framework for supply chain risk management in the semiconductor industry by integrating the Functional Resonance Analysis Method (FRAM), the Analytical Hierarchy Process (AHP), and Explainable Artificial Intelligence (XAI). The framework is designed for decision-making under conditions of geopolitical fragmentation, where supply concentration, export control regimes, and industrial policy interventions, such as the United States CHIPS and Science Act and the European Critical Raw Materials Act, jointly shape the strategic space available to manufacturers, policymakers, and regulators. We introduce two theoretical results, the Geopolitical Concentration Risk Propagation Theorem, which formalizes how supply concentration amplifies systemic risk through the supply chain network, and the Rare Earth Criticality Equilibrium Theorem, which establishes conditions for stable multi-sourcing strategies under export control uncertainty. The framework is validated using real-world data from the U.S. Geological Survey (USGS), the International Energy Agency (IEA), and the Semiconductor Industry Association (SIA), covering the period 2020–2024. The empirical analysis is built on a curated dataset of 15 critical minerals across 47 countries, with clearly specified training and out-of-sample validation partitions, and risk-prediction accuracy metrics aligned with the supply chain risk management literature. The framework achieves a 34% improvement in risk prediction accuracy compared with traditional AHP-based approaches, and SHAP-based explanations ensure transparency of the resulting policy guidance, with LIME-SHAP agreement exceeding 92% across the test scenarios considered. The findings provide actionable insights for navigating the increasingly fragmented semiconductor landscape while embedding rare earth dependencies, low-carbon transition needs, and trade policy dynamics into a single, interpretable decision architecture.