Berlin School of Economics and Law

The foundational formulation of supply chain viability in intertwined supply networks (ISN) is based on a dynamic game-theoretic modelling of a biological prey-predator system (i.e., the trophic chain) with associated viability kernel computation. However, two important aspects–network structural dynamics and decentralized coordination–have not been considered. In this study, we close this research gap by extending the foundational model in two ways. First, we extend the general viability model analytically through a control-theoretic lens, explicitly accounting for structural dynamics control. Second, we introduce a multi-agent system (MAS) grounded in the differential game of the foundational model and demonstrate, both conceptually and numerically, how viability is quantified at the agent level. Conceptually, the MAS serves as a micro-foundation for viability theory, translating abstract set-valued dynamics into agent-level decision processes. This allows to connect two major determinants of supply chain viability, i.e., engineered variety (control model) and emergent adaptive behaviors (agent-based model). In addition, we introduce time-to-exit-viability as a performance metric to quantify the conditions under which a network fails while preserving viability. The proposed formulations enable analysis of how decentralized decisions affect supply chain viability, the role of network structure in sustaining survivability, and coordination failures and resulting viability losses. In particular, we show that the viability kernel characterizes potential survivability and does not necessarily guarantee viability under non-coordinated decision-making.

In the event of a disruption at a primary supplier, the manufacturer's sourcing and production operations are inevitably affected, necessitating a response strategy based on the estimated disruption duration. This article derives the optimal production policy for each response strategy, aiming to minimize total costs over a planning horizon within a dynamic programming framework. We characterize these production policies across various model parameters, particularly focusing on the estimated disruption duration and initial inventory level. Our analysis identifies when the passive acceptance strategy and the backup strategy each hold an advantage, and how the cost benefits of each strategy vary with different model parameters. Additionally, we examine the consequences of overestimating or underestimating the disruption duration, providing insights into how the chosen response strategy and production policy, based on the estimated duration, diverge from the true optimal ones as the actual duration unfolds. We also highlight the cost implications of these discrepancies. Finally, in scenarios where the true duration follows a probabilistic distribution, our numerical experiments demonstrate how strategy selection and the cost of misestimation are influenced by the mean of the true disruption duration and the initial inventory level. These analyses offer decision-makers valuable guidance on when to act or wait in the face of disruptions.

In this study, we describe Ford’s practices and propose three industry-based frameworks for supply chain digital twin (SCDT) design and implementation at scale. First, a generalized three-layer framework for the design of SCDTs based on Ford's approach is developed. The layers are intracompany, Tier-1 network, and deep-tier network, classified based on data visibility. We describe how digital twins can enhance operational performance and be utilized for resilience stress testing. Second, generalized frameworks of SCDT implementation are shown composed of two dimensions, i.e., implementation scale and implementation scope. The three-stage implementation scale framework proposes a roadmap for transition from data-driven organizations to digital twin-driven management systems. The four-level implementation scope framework encompasses product, process, organization, and extended network levels, with a focus on the key role of the data analytics department in deploying SCDTs. We then generalize four fundamental principles for SCDTs: (i): object-driven and data-driven design and adaptation, (ii) visibility as the central angle of digital twin design and technology, (iii) digital twins are integrators of data and knowledge, and (iv) SCDT continuous adaptation. To the best of our knowledge, our paper is the first in the literature to report on the design and deployment of an SCDT at scale, which can be useful for academics and practitioners alike. We conclude that a properly developed SCDT can enable strategic and operational performance improvements, end-to-end visibility, agentic AI integration in decision-making, and supply chain stress testing, as well as create a new approach to managing the supply chain.

Artificial intelligence (AI) companies and their rhetoric infringe on academia in harmful ways, mirroring past uncritical acceptance of industry logics, such as those of tobacco and petroleum. In this position piece, we tease apart and explain why phrases like ‘generative AI’ impede scholarly discussion because by design these expressions are used to dazzle and sidestep scrutiny. Furthermore, we contend with the AI industry’s logics to enable rejecting frames such as: that we must embrace the future, that this hype cycle is unique, that anthropomorphism and circular reasoning hold water when discussing AI systems, and that students are now all cheating or all need to use AI. To these ends, we expound on why universities must take their role seriously to a) counter the AI industry’s marketing, hype, and harm; and to b) safeguard higher education, critical thinking, expertise, academic freedom, and scientific integrity. For each point we raise, we include pointers to relevant work to further inform and convince our colleagues.