The Formation of Production Networks: How Supply Chains Arise from Simple Learning with Minimal Information

2025-04-22

AI Summary

This paper investigates how heterogeneous firms autonomously determine prices, output levels, and upstream input procurement (type and quantity) via simplified reinforcement learning—without assuming market equilibrium or possessing complete technological information—thereby endogenously evolving stable yet reconfigurable production networks. Method: We develop a generative model grounded in dynamic input-output interactions, wherein firms learn and optimize solely from local feedback signals. Contribution/Results: Our approach is the first to achieve fully endogenous evolution of production networks without global equilibrium or perfect information assumptions. The model supports adaptive network reconfiguration under multiple exogenous shocks—including demand fluctuations, node exit, productivity shocks, and technological upgrades. Simulation results successfully replicate empirically observed hierarchical and modular structures of real-world supply chains and quantitatively identify asymmetric shock propagation across upstream and downstream tiers.

Technology Category

• Search and Optimization: Learning to Search
• Planning, Routing, and Scheduling: Learning for Planning and Scheduling
• Game Theory and Economic Paradigms: Adversarial Learning

Application Category

• Economics, Online Markets and Human Computation: Social networks and social learning
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

We develop a model where firms determine the price at which they sell their differentiable goods, the volume that they produce, and the inputs (types and amounts) that they purchase from other firms. A steady-state production network emerges endogenously without resorting to assumptions such as equilibrium or perfect knowledge about production technologies. Through a simple version of reinforcement learning, firms with heterogeneous technologies cope with uncertainty and maximize profits. Due to this learning process, firms can adapt to shocks such as demand shifts, suppliers/clients closure, productivity changes, and production technology modifications; effectively reshaping the production network. To demonstrate the potential of this model, we analyze the upstream and downstream impact of demand and productivity shocks.

Problem

• Model explains endogenous formation of production networks
• Firms use reinforcement learning to adapt to uncertainty
• Analyzes impact of demand and productivity shocks

Innovation

• Reinforcement learning for profit maximization
• Endogenous steady-state production networks
• Adaptation to shocks via learning