A differentiable model of supply-chain shocks

TL;DR

This paper introduces a differentiable approach to calibrate agent-based models of supply chains, significantly speeding up the process and enabling large-scale modeling of global supply networks.

Contribution

It demonstrates that using GPUs and automatic differentiation can accelerate ABM calibration by over 1000 times, facilitating scalable supply chain modeling.

Findings

Speed-ups of over 3 orders of magnitude in calibration time

Feasibility of modeling entire global supply networks

Potential for improved supply chain shock analysis

Abstract

Modelling how shocks propagate in supply chains is an increasingly important challenge in economics. Its relevance has been highlighted in recent years by events such as Covid-19 and the Russian invasion of Ukraine. Agent-based models (ABMs) are a promising approach for this problem. However, calibrating them is hard. We show empirically that it is possible to achieve speed ups of over 3 orders of magnitude when calibrating ABMs of supply networks by running them on GPUs and using automatic differentiation, compared to non-differentiable baselines. This opens the door to scaling ABMs to model the whole global supply network.