The Behaviour of Supply-Chain Networks under Targeted Export Controls

Abstract

The October 7, 2022, Export Controls restricted the flow of advanced semiconductor chips to China, imposing a sector-wide rule on all firms at once, rather than the firm-specific controls earlier. This provides us with a policy shock to base our analysis on. Using data from FactSet Revere on bilateral counterparty relationships, and a continuous DiD design with binarised treatment following Callaway, Goodman-Bacon, and Sant’Anna 2024, we document how the resulting decoupling occurred at the firm level, from who supplies to whom. We define treatment as pre-shock exposure to the relevant channel, exposure to Chinese customers for US firms and exposure to US suppliers for Chinese firms. There are two mechanisms, Stagnation, through formation differentials that open up after the shock (-1.4 pp on the customer side, -3.9 pp on the supplier side, p<0.001), while termination differentials remain null. The bilateral share decline therefore emerges because new cross-border ties do not form, not because existing ties are cut. Localisation: Chinese firms shifted towards local supply chains at scale (+11.6pp ATT on Chinese Supplier Share, and +19.6 pp amongst the most exposed firms), with approximately 83% of the lost US-Supplier Share absorbed locally (in share-point terms). We observe that relationships are being redirected toward countries in the U.S.- aligned manufacturing bloc rather than toward destinations closely tied to China, which aligns with Farrell and Newman’s (2019) idea that chokepoints in global networks can be weaponised. Concentration follows for low-to-moderate-exposed firms, while the most-exposed quartile reroutes and escapes. Lastly, we evaluate robustness through pre-trend testing, placebo dates, placebo exposures, alternative estimators, alternative control groups, multiple-testing corrections and share decomposition tests.