De-risking from China: how US firms reconfigure sourcing through decoupling, diversification and friendshoring across geopolitical disruptions

Purpose We examine whether US firms use three location strategies—decoupling, diversification, and friendshoring—to de-risk their sourcing portfolios. We investigate how the use of these strategies varies across bilateral (2018 US–China trade war), systemic (COVID-19 pandemic) and alliance-based (Russia–Ukraine war) shocks. We also test whether firms operating in high-tech industries de-risk more aggressively. Design/methodology/approach Using Panjiva US seaborne import records, we build firm-year measures of the three strategies for 1,565 US firms over 2011–2024. We then estimate fixed-effects models that interact firm's pre-disruption China exposure with the three disruption periods and a high-tech indicator. Findings Highly China-exposed firms decouple during all three disruptions, with decoupling switching on sharply at the trade war. Diversification activates specifically at the COVID-19 pandemic. Friendshoring peaks during the COVID-19 pandemic and reverses once the war begins. Sourcing portfolios appear more diversified and politically aligned, but these gains vanish once the China sourcing share is excluded, revealing them as mechanical by-products of decoupling. High-tech firms generally do not de-risk more than their peers. Originality/value We move the study of de-risking from perception-based case evidence to large-scale observed sourcing behavior. We conceptualize decoupling, diversification, and friendshoring as three sourcing location strategies and match each to its most salient disruption type, laying the groundwork for a contingency theory of strategy–shock fit. We further show that portfolio-wide indices of diversification and political alignment may overstate genuine de-risking because they incorporate the effects of decoupling. Moreover, policymakers' ambitions to de-risk high-tech industries and promote friendshoring have generated only a limited corporate response.