Supply-Chain Disruption, Supplier-Country Health, and the Reallocation of US Import-Value Shares: Monthly Evidence Under Timing-Disciplined Measurement

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Supply-chain disruptions since 2018 renewed attention to supplier-country conditions; this paper examines whether deterioration in those conditions precedes changes in US import-value shares when each measure is dated to match the question asked. From 564,983 deduplicated news records (2018–2026) and direct-source IMF, COMTRADE, and BIS data for 62 supplier economies, we build a Country Health Score under two datings: reference-dated, assigned to the months the data describe, and availability-lagged, shifted by one-to-three-month frontier-lag proxies. Among 29 dominant commodity–country cells across 14 countries, reference-dated deterioration is followed by within-commodity share erosion over three to twelve months, concentrated after 2022 and surviving Holm and wild-cluster inference at every horizon (largest Holm or wild-cluster p=0.014). Excluding Canada and Mexico, the association is solid at three months and suggestive at six. The association survives a pre-specified mechanical-channel gate (re-estimation without the export-coverage component) and a timing placebo; conflict-coverage spikes are a second, timelier predictor (post hoc). The same regression on the availability-lagged composite is null, and the corresponding coefficients differ formally at three and six months, consistent with timing-misalignment attenuation. Return regressions delimit rather than support the claim: no predictive relationship replicates out of sample. These results document a timing-consistent association, not an identified causal effect. Observations postdating the frozen analysis plan will provide the first fully prospective test.