Coercion Without an Author

Economic coercion is priced, deterred and remedied only when it can be attributed. Every instrument the international legal order possesses for responding to it — the prohibition on intervention, the secondary rules of state responsibility, countermeasures, the security exceptions of the trading system, and the domestic sanctions architecture built on top of them — presupposes a provable author. This paper argues that the presupposition no longer holds, measures what remains when it fails, and finds that the last consolation the literature offers is not supported by the data. The doctrinal argument is that six routes exist and five fail at the same point. The prohibition on intervention fails because purpose cannot be shown; state responsibility fails because control cannot be shown to the standard Bosnian Genocide preserved; the security exception fails because motive cannot be tested, and its jurisprudence since 2019 consists of one adopted panel report and six appealed into a vacancy; countermeasures fail because the predicate cannot be established to a standard that makes the response safe; and non-violation nullification, which requires no wrongfulness at all, still requires the measure to be shown to have been applied by the respondent, which is the half of the problem that matters. Only creeping expropriation reasons from effect — to the characterisation of acts whose authorship is not in dispute, never to authorship — and it fails on standing and forum rather than on proof. The paper's worked case is a pair. In China — Measures Concerning Trade in Goods and Services the European Union brought a complaint over the treatment of Lithuanian trade, had the panel's work suspended twice, the second time three days after it resumed one day before the lapse deadline, and withdrew the complaint on 1 December 2025 without a report or a finding. In China — Enforcement of Intellectual Property Rights the same complainant established an unwritten Chinese policy from a pattern of individual acts and obtained a binding award. The difference is not written against unwritten. It is whether the conduct leaves a documentary trace in the hands of someone other than the acting state. Judicial orders do; customs discretion does not. That criterion is stated ex ante, it is falsifiable, and it locates the boundary of the paper's own claim. The formal argument yields a result that is narrower and more uncomfortable than deterrence failure. Where the attainable probability of attribution falls below the point at which a proportionate countermeasure can deter, the countermeasure channel closes — it needs a provable wrongful act and is capped by proportionality — but untargeted retorsion does not, because lawful unfriendly conduct requires no attribution and no predicate. Retaliation does not stop; it migrates to instruments that cannot distinguish the responsible from the rest. The point is not that the cap is low but that it is legal: a chosen rule with a lawful substitute permanently beside it. The European Union's anti-coercion regulation is that structure in legislative form — a determination of coercion as the trigger, non-performance of international obligations permitted only "insofar as" wrongfulness obtains, and every listed measure drafted so that it need not be treaty-inconsistent. The empirical work is original and its headline finding cuts against the paper's own framing. Bilateral trade in 2024 at six-digit tariff-line resolution — 11,250,411 flows covering USD 22.94 trillion — is reduced to 323,929 importer-product cells and scored for concentration, world supply grip and the evidenced availability of substitutes. Severe chokepoints in production inputs are rare: 1,533 cells of 233,666, carrying 0.77 per cent of input trade by value, an order of magnitude consistent with the European Commission's narrower 2021 screen and here extended to the whole of world trade. Where they exist, one state is the leading supplier in 73.3 per cent of them. The second empirical result is negative and it removes the standard consolation. If deterrence survives below the threshold at all, it survives through mutual exposure — a target holding counter-leverage over the state it depends on can respond without alleging anything. That reciprocity is not there. The obvious test has no power: 21 of the 23 states China depends on are dependent on China in return, but 142 of the 152 states holding any severe dependency are, so coincidence predicts 21.5. Excluding China and testing the remainder against a degree-preserving null gives 38 reciprocated dependencies against 37.5 expected. Magnitudes point the same way: across the nineteen mutual pairs the median imbalance is fifteen to one, and two are within a factor of two. Reciprocal severe dependency occurs at exactly the rate the degree distribution predicts. The mutual-hostage reading of interdependence is not supported at tariff-line resolution. The paper closes with an evidential mechanism: where a measurable interruption coincides with a demonstrable grip on world supply and with the published absence of a verifiable ordinary explanation, a tribunal may draw an inference of fact in the manner Corfu Channel permits where the respondent holds the material. It shifts an evidential burden, not the legal burden and not the standard of proof, and it is drafted against the passage in Corfu Channel which says that exclusive control alone does neither. A companion paper applies the same cost structure to markets rather than to law. The two share a structure and not a cause, and neither depends on the other. Preprint; not yet peer reviewed.