The Capacity Complaint

Analysis · 28 August 2026 · predictions registered before research · honesty score −11, reported at a loss

Every previous trade remedy could be answered by changing what you ship. A complaint about unused capacity can only be answered by scrapping the plant — which means it cannot be answered at all, and was probably never meant to be.

28 August 2026

The descriptive analysis of this instrument was published before this article was conceived, and it belongs to Johannes Fritz.

Writing for Global Trade Alert on 12 March, a day after the notice issued, Fritz measured what it contains: 33 evidentiary indicators cited against only seven named policy interventions across the sixteen targeted economies, with ten economies cited for symptoms alone and no identified cause; a systematic divergence from USTR's own 2025 National Trade Estimate, in which fourteen of the sixteen targets received no overcapacity analysis at all; and the observation that the notice recasts the 1988 Omnibus Trade Act's macroeconomic surplus concern as a manufacturing-capacity offence. Those findings are his.

What follows is narrower: an argument about what kind of instrument that evidentiary structure produces.

What happened

On 11 March, USTR initiated Section 301 investigations into "acts, policies, and practices of certain economies relating to structural excess capacity and production" in manufacturing, naming China, the European Union, and fifteen other economies — Singapore, Switzerland, Norway, Indonesia, Malaysia, Cambodia, Thailand, Korea, Vietnam, Taiwan, Bangladesh, Mexico, Japan and India.

The notice's illustrative sector list runs to twenty-one entries, from aluminium to transportation equipment. Comments closed 15 April and hearings ran from 5 May. The administration's target date for completing the investigations was 24 July; that date passed with no remedies announced under this investigation. The Section 301 tariffs that took effect that day arose from a separate forced-labour investigation. The excess-capacity determination is still pending, with trade bar commentary expecting action late in the year.

The evidence is estimates

Trade remedies have historically been evidenced by customs records. A dumping margin is computed from transaction prices, a countervailing duty from a measured subsidy, a safeguard from an import surge counted at the border. Each of those figures has a paper trail a respondent can attack.

The notice's evidentiary base is different in kind. Its central figures are utilisation rates: global manufacturing capacity utilisation "between 75.0 and 75.9 percent" — sourced, in the notice's own words, to "U.S. government estimates" — against a healthy benchmark of "approximately 80 percent," a benchmark whose cited source is a US steel proclamation. The country sections continue in the same register: China at 74.4 percent utilisation, German chemicals at 72.7, Ireland at 72.7, Thai manufacturing below 60, Bangladeshi cement below 40, European auto plants at 55.

Several of those are estimates of other economies' unused capacity, which is a quantity no customs office records anywhere, because it describes goods that were never made.

The Singapore episode shows why that matters. When the investigation was announced with claims about Singapore's surplus, Singapore's government disputed the figure publicly within two days, pointing to its own trade data showing a deficit with the United States. USTR's own country page reports a US goods surplus with Singapore of $3.4 billion in 2025.

That dispute was possible because flow claims are checkable against customs records both sides keep. A utilisation estimate has no such record to check against. An economy accused of underused capacity cannot produce an invoice proving the capacity was used, and the notice does not disclose its estimation method. The evidentiary shift is therefore also a contestability shift: the further the instrument moves from flows toward capacity, the less the accused can dispute.

The compliance asymmetry

Every prior trade-remedy complaint could be mooted by changing behaviour at the border. Accused of dumping, raise prices. Accused of subsidised exports, absorb the duty or drop the subsidy. Accused of a surge, restrain exports — the 1980s were built on such restraints.

A capacity complaint cannot be answered that way. Exporting less does not reduce capacity; it raises the unused share, which is the offence as the notice defines it.

The only compliance that answers the complaint as stated is scrapping plant. That means destroying capital, most of it privately owned, in some sectors — Norwegian seafood, Bangladeshi cement — barely connected to any US import flow the notice identifies.

The one government the notice can cite as accepting capacity reduction is Korea, in petrochemicals, and that acknowledgment was made for domestic restructuring reasons. The notice deploys it as evidence against Korea rather than crediting it as compliance. The EU's response was to accept the concept and refuse the premise: it "shares the United States' concern" on overcapacity and "does not consider itself a contributor."

This is the evidentiary and compliance structure of sanctions rather than of trade remedies. Sanctions price a counterfactual — what the target could do, assessed by the sanctioning state's own estimates — and are lifted through negotiation rather than through documented behavioural compliance. The excess-capacity investigation reproduces both features inside a trade statute.

That reading is an inference and should be labelled as one. USTR describes the same structure as an investigation into unreasonable practices burdening US commerce.

The precedent, and what is without one

Capacity-premised trade action is not new. The 2018 Section 232 steel action was premised on global overcapacity, and the February 2025 steel proclamation names eighty percent as the target utilisation rate — the same benchmark the 2026 notice cites for all manufacturing. The 1986 US–Japan Semiconductor Arrangement resolved an overcapacity-driven dispute with managed prices and market-share undertakings. There are forty years of precedent.

What the precedents do not contain is capacity as the named object of investigation in itself, applied to twenty-one sectors and seventeen customs territories at once, including economies with which the United States runs a goods surplus.

Singapore is investigated for overcapacity while buying more US goods than it sells. Norway's bilateral surplus is $1.9 billion — a rounding error in US trade — and its cited practices concern how its sovereign wealth fund recycles oil revenue. Japan is included while running a global goods trade deficit of roughly $36 billion.

The notice anticipates the objection and concedes it in advance: excess capacity can be found "even if a given economy might experience balanced trade or have an overall goods trade deficit with the United States or with the world."

Under that sentence, no trade position exonerates.

What resolution will look like

If the argument above is right, the instrument cannot function as a remedy, because the remedy it implies — foreign capital destruction on the strength of US utilisation estimates — is one no target can rationally deliver and no tariff can force.

What a non-compliable demand can do is convert into negotiating leverage. A complaint that cannot be satisfied can always be traded away.

So the prediction is this: the investigation resolves primarily through bilateral settlements whose content is mostly not capacity — purchase commitments, investment pledges, enforcement cooperation, tariff schedules — with capacity language surviving in preambles.

The falsifier is concrete. A final Section 301 action whose relief is conditioned on measured, verified capacity closure in the target economy would sink this argument, and the failure would be checkable against this paragraph.

Five months of negotiation have so far produced no reported capacity-reduction offer from any target. That is worth a discount rather than a victory lap: negotiations are opaque, the window is short, and reports that partners are trading "policy changes on overcapacity" for tariff forbearance could yet turn out to include capacity measures.

The weakest joint

The sanctions-convergence claim is the one most likely to be wrong.

The contestability shift, the compliance asymmetry and the settlement prediction could all hold while it fails as a category claim. If the final action is conventional flow-keyed tariffs, the capacity framing was justification rather than architecture, and this article will have overread the notice's evidentiary register as its operating logic.

Its fate is decided by a document that does not exist yet — the determination due late in 2026 — and this piece should be reread against it.

Sources

1. Initiation of Section 301 Investigations, 91 FR 12886 — USTR, 17 Mar 2026. Read in full; all notice quotations and country figures.
2. Johannes Fritz, "What the Section 301 overcapacity investigation covers, and what it does not" — Global Trade Alert, 12 Mar 2026.
3. USTR press release on initiation, Mar 2026.
4. White & Case client alert on the 16-economy investigation, Mar 2026.
5. Holland & Knight, "And the (Tariff) Beat Goes On", Jul 2026.
6. CRS, "Legal Authority for Section 301 Tariffs to Address Forced Labor and Excess Manufacturing Capacity," LSB11460.
7. CRS, "Section 232 Investigations: Overview and Issues for Congress," R45249.
8. USTR Singapore country page.
9. Singapore says it runs trade deficit with US, contradicting Section 301 probe claim — Malay Mail, 13 Mar 2026. State position, as reported.
10. European Commission CVDs on Chinese BEVs, IP/24/5589; Implementing Regulation (EU) 2024/2754; WTO DS630. Consulted at secondary-source depth.
11. EU response to the investigation, via press summaries of official statements.
12. 1986 US–Japan Semiconductor Agreement, historical account — Hart, Indiana University.

Produced under provenance discipline. Falsifiable predictions were registered to a signed chain before any research began, verdicts were recorded before any prose existed, and the registered thesis survived; one prediction broke on its discriminator with its inference intact. Honesty score −11, reported at a loss: one point for every research step the platform could not carry. The run is sealed on the host that took the writes. The chain establishes the order in which the work was recorded. It does not establish that any source cited here is correct.

sha256 3c5421a91463c2a3b281cdd37839b08fd4bd82a8285fc5e1c72be495c3e315a1