Is the ban leaking? Antimony trans-shipment after China's Dec-2024 US export ban
Trade-flow companion to the price wedge (R72). The wedge measures scarcity (what a controlled material costs ex-China). This case measures the other half: is the control actually holding, or is banned material re-entering the US through a laundered origin? When a material is worth origin-laundering, the shortage is real and official bilateral trade data is actively misleading. This is a DUAL-SCORE / alternative-track signal — never folded into any Tier-1 exposure score. The divergence between "who China says it stopped selling to" and "who suddenly started selling to the US" IS the signal. Research, not investment advice; origin-relabelling is INFERRED from statistical implausibility plus public ownership records, never asserted as smuggling on any single shipment.
Live chart + method: [/situation-room → "Is the ban leaking?"](/situation-room).
Verdict
On 3 December 2024 China's MOFCOM (Announcement 46) banned exports of antimony (with gallium and germanium) to the United States — the first time a sole-dominant supplier cut off a critical-mineral input bilaterally with no Western refining alternative inside the ban window. US customs data since shows the textbook trans-shipment fingerprint:
| Antimony oxides (Sb₂O₃), US imports, gross wt | 12 mo pre-ban (Dec-23→Nov-24) | 12 mo post-ban (Dec-24→Nov-25) | move |
|---|---|---|---|
| China (the banned origin) | 14,687 t | 2,146 t | −85% |
| — China's share of all US oxide imports | 53% | 6% | collapsed |
| Thailand (≈0% of world antimony) | 1,820 t | 3,821 t | +110% |
| Mexico (≈0% mine, one toll smelter) | 200 t | 1,278 t | +539% |
| — Thailand + Mexico combined share | 7% | 14% | doubled |
| World total (panic restocking) | 27,553 t | 37,431 t | +36% |
Source: UN Comtrade (reporter USA, monthly, HS 282580), pulled via the free public preview API; committed artifact `data/intelligence/antimony-circumvention.json`. Production shares: USGS Mineral Commodity Summaries 2025/2026.
China's direct line to the US did not just fall — it near-vanished (several 2025 months under 40 t, Dec-2025 = 0 t). Meanwhile the fastest-growing origins are exactly the ones that cannot produce the material: Thailand appears in neither the USGS antimony mine table nor the refining table, yet became a 600–800 t/month supplier in Q1-2025; Mexico mines effectively no antimony and has a single toll smelter (US Antimony Corp, Madero) that can process feedstock of any origin. A surge from a country with no domestic ore is the classic relabelling tell.
Recomputed 2026-10-03, freshest available month (Jul-2026) plus the 7 months already on file since the original post-ban window closed. The frozen pre/post-ban table above is unchanged — this is what happened in the 8 months after it (Dec-2025 → Jul-2026, HS 282580 oxide, monthly):
| Origin | Dec-2025 → Jul-2026 total | Monthly avg | vs. original post-ban avg (Dec-24→Nov-25) |
|---|---|---|---|
| China | 350 t | ~44 t/mo | ↓ a further ~75% below the already-collapsed 179 t/mo |
| Thailand | 3,556 t | ~445 t/mo | ↑ ~40% above its own post-ban average of 318 t/mo — the surge has grown, not faded |
| Mexico | 0 t | 0 t/mo | gone dark: zero in all 8 months, both HS 282580 (oxide) and HS 811010 (unwrought metal) — after 1,278 t oxide in the original post-ban window |
One of the two implausible-origin legs has shut completely; the other has not reverted to baseline, it has grown. That is evidence against the "panic restocking settles down" reading and for the corporate-pipe reading below — Thailand is the leg the Reuters-traced Youngsun group pipe runs through (Guangxi Youngsun → Thai Unipet → Youngsun & Essen, Texas); Mexico's single toll smelter had no equivalent standing group relationship, which is consistent with it being the leg that stopped. Source: same Comtrade reporter/HS lines, periods 202512–202607; Jul-2026 pulled via the subscription API (`COMTRADE_API_KEY`), Dec-2025–Jun-2026 already in the committed artifact.
The corporate pipe (the common-ownership tell)
The strongest signal is not implausible volume alone — it is common ownership across the origin-country exporter and the US importer. Reuters (July 2025) traced one vertically-integrated group:
> Guangxi Youngsun Chemicals (China — producer) → Thai Unipet Industries > (Thailand subsidiary — the post-ban exporter of record) → Youngsun & Essen > Corp (Texas — importer / distributor of record).
Same group, name-consistent across the chain. A producer with a pre-existing "domestic-and-overseas dual-track" model simply re-routed internal group volume through its own offshore leg. That is harder to regulate than random trans-shipment — each entity obeys its own jurisdiction's law, so the corporate structure itself is the workaround. The volume-implausibility layer says "a ban is leaking"; the ownership layer says "and here is the corporate pipe it is leaking through."
Entity chain: Reuters investigation, July 2025 (traced consignor/consignee names across the chain) plus the companies' own public materials. Presented as the worked, publicly-reported example of the mechanism — NOT a per-shipment accusation beyond what the public record states.
Why it matters for the buyer
1. Official bilateral data lies during a control. A risk team reading "US no longer imports antimony from China" would conclude the dependency is severed. The flow data says the dependency is intact — it just wears a Thai or Mexican label now. Any exposure model keyed on stated country-of-origin is compromised for a controlled material; the implausible-origin cross-check is the correction. 2. It confirms the control is binding. Origin-laundering only happens when the material is genuinely scarce and valuable ex-China — the same thing the price wedge prices. Flow re-routing and a widening wedge are two independent readings of one fact: the tap is real. 3. It is a leading indicator of enforcement risk. As US Customs and the EU tighten origin-verification, the laundered leg is where a supply chain breaks next. A buyer sourcing antimony oxide "from Thailand" today should know the probable true origin — and the entity behind it.
Method & honesty rails
- Trade data: UN Comtrade, US imports, monthly, HS 282580 (antimony oxides)
and HS 811010 (unwrought antimony). Net weight is gross product weight of the oxide, not contained Sb. Late months lag ~2–3 months and are omitted.
- Recompute replication (2026-10-03):
reporterCode=842&flowCode=M, HS
282580/811010, periods 202512–202607, subscription API (https://comtradeapi.un.org/data/v1/get/C/M/HS, no partner filter — the response carries the full partner breakdown in one call), deduped with scripts/py/comtrade_dims.dedupe_dimensions (max-per-partner across partner2/mot/customs combinations, never summed). Aug-2026 onward not yet reported as of this recompute.
- Alternative-track only: this never touches
buyerRelativeScoreor the
base exposure — it sits next to them, exactly like the China–West price wedge.
- Inference, not accusation: relabelling is inferred from an origin's ~zero
USGS production plus public ownership records. No individual shipment is asserted to be illegal beyond what the cited public record (Reuters, company filings) states.
- Generalises (R72 Slice 2): the same two-layer detector — (a) imports from
origin X of controlled material M exceed X's plausible production baseline post-control; (b) exporter-of-record ↔ importer-of-record common ownership — applies across the MOFCOM-controlled set (Ga, Ge, graphite, magnet REs). This antimony case is the hand-traced template.