Is the control leaking? Silver after China's Jan-2026 state-trading channel

Trade-flow companion to the price wedge (R72). This is a DUAL-SCORE / alternative-track signal — never folded into any Tier-1 exposure score. Research, not investment advice; origin-relabelling is INFERRED from statistical implausibility plus public ownership records, never asserted as smuggling on any single shipment.

Verdict

The control is not leaking. It is running in the opposite direction from every other row in this register — and the two-layer detector, applied naively, would have got this case exactly backwards.

China narrowed who may export silver, not how much. Its direct supply to the West then collapsed — but no implausible origin appeared to replace it. What appeared instead was a reversal: in the first five months of the control, Hong Kong stopped being the outlet for Chinese silver and became the inlet for Western silver, pulling ~1.9 kt of London, New York and Dubai vault metal into the China complex and re-exporting 799 t of it onward into the mainland.

Layer 1 returns negative (every replacing origin is a genuine producer or refiner). Layer 2 returns negative (no namesake, no shared ownership, none needed — nothing here is unlawful). The finding is the direction of flow.

The control (what it actually does — and does not do)

MOFCOM Announcement No. 68 of 2025 (商务部公告2025年第68号, 26-Oct-2025) sets the application conditions for state-trading enterprises (国营贸易) authorised to export tungsten, antimony and silver in the 2026–2027 cycle, applying to shipments dated 1 January 2026 onward. Silver is in the STE channel for the first time. Qualification requires, verbatim, 2022–2024 silver export performance (「2022—2024年每年均有白银出口实绩」) or, for new applicants, 2024 output of ≥80 t (≥40 t for western regions), plus ISO 9000 and ISO 14000 certification. Applications closed 12 November 2025. MOFCOM published the approved list on 30 December 2025: 15 firms for tungsten, 11 for antimony, 44 for silver.

Two corrections to the obvious reading, both from the primary text:

• It is a channel control, not a volume control. MOFCOM Announcement No. 69

of 2025, which sets 2026 export quota totals, states 「2026年继续暂停磷矿石、白银 出口配额管理」 — quota administration for phosphate rock and silver remains suspended in 2026. There is no silver tonnage cap. Anyone describing this as a "silver export quota" is describing something the register does not contain.

• The list is not closed by construction. Announcement 68 publishes an

application route and a deadline; 44 silver exporters were approved. The restriction is on standing, not on access.

Layer 1 — who replaced China? (all genuine producers)

United Kingdom, imports of unwrought silver (HS 710691), Jan–May, net weight

India, imports of semi-manufactured silver (HS 710692), Jan–May, net weight

The control bit hard on the declared Chinese line — −93% into London, −81% into India. But the replacements are Kazakhstan, the US, Canada, Mexico, Poland and Korea: every one a producing or refining jurisdiction in the USGS table. There is no ~0%-capacity surger. India's total barely moved (−14%) because real metal from real refiners walked straight into the gap.

Why layer 1 must be switched OFF here

Run the standard volume-implausibility test on this table and it flags the wrong countries, loudly:

• Canada shipped 760 t into the UK and India in five months — **253% of its

entire annual mine production** (300 t).

• Kazakhstan shipped 1,086.8 t into the UK alone — **109% of a full year's

mine output** (1,000 t), in five months.

• The United Kingdom mines no silver at all and exported 4,106 t in the

same window.

• The UAE mines no silver and put 269.3 t into Hong Kong (below).

On any other row in this register that pattern is the fingerprint. Here it is the normal state of the market. Refined silver is a fungible bullion good; customs origin on a bullion consignment is routinely the last vaulting jurisdiction, not the mine, and London, Zurich, New York, Toronto, Singapore and Dubai lawfully move above-ground stock that was mined decades ago and elsewhere. A five-month flow exceeding annual mine output is evidence of a vault transfer, not of laundering.

This is the case's methodological contribution: for a fungible, vaulted, above-ground-stock commodity, layer 1 is not merely weak — it is invalid, and it produces false positives on exactly the jurisdictions that are behaving lawfully. The discriminators that survive are direction and the form split.

The finding — the pipe reversed

Hong Kong, imports of unwrought silver (HS 710691), Jan–May, net weight

Three origins that together sent Hong Kong 1.2 t in early 2025 sent it 1,892.9 t in early 2026.

Hong Kong, exports of unwrought silver, Jan–May, net weight — the onward leg

Hong Kong's historic job was to take China's silver out: 99.47% of China's 4,246.7 t of unwrought silver exports went to Hong Kong in 2024, the last year China reported. In the first five months under the control, Hong Kong's single largest silver customer is the mainland itself.

Mirror confirmation of the London leg

The new London→Hong Kong corridor is visible from both reporters, which is the test that matters when one side is a re-export hub:

United Kingdom, exports of unwrought silver, Jan–May, net weight

Hong Kong books 1,082.5 t inbound from the UK; the UK books 1,582.1 t outbound to Hong Kong — a 499.6 t mirror gap (Hong Kong reports 68% of the UK figure), consistent with in-transit timing and re-export accounting. Both sides agree a corridor that did not exist in 2025 carried four figures of tonnes in 2026. Neither side is used alone.

Note also what the UK table says about the baseline: early 2025 was itself an anomaly — London was draining 3,898.7 t to New York ahead of feared US metals tariffs. The 2026 window does not show a normal market reverting; it shows one extraordinary flow (London→New York) being replaced by another (London→Hong Kong).

Mode F test — did the book migrate to an uncontrolled form?

If a channel control bit, the export book could shift into a form or heading the catalogue does not name. It did not:

• Silver powder (HS 710610) — the PV-metallisation form. Japan's imports

from China rose 1.5 t → 5.4 t (+257%); the US and India buy essentially no Chinese powder in either window. China's powder line is small (6 t worldwide in 2024, down from 26 t in 2022) and is not carrying a diverted book.

• Hong Kong's powder exports to the mainland rose 4.0 t → 7.5 t (+88%), and

its powder imports rose 0.9 t → 9.4 t — powder is moving into the complex too, the same direction as the bullion.

No mode-F signature. The forms move together, inward.

The line no silver control reaches

China exported 12,445,358 t of photovoltaic cells and modules (HS 854143) in 2024. Chinese PV metallisation is the largest single industrial use of silver on earth, and that silver leaves the country inside the module, under a heading no metals control touches. This is the structural bypass — and it is entirely lawful, which is why it is stated here as the denominator rather than filed as a circumvention mode. The contained-silver tonnage in that line is NOT computed in this case: doing it honestly needs a sourced g/W loading series, and an unsourced loading assumption would invent a number next to real ones.

What it implies for the Tier-1 blind spot

Every exposure score in this product asks who buys from the controlled origin. On silver in 2026 that question is the wrong way round. A company scored on "China supply dependency" reads safer each month as China's export line falls — while the actual risk, that the world's above-ground silver is being pulled into a jurisdiction that has just narrowed who may ship it out, does not appear in the score at all. A collapsing bilateral import line is being generated by a control and by demand absorption simultaneously, and the exposure score cannot tell those two apart. Alternative-track only; this is not folded into any Tier-1 number.

Open questions (labelled, not filled)

• Kazakhstan → UK (1,086.8 t, 109% of annual mine output). Kazakhstan is a

genuine producer and refiner, so this is not a layer-1 flag on its own. Whether any of it is re-exported Russian metal is not evidenced here and is not claimed. Russia's own line into Hong Kong rose 9.0 t → 78.7 t.

• Round-tripping. Some part of the 799 t Hong Kong re-exported to the

mainland may be Chinese-origin metal returning. Hong Kong's export statistics do not separate re-exports by original origin on this line.

• China's own books. China has not reported to Comtrade since 2024, so every

post-control figure above is mirror data — the importer's own reported imports. That is the correct jurisdiction to query anyway, and here it is the only one available.

Method & honesty rails

• Trade data: UN Comtrade free public preview API

(comtradeapi.un.org/public/v1/preview/C/{A,M}/HS), HS 710610 / 710691 / 710692 / 854143. Pulled by scripts/py/iptm/fetch_silver_circumvention.py → data/intelligence/silver-circumvention.json. Rows dimension-reduced via scripts/py/comtrade_dims.py (summing raw rows multiplies by the partner2/mot/customs combination); the UK export leg was re-pulled through the same de-duplicator after a naive sum produced an impossible WORLD total.

• Weight, never value. Every figure is net weight in tonnes. Silver's dollar

price moved sharply across this window, so value comparisons would show moves that are pure price. No cell is zero-filled; months that failed to return are recorded as absent in the artifact (none were absent in the windows shown).

• Windows: Jan–May 2025 vs Jan–May 2026 — the same five calendar months

either side of the 1-Jan-2026 effective date, so no seasonal splice. May 2026 is the last month the thinnest reporter publishes.

• Production baseline: USGS Mineral Commodity Summaries 2025, silver

chapter (2024 estimates). It lists no mine production for the UAE, Hong Kong or the United Kingdom.

• Inference, not accusation. Nothing in this case alleges unlawful conduct by

anyone. No ownership hop is asserted because none was found and none is required — the flows described are lawful trade.

• Alternative-track only: this never touches buyerRelativeScore or the base

exposure.

Sources

• MOFCOM Announcement No. 68 of 2025 (primary, Chinese) —

https://www.mofcom.gov.cn/zwgk/zcfb/art/2025/art_a1cf9c25add141c18876309e1152c131.html

• MOFCOM, 2026–2027 approved STE exporter list, 30-Dec-2025 (primary, Chinese) —

https://www.mofcom.gov.cn/zwgk/zcfb/art/2025/art_5e7c9b75a7d94a7aab4f168c7e378735.html

• MOFCOM Announcement No. 69 of 2025, 2026 export quota totals — the silver quota

suspension (primary, Chinese) — https://www.mofcom.gov.cn/zcfb/dwmygl/art/2025/art_d939f5e193604b1aaf76aacc441a16f4.html

• UN Comtrade (free public preview API) — all trade tables above
• USGS, Mineral Commodity Summaries 2025, silver —

https://pubs.usgs.gov/periodicals/mcs2025/mcs2025-silver.pdf