Is the preference leaking? Gold declared as platinum alloy into India

Trade-flow companion to the price wedge (R72), and the corpus's first *preference-side* circumvention row. Every other case in this file asks whether a restriction is holding. This one asks the mirror question: whether a tariff preference is being filled with a good it was never written for. A DUAL-SCORE / alternative-track signal — never folded into any Tier-1 exposure score. Research, not investment advice; misdeclaration is INFERRED from statistical implausibility plus the buyer government's own remedial instruments, never asserted as an offence against any named party or shipment.

Verdict

India's platinum imports carry a payload that is not platinum. Between 2021 and 2025, India's declared imports of unwrought and semi-manufactured platinum (HS 711011 + 711019) rose from $195m to $3,011m — a 15× increase — while imports from every country that actually mines platinum stayed flat. The entire increase came from four countries that the USGS names as producing no platinum at all: the UAE (90.3% of India's platinum imports in 2022, 68.5% in 2024), then — within nine months of India restricting the UAE route — Thailand (85.0% in 2025), Tanzania (9.2%) and Indonesia (1.9%).

The payload is identifiable from price alone. Genuine producers sell to India at $30–37k/kg in every year of the window. The surging origins invoice at $77–84k/kg — within 1–2% of the LBMA average gold price. The good moving under India's platinum preference is priced as gold because it substantially is gold: an internal customs assessment quoted by The Reporters' Collective records "platinum alloy sheets containing gold more than 90%".

This is not a covert route. It was lawful on its face — World Customs Organization classification rules let a gold-dominant alloy be declared a platinum alloy on a low single-digit platinum content — and the buyer government closed it the only way such a route can be closed: by redefining the tariff line, four times in thirteen months. Each closure is visible in the customs data as an origin rotation rather than a volume decline. The preference did not stop leaking; it changed flag.

Recomputed 2026-10-02: the flag changed a fifth time, and this time we have the data. The HS 7110 route (this section) is now fully closed from all three surge origins — Thailand, Tanzania and Indonesia reported zero further 7110 platinum-alloy shipments to India after mid-2025. The leak did not stop; it jumped headings, to HS 7114 ("articles of goldsmiths'/silversmiths' wares"). Thailand alone declared $1.245bn of HS 711419 ("other precious metal, not silver") to India in January–February 2026, then $34.7k in March and zero since — a 99.995% drop in one month. See "Recomputed 2026-10-02 — the leak jumps to HS 7114" below.

Layer 1 — volume and unit-value implausibility

India's imports of platinum, HS 711011 (unwrought/powder) + 711019 (other semi-manufactured), UN Comtrade, annual, value and declared net weight:

Source: UN Comtrade subscription API, reporter India (699), flow M, HS 711011 + 711019, periods 2021–2025, canonical-row selection per scripts/py/comtrade_dims.py (aggregate-if-present-else-max, to avoid the partner2/mot/customs replication trap).

The zero-capacity denominator. USGS Mineral Commodity Summaries 2026, Platinum-Group Metals, names the world's platinum mine producers as the United States, Canada, Russia, South Africa and Zimbabwe, plus "Other countries" at 3,860 kg (2024). World platinum mine production was 179,000 kg in 2024 and 170,000 kg (est.) in 2025. The 2025 edition lists candidate producing countries as Australia, Canada, China, Colombia, Ethiopia, Finland, Russia, Serbia, South Africa, the United States and Zimbabwe. The UAE, Thailand, Tanzania and Indonesia appear in neither edition, in neither list. None of the four has a platinum mine or a primary PGM refinery of record.

The unit-value test is the decisive leg, and it is internal to the dataset. The control group is in the same table: South Africa, the UK and Germany invoice India at $30.3–36.9k/kg in every year 2021–2025 — that is what platinum costs. The four surging origins invoice at $77.2–83.6k/kg. Against the LBMA average gold price of $2,386/oz in 2024 ($76,711/kg), the 2024 surge cells price at:

• Thailand $78.4k/kg = 102% of the gold price
• Tanzania $77.3k/kg = 101% of the gold price
• Indonesia $77.2k/kg = 101% of the gold price

Three origins, three currencies of evidence, one number: the consignments are invoiced as gold to within measurement noise, and they are 2.5× the price of the platinum they were declared as. For 2025, against an LBMA H1-2025 average of $3,070.86/oz ($98,730/kg), Thailand's $83.6k/kg is 85% of gold — consistent with a gold-dominant alloy rather than pure gold, which is exactly what the customs assessment describes.

Volume, second. Thailand's 2025 declared 30.59 t is 18% of world platinum mine production from a country with no mine. That alone is disqualifying, and it does not depend on any contested cell.

Two anomalous cells, flagged not smoothed

The UAE 2023 ($3.4k/kg) and 2024 ($9.4k/kg) unit values match neither platinum nor gold, and are reported here as anomalies rather than findings. Two readings, and the conclusion survives both:

1. The declared net weight is overstated for those cells, in which case the value line is the reliable one — and it is independently corroborated (below) — while the weight is unusable. 2. The weight is real, in which case the UAE's declared 2024 figure of 185.55 t is 104% of the entire world's platinum mine production (179,000 kg), and the consignments cannot be platinum in any quantity approaching the declaration.

Either the weight is wrong or the commodity is wrong. We do not know which, and we do not publish the 104% as a finding — it is the second branch of a disjunction, not a measurement. The GATE-1 core of this case rests on the 2024 Thailand/Tanzania/Indonesia cells and the 2025 Thailand cell, all four of which are internally consistent.

Independent cross-check on the UAE leg

Two fully independent datasets agree to 1.4%:

The Reporters' Collective further reports that of that ₹24,000 crore, an estimated ₹21,000 crore was actually gold, and about ₹1,700 crore in taxes was avoided — with over ₹10,000 crore arriving in the two months before the July-2024 budget, which is consistent with this dataset's 2024 UAE value spike of $1,736.8m.

Recomputed 2026-10-02 — the HS 7110 route closes completely, and the leak jumps to HS 7114

UN Comtrade subscription API, reporter India (699), HS 711011+711019, monthly, partners UAE/Thailand/Tanzania/Indonesia (plus controls South Africa/UK/Germany), deduped per scripts/py/comtrade_dims.dedupe_dimensions (max-per-corridor, never summed). Two findings not visible in the annual table above.

1. The three surge origins went silent on HS 7110 five months before this case was written, not after. Pulling Jan–Jul 2025 alone reproduces each surge country's entire 2025 annual total exactly — Thailand $2,558.02m/30.590t, Tanzania $276.88m/3.450t, Indonesia $58.38m/0.714t, all matching the full-year figures in the table above to the dollar. The controls do not show this: South Africa's Jan–Jul share of its 2025 annual total is 59%, UK 79%, Germany 84% — ordinary scatter for small, steady volumes, not a hard stop. All three surge countries report zero further HS 711011/711019 shipments to India for the rest of 2025 and for every month of Jan–Jul 2026 pulled for this update — the most likely closing event is DGFT Notification 18/2025-26 (17-Jun-2025), which extended the gold-content restriction beyond platinum to the sibling PGM alloys; its text was already in the instrument table, but this is the first confirmation that the base platinum-alloy heading went quiet, not just its Pd/Rh/Ir siblings.

2. The leak reopened in January 2026 under HS 7114 ("articles of goldsmiths'/silversmiths' wares"), exactly the heading the original case flagged as unmeasured, and it is not small. HS 711419 ("other precious metal, not silver"), Thailand → India:

Two months, $1.245bn, at a unit value 1.4–1.7× the period's gold price — consistent with finished jewelry (fabrication premium over bullion), the same direction as the original platinum-alloy unit-value test but a different good. Then a 99.995% drop in March and zero for the four months after. The timing matches DGFT Notification No. 58/2025-26 (5-Feb-2026, restricting ITC-HS 71141920 "articles of platinum" specifically) biting within roughly one month — faster than the 7110 closures above, consistent with an already-aware regulator watching the heading it had just named.

A second, smaller, independently-corroborating rotation: HS 711411 ("silversmiths' wares of silver"), Thailand → India, $28.58m/12.069t (Jan-2026), $28.33m/10.402t (Feb-2026), $19.63m/7.506t (Mar-2026) — $76.5m over 30.0 tonnes in three months, unit value ~$2.4-2.7k/kg, consistent with silver (not gold-alloy) jewelry. This is the primary-sourced figure behind the GTRI/Tribune India citation below, which gave the same rotation as "4 t → 40 t" without dating it — the dating is new here.

Neither 711419 nor 711411 shows any Thailand volume in 2025 (checked Jan-2025 through Dec-2025); both are new in January 2026. Not traced here: whether the same importers/petitioners named in Layer 2 below are party to the 7114 leg — no ownership claim is made.

Layer 2 — the ownership tell is N, and its absence is informative

No common-ownership chain is traced, and none is expected. This is the same reasoning the mode-C⁰ (Russian seafood) row established, applied to a preference: nobody builds a namesake shell to do a lawful thing. Under WCO classification rules the declaration was, on its face, correct — the goods genuinely shipped from the UAE and Thailand, origin was never falsified, and the value-addition requirement was satisfiable with a certificate. A route that needs no misstatement needs no shell. The ownership fingerprint is a signature of evasion; this is arbitrage, and layer 2 is structurally silent on it.

What exists instead is court process, the same substitute the mode-E (AI-accelerator) row had to fall back on. Two companies are named in The Reporters' Collective's account as petitioners in Delhi High Court proceedings concerning the classification: MD Overseas Private Limited and Ausil Corporation Private Limited. Naming them here records that the classification was litigated in public — it is not an assertion that either company did anything unlawful, and litigating a customs classification is an ordinary and lawful act. Other participants are not publicly named.

The instrument chain — a four-step tariff-line chase

The on/off signature is clean, and it is the signature the mode-F (India rice) row established: when Notification 60/2024-25 cut the UAE line, UAE volume did not merely fall — its unit value reverted to the genuine-platinum band ($9.4k/kg → $44.1k/kg on 0.18 t). The cheap disguised payload left and a sliver of real platinum remained. Reversion of unit value on closure is the proof the price signal was the payload, not noise.

And the flow did not stop. Within the same year the origin rotated to Thailand under the India–ASEAN FTA — an independent preference India could not close by amending a bilateral — taking the route from a 10.75% platinum-alloy duty against 15% on gold. GTRI's account (via Tribune India) records the same rotation in the sibling metal: silver jewellery imports from Thailand rose ~10×, from 4 t to 40 t, with Thailand's share going 78% → 98%.

Transmission chain

`` gold (any origin, world market) → alloyed in UAE / Thailand with a low single-digit % of platinum → declared under HS 7110 as "platinum alloy" (lawful under WCO classification) → enters India at the CEPA / AIFTA platinum preference (8.15–10.75%) instead of the gold rate (15–18.45%) → refined in India; the gold is recovered ⇒ duty differential captured; India's platinum import series is corrupted ``

Origin is correct at every hop. What is false is the product identity — and until March 2025 it was not even false, merely permitted.

Why this is a new mode — (K) preference-line alloying

Modes A–J all describe escaping a restriction. This route escapes nothing: it qualifies for something. The good is physically engineered — by adding a trace of an expensive metal to a cheaper one, the inverse of the usual economics — so that it falls inside a preferential tariff line rather than outside a penalised one. It is mode F run in reverse: F shifts a physical parameter to leave a scope; K shifts one to enter a concession.

Three properties distinguish it, and all three are transferable:

1. The detector is the unit value, not the volume. Origin is true, so the implausible-origin test is weak (it fires, but only because these countries have no mines at all). The price test is decisive and needs no external benchmark: the genuine producers are in the same series, at a stable price, every year. Where a preference is defined by composition, read $/kg against the honest suppliers in the same table. 2. Lawfulness inverts layer 2. No shell, no relabel, no ownership pipe — because nothing needs hiding. An ownership tell of N is a prediction of mode K, not a gap in the research. 3. The remedy is a redefinition, so the leak rotates rather than stops. A restriction is closed by naming an origin; a preference can only be closed by redefining the good. India redefined it four times in thirteen months, and each redefinition moved the flow to the next tariff line or the next preference-granting partner. Count the buyer's own amendment cadence: a government amending the same heading repeatedly is publishing the fact that the route is still open.

Corollary for the register: this is the first row where the buyer's remedial instrument is the best evidence of the circumvention. The PIB release for Notification 18/2025-26 states the purpose in the government's own words — closing a route for "importing gold in disguised alloy form."

What it implies for the Tier-1 blind spot

• A country's platinum import series can be mostly gold. Any exposure metric

that reads India's HS 7110 line as platinum demand overstates it by roughly an order of magnitude in 2022, 2024 and 2025. India's genuine platinum imports ran 0.9–1.6 t/yr across the window; the declared series peaked at 197.67 t.

• Preference-side leakage is invisible to a register scoped to restrictions.

Nothing here is an export control. The governing instrument is an FTA schedule, and the decisive fact is a classification rule at the WCO, which no sanctions or export-control feed carries.

• The honest-supplier control group is an asset. Because genuine producers

keep shipping small, correctly priced volumes, every composition-defined preference line carries its own built-in benchmark. This is cheap to run corpus-wide and should be.

Caveats

• Inference, not proof. Misdeclaration is inferred from unit-value

divergence, zero mine capacity, and the buyer government's own remedial instruments. No claim is made about any individual shipment, importer, or exporter. The two named companies were petitioners in litigation; that is not an allegation against them.

• Two anomalous cells (UAE 2023, UAE 2024 unit values) are unresolved and

labelled above. The case does not rest on them.

• Threshold discrepancy, unresolved. Public reporting gives the WCO

platinum-content threshold as 1% (DD News; Tribune India/GTRI) and as 2% by weight (The Reporters' Collective; Business Standard). These are not reconciled here and the primary WCO/ITC-HS text has not been read for this case. The DGFT remedial thresholds are firm from the register: <99% platinum purity (Not. 60/2024-25) and >1% gold content (Not. 18/2025-26).

• The Tanzania and Indonesia preference bases are not traced. Both invoice at

the gold price and neither mines platinum, so the payload finding holds; which preference instrument they enter under is an open question (Tanzania is an LDC and a plausible India DFTP beneficiary; Indonesia is an AIFTA party — neither verified here).

• The 7114 hop is now measured (2026-10-02): $1.245bn, two months, then zero.

See the "Recomputed 2026-10-02" subsection in Layer 1. Comtrade's HS6 codes (711411, 711419, 711420) are coarser than India's ITC-HS 71141920, so this figure is the whole 6-digit basket India→Thailand, not proof every dollar of it is the gold-alloy good the DGFT notification named — the unit-value test (1.4-1.7x the gold price) is consistent with that good but not definitive at this resolution.

• Data lags. Monthly Comtrade data for India now runs through 2026-07

(pulled 2026-10-02); 2026-08 onward is not yet reported. The Apr-2026 notification's effect cannot be distinguished from the Feb-2026 one in this window since the 7114 leg was already at zero before April.

Sources

• UN Comtrade — India (699) imports, HS 711011 + 711019, annual 2021–2025.

https://comtradeapi.un.org/ (subscription API; free preview tier carries the same data at a 500-row cap)

• UN Comtrade (recomputed 2026-10-02) — India (699) imports, HS 711011 +

711019 monthly Jan 2025–Jul 2026 (confirms the H2-2025 close on the surge origins), and HS 711411 + 711419 + 711420 monthly Jan 2025–Jul 2026 (the 7114 migration), partners UAE/Thailand/Tanzania/Indonesia + South Africa/UK/Germany controls, comtradeapi.un.org/data/v1/get/C/M/HS, deduped per scripts/py/comtrade_dims.dedupe_dimensions.

• USGS, Mineral Commodity Summaries 2026 — Platinum-Group Metals (world mine

production 2024/2025e; producer list). https://pubs.usgs.gov/periodicals/mcs2026/mcs2026-platinum-group.pdf

• USGS, Mineral Commodity Summaries 2025 — Platinum-Group Metals (2023

production; candidate producing countries). https://pubs.usgs.gov/periodicals/mcs2025/mcs2025-platinum-group.pdf

• PIB (Government of India) — "Import Restriction on Certain Precious Metal

Alloys Containing Gold", on DGFT Notification No. 18/2025-26 and the prior Notification No. 60/2024-25 of 5 March 2025. https://www.pib.gov.in/PressReleseDetailm.aspx?PRID=2137846

• DGFT Notification No. 58/2025-26 (5 Feb 2026), articles of platinum.

https://content.dgft.gov.in/Website/dgftprod/7e9f69cb-93ac-4cca-bae3-1aafeb109436/Notification%20No.%2058%20English_0001.pdf

• DGFT Notification No. 03/2026-27 (2 Apr 2026), Chapter 71.

https://content.dgft.gov.in/Website/dgftprod/4b3ce088-3baa-45ab-98e3-7e52f85c747b/Notification-chp-71-English.pdf

• DD News (Government of India) — "India imposes import restrictions on

platinum alloys to curb misuse of trade agreements". https://ddnews.gov.in/en/india-imposes-import-restrictions-on-platinum-alloys-to-curb-misuse-of-trade-agreements/

• The Reporters' Collective — "Govt Knew of a Loophole Enabling Traders to

Import Gold Disguised as Cheaper Alloy" (₹24,000 crore / ₹21,000 crore gold / ₹1,700 crore tax; CEPA 8.15% vs gold 18.45%; internal customs assessment quote; Delhi High Court petitioners). https://www.reporters-collective.in/trc/govt-knew-of-loophole-enabling-traders-to-import-gold-disguised-as-cheaper-alloy

• Tribune India / GTRI — WCO 1% platinum-alloy rule; 4 t → 40 t Thai silver

jewellery rotation; India–ASEAN FTA arbitrage; 10.75% vs 15% duty. https://www.tribuneindia.com/news/business/india-puts-restrictions-on-platinum-alloy-imports

• Business Standard / GTRI — India–UAE CEPA precious-metal concerns, 3%

value-addition rule of origin. https://www.business-standard.com/industry/news/india-uae-trade-pact-gtri-expresses-concerns-over-precious-metal-imports-124071600663_1.html

• LBMA gold price averages — 2024 annual $2,386/oz; H1-2025 $3,070.86/oz.

https://www.lbma.org.uk/ (via World Gold Council price data, https://www.gold.org/goldhub/data/gold-prices)

Related

• Anchor action: 2025-06-17-india-dgft-pgm-alloys-import-licensing
• Sibling instruments: 2026-02-05-india-dgft-platinum-articles-import-licensing,

2026-04-02-india-dgft-chapter-71-precious-metals-import-restriction, 2025-06-17-india-dgft-cth-2843-colloidal-precious-metals-import-licensing

• Mode kin: mode F (specification/scope shifting — steel, EVs, rice), whose

reverse this is; mode C⁰ (Russian seafood) for the informative-N ownership argument; mode E (AI accelerators) for court-process-as-evidence.

• Rollup: [circumvention-signals.md](../circumvention-signals.md)