Is the ban leaking? Russian rough diamonds after the G7/EU ban
Alternative-track signal (R72). Runs beside the Tier-1 exposure score and is never folded into it. Research, not investment advice. Origin-relabelling is inferred from statistical implausibility, never asserted of any single parcel.
Verdict
At the EU border the ban holds; the laundering fingerprint sits one hop upstream, and in 2025 it stopped.
- Antwerp's door held. Belgium's imports of non-industrial rough (HS 710231)
from Russia went 12.1 M ct (2022) → 0 (2024, 2025). The obvious transit hub did not take over either: UAE → Belgium fell 8.4 M ct → 0.10 M ct (−98.8 %). No zero-production origin surged into Belgium.
- The relabel ran through Armenia, which mines no diamonds (Kimberley Process
2024: production 0.00 ct). Armenia's rough imports from Russia rose 0.24 M ct (2021) → 2.22 M ct (2024), ×9.1. Its rough exports to the UAE rose $3.1 M → $197.3 M, ×64. In 2024 Armenia exported 2,821,157 ct of rough against 2,824,094 ct imported. That is 99.9 % carat-for-carat, which fits a pass-through rather than a cutting industry.
- Then the pipe closed. In 2025 Armenia reports zero rough imports from
Russia, and its exports to the UAE are $8.1 M (−96 %). The data doesn't say why.
- What this means for the product: a Tier-1 origin screen that reads
"Belgium: 0 % Russian" is correct and still misses the point. Russian rough (45 % of world carats, KP 2024) entered the UAE market under an Armenian flag from 2022 to 2024. That is the market the G7 traceability regime exists to police.
The implausible-origin tables
Source for every trade figure: UN Comtrade public API, HS 710231 (diamonds, non-industrial, unworked or simply sawn/cleaved/bruted), annual, reporter-side, rows reduced with scripts/py/comtrade_dims.py::dedupe_dimensions. Carats are Comtrade altQty. Pulled 2026-09-30.
1. The EU buyer: Belgium's imports (reporter 056)
| Origin | 2022 | 2023 | 2024 | 2025 |
|---|---|---|---|---|
| World | $8,064.0 M / 48.24 M ct | $5,351.1 M | $3,265.4 M | $2,440.5 M |
| Russia | $1,360.0 M / 12.11 M ct (16.9 %) | $308.3 M / 1.92 M ct | 0 | 0 |
| UAE (transit hub) | $1,640.2 M / 8.40 M ct (20.3 %) | $1,067.1 M / 5.60 M ct | $195.2 M / 1.20 M ct | $27.0 M / 0.10 M ct (1.1 %) |
| Botswana | $1,177.2 M (14.6 %) | $864.2 M | $618.9 M | $849.2 M (34.8 %) |
| Angola | $719.6 M (8.9 %) | $580.1 M | $548.0 M | $521.1 M (21.4 %) |
| DR Congo | $68.5 M (0.8 %) | $33.6 M | $277.1 M | $283.3 M (11.6 %) |
The replacement origins that gained share (Botswana, Angola, DR Congo) are all mining countries. Belgium shows no Layer-1 fingerprint.
2. The relabel: Armenia (reporter 051), rough in vs rough out
| Flow | 2021 | 2022 | 2023 | 2024 | 2025 |
|---|---|---|---|---|---|
| Imports from Russia | $18.5 M / 243,443 ct | $38.7 M / 341,843 ct | $122.9 M / 1,239,180 ct | $231.1 M / 2,224,200 ct | 0 |
| Imports, all origins | $68.6 M / 643,502 ct | $142.3 M / 1,254,342 ct | $287.5 M / 2,501,408 ct | $352.2 M / 2,824,094 ct | $56.1 M / 86,484 ct |
| Russia share of Armenian imports (ct) | 37.8 % | 27.3 % | 49.5 % | 78.8 % | 0 % |
| Exports to UAE | $3.1 M / 293,002 ct | $50.4 M / 1,116,819 ct | $141.0 M / 2,549,910 ct | $197.3 M / 2,145,540 ct | $8.1 M / 35,806 ct |
| Exports, all destinations | $5.0 M / 294,533 ct | $57.1 M / 1,122,536 ct | $154.4 M / 2,678,903 ct | $235.2 M / 2,821,157 ct | $8.3 M / 36,127 ct |
| Carats out ÷ carats in | 46 % | 89 % | 107 % | 99.9 % | 42 % |
Other supplier lines in Armenia's import data are also implausible. Curaçao (M49 531) sent $80.4 M in 2024, and Belarus sent $24.2 M in 2022. Neither mines diamonds. They are listed here as open questions, not claims.
Mirror check (UAE, reporter 784): the UAE reported imports from Armenia of $49.3 M (2022) and $225.6 M (2023), against Armenia's $50.4 M and $141.0 M. The UAE also reported $1,503.3 M of rough from Russia directly in 2023, up from $1,388.1 M. Direct Russia→Dubai trade was never banned. The UAE has not reported 2024 or 2025 to Comtrade.
3. Production baseline (Kimberley Process, 2024 Global Summary)
| Production | Rough imports | Rough exports | |
|---|---|---|---|
| Armenia | 0.00 ct | 9,378,104 ct / $2,073.0 M | 3,924,377 ct / $250.7 M |
| Russian Federation | 58,543,796 ct / $5,084.6 M | — | 30,202,166 ct / $2,607.6 M |
| World | 129,124,016 ct / $11,282.3 M |
Source: kimberleyprocessstatistics.org, 2024GlobalSummary.pdf (public statistics). Russia = 45.3 % of world carats, 45.1 % of value.
Unreconciled: KP records Armenia importing 9.38 M ct / $2.07 bn in 2024, about 3× the carats and 6× the value in Comtrade HS 710231 ($352.2 M / 2.82 M ct). KP covers all rough (7102.10/.21/.31), and free-zone or transit treatment may differ. We have not reconciled this gap. If KP is closer to right, the Armenian throughput above is a floor.
Transmission chain
`` Russia (45 % of world rough carats; US-banned 2022 under EO 14068, EU direct ban 1-Jan-2024, indirect ban phased from 1-Mar-2024) │ rough, HS 710231: 0.24 M ct (2021) → 2.22 M ct (2024) ▼ Armenia (0.00 ct mine production; ~99.9 % carat pass-through in 2024) │ exported as rough, "Armenia" as last-export country ▼ UAE (not a sanctioning state; also takes ~$1.5 bn/yr direct from Russia) │ mixed-parcel re-export; cutting mostly in India ▼ G7/EU buyer: rough blocked at Antwerp (UAE→BE −98.8 %); the residual risk is polished stones cut from these parcels ``
The weak link is the last-export-country convention on rough parcels. Once Russian goods are mixed in Dubai with Armenian-flagged goods, the origin can only be recovered by provenance tracing, not by customs statistics. The EU's indirect ban and the G7 traceability regime target exactly that gap.
What the Tier-1 score cannot see
A company screen that asks "what share of your supplier's imports is Russian origin?" returns zero for every Antwerp-sourced parcel. The signal is upstream and in a non-sanctioning hub, so it shows up only in the divergence between Armenia's zero production and its 2.8 M ct of exports. It stays on this alternative track, as the dual-score rule requires.
Common-ownership tell
Not traced — N. This tick did not identify the Armenian importers/exporters or check them against the Armenian State Register of Legal Entities for Russian ownership (e.g. ALROSA affiliates). Open question, and the next step if this case is refreshed.
Caveats
- Inference, not proof. The pattern fits relabelling. Any single Armenian
parcel may be legitimate: Armenia has a historic cutting sector, and Russian rough into Armenia is not illegal under Armenian law.
- The Armenian leg may not breach any G7 rule directly. The UAE is not a
sanctioning jurisdiction. The circumvention risk lies in the onward stones that reach G7 buyers, and this dataset does not follow those.
- Polished not measured. The G7 indirect ban targets polished stones cut from
Russian rough (HS 710239). This case measures rough only.
- Data gaps. The UAE has no 2024–2025 Comtrade reports. The 2025 Armenian
collapse is unexplained. It could be enforcement, a reclassification, or a reporting change. The KP vs Comtrade gap (above) is unreconciled.
- Register-state: "no Layer-1 surge into Belgium" describes the reporter
data we pulled. It does not mean no Russian stone reached the EU.