Is the ban leaking? Western aircraft parts into Russia after the Feb-2022 FDP rule

Trade-flow companion to the price wedge (R72). This is a DUAL-SCORE / alternative-track signal — never folded into any Tier-1 exposure score. The divergence between "who the West says it stopped selling to" and "who suddenly started selling to the buyer" IS the signal. Research, not investment advice; diversion is INFERRED from statistical implausibility plus public licensing and designation records, never asserted as smuggling on any single shipment.

Verdict

The aviation control is mode E — end-user falsification / pass-through purchaser, the reverse-direction mode where the control sits on the buyer and the restricted good is the controlling state's own export. It is the corpus's fourth mode-E instance, and it resolves the mode's single biggest weakness.

Mode E's stated flaw is that its denominator is soft: "you cannot deploy this much" depends on installed capacity, which a genuine buildout in the transit country can inflate as cover. Aviation has no such softness, because a Western aircraft part is not merely bought — it must be released to service by a maintenance organisation the regulator has approved by name. EASA publishes that register, free, and the last pre-control snapshot is dated 30 April 2021 — before the measure it is used to test:

Source: EASA-IFP "List of Valid Foreign Part 145 organisations", data of 30-Apr-2021; 332 approvals parsed across 56 countries. Committed artifact `data/intelligence/aircraft-parts-circumvention.json`.

A jurisdiction with zero approvals has no lawful path to consume Western aircraft parts at scale. Whatever it imports is passing through. That is a binary the AI-accelerator and machine-tool mode-E rows never had.

Run the trade data against that denominator and Kyrgyzstan is the fingerprint:

Source: UN Comtrade free public preview API, annual, HS; reporter Kyrgyzstan. 2019 baseline for the Russia line: US$32,183.

Both legs move together, which is the pass-through shape: intake from the West +291% (2021→2023) funding a Russia line that goes from US$5,716 to US$2.56m — a 447× increase — in a country that cannot legally fit any of it to an aircraft. Armenia shows the same shape from a smaller base: airframe exports to Russia US$0 in both 2019 and 2021 → US$445,554 in 2022, which was 99.2% of Armenia's entire world export book for the line, while its US intake rose from US$77,020 to US$382,419.

The control

Three instruments, all effective within 48 hours of the invasion:

• US — BIS interim final rule 87 FR 12226 (published 3 Mar 2022,

effective 24 Feb 2022), adding EAR § 746.8: a licence requirement with a policy of denial for CCL Categories 3–9 (including aerospace, avionics and propulsion) to Russia, plus two Foreign Direct Product rules (§ 734.9(f), § 734.9(g)) extending US jurisdiction to foreign-made items that are the direct product of US technology.

• EU — Council Regulation (EU) 2022/328 of 25 February 2022, inserting

Article 3c into Reg 833/2014: "It shall be prohibited to sell, supply, transfer or export, directly or indirectly, goods and technology suited for use in aviation or the space industry, as listed in Annex XI, whether or not originating in the Union, to any natural or legal person, entity or body in Russia or for use in Russia."

• UK — The Russia (Sanctions) (EU Exit) (Amendment) (No. 4) Regulations 2022,

a parallel aircraft/parts prohibition.

Why this control inverts the detector

Every origin-control case in this corpus (antimony, gallium, graphite) asks: which ~0%-capacity country suddenly claims to produce this? The evader's move is to change the goods' country of origin.

The aviation measures are destination controls that travel with the item's technology origin. The FDP rule and Art. 3c's "whether or not originating in the Union" attach to the part wherever it goes and whoever re-exports it. A Boeing 737 brake assembly is a US-origin item in Dubai exactly as it is in Seattle, so relabelling origin launders nothing. What gets laundered is the end destination — a field that exists in no customs series anywhere.

The fingerprint therefore inverts from origin-implausibility to destination-implausibility, and the denominator changes from refining capacity to lawful consumption capacity. That is where the Part-145 register does the work.

> New rule — the licence-register denominator. Mode E's denominator is soft > only when you look for it in economic data (installed capacity, baseline > demand), which a genuine buildout can inflate. Look instead for a > regulator-published licence register in the controlled good's own domain: > airworthiness approvals, pharmacy licences, reactor operating licences, spectrum > grants. Where one exists, mode E's test hardens from "you cannot deploy this > much" to "you cannot lawfully consume any of this," and the register is > usually free, named, and dated before the control.

The comparison set — large is not the same as implausible

The UAE moves by far the most absolute value, and the engine line tells the same story for China: exports of HS 8411.91 (parts of turbojets and turboprops) to Russia went US$1,069,356 (2021) → US$65,359,741 (2023) → US$87,605,998 (2024), an 82× increase, while China's own intake of those parts from the world rose from US$2.85bn to US$6.46bn. Those are the substantive channels by volume.

But neither is implausible. Both jurisdictions hold large genuine maintenance industries, and 3–4% of a real global book going to a large neighbouring aviation market is not, on its own, evidence of anything. The implausibility test fires on Kyrgyzstan and Armenia precisely because it is scale-free: it asks what fraction of a jurisdiction's book is one destination, against whether that jurisdiction may lawfully touch the goods at all.

Türkiye: the honest negative, and the paper-flag corollary

Türkiye is the corpus's cleanest negative on a jurisdiction that everyone names. Its Russia line is flat and never exceeds 0.25% of its world book in any year 2019–2025 (peak 0.22%, in 2021 — before the control), and its intake of US airframe parts fell from US$316.9m (2021) to US$15.1m (2023).

Yet the US Department of Justice established that Oleg Patsulya and Vasilii Besedin, through MIC P&I, LLC, "attempted to deceive U.S. industry and government officials by claiming that the end user for their purchases was in Turkey rather than Russia, and then attempted to illicitly transship items through the Maldives," receiving at least US$4,582,288 from Russian airlines through Turkish bank accounts. Patsulya was sentenced to 70 months.

The two facts are not in tension — together they are the sharpest lesson in this case:

> The paper-flag corollary to mode E. A falsely declared destination leaves > no trace in the declared country's trade statistics, because the goods never > arrive there. Türkiye was the lie, not the route. Only physical routing > (the Maldives) is capable of showing up in customs data at all — and only if > the routing jurisdiction reports. Never treat a country named in an enforcement > narrative as a country the trade data should light up.

The unobservable legs — labelled, not filled

• Maldives. Zero EASA approvals, and airframe-parts intake from the world of

US$34.6m (2021) → US$39.2m (2023) → US$50.7m (2025), with US-origin intake doubling in the control year (US$8,943,307 → US$19,899,792). It reports an export line for one year only (2021, US$23,763). The onward leg — the leg DOJ says was used — is structurally unobservable in Comtrade. This is a gap, not a finding: we cannot say where those parts went.

• India. Reports nothing at HS 8807.30 or 8411.91 in either direction in

the free preview API for any year 2019–2025. Public reporting has described Indian intermediaries in this trade; this case can neither confirm nor quantify that, and does not assert it.

• Russia itself. Suspended publication of its customs data in April 2022.

There is no Russian import line at any price. Every Russia figure here is mirror data — the partner's reported export. Mirror data understates by construction: a consignment declared for a third destination and diverted never appears on a hub→RU line. These are floor estimates.

• UAE 2024–2025 and China 2025 had not filed these lines at the time of

the pull. They are shown as not reported, never as zero.

Layer 2 — the common-ownership tell returns N, and that is structural

No ownership hop is asserted here, because none could be traced to a registry. That absence is informative rather than inconclusive, and it follows from the mode:

> In an origin control, the evader must be the dominant producer or its > affiliate — nobody else has the molecules — so ownership is load-bearing and a > namesake shell abroad is the strongest single tell (Youngsun → Thai Unipet). > In a destination control the item is freely available from thousands of > distributors worldwide. The evader needs no relationship to the producer at > all — only a forwarding function. So the ownership tell decays into a much > weaker freshly-incorporated-forwarder tell.

What the public designation record does show, verbatim from primary sources:

Cargoline was founded in March 2022 — one month after the control took effect. That is the temporal analogue of the ownership tell, and it is the one layer-2 signal that survives in a destination control: not who owns the forwarder, but when the forwarder came into existence relative to the measure. Note also that Ukon's named CEO is an Azerbaijani national, not a Russian one — consistent with the forwarding-function reading and with the absence of an ownership pipe.

The enforcement grade — an externally checked result

The corpus's most valuable results are the ones a regulator grades independently. This is one:

1. The detector points at Kyrgyzstan and Armenia on 2022–23 data. 2. BIS added 71 entities on 19 May 2023 under exactly three destinations — Russia, Armenia and Kyrgyzstan (FR Doc 2023-10684), the Armenian and Kyrgyz entries "for facilitating diversion of controlled goods to Russia". OFAC designated Cargoline on 20 Jul 2023 and Ukon on 23 Feb 2024. 3. The flow then dies: Kyrgyzstan's airframe line to Russia falls from US$2,556,275 (2023) to US$192,150 (2024) and its world book for the line collapses from US$2.91m to US$249,233; Armenia's falls from US$445,554 (2022) to US$835 (2024), and its engine line from US$1,076,443 to US$394.

The detector selected the jurisdictions the regulators then charged, on data it was not fitted to, and the charged route closed. That is the same shape as the plywood mode-C⁻ result — a test reproducing an outcome it did not know.

…and then it moved headings

The route did not die. It migrated:

Kyrgyzstan's 2025 engine-parts line to Russia is US$4.76m — larger than its airframe line ever was at its 2023 peak, and it is 88% of the country's entire world book for the heading, in a jurisdiction that still holds zero airworthiness approvals.

> The enforcement-migration tell. This is not mode F. Mode F is the exporter > shifting a physical parameter to fall outside a measure's scope; here the > scope never changed — both headings were prohibited throughout. What moved was > enforcement attention. When a designation round closes a route, re-run the > detector across adjacent headings in the same prohibition before recording the > control as holding. A flow that vanishes from the line you are watching has not > necessarily stopped.

The transmission chain

`` US / EU manufacturers (US-origin content — FDP travels with the part) │ lawful sale to a third-country buyer on a declared end-use ▼ Pass-through jurisdiction with ZERO airworthiness approvals (Kyrgyzstan, Armenia; Maldives leg unobservable) • incorporated after the control in at least one designated case • no lawful path to fit the part to any aircraft ▼ Russian airlines / MRO (30 EASA Part-145 approvals severed by the control — Aeroflot, Rossiya, S7 Engineering among them) ``

What it implies for the Tier-1 blind spot

A Tier-1 exposure score reads official bilateral trade and sees a Western aerospace supply chain that stopped at the Russian border in February 2022. Three things that score cannot see:

1. The end-user field does not exist in customs data. Mode E's diagnostic is structurally absent from every trade series on earth. A score built on origin and destination codes is not merely incomplete here — it is reading a field that cannot carry the signal. 2. Country-level aggregation destroys the tell. Kyrgyzstan's entire airframe book is a rounding error against the UAE's; only the ratio (88% of a small book) and the licence register (zero approvals) fire. A score that weights by trade value will rank the UAE first and never see Kyrgyzstan. 3. Route closure is not flow closure. The 2024 collapse would read as "control holding" to any monitor watching HS 8807. The 2025 engine line says otherwise.

Caveats

• Inference, not proof. Nothing here asserts that any specific shipment was

unlawful. Implausibility is a statistical property of aggregates; individual consignments may be entirely licensed and legitimate.

• Mirror data only, and it understates. Russia publishes nothing since Apr-2022;

every Russia figure is a partner's reported export and is a floor.

• HS splice. HS2022 moved aircraft parts from heading 8803 to the new heading

8807 on 1 Jan 2022. 880330 (2019–2021) and 880730 (2022–2025) are the same line either side of that break and are spliced here with the break labelled; HS 8411.91 is unchanged across editions and is the unspliced control series.

• Reporting gaps are labelled, never zeroed. UAE 2024–25, China 2025 and India

throughout are shown as not reported. The Maldives export leg does not exist in the data at all.

• The Part-145 denominator is EASA's, not the FAA's. A jurisdiction with zero

EASA approvals could in principle hold FAA or national approvals; the EASA register is used because it is free, published, and dated before the control. It bounds EU-facing lawful consumption, which is the relevant bound for Art. 3c, and is used here as a strong indicator rather than a legal finding.

• 2025 is incomplete across several reporters and will revise.

Sources

• UN Comtrade free public preview API (comtradeapi.un.org/public/v1/preview/C/A/HS),

annual, HS — reporters AE, CN, TR, KG, AM, KZ, UZ, IN, MV; HS 8803.30/8807.30 and 8411.91; flows X and M. Committed artifact data/intelligence/aircraft-parts-circumvention.json; fetcher scripts/py/iptm/fetch_aircraft_parts_circumvention.py.

• EASA-IFP, List of Valid Foreign Part 145 organisations, data of 30-Apr-2021 —

<https://www.easa.europa.eu/sites/default/files/dfu/EASA-IFP-ListofValidForeignPart145organisations(WEB)_2021-04-30-11-04-06.pdf>

• US BIS, Implementation of Sanctions Against Russia Under the EAR, 87 FR 12226,

FR Doc 2022-04300 (3 Mar 2022, effective 24 Feb 2022).

• Council Regulation (EU) 2022/328 of 25 February 2022 (inserting Art. 3c into

Reg (EU) No 833/2014), CELEX 32022R0328.

• US BIS, Addition of Entities to the Entity List, FR Doc 2023-10684 (19 May 2023).
• US Treasury press release jy1636 (20 Jul 2023) — OSOO Kargolayn.
• US Treasury press release jy2117 (23 Feb 2024) — Ukon, Linker FZE.
• US DOJ — United States v. Oleg Patsulya and Vasilii Besedin (D. Ariz.), MIC P&I, LLC.

Anchor action: `2022-02-24-us-bis-ear-russia-ccl-fdp-controls`.