Is the tariff leaking? Russian nitrogen after the EU's Jul-2025 fertiliser duty

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 EU says it stopped buying fertiliser from" and where the nitrogen actually still comes from IS the signal. Research, not investment advice; nothing here is asserted to be unlawful — the central finding is that no rule was broken at all.

The anchor action closed with an open question: "whether Russian/Belarusian fertiliser volumes are being re-routed through third countries to circumvent the tariff." This case answers it. The answer is no — and the reason that answer is not reassuring is the point.

Verdict

On 1 July 2025 Regulation (EU) 2025/1227 put an escalating per-tonne duty on Russian and Belarusian fertilisers, on top of the existing 6.5% ad valorem: €40/t on CN 3102 (nitrogenous) and €45/t on CN 3105 (compound), stepping to €315/t and €430/t from 1 July 2028. The walled line duly collapsed. But the duty is drawn at Chapter 31 — fertilisers, and the nitrogen arrives in Chapter 28.

Source: Eurostat COMEXT DS-045409 (reporter EU27_2020, flow=1 imports, QUANTITY_IN_100KG), monthly, matched 11-month Jul→May windows — the regulation's own tariff-year clock, truncated to the latest published month (May-2026). Contained-N conversion factors in Method below.

Read on the tariffed line alone, this is a triumph: Russia's share of extra-EU nitrogenous-fertiliser imports fell 28.6% → 9.4%, UAN fell −98%, ammonium nitrate −84%, and Belarus went to literal zero on UAN. Read in contained nitrogen — the unit that actually measures the dependency — the EU cut its Russian nitrogen intake by about half, not by three-quarters, and the majority of what survives now arrives as a chemical the regulation does not mention. Russia remains the EU's #2 ammonia supplier at 25.6%, behind Algeria.

Layer 1 — the implausible-origin test does not fire. Anywhere.

Every origin that gained EU share is a real ammonia producer, checked against the USGS Mineral Commodity Summaries 2026 world ammonia table (kt contained N):

There is no ~0%-capacity origin anywhere in the table. Not one. On a control this severe against a supplier this dominant, the antimony/magnesium fingerprint — a phantom flag from a country that cannot make the material — is simply absent.

Georgia is the one origin worth the second layer, and the second layer clears it. A 32% share of EU ammonium-nitrate imports from a small country on Russia's border is exactly the shape that should trigger suspicion. It survives the test: Rustavi Azot is a genuine gas-based plant with ~220 kt/yr ammonia and 500–550 kt/yr ammonium nitrate nameplate — Georgia's 143 kt EU line is well inside it — running on natural gas since 1960, and it has been owned by Indorama Corporation (Singapore-domiciled) since 10 January 2023, with modernisation financed by EBRD, ADB and FMO. Development-bank due diligence and a Singaporean parent is the opposite of the Youngsun→Thai Unipet shell pattern. This is the second time in the corpus (after Israel on magnesium) that layer 2 is used to clear an implausible-looking origin rather than convict one.

Layer 2 — there is no ownership pipe, because nothing is disguised

No shell, no relabel, no transit country, no falsified origin. The Russian ammonia is declared as Russian, lawfully, at the border. It lands in exactly three member states:

Source: Eurostat COMEXT DS-045409, reporter = each member state, partner=RU. The three legs reconcile exactly to the EU27 aggregate.

Lithuania's line rose 22% in the tariff year, and the mechanism is on the public record rather than inferred. LRT (Lithuanian public broadcaster) reports that from 2020 to 2025 between 91% and nearly 100% of all ammonia imported into Lithuania was Russian, and that AB Achema — the country's nitrogen producer — bought Russian ammonia to meet contractual obligations to its own customers, having suspended its gas-based ammonia unit because European gas was too expensive and imported product too cheap. Achema's stated hope was that the 1 July 2025 tariff and CBAM would fix the market.

That is the whole mechanism, said out loud by the affected party: an EU plant shuts its own ammonia unit, buys Russian ammonia instead, and converts it into fertiliser that is EU-origin. The finished product never crosses a border, so it appears in no import statistic at all — not under a phantom flag, but under no flag, as domestic output.

Mode G — upstream-of-scope precursor passthrough (new)

The corpus already has C (transformation wash abroad), C⁻ (adjudicated-failed), C⁰ (pre-emptively voided) and F (specification/scope shifting). This is none of them:

• Not C. In mode C the transformation happens in a third country to acquire

a foreign flag (Russian crude → Indian diesel; Xinjiang lint → Vietnamese apparel). Here it happens inside the controlling jurisdiction itself, and the output is domestic, so there is no origin line to read and no input-ratio to compute against a surger.

• Not F. In mode F the exporter changes something — a plate width, a trace

alloy — to step outside a scope description. Here nothing moved. The Russian ammonia line was already running at ~590 kt/yr before the tariff and is running at ~530 kt/yr after. The route was never opened in response to the control; it was simply never covered by it.

> Mode G — upstream-of-scope precursor passthrough. The control walls a > downstream tariff heading. The dependency continues to arrive, unchanged and > correctly declared, under an untouched upstream heading, and the buyer's > own domestic industry performs the transformation inside the wall. Layer 1 > cannot fire (no implausible origin), layer 2 has nothing to find (no > deception), and the finished good never appears in trade data (it is domestic > production). The only detector that works is re-denominating both headings > into the common contained element.

Three transferable rules

(i) The contained-element rule. Never grade a control in tonnes of the walled product. Grade it in the element the dependency is actually made of. Here the same event reads as −71% (walled line, contained N), −52% (all Russian N) or −85% (Russian urea, gross weight) depending purely on the denominator chosen — and only the middle number describes the dependency. Before quoting any control's effect, ask what the controlled material's precursor heading is and whether the instrument names it.

(ii) Check the escalator against the line it will actually hit. The duty still has an 8× increase ahead of it — €40/t today, €315/t from 1 July 2028 — and it is aimed at a line that is already empty: the lowest rung alone removed 71% of the contained N. Meanwhile the heading carrying the majority of the surviving Russian nitrogen is not in the escalator at any rung. The predictable result of ramping to €315/t is not less Russian nitrogen; it is Russian nitrogen concentrating almost entirely into CN 2814. A buyer reading the 2028 headline as "Russian fertiliser is finished in Europe" will be wrong in a specific, forecastable way.

(iii) The volume circuit-breaker never fired — so don't credit it. Article 1(3) provides that if CN 3102 imports from RU/BY exceed 2.7 Mt in Jul-2025→Jun-2026, the Commission must impose the maximum €315/t rate within 21 days. Cumulative Russian CN 3102 arrivals over Jul-2025→May-2026 were ~772 kt — 29% of the trigger. The cap was never remotely binding, so the collapse is attributable to the €40/t bottom rung (≈11% on urea at prevailing prices, on top of 6.5% ad valorem), not to importers rationing against a quota. A modest ad-quantum duty was sufficient; the escalator is, for this line, largely redundant.

Dating the break — and the confounder that is not the tariff

The monthly series contains two cliffs, each preceded by a front-load, and only the first belongs to this instrument (kt, EU27 imports):

Cliff 1 (Jul-2025) is the tariff — Russia-specific, landing exactly on the effective date, preceded by a textbook June front-load (Russian NPK at its series high the month before the duty starts).

Cliff 2 (Jan-2026) is not. It hits every origin — the whole extra-EU CN 3102 book falls 2,574 → 230 kt (−91%) after a December spike to more than double any other month in the series, with Egypt and China front-loading just as hard as anyone. It coincides with the CBAM definitive phase (1 January 2026), which covers fertilisers. Whatever its precise mechanism — the 2026 adjustment factor is only 2.5%, too small to price a 91% collapse, so the binding constraint is more plausibly the authorised-declarant gate than the certificate cost — it is origin-neutral and therefore not evidence about the Russia tariff. Logged as a confounder, not resolved here.

And a fourth transferable tell falls out of that row: storability, not scope, governs front-loading. Ammonia is also a CBAM good, and its line does not move at either cliff — because solid urea can be warehoused ahead of a deadline and anhydrous ammonia, needing refrigerated pressure storage, cannot. A front-load spike measures what the trade can stockpile, not what the rule covers.

Why it matters for the buyer

1. The headline overstates the de-risking by roughly a third. "Russian share of EU nitrogen fertiliser: 28.6% → 9.4%" is true and is the number that will be quoted. In contained nitrogen the cut is ~52%, and 57% of the residual sits in a heading nobody is watching. Any exposure model keyed to the controlled tariff heading is measuring the instrument, not the dependency. 2. The wash can be domestic. Every other transformation case in this corpus sends you looking abroad for the laundering step. Here the laundering step is a plant in Antwerp, Klaipėda or Rouen, and its output is counted as European production. When a control is drawn at a downstream heading, the importing country's own industry is the most likely transformation site — and the one that leaves no trace in any origin statistic. 3. It is a forecastable enforcement gap, not a historical one. The escalator ramps for three more years against an empty line. Extending scope to CN 2814 is the single change that would alter the outcome, and it has not been proposed. Watch for it; until it happens, the 2028 "Russian fertiliser banned in all but name" story will be substantially false.

Method & honesty rails

• Trade data: Eurostat COMEXT DS-045409, reporter EU27_2020 (and individual

member states for the ammonia landing table), flow 1 (imports), indicator QUANTITY_IN_100KG, monthly. Windows are matched 11-month Jul→May spans so the pre/post comparison runs on the regulation's own tariff-year boundary without padding the post period; COMEXT publishes to May-2026 at the time of writing and the latest months revise.

• Contained-N factors (commercial grades, stated so the reader can vary

them): urea 46.0%, ammonium sulphate 21.0%, AS-AN double salts 26%, ammonium nitrate 34.5%, CAN 27.0%, sodium nitrate 16.0%, AN/CaN double salts 27.0%, UAN 32.0%, other 26%, anhydrous ammonia 82.2%. CN 3105 (compound NPK) is excluded from the contained-N table — its N content varies by blend and including it would require a grade assumption the data does not support. Its gross-weight move (−61%) is reported separately and not netted in.

• The one inferential leg, labelled: that the imported Russian ammonia is

converted into EU fertiliser is an inference from end-use, not a traced flow. Customs data ends at the border; no per-shipment consumption record is public. The Lithuanian leg is the exception — there the producer's own purchase of Russian ammonia to serve fertiliser contracts is on the public record via LRT. Ammonia does have non-fertiliser industrial uses, so the conversion share is not claimed to be 100%.

• Nothing here is an allegation. No rule was broken. Reg 2025/1227 does not

cover CN 2814 and importing Russian ammonia is lawful. The finding is about instrument scope, not evasion — which is precisely why layer 2 finds no ownership pipe.

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

exposure — it sits next to them, exactly like the China–West price wedge.

Sources

• Regulation (EU) 2025/1227 — EUR-Lex Official Journal text (Annex II CN codes; Art. 1(3) volume thresholds)
• Council of the EU — press release, 12 June 2025
• European Commission Access2Markets — duty schedule notice
• Eurostat COMEXT DS-045409 — dissemination API
• USGS Mineral Commodity Summaries 2026 — Nitrogen (Fixed)—Ammonia
• Rustavi Azot — company profile; EBRD, modernisation of fertiliser production in Georgia (2025); Global Energy Monitor — Rustavi Azot Ammonia Plant
• LRT (Lithuanian national broadcaster) reporting on Lithuania's Russian ammonia intake, as syndicated; Achema production suspension and restart, ICIS via World Fertilizer; AB Achema plant profile, Ammonia Energy Association
• Register anchor: docs/iptm/actions/2025-06-20-eu-regulation-2025-1227-russia-belarus-agri-fertiliser-tariffs.md; confounder: docs/iptm/actions/2026-01-01-eu-cbam-definitive-phase.md