Is the ban leaking? Sulphur after Russia's Nov-2025 export ban

Trade-flow companion to the price wedge (R72). 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

No laundering fingerprint — and that is the finding. Russia banned exports of liquid, granulated and lump sulphur (HS 2503) effective 1 November 2025. The direct line died exactly as advertised: Brazil's imports from Russia fell −84% over the matched six-month window and hit zero in every month from February 2026 onward. No phantom flag appeared to carry it. Instead the volume was picked up by Kazakhstan and Turkmenistan — genuine producers, whose combined share of Brazil's book went from 23.5% to 42.3% while the market shrank 35%.

Both layers of the detector fail here, by construction:

• Layer 1 (volume-implausibility) cannot fire. Kazakhstan produces roughly

4 Mt/yr of sulphur and supplied over 20% of world sulphur trade at its 2024 peak — the 284 kt it shipped Brazil in six months is nowhere near its ceiling.

• Layer 2 (common-ownership tell) has nothing to find. Russia's own ban text

exempts exports to Eurasian Economic Union members and exempts international transit shipments outright. Nobody incorporates a namesake shell to do a thing the control explicitly legalises — the same logic the Russian-seafood case established for the Mar-2022→Feb-2024 window.

What the corridor does have is a second lever, and Russia pulled it. On 24 May 2026 Roszheldor — the Federal Agency for Railway Transport, not a trade-policy body — issued Order No. 587, "On the temporary suspension of loading and transportation," blocking Kazakh sulphur moving by rail to Russian seaports and border crossings; consignments already loaded were redirected to storage inside Russia. Kazakhstan is landlocked. Roughly 2.3 Mt of its sulphur crossed Russian Railways in H1-2025 alone (Argus, via Oninvest), and the Baltic outlet at Ust-Luga went dark. On 26 June 2026 Kazakhstan's own Energy Ministry signed Order No. 1363, suspending sulphur exports from 27 June — exempting shipments destined for railway stations inside the Russian Federation. On 24 July 2026 Roszheldor Order No. 926 cancelled Order No. 587, on the instruction of First Deputy Prime Minister D.V. Manturov.

Call this mode D⁰ — corridor capture, and read it as mode D run backwards. Mode D (Belarusian potash) is the target's move: a landlocked sanctioned producer whose route is severed reroutes through a co-sanctioned neighbour, and because that neighbour is a genuine co-producer, the volume-implausibility test is structurally defeated. D⁰ is the controlling state's move on the same geometry: the substitute origin is a genuine producer whose only route to market crosses the controlling state's territory, so the controlling state can throttle the substitute without issuing an export control at all. Both modes defeat layer 1 the same way — a real producer's flag never looks implausible — and both say the same thing about what to measure. Origin is diversified; the route is not.

The instrument cascade

Two of the seven rows are not trade-policy instruments. A register that watches export controls, sanctions and tariffs logs rows 1, 3, 4 and 6 and misses rows 5 and 7 entirely — and rows 5 and 7 are the ones that actually moved the tonnage. The enforcement lever left the trade-policy namespace and reappeared as a railway loading order, reversible at a deputy prime minister's word.

The flow — Brazil, the freshest large maritime buyer

Matched six-month windows, both fully reported, both ending before the 24 May rail order — so this table isolates the export ban from the corridor closure.

Source: UN Comtrade free public preview API — reporter Brazil (76), monthly, flow M, HS 2503, `comtradeapi.un.org/public/v1/preview/C/M/HS`. Figures are net weight, collapsing the API's `motCode` mode-of-transport split back to the `motCode=0` total. Cross-check: the pre-window annualises to ~2.4 Mt/yr against Comtrade's independently reported Brazil 2024 annual of 2,389,273 t.

Kazakhstan did not merely grow — it went from Brazil's #3 supplier (13.8%) to its #1 (31.6%), passing both the United States and Saudi Arabia, in a market that lost a third of its volume. That is the carve-out corridor filling in, on schedule, in plain sight, entirely lawfully.

The monthly series — and the honest limit on it

Tonnes, net weight. Same Comtrade query as above.

The Russian line's four-month run of exact zeros from February 2026 is unambiguous. The Kazakh line's May–June zeros are not attributable to Order No. 587 on this data, and we will not claim they are, for three reasons:

1. Timing runs the wrong way. A Brazilian import clearance in May reflects a Black Sea or Baltic loading roughly four to six weeks earlier — i.e. before the 24 May order. Order No. 587's trade-data footprint lands in July–September 2026 data, not yet published. 2. The world column collapsed too. Brazil's total book fell to 59 kt and 79 kt in those two months, a quarter of its normal run-rate. A zero inside a near-zero month is weak evidence about any single origin. (This is the discipline the Russian-seafood case banked: whenever a buyer line collapses, pull the total before calling it a shock.) 3. Zeros are within this series' noise. The Kazakh line already printed 3,052 t (Feb-25), 5,740 t (Jun-25) and 1,204 t (Dec-25). Sulphur moves in lumpy parcels.

Falsifiable forward test. Order No. 587 ran 24 May → 24 July 2026, and is closed. If mode D⁰ is real, Kazakh-origin sulphur into maritime buyers (Brazil, Indonesia, Tunisia, Morocco) should show a June–August 2026 trough followed by recovery from September 2026, with no compensating surge from any ~0%-capacity origin in between. If instead a new origin with no sour-gas processing appears in that gap, mode D⁰ is wrong and this is an ordinary mode-A relabel. Re-pull the same query after the Q3-2026 releases land.

The confounder, stated up front

This window is not clean, and reading it as a pure export-ban experiment would be wrong. The Middle East war that began 28 February 2026 and the ensuing Strait of Hormuz blockade trapped roughly half of world seaborne sulphur exports; CRU described "a naval blockade of the Strait of Hormuz" driving sulphur and sulphuric-acid prices to record highs, and Platts assessed FOB Baltic granular sulphur at $800/mt as late as 23 July 2026. In Brazil's book the Gulf bloc fell −60% — a far larger absolute displacement than the Russian ban caused.

India, a ~96%-Gulf-facing buyer, gives the confounder its own clean reading and a useful control:

Same Comtrade query, reporter India (699), matched six-month windows — India had not reported May or June 2026 at pull time.

Canada appearing from nothing at 11.5% of India's book is the tell that this is a genuine-substitution market, not a laundering market: at $800+/mt, Alberta sour-gas sulphur is worth railing to Vancouver and shipping to Kandla. Long-haul real supply moves before phantom flags do.

Indonesia — the third buyer, and the decisive negative. Indonesia is the market where sulphur demand is least elastic (HPAL nickel leaching) and it sits entirely off the Caspian corridor. Its substitution set is the same shape as India's:

Same Comtrade query, reporter Indonesia (360), matched seven-month windows; June-2026 not yet reported. Indonesia's total fell only −4% against Brazil's −35% — HPAL leaching demand does not ration, it pays. That is why Indonesia is the buyer where a phantom flag would have been most profitable to fly, and none appeared.

Across three independent buyers on three continents, every origin that gained share is a genuine sulphur producer: Kazakhstan, Turkmenistan, Canada, the United States, India. Not one ~0%-capacity origin appears anywhere in the substitution set. That is as clean a layer-1 negative as the corpus has recorded — and it is what makes the corridor, rather than the flag, the thing worth watching.

Why it matters for the buyer

1. Counting origins is not measuring diversification. A risk model that saw Brazil's sulphur book go from 5% Russian to 1% Russian, with Kazakhstan and Turkmenistan filling in, would score the dependency as reduced. The physical reality is the opposite: the book went from 23.5% to 42.3% Caspian, and every Caspian tonne — Russian, Kazakh or Turkmen — leaves through Russian rail and Russian ports. Concentration rose while the origin count improved. The unit of exposure is the corridor, not the flag. 2. The map test is an announcement-time detector. Like the C⁰ seafood case, mode D⁰ is visible before any trade data exists — but you read a map, not a clause. When a control names an exemption corridor, ask: does the exempted or substitute origin's export route cross the controlling state's territory? If yes, the substitute is transit-captive and the diversification is nominal. Kazakhstan's alternatives — the Trans-Caspian/Middle Corridor via Azerbaijan and Georgia — exist but did not absorb 2.3 Mt per half-year in 2026. 3. The lever can sit outside the instruments you monitor. Orders No. 587 and No. 926 are railway loading orders. They carry no tariff line, no entity list, no licence regime, and would not appear in a sanctions or export-control feed. Any monitoring product scoped to trade-policy instruments has a blind spot exactly the width of a railway — and mode D⁰ says that blind spot is where the decisive action happens once the formal control is already in place. 4. Reversibility is the point. Order No. 587 was cancelled after two months on a deputy prime minister's instruction. The corridor is therefore not closed; it is a standing option held over roughly 4 Mt/yr of Kazakh and an unquantified volume of Turkmen sulphur — which is precisely the position the phosphate and battery-metals chains now sit downstream of.

Method & honesty rails

• Trade data: UN Comtrade free public preview API, monthly, flow M, HS 2503

(sulphur of all kinds, other than sublimed/precipitated/colloidal), reporters Brazil (76) and India (699). Net weight is gross product weight. The API returns one row per mode-of-transport code plus a motCode=0 total; figures here take the motCode=0 total per partner-month (summing the raw rows double-counts). World totals are the sum of named partners, not the partnerCode=0 row, because that row is missing in four reporter-months (India 2025-12 and 2026-04; Indonesia 2025-06 and 2026-01) — using it there would have understated the denominator and inflated every share. Brazil, the case's core evidence, has both figures agreeing in all sixteen months. Buyer-side reporting is used deliberately — Russia has not published detailed customs data since 2022, so every Russian figure here is mirror data reported by the importer.

• Reporting completeness: Morocco — the world's largest sulphur importer — had

published nothing for any 2026 month at pull time and is therefore absent. India had not published May or June 2026; Indonesia had not published June 2026. Windows are matched explicitly wherever this bites, and the mismatch is flagged in-table rather than smoothed over.

• One test we could not run. Kazakhstan also has an eastward rail route to China

that does not cross Russia, so "did the blocked volume go east instead?" is the natural companion question. It is unanswerable here: China's last month in the free Comtrade preview for HS 2503 is December 2024. For scale, China's Kazakh line in the two months that do exist was 1,568 t (Nov-24) and 952 t (Dec-24) against a ~700 kt/month total book — so the eastward valve was negligible before the ban, and whether it opened after is an open question, not a negative result. Tunisia returned no data at all. Flagged rather than dropped.

• Policy text: Resolutions No. 1730 and No. 350 are quoted from the Russian

Government's own English-language announcements. Note a discrepancy worth carrying: the register's anchor action attributes the original ban to Resolution No. 1470 of 28 Oct 2025, while government.ru's announcement names Resolution No 1730 dated 1 November 2025. This case uses the primary text; the register entry should be reconciled.

• Inference, not accusation. No shipment, entity or person is alleged to have

evaded anything. The ownership layer returns N, and — as on Russian seafood — that N is substantive rather than a gap: the corridor was exempted in the control's own text, so there was nothing to evade and no reason to build a shell.

• Alternative-track only: never touches buyerRelativeScore or the base exposure.
• Contested attribution: the May–June 2026 Kazakh zeros are reported as *not yet

attributable*, with a dated forward test above. Neither Russian nor Kazakh officials publicly explained Order No. 587; the reasons reported are analyst inference (domestic redirection after strikes on Russian processing plants; or prioritising higher-margin cargo over low-margin transit), and are labelled as such.

Sources

• Russian Government, *"Government introduces a temporary ban on exports of technical

sulphur…"* — Resolution No 1730, 1 Nov 2025 — http://government.ru/en/docs/56821/

• Russian Government, *"Government extends temporary ban on exporting industrial

sulfur…"* — Resolution No 350, 31 Mar 2026 — http://government.ru/en/docs/58223/

• Oninvest, "Shifting priorities: why Russia blocked sulfur exports from Kazakhstan"

(Roszheldor order of 24 May 2026; 2.3 Mt H1-2025 RZD transit, per Argus; TCO/NCOC as the affected exporters) — https://en.oninvest.com/article/shifting-priorities-why-russia-blocked-sulfur-exports-from-kazakhstan

• The Diplomatic Insight, "Kazakhstan's Sulfur Exports Face a Transit Bottleneck"

(>20% of world sulphur trade at 2024 peak; loaded railcars redirected to storage in Russia; Middle Corridor alternative) — https://thediplomaticinsight.com/kazakhstan-sulfur-export-isolated-russia/

• Shanghai Metals Market, *"Kazakhstan follows Russia in full suspension of sulfur

exports"* (KZ MoE Order No. 1363, signed 26 Jun 2026, effective 27 Jun; Russia-bound rail-station exemption) — https://news.metal.com/newscontent/103978838-smm-flash-kazakhstan-follows-russia-in-full-suspension-of-sulfur-exports

• Hellenic Shipping News / S&P Global Platts, *"Russia lifts rail restrictions on

Kazakhstan sulfur exports"* (Roszheldor Order No. 926 of 24 Jul 2026 nullifying Order No. 587, on First Deputy PM D.V. Manturov's instruction; Ust-Luga; FOB Baltic granular $800/mt, 23 Jul 2026) — https://www.hellenicshippingnews.com/russia-lifts-rail-restrictions-on-kazakhstan-sulfur-exports/

• CRU Group, "Middle East conflict pushes sulphur and acid prices to new heights"

(naval blockade of the Strait of Hormuz; record sulphur and sulphuric-acid prices) — https://www.crugroup.com/en/communities/thought-leadership/2026/middle-east-conflict-pushes-sulphur-and-acid-prices-to-new-heights/

• Turkish Export General Directorate circular of 6 Apr 2026 (GTP 2503 export ban,

effective 7 Apr–30 Sep 2026) — https://orgtr.org/kukurt-gtp2503-ihracati/

• UN Comtrade, free public preview API — https://comtradeapi.un.org/public/v1/preview/C/M/HS