Is the ban leaking? Phosphate (DAP/MAP) after China's Dec-2025 export suspension

Trade-flow companion to the price wedge (R72), first FOOD-SECURITY / fertiliser entry. The wedge measures scarcity (what a controlled material costs ex-China). This case measures the other half: is the control actually holding, or is banned material re-entering buyers through a laundered origin? This is a DUAL-SCORE / alternative-track signal — never folded into any Tier-1 exposure score. The divergence between "who China says it stopped selling" and "who suddenly started selling" IS the signal. Research, not investment advice.

Verdict up front: the classic two-layer detector is NEGATIVE here. Phosphate is the fertiliser-axis sibling of bismuth / tellurium / molybdenum — a control that bites rather than leaks. China's export line collapsed; the gap was filled by genuine producers (Morocco/OCP, Saudi/Ma'aden, Russia, Jordan), only partially, so the dependency surfaces as physical shortage + a ~50% price spike, not a phantom-flag re-route. There is no ~0%-capacity transit surger and no common-ownership pipe — because bulk fertiliser cannot re-label (the Wa-State-tin lesson) and real non-Chinese producers exist (the un-gallium). The NEW element is a policy-side signature: China widened the ban serially (DAP → MAP → NPK compound → urea), the food-security analogue of the solar "whack-a-mole," but within one origin on the product-form axis rather than country-to-country. Whether exporters actively reclassified DAP as NPK to evade is not sourced — it is flagged as an open question, not evidence.

The control

On 11–12 December 2025 China's NDRC convened major phosphate producers (Yunnan Phosphorus, Hubei Xingfa, Guizhou Kailin, Wengfu) and the CPIA/CAMPA associations and instructed them to suspend phosphate fertiliser exports "in principle" until August 2026 — DAP, MAP, TSP and compound NPK — to secure domestic supply for the 2026 spring planting season (NDRC Notice 发改经贸〔2026〕 149号 reinforced it on 5 Feb 2026). China supplied ~40% of global phosphate fertiliser exports pre-ban and holds ~40% of the world's phosphate-rock reserves and ~40% of mine output (~80 Mt of ~200 Mt, 2022) — the most concentrated single-origin position of any major fertiliser. Chinese phosphate exports had already fallen ~10.0 Mt (2021) → ~6.6 Mt (2024) under a multi-year mining-cap and licensing regime; the Q1 window went ~785,000 t (3-yr avg) → ~111,000 t (Q1-2025), −86%; the Dec-2025 suspension zeroes the residual. (Anchor action: `2025-12-12-china-ndrc-phosphate-fertilizer-export- suspension`; volumes: NDRC/S&P Global via the action, Vinachem, Hermes-Kalamos.)

The implausible-origin test — REJECTED (no phantom flag)

Read through India, the world's largest DAP importer (~60% of its P requirement imported) and the buyer where China's collapse is cleanest:

Source: India DGCIS / Dept. of Fertilizers, compiled by fertilizerfield.com; cross-checked by Takshashila Institution ("China 22.28 lakh MT FY2023-24 → 8.47 lakh MT FY2024-25 → ~0 FY2025-26"). Shares approximate; fiscal- vs calendar-year labelling not fully harmonised across the two sources — treat the direction, not the last decimal.

China's line collapsed −65% (2023→2024) and to ~zero in 2025 — the textbook direct-channel severance. But unlike antimony, the gap did not reappear under a ~0%-capacity flag. Every origin that grew or held — Saudi Arabia, Morocco, Jordan, Russia — is a genuine top-five phosphate producer with real mines and refineries. There is no Thailand-shaped surger because you cannot re-consign a Panamax of granular DAP through a country that has no phosphate rock. And the substitution is incomplete: India's total DAP imports fell 67.79 → 46.22 lakh MT (−32%), inventories roughly halved (2.16 Mt → 1.24 Mt, 1 June 2024→2025), and prices jumped — Jordan/Saudi June-2025 deals at $781.5 / $810 per tonne CFR vs a prior $515–525 (+50–58%). The ban is binding; it is just binding through scarcity and price, exactly what the wedge prices — not through a laundered origin.

Why there is no fingerprint (and that is the finding)

1. Bulk fertiliser cannot re-label. Like Wa-State tin concentrate crossing a land border, a bulk-commodity cargo carries its origin in its bill of lading and its economics; there is no margin to route DAP through a phantom entrepôt. Mode-A is structurally unavailable. 2. Genuine ex-China producers exist. Unlike gallium (China ~98%, no Western primary), phosphate has Morocco (largest reserves on earth), Russia, Saudi Arabia and Jordan. A Chinese cut is absorbed by real capacity (the bismuth pattern), not laundered. The catch: those producers were already near max capacity, with new tonnage not online until 2027–2028, so absorption is partial and the residual is price. 3. No common-ownership pipe (tell = N). No namesake shell, no relocated Chinese-owned capacity abroad. The risk owner is not a Western compliance team screening origin — it is the importing farmer and finance ministry (India, Bangladesh, Brazil, Kenya) facing a physical availability + subsidy-cost shock for the 2026 planting season.

The one NEW signature: serial product-scope widening (inference, open)

China did not stop at DAP. The scope widened on a schedule: phosphate suspension (DAP/MAP/TSP/NPK, 11-12 Dec 2025) → nitrogen-potassium blends and further phosphate varieties added mid-March 2026 (a near "zero-export" regime effective ~19 Mar 2026, leaving essentially only ammonium sulfate freely exportable; half-to-three-quarters of China's fertiliser exports, up to ~40 Mt, now restricted). This is the food-security analogue of the solar "whack-a-mole," but inverted: solar's origin hopped country-to-country to stay ahead of a trade remedy; here the controlling state itself hops product-to-product to stay ahead of substitution. The pattern is consistent with product-scope arbitrage — exporters shifting into whatever HS line remains un-banned (e.g. declaring phosphate content inside a compound-NPK line, HS 3105.20, rather than DAP, HS 3105.30) — but no public source documents an actual reclassification flow, and the serial widening is equally explained by a simply escalating domestic-supply drive. Labelled as an open question, not a proven mode. If a customs dataset later shows China's NPK/compound line rising as its DAP line fell in the Jan–Aug 2026 window, this lifts to a genuine intra-origin fingerprint; until then it is a watch-item.

The demand-side twist (structural, not circumvention)

Part of why China is keeping the molecule home is not agriculture at all: China is the dominant maker of LFP (lithium-iron-phosphate) battery cathodes, which consume purified phosphoric acid, and reporting on the suspension explicitly ties the export cut to reserving phosphate for EV-battery use as well as food. The same phosphate unit is now contested between the food system and the energy transition inside the world's dominant supplier — a structural tightener that sits beneath the export control and will outlast the August-2026 window. (Noted as reported reasoning, not a quantified diversion.)

Why it matters for the buyer

1. A binding control need not leak to be dangerous. The absence of a phantom flag here is not resilience — it is a hard physical shortage landing on import-dependent economies, with the exposure priced in DAP CFR and in fertiliser-subsidy budgets, not hidden behind a relabelled origin. 2. This is where the detector's assumptions flip. The two-layer method presupposes a material worth laundering into a screening buyer. Phosphate is a bulk food input flowing to price-taking importers; the honest signal is the direct-line collapse + the price wedge + the serial scope-widening, read exactly as the tin and molybdenum cases read their silent flow legs. 3. The chokepoint is upstream and structural. China's ~40% of reserves + ~40% of output + LFP demand means the 2027–2028 capacity additions (OCP, Ma'aden, US DPA build-out under E.O. 14387) are the real story to watch, not a transit surger that will never appear.

Method & honesty rails

• Trade data: India DGCIS/DoF import-by-origin (fertilizerfield, Takshashila);

China export volumes NDRC/S&P/Vinachem/Hermes-Kalamos; HS 3105.30 (DAP), 3105.40 (MAP), 3105.20 (compound NPK), phosphate rock 2510, phosphoric acid 2809. Figures are annual and directional; India fiscal/calendar labelling is not fully harmonised between the two compilers — the −65%/−32% moves are robust, the last decimals are not.

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

exposure — it sits beside them, like the price wedge.

• Detector-negative, GATE 0: every figure traces to a named public source,

but no circumvention fingerprint is proven — mode-A is rejected, the ownership tell is N, and the product-scope-arbitrage signature is inference pending 2026 customs data. The value is the mapped blind spot (the first fertiliser/food-security chokepoint in the corpus) and the honest classification of a control that bites without leaking.

• Inference, not accusation: no origin-relabel or evasion flow is asserted;

the serial scope-widening is reported policy, its arbitrage interpretation an explicit open question.