Global Flour & Wheat Intelligence
UAE strategic decision dashboard · ADAFSA · Panel brief — Jul 2026
From the world balance sheet to the UAE loaf
A decision-maker's view of who grows, ships and buys wheat & flour — where the UAE sits in that map, the live risks and opportunities moving price and quantity, the forward outlook, and what the panel recommends the UAE do next.
Official — internal review onlyInterim build · not an official ADAFSA system
Interim reference build on private infrastructure (khamis.tech). Figures are reference/indicative pending ADAFSA data-platform integration.
🌍
!▲ +5.5%
843.8 MMT
World wheat production 2025/26 — a record crop
Yet exporter stocks stay tight (~11.3%).
📊
!▲ +1.7pt
33.8%
Global stocks-to-use 2025/26
Comfortable headline — China holds ~½ of it.
💵
!2-yr high
~$6.75/bu
CBOT wheat, mid-Jul 2026 (≈ $250/t)
Firmed on Black Sea & Hormuz risk.
🇦🇪
!~0% grown
1.87 MMT
UAE wheat imports MY2024/25 (~1% of world trade)
100% of wheat imported; ~3.8 mo cover.
🚢
!~flat
216.5 MMT
World wheat exports 2025/26 (~26% of the crop)
Broadly flat while output hit a record — exporters stay tight.
Decision requested
Commit 382,200 MT — 2.6 months of national wheat use — in 4 dated tranches for December-to-March discharge, with the first tender closing 30 September 2026.
If declined: the same 382,200 MT is still bought — but at roughly 45 days of notice instead of 77, which the World Bank tender model prices at about +$6–11/t (≈$2–5M on this tonnage, indicative) — and it is bought after 15 Feb 2027, when Russia is back under quota and a weekly floating duty, with the southern-hemisphere crop already sold.
The window we are committing into
203 days
to 15 Feb 2027, when Russia’s 20 MMT export quota re-engages · southern-hemisphere discharge Dec–Mar
cover ~3.75 mo today, ~3.4 mo on USDA’s own 2026/27 balance · planning basis $285/t, band $260–315/t (indicative)
as of 27 Jul 2026 · every figure carries a vintage — re-derive at WASDE-674, 12 Aug 2026
Reason 1 — the option expires on a date, not at a price
15 Feb 2027
Russia’s 20 MMT export quota lapsed on 30 June and does not re-engage until 15 February 2027 — a dated, unrestricted window on the origin that supplied 42.5% of our last recorded half-year of arrivals — and the southern-hemisphere crop, which is how we diversify away from that share, discharges inside the same window. The window is also thinner than usual, with Australia at 26.7–28.0 MMT against 35.99 last season and a very strong El Niño forecast to peak over its harvest, so tonnage inside it is allocated first-come — and none of that reasoning requires a price view.
Reason 2 — the forecast cannot carry a price bet
44 MMT of published forecast error
USDA publishes its own July track record — a 2-in-3 band of ±2.7% on world production, 44 MMT wide, three times the 14.5 MMT the three agencies actually differ by — so their agreement measures agreement, not accuracy. At this horizon wheat futures score 0.91–1.05 against a simple no-change forecast, statistically indistinguishable from assuming today’s price holds, which is exactly why the commitment has to be dated rather than priced.
Reason 3 — the asymmetry, priced
$11M once against $143–220M a year
If we commit at the $285/t planning anchor and the price falls to the World Bank’s own CY2027 forecast of $260/t, the mark-to-market loss on the committed tonnage is $11M once — while the 1.43 MMT we still buy at spot lands $36M/yr cheaper, leaving the country roughly $25M ahead in the year we are wrong. If we do not commit and a single named disruption fires, the modelled cost is $143–220M a year, and replenishment runs 63 to 138 days from bid close to discharge, so the cover cannot be rebuilt inside the event at any price.
Actuals — Russian quota 20 MMT, in force 15 Feb–30 Jun 2026, resolution signed 24 Dec 2025 (government.ru). Australia 26.7 MMT (ABARES, 2 Jun 2026) vs 28.0 (USDA WASDE-673, 10 Jul 2026); 2025/26 was 35.99. El Niño very strong, peak Oct–Dec 2026 (IRI/CPC, 20 Jul 2026). USDA July reliability RMSE 2.7%, 90% CI 4.6% (WASDE-673, pp. 35–37). Futures skill 0.91–1.05 vs a no-change forecast (Reeve & Vigfusson, Fed IFDP 1025, 2011). HRW Gulf FOB $286.0/t Jun-2026 actual; $260/t CY2027 forecast (World Bank Pink Sheet, 2 Jul 2026; CMO Apr 2026 vintage). Lead times 63–138 days bid-to-discharge (Saudi GFSA tender, 27 Feb 2026). Urgency coefficient +7 days = −$1.05/t (World Bank/FAO/WFP 2025, Table 12 — estimated on MENA state tender agencies, not UAE private millers; an analogue, not a measurement of the UAE). Derived and indicative — the 44 MMT band (USDA’s published error applied to its own 819.97 MMT forecast), 382,200 MT and 2.6 months (at 147,000 MT/mo total wheat use — the sum of the four windows’ own tonnages, not a round quarter-year), the 42.5% Russian share (USDA GAIN TC2026-0005, 17 Apr 2026), every dollar figure on 1.87 MMT of imports, and the $143–220M disruption range. The tranche schedule, tonnage, owner and dates are FlourSight proposals adopted by no authority.
Evidence — the commitment calendar, the forecast error bands and the price series behind the three reasons above
6 windows to June 2027 — every date published, every price a band
This is the object of the ask: approve the calendar, not a price view. Hover a move code for its title.
Dates are the publishers’ own — WASDE release calendar, ABARES quarterly, Russian export-duty resolution of 24 Dec 2025, AMIS harvest windows. Tonnages, tranche sizing, owners and the cost-of-waiting figures are FlourSight proposals and derived estimates, indicative, adopted by no authority.
Agencies differ by 14.5 MMT; the forecaster’s own error band is 44
Agency agreement measures agreement, not accuracy — USDA publishes its own July track record
USDA WASDE-673, 10 Jul 2026, pp. 35–37 “Reliability of July Projections” — RMSE 2.7% on world production, 9.1% on ending stocks; 90% confidence intervals 4.6% and 15.3%. Agency spread from USDA, IGC and FAO/AMIS current vintages. The error band is the forecaster’s published record, not our estimate.
Being wrong costs $11M once; being unprepared costs $143–220M a year
The two directions are not symmetric — which is the whole case for committing
Derived, indicative, on 1.87 MMT of imports and a $285/t planning anchor: mark-to-market against the World Bank CY2027 forecast of $260/t; disruption range from the deck’s own scenario engine. Replenishment 63–138 days bid-to-discharge (Saudi GFSA tender, 27 Feb 2026).
Every $25/t across the planning band is $48M a year on the subsidy line
The currency leadership approves in — $/t × 1.87 MMT
Derived, indicative: annual import bill = price × 1.87 MMT (USDA GAIN TC2026-0005). Under the Abu Dhabi fixed-price flour programme the movement lands on the government subsidy line, not on bread prices. Band $260–315/t, central ~$285/t, vintage 27 Jul 2026.
4 conditions would force a re-plan — each one is a live Risk trigger
A forecast without a falsification condition is an opinion
Release dates are the publishers’ own: FAO food-price index ~7 Aug 2026; WASDE-674 on 12 Aug 2026; IGC late Aug; Russian duty every Friday effective the following Wednesday. Falsification thresholds are FlourSight proposals. Every figure on this tab carries a vintage — a signal built on the 10 Jul WASDE expires at the 12 Aug WASDE.