The price of a probability
On 28 February 2026 Israel and the United States struck Iran. Iran closed the Strait of Hormuz the same day and it has not reopened. Four banks have now put numbers against that war in their credit-risk disclosures, and the numbers do not agree.
The same method, turned towards a commercial reader. The rest of this site asks what a piece of infrastructure or a corporate network means for a state. This asks what a geopolitical judgement costs a lender, in a number the lender has already published. The discipline is identical: primary sources only, every figure traced, confidence stated separately from probability, and the falsifiers written down before the conclusion.
The war, and why it reaches a balance sheet
Israel and the United States began a series of strikes against Iran on 28 February 2026, saying they were targeting Iran's military infrastructure and its nuclear and ballistic missile programme, and calling for a change of leadership. Iran's Supreme Leader was killed. Iran struck back at Israel, at United States bases across the region, and at military and civilian sites in Arab states hosting American forces. British bases in Bahrain, Qatar and Cyprus were among those attacked.
The same day Iran declared the Strait of Hormuz closed to all vessels, later qualifying that to admit ships of "non-hostile" states. On 13 April the United States began a counter-blockade of Iranian ports. Both blockades remain in place.
The scale is measurable rather than rhetorical. Around 3,000 vessels used the strait each month before the conflict. World Trade Organization tracking shows a 95 per cent reduction in ships carrying crude oil to and from Persian Gulf ports and a 99 per cent reduction in LNG carriers since February.
The Strait of Hormuz carried twenty million barrels a day in 2024, about a fifth of global petroleum liquids consumption and more than a quarter of seaborne oil trade. The US Energy Information Administration's own conclusion: "Most volumes that transit the strait have no alternative means of exiting the region, although there are some pipeline alternatives."
Against that, the EIA estimates about 2.6 million barrels a day of Saudi and UAE pipeline capacity could be available to bypass the strait in a supply disruption. The pipelines exist, but both were already being used more heavily day to day, which limits the headroom available to reroute.
Stated carefully, because it is easy to overstate. That 2.6 million figure is spare capacity available in a disruption, not the total volume that has a physical route. Subtracting it from twenty million and calling the remainder stranded would confuse two different quantities. An earlier draft of this work did exactly that, and the arithmetic was removed.
And "closed" is the wrong word for what Iran is now doing. Tehran is running a tiered permission system, reportedly with tolls, admitting ships of specific countries sometimes for a fee, and has proposed a new body to manage the strait and Iranian control of wider parts of it. This is not a blockade so much as the assertion of a tollgate — which matters for how the conflict ends, because a dispute about who collects money for passage is a claim to sovereignty over a waterway.
A disruption of that scale is an energy price event, a fiscal event for every Gulf exporter, and therefore a credit event for every bank lending into them. It reaches banks with no Gulf exposure at all, through the price of energy and the cost of freight.
Why this is answerable at all
Under IFRS 9 a bank must estimate credit losses it expects but has not yet incurred. To do that it builds several pictures of the future and assigns each a probability. Move weight onto the darker pictures and the estimate rises, and that rise is a cost in the current period even though no borrower has missed a payment.
So a geopolitical judgement has a disclosed, auditable route into earnings. Not a vague claim that instability is bad for business: an actual figure, in an actual filing, produced by someone deciding the world got riskier.
What the four banks did
| Bank | Downside weighting | At 30 June 2026 | Disclosed against the conflict |
|---|---|---|---|
| Standard Chartered | 41% → 60% | Held | $234m, decomposed line by line |
| First Abu Dhabi Bank | 30% → 50% in Q1 | Held, plus a further AED 350m in Q2 | AED 650m, single figure |
| Emirates NBD | → 100% retail, 80% corporate, in Q1 | Reverted to 30% | Nil additional at 30 June. No single conflict-attributable figure disclosed |
| HSBC | Not disclosed | Not disclosed | $0.3bn in Q1; at the half year a forward stress range instead, not conflict-specific |
Standard Chartered's $234 million is about 4.9 per cent of a record half-year profit before tax of $4.8 billion. Material to the impairment line, immaterial to the result. Nothing here suggests the conflict damaged any of these banks' half years — but immaterial to the result is not the same as immaterial to a borrower.
The finding
The four banks did not translate the same geopolitical shock into a common provisioning response. H1 2026 impairment therefore cannot be read as a pure measure of underlying credit deterioration. It also embeds differences in management judgement, portfolio composition, modelling choices and regulatory treatment.
The tempting version of this is that the banks disagree, so provisioning measures management rather than risk. That claim is too strong and this page does not make it. Two banks can face different books, different borrower mixes and different regulatory permissions, and rationally arrive at different numbers. A shared regulator does not mean a shared risk, and opposite provisioning decisions do not require one bank to be wrong.
What the divergence does establish is narrower and more useful: a headline impairment figure, compared across institutions, is carrying more than credit.
What you can check, and what you cannot
Five things drive the divergence. They are not equally visible from outside the bank, and that determines what any outside reader can honestly conclude.
| Cause of divergence | Observable? | Where to look |
|---|---|---|
| Scenario weightings | Yes, disclosed | IFRS 9 notes in interim statements |
| Regulatory treatment | Yes, disclosed and dated | Central bank announcements; the bank's own relief-measures note |
| Scenario architecture | Partly — narrative only | Risk review commentary |
| Portfolio and borrower mix | Partly — segment level | Pillar 3 and segmental disclosure |
| Model structure and judgemental overlays | Largely not | Not disclosed at a comparable level |
Because the last two are not observable, this work cannot separate how much of the divergence is judgement and how much is structure. It can establish that both are present in the reported number. Saying so is the difference between an analysis and an assertion.
The regulatory asymmetry
The two UAE banks were operating inside a Central Bank of the UAE Financial Institution Resilience Package, introduced on 17 March 2026. It cut the countercyclical buffer to zero, released a point of the capital conservation buffer, and allowed liquidity ratios to run below their normal minima.
It also allowed flexibility in how restructured borrowers are staged, "compensated for by additional provisions at the portfolio level", in FAB's own words. Standard Chartered, outside any such regime, re-staged individual exposures.
So staging is not comparable across these banks in this window. The relevant axis is not international versus regional. It is whether a bank was operating inside a temporary official support regime or outside one.
What that regime is not evidence of: that deterioration is being actively suppressed. Stable staging is equally consistent with borrowers not yet having deteriorated. The comparability problem is certain; the suppression is not, and cannot be established from public disclosure. 30 September 2026 is the current end-date the UAE banks disclose for the extended measures, though the package has capital, liquidity and credit-risk components and the filings do not establish that every concession runs to the same date.
Is 30 per cent right?
Standard Chartered published a scenario weighting and separately quantified the conflict-related impairment effect. Its new downside scenario, Sustained Middle East Conflict, carries 30 per cent. Mapped onto the UK Professional Head of Intelligence Assessment yardstick for comparison, that sits in the unlikely band.
One caution on that mapping. The PHIA scale is used here as a translation device. An IFRS 9 scenario weighting and an intelligence probability judgement are not methodologically identical — the first is an input to a loss calculation that also does work capturing the shape of a distribution, the second a statement about whether an event occurs. They are compared because the banks call their weightings probabilities and move them in response to world events, not because the two frames are equivalent. What follows challenges the weighting, not the $234 million.
That the Sustained Middle East Conflict scenario describes Standard Chartered's operating environment through the FY2026 reporting period: highly likely, 80 to 90 per cent. Confidence: medium.
High confidence on the structural mechanics — available bypass capacity cannot replace normal flows, and both blockades are in place. Medium on political duration, because the counter-evidence is real and has not been discounted.
Three things happened after every bank in the sample had closed its books on 30 June. In August Iran appointed a hardliner to run national security, whose stated conditions for reopening Hormuz are wider than what the June memorandum contained. On 19 August the United Arab Emirates announced it would cut off all trade and all commercial and financial transactions with Iran — on WTO trade data cited by the House of Commons Library, Iran is the source of 30 per cent of UAE imports, its largest single source. On 24 August the United States launched an expanded sanctions regime targeting Iranian technology, oil, gold and shipping.
Emirates NBD reverted to normal scenario weightings and held no additional overlay at 30 June. Seven weeks later its own government announced it was severing its largest import relationship.
And one piece of evidence does not depend on hindsight at all. On 21 May, five weeks before the reporting date, the UAE's own state oil company said full flows through Hormuz would not resume until 2027, even if a deal were reached quickly. That was public, and available to every bank in this sample when it set its assumptions.
So the claim is not that any bank was wrong in June. On the information available at the reporting date, a downside weighting in the unlikely band was defensible. The claim is that it is not defensible now. A scenario is a tail risk when it might happen. This one is happening.
A question I cannot answer and a risk committee can
Standard Chartered's new scenario is defined in its filing as a prolonged crisis producing "sustained energy supply disruptions and elevated global commodity prices". That is a claim about volume.
My prior was that this would be the wrong variable. The closest recent precedent, the 2024 Red Sea disruption examined in my work on Mediterranean terminals, operated mainly through routing, insurance, freight and price. Cargo still arrived. It arrived later, further round and considerably dearer, and the volume that left one terminal reappeared inside the same operator's network.
That prior does not transfer, and the reason is the point. The Red Sea had the Cape, and it could absorb essentially all the displaced cargo. Available bypass capacity at Hormuz is far too small to replace normal flows. So for this chokepoint supply volume is plausibly the right variable, and Standard Chartered may have specified its scenario correctly.
Does our geopolitical stress scenario model physical supply loss, or the cost of keeping supply moving — and have we checked which one this particular chokepoint produces?
The finding that survives is more useful than the one it replaces. Chokepoints are not interchangeable. The transmission channel depends on whether an alternative route exists, and a scenario built by analogy to the last crisis would have modelled the wrong thing. I am not claiming this model is mis-specified — I cannot see the variables, only the narrative describing them. That question is one a committee can answer internally in an afternoon, and it is worth asking.
The sovereign channel
Standard Chartered attributed part of its $159 million of overlays to "the estimated probability weighted impact of sovereign downgrades across several footprint markets". It does not name them in that passage. Elsewhere, in a footnote to a geographic exposure table, the bank defines its Middle East as Bahrain, Egypt, Iraq, Jordan, Oman, Qatar, Saudi Arabia and the United Arab Emirates — substantially the same set of states drawn into the conflict.
The limit of that inference, stated. The eight-country definition appears in a different table. The overlap is evidence of where the risk plausibly sits, not proof of which sovereigns the overlay was struck against. No credit rating is asserted anywhere in this work. Ratings could not be verified at source and are treated as a forward indicator instead — and agencies act after fiscal consequences become visible, so the channel confirms rather than warns.
What would prove this assessment wrong
- Hormuz reopens on terms and both blockades lift. The scenario resolves and 30 per cent was defensible throughout.
- Standard Chartered releases overlay at Q3 or the full year, in which case the weighting was right and this assessment was wrong.
- Emirates NBD requires no re-provisioning despite the UAE trade severance, in which case the reversion was correct and the timing argument fails.
- Pakistan's mediation produces a settlement within two quarters.
- Standard Chartered's next disclosure specifies the transmission variable and it turns out to be price and routing rather than volume, in which case the scenario question above dissolves.
What happens next, and when
30 September 2026
The disclosed end-date for the CBUAE package's extended measures. Extension would indicate the central bank still sees a case for extraordinary resilience measures. Expiry would remove part of the mechanism currently dampening the staging signal, making subsequent UAE disclosures more informative.
Q3 and full-year reporting
Whether the divergence unwinds, or becomes visible in borrower performance once temporary regulatory support and first-half overlays roll forward. That is the test, and it has not happened yet.
Sourcing
Every bank figure on this page is taken from the institution's own filing or results release, read at source. Where a document could not be reached at the company's own site, it was obtained from the exchange: FAB's interim statements come from the Abu Dhabi Securities Exchange. Conflict facts are from House of Commons Library research briefings CBP-10521, CBP-10636, CBP-10637 and CBP-11075. Chokepoint figures are from the US Energy Information Administration, and describe 2024 as a pre-war baseline; current traffic is from the WTO's Strait of Hormuz trade tracker. Figures reported only in secondary coverage have been excluded rather than published with a caveat.
The full assessment is below, and so is the method note that goes with it: how the figures were found, why judgements are graded the way they are, and the places where an earlier draft of my own reasoning was wrong and had to be corrected. The underlying source register, with the passages read and the arithmetic, is available on request.