Treasury put a $9 billion figure on Iranian shadow banking through U.S. correspondents
A trend study of 2024 reporting estimates about $9 billion in potential Iranian shadow-banking activity that touched American correspondent accounts. It is built from bank reports, not from a single verdict.
A number from reports, not from a trial
Treasury analysts estimated roughly $9 billion in potential Iranian shadow-banking activity that touched U.S. correspondent accounts during 2024. The figure comes from bank reporting. It is not a single prosecution and it is not a court judgment that any named bank laundered that sum. Faircroft will keep saying that, because a round number of this size gets detached from its footnote within a week. The footnote is the whole briefing. Potential. Touched correspondent accounts. Built from reports. Year: 2024. Published for committees that need something more precise than “Iran risk is elevated.” The split inside the estimate is what makes it usable. Shell companies with limited operating substance accounted for about $5 billion of the flagged patterns. Oil-marketing fronts clustered in the United Arab Emirates and Singapore accounted for about another $4 billion. Technology-procurement-style firms added hundreds of millions more. Those are buckets, not indictments of every company in Dubai or Singapore that sells oil or chips. They are a map of where the reports clustered.
The counterparty is often unwitting
The risk in this study is pattern and speed, not a villain who announces himself. A regional company that suddenly behaves like a commodity trader deserves the same look as a client booked in a jurisdiction everyone already calls high risk. Many of the counterparties in these flows do not know the story at the other end. That does not make the flow fine. It means your control cannot wait for a name you recognise. Hollow revenue, a business purpose that changes in a quarter, round-dollar wires between companies that do not trade with each other: those are the scenarios. Legacy typologies built for cash from the drug trade will sit quiet while this moves. Private-banking teams need the same briefing as the correspondent desk. Wealth clients sometimes fund the same structures through personal accounts. A study that is briefed only to the payments floor will be missed by the coverage officer who just approved a limit for a family office that “does a bit of commodities.” One page is enough for a board. Which business lines feed the corridors named in the study. What changed in scenarios after the study was published. Who owns each bucket. If the answer is that nothing changed, write that down. A gap you found yourself is cheaper than the same gap found in an exam.
Language in the reports you file
Supervisors who have read the study will look for whether your suspicious-activity narratives use the same red flags, in plain words: shell, limited substance, oil front, procurement, speed. You do not copy the Treasury prose into a report about a different customer. You train analysts so that, when they see the pattern, they can describe it in a way an examiner recognises. Quantitative studies go stale. Revisit the scenario library within two quarters so the thresholds are not a fossil of last year’s round numbers. Internal teams can replay a cut-down version on anonymised data. The goal is not to reproduce $9 billion. The goal is to show that an analogous pattern would have produced an alert.
What the study is not
It is not a sanctions designation. It is not a list of banks to exit. It is not proof that $9 billion was the property of the Iranian state. It is a trend analysis of 2024 correspondent reporting, with a methodology you should read before you quote the headline in a committee. Quoting it without the word “potential” is how a briefing becomes false. Refusing to quote it because it is not a verdict is how a correspondent book stays calibrated to 2019.
The 2026 use
Attach the study. Attach your one-page map. Attach the date you retuned scenarios, or the decision that you did not need to, with the reason. Ask wealth and correspondent owners to sign the same minute. Oil, shells, and procurement are three lines. If your book has none of them, say none, and say how you checked. A study this large will be in the pack of every examiner who covers dollar correspondents for the next year. The only embarrassing answer is that you heard the number and did not open the document.
A retune that is specific enough to test
Pick one bucket and write the scenario in a sentence a model owner can implement. Shells: a company with thin reported revenue receiving round payments from unrelated companies and paying them out within days. Oil fronts: a trader whose stated business is local services and whose payments reference cargo, ports, or intermediaries in the Emirates or Singapore. Procurement: a buyer of electronics or machine parts whose counterparties change every month and whose volumes do not match its staff or its premises. You do not need all three live in a fortnight. You need one live, sampled, and a date for the other two. Sampling means fifty payments, or whatever your volume allows, with a note of how many alerted and whether the narrative would make sense to someone who has read the study. Wealth coverage gets the same sentence. If a personal account or a family vehicle is the source of funds for a company in those buckets, the correspondent alert and the wealth alert should meet in one case, not in two systems that never speak. Put that join in the minute. The $9 billion figure will be quoted at you. Answer with the bucket you retuned, the sample size, and the date. That is a shorter answer than the headline, and it is the one that holds. Revisit it two quarters later so the 2024 patterns are not the only ones the engine still knows. The study itself says the figures age. Your minute should show that you expected that. Keep the $5 billion shell bucket and the $4 billion oil bucket as separate lines, so nobody briefs the board on a single undifferentiated Iran number. The procurement residual is smaller and still worth its own line if that business is in the book. Do not fold it into the oil number to make the slide simpler.