A full adjustment leaves Citi above its group capital requirement. Its history of control failures makes the supervisory consequences harder to dismiss.
A Citi UK securities unit reportedly measured some trades with a risk model that regulators had not approved for those types of transactions. Reuters says the resulting understatement of risk-weighted assets was about $35 billion, citing an anonymous source. Citi acknowledges an internal capital modeling issue at a European trading subsidiary, says it fixed the problem during summer 2026 and injected capital to restore local buffers. Citi says the subsidiary remained above regulatory minimums, with no impact on the holding company or main operating bank. The $35 billion describes measured risk, rather than money disappearing from the bank. That difference is where the investment argument begins.
We would refuse to sell Citi as though the reported $35 billion were an equity loss. We predict group Standardized risk-weighted assets at 31 December 2026 will remain below $1,267,261 million, the September balance plus the entire reported error. The full scenario leaves a group capital cushion, while Citi's record with supervisors makes another modeling mistake expensive beyond its CET1 ratio. Even a comfortable buffer buys little trust from the people who police the models.
A model assigns measured risk to trades; those assets sit below equity in the regulatory ratio. The Denominator Bridge is the difference between adding measured risk to that denominator and removing actual equity from the numerator. It matters here because a $35 billion change in the risk calculation calls for extra capital at the applicable required percentage, rather than a matching $35 billion payment. A weaker ratio restricts the freedom to deploy capital, even when equity dollars stay put.
At 30 September 2026, Citi held $157,415 million of common equity Tier 1 capital against $1,232,261 million of Standardized risk-weighted assets. Its reported CET1 ratio was 12.77%, with an 11.6% requirement. In a full addition scenario, $1,232,261 million plus the reported $35,000 million equals $1,267,261 million in risk-weighted assets; $157,415 million divided by $1,267,261 million gives 12.42%. That is a 0.35 percentage point fall. An 11.6% requirement on $35,000 million adds $4,060 million to required capital, because 11.6% of $35,000 million is $4,060 million. This is a buffer problem with a scale that investors can price.
The September cushion was $157,415 million less 11.6% of $1,232,261 million, or $14,473 million. With all $35,000 million added, $157,415 million less 11.6% of $1,267,261 million leaves $10,413 million. The result leaves Citi above its stated group requirement even after the full adjustment scenario. A regulator can demand remediation while the bank still has capital to cover the added requirement. That headroom supports future risk, rather than immediate cash distributions to owners.
Citi also reported $1,312,128 million of Advanced Approaches risk-weighted assets at 30 September 2026. The reported $35,000 million is 2.84% of its $1,232,261 million Standardized base and 2.67% of its $1,312,128 million Advanced base. The Standardized measure carries the bank's 11.6% required CET1 ratio, so its denominator determines our capital judgment. Using one approach's RWA base to judge the other's required capital would misprice the hurdle investors care about.
Citi's 30 June 2026 filing put Standardized RWA at $1,232,309 million, compared with $1,232,261 million at 30 September. That near-flat group line sits alongside Citi's statement that it found and resolved the subsidiary problem during the summer. Our scenario adds the entire Reuters amount above September, placing a demanding assumption against the reported buffer. A sudden rise in group RWA would press on future capital flexibility; a steady denominator leaves the $35 billion headline far removed from shareholder dilution.
The September filing identifies Citigroup Global Markets Limited as a UK PRA-registered broker-dealer with $30 billion of regulatory capital at 30 September 2026, up from $27 billion at 30 June 2026. Citi describes the affected company to Reuters only as a European trading subsidiary, and says it injected capital there. Those two balances describe CGML's regulatory capital, while Citi's statement identifies the affected subsidiary only in general terms. Capital held inside a supervised subsidiary serves its local buffer first, limiting the group's freedom to move resources wherever returns look best.
In July 2024, the Federal Reserve cited "insufficient progress remediating its problems with data quality management" and said its 2020 action against Citi remained in effect; the Board and OCC imposed approximately $135.6 million in combined penalties. The Bank of England reported in May 2024 that Citigroup Global Markets Limited had received repeated supervisory communication about its trading controls. The PRA imposed £33,880,000 and the FCA £27,766,200 for trading controls failures. The strongest case against our view is that another modeling lapse prolongs supervisory restrictions even with adequate CET1 capital. The price is management freedom and time, not a sudden $35 billion equity loss.
A skeptical supervisor has a sharper objection than a bad reputation: the year-end denominator itself can go against us. Between 31 December 2025 and 30 September 2026, Citi's Standardized RWA rose from $1,192,174 million to $1,232,261 million, a $40,087 million increase. If the full $35,000 million correction arrives in the final quarter on top of September's balance, December reaches $1,267,261 million before ordinary business movements. A corrected model can change the measured risk abruptly, even while the equity stock remains fixed. Our answer is the comparable three-month Standardized movement from $1,232,309 million in June to $1,232,261 million in September, while Citi says it fixed the subsidiary issue in summer 2026. That nearly level quarter favors a smaller year-end capital hit, although the December denominator still decides our bet.
Metro Bank's 2019 Pillar 3 account gives the capital danger a price. Metro increased risk-weighted assets by £900 million following a review and completed a £375 million equity raise in May 2019. Against £9,147 million of RWA at 31 December 2019, that correction was 9.8%, calculated as £900 million over £9,147 million. For Citi at 30 September 2026, the equivalent comparison is $35,000 million over $1,232,261 million, or 2.84%. Metro turned a classification error into an equity financing, something shareholders paid for directly. Citi's smaller relative adjustment and $10,413 million scenario headroom make that funding outcome less persuasive here, without erasing the supervisory warning.
The Denominator Bridge still matters after those regulatory examples: it turns the reported $35 billion into a $4,060 million extra capital requirement in our full addition scenario. The Fed's 2020 action and the PRA's 2024 controls findings make supervisory trust a separate cost, one borne through restrictions and management attention. We would demand a discount for that history while resisting an equity-hole narrative. The shareholder price should reflect that enforcement history without pretending that $35 billion of common equity has disappeared.
We would keep a governance discount on Citi but avoid a capital-panic sale because we predict its group Standardized RWA at 31 December 2026 will stay below $1,267,261 million. We would be wrong if Citi's 14 January 2027 results report 31 December 2026 Standardized RWA of $1,267,261 million or higher.
