Swedbank Finally Gets the IRB Keys — and Still Won't Say What It Will Do With Them

Swedbank Finally Gets the IRB Keys

For roughly two years, Swedbank's capital story has been a waiting game dressed up as a strategy. The buffer sat stubbornly far above the 100–300bp management range, and every quarter the same explanation came back: we can't release capital until the model approvals land, and until the U.S. investigations are closed. This quarter, one of those two doors finally opened — and the other one turned out to have been open for a while.

The unlock is real, and it is arithmetic

Swedbank received approval for its Swedish mortgage IRB model (implemented in Q4) and, after quarter-end, its Baltic corporate model (implemented in Q1). The CFO walked the bridge explicitly: the CET1 requirement falls from 15.1% to 14.1% at year-end, and the buffer picks up roughly 70bp from the mortgage model, 10–12bp from moving sovereign exposures to the standardized approach, 3–5bp from the Entercard non-performing book sale, minus about 6bp from the Pillar 2-to-Pillar 1 reshuffle. Then Q1 gives back 25bp via an Article 3 add-on on the Baltic corporate model while amendments are made, offset by 8–10bp from the performing book sale.

Summarizing both quarters, we expect a CET1 buffer benefit of approximately 60 basis points over the coming two quarters.

That puts the pro-forma buffer near 290bp in Q4 and 275bp in Q1 — still above the 200bp midpoint management says it targets, but no longer absurdly so. This is the first time in the entire archive where the company has put hard, dated numbers on the IRB model pipeline rather than gesturing at 'the timeline is not owned by us.' Compare that to July, when Jon Lidefelt said plainly that “the IRB overhaul has taken longer for us and others than what we expected, and we don't fully know the timeline.” — Jon Lidefelt, CFO · 2026-07-17

Worth flagging: this is not a Swedbank idiosyncrasy. The market's single highest-momentum keyword this quarter is IRB model, and on the same day SEB disclosed Baltic IRB models approvals and Handelsbanken discussed new PD model approvals for household exposures. The Nordic regulatory unlock is a sector wave — Swedbank is riding it, not creating it.

The answer that has not changed in four quarters

Here is the tension. After walking through the cleanest capital bridge the company has ever presented, Jens Henriksson was asked — twice — whether excess capital would come back via special dividend or buyback. The reply was verbatim the same construction used in October 2025, January 2026 and July 2026: a 100–300bp range, a 200bp target in the 15/27 plan, a 60–70% dividend policy, and “capital release above continues to be a judgment call.” — Jens Henriksson, CEO · 2026-10-22

The framing has shifted subtly but tellingly. In October 2025, that sentence ended with the U.S. investigations named as the largest uncertainty: “the timing of further capital release continues to be a judgment call depending on the many uncertainties, where the long-running U.S. investigations is the largest one.” — Jens Henriksson, Chief Executive Officer · 2025-10-23 By July, with the DOJ and SEC behind them, it became “With all the U.S. investigations behind us, the uncertainties have, of course, diminished, and we have no intention of holding more capital than necessary.” — Jens Henriksson, CEO · 2026-07-17 Now the investigations are gone from the capital answer entirely — the DFS settlement of SEK 483 million is booked and done — and still the answer is deferred to the board at the Q4 report. The uncertainty was never really the investigations. It was the IRB timeline; and now that is resolved, the deferral has simply moved one quarter forward.

The one genuinely new disclosure is that SRTs are creeping closer to the toolkit: “in general that it is a good tool to have in the toolbox. I guess in the banking environment, the importance of it might increase.” — Jon Lidefelt, CFO · 2026-10-22 That is a hedge, not a plan — but it is the first time the subject has been volunteered rather than extracted.

Credit: the quiet de-risking nobody is pricing

Buried under the capital noise is a portfolio reshaping. Credit impairments were a net reversal of SEK 701 million, of which Entercard contributed SEK 421 million — including a SEK 543 million revaluation gain on the NPL sale. Strip that out and you still have SEK 158 million of net releases, driven by a better macro scenario and the release of the post model adjustment, now down to SEK 93 million. The Credit impairments keyword is the company's top-ranked theme this quarter, and the explanation is refreshingly blunt: “we always tend to be rather prudent when it comes to credit impairments... we previously taken provisions for potential worse outcomes that have now not materialized.” — Jon Lidefelt, CFO · 2026-10-22

The Lithuania question — 53bp of impairments in one country — was met with the standard 'a few individual corporates' answer, unchanged from July's identical framing. Fair enough. But the structural point is that Entercard is being dismantled in two tranches, which permanently lowers group risk density and hands back capital at the same time. That is a cleaner story than the headline reversal suggests.

The plumbing: funding costs lead, assets follow

Net interest income rose 3% sequentially, but the quality was mixed. The Treasury line took negative revaluations — “If you look at the Treasury NII, there is volatility in this over time. It is revaluations.” — Jon Lidefelt, CFO · 2026-10-22 — while the real drag was the familiar mechanism of STIBOR rising ahead of any Riksbank move, so funding repriced before mortgages did. This is the same three-month Sweden / six-month Baltics lag the company has explained on every call since April 2025. It is not new; it is structural.

What is working is the Baltics, where rising EURIBOR lifted NII 11%, and Swedish mortgages, where Swedbank captured 19% of total market growth through its own channels in July–August against a back book share nearer 18%. The CEO noted corporate lending growth in Sweden was roughly half real estate — “about half comes from real estate sector, and we have a very strong offering there” — Jens Henriksson, CEO · 2026-10-22 — and the deposit mix shift that terrified investors two years ago is now described as small and confined to the Baltics.

What stopped being said

Two themes have vanished from the transcript. VAT recoveries, which dominated cost discussions for four straight quarters and repeatedly beat guidance, are absent entirely — replaced by extraordinary costs of SEK 673 million, including the DFS settlement. And the U.S. AML saga, the gravitational centre of every capital question from 2024 through mid-2026, is now a one-line accounting entry. The cost base that those refunds flattered is now carrying restructuring instead: the SEK 27.5 billion guidance holds, but the underlying 0.37 cost-income ratio and 16% ex-items ROE are doing the heavy lifting.

The story, then, is a bank that has finally removed its two biggest self-imposed constraints and is choosing, once again, to sit on the capital a little longer. That is defensible — the remaining Baltic retail model is still outstanding and the world is uncertain — but it is also the fourth consecutive quarter of the same answer to the same question, and the market has heard it before.