Part II: Basel III Endgame Re-Cut – Winners, Losers, and What Really Changed

In Part I of this series, we established the foundational frame: the March 19, 2026 Federal Reserve package is not four separate workstreams but a single coordinated recalibration of the U.S. capital stack — stress testing, supplementary leverage, Basel III risk-based capital, and the G-SIB surcharge operating as one integrated architecture. We showed that the headline −4.8% CET1 average for G-SIBs masks a rebalancing where capital rises in some stripes and falls in others, with the net outcome entirely dependent on business mix.

Part II takes that architectural frame and decomposes it into the four risk stripes that actually determine whether a specific institution wins or loses: credit, market risk (FRTB), operational risk, and CVA. This is the analytical backbone of the series — the paper that translates the integrated architecture into firm-specific, desk-level capital outcomes and reveals why the gap between the biggest winner and the biggest loser spans a staggering 17 percentage points.

The number nobody is quoting — but every CEO needs

The range of CET1 impacts across the nine Category I and II bank holding companies runs from approximately +13% to −4%. The standard deviation — 8.7% — nearly doubles the mean. A firm at +13% and a firm at −4% are both inside the −4.8% average. Neither should be managing to it. Average impact analysis is essentially useless for capital strategy. Everything in this paper explains why that spread exists.

The philosophy changed, not just the calibration

The paper opens with a critical distinction most commentary is already blurring. The 2023 proposal was, by the agencies’ own subsequent admission, a gold-plated implementation — one that would have increased aggregate capital for the largest U.S. banks by approximately 16%, with the ISDA/SIFMA quantitative impact study finding that the trading book component alone would have increased market risk and CVA RWA by 129%. The 2026 re-proposal starts from a fundamentally different premise: implementing the 2017 Basel standards as written, with targeted U.S. adaptations — not exceeding them. That is not a calibration tweak. It is a structural repositioning.

Key points of the paper:

• Credit risk is the largest story nobody is telling. Despite receiving less attention than FRTB and CVA, the credit risk reduction — −10.0% in CET1 terms — is the single biggest positive driver in the entire package. Residential real estate RWAs fall 30.4% under the standardised approach. SFT minimum haircuts are dropped entirely. The 65% investment-grade risk weight is extended to internally-rated corporates. For commercial banks, regional lenders, and mortgage originators, this is the most material number in the document — and it is being buried under trading book commentary.
• FRTB: a genuine reduction from 2023, but still an increase from today. Market risk RWAs rise from $453 billion to $546 billion — a 20.5% increase from current rules. The 2026 re-proposal moderates the extreme of the 2023 proposal through six key modifications including full diversification for Type A NMRFs, removal of the Spearman Correlation test, and a cap on IMA capital at SA levels. But for structured credit desks, correlation books, and bespoke securitisation positions — where IMA eligibility is hardest to achieve — the SA floor remains the binding constraint.
• The ILM removal transforms operational risk economics. The 2023 proposal’s Internal Loss Multiplier, floored at 1, structurally over-capitalised custody and investment management franchises. The 2026 re-proposal removes the ILM entirely — the single largest operational risk change in the document for firms like BNY Mellon and State Street. Aggregate operational risk RWA for Category I and II firms falls from $1,745 billion to $927 billion under the new Expanded Risk-Based Approach. But critically, the paper demonstrates that you cannot analyse this in isolation: the Fed explicitly co-calibrated the Basel III op risk increase with a corresponding stress capital buffer reduction. Nearly every standalone Basel III commentary is missing this interdependence.
• CVA nearly doubles from a near-zero base. The first standalone CVA capital charge in the U.S. framework produces a 96% increase in CVA CET1 requirements against the standardised approach baseline. The 2026 scope is narrower than 2023 — client-facing derivative transactions are excluded, a material change for clearing intermediaries and prime brokers — but the repricing event for bilateral OTC derivatives desks, XVA economics, and counterparty selection is structural and immediate.

The three-way regulatory arbitrage verdict

The paper includes a full cross-jurisdictional comparison of all four risk stripes across the U.S. (March 2026 NPR), the EU (CRR3), and the UK (PRA PS12/6). Three of four stripes produce a UK disadvantage versus at least one other major regime. On operational risk, the EU and UK share the penalty equally — both structurally more costly than the U.S. framework. On CVA, the UK is the most punitive regime without ambiguity: NFC and pension fund exemptions retained in the EU and effectively carved out in the U.S. are removed entirely under UK Basel 3.1. For institutions operating cross-border books, this section alone reframes booking location economics.

The integrated lens: why standalone analysis is now structurally flawed

The paper closes with the argument that began in Part I and intensifies here: the Fed designed the Basel III NPR and the stress testing NPR as a jointly calibrated package. Modelling one without the other overstates the operational risk and market risk impact for Category I and II firms by a material margin. The four highest-stakes comment-period questions — CVA scope, AOCI boundaries, mortgage servicing asset risk weights, and IMA eligibility for structured products desks — are all live, and the June 18, 2026 deadline is the last genuine opportunity to shape the outcome.

Fill in the form below to receive the full paper — including the definitive winners-and-losers matrix by business model and risk stripe, the three risk-stripe interactions nobody is modelling, the complete U.S. vs. EU vs. UK comparison tables, and the four comment-period questions where the agencies are genuinely receptive to quantitative input.