The longest and most bitterly contested financial regulation fight since the 2008 crisis is quietly reaching its conclusion, and the banks have won almost everything they asked for.
On 19 March 2026, the Federal Reserve Board, the Office of the Comptroller of the Currency and the Federal Deposit Insurance Corporation jointly issued three proposals that would rebuild the American bank capital framework. The proposals formally rescind the 2023 Basel III endgame framework, the rule that would have sharply raised the amount of capital large banks must hold. The comment period closed on 18 June 2026. The agencies are now writing the final text.
The 2023 rule never took effect. It was withdrawn after opposition from the banking industry, from members of Congress and, unusually, from sitting Federal Reserve governors. What replaces it is described by the agencies as a modernisation of the framework. In substance, it lowers required capital for most large institutions relative to what was proposed three years ago.
What the three proposals actually contain
The package is deliberately structured as three interlocking parts rather than one rule, which makes it harder to summarise and harder to oppose as a single object.
- An Expanded Risk-Based Approach, applying to the largest and most internationally active institutions, described in the framework as Category I and Category II banking organisations.
- A revised Standardised Approach governing how US banking organisations calculate and report risk-weighted assets, which is the denominator in every capital ratio that matters.
- A modified method for calculating the surcharge applied to global systemically important banks, the extra capital layer imposed on the handful of firms whose failure would be most damaging.
The third item is the one with the largest practical effect on the biggest institutions. The surcharge sits on top of every other requirement, and a change in how it is calculated flows directly into how much capital a bank must hold against the same book of business.
How the 2023 rule died
To understand what is being finalised now, it helps to understand what was rejected.
The 2023 proposal was the American implementation of revisions agreed internationally by the Basel Committee in 2017. But the US version departed substantially from the international text. It was more conservative in several places, and the agencies estimated it would raise capital requirements for the largest banks by roughly a fifth.
The reaction was unprecedented in scale. The industry mounted an extensive public campaign. Members of Congress from both parties objected. Most significantly, the proposal failed to command unanimous support inside the Federal Reserve itself, with governors publicly dissenting. A rule that cannot hold its own board is not a rule that survives a comment period.
The banks argued that higher capital requirements would reduce lending, push activity toward less regulated non-bank institutions, and disadvantage US firms against European and Japanese competitors implementing a softer version of the same accord. Supporters of the rule argued that capital is what allows a bank to absorb losses without public support, and that the 2023 crisis among regional banks had demonstrated the cost of thin buffers.
Both arguments have merit. The 2026 proposals resolve them decisively in one direction.
Why this connects to the stress test
Capital requirements in the United States are set by two mechanisms working together. The risk-based framework being rewritten now provides the floor. The annual supervisory stress test provides a bank-specific layer on top, known as the stress capital buffer.
That second mechanism is also in transition. All 32 participating bank holding companies cleared the 2026 supervisory stress test published on 24 June, absorbing more than $708 billion of hypothetical losses while remaining above their minimum requirements, with aggregate capital falling 1.6 percentage points. But as our report on that exercise set out, the Board decided in February that the 2026 results would be disclosed without feeding into the capital framework.
So both halves of the machinery are suspended at once. The stress test ran but does not bind. The risk-based rules are being rewritten but are not final. Banks are currently operating under requirements derived from a framework that everyone involved agrees is being replaced.
What this means in practice
Capital is the least glamorous and most consequential number in banking. It determines how much a bank can lend against a given equity base, how much it can return to shareholders, and how much loss it can absorb before it needs help.
Lower requirements have three predictable effects. They increase return on equity, because the same profit sits on a smaller capital base. They free capacity for buybacks and dividends. And they reduce the buffer available when losses arrive.
The first two are immediate and measurable. The third is invisible until it matters, which is the recurring difficulty in this debate. The cost of holding capital is paid every quarter and is easy to quantify. The benefit is paid once, during a crisis that may not arrive for a decade, and cannot be observed at all if the capital works as intended.
The timing question nobody is discussing
There is an uncomfortable coincidence in the calendar that deserves more attention than it is receiving.
Bank capital requirements are being loosened at the same moment that lending to artificial intelligence infrastructure is expanding at record speed. AI-related debt issuance is on track to approach $570 billion this year, with a growing share moving through private credit and structures that sit outside the traditional banking book. We examined that shift in our analysis of the credit market behind the build-out.
Banks are exposed to that lending in several ways that do not appear as direct loans: through credit facilities to private credit funds, through warehouse lines, through counterparty exposures and through the securities they hold. A capital framework calibrated on historical loss patterns may not fully capture concentration in a sector that did not exist at meaningful scale during any previous downturn.
This is not a prediction of failure. It is an observation that the buffer is being reduced and the exposure is new at the same time.
What to watch as the rules are finalised
- Whether the final surcharge methodology for the largest banks matches the proposal or is softened further after industry comment.
- When the stress capital buffer is reconnected to the risk-based framework, currently expected to affect requirements from 2027.
- Whether the agencies publish updated estimates of the aggregate capital effect, and how those compare with the roughly 20% increase the 2023 rule implied.
- Whether any Federal Reserve governor dissents from the final rule, which would signal the internal disagreement of 2023 has not been resolved so much as reversed.
- How quickly banks move to buybacks once requirements are known, which is the clearest measure of how much capacity the change actually releases.
Outlook
The direction is now settled. American bank capital requirements are going down rather than up, and the remaining questions concern magnitude and timing rather than direction.
Whether that is prudent depends on a judgement no regulator can make with confidence: whether the banking system is currently over-capitalised relative to the risks it actually carries, or whether the risks have simply moved somewhere less visible. The industry argues the first. The structure of the credit market in 2026, with a great deal of leverage sitting in private vehicles connected to banks by facilities rather than by loans, is a reasonable argument for the second.
The answer will not be known until the next serious downturn, which is precisely the problem with capital regulation and always has been.
About the data: The three capital proposals described here were issued jointly by the Federal Reserve Board, the Office of the Comptroller of the Currency and the Federal Deposit Insurance Corporation on 19 March 2026, with comments due 18 June 2026. They are proposals and had not been finalised at the time of publication; final rules may differ. Stress test results, including the number of participating firms, aggregate hypothetical losses and the decline in aggregate capital, are from the Federal Reserve 2026 supervisory stress test published 24 June 2026. The estimated capital increase associated with the withdrawn 2023 framework is drawn from the agencies own published estimates at the time. AI-related debt issuance figures are investment bank estimates rather than official statistics.
Reader note
This article is general information, not personalized financial advice. Read our Financial Disclaimer.