Federal Reserve finalizes changes to large-bank stress tests

The Federal Reserve Board finalized two changes to its large-bank stress tests in the United States on September 30, 2026. The rules introduce annual public input on parts of the test and a two-year average for a capital requirement tied to test results. The Fed says the changes should make that requirement less volatile, but their effect has yet to be measured.

The Board uses supervisory stress tests to assess whether large banks could absorb losses in a severe recession and keep lending to households and businesses. Their results feed into each affected bank’s stress capital buffer, one component of the capital it must hold. The new rules change both how the Fed develops the tests and how it uses their results to set that buffer.

What changes in the Fed’s annual stress tests

Under the first final rule, the Fed must invite public input each year on its stress-test scenarios and any material changes to its models. The rule also updates the framework for designing hypothetical scenarios and adopts models for the 2027 test. Those requirements make consultation a recurring part of the process, rather than treating the test’s design as a matter explained only after a result is issued.

Banks with large trading books will face two global market shock components each year. These components apply shocks to specified markets. For each firm, the Fed will use the component that produces the larger losses when calculating its stress-test result. That choice matters because trading exposures differ across banks: the rule calls for the more severe of the two calculated outcomes for each affected firm.

The Fed estimates that the changes together are likely to cut year-to-year volatility in capital requirements by about 50%, without materially changing aggregate capital requirements. Those are the Board’s projections, not observed effects of the new rules. Individual banks can still have different results because the tests measure their own exposures and projected losses.

Why the Fed is changing the process

In a September 18 speech, Vice Chair for Supervision Michelle W. Bowman described the stress capital buffer as the way supervisory testing enters large-bank capital requirements. She said the tests estimate losses, revenue and capital levels under a hypothetical severe recession. She also said the existing framework had faced criticism over limited transparency, swings between years and the absence of a meaningful appeals process.

Bowman outlined planned disclosures covering model equations, variables, coefficients, assumptions, limitations and the reasoning behind model choices. She said the Fed would provide more detail about how it designs scenarios. Her speech described the changes before the Board’s final vote; the September 30 announcement confirms the annual public-input requirement and the adoption of the 2027 models.

The September decision follows an earlier effort to open the testing process. Bowman said the Board committed in December 2024 to disclose and seek comment on models used to project banks’ losses and revenues. Reuters reported that the finalized framework largely mirrors changes the central bank had proposed previously, after years of industry complaints about how opaque and subjective the tests were. That account describes the debate; it does not establish how the final rules will perform.

When the changes apply and what remains open

The Board has specified the 2027 testing cycle for the adopted models and 2028 for the start of two-year averaging in stress capital buffer requirements. Bowman also said the planned annual effective date for the buffer would move from October 1 to January 1 of the following year, giving firms more time to implement a resulting requirement. The announcement links to the final notices and an effective-dates document for the detailed timetable.

A separate change to the Fed’s noninterest-income model is still a proposal. The Board says it would better capture differences in how banks earn fee income and would replace the current model if adopted. Comments are due 60 days after the proposal appears in the Federal Register; the September 30 announcement does not give a calendar deadline. Its final terms therefore remain open.

For banks, the immediate change is greater visibility into the test design and a stated path to smoother capital-buffer calculations. For the public, the promised annual comment process offers a way to examine scenarios and material model changes before they are used. Whether the Fed’s estimated reduction in volatility is realized will become clear only after the new calculations take effect.