The Federal Reserve has finalized a broad overhaul of its large-bank stress-testing framework that will average two years of stress-test results, open material model changes and annual scenarios to public input, alter the stress-test calendar and change how trading-book shocks are constructed.
The Fed said the package is expected to reduce year-to-year volatility in stress-test-related capital requirements by approximately 50% without materially changing aggregate capital requirements, according to its September 30 final-rule announcement. But a close reading of the underlying rules shows that the 50% figure is broader than the effect of two-year averaging alone.
The separate final rule governing the stress capital buffer calculation estimates that averaging itself would have reduced the average absolute annual change in firms’ stress capital buffers from 64 basis points to 47 basis points in the Fed’s historical analysis, a reduction of about 27%.
The approximately 50% estimate reflects the broader set of final stress-test rules, model changes and scenario changes rather than averaging by itself. That distinction matters because the final package changes considerably more than the formula used to calculate one capital buffer.
The Fed will average two stress tests, but not immediately
A bank’s stress capital buffer, or SCB, is the portion of its capital requirement that is tied directly to its performance under the Fed’s hypothetical severe recession. Under the existing framework, the calculation uses the current stress test’s projected decline in the bank’s common equity tier 1 capital ratio, adds planned common-stock dividends and applies a 2.5% minimum floor.
The new rule changes the first part of that calculation. For firms tested in consecutive years, the Fed will average the stress-capital declines generated by the two most recent annual supervisory stress tests before adding the current dividend component and applying the floor.
In simple terms, one unusually severe or unusually mild annual result will no longer flow completely into the bank’s SCB at once. Half of the calculation will come from the latest applicable test and half from the preceding one. That mechanism was already contained in the Fed’s April 2025 proposal to reduce SCB volatility, but the implementation date has changed substantially.
The proposal contemplated applying averaging beginning with the 2025 stress test, combining the 2024 and 2025 stress-capital declines. The final rule does not do that. Firms remain under SCB requirements without results averaging until January 1, 2029.
The first averaged requirements will combine the 2027 and 2028 stress-capital decline components. The delay means the Fed’s new annual capital-requirement timetable starts before the averaging mechanism itself.
Requirements generated from the 2027 test will take effect January 1, 2028, rather than October 1 under the previous framework. But two-year averaging first affects requirements taking effect January 1, 2029. The Fed says the delay ensures that both stress-test results entering the first average were generated using models informed by public input.
For readers following the Fed’s broader overhaul of bank supervision, Investozora previously examined Vice Chair for Supervision Michelle Bowman’s plans for the Federal Reserve stress-test overhaul.
The 50% figure is not the effect of averaging alone
The final documents provide a more precise picture of the volatility claim than the headline announcement does. When the Fed proposed two-year averaging in April 2025, its historical analysis estimated that average annual changes in SCB requirements would fall from 65 basis points to 54 basis points, about 17%.
The final volatility rule reran the analysis with additional data, including the 2025 stress-capital decline components and revised 2024 results. Under that analysis, average year-over-year changes fall from 64 basis points to 47 basis points, or about 27%. The average SCB level changes only slightly, from 3.83% under the current framework to 3.80% under the final rule.
Investozora analysis: The Fed’s approximately 50% volatility estimate therefore should not be read as saying that two-year averaging alone cuts SCB volatility in half. The final rule’s own economic analysis assigns a roughly 27% reduction to that mechanism. The larger figure comes from the combined stress-testing package, including changes to models and scenario design.
That is also why the two figures are not contradictory: they measure different packages of changes. The broader model analysis finds that the adopted changes would have reduced aggregate SCB requirements by approximately 1% of required common equity tier 1 capital when applied to the 2024, 2025 and 2026 stress-testing cycles. The Fed cautions that the exact impact will vary with the scenario and individual firms’ data.
The final rule reversed a proposed September 30 testing date
Another material change appears in the stress-test calendar. The 2025 transparency proposal contemplated moving the stress-test “jump-off” date — effectively the balance-sheet starting point for the exercise from December 31 to September 30. The final rule abandons that change and retains December 31.
The 208-page final transparency and accountability rule says commenters raised operational concerns and warned that September 30 could introduce additional seasonal variability. The Fed instead redesigned the calendar around the existing year-end starting point.
Proposed scenarios are now scheduled for disclosure by January 10, with at least 30 days for public input, while final scenarios move from February 15 to February 28. Capital-plan submissions move from April 5 to April 30. Final models are scheduled for disclosure by May 15, while final SCB requirements move from August 31 to September 30 and take effect January 1 rather than October 1.
That creates a longer chain between the balance-sheet date underlying the test and the date the resulting capital requirement becomes effective, trading some immediacy for additional public review and implementation time.
Material model changes now have a defined public-input threshold
The final framework also establishes a formal process for identifying model changes important enough to require public input before implementation. A change can qualify as material if the Fed estimates that it would alter an individual covered firm’s projected post-stress CET1 ratio by at least 20 basis points or meet the separate 10-basis-point aggregate threshold specified by the rule.
The final rule also expands the population considered when determining materiality. Instead of looking only at firms participating in the upcoming test, the Fed will consider qualifying firms with at least $100 billion in consolidated assets that remain subject to the stress-test rules, including certain firms not scheduled to participate in that particular year’s test.
Material changes must generally be published by August 31 before the relevant test and receive at least 30 days of public input. This formalizes a process that previously did not provide an annual pre-implementation public-comment stage for material model changes.
Trading banks will face two global market shocks
The final package also changes the global market shock, or GMS, applied to firms with substantial trading activity. Instead of relying on a single GMS outcome, the Fed expects to construct two market-shock scenarios using the same as-of date and use whichever produces the larger loss for each affected firm when calculating its stress-test result.
At the same time, the Board declined to add specific digital-asset shocks. The final rule says the FR Y-14Q data currently do not capture banks’ digital-asset exposures at sufficient granularity to construct those shocks and that collecting the additional information would increase reporting burden. The Fed left open the possibility of reconsidering the issue later.
What the final rules do not establish
The documents do not establish that large banks’ required capital will fall by 50%. The 50% estimate concerns volatility — the size of year-to-year changes, rather than the level of required capital.
Nor does the final package guarantee that volatility will fall by exactly 50% in future stress tests. The number comes from the Fed’s analysis of how the combined changes would have behaved using historical stress-test cycles and assumptions. Actual future results will still depend on banks’ balance sheets, supervisory models and the scenarios used in each test.
The averaging rule itself also carries a documented tradeoff. Because half of the relevant stress-capital decline can come from the previous year’s test, the requirement responds more slowly to changes in a firm’s current risk profile. The Fed explicitly identifies that loss of timeliness as a cost of averaging.
The next important dividing line is therefore not September 30’s announcement but implementation: the revised stress-test process begins feeding into the 2027 cycle, SCB requirements shift to a January 1 effective date beginning in 2028, and actual two-year results averaging first enters the capital requirement effective January 1, 2029.
