NextFin News - The Federal Reserve will consider final revisions to its bank stress-testing framework within weeks, Vice Chair for Supervision Michelle W. Bowman said on September 18, 2026, capping a multiyear overhaul that she said will "finally close the book on an opaque and unnecessarily unpredictable framework." The coming package pairs two rules that cut stress-capital-buffer volatility in half with new transparency mandates and a third, non-capital supervisory exercise designed to make stress testing a sharper tool for spotting firm-specific risk.
The stakes are concrete. The stress capital buffer (SCB) is the bank-specific slice of capital that moves every year with the annual stress test, calculated as the larger of 2.5 percent or a bank's projected peak-to-trough Common Equity Tier 1 decline in the severely adverse scenario plus four quarters of planned dividends as a share of risk-weighted assets. It is the single most consequential output of the annual exercise, and a framework that can swing a bank's binding capital floor by hundreds of basis points year to year drives behavior: capital planning, dividend policy, buybacks, even business mix. Bowman's answer — smoothing the buffer, publishing the models, and splitting capital-setting from supervision — is the Fed's bid to make the post-crisis stress test legally defensible and market-credible at the same time.
The Situation: A Deadline, a Lawsuit, and a System That Just Passed
Bowman delivered the remarks at the Luncheon of the Lord Mayor of the City of London at Mansion House, standing opposite the Bank of England — a deliberate stage for a speech about the legitimacy of bank capital regulation. The timeline she set is tight:
In the coming weeks, the Federal Reserve Board will consider final revisions to the Board's stress testing framework.
And before the end of 2026, she expects the Board to finalize a broader set of risk-based capital reforms for large and small banks plus improvements to the global systemically important bank (G-SIB) surcharge. In totality, she said, "these reforms create a durable and lasting framework to produce capital requirements that are rational, robust, aligned with risk, and transparent."
The announcement is the last act of a process that began in earnest in December 2024, when the Board committed to disclose — and then seek public comment on — all models used to determine banks' hypothetical losses and revenues, and to open the scenario-design process to comment beginning with the 2026 stress test. Those commitments were not made in a vacuum. Days after the December 2024 announcement, the American Bankers Association, the Bank Policy Institute, the U.S. Chamber of Commerce and two Ohio banking groups sued the Federal Reserve in the U.S. District Court for the Southern District of Ohio, arguing the stress-testing framework was adopted in secret and violated the Administrative Procedure Act. The Fed and the plaintiffs then jointly moved to pause the litigation while the rulemaking ran its course. Bowman's speech is the Fed's answer to that challenge: transparency, codified in rule text.
The backdrop is a banking system that just cleared its most recent exam with room to spare. The 2026 supervisory stress test, released June 24, found all 32 participating banks remained above their minimum capital requirements under a severely adverse scenario in which the unemployment rate rises 5.5 percentage points to 10 percent, real GDP contracts 4.6 percent, house prices fall 30 percent, commercial real estate prices drop 39 percent, and equity prices tumble 58 percent. Despite absorbing more than $708 billion in total losses under the scenario, aggregate CET1 capital declined only 1.6 percentage points, from 12.8 percent to a stressed minimum of 11.2 percent, before recovering to 12.7 percent at the end of the projection horizon. The projected losses included roughly $200 billion in credit card losses, $160 billion in commercial and industrial loan losses, and $75 billion in commercial real estate losses.
Yet the test that produced those results will not set new capital requirements until 2027. In February 2026, the Board voted to maintain current SCB requirements through September 30, 2027, so the June results would not force banks to raise fresh capital. That freeze is what allowed the largest lenders to move immediately to shareholder returns: JPMorgan Chase raised its quarterly dividend to $1.65 from $1.50 and authorized a $50 billion repurchase program; Goldman Sachs lifted its payout to $5.00 from $4.50; Morgan Stanley boosted its dividend 15 percent to $1.15 and reauthorized $20 billion in buybacks; Wells Fargo said it intends to raise its quarterly dividend 11 percent to $0.50 from $0.45, pending board approval. The 2026 capital-requirements document confirms the spread in binding buffers: JPMorgan and Wells Fargo sit at the 2.5 percent floor, Goldman Sachs at 3.4 percent, and Citigroup at 3.6 percent.
The Two Rules: Smoothing Capital Without Cutting It
The first lever is mechanical. The second final rule reduces SCB volatility by averaging the results of a bank's two most recent annual stress tests. Instead of one year's scenario — with its one-off unemployment path and its single global market shock — setting the entire buffer, the buffer will reflect two consecutive cycles. Bowman put the arithmetic plainly:
Together, these rules reduce SCB volatility by half without materially changing aggregate levels of required capital.
That framing is the crux of the whole overhaul. This is not a capital cut; it is a capital-smoothing device. Aggregate required capital is roughly unchanged, but the year-to-year path is flattened. For bank treasurers, that changes the variance of the constraint, not its level — and variance is what makes capital planning painful. A buffer that can jump because of a single severe scenario design forces banks to hold excess capital through the good years as insurance against a bad test. Smoothing cuts that precautionary hoard without lowering the floor.
The second lever is timing. The annual effective date of the SCB requirement moves from October 1 to January 1 of the following year. Under the current calendar, the Fed publishes results in late June and banks must be living with the new buffer by October 1 — roughly three months to adjust capital plans, dividend announcements, and balance-sheet positioning. Pushing the effective date to January 1 aligns the requirement to the calendar year and gives firms an additional quarter to implement. The change "aligns compliance to calendar year and provides additional time to implement the resulting capital requirements," the speech said.
Two comments the Fed received — and is signaling it will adopt — show how far the redesign goes toward pre-commitment. One recommends freezing firm balance sheets at a specified date before the scenarios are released, removing the game of banks reshaping their books once they see the test. The other recommends running two global market shock scenarios on the same as-of date and using the larger loss in the SCB calculation. "This approach will reduce volatility in stress test results," Bowman said of the latter. "I look forward to considering these approaches in the final rule." The direction is unmistakable: less discretion, more rules set in advance.
Transparency as Legal Armor
The first final rule — the Enhanced Transparency and Public Accountability rule — is the one that answers the lawsuit. It requires the Board to publish detailed information on the stress test models and the scenario-design process. Model disclosures will include the equations, variables, coefficients, underlying assumptions, limitations, and decisionmaking rationale for each model, plus alternative model considerations and the specific model changes the Board plans to implement for the 2027 stress test.
The scenario side is equally consequential. Currently, the Fed publishes design guides for only two variables — the unemployment rate and housing prices and their change over the projection horizon. Those two guides will be enhanced, and additional detailed guides for variables tied to the U.S. and international economies will be developed. The Fed will also disclose details about variable paths in the macroeconomic model and the framework and model components used to create the global market shock. The public, in effect, gets to see the machinery that determines how much capital the largest banks in the world must hold.
Why does this matter beyond good governance? Because a stress test that cannot explain its own math is a stress test that cannot survive a court challenge. The plaintiffs in the 2024 suit did not argue that stress testing is bad policy; they argued the models were "adopted in secret" and produced "vacillating and unexplained requirements." Publishing equations and coefficients converts a black box into a reviewable record. It also lets banks and their creditors model their own buffers with some confidence — which is precisely what reduces the opacity premium that the market has quietly been charging bank valuations for years.
There is a cost, and it is built into the design. Transparency reduces the Fed's freedom of maneuver. A model published in advance can be gamed; a scenario guide telegraphs the shock. That is the classic trade-off in supervisory design: predictability buys legitimacy, but predictability also buys adaptability. The Fed's answer is to separate the capital-setting exercise from the exploratory one — which is exactly what the third pillar does.
The Third Pillar: Supervision Without Capital Consequence
Alongside the two final rules, Bowman recommended that the Board consider a third proposal revising the 2027 test models, and previewed an expanded forward-looking approach. These exercises will be, in her words, "targeted to identify risks and vulnerabilities," leveraging the underlying data from the capital-based tests:
Importantly, the results of these forward-looking tests would not affect capital requirements.
The toolbox has three instruments. Scenario analysis would pair supervisory stress scenarios with open dialogue between examiners and firms about what the firms' own analysis shows. Reverse stress testing asks each firm to design the scenario that would materially impair its own financial condition — a useful inversion, because the scenario that breaks one balance sheet is often the one that matters for a peer with similar exposures. The goal, in Bowman's words, is to "deepen our understanding of exposures to material financial and nonfinancial risks and their resiliency to those risks."
This is the structural insight of the overhaul: the capital-setting stress test becomes narrower, more mechanical, and more transparent, while the supervisory stress test becomes broader, more qualitative, and more probing. The first tells the market how much capital a bank must hold; the second tells the supervisor where the bodies are buried. Keeping the second outside the capital formula is what lets the Fed ask harder questions without triggering automatic capital consequences — and without inviting another Administrative Procedure Act challenge over unpublished models.
This work will not replace the judgment of our examiners or the responsibility of banks to manage their own risks—implemented effectively, it will sharpen both.
What the Market Is Pricing — and What It Is Not
Bank equities have spent 2026 rallying into exactly this outcome. The trade has been straightforward: stress-test volatility falls, capital requirements become more predictable, and the capital that banks no longer need to hold as a precaution flows back to shareholders through dividends and buybacks. The June 2026 test already set the template — because SCB requirements were frozen while the framework was revised, banks did not wait to announce capital returns, and the sector's biggest names moved within hours of the results.
The second-order question is whether the market has priced the right thing. The conventional read is that lower SCB volatility means lower capital and higher payouts. That is the first-order effect, and it is already in bank valuations. The second-order effect runs a different channel: a more predictable, legally durable framework reduces the tail risk of a sudden, litigation-driven capital demand — the kind of regulatory uncertainty that compresses bank price-to-book multiples even when earnings are strong. In that sense, the value of the overhaul is not primarily the capital it frees; it is the reduction in the discount rate the market applies to bank earnings because the rules might change without warning. A rule that survives court is worth more than a rule that is stricter but gets tied up for years.
That is also why the sequencing matters. Finalizing the stress-test rules in the coming weeks, then the risk-based capital and G-SIB surcharge reforms before year-end, gives the Fed a complete capital package heading into 2027. The Basel III "endgame" proposals have hung over the sector for years; a finished, defensible U.S. framework lets banks plan multi-year balance sheets rather than hedge against regulatory surprise. The market is not just pricing less capital volatility; it is pricing the end of a regulatory cliffhanger.
The Counter-Thesis: Is Smoothing Just Loosening by Another Name?
The strongest case against this overhaul is that smoothing is a form of loosening in disguise. A bank that would have faced a sharply higher SCB in a bad year now averages that year with a benign one; the peak constraint is lower, and the system carries less capital precisely when it is most likely to need it. Averaging two years of tests also means the buffer reflects a recession that is, by construction, partially in the rear-view mirror. If the next shock arrives faster than the averaging window can capture, the capital built during calm years may prove insufficient — not because the math is wrong, but because the math is slow. Creditors who relied on the annual test as a hard, comparable signal of resilience would see that signal move more slowly, and more of the information burden would shift to a supervisory dialogue that is, by design, less public.
There is force to that objection, but it misses what the Fed is actually doing. Aggregate required capital is not being cut; it is being smoothed. The 2.5 percent floor remains, the 4.5 percent CET1 minimum remains, and the G-SIB surcharge sits on top. More importantly, the forward-looking supervisory exercises are designed to catch the fast-moving risk that the smoothed capital formula might miss — the commercial-real-estate concentration, the leverage in nonbank intermediaries, the funding mismatch that a two-year average cannot see. The capital floor is the shield; the qualitative supervision is the early-warning radar. One without the other is what produced the last two decades of boom and bust in bank regulation.
The honest vulnerability in Bowman's framework is different, and it is procedural. The Fed is promising final rules "in the coming weeks" and a full capital package by year-end. If the Board misses those dates — if the transparency rule triggers another round of comment, or if the G-SIB surcharge reform reopens old fights — the credibility of the "predictable framework" narrative takes a hit before it is even established. The market is not pricing delay risk; it is pricing delivery.
What Comes Next: Beneficiaries, the Exposed, and the Signal to Watch
The beneficiaries are the banks with the most volatile SCBs — the global dealers whose trading books swing the global market shock, and the lenders with concentrated exposures that produce lumpy stress losses. A smoother buffer is worth more to a firm whose requirement jumps three percentage points in a bad year than to one that sits at the 2.5 percent floor every year. The exposed are the creditors and counterparties who relied on the annual test as a hard signal: if buffers move more slowly, the market signal moves more slowly too, and more of the information burden shifts to the supervisory dialogue — which, by design, is less public.
Split by horizon, the picture is mixed. In the short term, the announcement is a tailwind for bank equities: less regulatory uncertainty, a clearer path to capital returns, and a Fed that is visibly finishing what it started. Over the medium term, the test is implementation — whether the averaged SCB actually reduces volatility without eroding the buffer, and whether the model disclosures are detailed enough to satisfy both the courts and the market. Over the long term, the question is whether a transparent, pre-committed framework proves more resilient in the next crisis than a discretionary one. The historical record is not encouraging: rules written for the last crisis rarely fit the next one. Bowman's answer — a mechanical capital test plus a flexible supervisory overlay — is an attempt to have both. It is the right design, provided the supervisory side is given teeth.
Three scenarios frame the path ahead. In the base case, the two rules are finalized in the coming weeks with the two-year averaging and the January 1 effective date intact; SCB volatility falls roughly as projected; bank capital returns continue at the 2026 pace; and the forward-looking exercises begin in 2027 without affecting capital. In the upside case, the model disclosures exceed expectations, the litigation is formally resolved, and the completed year-end capital package — risk-based rules plus G-SIB surcharge — removes the last overhang on bank valuations, setting up a multi-year re-rating of the sector. In the downside case, finalization slips past the stated timeline, or the averaged buffer proves too slow in a fast-moving downturn, forcing the Fed to choose between its predictability pledge and a swift capital response.
The falsifying signal is specific and sits in the rule text, not the speech: if the final rules do not include the two-year averaging of SCB results and the January 1 effective date — or if aggregate required capital falls materially rather than staying roughly flat — then the "smoothing, not loosening" thesis is wrong, and the market should read this as deregulation, not reform. Watch the rule text, not the rhetoric.
Bowman is not cutting capital; she is cutting uncertainty. And in bank regulation, uncertainty has been the more expensive of the two.
Explore more exclusive insights at nextfin.ai.
