LONDON — The Federal Reserve Board will consider final revisions to its bank stress testing framework "in the coming weeks," Vice Chair for Supervision Michelle W. Bowman said on Sept. 18, 2026, framing the upcoming action as the culmination of a multiyear effort to address what she called fundamental flaws in the way U.S. bank capital requirements and supervision have worked since the stress testing program was introduced in the aftermath of the financial crisis.

Bowman delivered the speech, titled "The Final Chapter on Modernizing Bank Regulatory Stress Testing," at the Lord Mayor of the City of London's luncheon at Mansion House. The setting, she noted, is fitting: located opposite the Bank of England and beside the Royal Exchange, Mansion House has long stood at the center of London's civic and financial life — a reminder that banking has always depended on public confidence as much as on private enterprise. Her prepared remarks were published on the Federal Reserve's website, with a PDF version also available.

"It is an honor and a pleasure to join you to discuss our efforts to modernize U.S. banking regulations and supervision," Bowman said, describing the stress test framework as one that "creates a legally sound regulatory process for setting capital requirements and enhancing supervision."

Three Actions in the Weeks Ahead

According to Bowman, the final revisions the Board will consider are intended to improve transparency and public accountability, strengthen the reliability and accuracy of the models, and reduce the volatility of the capital requirements that flow from the annual stress test results. In the U.S. framework, the Board incorporates forward-looking supervisory stress testing into large bank capital requirements through stress capital buffer (SCB).

Separately, Bowman said she expects that before the end of the year the Board will finalize reforms to risk-based capital requirements for both large and small banks, along with improvements to the global systemically important bank (GISB) surcharge. Taken together, she argued, these reforms will create "a durable and lasting framework" that produces capital requirements that are "rational, robust, aligned with risk, and transparent."

Bowman described stress testing as a valuable mechanism for gauging large-bank resilience. Hypothetical scenarios, she explained, allow the public and Federal Reserve supervisors to assess whether the nation's largest banks hold sufficient capital to absorb losses and continue lending to households and businesses through times of severe stress. At the same time, she acknowledged that the tool "is not without flaws." Under her direction, she said, staff have worked to improve the framework and its implementation so that stress testing is "credible, effective, and fair," with the goal of eliminating opacity and unpredictability.

Difficult Lessons and a Change of Course

The program she outlined represents the culmination of a multiyear effort to address fundamental flaws in the approach to bank capital requirements and supervision that has prevailed since stress testing was introduced following the financial crisis, Bowman said. In an effort to address legal challenges and make overall improvements, the Federal Reserve has incorporated lessons learned from the outset of the program — some of them difficult, she acknowledged, "especially for an institution like the Federal Reserve Board."

Learning and incorporating those lessons, she argued, is "absolutely critical to the legitimacy of the stress test results." Among the imperatives she cited: process and model transparency, public accountability, supportable and repeatable outcomes, and proactive regulatory action. Once the reforms are finalized, she said, the Fed "will finally close the book on an opaque and unnecessarily unpredictable framework" and can begin anew with one that complies with U.S. administrative procedure laws, better promotes the stability of the financial system, and better ensures the safety and soundness of the largest banks.

Bowman also provided context on how the Fed arrived at this point. The Board made a public commitment to seek comment on comprehensive changes to the supervisory stress test and to make public input an ongoing feature of the process. In December 2024, the Board committed to the basic contours of those changes, including disclosing — and then seeking public comment on — all models used to determine the hypothetical losses and revenues of the banks subject to the tests. The commitment also covers changes to the framework that guides the design of hypothetical scenarios and, beginning with the 2026 stress test, the hypothetical scenarios themselves.

Despite that progress, Bowman was blunt about the program's history. Since its inception, she said, the Board's stress testing program has been broadly criticized for its limited transparency, its unreasonable year-over-year volatility, and the absence of any meaningful appeals process. The Board had been keenly aware of these deficiencies for many years, she argued, but considered them "a feature, not a bug," and even after mounting threats of litigation were realized, it did not take meaningful remedial action. "Upon my confirmation as Vice Chair for Supervision, I swiftly changed course," Bowman said.

Rule One: Transparency and Public Accountability

Last year, the Board issued two proposals designed to increase the transparency and public accountability of the capital-based stress test while maintaining its dynamism. The existing framework — including the stress test models, the scenario design framework, and the specific scenarios — is not adequately disclosed or subject to public comment. Bowman said this lack of transparency can lead to uncertainty for banks in capital planning, potential misalignment of capital requirements with actual risks, and limited public understanding and scrutiny of the stress testing process.

The first of the two measures is the Enhanced Transparency and Public Accountability final rule. As proposed, it requires the Board to publish detailed information on the stress test models and on the scenario design process. The model disclosures would publish the equations, variables, coefficients, underlying assumptions, limitations, and decision-making rationale for each model, as well as alternative model considerations and the specific model changes the Board plans to implement for the 2027 stress test. In Bowman's view, this appropriately creates public accountability for the models that determine losses under stress and lead to the calibration of the resulting capital requirement.

Publishing the scenario design process will also enhance and expand the accompanying guides for scenario variables. Currently, guides are provided for only two variables — the unemployment rate and housing prices, along with their change over the test projection horizon. Those two guides will be enhanced, and additional detailed guides will be developed for variables related to the U.S. and international economies. The Fed will also disclose details about variable paths in the macroeconomic model, the framework and model components used to create the global market shock, and other changes in how scenarios are designed. The aim, Bowman said, is to enable the public to access all relevant information needed to properly evaluate the hypothetical scenario each year.

Rule Two: Averaging Test Results to Reduce Volatility

The second final rule addresses the volatility of the SCB requirement by averaging the results of a bank's two most recent annual stress tests. These exercises examine the effect of stress capital levels on the resilience of large banks, estimating losses, revenue, and capital levels under a hypothetical severe recession scenario. The results are used to determine the calibration of the SCB, which is one component of the amount of capital large banks must hold to provide a buffer against potential losses. Bowman attributed the significant volatility of past results to changes in annual test administration and in hypothetical scenario design.

The new approach reduces that volatility by averaging results over two consecutive years. In addition, the annual effective date of the SCB requirement will be delayed from Oct. 1 to Jan. 1 of the following year, aligning compliance with the calendar year and providing banks additional time to implement the resulting capital requirements. Taken together, Bowman said, the two rules reduce SCB volatility by half without materially changing aggregate levels of required capital.

She also highlighted two of the comments the Fed received during the process. The first addresses risk sensitivity and recommends establishing a specified date to freeze firm balance sheets prior to the release of the proposed scenarios shortly thereafter. The second improves risk capture by recommending that two global market shock scenarios be incorporated on the same as-of date, with the larger loss used to calculate the SCB — an approach that would reduce volatility in stress test results. "I look forward to considering these approaches in the final rule," Bowman said.

A Revised 2027 Model for Noninterest Income

In addition to finalizing the two rules, Bowman said she will recommend that the Board consider a third proposal to revise the 2027 test models — an approach that was also informed by public feedback, which she said once again underscores the value of the comment process.

That proposal will seek public comment on a revised model for noninterest income, the fee and trading revenue that banks earn from activities including wealth management, investment banking, and making. The revised model is intended to better capture business diversity across firms and would replace the existing noninterest income model for the 2027 stress test.

Expected Benefits and the Case for Risk Sensitivity

Addressing long-standing criticisms is only one motivation for the work to improve the tests, Bowman said, and she listed a number of further benefits. First, greater transparency increases public accountability, instills confidence in the fairness of the supervisory test, and strengthens market discipline: with a clearer view, the public — including investors, counterparties, and rating agencies — can better assess a firm's risk profile. Second, the improved disclosure process will lead to better models because it creates a new mechanism for obtaining public feedback. Finally, firms will better understand how their risk profiles factor into their capital requirements, and combined with lower year-over-year volatility, that will allow them to plan capital and business decisions more effectively.

The rationale for the changes, she said, is straightforward. A risk-sensitive stress testing framework ensures that SCB requirements align with risk-taking behavior: if firms take additional risks, they should expect those risks to be reflected in greater losses in the stress test, which in turn would lead to higher capital requirements — and the reverse is also true. Combined with minimum risk-based capital requirements, risk-sensitive SCBs ensure that large banks remain resilient and can continue lending under severely stressful conditions. Bowman added that these improvements, which stem from the public comments the Fed received, further underscore the value of greater transparency and accountability in the stress testing framework.

Beyond Capital: Stress Testing as a Supervisory Lens

Bowman then looked ahead to an expanded role for stress testing in supervision. To date, she noted, reform efforts have focused on bringing transparency and public accountability to the capital-related stress test, but stress testing is also intended to more broadly improve the resilience of the financial system and increase the public's confidence that the banking system will hold up to a wide range of shocks.

In theory, she said, stress tests should identify unique firm vulnerabilities to material financial and nonfinancial risks, and provide insight into the stressful scenarios that would cause those risks to materialize — allowing both firms and supervisors to understand the scenarios that pose material threats to a firm's safety and soundness and to U.S. financial stability.

She emphasized that this is not a novel concept or new approach. Large banks already deploy a range of stress testing exercises, including scenario analyses and enterprise-wide stress tests, and use the results for their own business purposes: informing risk management, business planning, and continuity of operations during times of economic and financial stress. "Just as banks rely on stress testing to manage their own risk, the Fed's stress tests should do more for our supervisors than calculate a single capital requirement," Bowman said. In her view, Fed stress tests should also identify a firm's vulnerability to various material financial and nonfinancial risks before those risks emerge, helping to determine how best to focus supervisory attention across the large bank portfolio. Because these exercises would serve principally as forward-looking supervisory tools, their results would not be made public.

She said she envisions the approach being implemented in a number of ways, and highlighted those she sees as most likely to take shape. Additional scenarios that stress bank balance sheets could serve as a forward-looking tool for the early identification of idiosyncratic and emerging risks — precisely the risks that could threaten the safety and soundness of a firm or U.S. financial stability. The results of such analyses, she added, would help prioritize the supervisory approach toward material financial and nonfinancial risks.

The Silicon Valley Bank Counterfactual

Bowman used the Silicon Valley Bank (SVB) experience to illustrate the potential value of this approach. Had the Fed applied this type of analysis, she said, supervisors could have measured the effect of a range of stressful scenarios on SVB's portfolios, exposures, and capital and liquidity positions. A plausible recession scenario — with rising interest rates and investment securities measured at fair value — would have shown the firm's capital falling below minimum required levels as early as the fourth quarter of 2021.

In her view, the early identification of the firm's vulnerability to interest rate risk and its concentration of uninsured deposits would have made it "nearly impossible" to defer prompt and decisive action to require the firm to reduce its vulnerabilities. Combined with other supervisory evidence, she said, it would have prompted earlier action. Analysis of this kind would inform supervisors where to direct their attention and encourage timely escalation. Such forward-looking analysis could also be run on part of a bank's portfolio using bespoke models and assumptions, better informing and directing examinations.

Dialogue With Firms and Reverse Stress Testing

The SVB example, Bowman said, illustrates that scenario analysis is powerful on its own — and becomes even more powerful when it is not a one-way exercise. Banks are already running their own versions of this work internally, and she said she believes there is "tremendous value in comparing notes."

Open dialogue between firms and supervisors, she argued, promotes mutual learning about firm-specific risks, alternative scenarios pertinent to a firm's unique risks, and second-order effects for the broader banking system. Large banks design and assess — in some cases on a daily basis — the most severe scenarios they may face. Supervisors, in turn, will engage to understand firms' vulnerability to the risks uncovered by their own analyses and contrast those findings with the Fed's stress test analyses of the same banks.

Enterprise-wide stress tests offer a concrete example of the practice: they translate a narrative about a firm's risks, combined with severe macroeconomic and financial drivers, into projected losses, net revenues, and capital levels under stress. Depending on its business model, a firm might use the exercise to probe a specific vulnerability — such as concentrated exposure to a single counterparty or to private credit — or a combination of risks that quietly erodes its financial health over time. These, Bowman noted, are precisely the kinds of risks that can build over multiple quarters without triggering an alarm — which is why direct dialogue between Fed supervisors and firms about what their own analyses show must be encouraged.

The final approach Bowman previewed is reverse stress testing, in which firms design a scenario that would materially impair their own financial condition. The exercise provides supervisory staff with valuable insights into a bank's risk management, solvency, and liquidity under stress, as well as the firm's strategy and plan to mitigate risks during stress. Reverse stress testing can also help address gaps in supervisors' understanding of other firms' risk, because a severely stressful scenario for one firm can also be severe for firms with similar balance sheet characteristics. Bowman called the exercise an important opportunity for supervisors and firms to engage on firm-specific vulnerabilities and potential spillovers to the broader U.S. financial system.

A "New Era" for Stress Testing

Stepping back, Bowman said the theoretical purpose of stress testing is not simply to set capital requirements. From its roots as a reform emerging from the global financial crisis, stress testing has always been intended to resilience and confidence in the U.S. banking system. Now, nearly two decades later, she said it is her responsibility as the Board's Vice Chair for Supervision to reinforce the Board's commitment to regulatory stress testing and to enhance transparency and public accountability in the regulatory process — while making capital-related stress testing more accurate, effective, predictable, and fair.

"As we count down to finalization of these rules, we are ready to begin a new era for stress testing: using this powerful tool to enhance supervision," Bowman said. The work, she stressed, will not replace the judgment of the Fed's examiners or the responsibility of banks to manage their own risks; implemented effectively, it will sharpen both. In the coming months, as the proposals are finalized, she said she will share more about how supervisors will explore scenario analysis and open dialogue with firms, and how reverse stress testing can become a regular part of the supervisory process for the largest banks.

A footnote to the prepared text states that the views expressed are Bowman's own and are not necessarily those of her colleagues on the Federal Reserve Board or the Federal Open Market Committee.