Citizen Portal
Sign In

Get AI Briefings, Transcripts & Alerts on Local & National Government Meetings — Forever.

Federal Reserve board votes to publish stress‑test models, scenario guidance and 2026 scenario for public comment

Federal Reserve Board · October 24, 2025

Loading...

AI-Generated Content: All content on this page was generated by AI to highlight key points from the meeting. For complete details and context, we recommend watching the full video. so we can fix them.

Summary

The Federal Reserve Board approved a package to disclose supervisory stress‑test model descriptions, codify scenario‑design guidance and publish the 2026 stress‑test scenario for public comment, over a single dissent warning the move could weaken the test by enabling gaming.

The Federal Reserve Board on Wednesday voted to publish for public comment detailed descriptions of the models used in its supervisory stress tests, a policy statement setting out how stress‑test scenarios are designed, and the proposed severely adverse scenario for the 2026 stress test. The package also would codify an annual process to publish future scenarios and material model changes for public comment.

Board members said the move aims to increase transparency and predictability in a decade‑old supervisory program that links stress‑test results to firms’ stress capital buffers. Vice Chair for Supervision Bowman told colleagues the proposals fulfill a December commitment to disclose the models and invite public input, and he framed the changes as necessary to improve due process and public accountability.

In a staff presentation, the board was asked to publish full model descriptions — including equations, variables, coefficients, assumptions and limitations — and to invite comment on planned 2026 model changes intended to address sensitivities in pre‑provision net revenues and improve operational‑risk and securities modeling. The package would also expand scenario‑design guidance beyond the current focus on unemployment and house prices to a broader set of U.S. and international variables, simplify the global market shock (reducing the set of published shocks while publishing mapping documentation), and change the stress‑test ‘‘jump‑off’’ date from Dec. 31 to Sept. 30 so scenarios are public after firms finalize balance‑sheet data.

Staff estimated the aggregate effect of the proposed model and scenario revisions would not materially change capital requirements for firms subject to the supervisory stress test across a range of scenarios. The presentation also proposed revisions to reporting forms that would remove items no longer needed for the test and add targeted fields to improve risk capture; staff said the removals would reduce supporting documentation burden by over 10,000 pages on average per firm.

Board members asked detailed questions. Vice Chair Jefferson raised concerns that moving the jump‑off date and a separate averaging proposal could make inputs ‘‘stale’’ — relying on balance‑sheet data a year or more old — and asked whether transparency could substitute for averaging in reducing volatility. Staff said the chosen as‑of date limits firms’ incentives to shift balance sheets in response to scenarios while acknowledging trade‑offs between freshness of inputs and predictability.

Members pressed on modeling choices: a proposed revision to the corporate‑loan probability‑of‑default estimation uses a broader sample and more granular post‑crisis data, which staff said lowers estimated losses for sound statistical reasons; some market‑risk and fair‑value assumptions were described as less conservative based on broader bank data; and shortening liquidity horizons in the global market shock to concentrate assumed realizations within one quarter would reduce outlier outcomes on both gains and losses. Staff said some easing in particular assumptions would be largely offset by other changes, muting net capital impacts.

Not all members supported the package. Governor Barr said he could not support the changes, arguing that disclosure of models and scenarios would make the stress test weaker and less credible by enabling banks to game the rules and reduce voluntary buffers. He warned the changes would be difficult to reverse once models are disclosed.

A motion to approve publication for comment — including a notice of proposed rulemaking on enhanced transparency and public accountability, an invitation to comment on the 2026 scenario, and publication in the Federal Register — was moved, seconded and carried by roll call. Recorded votes: Vice Chair Jefferson — yes; Vice Chair for Supervision Bowman — yes; Governor Waller — yes; Governor Cook — yes; Governor Barr — no; Governor Myron — yes; Chair Powell — yes. The board encouraged public comment during the posted comment period and adjourned the meeting.

The board’s action initiates a public comment period that staff and members said will inform any future rulemaking or final model changes. The proposals do not represent Federal Reserve forecasts; staff asked for public input on the narrative and calibration of the 2026 scenario and on the trade‑offs between predictability, volatility reduction, and risks of gaming.