Proposed Rule2025-20211
Enhanced Transparency and Public Accountability of the Supervisory Stress Test Models and Scenarios; Modifications to the Capital Planning and Stress Capital Buffer Requirement Rule, Enhanced Prudential Standards Rule, and Regulation LL
Primary source
Metadata and text below are from the Federal Register, a public-domain U.S. government work. Always verify the official published version before relying on it for any legal matter.
Published
November 18, 2025
Issuing agencies
Federal Reserve System
Abstract
The Board is inviting public comment on the models used to conduct the Board's supervisory stress test, changes to those models to be implemented in the 2026 stress test, and proposed changes to enhance the transparency and public accountability of the Board's stress testing framework (the proposal). The proposal would amend the Policy Statement on the Scenario Design Framework for Stress Testing, including to implement guides for additional scenario variables, and the Stress Testing Policy Statement. The proposal would also codify an enhanced disclosure process under which the Board would annually publish comprehensive documentation on the stress test models, invite public comment on any material changes that the Board seeks to make to those models, and annually publish the stress test scenarios for comment. Lastly, the proposal would make changes to the FR Y-14A/Q/M to remove items that are no longer needed to conduct the supervisory stress test and to collect additional data to support the stress test models and improve risk capture.
Full Text
<html>
<head>
<title>Federal Register, Volume 90 Issue 220 (Tuesday, November 18, 2025)</title>
</head>
<body><pre>
[Federal Register Volume 90, Number 220 (Tuesday, November 18, 2025)]
[Proposed Rules]
[Pages 51856-51953]
From the Federal Register Online via the Government Publishing Office [<a href="http://www.gpo.gov">www.gpo.gov</a>]
[FR Doc No: 2025-20211]
[[Page 51855]]
Vol. 90
Tuesday,
No. 220
November 18, 2025
Part III
Federal Reserve System
-----------------------------------------------------------------------
12 CFR Parts 225, 238, and 252
Enhanced Transparency and Public Accountability of the Supervisory
Stress Test Models and Scenarios; Modifications to the Capital Planning
and Stress Capital Buffer Requirement Rule, Enhanced Prudential
Standards Rule, and Regulation LL; Proposed Rule
Federal Register / Vol. 90, No. 220 / Tuesday, November 18, 2025 /
Proposed Rules
[[Page 51856]]
-----------------------------------------------------------------------
FEDERAL RESERVE SYSTEM
12 CFR Parts 225, 238, and 252
[Regulations Y, LL, and YY; Docket No. R-1873]
RIN 7100-AH05
Enhanced Transparency and Public Accountability of the
Supervisory Stress Test Models and Scenarios; Modifications to the
Capital Planning and Stress Capital Buffer Requirement Rule, Enhanced
Prudential Standards Rule, and Regulation LL
AGENCY: Board of Governors of the Federal Reserve System (Board).
ACTION: Notice of Proposed Rulemaking.
-----------------------------------------------------------------------
SUMMARY: The Board is inviting public comment on the models used to
conduct the Board's supervisory stress test, changes to those models to
be implemented in the 2026 stress test, and proposed changes to enhance
the transparency and public accountability of the Board's stress
testing framework (the proposal). The proposal would amend the Policy
Statement on the Scenario Design Framework for Stress Testing,
including to implement guides for additional scenario variables, and
the Stress Testing Policy Statement. The proposal would also codify an
enhanced disclosure process under which the Board would annually
publish comprehensive documentation on the stress test models, invite
public comment on any material changes that the Board seeks to make to
those models, and annually publish the stress test scenarios for
comment. Lastly, the proposal would make changes to the FR Y-14A/Q/M to
remove items that are no longer needed to conduct the supervisory
stress test and to collect additional data to support the stress test
models and improve risk capture.
DATES: Comments must be received on or before January 22, 2026.
ADDRESSES: You may submit comments, identified by Docket No. R-1873 and
RIN 7100-AH05, by any of the following methods:
<bullet> Agency website: <a href="https://www.federalreserve.gov/apps/proposals/">https://www.federalreserve.gov/apps/proposals/</a>. Follow the instructions for submitting comments, including
attachments. Preferred Method.
<bullet> Mail: Benjamin W. McDonough, Deputy Secretary, Board of
Governors of the Federal Reserve System, 20th Street and Constitution
Avenue NW, Washington, DC 20551.
<bullet> Hand Delivery/Courier: Same as mailing address.
<bullet> Other Means: <a href="/cdn-cgi/l/email-protection#d3a3a6b1bfbab0b0bcbebeb6bda7a093b5a1b1fdb4bca5"><span class="__cf_email__" data-cfemail="dbabaeb9b7b2b8b8b4b6b6beb5afa89bbda9b9f5bcb4ad">[email protected]</span></a>. You must include the
docket number in the subject line of the message.
Comments received are subject to public disclosure. In general,
comments received will be made available on the Board's website at
<a href="https://www.federalreserve.gov/apps/proposals/">https://www.federalreserve.gov/apps/proposals/</a> without change and will
not be modified to remove personal or business information including
confidential, contact, or other identifying information. Comments
should not include any information such as confidential information
that would be not appropriate for public disclosure. Public comments
may also be viewed electronically or in person in Room M-4365A, 2001 C
St. NW, Washington, DC 20551, between 9 a.m. and 5 p.m. during Federal
business weekdays.
FOR FURTHER INFORMATION CONTACT: Doriana Ruffino, Assistant Director,
(202) 452-5235, Hillel Kipnis, Assistant Director, (202) 452-2924, John
Simone, Lead Financial Institution Policy Analyst, (202) 245-4256, Ben
Ranish, Principal Economist, (202) 973-6964, Nathan Palmer, Senior
Economist, (202) 785-6089, and Theo Pistner, Financial Institution and
Policy Analyst II, (202) 941-1825, Division of Supervision and
Regulation; William Bassett, Senior Associate Director, (202) 736-5644,
Bora Durdu, Deputy Associate Director, (202) 452-3755, Elena
Afanasyeva, Principal Economist, (202) 736-1971, Levent Altinoglu,
Principal Economist, (202) 721-4503, and Sam Jerow, Senior Financial
Analyst, (202) 245-4299, Division of Financial Stability; Asad Kudiya,
Associate General Counsel, (202) 360-6887, Julie Anthony, Senior
Special Counsel, (202) 658-9400, Jonah Kind, Senior Counsel, (202) 452-
2045, Brian Kesten, Senior Counsel, (202) 843-4079, Katherine Di
Lucido, Senior Attorney, (202) 253-5994, Legal Division. Board of
Governors of the Federal Reserve System, 20th Street and Constitution
Avenue NW, Washington, DC 20551. For users of TDD-TYY, please call 711
from any telephone, anywhere in the United States.
SUPPLEMENTARY INFORMATION:
Table of Contents
I. Introduction
II. Background on Stress Testing Framework, Stress Test Models, and
Scenario Design Framework
A. Stress Testing Framework
B. Prior Supervisory Stress Disclosures and Policy Statements
C. Supervisory Stress Test Modeling Framework
D. Stress Test Models
E. Summary of the Proposal
F. Purpose of the Proposal
III. Overview of the Stress Test Modeling Framework
A. Supervisory Stress Test Models
B. Supervisory Stress Test Scenarios
C. Data Used in Stress Testing
IV. Enhanced Disclosure Process
A. Annual Disclosure of Models
B. Model Changes
C. Material Model Changes
D. Annual Disclosure of Scenarios
E. Stress Capital Buffer Requirement Reconsideration Process
V. Revisions to the Stress Testing Policy Statement
A. Future Supervisory Stress Test Results Disclosures
B. Other Revisions to the Stress Testing Policy Statement
VI. Other Revisions to the Stress Testing and Capital Plan Rules
A. Stress Test Jump-Off Date Change
B. Global Market Shock Date
C. Amendment to the Dividend Add-On Component Calculation
VII. Revisions to the FR Y-14A/Q/M
VIII. Proposed Changes to the Stress Test Modeling Framework
A. Proposed Changes to Stress Test Models
B. Analysis of Proposed Model Changes
IX. Proposed Changes to the Scenario Design Policy Statement
A. Changes to the Background and Overview and Scope Sections
B. Changes to the Content of the Stress Test Scenarios Section
C. Approach for Formulating Macroeconomic Assumptions in the
Baseline Scenario
D. Scenario Narrative: Refinement to the Recession Approach
E. Changes to Construction of Certain Variables in the Severely
Adverse Scenario
F. Scenario Design Principles Derived From Stress Testing
Literature: Severity, Credibility, and Procyclicality
G. Description of Variable Guides in the Severely Adverse
Scenario
H. Global Market Shock
X. Economic Analysis
XI. Administrative Law Matters
A. Paperwork Reduction Act Analysis
B. Regulatory Flexibility Act Analysis
C. Plain Language
D. Providing Accountability Through Transparency Act of 2023
I. Introduction
In December 2024, the Board announced that it would propose
significant changes to improve the transparency of the supervisory
stress test and reduce the volatility of resulting capital
requirements.\1\ The Board noted it planned to propose changes to
disclose and seek public comment on the models that determine the
[[Page 51857]]
hypothetical losses and revenue of banks under stress and ensure that
the public can comment on the hypothetical scenarios used annually for
the test, before the scenarios are finalized. With this proposal, the
Board is inviting public comment on the comprehensive model
documentation for the 2026 stress test, as well as proposed changes to
the models relative to the 2025 stress test. The comprehensive model
documentation is available at <a href="https://www.federalreserve.gov/supervisionreg/dfa-stress-tests-2026.htm">https://www.federalreserve.gov/supervisionreg/dfa-stress-tests-2026.htm</a>. The Board is inviting comment
on the proposed scenarios for the 2026 stress test through a separate
notice.
---------------------------------------------------------------------------
\1\ See Board, Press Release (Dec. 23, 2024), <a href="https://www.federalreserve.gov/newsevents/pressreleases/bcreg20241223a.htm">https://www.federalreserve.gov/newsevents/pressreleases/bcreg20241223a.htm</a>.
In February 2025, the Board reiterated its previous announcement
that it would begin the public comment process on changes to the
supervisory stress test. See Board, Press Release (Feb. 5, 2025),
<a href="https://www.federalreserve.gov/newsevents/pressreleases/bcreg20250205a.htm">https://www.federalreserve.gov/newsevents/pressreleases/bcreg20250205a.htm</a>.
---------------------------------------------------------------------------
This proposal seeks to improve the transparency and public
accountability of the supervisory stress test, while ensuring that the
stress test remains an effective tool for understanding and assessing
risk and retaining appropriate risk sensitivity and risk capture in
capital requirements.
The Board periodically reviews its regulations, including
transparency efforts surrounding its regulations, to ensure they
continue to achieve their goals in an effective and efficient manner.
In addition to the changes discussed herein, the Board is also
considering the effectiveness of its regulatory capital and capital
planning requirements for large firms to ensure they remain cohesive
and effective, maintain the resilience of the banking sector, and
minimize any unnecessary burden. If appropriate, the Board will make
changes to its rules through the public notice and comment process.
Question 1: The Board seeks comment on all aspects of the proposal.
What, if any, other elements of the supervisory stress test framework
should the Board consider amending to improve the transparency, public
accountability, and effectiveness of the supervisory stress test? For
example, the Board could instead transliterate the models used to
conduct the stress test and codify these transliterations in its
regulations. What would be the advantages and disadvantages of this
approach or other approaches the Board could consider?
II. Background on Stress Testing Framework, Stress Test Models, and
Scenario Design Framework
A. Stress Testing Framework
Congress enacted the Dodd-Frank Wall Street Reform and Consumer
Protection Act (Dodd-Frank Act) in the wake of the 2007-09 financial
crisis.\2\ Section 165 of the Dodd-Frank Act, as amended by section 401
of the Economic Growth, Regulatory Relief, and Consumer Protection
Act,\3\ requires the Board to establish enhanced prudential standards
for nonbank financial companies supervised by the Board and bank
holding companies with $250 billion or more in total consolidated
assets.\4\ The purpose of these enhanced prudential standards is to
prevent or mitigate risks to the financial stability of the United
States that could arise from the material financial distress or
failure, or ongoing activities, of large, interconnected financial
institutions.
---------------------------------------------------------------------------
\2\ Dodd-Frank Wall Street Reform and Consumer Protection Act,
Public Law 111-203, 124 Stat. 1376 (2010).
\3\ Economic Growth, Regulatory Relief, and Consumer Protection
Act, Public Law 115-174, 132 Stat. 1296 (2018).
\4\ See 12 U.S.C. 5365(a). In addition, the International
Lending Supervision Act of 1983 provides the Board with broad
discretionary authority to set minimum capital levels for state
member banks and certain affiliates of insured depository
institutions, including holding companies, supervised by the Board.
See 12 U.S.C. 3902(1); 3907(a); 3909(a). Under section 5(b) of the
Bank Holding Company Act of 1956 (Bank Holding Company Act), the
Board may issue such regulations and orders relating to capital
requirements of bank holding companies as may be necessary for the
Board to carry out the purposes of the Bank Holding Company Act. 12
U.S.C. 1844(b). Foreign banking organizations with a U.S. branch,
agency, or commercial lending company subsidiary are made subject by
the International Banking Act of 1978 (International Banking Act) to
the provisions of the Bank Holding Company Act in the same manner as
bank holding companies, see 12 U.S.C. 3106; therefore, the Board is
also authorized under section 5(b) of the Bank Holding Company Act
to impose these requirements on those foreign banking organizations,
including on their U.S. operations. Similarly, with regard to
savings and loan holding companies, section 10(g) of the Home
Owners' Loan Act authorizes the Board to issue such regulations and
orders relating to capital requirements as the Board deems necessary
and appropriate to carry out the purposes of the Home Owners' Loan
Act. See 12 U.S.C. 1467a(g)(1).
---------------------------------------------------------------------------
Section 165(i)(1) of the Dodd-Frank Act requires the Board to
conduct an annual supervisory stress test of nonbank financial
companies supervised by the Board and bank holding companies with $250
billion or more in total consolidated assets to evaluate whether the
firm has the capital, on a total consolidated basis, necessary to
absorb losses as a result of adverse economic conditions.\5\ Section
401(e) of the Economic Growth, Regulatory Relief, and Consumer
Protection Act requires the Board to conduct periodic stress tests for
bank holding companies with total consolidated assets between $100
billion and $250 billion.\6\ Section 165(i)(1) of the Dodd-Frank Act
requires the Board to publish a summary of the supervisory stress test
results.\7\ In 2012, the Board adopted a final rule implementing the
stress test requirements established in the Dodd-Frank Act.\8\
---------------------------------------------------------------------------
\5\ 12 U.S.C. 5365(i)(1).
\6\ 12 U.S.C. 5365 note (Supervisory Stress Test).
\7\ 12 U.S.C. 5365(i)(1)(B)(v).
\8\ See 77 FR 62378 (Oct. 12, 2012).
---------------------------------------------------------------------------
The Dodd-Frank Act also requires bank holding companies with $250
billion or more in total consolidated assets, as well as nonbank
financial companies supervised by the Board, to conduct company-run
stress tests on a periodic basis.\9\ Under the Board's rules, firms
subject to Category I, II, or III standards must conduct company-run
stress tests.\10\ Company-run stress tests provide forward-looking
information to supervisors to assist in their overall assessments of a
firm's capital adequacy, help to better identify downside risks and the
potential impact of adverse outcomes on the firm`s capital adequacy,
and assist in achieving the financial stability goals of the Dodd-Frank
Act. Further, the company-run stress tests help improve firms' stress
testing practices with respect to their own internal assessments of
capital adequacy and overall capital planning.
---------------------------------------------------------------------------
\9\ 12 U.S.C. 5365(i)(2).
\10\ See 84 FR 59032 (Nov. 1, 2019); 12 CFR 238.142; 12 CFR
252.53. State member banks with average total consolidated assets of
greater than $250 billion must also conduct company-run stress
tests. 12 CFR 252.13.
---------------------------------------------------------------------------
Each June, the Board publishes the results of its annual
supervisory stress test, including each firm's projected capital
ratios, pre-tax net income, losses, revenues, and expenses, under
hypothetical, severely adverse economic and financial conditions.\11\
These disclosures provide the public with valuable information about
each firm's financial condition and the ability of
[[Page 51858]]
each firm to absorb losses considering a stressful economic
environment.
---------------------------------------------------------------------------
\11\ A firm subject to Category I through III standards must
participate in the supervisory stress test every year, while a firm
subject to Category IV standards is generally required to
participate only every other year. See 12 CFR 217.2; 12 CFR 238.10;
12 CFR 252.5; 84 FR 59032 (Nov. 1, 2019). In 2019, the Board adopted
rules establishing four categories of prudential standards for U.S.
banking organizations with total consolidated assets of $100 billion
or more and foreign banking organizations with combined U.S. assets
of $100 billion or more. See 12 CFR 217.2; 12 CFR 238.10; 12 CFR
252.5; 84 FR 59032 (Nov. 1, 2019). Category I standards apply to
U.S. GSIBs and their depository institution subsidiaries. Category
II standards apply to banking organizations with at least $700
billion in total consolidated assets or at least $75 billion in
cross-jurisdictional activity and their depository institution
subsidiaries. Category III standards apply to banking organizations
with total consolidated assets of at least $250 billion or at least
$75 billion in weighted short-term wholesale funding, nonbank
assets, or off-balance sheet exposure and their depository
institution subsidiaries. Category IV standards apply to banking
organizations with total consolidated assets of at least $100
billion that do not meet the thresholds for a higher category and
their depository institution subsidiaries.
---------------------------------------------------------------------------
Following the 2007-09 financial crisis, the Board also made changes
to its capital rule to address weaknesses observed during the
crisis.\12\ These changes included the establishment of a minimum
common equity tier 1 capital requirement and a fixed capital
conservation buffer equal to 2.5 percent of risk-weighted assets.\13\
Large firms also became subject to a countercyclical capital buffer
requirement, and the largest and most systemically important firms--
global systemically important bank holding companies, or GSIBs--became
subject to an additional capital buffer based on a measure of their
systemic risk, the GSIB surcharge.\14\ In 2020, the Board adopted the
stress capital buffer requirement for certain firms.\15\ Because a
firm's stress capital buffer requirement is informed by the firm's
performance under the hypothetical economic conditions modeled by the
supervisory stress test, each firm's stress capital buffer requirement
is tailored to its risk profile.
---------------------------------------------------------------------------
\12\ See generally 12 CFR part 217.
\13\ See 78 FR 62018 (Oct. 11, 2013); 12 CFR 217.11.
\14\ See 80 FR 49082 (Aug. 14, 2015).
\15\ In 2020, the Board finalized a rule to integrate
supervisory stress test results into the capital framework, through
the stress capital buffer requirement. See 85 FR 15576 (Mar. 18,
2020). The stress capital buffer requirement is calculated as the
difference between a firm's starting and lowest projected common
equity tier 1 capital ratio under the severely adverse scenario in
the supervisory stress test plus four quarters of planned common
stock dividends, expressed as a percentage of risk-weighted assets.
See 12 CFR 225.8(f); 12 CFR 238.170(f). The stress capital buffer
requirement framework generally applies to firms with $100 billion
or more in total consolidated assets.
---------------------------------------------------------------------------
Supervisory stress testing and stronger capital requirements have
significantly improved the resilience of the U.S. banking system. Since
2009, the common equity capital ratios of firms subject to the test
have more than doubled, with common equity capital of such firms
increasing by over $1 trillion.\16\ Since 2020, the supervisory stress
test results have also informed a firm's stress capital buffer
requirement. Greater transparency would allow firms to better
understand the capital requirements associated with investment and
expansion of different business lines and would facilitate more
effective long-term capital planning. This, in turn, could enhance
firms' ability to supply credit to households and businesses,
ultimately supporting economic growth and financial stability.
---------------------------------------------------------------------------
\16\ Based on FR Y-9C (Consolidated Financial Statements for
Holding Companies) filings.
---------------------------------------------------------------------------
B. Prior Supervisory Stress Disclosures and Policy Statements
In addition to the annual stress test results disclosure, the Board
has historically published some information about the supervisory
stress test scenarios and models.
Scenarios
The Board's stress test rules provide that the Board will notify
firms, by no later than February 15 of each year, of the scenarios that
the Board will apply to conduct its annual supervisory stress test and
that firms must use to conduct their company-run stress tests.\17\ The
Board also provides a narrative description of the scenarios no later
than February 15 of each calendar year.\18\
---------------------------------------------------------------------------
\17\ See 12 CFR 238.132(b); 12 CFR 238.143(b); 12 CFR 252.14(b);
12 CFR 252.44(b); 12 CFR 252.54(b).
\18\ See, e.g., Board, 2025 Stress Test Scenarios (Feb. 2025),
<a href="https://www.federalreserve.gov/publications/files/2025-stress-test-scenarios-20250205.pdf">https://www.federalreserve.gov/publications/files/2025-stress-test-scenarios-20250205.pdf</a>.
---------------------------------------------------------------------------
In 2013, the Board increased the transparency of the scenarios by
finalizing the Policy Statement on the Scenario Design Framework for
Stress Testing (Scenario Design Policy Statement), which articulated
the Board's approach to scenario design for the supervisory and
company-run stress tests, outlining the characteristics of the stress
test scenarios, and explaining the considerations and procedures that
underlie the formulation of these scenarios.\19\ The Scenario Design
Policy Statement also described the baseline and severely adverse
scenarios, the Board's approach for developing these two macroeconomic
scenarios, and the approach for developing any additional components of
the stress test scenarios. The Scenario Design Policy Statement
explained that the severely adverse scenario is designed to reflect
conditions that have characterized post-war U.S. recessions (the
recession approach). Historically, recessions have typically featured
increases in the unemployment rate, contractions in aggregate incomes
and economic activity, and declines in inflation and interest rates.
---------------------------------------------------------------------------
\19\ 12 CFR part 252, Appendix A.
---------------------------------------------------------------------------
In the 2013 Scenario Design Policy Statement, the Board explained
that, in light of the typical co-movement of measures of economic
activity during economic downturns, such as the unemployment rate and
gross domestic product, the Board would first specify a path for the
unemployment rate and then develops paths for other measures of
activity broadly consistent with the course of the unemployment rate in
developing the severely adverse scenario. The 2013 Scenario Design
Policy Statement also stated that economic variables included in the
scenarios may change over time, and that the Board may augment the
recession approach with certain salient risks, which would involve
incorporating features that address aspects of the current economic or
financial market environment that represent higher-than-normal risks to
the condition of the banking system.
In 2019, the Board updated the Scenario Design Policy Statement,
which increased the transparency and predictability of the scenarios by
allowing for a smaller-than-usual increase in unemployment if the
stress test were to occur during an economic downturn, a change that
would pass through to reduced severity of other key scenario variables
due to the deference given to historical correlations. The 2019 update
also introduced a formula with countercyclical features to guide the
evolution of the ratio of housing prices to disposable income in the
scenario, which provided more predictability in the way that the stress
test would treat business lines affected by changes in house prices.
However, the Board believes that the design of scenarios could be made
more transparent and predictable by providing additional guides for
certain macroeconomic variables, and by disclosing additional detailed
information on the methodology used to create the global market shock
component of the severely adverse scenario, as described below.
a. Trading and Counterparty Components
For a subset of firms, the severely adverse scenario also includes
two additional components: the global market shock component and the
largest counterparty default component.\20\ The global market shock
component is a set of hypothetical shocks to a large set of risk
factors reflecting general market distress and heightened uncertainty.
A firm with significant trading activity must consider the global
market shock component as part of its severely adverse scenario and
recognize associated losses in the first quarter of the projection
horizon.\21\ The global
[[Page 51859]]
market shock component is applied to asset positions held by the firms
on a given as-of date.\22\ In addition, for certain large and highly
interconnected firms, the same global market shock component is applied
to counterparty exposures under the largest counterparty default
component.\23\ The largest counterparty default component is intended
to assess the potential losses and capital impact associated with the
default of the largest counterparty of each applicable firm, and the
as-of date aligns with that of the global market shock component.
---------------------------------------------------------------------------
\20\ See 12 CFR 238.143(b)(2)(i); 12 CFR 252.54(b)(2)(i). For
more information on the scenarios and components, see Board, 2025
Stress Test Scenarios (Feb. 2025), <a href="https://www.federalreserve.gov/publications/files/2025-stress-test-scenarios-20250205.pdf">https://www.federalreserve.gov/publications/files/2025-stress-test-scenarios-20250205.pdf</a>.
\21\ The global market shock component applies to firms subject
to Category I, II, and III standards that have aggregate trading
assets and liabilities of $50 billion or more, or trading assets and
liabilities equal to or greater than 10 percent of total
consolidated assets. See 12 CFR 238.143(b)(2)(i); 12 CFR
252.54(b)(2)(i).
\22\ Under the Board's current stress test rules, the global
market shock as-of date must occur between October 1 and March 1.
See 12 CFR 238.143(b)(2)(i); 12 CFR 252.14(b)(2)(i); 12 CFR
252.54(b)(2)(i).
\23\ The largest counterparty default component generally
applies to all firms subject to the global market shock component,
as well as firms with substantial processing and custodial
operations.
---------------------------------------------------------------------------
The design and specification of the global market shock component
differs from the design and specification of the severely adverse
scenario in several respects. First, in alignment with U.S. generally
accepted accounting principles (U.S. GAAP), profits and losses from
trading and counterparty credit positions are measured in mark-to-
market accounting terms in the global market shock, while revenues and
losses from traditional banking activities, as generated under
macroeconomic scenarios, are generally measured using the accrual
accounting method. Second, the timing of loss recognition differs
between the global market shock and the severely adverse macroeconomic
scenario. The global market shock affects the mark-to-market value of
trading positions and counterparty credit losses in the first quarter
of the severely adverse scenario. This timing is based on an
observation that market dislocations can happen rapidly and
unpredictably at any time under stressed conditions. In addition, the
severely adverse scenario is applied as of December 31 of each year
(the jump-off date), whereas the global market shock as-of date changes
every year (within the window specified in the Board's stress test
rules) and does not necessarily coincide with the year-end. This timing
is also based on a scenario assumption that market dislocations can
happen rapidly and unpredictably at any time during the scenario
horizon. Recognizing the global market shock in the first quarter helps
ensure that potential losses from trading and counterparty exposures
are incorporated into firms' capital ratios in each quarter of the
severely adverse scenario.
Models
Prior to 2019, the annual stress test results disclosure document
contained an appendix describing the Board's supervisory stress test
models.\24\ In 2019, the Board increased the transparency of the
supervisory stress test models by finalizing the Stress Testing Policy
Statement \25\ and the Enhanced Disclosure of the Models Used in the
Federal Reserve's Supervisory Stress Test (Enhanced Model
Disclosure).\26\ The Stress Testing Policy Statement describes the
Board's policies and procedures that guide the development,
implementation, and validation of the models.\27\ The Stress Testing
Policy Statement also describes the Board's principles for stress test
model design, namely that the system of models used in the supervisory
stress test should result in projections that are (1) independent of
firm projections; (2) forward-looking in that they project future
losses and revenue; (3) consistent and comparable across firms; (4)
generated from simple approaches, where appropriate; (5) robust and
stable; (6) conservative; and (7) able to capture the effect of severe
economic stress. The Board has developed stress test models in
accordance with these principles, which are the foundation for the
stress test modeling decisions described in the comprehensive
documentation of the supervisory stress test models that the Board is
publishing in conjunction with this proposal.
---------------------------------------------------------------------------
\24\ See, e.g., Board, 2018 Supervisory Stress Test Results
(Jun. 2018), <a href="https://www.federalreserve.gov/publications/files/2018-dfast-methodology-results-20180621.pdf">https://www.federalreserve.gov/publications/files/2018-dfast-methodology-results-20180621.pdf</a>.
\25\ See 84 FR 6664 (Feb. 28, 2019).
\26\ See 84 FR 6784 (Feb. 28, 2019).
\27\ See 12 CFR 252, Appendix B.
---------------------------------------------------------------------------
The Enhanced Model Disclosure supplemented prior public
descriptions of the stress test models by providing some information
about their structure and by including a list of key variables that
influence the results of each model.\28\ However, the Board believes
more detailed information, beyond what is in the current Enhanced Model
Disclosure, would improve the ability of firms to accurately assess how
changes in their business activities might impact their supervisory
stress test results and, relatedly, their stress capital buffer
requirements and overall capital requirements.
---------------------------------------------------------------------------
\28\ See, e.g., Board, 2025 Supervisory Stress Test Methodology
(Jun. 2025), <a href="https://www.federalreserve.gov/publications/files/2025-june-supervisory-stress-test-methodology.pdf">https://www.federalreserve.gov/publications/files/2025-june-supervisory-stress-test-methodology.pdf</a>.
---------------------------------------------------------------------------
C. Supervisory Stress Test Modeling Framework
The Board's stress test models take macroeconomic variables from
the Board's severely adverse scenario and firm data as inputs to
produce each firm's projected capital ratios over a nine-quarter
horizon. The projected common equity tier 1 capital ratio is used to
inform each firm's stress capital buffer requirement, which becomes
part of a firm's capital conservation buffer.
The stress test models are intended to capture how a firm's
regulatory capital would be affected by the macroeconomic and financial
conditions described in the stress test scenarios, given the
characteristics of the firm's business model and balance sheet
composition. The Board uses a variety of statistical modeling
techniques to produce the stress test results, including multivariate
regression, which uses relationships in historical data to produce
projections of a variable (such as a loss given default). These models
are represented by a set of formulas and coefficients that produce the
projections.
The Board estimates the effect of the severely adverse scenario on
the regulatory capital ratios of firms by projecting revenues,
expenses, and losses for each firm over a nine-quarter projection
horizon (projection horizon). The projection horizon spans nine
quarters to ensure that the firms can continue to provide credit and
serve as financial intermediaries despite several quarters of adverse
economic conditions, as well as to promote the forward-looking nature
of capital planning by firms.
Projected net income, adjusted for the effect of taxes, is combined
with assumptions regarding capital actions and other changes to
regulatory capital to produce post-stress capital ratios. The Board's
approach to modeling supervisory stress test results, including the
calculation of post-stress capital ratios, is generally in alignment
with U.S. GAAP and the regulatory capital framework.\29\ However, the
stress test models may deviate from U.S. GAAP and the regulatory
capital framework, as circumstances warrant.
---------------------------------------------------------------------------
\29\ See generally 12 CFR part 217.
---------------------------------------------------------------------------
The Board established the Stress Testing Policy Statement modeling
principles to ensure that the models are well suited for their purpose
in the regulatory framework. In some cases, the Board's adherence to
the principles limits modeling choices and results in certain common
limitations across similarly constructed component
[[Page 51860]]
models. For instance, consistent with the principles of independence,
consistency and comparability, and simplicity, models are not designed
to capture all firm-specific nuances, future strategic initiatives, or
planned capital actions. Additionally, models may be limited by their
reliance on historic relationships and by the nature of the data
captured in firms' regulatory reports. Detailed assumptions and
limitations for the models are discussed in the comprehensive
documentation, which is available at <a href="https://www.federalreserve.gov/supervisionreg/dfa-stress-tests-2026.htm">https://www.federalreserve.gov/supervisionreg/dfa-stress-tests-2026.htm</a>.
Under the Stress Testing Policy Statement, the Board's projections
also assume that a firm's balance sheet remains unchanged throughout
the projection horizon.\30\ This assumption seeks to help ensure that a
firm cannot ``shrink to health'' and that it remains sufficiently
capitalized to accommodate credit demand in a severe downturn.
---------------------------------------------------------------------------
\30\ See 12 CFR 252, Appendix B, section 2.7.
---------------------------------------------------------------------------
D. Stress Test Models
The Board's stress test models comprise twenty-one component models
that, when aggregated, produce projected regulatory capital ratios for
each firm (see Table 1 below). The models can be grouped into four
categories: credit risk, market risk, net revenue, and aggregation.
Credit risk models capture losses associated with retail and wholesale
loans that are held at amortized cost. Market risk models capture
losses associated with trading and counterparty exposures, securities,
and other assets held at fair value. Net revenue models capture income
and expenses, including those related to operational risk, earned or
incurred by a firm. Positive pre-provision net revenue offsets credit
and market risk losses in the calculation of a firm's pre-tax net
income. Aggregation models calculate a firm's pre-tax net income, which
is then adjusted for other elements such as taxes and regulatory
capital deductions to arrive at the projection of a firm's regulatory
capital, which is used to calculate a firm's projected capital ratios.
Additional detail about these component models is provided in Section
III.A of this Supplementary Information and the comprehensive model
documentation available at <a href="https://www.federalreserve.gov/supervisionreg/dfa-stress-tests-2026.htm">https://www.federalreserve.gov/supervisionreg/dfa-stress-tests-2026.htm</a>.\31\
---------------------------------------------------------------------------
\31\ See also Board, 2025 Supervisory Stress Test Methodology
(Jun. 2025), <a href="https://www.federalreserve.gov/publications/files/2025-june-supervisory-stress-test-methodology.pdf">https://www.federalreserve.gov/publications/files/2025-june-supervisory-stress-test-methodology.pdf</a>.
---------------------------------------------------------------------------
[[Page 51861]]
[GRAPHIC] [TIFF OMITTED] TP18NO25.035
[[Page 51862]]
[GRAPHIC] [TIFF OMITTED] TP18NO25.036
E. Summary of the Proposal
---------------------------------------------------------------------------
\32\ The Trading Issuer Default Loss Model, Trading Profit and
Loss Model, Credit Valuation Adjustment Model, and Largest
Counterparty Default Model apply only to a subset of firms. See
Section II.B of this Supplementary Information.
---------------------------------------------------------------------------
The Board is publishing comprehensive documentation on the stress
test models on the Board's website, at <a href="https://www.federalreserve.gov/supervisionreg/dfa-stress-tests-2026.htm">https://www.federalreserve.gov/supervisionreg/dfa-stress-tests-2026.htm</a>. This model documentation
contains information on the models that together produce the results of
the supervisory stress test. The model documentation includes the
equations, variables, and coefficients used in each model (where
applicable); assumptions and limitations of each model; rationales for
modeling decisions; and discussions of alternative models. Section
VIII.A of this Supplementary Information summarizes changes to the
models, relative to the 2025 stress test, that the Board plans to
implement in the 2026 stress test cycle; section VIII.B of this
Supplementary Information contains an analysis of the potential effects
of these proposed model changes. Detailed documentation on these
changes is also provided on the Board's website, at <a href="https://www.federalreserve.gov/supervisionreg/dfa-stress-tests-2026.htm">https://www.federalreserve.gov/supervisionreg/dfa-stress-tests-2026.htm</a>. As
part of this proposal, the Board is inviting public comment on the
stress test models and these changes.
In addition, the Board is proposing to codify an enhanced
disclosure process that would build on the previous efforts that the
Board has made to increase the transparency and public accountability
of the supervisory stress test. Under this enhanced disclosure process,
the Board would annually publish comprehensive model documentation on
the stress test models, invite public comment on any material changes
that the Board seeks to make to those models, and annually publish the
stress test scenarios for comment. The Board would also commit to
responding to substantive public comments on any material model changes
before implementing them. The proposal would revise the Stress Testing
Policy Statement to align with this enhanced disclosure process, as
well as to amend the Board's general policy related to disclosing
additional information directly to a firm about that firm's supervisory
stress test results. To accommodate the annual comment process on the
scenarios, the proposal would shift the jump-off date of the
supervisory and company-run stress tests from December 31 to September
30.
Additionally, this proposal would amend the Scenario Design Policy
Statement in several ways. The Board would include in the Scenario
Design Policy Statement detailed descriptions of additional guides that
are used to inform the Board's choice of the values of the scenario
variables along their scenario paths. The guides are designed to
balance the competing objectives of predictability and transparency, on
the one hand, with the severity and relevance of the macroeconomic and
financial market scenarios, on the other hand. Most of the proposed
guides also incorporate features similar to the range of options in the
existing unemployment guide or the automatic adjustment of the house
price path to current housing market conditions in the existing house
price guide. This approach would allow the Board to continue to adjust
the severity of those variables as necessary to avoid inducing greater
procyclicality in the financial system and macroeconomy.
Similarly, the Board is proposing to incorporate additional
information into the Scenario Design Policy Statement about the
framework used to create the global market shock component of the
severely adverse scenario. This information includes, but is not
limited to, details on the logic underlying the severity of the shocks
and a description of the processes used to generate the shock values.
The Board is also proposing to update the global market shock
methodology to simplify the scenario and better align certain elements
of the global market shock with the nature of an ``instantaneous''
shock. The proposal would also make revisions to the stress test rules
to improve the risk capture of the supervisory stress test by widening
the
[[Page 51863]]
as-of date window for the global market shock.
Finally, the proposal would make changes to the FR Y-14A/Q/M
reports to remove items and documentation requirements that are no
longer needed to conduct the supervisory stress test, as well as to
collect additional data to improve risk capture.
F. Purpose of the Proposal
The purpose of this proposal is to provide the public with more
information about the stress test models and scenarios and to help
ensure that the public has an opportunity to comment on the models and
scenarios. While the Board has increased the transparency of the stress
test models over time, disclosing additional information about the
stress test models and their underlying methodologies will further
increase transparency and improve public accountability.
Publishing detailed descriptions of the stress test models for
comment, as well as committing to future enhanced disclosures, has
benefits. First, the increase in transparency would increase public
accountability and instill confidence in the fairness of the
supervisory stress tests. Second, the disclosure process would create a
new mechanism for obtaining feedback from the public, including
academics, financial analysts, and firms, on the design and
specifications of the models, which should lead to model improvements.
Third, a firm would have a better sense of how its risk profile would
factor into its stress capital buffer requirement, which would reduce
the likelihood of unanticipated stress test results and allow for
better capital and business planning by firms. Finally, the public
disclosure of additional information about supervisory stress tests
should strengthen market discipline, because investors, counterparties,
and rating agencies would be able to better assess a firm's risk
profile.\33\ The costs and benefits of this proposal are described more
thoroughly in Section X of this Supplementary Information.
---------------------------------------------------------------------------
\33\ See, e.g., N. Gambetta, M.A. Garc[iacute]a-Benau, & A.
Zorio-Grima, Stress test impact and bank risk profile: Evidence from
macro stress testing in Europe, 61 Intl. Rev. of Econ. & Fin 347-54
(2019); I. Goldstein & Y. Leitner, ``Stress test disclosure: theory,
practice, and new perspectives,'' Handbook of Financial Stress
Testing 208-223 (2022).
---------------------------------------------------------------------------
With respect to the proposed amendments to the Scenario Design
Policy Statement, this proposal also builds on the contents of the
current Scenario Design Policy Statement and would amend it to provide
additional transparency, public accountability, and predictability in
the variable paths. The changes would support the Board in developing
scenarios, inviting comment on those scenarios, incorporating input
from commenters, and maintaining the current schedule for release of
the final scenarios. Despite the increased predictability in the
scenarios, the new framework would remain flexible enough to suitably
assess whether firms can maintain an adequate amount of loss-absorbing
capital to stay above minimum regulatory requirements and continue
financial intermediation during periods of stress, as well as adjust
features that might add to existing procyclicality in the financial
system, as appropriate. In practice, the scenarios resulting from the
revised framework are expected to remain consistent with the current
Scenario Design Policy Statement and should not result, on average over
a typical business cycle, in materially different scenarios than would
have been designed previously.
Additionally, the proposal would simplify the design of the global
market shock component and incorporate additional information on the
development process into the Scenario Design Policy Statement, which
outlines the Board's approaches to designing market shocks, including
important considerations for scenario design, possible approaches to
developing scenarios, and a development strategy for implementing the
preferred approach. Taken together, these changes would improve
transparency, public accountability, and predictability of the
supervisory scenarios, while ensuring the supervisory stress test's
ability to capture changes in the risks in the financial industry over
time.
III. Overview of the Stress Test Modeling Framework
As summarized in Section II.D of this Supplementary Information,
the Board estimates the effect of the scenarios on the regulatory
capital ratios of firms participating in the stress test by projecting
net income and other components of regulatory capital for each firm
over a nine-quarter projection horizon. To do so, the Board uses
twenty-one component models, the macroeconomic variables from the
Board's severely adverse scenario, and firm data. This section provides
an overview of the component models the Board used to run the 2025
supervisory stress test. See Table 1 in Section II.D of this
Supplementary Information.
A. Supervisory Stress Test Models
The Board calculates projected pre-tax net income by combining
projections of pre-provision net revenue,\34\ provisions for credit
losses,\35\ and other gains or losses.\36\ Each component of pre-tax
net income is described below.
---------------------------------------------------------------------------
\34\ Pre-provision net revenue includes, among other items,
income from mortgage servicing rights, losses from operational risk
events, and other real estate owned costs.
\35\ For firms that have adopted Accounting Standards Update
(ASU) 2016-13, the Federal Reserve incorporates its projection of
expected credit losses on securities in the allowance for credit
losses, in accordance with Financial Accounting Standards Board
(FASB), Financial Instruments--Credit Losses (Topic 326). See FASB
ASU No. 2016-13, ``Financial Instruments--Credit Losses (Topic 326):
Measurement of Credit Losses on Financial Instruments.''
\36\ Other gains or losses include losses on held-for-sale
loans, loans measured under the fair-value option, and loan hedges.
---------------------------------------------------------------------------
Pre-Provision Net Revenue
Pre-provision net revenue is defined as net interest income
(interest income minus interest expense) plus noninterest income minus
noninterest expense. Consistent with U.S. GAAP, these projections
include projected losses due to operational risk events and expenses
related to the disposition of other real estate owned.\37\ The Board
projects most components of pre-provision net revenue using models that
relate specific revenue and non-provision-related expenses to the
characteristics of firms and to macroeconomic variables. These include
eight components of interest income, seven components of interest
expense, six components of noninterest income, and three components of
noninterest expense. The Board separately projects losses from
operational risk and other real estate owned expenses. Operational risk
is defined as ``the risk of loss resulting from inadequate or failed
internal processes, people and systems or from external events.'' \38\
Other real estate owned expenses are expenses related to the
disposition of real estate owned properties and stem from losses on
first-lien mortgages.
---------------------------------------------------------------------------
\37\ However, pre-provision net revenue projections do not
include debt valuation adjustments, which are not included in
regulatory capital.
\38\ 12 CFR 217.101 ``Operational risk.''
---------------------------------------------------------------------------
Loan Losses and Provisions on Loans Measured at Amortized Cost
The Board typically projects losses using one of two modeling
approaches: the expected-loss approach or the net charge-off approach.
Generally, under the expected loss approach, expected losses are
estimated by projecting the probability of default, loss given default,
and exposure at default for each quarter of the projection horizon.
Expected losses in each quarter are the product of these three
components. Under the net
[[Page 51864]]
charge-off approach, losses are projected using historical behavior of
net charge-offs as a function of macroeconomic and financial market
conditions and loan portfolio characteristics.\39\
---------------------------------------------------------------------------
\39\ Entire loans or portions of loans may be charged off if a
firm believes that the loan will not be repaid. If an amount that is
charged off is ultimately repaid by the borrower, then that repaid
amount is added to a firm's income as a recovery. Net charge-offs
are total charge-offs less any recoveries.
---------------------------------------------------------------------------
The Board estimates losses for loans measured at amortized cost
separately for different categories of loans, based on the type of
obligor, collateral, and loan structure. The individual loan types
modeled can broadly be divided into (1) retail loans, including various
types of residential mortgages, credit cards, student loans, auto
loans, small business loans, and other consumer loans; and (2)
wholesale loans, such as commercial and industrial loans and commercial
real estate loans. For most loan types, losses in quarter t are
estimated as the product of the projected probability of default in
quarter t, the loss given default in quarter t, and exposure at default
in quarter t.
The probability of default component measures the likelihood that a
borrower enters default status during a given quarter t. The other two
components capture the lender's net loss on the loan if the borrower
enters default. The loss given default component measures the
percentage of the loan balance that the lender will not be able to
recover after the borrower enters default, and the exposure at default
component measures the total expected outstanding loan balance at the
time of default.\40\
---------------------------------------------------------------------------
\40\ When applicable, loan loss models may factor in shared-loss
agreements with the Federal Deposit Insurance Corporation.
---------------------------------------------------------------------------
The Board's definition of default, for stress test modeling
purposes, may vary for different types of loans and may differ from
general industry definitions or classifications. The Board generally
models probability of default as a function of loan characteristics and
economic conditions. The Board typically models loss given default
based on historical data, and modeling approaches vary for different
types of loans. For certain loan types, the Board models loss given
default as a function of borrower, collateral, or loan characteristics
and the macroeconomic variables from the supervisory scenarios. For
other loan types, the Board assumes loss given default is a fixed
percentage of the loan balance for all loans in a category. The
approach to modeling exposure at default also varies by loan type and
depends on whether the loan is a term loan or a line of credit.
For certain retail loan categories, projections capture the
historical behavior of net charge-offs as a function of macroeconomic
and financial market conditions and loan portfolio characteristics. The
Board then uses these stress test models to project future charge-offs
consistent with the evolution of macroeconomic conditions under the
severely adverse scenario. To project losses, the projected net charge-
off rate is applied to projected loan balances.
Losses on loans are then projected to flow into net income through
provisions for loan and lease losses (for simplicity, provisions for
loan losses). Provisions for loan losses reflect funds set aside to
cover loan losses that a firm expects to incur in a predetermined
future window. Provisions for loan losses feed into the allowance for
loan losses, which serves as a contra asset on a firm's balance sheet.
The charged-off amount of a loan reduces the outstanding balance of the
loan while also reducing the allowance for loan losses (that is,
charge-offs do not reduce a firm's total assets). Generally, provisions
for loan losses for each projected quarter in the supervisory stress
test equal projected losses on loans for the quarter plus the change in
the allowance for loan losses needed to cover the subsequent four
quarters of expected loan losses. This calculation incorporates the
allowance for loan losses established by the firm as of the jump-off
date of the stress test exercise.
Current Expected Credit Losses Framework
On January 1, 2020, most large and mid-sized U.S. banks adopted the
Current Expected Credit Losses (CECL) standard for calculating
allowances.\41\ CECL superseded the incurred loss accounting standard,
which was a backward-looking measure that enabled firms to calculate
allowances based on historical loss data and current economic
conditions. CECL, by contrast, is a forward-looking measure that
requires firms to estimate lifetime losses based on reasonable
estimates of future economic conditions. In October 2024, the Board
announced that it would continue to evaluate future enhancements to the
supervisory stress test approach for the incorporation of CECL.\42\
---------------------------------------------------------------------------
\41\ See FASB ASU No. 2016-13, ``Financial Instruments--Credit
Losses (Topic 326): Measurement of Credit Losses on Financial
Instruments.''
\42\ See Q(DST0030) (Oct. 9, 2024) and Q(DST0029) (Dec. 15,
2023), <a href="https://www.federalreserve.gov/publications/ccar-qas/comprehensive-capital-analysis-and-review-questions-and-answers.htm">https://www.federalreserve.gov/publications/ccar-qas/comprehensive-capital-analysis-and-review-questions-and-answers.htm</a>.
---------------------------------------------------------------------------
The Board is not proposing to implement CECL into the supervisory
stress testing framework as a part of this proposal. The allowance
calculation framework currently used in the supervisory stress test is
already forward-looking: it projects loan loss provisions four quarters
ahead. This approach aligns with the Board's modeling principle of
simplicity as it requires fewer assumptions than would be required to
determine provisions under CECL. In addition, in aggregate, the
cumulative loan loss provisions under the supervisory severely adverse
scenario are similar to provision projections submitted by the firms
that have adopted CECL. Should the Board decide to implement CECL into
the supervisory stress testing framework, it would seek public comment
prior to implementation, as it would likely be a material model change
as defined in this proposal.
Question 2: What factors should the Board consider when determining
whether to implement CECL into the supervisory stress testing framework
and why?
Question 3: What would be the advantages and disadvantages of
incorporating CECL into the supervisory stress testing framework?
Losses on Loans Measured on a Fair Value Basis
Certain loans are accounted for on a fair value basis instead of on
an amortized cost basis. If a loan is accounted for using the fair
value option, it is marked to market, and the accounting value of the
loan changes as a function of changes in market risk factors and
fundamentals. Similarly, loans that are held for sale are accounted for
at the lower of cost or market value. The stress test models for these
asset classes project gains and losses over the nine-quarter projection
horizon, net of any hedges, using the scenario-specific path of
interest rates and credit spreads. The Board uses different models to
estimate gains and losses on wholesale loans and retail loans that are
accounted for on a fair value basis since these loans have different
risk characteristics. However, these models all generally project gains
and losses over the nine-quarter projection horizon, net of hedges, by
applying the scenario-specific interest rate and credit spread shocks
to loan yields.
Losses on Securities
A firm's balance sheet typically contains holdings of two types of
securities related to investment activities: available-for-sale and
held-to-
[[Page 51865]]
maturity. Available-for-sale and held-to-maturity securities are
generally held at fair value and amortized cost, respectively, on a
firm's balance sheet. The Board estimates two types of losses on
securities related to investment activities.\43\
---------------------------------------------------------------------------
\43\ This portfolio does not include securities held for
trading. Losses on these securities are projected by the Trading
Profit and Loss Model that projects gains and losses on trading
exposures.
---------------------------------------------------------------------------
For debt securities classified as available-for-sale, projected
fluctuations in the fair value of the securities due to changes in
interest rates and other factors will result in unrealized gains or
losses that are recognized in capital for some firms through other
comprehensive income. Under U.S. GAAP, unrealized gains and losses on
available-for-sale debt securities are reflected in accumulated other
comprehensive income and do not flow through net income.\44\ Under the
regulatory capital rule, accumulated other comprehensive income must be
incorporated into common equity tier 1 capital for certain firms.
Unrealized gains and losses are calculated as the difference between
each security's fair value and its amortized cost. The amortized cost
of each available-for-sale debt security is equivalent to the purchase
price of the debt security, which is periodically adjusted if the debt
security was purchased at a price other than par or face value, has a
principal repayment, or has an impairment recognized in earnings.\45\
---------------------------------------------------------------------------
\44\ Unrealized gains and losses on equity securities are
recognized in net income and affect regulatory capital for all
firms. See FASB ASU No. 2016-01, ``Financial Instruments--Overall
(Subtopic 825-10): Recognition and Measurement of Financial Assets
and Financial Liabilities.''
\45\ The fair value of each available-for-sale security is
projected over the nine quarter projection horizon using either a
present value calculation, a full revaluation using a security-
specific discounted cash flow model, or a duration-based approach,
depending on the asset class.
---------------------------------------------------------------------------
Credit losses on available-for-sale and held-to-maturity securities
may be also recorded. Except for certain government-backed obligations,
both available-for-sale and held-to-maturity securities are at risk of
incurring credit losses.\46\ The stress test models project security-
level credit losses, using as an input the projected fair value for
each security over the nine-quarter projection horizon under the
severely adverse scenario. Credit losses on securities are included in
the projection of provisions.
---------------------------------------------------------------------------
\46\ Certain government-backed securities, such as U.S.
Treasuries, U.S. government agency obligations, U.S. government
agency or government-sponsored enterprise mortgage-backed
securities, federally backed student loan asset-backed securities,
and pre-refunded municipal bonds, are assumed not to be subject to
credit losses.
---------------------------------------------------------------------------
Projected other comprehensive income gains or losses from
available-for-sale debt securities are computed directly from the
projected change in fair value, taking into account credit losses and
applicable interest-rate hedges on securities. All debt securities held
in the available-for-sale portfolio are subject to other comprehensive
income losses.
Losses on Private Equity Exposures
The Board projects the value of private equity investments in
response to the severely adverse scenario of the supervisory stress
test.\47\ The Private Equity Model assigns losses and recoveries based
on changes in fair value, recognized in net income for all positions,
regardless of their individual accounting elections. While U.S. GAAP
allows for private equity to be carried under a variety of accounting
measures, the different accounting methods are generally not reflective
of fundamental risk differences--fair value is typically realized upon
the orderly sale of a given private equity investment, irrespective of
its accounting treatment during the holding period.\48\
---------------------------------------------------------------------------
\47\ The Board projects private equity losses only for firms
that are required to submit FR Y-14Q, Schedule F.24 (Private Equity)
because private equity exposures are reported on that schedule.
Currently, Schedule F.24 is required to be reported by firms subject
to Category I through III standards that have, on average, aggregate
trading assets and liabilities of $50 billion or more, or aggregate
trading assets and liabilities equal to 10 percent or more of total
consolidated assets. As discussed in Section XI.A of this
Supplementary Information, the Board is proposing to modify the
threshold for Schedule F.24 to align with other banking book
schedules.
\48\ Unlike a bond or loan, private equity investments generally
cannot be redeemed by holding to maturity and are therefore
fundamentally exposed to market risk at exit.
---------------------------------------------------------------------------
Losses on Trading Exposures
The trading stress test models cover a wide range of a firm's
exposure to asset classes such as public equity, foreign exchange,
interest rates, commodities, securitized products, traded credit (for
example, municipal securities, auction rate securities, corporate
credit, and sovereign credit), and other fair-value assets. Loss
projections are constructed by applying the market risk factor
movements specified in the global market shock component \49\ to market
values of firm-provided positions and risk factor sensitivities.\50\
The global market shock only applies to a subset of firms, as described
in Section II.B.a of this Supplementary Information. In addition, the
global market shock component is applied to firm counterparty exposures
to generate losses due to changes in credit valuation adjustment, which
is a change to the market value of an exposure (for example, a
derivative) to account for the risk that the counterparty defaults on
its obligation. Trading and credit valuation adjustment losses are
calculated only for firms subject to the global market shock component.
In contrast to the nine-quarter evolution of losses for other parts of
the supervisory stress test, and as previously described, these losses
are estimated and applied in the first quarter of the projection
horizon. This timing is based on the observation that market
dislocations can happen rapidly and unpredictably any time under stress
conditions. It also ensures that potential losses from trading and
counterparty exposures are incorporated into a firm's capital ratio at
all points in the projection horizon.
---------------------------------------------------------------------------
\49\ See Section II.B.a of this Supplementary Information.
\50\ The supervisory trading models are also used to calculate
gains or losses on firms' portfolios of hedges on credit valuation
adjustment exposures.
---------------------------------------------------------------------------
The Board separately estimates the risk of losses arising from the
default of issuers of debt securities held for trading. These losses
account for concentration risk in corporate, sovereign, agency, and
municipal credit positions. In contrast to the trading losses described
above, these losses are applied in each of the nine quarters of the
projection horizon to capture the risk that several quarters of
stressful economic conditions may cause additional issuers of debt
securities to default, which aligns with the Board's principle of
conservatism from the Stress Testing Policy Statement.
Largest Counterparty Default Losses
The largest counterparty default component is applied to firms with
substantial trading or custodial operations. This component captures
the risk of loss due to the unexpected default of the counterparty
whose default on derivatives and securities financing transactions,
with exposures revalued by applying the global market shock component,
would generate the largest stressed losses for a firm. Consistent with
the Board's modeling principles and with the losses associated with the
global market shock component, losses associated with the largest
counterparty default component are recognized in the first quarter of
the projection horizon.
Balance Projections and the Calculation of Regulatory Capital Ratios
As described above, the Board assumes that a firm takes actions to
maintain its current level of assets, including its investment
securities, trading assets, and loans, over the
[[Page 51866]]
projection horizon. The Board also assumes that a firm's risk-weighted
assets and leverage ratio denominators remain unchanged over the
projection horizon, except that the Board will account for changes
primarily related to the calculation of regulatory capital or due to
changes to the Board's regulations.\51\
---------------------------------------------------------------------------
\51\ See 12 CFR 252, Appendix B, section 3.4; Board, Press
Release (Mar. 4, 2020), <a href="https://www.federalreserve.gov/newsevents/pressreleases/bcreg20200304a.htm">https://www.federalreserve.gov/newsevents/pressreleases/bcreg20200304a.htm</a>.
---------------------------------------------------------------------------
The Board includes five regulatory capital ratios in the
supervisory stress test: (1) common equity tier 1 risk-based capital,
(2) tier 1 risk-based capital, (3) total risk-based capital, (4) tier 1
leverage, and (5) supplementary leverage. A firm's post-stress
regulatory capital ratios are projected in accordance with the Board's
regulatory capital rule using the Board's projections of pre-tax net
income and other scenario-dependent components of the regulatory
capital ratios. Pre-tax net income and the other scenario-dependent
components of the regulatory capital ratios are combined with
additional information, including assumptions about taxes and capital
distributions, to project post-stress measures of regulatory capital.
In those calculations, the Board adjusts pre-tax net income to account
for taxes and other components of net income, such as income
attributable to minority interests, to arrive at after-tax net income.
The Board calculates the change in equity capital over the projection
horizon by combining projected after-tax net income with changes in
other comprehensive income, assumed capital distributions, and other
components of equity capital. The path of regulatory capital measures
over the projection horizon is calculated by combining the projected
change in equity capital with the firm's starting capital position and
accounting for other adjustments to regulatory capital specified in the
Board's regulatory capital framework.\52\ The denominator of each
firm's risk-based capital ratios is based on a firm's standardized
approach for calculating risk-weighted assets on the jump-off date of
the supervisory stress test, and may change for each quarter of the
projection horizon to account for adjustments specified in the capital
rule (for example, adjustments due to the thresholds for deducting
certain deferred tax assets).
---------------------------------------------------------------------------
\52\ The regulatory capital framework specifies that regulatory
capital ratios account for items subject to adjustment or deduction
in regulatory capital, limits the recognition of certain assets that
are less loss-absorbing, and imposes other restrictions. See
generally 12 CFR part 217.
---------------------------------------------------------------------------
B. Supervisory Stress Test Scenarios
The Board conducts the supervisory stress test using two
scenarios--the baseline and severely adverse. The severely adverse
scenario describes a hypothetical set of conditions designed to assess
the strength and resilience of firms in a severely adverse economic
environment and includes 28 variables that are disclosed by the Board
each year prior to the supervisory stress test. Some variables describe
economic developments within the United States while others describe
developments in foreign countries.\53\ These variables serve as an
input to the calculation of supervisory stress test results for all
firms. As discussed above, for a subset of firms, the severely adverse
scenario also includes two additional components: the global market
shock component and the largest counterparty default component. The
scenarios and associated components are developed solely for
supervisory stress testing purposes and do not represent economic
forecasts of the Board.
---------------------------------------------------------------------------
\53\ For a description of the macroeconomic variables applicable
to the 2025 supervisory stress test, see Board, 2025 Stress Test
Scenarios (Feb. 2025), <a href="https://www.federalreserve.gov/publications/files/2025-stress-test-scenarios-20250205.pdf">https://www.federalreserve.gov/publications/files/2025-stress-test-scenarios-20250205.pdf</a>.
---------------------------------------------------------------------------
Geographic Variation of Macroeconomic Variables
While the Board projects the paths of macroeconomic variables at
the national level, the Board uses regional-level (that is, state- and/
or county-level) macroeconomic variables in the stress test models to
project losses on certain loans held for investment at amortized
cost.\54\ In general, model outputs are demonstrably impacted by the
macroeconomic environment, as both probability of default and loss
given default increase during periods of economic stress. Importantly,
the macroeconomic environment can also vary notably across geography,
in addition to across time. For instance, during the 2007-2009 crisis
period, housing prices fell more sharply in certain geographies
compared to others. Accordingly, historical loss rates in many loan
categories were higher during this period in geographies where housing
prices fell more sharply.
---------------------------------------------------------------------------
\54\ Specifically, the Board uses regional-level macroeconomic
variables in the First Lien Model, the Home Equity Model, the Credit
Card Model, the Auto Model, and the Commercial Real Estate Model.
---------------------------------------------------------------------------
Therefore, to account for the impacts of different macroeconomic
environments across geographies on historical loan performance, the
Board calibrates model parameters in certain stress test models using
regional macroeconomic variables as opposed to national macroeconomic
variables. For example, the unemployment rate used in an applicable
model may be the state level unemployment rate, while the house price
index values used in the model may be the county-level house price
indices or, in the case of loans in counties where a house price index
is not projected, a state-level house price index.\55\ Analysis
performed by the Board demonstrates that a certain model's statistical
fit and sensitivity to the macroeconomic environment may perform better
when using regional-level variables compared to when using only
national-level variables. The use of regional-level variables is
described in each applicable model section of the comprehensive model
documentation.
---------------------------------------------------------------------------
\55\ Certain variables do not vary based on geography. For
example, interest rates are typically set by national and not
regional markets. For these variables, the Board uses the national-
level paths in the models.
---------------------------------------------------------------------------
However, because the severely adverse scenario only includes
national-level variable paths, the Board derives the paths of regional-
level variables from the paths of national-level variables. The Board
employs a simple approach to calculating the paths of regional-level
variables in that these variables have the same percentage change (in
the case of an index variable) or level change (in the case of non-
index variables) as the national-level variables, but the starting
points are the regional-level values, not the national-level values.
For example, the projected path of the house price index is assumed to
have the same percentage change in a given quarter as the percentage
change of the national house price index,\56\ and the projected path of
unemployment rate is assumed to have the same level change in a given
quarter as the level change of the national unemployment rate.\57\ The
use of percentage changes for home price indices and level changes for
unemployment rates avoids accentuating differences in the macroeconomic
environment observed immediately prior to the beginning of the
scenario, which could lead to large discrepancies in projected variable
paths across geographies during the severely adverse scenario.
---------------------------------------------------------------------------
\56\ The house price index used in the supervisory stress test
scenarios is set to be equal to 100 in January 2000. This choice of
index month is arbitrary and does not reflect any underlying
economic importance of this period.
\57\ For example, if the national unemployment rate increases by
0.5 percentage points in a given quarter, the state-level
unemployment rate would be projected to increase by 0.5 percentage
points in that quarter as well.
---------------------------------------------------------------------------
These simple, uniform policies for allocating changes to the
national
[[Page 51867]]
macroeconomic environment at the regional level ensure that loans to
borrowers in certain geographies are not unduly favored or penalized.
While it is plausible that certain geographies may experience more
volatility than others in terms of the macroeconomic environment, the
Board does not estimate such volatility to differentiate scenarios
across geography, to avoid making assumptions about the severity of a
hypothetical recession across different regions.
The Board also uses historical regional data to produce model
projections. While the regional scenarios are projected based on the
national path, the Board retains variation in the historical regional
macroeconomic variables.\58\ The Board may also use historical regional
macroeconomic variables in the models to calculate the appreciation in
house prices since origination (which may be needed to calculate loan-
to-value ratios), or the Board may use regional macroeconomic variables
to calculate year-over-year changes in the variables. Alternatively,
the Board could replace all historical values with their national
equivalent when projecting losses, thus applying a truly uniform
treatment across geographies. While this alternative would have the
benefit of maximizing geographic consistency, it would ignore
meaningful variation in the historical environment and thereby reduce
the predictive power of the model. For instance, if a given geography
has had higher house price appreciation since its origination date
compared to the national average, without incorporating these
historical values into the macroeconomic data used to project losses
the model would understate the level of equity the borrower has as of
the beginning of the projection period. The Board has therefore
developed this hybrid approach to estimating losses in the supervisory
stress test, in which it applies a uniform treatment to projected
values of macroeconomic variables across geographies, while also
retaining historical differences across geographies. This methodology
allows for the incorporation of all available historical data needed to
produce accurate projections, while avoiding the need to make
assumptions about which geographies will have more or less severe
macroeconomic paths during a hypothetical recession. Further discussion
of how the Board's models account for geographic variation in
variables, including a proposed change to the Board's modeling
approach, is included in the comprehensive model documentation,
available at <a href="https://www.federalreserve.gov/supervisionreg/dfa-stress-tests-2026.htm">https://www.federalreserve.gov/supervisionreg/dfa-stress-tests-2026.htm</a>.
---------------------------------------------------------------------------
\58\ The historical regional unemployment rate and house price
index data are seasonally adjusted using the X11 procedure when a
seasonally adjusted version of these series is not available from
the source data. Seasonal adjustment is applied for consistency and
comparability with the published national scenario variables. For
more information about the X11 procedure developed by the U.S.
Census Bureau, see Shiskin J., Young A., and Musgrave, J., 1967. The
X-11 Variant of the Census Method II Seasonal Adjustment Program.
U.S. Department of Commerce, Bureau of the Census.
---------------------------------------------------------------------------
Question 4: What are the advantages and disadvantages of the
Board's treatment of regional (i.e., state and county) macroeconomic
variables in the credit risk models?
Question 5: What alternatives should the Board consider to the
approach outlined above for defining state and county macroeconomic
variables based on the national variables included in the scenarios?
What would be the advantages and disadvantages of these alternatives?
Auxiliary Variables
In addition to the 28 variables that the Board discloses each year,
the Board also generates paths for a limited number of other variables
that are used in the supervisory stress test. These variables, known as
auxiliary variables, are not disclosed by the Board because their paths
are based on the paths of the 28 disclosed variables (that is, the
paths are contingent upon movements in the 28 disclosed variables). For
example, the path of Mexico's gross domestic product (GDP) growth rate
is a function of the GDP growth rate paths of other country blocs that
are disclosed. Some models use these auxiliary variables, as described
in the applicable model sections of the comprehensive model
documentation available at <a href="https://www.federalreserve.gov/supervisionreg/dfa-stress-tests-2026.htm">https://www.federalreserve.gov/supervisionreg/dfa-stress-tests-2026.htm</a>.\59\
---------------------------------------------------------------------------
\59\ Detailed descriptions of the process for creating the paths
of auxiliary variables are included in the applicable model
documentation.
---------------------------------------------------------------------------
C. Data Used in Stress Testing
Input Data
The Board generally develops and implements the models with data it
collects on regulatory reports as well as proprietary third-party
industry data. Most of the data used in the supervisory stress test
projections are collected through the Capital Assessments and Stress
Testing regulatory report (FR Y-14), which includes a set of annual (FR
Y-14A), quarterly (FRY-14Q), and monthly (FRY-14M) schedules.\60\
---------------------------------------------------------------------------
\60\ The FR Y-14 report forms and instructions are available on
the Board's website at <a href="https://www.federalreserve.gov/apps/reportforms/default.aspx">https://www.federalreserve.gov/apps/reportforms/default.aspx</a>.
---------------------------------------------------------------------------
A firm must submit detailed loan and securities information for all
material portfolios on the FR Y-14Q and FR Y-14M. The definition of a
material portfolio for purposes of FR Y-14 reporting is based on a
firm's size and complexity.\61\ Portfolio categories are defined in the
FR Y-14M and FR Y-14Q reporting instructions. Each firm has the option
to submit the relevant data schedule for a given portfolio that does
not meet the materiality threshold as defined in the instructions. If a
firm does not submit data on its immaterial portfolio(s), the Board
will assign to that portfolio the median loss rate estimated across the
set of firms with material portfolios. This loss assumption adheres to
the principle of simplicity, as well as the principle of consistency
and comparability, from the Stress Testing Policy Statement.
---------------------------------------------------------------------------
\61\ Specifically, the definition of a material portfolio varies
depending upon a firm's categorization in the risk-based category
framework adopted by the Board for determining prudential standards.
See 12 CFR 238.10; 12 CFR 252.5.
---------------------------------------------------------------------------
While each firm is responsible for ensuring the completeness and
accuracy of data provided in the FR Y-14 reports, the Board makes
efforts to validate firm-reported data and requests resubmissions of
data where errors are identified. If data quality remains deficient
after resubmission, the Board applies conservative assumptions to a
particular portfolio or to specific data, depending on the severity of
deficiencies. For example, if the Board deems the quality of a firm's
submitted data too deficient to produce a stress test model estimate
for a particular portfolio, then the Board assigns a high loss rate
(for example, 90th percentile) or a conservative pre-provision net
revenue rate (for example, 10th percentile) to the portfolio balances
based on supervisory stress test projections of portfolio losses or
pre-provision net revenue for other firms.\62\ If data that are direct
inputs to stress test models are missing or reported erroneously but
the problem is isolated in such a way that the existing supervisory
framework can still be used, the Board assigns a conservative value
(for example, 10th or 90th percentile) to the specific data based on
all available data reported by firms. These assumptions are consistent
with the Board's principle of conservatism and policies on the
treatment of immaterial portfolios and missing or erroneous
[[Page 51868]]
data, as described in the Stress Testing Policy Statement.
---------------------------------------------------------------------------
\62\ Prior to assigning a conservate loss or revenue rate to
produce a firm's stress test results, the Board consults with a firm
that submits deficient data in order to determine whether the
applicable data issue can be remedied.
---------------------------------------------------------------------------
Additionally, certain stress test model projections rely on data
from the Consolidated Financial Statements for Holding Companies
regulatory report (FR Y-9C), which contains consolidated income
statement and balance sheet information for each firm subject to the
stress test. The FR Y-9C also includes off-balance sheet items and
other supporting schedules, such as the components of risk-weighted
assets and regulatory capital, that may be used in the stress test
models.
In limited circumstances, the Board also uses data provided by
third parties in the development and execution of the supervisory
stress test. The comprehensive model documentation identifies these
instances. The scenario data discussed above is also an input into the
stress test projections.
Data Preparation and Adjustments
a. Data Preparation
The data inputs the Board uses may not be initially suitable for
use in the stress test models. In these cases, the Board takes several
steps to prepare the data for use in the stress test models. The
specific steps for each model are discussed in the applicable model
descriptions within the comprehensive model documentation, though
generally data are prepared for use in the models for two purposes: to
remove outliers from the sample and to seasonally adjust the data.
These adjustments help ensure that the model results are reasonable.
The Board may remove outliers or data that are not applicable to
the model from the sample to facilitate more usable results. For
example, if a commercial real estate loan has a unusually high loan-to-
value (LTV) ratio (over 150 percent at origination), then data for that
loan are not included in the Commercial Real Estate Model because its
inclusion may produce unreliable results. Additionally, if first lien
mortgages are insured by the Federal Housing Administration or
Department of Veterans Affairs, then they are excluded from the First
Lien Model because these loans would not generate losses in the
supervisory stress test, as they are assumed to be fully insured by the
U.S. government. In both examples, the model output is more sensible
and more reflective of a firm's risk profile because of these
adjustments.
The Board also may seasonally adjust data, where appropriate. For
example, the vacancy rate of hotel commercial real estate exposures may
fluctuate on a seasonal cycle, with the vacancy rate moving higher or
lower in certain months based on a somewhat predictable pattern.
Because the vacancy rate can be an important variable for calculating
losses on hotel commercial real estate loans, this rate is seasonally
adjusted to ensure that the Commercial Real Estate Model produces more
stable results.
These types of data preparation steps help ensure that the Board's
models produce more reasonable results and that they align with the
principles in the Stress Testing Policy Statement in that they generate
consistent and robust projections. The Board therefore expects to
continue to use these data preparation steps, where appropriate, as
they are integral to the supervisory stress test process.
b. Data Adjustments
Data inputs are integral to generating the output of the stress
test models, which is a key component of a firm's stress capital buffer
requirement. The Board's Stress Testing Policy Statement notes that the
Board does not use data submitted by one or some of the firms unless
comparable data can be collected from all the firms that have material
exposure in a given area when generating supervisory stress test
projections.\63\ However, situations may arise where adjustments to a
firm's data would make the results more reasonable, and therefore
better calibrate a firm's stress capital buffer requirement to its risk
profile. The Board expects to continue to make these adjustments going
forward, where appropriate. Examples of when the Board may apply these
adjustments are described below.
---------------------------------------------------------------------------
\63\ See 12 CFR 252, Appendix B, section 2.8.
---------------------------------------------------------------------------
For example, the Board may apply a data adjustment where there is
missing or deficient firm-provided data, or where a firm uses
divestiture accounting. As described above, if the Board deems the
quality of a firm's submitted data too deficient to produce a stress
test model estimate for a particular portfolio, then the Board assigns
a conservative loss rate (for example, 90th percentile) or a
conservative pre-provision net revenue rate (for example, 10th
percentile) to the portfolio balances based on supervisory stress test
projections of portfolio losses or pre-provision net revenue for other
firms. If data that are direct inputs to stress test models are missing
or reported erroneously but the problem is isolated in such a way that
the existing supervisory framework can still be used, the Board assigns
a conservative value to the specific data based on all available data
reported by firms.
Additionally, when a firm sells assets or businesses, it may use
divestiture accounting in its financial statements until the sale is
consummated. Under divestiture accounting, a firm may list divested
assets as discontinued operations, classify them as held for sale or
available for sale instead of held for investment or held to maturity,
and report revenues as income from discontinued operations. The
accounting classification can be important for the supervisory stress
test as it may determine which model stresses the assets or income. For
example, in the 2025 supervisory stress test, the Board adjusted
certain input data that had been reclassified due to divestiture
accounting to improve projections of loan losses and related income to
ensure consistent treatment across firms with similar risks.
IV. Enhanced Disclosure Process
The Board is proposing to codify an enhanced disclosure process
under which the Board would annually publish comprehensive
documentation on the stress test models, invite public comment on any
material changes that the Board seeks to make to those models, and
annually publish the stress test scenarios for comment.
A. Annual Disclosure of Models
Under the proposal, the Board would annually publish comprehensive
documentation on the stress test models, similar to the comprehensive
documentation the Board is publishing with this proposal at <a href="https://www.federalreserve.gov/supervisionreg/dfa-stress-tests-2026.htm">https://www.federalreserve.gov/supervisionreg/dfa-stress-tests-2026.htm</a>. The
Board would be required to publish this comprehensive documentation by
May 15 of the year in which the stress test is performed, and the
models described in the documentation would be used to produce the
stress test results disclosed by the Board by June 30 of that year. In
addition, the Board would seek public comment, and respond to such
public comment, on any material changes to the models before
implementing those changes in a stress test. Material model changes are
discussed in more detail in Section IV.B of this Supplementary
Information. To implement this enhanced disclosure process, the Board
is proposing to revise Regulations YY and LL, as well as the Stress
Testing Policy Statement.
For example, if the Board did not seek to make any material model
changes to its stress test models for the 2027 supervisory stress test,
then it would publish the comprehensive model documentation used in the
2027 stress test cycle by May 15, 2027. This
[[Page 51869]]
documentation would identify any changes (relative to the models used
in the 2026 stress test), including technical, non-material changes to
the models to improve performance. This process would allow the public
to review the changes, as well as comprehensive documentation on the
models used in the 2027 stress test cycle, before the release of the
stress test results.
As an alternative example, if the Board sought to implement a
material model change (as discussed in Section IV.B of this
Supplementary Information) in the 2027 supervisory stress test, then
the Board would seek comment on the proposed change, consider and
respond to public feedback, and, then implement, defer, or reject the
material model change for the 2027 stress test cycle. If the Board
sought to implement the material model change in the 2027 stress test,
the Board would republish updated model documentation before or
simultaneously with the annual publication of comprehensive model
documentation (i.e., by May 15, 2027). This process for material model
changes would increase the transparency of the Board's stress testing
model framework and ensure that the public has the opportunity to
comment on material model changes before they are used in the next
stress test cycle.
Question 6: How else could the Board enhance the transparency and
public accountability of its stress test models? For instance, what
additional information regarding the stress test models, if any, should
the Board provide, and why?
Question 7: How else could the Board facilitate public
participation in model development? For example, the Board could invite
comment on all model changes, rather than only material model changes,
before implementing them in the stress test. Under such an approach,
the Board could make an exception for technical or other types of
ministerial changes. Such a process would limit the Board's flexibility
to revise models due to unforeseen events and circumstances. What are
the advantages and disadvantages of this expanded approach or other
approaches to facilitate public participation in model development? How
should the Board balance transparency and public accountability with
model dynamism and operational burden?
Question 8: What are the advantages and disadvantages of inviting
public comment, and committing to responding to comments, on material
model changes before the Board implements them in the subsequent stress
test?
Question 9: What are the advantages and disadvantages of publishing
the comprehensive model documentation by May 15 of each stress test
cycle? For example, does this timeline provide enough time for the
public to review any changes made by the Board to confirm they are not
material? Should the Board consider publishing the comprehensive model
documentation earlier at an earlier date, such as April 5, or a later
date, such as June 30? What would be the advantages or disadvantages of
publishing the comprehensive model documentation earlier or later?
Question 10: The Board is not currently publishing the results of
its internal model validation process. What would be the advantages and
disadvantages of publishing these results or providing more information
about its internal model validation process?
B. Model Changes
The proposed rule would define a ``model change'' to mean ``the
introduction of a new model or a conceptual change to an existing
model.'' \64\ Conceptual changes to existing models would include
changes to model assumptions, incorporation of a new statistical
technique to estimate loss, or the addition or deletion of any model
components or sub-components that currently inform a firm's stress
capital buffer requirement.
---------------------------------------------------------------------------
\64\ As discussed in Section II.D of this Supplementary
Information, there are twenty-one component models that comprise the
stress test models. A ``new model'' would mean a model that fully
replaces one of these twenty-one component models or is added to the
modeling suite (e.g., a 22nd component model). For purposes of
assessing materiality, as discussed in Section IV.C of this
Supplementary Information, model changes would not be aggregated or
netted across the component models.
---------------------------------------------------------------------------
Model changes would not include changes resulting from updates or
adjustments to input data, such as firm data, third-party vendor data,
and scenario data, including any re-estimation based on this data, as
well as changes related to the mechanical implementation of federal,
state, or local laws that are directly embedded in a stress test model
(e.g., the federal statutory tax rate).\65\ As is current practice, the
Board would continue to implement model changes related to changes in
accounting definitions or regulatory capital rules and model parameter
re-estimation based on newly available data with immediate effect.
These types of adjustments would not be considered model changes since
they do not substantively change the form of the stress test models as
described in the documentation. For example, the Board re-estimates
many of its models with updated data each year when it runs the
supervisory stress test. This re-estimation may result in changes to
the statistical coefficients produced by some of the models, even
though the Board has made no conceptual changes to the models. Under
the proposed definition of model change, such re-estimation would not
be viewed as a model change because the resulting changes stem solely
from updated data and not from a conceptual change to the models. In
contrast, the introduction or revision of a legal requirement that
causes a conceptual change to a model could be considered a model
change, and the Board would seek public comment before implementing
such a change if it met the proposed definition of a material model
change.
---------------------------------------------------------------------------
\65\ Re-estimation comprises updates to model parameters based
on consideration of different input data (e.g., incorporating the
most recent year's data as a model input, or incorporating data from
new stress test entrants or from mergers).
---------------------------------------------------------------------------
Question 11: What other types of changes to the supervisory stress
testing framework could the Board consider including in the definition
of ``model change''? What are the advantages and disadvantages of
broadening or narrowing the definition of ``model change''? For
example, should the Board define ``model changes'' to include changes
that result from new or updated input data, or changes that result from
using a new, third-party data source?
C. Material Model Changes
Each year, the Board refines and enhances its stress test models to
reflect advances in modeling techniques, respond to model validation
findings, incorporate richer and more detailed data, or identify more
stable models or models with improved performance, particularly under
stressful economic conditions. These changes may include re-
specification of models based on performance testing, benchmarking, and
other targeted changes used to produce projections.\66\ This process is
an important aspect of the modeling framework to help ensure that the
stress test models capture changes in borrower and lender behavior and
bank business practices. These model changes also help ensure that the
models are able to remain dynamic (i.e., can be enhanced to capture
emerging risks), produce
[[Page 51870]]
reasonable results, identify salient risks at firms, and maintain an
optimal level of robustness and stability.
---------------------------------------------------------------------------
\66\ Benchmarking is the process of evaluating a model's
performance by comparing its outputs and other performance metrics
against a specific standard, baseline, or the output and performance
of other comparable models or relevant data sources.
---------------------------------------------------------------------------
In addition, the Board must sometimes make changes to its stress
test models while it is running the stress test in response to
unforeseen events or circumstances to ensure that model output is
reasonable. For example, during the COVID-19 pandemic, the vacancy
rates for hotel properties were unprecedented and the Board made
certain adjustments to yield sensible commercial real estate loan
losses in the model output. Without making these in-cycle changes, the
results of the stress test would have been irrational and led to stress
capital buffer requirements that were not commensurate with applicable
firms' risk profile.
Under the proposed enhanced disclosure process, if these changes
are not material, as defined below, the Board would publish these model
changes by May 15 of the year in which the stress test is performed. To
balance the benefit of public feedback with the operational and
resource costs of seeking such feedback and to allow the Board to make
timely model adjustments to ensure reasonable results, the Board would
not formally invite public comment on these non-material model changes
before implementing them in the stress test; however, the Board
welcomes public feedback on these and all other aspects of the stress
test models once they are published. Notably, the Board would not
implement any in-cycle adjustments that are considered material model
changes prior to seeking public comment on the adjustment. In addition,
the Board would review and respond to all substantive public comments
on material model changes before implementing them in the stress test.
As discussed above, the Board is proposing to publish for comment
all material model changes and respond to all substantive comments on
such material model changes before implementing them in the stress
test. For example, if the Board sought to implement a new statistical
technique that would result in a material model change, then the Board
would seek public comment prior to implementing either of those
changes.\67\ The Board is proposing to define a ``material model
change'' as a model change that could have, in the Board's estimation,
an impact on the post-stress common equity tier 1 capital ratio of any
firm, or on the average post-stress common equity tier 1 capital ratios
of all firms required to participate in the upcoming stress test cycle,
based on the prior year's severely adverse scenario and prior year's
input data, equal to (i) a change of 20 basis points or more in the
projected common equity tier 1 ratio of any firm participating in the
upcoming stress test cycle; or (ii) a change of 10 basis points or more
in the average of the absolute value of each firm's change in projected
common equity tier 1 ratio.\68\ The Board proposes to apply this
definition of a material model change across both Regulation YY and
Regulation LL, such that the individual materiality threshold would
apply to all firms required to participate in the next stress test
under either regulation, and such that the Board's estimation of
whether a change meets the aggregate materiality threshold would be
determined across all firms required to participate in the next stress
test under either regulation.
---------------------------------------------------------------------------
\67\ For purposes of assessing materiality, model changes would
not be aggregated or netted across the component models. For
example, if the Board proposed a change to both the Pre-Provision
Net Revenue Model and Corporate Model in the same stress test cycle,
the Board would estimate the effects of each change separately for
purposes of determining materiality. Similarly, for purposes of
assessing materiality, model changes would not be aggregated or
netted within component models. For example, if the Board proposed
two changes to a component model, the Board would evaluate the
materiality of each change separately.
\68\ The Board would take the absolute value of each firm's
change in projected common equity tier 1 ratio, then average those
values. If the average is 10 basis points or greater, the change
would constitute a material model change.
---------------------------------------------------------------------------
The Board is proposing to use the threshold of a 20 basis point
change in the common equity tier 1 capital ratio for individual firms
in the definition of material model change because that threshold would
ensure that the public would be able to comment on any change that
would be likely to affect a firm's stress capital buffer requirement.
Considering the history of recent model changes, a threshold of 20
basis points would generally capture model changes that involve
conceptual enhancements to model specifications, such as to incorporate
improved modeling techniques or to capture emerging risks, while
scoping out those that are simpler model refinements, such as those
implemented to ensure that the models maintain consistency given
changing requirements (e.g., refinements made to accommodate the
transition from the London Interbank Offered Rate to SOFR). Therefore,
changes of smaller magnitudes would be unlikely to impact a firm's
stress capital buffer requirement, particularly if the proposed two-
year averaging approach to calculate a firm's stress capital buffer
requirement is adopted.\69\ If the two-year averaging approach is not
finalized or not finalized as proposed, the Board would consider a
lower individual materiality threshold of 10 basis points, which would
ensure that the public would be able to comment on any change that
would be likely to affect a firm's stress capital buffer requirement
without two-year averaging.
---------------------------------------------------------------------------
\69\ See 90 FR 16843 (Apr. 22, 2025).
---------------------------------------------------------------------------
The Board is proposing the threshold of a 10 basis point average
change in the absolute value of the change to each firm's projected
common equity tier 1 capital ratio in case a model change has minimal
individual impacts, but has a notable aggregate impact on firms
required to participate in the upcoming stress test. The Board selected
10 basis points for this aggregate prong because a model change of this
size would be likely to impact the aggregate projected common equity
tier 1 capital ratio, which is a salient measure of the health of the
banking system. A change that satisfies either of these materiality
thresholds would be considered a material model change.
Question 12: What are the advantages and disadvantages of this
definition of a material model change? What alternative quantitative
thresholds for materiality, if any, should the Board consider, and why?
For example, in assessing the materiality of a model change, as
described in the Stress Testing Policy Statement, the Federal Reserve
currently considers a change to be highly material if it would result
in a change in the common equity tier 1 capital ratio of 50 basis
points or more for one or more firms, relative to the model used in
prior years' supervisory exercises. What would be the advantages and
disadvantages of this or other alternative standards?
Question 13: What alternative definitions of materiality, if any,
should the Board consider? For example, the Board could consider the
impact of a change on a firm's pre-tax net income, rather than its
common equity tier 1 ratio. What are the advantages and disadvantages
of such alternative definitions?
Question 14: Under the proposal, for purposes of assessing the
materiality of a model change, the Board would not aggregate or net the
impact across or within component models. What forms of netting or
aggregation, if any, would be most appropriate and why? What would be
the advantages and disadvantages of netting or aggregating model
changes across or within component models to assess materiality? If the
Board were to net or
[[Page 51871]]
aggregate model changes, should the Board consider alternative
materiality thresholds? For example, the Board could consider an
alternative definition of materiality that considers the aggregate
impact of all of the model changes the Board intends to implement in a
future stress test cycle. Alternatively, the Board could aggregate the
impacts of all model changes to a given suite of models (e.g., credit
risk models) instead of considering the individual impacts of model
changes to the Auto Loan Model and the Commercial Real Estate Model.
Question 15: What are the advantages and disadvantages of inviting
and responding to public comment on material model changes before
implementing those changes? The proposal does not currently specify the
length of the comment period. What are the advantages and disadvantages
of a set length for the comment period (e.g., 30-day, 60-day, etc.)?
When considering the appropriate length of the comment period, how
should the Board evaluate trade-offs, for instance, between ensuring
that the public has ample time to consider and comment on material
model changes and ensuring that the stress test results are released by
June 30?
Question 16: If the Board does not adopt its proposal to calculate
a firm's stress capital buffer requirement by averaging stress test
results over two consecutive years, should the Board consider a lower
threshold to determine materiality, such as 10 basis points for the
individual firm threshold instead of the proposed 20 basis points? What
would be the advantages and disadvantages of a lower threshold?
D. Annual Disclosure of Scenarios
Under the proposal, the Board would annually publish for comment
the proposed stress test scenarios by October 15 of the calendar year
prior to the stress test, for at least a 30-day period. The timing of
the release and duration of the comment period will allow for
sufficient time to respond to comments and finalize the scenarios
within the current window for publishing final scenarios by February 15
in each annual stress test cycle.\70\ The disclosure of the annual
scenarios for comment, along with the implementation of additional
scenario variable guides and revisions to the Scenario Design Policy
Statement, would meaningfully improve the transparency, public
accountability, and predictability of the annual stress tests.
---------------------------------------------------------------------------
\70\ Trading or other components of the scenarios, and any
additional scenarios used by the Board, would continue to be
communicated by March 1 of the calendar year in which the stress
test is performed. 12 CFR 238.132(b); 12 CFR 238.143(b)(2)(i); 12
CFR 252.14(b)(2)(i); 12 CFR 252.44(b); 12 CFR 252.54(b)(2)(i).
---------------------------------------------------------------------------
The publication of macroeconomic scenarios in October would use
nowcasts, which are projections under baseline conditions, to determine
the jump-off points for the proposed scenario variable paths. The final
scenarios would be updated to include actual data. The paths of
scenario variables may be adjusted to some extent between the initial
scenario publication and the finalized scenario to reflect these
updated values.
By designing and publishing the guides described in Section IX.G of
this Supplementary Information, the Board expects that the annual
severely adverse scenarios will generally be more consistent and
predictable year-to-year. As a result, the Board weighed whether
publishing the annual scenarios for comment in a typical year would
contribute meaningful additional accountability that would improve the
stress test program, and whether the Board should limit publication of
the annual scenarios for comment to situations where the Board is
proposing to incorporate a salient risk into the scenarios that is not
described in this proposal. However, in the interest of enhancing
transparency and public accountability, the Board determined to
maintain its current practice of publishing its annual scenarios and,
further, to propose changes to Regulations LL and YY formalizing this
disclosure process.
Additionally, the Board plans to maintain its current practice of
disclosing the final scenarios only after firms' portfolios are fixed,
as disclosure of the final scenarios prior to the jump-off date of the
stress test could incentivize firms to modify their businesses to
minimize losses in the supervisory stress test without changing the
actual risk profile of the firms. Therefore, the Board is proposing to
move the jump-off date of the stress test from December 31 to September
30. This proposed change is discussed in greater detail in Section VI.A
of this Supplementary Information.
Finally, as described in Section VI.B of this Supplementary
Information, the Board is proposing to change the as-of date window for
the global market shock to occur between October 1 of the calendar year
two years prior to the year in which the stress test is performed to
October 1 of the year prior to the year in which the stress test is
performed. Therefore, the Board anticipates that the global market
shock as-of date will have already occurred for most future proposals
regarding the initial disclosure of the stress test scenarios. However,
the Board has not yet announced the global market shock as-of date for
the 2026 stress test and so cannot provide the exact relative shock
values for certain global market shock variables since the relative
shock values are a function of the actual data on the as-of date.
For relative shocks associated with the 2026 global market shock,
the data on the global market shock as-of date would be applied to
determine relative shock values, which will be disclosed as part of the
finalized scenarios. For example, if the Board proposes a shock to the
BBB corporate spread of 200 basis points and the BBB corporate spread
market level on the global market shock as-of-date is 400 basis points,
then the relative shock to the BBB corporate spread would be 200/400,
or 50 percent, for the 2026 global market shock.
Question 17: How should the Board publish the annual scenario for
comment? For example, the Board could publish the scenario on the
Board's website or include the text and supporting materials in a
Federal Register notice. Alternatively, the Board could consider
codifying each annual scenario as a part of Regulation YY. What would
be the advantages and disadvantages of these options or other
alternatives?
Question 18: What are the advantages and disadvantages of
publishing the annual scenarios for comment prior to the jump-off date
of the annual stress test cycle?
Question 19: What are the advantages and disadvantages of a 30-day
comment period? Should the Board consider an alternative comment period
length? If so, how long should the comment period be (e.g., 45 days, 60
days, etc.)? When considering the appropriate length of the comment
period, how should the Board evaluate trade-offs, for instance, between
ensuring that the public has ample time to consider and comment on
annual scenarios and ensuring that the stress test scenarios can be
finalized before February 15?
Question 20: How should the Board analyze comments received from
the public on proposed scenarios? What types of information would be
helpful to commenters in order to understand how the Board incorporates
comments received on proposed scenarios before finalizing the annual
scenarios?
E. Stress Capital Buffer Requirement Reconsideration Process
Under the Board's capital plan rule, a firm may request
reconsideration of the calculation of its preliminary stress
[[Page 51872]]
capital buffer requirement within 15 calendar days of receiving notice
of the preliminary requirement.\71\ A request for reconsideration may
include a request for an informal hearing on the firm's request for
reconsideration; the Board may, in its sole discretion, order an
informal hearing if the Board finds that a hearing is appropriate or
necessary to resolve disputes regarding material issues of fact.\72\
The Board is not proposing to change this reconsideration process.\73\
However, the Board is requesting public input on potential enhancements
to the stress capital buffer requirement reconsideration process. In
particular, the Board seeks public input on the following question:
---------------------------------------------------------------------------
\71\ 12 CFR 225.8(h)(2)(i); 12 CFR 225.8(i)(2); 12 CFR
238.170(h)(2)(i); 12 CFR 238.170(i)(2).
\72\ 12 CFR 225.8(i)(3)(ii); 12 CFR 225.8(i)(4); 12 CFR
238.170(i)(3)(ii); 12 CFR 238.170(i)(4).
\73\ Model adjustments made in response to a reconsideration
request granted by the Board would not be considered model changes
under the proposed enhanced disclosure process.
---------------------------------------------------------------------------
Question 21: What enhancements, if any, should the Board consider
making to its reconsideration request process? For example, the Board
could allow firms more time to request reconsideration of their
results, broaden or narrow the grounds for and scope of review, and/or
modify existing reconsideration request requirements in light of the
publication of the comprehensive model documentation and proposed
enhanced disclosure process. What would be the advantages and
disadvantages of these enhancements? What other changes, if any, should
the Board consider making to the reconsideration requirements and
procedures? What would be the advantages and disadvantages of those
changes?
V. Revisions to the Stress Testing Policy Statement
The Board is also proposing certain changes to the Stress Testing
Policy Statement to (i) amend the section related to disclosure of
information related to the stress test; and (ii) to align the Stress
Testing Policy Statement with the proposed enhanced disclosure process.
A. Future Supervisory Stress Test Results Disclosures
The Board's Stress Testing Policy Statement states that, in
general, the Board does not share information regarding supervisory
stress test results with firms that is not made available to the
broader public. However, providing additional details to a firm about
its own results could provide the firm with additional visibility into
its stressed revenue and loss projections, including any underlying
risks, and improve the firm's understanding of its stress capital
buffer requirement. For example, additional results information would
allow a firm to better understand how the stress test translates their
balance sheet and income information into projected losses and revenue,
which could help them better plan their business and understand the
risk of their exposures. To provide additional transparency, the Board
is therefore proposing to revise the Stress Testing Policy Statement to
clarify that the Board will generally disclose information directly to
a firm about the firm's supervisory stress test results that is not
available to the broader public, so long as the Board discloses similar
information to the other firms participating in a given stress test
cycle. For example, the Board may provide a firm's common equity tier 1
capital ratio during all quarters of the projection horizon. Providing
firm-specific results directly to the affected firms even when that
information is not disclosed to the broader public would allow firms to
better understand their results while preventing potentially sensitive
information about a firm from being shared with competitors. The Board
would continue to disclose the supervisory stress test results to the
public.
Question 22: What are the advantages and disadvantages of revising
the Stress Testing Policy Statement to clarify that the Board will
generally share non-public information about a firm's results directly
with a firm (provided that the Board is disclosing similar information
to other participating firms)?
B. Other Revisions to the Stress Testing Policy Statement
In addition, the Board is proposing to revise the Stress Testing
Policy Statement to align it with the proposed enhanced disclosure
process. For example, the Board is proposing to state that, during
model development, it invites, evaluates, and responds to substantive
public input on the stress test models. The Board is also proposing to
revise the Stress Testing Policy Statement to clarify that its public
disclosures about the stress test will now include comprehensive
descriptions of the models and changes to those models.
Question 23: What other changes could the Board make to the Stress
Testing Policy Statement to reflect the enhanced transparency of the
supervisory stress test or to supplement the Board's efforts to make
the supervisory stress test more transparent and to facilitate public
participation? What are the advantages and disadvantages of such
changes?
VI. Other Revisions to the Stress Testing and Capital Plan Rules
The Board is also proposing to revise the stress testing and
capital plan rules to reflect the Board's efforts to disclose more
information about the stress test scenarios.
A. Stress Test Jump-Off Date Change
The Board is separately seeking comment on the proposed scenarios
for use in the 2026 supervisory stress test. In general, disclosure of
the proposed scenarios prior to the jump-off date of the supervisory
stress test could incent firms to temporarily modify their businesses
to affect the results of the stress test without changing the actual
risk profile of the firms. The Board recognizes that the increased
transparency around scenario design resulting from the disclosure of
additional guides and a macroeconomic model used in that process would
allow firms to anticipate the trajectories of key scenario variables.
Using this information, firms could adjust their portfolios to specific
aspects of the proposed scenarios in ways that would reduce measured
losses without reducing the actual riskiness of the portfolios. Such
changes to firm business profiles could also result in greater than
typical quarter-to-quarter variability in the banking books of firms.
To address this potential risk associated with increased
transparency, the Board proposes to modify the jump-off date of the
supervisory and company-run stress tests from December 31 to September
30, while leaving unchanged the other dates associated with publication
of the final scenario and stress test results.\74\ With respect to the
capital planning rules, the Board proposes accomplishing this change
through revision to the definition of ``planning horizon'' in
Regulation Y and Regulation LL. This change would allow the Board to
publish the scenario for comment after the jump-off date of the stress
test, preventing firms from adjusting their exposures based on the
stress test. However, this proposed change would introduce an
additional quarter of staleness to the stress test and
[[Page 51873]]
stress test results. This change would also affect firms' capital plan
submissions. Although the due date for firms' annual capital plan
submissions would be unchanged, because of the proposed update to the
definition of planning horizon, firms' capital plans would not project
out as far. While the Board weighs these risks and considers adjusting
the stress test jump-off date, the Board seeks input from the public
regarding whether these risks are outweighed by the value to firms and
the public by publishing scenarios prior to the jump-off date of the
supervisory and company-run stress tests. Therefore, the Board seeks
public comment on whether to propose such modifications to limit the
ability of firms to adjust their balance sheets in response to the
proposed scenario prior to the jump-off date of the stress test.
---------------------------------------------------------------------------
\74\ The Board has experience operating the annual supervisory
stress test with a September 30 jump-off date. See, e.g., Board,
2015 Supervisory Scenarios for Annual Stress Tests Required under
the Dodd-Frank Act Stress Testing Rules and the Capital Plan Rule
(Oct. 23, 2014), <a href="https://www.federalreserve.gov/newsevents/pressreleases/files/bcreg20141023a1.pdf">https://www.federalreserve.gov/newsevents/pressreleases/files/bcreg20141023a1.pdf</a>.
---------------------------------------------------------------------------
Question 24: What are the advantages and disadvantages of retaining
a jump-off date that would occur after the publication of the annual
scenario for comment? Should the Board consider retaining the December
31 jump-off date in order to promote transparency? Are there additional
risks or trade-offs that the Board should consider?
Question 25: What would be the advantages and disadvantages of
modifying the jump-off date of the stress test from December 31 to
September 30? If the Board were to modify the jump-off date, what other
changes should the Board consider making to the stress test timeline?
For example, what would be the advantages and disadvantages if the
Board were to change the timing of a firm's capital plan submission?
What would be the advantages and disadvantages of these changes?
Question 26: Should the Board consider modifying the jump-off date
of the stress test to a later date, rather than an earlier date, in
order to accommodate a public comment period?
B. Global Market Shock Date
The global market shock (GMS) is applied to market risk positions
held by the firms on a given as-of date, which, under the Board's
stress test rule, currently occurs between October 1 of the previous
year and March 1 of the year of a given stress test cycle.\75\ Under
the Board's regulations, the GMS can apply to both the supervisory
stress test and the company-run stress test for applicable firms. For
the supervisory stress test and the company-run stress test, the Board
must generally provide each affected firm with a description of the GMS
and with the specific GMS as-of date by March 1 of the year in which
the stress test occurs.\76\ For the company-run stress test, the Board
generally must also notify each affected firm by December 31 of year
preceding the stress test that the firm is required to include
additional components or scenarios in its company-run stress test.\77\
---------------------------------------------------------------------------
\75\ See, e.g., 12 CFR 238.143(b)(2)(i); 12 CFR 252.14(b)(2)(i);
12 CFR 252.54(b)(2)(i).
\76\ See, e.g., 12 CFR 238.132(b); 12 CFR 238.143(b)(2)(i); 12
CFR 252.14(b)(2)(i); 12 CFR 252.44(b); 12 CFR 252.54(b)(2)(i).
\77\ See, e.g., 12 CFR 238.143(b)(4)(i); 12 CFR 252.14(b)(4)(i);
12 CFR 252.54(b)(4)(i).
---------------------------------------------------------------------------
The Board selects a cycle-specific as-of date each year and,
typically, announces it to firms about two weeks later to ensure the
firms retain necessary data. The as-of date is expected to change from
year to year to avoid creating potential incentives for firms to take
temporary trading positions. However, there is a comprehensive date
selection process that, in practice, shortens the actual window during
which the GMS as-of date is generally selected. A wider date range
would allow the Board to capture a broader range of market risks across
different time periods, thereby improving the risk capture of the
global market shock. The Board therefore proposes revising the date
range for the GMS as-of date to occur between (inclusive of) October 1
of the calendar year two years prior to the year in which the stress
test is performed to (exclusive of) October 1 of the calendar year one
year prior to the year in which the stress test is performed. By way of
example, this change would mean that for the 2026 supervisory stress
test, the GMS as-of date could fall on any date between October 1,
2024, through September 30, 2025. The Board proposes using this date
range because it would allow the Board to choose from a full year's
worth of potential GMS as-of dates. Additionally, the proposed range
would include only dates prior to the release of the given stress test
cycle's GMS for notice and comment. Therefore, firms subject to the GMS
would not be able to use their knowledge of the GMS as-of date to
update their balance sheet positions or adjust their portfolios to
minimize stress losses without a commensurate reduction in risk
profile.
In conjunction with the proposal to change the GMS as-of date
window, the Board also proposes to change the date by which the Board
needs to notify affected firms of this as-of date from March 1 of the
year in which the stress test occurs. Unless it determines otherwise,
the Board must notify affected firms of the GMS as-of date by October
15 of the year prior to the year in which the stress test is performed.
The Board would continue to provide firms with a description of the
GMS, as finalized, by March 1 of the calendar year in which the stress
test occurs. Additionally, to conform to the proposed changes to the
stress test timeline, the Board proposes to change the date by which
the Board must notify firms that they are required to include
additional components or scenarios in their company-run stress test
from December 31 to September 30 of the year preceding the stress test.
This change would ensure that firms are aware of the components to
which they would be subject prior to the annual publication of the
scenarios for notice and comment.
Question 27: What are the advantages and disadvantages of modifying
the window for the GMS as-of date in the stress test from October 1 of
the calendar year one year prior to the year in which the stress test
is performed through March 1 of the year in which the stress test is
performed, to a date that is no earlier than October 1 of calendar year
two years prior to the year in which the stress test is performed and
that precedes October 1 of the calendar year one year prior to the year
in which the stress test is performed? What alternative GMS as-of date
ranges, if any, should the Board consider, and why? In addition to
changing the GMS as-of date window, what other changes, if any, should
the Board consider making to the stress test timeline? What effects, if
any, would changing the window for the GMS as-of date have on any other
aspects of the stress test or the stress test timeline?
Question 28: What are the advantages and disadvantages of the
proposed dates by which the Board would notify firms of the GMS as-of
date, provide a description of any associated components, and notify
firms of any additional components that they are required to include in
their supervisory and company-run stress tests? What alternative dates,
if any, should the Board consider for these activities and why? For
example, to better ensure that more stakeholders provide input into the
proposed GMS, the Board could wait until the scenarios are final before
notifying firms which components they must include in their company-run
run stress tests.
Question 29: The GMS only considers a firm's positions on one as-of
date and only under one set of shocks. Should the Board consider
alternative approaches to further increase the risk capture of the GMS,
such as applying the GMS to more than one as-of date or more than
[[Page 51874]]
one set of shocks for a given stress test? What would be the advantages
and disadvantages of these alternative approaches? What other
approaches should the Board consider to improve the risk capture of the
GMS and why?
C. Amendment to the Dividend Add-On Component Calculation
The dividend add-on component of the stress capital buffer
requirement currently comprises planned dividends in the fourth through
seventh quarters of the planning (or projection) horizon of the
supervisory stress test.\78\ Under the current framework, the planned
dividends that are incorporated in the stress capital buffer
requirement align with the effective date of the stress capital buffer
requirement (that is, October 1 generally is the first day of the
fourth quarter of the existing planning horizon) and last for the one-
year period through which the stress capital buffer requirement is
expected to be effective (that is, through the seventh quarter of the
existing planning horizon, after which the following year's stress
capital buffer requirement would be expected to take effect).
---------------------------------------------------------------------------
\78\ See 12 CFR 225.8(d)(16); 12 CFR 238.130. The planning (or
projection) horizon for the supervisory stress test is nine
consecutive quarters starting on the jump-off date of the
supervisory stress test.
---------------------------------------------------------------------------
As part of this rulemaking, the Board is proposing to change the
jump-off date of the stress test from December 31 to September 30. To
maintain alignment between the dividend add-on component of the stress
capital buffer requirement and the one-year period during which the
requirement typically is effective, the Board proposes to change the
dividend add-on component to cover dividends issued in quarters five
through eight, instead of quarters four through seven, of the planning
horizon of the supervisory stress test. This change involves updates to
the capital plan rules, at Regulation Y and Regulation LL, to any
references to the relevant quarters of the planning horizon.\79\ This
proposed revision is intended to maintain the alignment between the
dividend add-on component and the one-year period during which the
stress capital buffer requirement generally is effective, assuming the
proposal to move the jump-off date of the stress test to September 30
is adopted. If this aspect of the proposal is not adopted, then the
Board would not adjust the planning horizon period for planned
dividends.
---------------------------------------------------------------------------
\79\ 12 CFR 225.8(f)(2)(i)(C)(1); 12 CFR 225.8(f)(4); 12 CFR
225.8(h)(2)(ii)(A); 12 CFR 225.8(h)(2)(ii)(B); 12 CFR 225.8(k)(2);
12 CFR 238.170(f)(2)(i)(C)(1); 12 CFR 238.170(f)(4); 12 CFR
238.170(h)(2)(ii)(A); 12 CFR 238.170(h)(2)(ii)(B); 12 CFR
238.170(k)(2).
---------------------------------------------------------------------------
Such a change to the planning horizon period has likewise been
proposed as part of the Board's proposed Modifications to the Capital
Plan Rule and Stress Capital Buffer Requirement, issued in April 2025,
in order to similarly maintain alignment between that proposal's
updates to the effective date of a firm's stress capital buffer
requirement and the dividend add-on component.\80\ Should both
proposals be finalized as proposed, the Board would expect to adjust
the dividend add-on component of the stress capital buffer requirement
to maintain alignment between the dividend add-on component and the
one-year period in which the stress capital buffer requirement
generally is effective. In such an instance, the Board would expect to
change the dividend add-on component to cover dividends issued in
quarters six through nine of the planning horizon of the supervisory
stress test.
---------------------------------------------------------------------------
\80\ See 90 FR 16843 (Apr. 22, 2025).
---------------------------------------------------------------------------
Question 30: What would be the advantages and disadvantages of the
proposed change to the dividend add-on component of the stress capital
buffer requirement?
VII. Revisions to the FR Y-14A/Q/M
To reduce regulatory reporting burden, support the proposed model
changes, and improve risk capture, the Board is proposing several
revisions to the FR Y-14A/Q/M. To reduce regulatory reporting burden,
the Board is proposing to remove items and documentation requirements
that are no longer needed to conduct the supervisory stress test. For
example, the proposal would remove certain FR Y-14 supporting
documentation requirements that are no longer needed to assess a firm's
FR Y-14 submission. The Board also proposes to collect additional data
that would support the supervisory stress test models and improve risk
capture. For example, to capture data in a manner that aligns better
with the treatment of private equity under the macroeconomic scenario,
the proposal would include revisions for reporting private equity
exposures and associated hedges. Additionally, to broaden the
consideration of hedges and revenue and loss sharing agreements in the
stress test, the Board is proposing revisions that would capture more
data on various types of hedges or revenue and loss sharing agreements.
Lastly, the Board is proposing several minor revisions to clarify the
FR Y-14 instructions or align with the proposed changes to the stress
test timeline. The proposed revisions are described in Section XI.A of
this Supplementary Information.
VIII. Proposed Changes to the Stress Test Modeling Framework
The Board is proposing to use the models described in the documents
posted on the Board's website to generate results for the 2026
supervisory stress test. Included in these descriptions are some model
specifications that were not used to conduct the 2025 supervisory
stress test but are proposed to be used for the 2026 supervisory stress
test. These proposed model changes are summarized in Section
VIII.A.\81\ In addition, a detailed description of and rationale for
each of these proposed model changes is provided in a separate document
posted on the Board's website with the comprehensive model
documentation. Section VIII.B of this Supplementary Information
provides an analysis of the potential effects of the proposed changes.
Based on this analysis, implementing the proposed model changes and
proposed revisions to the global market shock scenario design in the
2024 and 2025 stress tests would have, independent of other factors,
increased the aggregate projected common equity tier 1 (CET1) stress
ratio, on average, by 29 basis points. This is equivalent to a
reduction in stress capital buffer requirements of approximately 23
basis points or approximately 2.2 percent of current required capital.
---------------------------------------------------------------------------
\81\ These proposed changes would constitute ``model changes''
under the proposed definition of ``model change,'' as discussed in
Section IV.B of this Supplementary Information.
---------------------------------------------------------------------------
A. Proposed Changes to Stress Test Models
The Board is proposing several changes to the supervisory stress
test models for the 2026 stress test, which are discussed in more
detail in the Model Changes document provided on the Board's website,
at <a href="https://www.federalreserve.gov/supervisionreg/dfa-stress-tests-2026.htm">https://www.federalreserve.gov/supervisionreg/dfa-stress-tests-2026.htm</a>. More significant proposed changes to the Pre-provision Net
Revenue and Operational Risk Models are described within the
comprehensive model documentation, also available at <a href="https://www.federalreserve.gov/supervisionreg/dfa-stress-tests-2026.htm">https://www.federalreserve.gov/supervisionreg/dfa-stress-tests-2026.htm</a>. The
Board is requesting comment on the proposed changes, together with the
model documentation.
With respect to the credit risk models, the Board is proposing to
change how it uses geography in scenario variables (First Lien, Home
Equity, Credit Cards,
[[Page 51875]]
Auto, and Commercial Real Estate Models); change how it treats
foreclosures under judicial supervision (First Lien and Home Equity
Models); change how it calculates loss given default for international
loans (Commercial Real Estate and Corporate Models); change how it
includes losses attributable to accrued interest and carrying costs
(First Lien and Home Equity Models); change how it uses multipliers in
the Provisions Model; revise the mortgage loss given default model in
the First Lien Model; revise the bank card model in the Credit Card
Model; change how it projects losses on auto leases in the Auto Model;
and update the probability of default, loss given default, and exposure
at default components in the Corporate Model.
With respect to the market risk models, the Board is proposing to
update several of its market risk models for the 2026 stress test,
including to simplify the Yield Curve Model; adjust its process for
projecting credit valuation adjustments for derivative positions in the
Credit Valuation Adjustment Model; lower the loss given default
assumption amount and loan equivalent factor parameter in the Fair
Value Option Model; update and simplify the Securities Model; and
exclude additional counterparties in the Largest Counterparty Default
Model.
With respect to the net revenue models, the Board is proposing an
alternative suite of pre-provision net revenue component models that
depart from the current panel regression-based approach. This
alternative suite is described in the Pre-provision Net Revenue Model
documentation, available at <a href="https://www.federalreserve.gov/supervisionreg/dfa-stress-tests-2026.htm">https://www.federalreserve.gov/supervisionreg/dfa-stress-tests-2026.htm</a>. The Board is also proposing
to discontinue the current regression model used to project operational
risk losses and instead project losses with a distributional model.
This alternative model is described in the Operational Risk Model
documentation, also available at <a href="https://www.federalreserve.gov/supervisionreg/dfa-stress-tests-2026.htm">https://www.federalreserve.gov/supervisionreg/dfa-stress-tests-2026.htm</a>.
Aggregate impacts on regulatory capital of the model changes
described above are small (see Table 2). Across risk stripes, the
proposed model changes would reduce credit, market, and operational
losses, which would be balanced by the effects of the proposed model
changes to the Pre-provision Net Revenue Model. Across firm categories,
GSIBs would observe modest increases in aggregate projected CET1 stress
ratio under the proposed changes. Firms subject to Category II-III
standards would also observe a modest increase in their projected CET1
stress ratio.
Question 31: The Board invites public comment on these proposed
model changes. What other changes, if any, should the Board consider
implementing in the 2026 stress test cycle, either instead of or in
addition to the proposed changes?
Question 32: What other information or data should the Board
consider to assess the quantitative economic impact of the proposed
model changes and why?
B. Analysis of Proposed Model Changes
To further enhance the transparency of the stress test models, this
section analyzes the potential effects of the proposed model changes
described in Section VIII.A of this Supplementary Information, and the
liquidity horizon revisions to the global market shock scenario design
described in Section IX.H of this Supplementary Information, that
inform the Board's determination of firms' stress capital buffer
requirements.
In aggregate, the stress test model and scenario changes are not
expected to materially change capital requirements for firms subject to
the supervisory stress test, across various stress scenarios and jump-
off conditions at the start of the test. To illustrate the effect of
these proposed model changes, this analysis averaged the impact of
these changes on the CET1 stress ratio for a balanced sample of 30
firms subject to the 2024 stress test and expected to participate in
the 2026 stress test, then aggregated the averages.\82\ The analysis
estimates that the proposed model and scenario changes, independent of
other models and components, could have resulted in an increase of 29
basis points in the average aggregate CET1 stress ratio. This is
equivalent to a reduction in stress capital buffer requirements of
approximately 23 basis points or approximately 2.2 percent of current
required capital. The analysis estimates that the model changes would
reduce stress capital buffer requirements by approximately 13 basis
points, and that the revisions to the global market shock scenario
design, described in Section IX.H of this Supplementary Information,
would reduce stress capital buffer requirements by approximately 10
basis points. For U.S. GSIBs, the analysis estimates a decline of 25
basis points of stress capital buffer requirements.
---------------------------------------------------------------------------
\82\ This analysis used the 2024 and 2025 scenarios,
respectively, and the same data used for those years' stress tests.
The estimated impact of these changes remains highly sensitive to
the stress test scenario and firm-specific data for each year. While
the precise impact will vary each year based on stress test
scenarios and specific firm data, Board analysis across a range of
conditions shows that capital requirements should remain essentially
unchanged.
---------------------------------------------------------------------------
As the U.S. banking system's 13.0 percent CET1 capital ratio (8.2
percent leverage ratio) is well within the estimated optimal range in
the literature,\83\ the net benefit of modest changes to the overall
level of banking system capital is small.\84\ However, as discussed
further below, the proposed model changes have varied effects on
capital requirements across loss type and firm category.
---------------------------------------------------------------------------
\83\ For discussions of optimal bank capital, see generally
Basel Committee, ``An Assessment of the Long-Term Economic Impact of
Stronger Capital and Liquidity Requirements'' (Aug. 2010), <a href="https://www.bis.org/publ/bcbs173.pdf">https://www.bis.org/publ/bcbs173.pdf</a> (``BCBS 2010 study''); see also I.
Fender & U. Lewrick, Adding it All Up: The Macroeconomic Impact of
Basel III and Outstanding Reform Issues, BIS Working Paper No. 591
(Nov. 2016) (``Fender and Lewrick (2016)''), <a href="https://www.bis.org/publ/work591.pdf">https://www.bis.org/publ/work591.pdf</a>; D. Miles et al., Optimal Bank Capital, 123 The
Econ J. 1, 29 Table 10 (Mar. 2013) (``Miles et al. (2013)''),
<a href="https://academic.oup.com/ej/article/123/567/1/5080596">https://academic.oup.com/ej/article/123/567/1/5080596</a>; M. Brooke et
al., Measuring the Macroeconomic Costs and Benefits of Higher UK
Bank Capital Requirements, Bank of England, Financial Stability
Paper No. 35 (Dec. 2015) (``Brooke et al. (2015)''), <a href="https://www.bankofengland.co.uk/-/media/boe/files/financial-stability-paper/2015/measuring-the-macroeconomic-costs-and-benefits-of.pdf">https://www.bankofengland.co.uk/-/media/boe/files/financial-stability-paper/2015/measuring-the-macroeconomic-costs-and-benefits-of.pdf</a>; S.
Firestone et al., An Empirical Economic Assessment of the Costs and
Benefits of Bank Capital in the United States, 101 Federal Reserve
Bank of St. Louis Rev. 203, 203-30 (2019) (``Firestone et al.
(2019)''), <a href="https://doi.org/10.20955/r.101.203-30">https://doi.org/10.20955/r.101.203-30</a>; B. Soederhuizen,
et al., Optimal Capital Ratios for Banks in the Euro Area, 69 J.
Fin. Stability, Art. No. 101164 (Dec. 2023) (``Soederhuizen et al.
(2023)''), <a href="https://doi.org/10.1016/j.jfs.2023.101164">https://doi.org/10.1016/j.jfs.2023.101164</a>; J. Barth & S.
Matteo Miller, Benefits and Costs of a Higher Bank `Leverage
Ratio','' 38 J. Fin. Stability 37, 37-52 (Oct. 2018) (``Barth and
Miller (2018)''), <a href="https://doi.org/10.1016/j.jfs.2018.07.001">https://doi.org/10.1016/j.jfs.2018.07.001</a>; J.
Dagher et al., Benefits and Costs of Bank Capital, IMF Staff
Discussion Note SND/16/04 (Mar. 2016) (``Dagher et al. (2016)''),
<a href="https://www.imf.org/external/pubs/ft/sdn/2016/sdn1604.pdf">https://www.imf.org/external/pubs/ft/sdn/2016/sdn1604.pdf</a>.
\84\ Ratios are based on the aggregate of all FR Y-9C filers as
of Q1 2025, which generally excludes holding companies with less
than $3 billion in consolidated assets and depository institutions
without parent holding companies. The aggregate CET1 ratio
additionally excludes holding companies that have opted in to the
Community Bank Leverage Ratio requirement, and reflects standardized
risk-weighted assets.
---------------------------------------------------------------------------
Based on this analysis, the proposed model changes are expected to
result in more risk-sensitive capital requirements, independent of
their effect on the level of requirements. Specifically, implementation
of the proposed model changes would render the models more stable,
likely reducing misalignment between firms' losses under stress and
their respective stress capital buffer requirements. To the extent that
the stress capital buffer requirements are affected by these proposed
model changes and are a part of firms' most-binding capital
constraint,\85\ the proposed model
[[Page 51876]]
changes would thereby improve the risk sensitivity--and efficiency and
effectiveness--of capital requirements.
---------------------------------------------------------------------------
\85\ The capital requirements of firms with stress losses plus
dividend add-ons reliably below the 2.5 percent capital conservation
buffer would be unaffected by the proposed model changes.
---------------------------------------------------------------------------
This analysis recognizes that the limited overall effect on
stressed CET1 capital ratios masks significant variation across the
different loss drivers. As shown in Table 2 below, the proposed model
changes could result in less severe credit, market, and operational
loss estimates--which would be driven by overhauling the wholesale
corporate probability of default model and discontinuing the
macroeconomic regression approach for operational risk loss estimation,
as described further in the Corporate Model and Operational Risk Model
descriptions. However, the proposed changes to the Pre-provision Net
Revenue Model would offset these loss reductions. By reducing the
reliance of net revenue projections on recent outcomes and relying more
on firm projections of net noninterest income, the projections of net
revenue would be more consistent with a stress scenario and would
better align with firms' projections.
[GRAPHIC] [TIFF OMITTED] TP18NO25.037
Table 3 below provides a separate analysis of estimates of stress
losses across firm types that are subject to the stress capital buffer
requirement. The analysis shows the reduction in hypothetical stress
losses is concentrated at larger firms.
[GRAPHIC] [TIFF OMITTED] TP18NO25.038
IX. Proposed Changes to the Scenario Design Policy Statement
The Board is also proposing to make several changes to the Scenario
Design Policy Statement. While many of these proposed changes are
technical in nature, this section identifies substantive changes and
requests comment on those proposed changes.
Question 33: The Board invites comment on all aspects of the
technical and substantive proposed revisions to the Scenario Design
Policy Statement. What are the advantages and disadvantages of these
proposed changes? What would be the advantages and disadvantages if the
Board were to consider describing the Board's expectations for
additional components of the scenario design framework?
A. Changes to the Background and Overview and Scope Sections
The Board is proposing to make limited changes to the first two
sections of the Scenario Design Policy Statement, which address
background and overview and scope topics, respectively. In the
background section, the Board would clarify that the stress tests
primarily focus on credit risk, operational risk, and market risk. The
inclusion of operational risk in this list helps clarify the Board's
continued focus on designing a supervisory tool that makes a valuable
forward-looking assessment of large financial companies' capital
adequacy under hypothetical economic and financial market conditions.
The Board would also clarify that it expects to provide only two
different sets of macroeconomic scenarios for both the supervisory and
company-run stress tests. These two sets of macroeconomic scenarios are
the baseline and severely adverse scenario. This change would clarify
the quantity of macroeconomic scenarios the Board expects to provide,
consistent with the removal of a separate adverse scenario.\86\
---------------------------------------------------------------------------
\86\ 84 FR 59032, 59061 (Nov. 1, 2019).
---------------------------------------------------------------------------
In the overview and scope section, the Board would make conforming
edits to the description of the organization of the Scenario Design
Policy Statement to reflect the changes discussed earlier in this
proposal.
Question 34: What additional changes, if any, should the Board
consider making to these sections, and why? What would be the
advantages and disadvantages of providing more than two scenarios? What
are the
[[Page 51877]]
advantages and disadvantages of the Board's continued focus on credit,
operational, and market risk?
B. Changes to the Content of the Stress Test Scenarios Section
The Board is proposing to make two general changes to this section,
which describes the Board's expectations for the content of the
published stress test scenarios.
First, as described below, this section would be amended to clarify
that the Board expects to generally publish two different macroeconomic
scenarios: the baseline and severely adverse scenarios. This section
would also be revised to clarify that the Board expects to invite
comment on severely adverse scenarios.
Second, as described in Section IX.H of this Supplementary
Information, the Board is proposing to make certain changes related to
the global market shock component. See Section IX.H of this
Supplementary Information for a discussion of those changes.
Question 35: What additional changes, if any, should the Board
consider making to these sections, and why?
C. Approach for Formulating Macroeconomic Assumptions in the Baseline
Scenario
The Board is proposing to provide additional details describing the
process by which the Board would set the paths of the variables in the
baseline and severely adverse scenarios. In particular, the amendments
reflect that the Board would post on the Board's website a description
of the macroeconomic model utilized to support the construction of the
baseline and severely adverse scenarios in the annual stress test. By
posting a description of this model (the ``macroeconomic model for
stress testing'') on the Board's website, the Board expects to improve
the transparency, public accountability, and predictability around the
Board's scenario design framework, particularly with respect to the
baseline scenario and certain variables in the severely adverse
scenario. The Board recognizes that, while these enhancements are
consistent with the Board's goal of increased transparency in the
supervisory stress test, they may constrain the design of the scenario
paths for some variables to follow those prescribed by the
macroeconomic model for stress testing. Nevertheless, the Board expects
that other aspects of the proposed changes to the Scenario Design
Policy Statement will preserve sufficient flexibility to allow the
Board to adjust the severity of the annual scenario based on relevant
indicators of economic and financial conditions and other emergent
procyclical factors. Importantly, the Board uses these models to
generate paths for the scenario variables only. These models are used
solely for stress testing purposes and the output is not a forecast of
the Board.
Question 36: What are the advantages and disadvantages of adopting
a macroeconomic model for stress testing to guide the selection of
certain variables in the baseline and severely adverse scenarios?
Question 37: What additional changes, if any, should the Board
consider making to this section, and why?
D. Scenario Narrative: Refinement to the Recession Approach
A number of considerations contribute to the Board's formulation of
the severely adverse scenario. As a starting point, the basic approach
adopted by the Board is the recession approach--the notion that the
Board will construct a scenario informed by the historical paths of
macroeconomic and financial market variables across post-war U.S.
recessions. However, different recessions have differed in important
respects, and a simple recreation of a given episode or an average over
all recessions would fail to reproduce important potential stressors to
firms' balance sheets. Hence, in applying the recession approach, the
Board develops a specific narrative characterizing the hypothetical
recession represented by the scenario to help inform the specific paths
for scenario variables. This narrative combined with data are then
modified to account for the Board's stress testing principle of
conservatism alongside other considerations offered by the literature
on stress testing including a goal to develop sufficient severity and
credibility of the scenarios, and a goal to not add sources of
procyclicality to the financial system, as described below.\87\ This
section gives an overview of these considerations and other details,
providing a common structure for the discussion outlined in the guides
for individual variables under this framework, in Section IX.G of this
Supplementary Information.
---------------------------------------------------------------------------
\87\ 12 CFR 252, Appendix B.
---------------------------------------------------------------------------
The Recession Approach
The Board intends to continue to use a recession approach to
develop the severely adverse scenario. Under the recession approach,
the Board expects to specify the future paths of variables to reflect
conditions that characterize post-war U.S. recessions, generating
either a typical or specific recreation of a post-war U.S. recession.
The Board chose this approach in developing past scenarios, and in the
Scenario Design Policy Statement, because it has observed that the
conditions that typically occur in recessions--such as increasing
unemployment, declining asset prices, and contracting loan demand--can
put significant stress on firms' balance sheets. This stress can occur
through a variety of channels, including higher loss provisions due to
increased delinquencies and defaults, losses on trading positions
through sharp moves in market prices, and lower bank income through
reduced loan originations. For these reasons, the Board expects that
the paths of economic and financial variables in the severely adverse
scenario should, at a minimum, resemble the paths of those variables
observed during a recession. The guide for each variable in this
framework reviews the movements of that variable across past recessions
and bases the formulation of its scenario path on that analysis. While
the recession approach provides a starting point for the formulation of
the scenario, recessions are not all the same. The length and depth of
recessions differ, as do the parts of the economy and financial markets
that are most affected, so the Board must include other considerations
in its scenario design.
The Scenario Narrative
Because recessions have differed in cause, character, and
consequence--from oil price shocks and housing slumps to asset-price
busts and pandemics, from short to long, and from mild to moderate to
severe--the Board augments the basic recession approach with an annual
scenario narrative. The annual scenario narrative provides qualitative
direction on how the Board builds that year's severely adverse
scenario.
While some specifics of the narrative may be adjusted on a year-to-
year basis to reflect developments in the macroeconomic and financial
environment, the overall narrative motivating scenario design will be
that of a sharp recession triggered by an adverse shock to financial
markets. Under the proposal, the Board expects that the macroeconomic
scenario used in the Board's annual supervisory severely adverse
scenario will begin with a sudden and significant increase in
uncertainty and associated rapid deterioration in risk appetite that
cause a spike in financial market volatility and a sharp decline in
many U.S. and
[[Page 51878]]
foreign financial assets. The resulting turmoil would disrupt funding
markets and lead to widespread deleveraging, including forced sales of
illiquid assets at fire sale prices by a range of financial firms and
some temporary breakdowns in the typical correlations between financial
asset prices. (Such sharp changes in financial conditions have been
observed previously in response to the outbreak of COVID-19 or regional
wars, the failure or distress of a large financial institution, or
sudden shifts in the economic policies in advanced economies.)
Under the Board's recession approach, the Board expects that,
although financial market functioning returns to normal within a few
months of the initial shock, uncertainty remains high and risk appetite
remains low for an extended period. The sustained flight to quality
would be expected to push down risk-free interest rates but keep credit
conditions tight and financial asset prices depressed for several
quarters. The market dysfunction would cause a contraction in the
supply of credit from other types of financial intermediaries that
would create demands on banks to provide substantial liquidity to
existing customers with formal credit lines. Banks would also make ad
hoc decisions to support customers without formal arrangements when
doing so could lead to lower losses on their existing loans.\88\ This
shift in demand for credit toward banks from other financial
intermediaries would lead to banks' balance sheets remaining constant
even as overall credit demand declines.\89\ This feature of the
scenario is supported by the stress testing principle of
conservatism.\90\ To that end, maintaining higher capital requirements
during periods of economic expansion ensures that stress tested firms
employ sufficient capital to absorb losses and support the economy
during a downturn.
---------------------------------------------------------------------------
\88\ For example, in June 2020 the Federal Financial
Institutions Examination Council issued interagency guidance to bank
examiners stating, ``examiners will not subject a . . . modified
loan to adverse classification solely because the value of the
underlying collateral has declined . . ., provided that the borrower
has ability to repay . . .'' See Interagency COVID-19 Examiner
Guidance, <a href="https://www.federalreserve.gov/newsevents/pressreleases/files/bcreg20200623a1.pdf">https://www.federalreserve.gov/newsevents/pressreleases/files/bcreg20200623a1.pdf</a>.
\89\ Commercial and industrial loans grew 20 percent in 2007 as
credit markets seized at the beginning of the 2007-2009 financial
crisis. See M. Bech & Tara Rice, Profits and Balance Sheet
Developments at U.S. Commercial Banks in 2008, 95 Fed. Rsrv. Bull.
A57-97 (2009), <a href="https://www.federalreserve.gov/pubs/Bulletin/2009/articles/bankprofit/default.htm">https://www.federalreserve.gov/pubs/Bulletin/2009/articles/bankprofit/default.htm</a>. For COVID-19, see H. Ennis & A.
Jarque, Bank Lending in the Time of COVID, Federal Reserve Bank of
Richmond Economic Brief No. 21-05 (Feb. 2021).
\90\ 12 CFR 252, Appendix B.
---------------------------------------------------------------------------
In the scenario, the news from financial markets would cause near-
immediate decisions by consumers to curtail spending and by businesses
to cut payroll and cancel planned investments, leading to a demand-
driven contraction in economic activity putting downward pressure on
inflation. The initial disruption to spending and employment along with
tightening credit conditions would trigger a negative feedback loop
that results in further declines in payrolls, investment, and spending
in subsequent quarters. With businesses shrinking or failing in the
scenario, demand for commercial real estate would decrease
significantly relative to supply, leading to large declines in
commercial property prices. Meanwhile, rising household financial
distress would lead to increased supply of homes for sale and reduced
household formation, which would depress residential real estate
markets.
The financial market dysfunction and deepening recession in the
United States would spill over to its major trading partners, including
the euro area, United Kingdom, Japan, and Developing Asia. Those areas
would experience declines in economic activity commensurate with the
global slowdown running from 2008 to 2010. Consistent with existing
stress testing principles, this scenario assumes that permanent
government stabilization programs (e.g., unemployment insurance) and
monetary policy in the United States and elsewhere would function
normally, but that there would be no extraordinary measures taken by
fiscal or financial authorities to support the economy or financial
markets during this time. The specific implications of this narrative
for scenario variables are detailed in each guide, but the narrative
interacts importantly with the recession approach: financial recessions
often exhibit different properties than other recessions, as they are
often steeper, deeper, and more drawn-out than typical, non-financial
recessions.\91\ Adopting this scenario narrative reflects a principle
of conservatism, and is in line with recommendations from the stress
testing literature, as discussed in Section IX.F of this Supplementary
Information.
---------------------------------------------------------------------------
\91\ See, e.g., C. Reinhart & K. Rogoff, This Time Is Different:
Eight Centuries of Financial Folly (2009).
---------------------------------------------------------------------------
Question 38: The Board invites comment on all aspects of how the
Board designs the scenario narrative in the annual stress test. What
are the advantages and disadvantages of adopting this financial
recession approach? What other approaches, if any, should the Board
consider adopting, and why? What adjustments, if any, to the financial
recession approach should the Board consider adopting, and why?
Adding Salient Risks to the Severely Adverse Scenario
Consistent with the Scenario Design Policy Statement, under this
proposal, the Board expects that the severely adverse scenario would be
developed to reflect the current level of vulnerabilities or risks to
the banking sector that are apparent in relevant indicators of economic
and financial conditions. The Board anticipates that the proposed
guides for certain scenario variables described below provide an
appropriate range of values to design the severely adverse scenario in
most years. The waxing and waning of relevant indicators of economic
and financial conditions will inform the Board's decisions about where
to set the value of those parameters within those ranges for each
variable.
The Board continues to expect that there will be some important
instances when it will be appropriate to augment the recession approach
with salient risks and to set variables values inside of, and in some
cases, outside of the ranges and values provided in the guides in the
Scenario Design Policy Statement. As a result, each year, the Board
will consider particular risks to the financial system and to the
domestic and international macroeconomic outlook identified by its
economists, bank supervisors, and financial market experts. The Board,
using its internal analysis and supervisory information and in
consultation with the Federal Deposit Insurance Corporation and the
Office of the Comptroller of the Currency, will then determine whether
any of those risks appear significantly more elevated than usual or,
conversely, whether risks are unusually low at a particular time, such
that they cannot be appropriately reflected by choosing values within
the ranges of the proposed guides. In those cases, which it expects to
be infrequent, the Board will make appropriate adjustments to the paths
of specific economic variables. These adjustments will not always be
reflected in the general severity of the recession and, thus, all
macroeconomic variables; rather, the adjustments will sometimes apply
to a subset of variables to reflect co-movements in these variables
that are historically less typical.
To assist the public in assessing the use of salient risks in the
scenario, the Board considered the following examples. A stress test
initiated in a
[[Page 51879]]
period of unusually high uncertainty and rapid deterioration in
economic and financial conditions, such as the first quarter of 2009 or
the first quarter of 2020, likely would prove challenging for the
ranges in this proposed framework. In each case, the prevailing
conditions made it plausible that key variables would settle beyond the
range of their previous peak or trough values, on which the guides for
the variables in the severely adverse scenario are calibrated. Although
the unemployment guide remained flexible enough to respond to the spike
in the unemployment rate to nearly 15 percent during the first months
of the COVID-19-related business closures in 2020, the paths of other
variables may have needed to be adjusted more severely if the economy
had not recovered as quickly as it did.
As another example, the Board may become increasingly concerned
about vulnerabilities related to a particular asset class that was
experiencing rapid and persistent price increases supported by
increasingly leveraged investors. Those circumstances existed in the
housing market in the early 2000s and may have tested the credibility
of a guide framework based solely on past performance of home prices,
given that up until then, the price index for homes the Board uses for
stress testing had rarely experienced a decline.\92\
---------------------------------------------------------------------------
\92\ The Board uses the Price Index for Owner-Occupied Real
Estate, Z.1 Release (Financial Accounts of the United States),
Federal Reserve Board (series FL075035243.Q).
---------------------------------------------------------------------------
Sometimes, the salient risk may arise within an asset class. The
Board most recently incorporated this type of salient risk in the 2024
stress test scenario. That year, the Board noted unusually high
vulnerabilities in types of commercial properties that could be most at
risk for a sustained drop in income and asset values due to the
prevalence of remote work.\93\
---------------------------------------------------------------------------
\93\ See Board, 2024 Stress Test Scenarios, ``Additional Key
Features of the Severely Adverse Scenario,'' at 12-13 (Feb. 2024),
<a href="https://www.federalreserve.gov/publications/files/2024-stress-test-scenarios-20240215.pdf">https://www.federalreserve.gov/publications/files/2024-stress-test-scenarios-20240215.pdf</a>.
---------------------------------------------------------------------------
The Board is proposing two changes to its consideration of salient
risks in the severely adverse scenario. First, the Board would remove
paragraph 4.2.4(d) from the Scenario Design Policy Statement. Removing
this paragraph could help improve the transparency of the scenario
design process by limiting the Board's expectations for considering
risks of uncertain significance. While this approach would reduce the
Board's ability to test for emerging and untested risks in the
financial system through the severely adverse scenario, the Board
expects that the remaining components of the Board's supervisory stress
test should be sufficient to establish a credible severely adverse
scenario.
Second, where the Board does consider salient risks in designing
the severely adverse scenario, the Board will endeavor to disclose and
explain the Board's reasoning in the Board's publication of the annual
stress test scenarios, and subsequently adjust those aspects of the
scenario, if necessary, in response to those comments.
Question 39: What are the advantages and disadvantages of the
Board's approach to considering salient risks? What additional or
alternative approaches, if any, should the Board consider for the
consideration of salient risks? What additional or alternative
circumstances should the Board take into account when evaluating
whether to consider salient risks, if any?
E. Changes to Construction of Certain Variables in the Severely Adverse
Scenario
As noted above, the Board finalized changes to the Scenario Design
Policy Statement in 2019 that established a guide that it would use in
setting the size of the maximum change in the unemployment rate and the
timing of its peak. The Scenario Design Policy Statement also
introduced a guide to govern the size of the maximum decline in house
prices in the severely adverse scenario. This proposal maintains those
features of the guides for those two variables, introduces guides that
will be used to set the changes in the values, and the timing of those
changes, for more variables in the severely adverse scenario, and
provides additional context for the path of each variable before it
reaches the maximum change. In addition, the Board is separately
disclosing a specific macroeconomic model that it proposes to use to
translate the paths of certain variables that are set using the
proposed guides into internally consistent projections for the
remaining variables, such as the 3-month Treasury bill rate, GDP,
Disposable Personal Income (DPI), and inflation.
In addition to updating existing guides for the unemployment rate
and house prices, the Board is proposing to establish a guide for each
of the following variables: equity prices; the VIX index; 5-year
Treasury yields; 10-year Treasury yields; BBB corporate bond yields;
mortgage rates; commercial real estate prices; and certain
international scenario values. These include all but one of the
remaining financial market variables typically included in the domestic
severely adverse scenario disclosure each year (the exception being the
3-month Treasury bill rate, as discussed below).
The Board uses guides to inform its determination of the behavior
of these financial market variables in the severely adverse scenario,
rather than model predictions, for several reasons. Although the
parameters of the guides are calibrated based on an analysis of
historical changes in those variables during recessions and the
resulting set of scenario paths typically would be consistent with
historical co-movements in those variables, using explicit forward-
looking models of these variables to determine scenario paths would be
inconsistent with several stress testing principles, such as simplicity
and transparency, as described below.
Under a model-driven approach to determine the paths of these
variables, each model would require the Board to identify, design,
test, explain, and publish additional assumptions, variables, formulas,
and parameters that would drive the results of the model. Models of
financial market variables can be particularly unreliable during
periods of severe stress like the environment envisioned by the
hypothetical severely adverse scenario.\94\ Thus, the model-driven
approach to determining these variables would contrast with the stress
testing principle of using simpler and more transparent approaches,
where appropriate.
---------------------------------------------------------------------------
\94\ T.C. Green & S. Figlewski, Market Risk and Model Risk for a
Financial Institution Writing Options, 54 J. Fin. 1465-99 (Dec.
1999).
---------------------------------------------------------------------------
The Board believes that the guide-based approach also better
achieves the stress testing principle of using a stable process that is
reliably able to capture the impact of economic stress. These simple,
transparent guides also will allow the Board to use its judgment at
times when it is necessary to account for conditions that are plausible
even if they have not been observed previously, consistent with the
stress testing principle of conservatism. Finally, the guides better
preserve the Board's ability to adjust the severity of the stress test
to avoid adding to procyclical forces, when doing so is appropriate and
consistent with fostering financial stability. The Board's judgment
about the appropriateness of the annual stress test scenarios will
reflect changes in the specific risks or vulnerabilities that the
Board, in consultation with the other federal banking agencies,
determines should be considered in the annual stress tests.\95\
---------------------------------------------------------------------------
\95\ See 84 FR 6651, 6656 (Feb. 28, 2019).
---------------------------------------------------------------------------
[[Page 51880]]
The paths for the remaining variables in the domestic scenario--
GDP, DPI, inflation, and the 3-month Treasury rate--will be informed by
the Board's macroeconomic model for stress testing.\96\ In contrast to
the guide-based approach described above for certain variables, the
Board uses a model-driven approach for these remaining variables
because they are particularly suited to model projections that are
simple to produce and explain. As explained in the model documentation
available on the Board's website, that model uses a set of well-studied
longer-run economic relationships that have proven to be useful in a
variety of economic conditions and modeling frameworks. These include
Okun's Law, a Phillips Curve, and an inertial Taylor Rule.\97\ The
Board acknowledges that increasing the predictability of the paths of
scenario variables in this way could reduce the dynamism of the stress
test or incent firms to optimize their portfolios in ways that reduce
capital requirements, perhaps without a commensurate reduction in risk.
However, the guides and the model are constructed to remain flexible
enough to ensure that the Board can adjust the severely adverse
scenario to capture emerging risks and changes in the level of systemic
risk since the previous stress test in a timely fashion. This
flexibility includes the ability to increase scenario severity when
systemic risks may have built up during robust economic expansions or
periods when risk appetite is high or to avoid adding sources of
procyclicality through the stress test. The proposal continues to
ensure that the scenarios maintain a minimum severity level, even when
economic and financial conditions are strained. Setting a floor for the
severity of the scenario is appropriate because risks that built up
during an economic expansion can persist at financial intermediaries
during downturns and because firms that are under stress sometimes take
imprudent risks that they believe will facilitate recovery.\98\
---------------------------------------------------------------------------
\96\ This approach is consistent with how the Board has designed
recent stress test scenarios. See id. at 6659.
\97\ See <a href="https://www.federalreserve.gov/supervisionreg/dfa-stress-tests-2026.htm">https://www.federalreserve.gov/supervisionreg/dfa-stress-tests-2026.htm</a>.
\98\ See J. Peek & E. Rosengren, Unnatural Selection, Perverse
Incentives and the Misallocation of Credit in Japan, 95 Am. Econ.
Rev. 1144-66 (2005).
---------------------------------------------------------------------------
The Board also considered that employing the guides or the
macroeconomic model for stress testing sometimes may reduce the
severity of some aspects of the scenario relative to what the currently
less-constrained scenario design process would achieve, and in other
cases it may result in higher severity for some aspects of the scenario
than might otherwise be the case. The flexibility in the guides should
be sufficient for the Board to account for those eventualities by
choosing offsetting values across multiple guides that create the
appropriate overall severity of the scenario.
Question 40: What are the advantages and disadvantages of using
guides and the macroeconomic model for stress testing to guide the
setting of scenario variables in the severely adverse scenario? What,
if any, alternatives to using a macroeconomic model to set the
projection paths of other variables should the Board consider?
F. Scenario Design Principles Derived from Stress Testing Literature:
Severity, Credibility, and Procyclicality
In designing the guides for the construction of the severely
adverse scenario presented in this framework, the Board is informed by
the stress testing literature, which provides certain principles for
scenario design,\99\ which are also reflected in the Board's Stress
Testing Policy Statement.\100\ First, the literature emphasizes the
need for adequately severe scenarios, even when the economy and
financial system are in a stressed condition--complementing the Board's
principle of conservatism.\101\ Second, the literature offers insights
on how historical data should inform the design of an adequately severe
scenario, augmenting the Board's recession approach. Third, the
literature highlights the need for stress tests to avoid adding to
other sources of procyclicality in the financial system. In explaining
the paths for variables in the severely adverse scenario, the guides
provide specific applications of these principles, while this
introduction provides an overview of their general meaning and
rationale.
---------------------------------------------------------------------------
\99\ Some of the well-known contributions are T. Schuermann,
Stress Testing Banks, 30 International Journal of Forecasting 717-28
(2014) (``Schermann (2014)''); and N. Liang, Well-Designed Stress
Test Scenarios Are Important for Financial Stability, Brookings
Institution Paper (Feb. 2, 2018) (``Liang (2018)''), <a href="https://www.brookings.edu/articles/well-designed-stress-test-scenarios-are-important-for-financial-stability">https://www.brookings.edu/articles/well-designed-stress-test-scenarios-are-important-for-financial-stability</a>.
\100\ See 12 CFR 252, Appendix B.
\101\ Id.
---------------------------------------------------------------------------
The first principle derived from the literature concerns the need
for sufficiently severe scenarios. Plainly, insufficient stress test
severity can lead to adverse outcomes. Inadequately assessed risks lead
to an underassessment of the associated credit losses and capital
needs--the basic source of failures of many financial institutions
during the 2007-2009 financial crisis which the Board's stress tests
are meant to avoid. Frame et al. (2015) provide an in-depth analysis of
how the assessment of risks (or stress test) conducted by the Office of
the Federal Housing Enterprise Oversight (OFHEO) actually contributed
to the failures of the Federal National Mortgage Association (Fannie
Mae) and the Federal Home Loan Mortgage Corporation (Freddie Mac).\102\
Importantly, stress tests must be adequately severe both in good times
and in bad.
---------------------------------------------------------------------------
\102\ See S. Frame, C. Gerardi, & P. Willen, The Failure of
Supervisory Stress Testing: Fannie Mae, Freddie Mac, and OFHEO,
Federal Reserve Bank of Boston Working Paper No. 15-4 (2015),
<a href="https://www.bostonfed.org/publications/research-department-working-paper/2015/the-failure-of-supervisory-stress-testing-fannie-mae-freddie-mac-and-ofheo.aspx">https://www.bostonfed.org/publications/research-department-working-paper/2015/the-failure-of-supervisory-stress-testing-fannie-mae-freddie-mac-and-ofheo.aspx</a>. OFHEO was the federal regulator of the
government-sponsored mortgage agencies, Fannie Mae and Freddie Mac.
---------------------------------------------------------------------------
In the context of stress testing during crises, in particular,
there are additional arguments against insufficient stress test
severity. Schuermann (2014) and Judge (2022) argue that insufficiently
severe stress test scenarios can erode credibility and trust and impede
timely and adequate policy responses to ongoing crisis developments,
thereby exacerbating a downturn.\103\ Bernanke (2013) also highlights
that stress tests in times of crisis should provide anxious investors
with credible information about prospective losses.\104\ This
literature points to the importance of sufficiently severe scenarios
for the health of the financial system, including by maintaining
credibility with the public and financial markets.
---------------------------------------------------------------------------
\103\ K. Judge, ``Stress Testing During Times of War,'' Handbook
of Financial Stress Testing (2022) (``Judge (2022)'').
\104\ B. Bernanke, ``Stress testing banks: What have we
learned?,'' Speech at the ``Maintaining Financial Stability: Holding
a Tiger by the Tail'' Conference (2013) (``Bernanke (2013)''),
<a href="http://www.federalreserve.gov/newsevents/speech/bernanke20130408a.htm">www.federalreserve.gov/newsevents/speech/bernanke20130408a.htm</a>.
---------------------------------------------------------------------------
Further evidence for the importance of sufficiently stressful
scenarios to maintaining public credibility comes from past U.S. stress
tests. For example, the rapid deterioration in the U.S. economy in
early 2009 led to realized unemployment rates that approached the peak
of the unemployment rate path in the severely adverse scenario used for
the Supervisory Capital Assessment Program (SCAP) in 2009.\105\ In
fact, the scenario peak for the unemployment rate hypothesized would
reach only 8.9 percent at the end of 2009, but as of
[[Page 51881]]
March 2009 the unemployment rate measured 8.5 percent and ultimately
the unemployment rate peaked at 10 percent in October of 2009.\106\
Because the results of the SCAP determined the amount of capital that
firms needed to raise in financial markets or through the Treasury's
Capital Assistance Program, a scenario that turned out to be
insufficiently severe could have left some firms undercapitalized and
failed to achieve the goal of stabilizing the financial system.\107\
---------------------------------------------------------------------------
\105\ See, e.g., E. Andrews & E. Dash, ``Government Offers
Details of Bank Stress Test,'' N.Y. Times (Feb. 25, 2009), <a href="https://archive.nytimes.com/www.nytimes.com/indexes/2009/02/26/todayspaper/index.html">https://archive.nytimes.com/www.nytimes.com/indexes/2009/02/26/todayspaper/index.html</a>.
\106\ A similar concern related to insufficient scenario
severity followed the announcement of the European Union's stress
tests in 2018, with the criticism that the assumptions were milder
than conditions in the 2007-2009 financial crisis. See F. Guarascio,
``EU's 2018 Stress Test too Mild, Spared Weaker States--Auditors'',
Reuters (Jul. 10, 2019), https://www.reuters.com/article/business/
eus-2018-bank-stress-test-too-mild-spared-weaker-states-auditors-
idUSKCN1U5113/
#:~:text=The%20auditors%20said%20last%20year's,their%20risk%20rather%
20than%20size.
\107\ An explanation of the synergy between the SCAP and CAP is
available here: Supervisory Capital Assessment Program & Capital
Assistance Program (SCAP and CAP), U.S. Department of the Treasury,
<a href="https://home.treasury.gov/data/troubled-assets-relief-program/bank-investment-programs/scap-and-cap">https://home.treasury.gov/data/troubled-assets-relief-program/bank-investment-programs/scap-and-cap</a>.
---------------------------------------------------------------------------
This example helps demonstrate the importance of the principle of
severity when considering historical data and current conditions in the
construction of an adequately severe scenario. While unemployment rates
are discussed at length in the unemployment guide below, the maximum
level of 8.9 percent specified in the 2009 SCAP, at the time, was well
beyond the level reached in most post-war recessions. At the time the
scenario was issued, a projected increase to 8.9 percent was thus very
severe compared to outcomes over the past quarter century, but
nonetheless proved lower than the actual realized peak in 2009.
That experience reinforces the need for the framework to support
variable paths that exceed levels observed in the historical data.
Choosing a historical scenario has a price--``it does not test for
anything new.'' \108\ While the recession approach dictates that
variable movements follow historical recessions, when current
conditions are already extreme, a credible scenario may replicate
historical recessions in terms of the size of movements previously
observed, leading to levels of variables that may exceed historical
levels. Several of the guides in this framework allow, at times, for
variables to exceed their historical range, either in levels or in
changes, in order to maintain adequate severity.
---------------------------------------------------------------------------
\108\ See Schuermann (2014), supra note 99.
---------------------------------------------------------------------------
Ultimately, no single scenario can account for all potential
contingencies. Therefore, the severely adverse scenario used in the
Board's annual stress test must be sufficiently severe to ensure that
banks will be resilient to a range of alternative and plausible
scenarios that could generate net losses that are of similar
magnitudes.\109\
---------------------------------------------------------------------------
\109\ See Liang (2018), supra note 99.
---------------------------------------------------------------------------
At the same time, the Board recognizes that the severity of the
annual stress tests potentially can have unintended effects on firms'
operations. For instance, the academic literature finds that stress
tests improve financial stability by reducing riskier bank
lending.\110\ Ensuring that firms are appropriately capitalized for the
risks they are taking is a goal of stress testing; however, if those
effects are not well aligned with the true riskiness of a particular
type of loan, then stress tests could unintentionally reduce banks'
credit supply. For instance, some evidence exists that counties in
which stress tested banks had high market share may have experienced a
lower supply of credit to small and young businesses, which are
generally considered riskier than established businesses but can
generate a disproportionate share of growth in employment and
income.\111\ However, other research concludes that businesses largely
offset the reduction in loans from banks that participate in the stress
tests with other sources of credit. Those sources include loans from
smaller banks not in the stress tests,\112\ debt issuance in capital
markets, or loans from nonbank financial institutions.\113\ Moreover,
these potential unintended effects on credit supply by stress tested
firms must be weighed against the benefits, discussed above, that more
credible stress tests bring to the economy and the financial system. By
ensuring that firms have sufficient quantity and quality of loss-
absorbing capital to cover the risks that they are taking, the stress
tests ensure the resilience and stability of the banking sector even in
circumstances when stresses take unexpected forms.
---------------------------------------------------------------------------
\110\ V. Acharya, A. Berger, & R. Roman, Lending implications of
U.S. bank stress tests: Costs or benefits?, 34 J. Fin.
Intermediation 58-90 (2018).
\111\ See S. Doerr, Stress Tests, Entrepreneurship, and
Innovation, 25 Rev. of Fin. 1609-1637 (Sep. 2021), <a href="https://doi.org/10.1093/rof/rfab007">https://doi.org/10.1093/rof/rfab007</a>.
\112\ See K. Cort[eacute]s et al., Stress tests and small
business lending, 136 J. Fin. Econ. 260-279 (2021) (``Cort[eacute]s
(2021)'').
\113\ See J. Berrospide & R. Edge, Bank capital buffers and
lending, firm financing and spending: What can be learned from five
years of stress test results?, 57 J. Fin. Intermediation 1010-61
(2024) (``Berrospide (2024)''); T. Davydiuk, T. Marchuk, & S. Rosen,
Direct lenders in the U.S. middle market, 162 J. Fin. Econ. (2024)
103946 (``Davydiuk (2024)'').
---------------------------------------------------------------------------
The balance of those advantages and disadvantages of scenario
severity can change over time. Losses at financial institutions are
more likely to arise when the economy slows. Profits are more robust
during periods of economic growth, in turn increasing resources
available to cover future losses. In other words, capital is naturally
procyclical, having an underlying tendency towards a positive
correlation with financial conditions. Moreover, when underlying
conditions are favorable and firm losses are low, firms sometimes
project forward an expectation for low losses, paving the way to take
more risk.\114\ Conversely, when conditions are bad, firms may
overcompensate and restrict credit even to otherwise creditworthy
borrowers, exacerbating the downturn. Thus, firms' behavior may amplify
underlying procyclicality.
---------------------------------------------------------------------------
\114\ See A. Berger & G. Udell, The institutional memory
hypothesis and the procyclicality of bank lending behavior, 13 J.
Fin. Intermediation 458-495 (2004) (``Berger (2004)''); A.
Greenspan, ``Challenges facing community banks,'' Remarks before the
Independent Community Bankers of America (Mar. 8, 2000) (``Greenspan
(2000)''), <a href="https://www.federalreserve.gov/boarddocs/speeches/2000/20000308.htm">https://www.federalreserve.gov/boarddocs/speeches/2000/20000308.htm</a>.
---------------------------------------------------------------------------
Stress tests could, through different designs, either amplify or
mitigate this procyclicality. If stress tests are always more severe in
bad times, despite an expectation that conditions could soon improve,
then this severity would add undue stress to the financial system,
reducing financial intermediation with negative implications for the
macroeconomy. That said, the purpose of the stress test scenarios is
not to serve as an explicit countercyclical offset to the financial
system, but rather to ensure that the firms are properly capitalized to
withstand severe economic and financial conditions. Hence, the Board
adopts a middle path, seeking to specify the severely adverse scenario
to avoid adding sources of procyclicality to the financial system,
neither explicitly mitigating any existing procyclical tendencies nor
magnifying them. Indeed, Kohn and Liang (2019) argue that the ability
to adjust elements that potent
[…truncated; see source link]Indexed from Federal Register on November 18, 2025.
This is legal information, not legal advice. Laws vary by jurisdiction and change frequently. Always verify current law with official sources and consult a licensed attorney in your jurisdiction for advice on your specific situation.