View original document

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

United States Government Accountability Office

GAO

Report to Congressional Committees

September 2010

TROUBLED ASSET
RELIEF PROGRAM
Bank Stress Test
Offers Lessons as
Regulators Take
Further Actions to
Strengthen
Supervisory Oversight

GAO-10-861

September 2010

TROUBLED ASSET RELIEF PROGRAM
Accountability Integrity Reliability

Highlights

Bank Stress Test Offers Lessons as Regulators Take
Further Actions to Strengthen Supervisory Oversight

Highlights of GAO-10-861, a report to
congressional committees

Why GAO Did This Study

What GAO Found

The Supervisory Capital
Assessment Program (SCAP) was
established under the Capital
Assistance Program (CAP)—a
component of the Troubled Asset
Relief Program (TARP)—to assess
whether the 19 largest U.S. bank
holding companies (BHC) had
enough capital to withstand a
severe economic downturn. Led by
the Board of Governors of the
Federal Reserve System (Federal
Reserve), federal bank regulators
conducted a stress test to
determine if these banks needed to
raise additional capital, either
privately or through CAP. This
report (1) describes the SCAP
process and participants’ views of
the process, (2) assesses SCAP’s
goals and results and BHCs’
performance, and (3) identifies
how regulators and the BHCs are
applying lessons learned from
SCAP. To do this work, GAO
reviewed SCAP documents,
analyzed financial data, and
interviewed regulatory, industry,
and BHC officials.

The SCAP process appeared to have been mostly successful in promoting
coordination, transparency, and capital adequacy. The process utilized an
organizational structure that facilitated coordination and communication
among regulatory staff from multiple disciplines and organizations and with
the BHCs. Because SCAP was designed to help restore confidence in the
banking industry, regulators took unusual steps to increase transparency by
releasing details of their methodology and sensitive BHC-specific results.
However, several participants criticized aspects of the SCAP process. For
example, some supervisory and bank industry officials stated that the Federal
Reserve was not transparent about the linkages between some of the test’s
assumptions and results. But most of the participants in SCAP agreed that
despite these views, coordination and communication were effective and
could serve as a model for future supervisory efforts. According to regulators,
the process resulted in a methodology that yielded credible results. By design,
the process helped to ensure that BHCs would be capitalized for a potentially
more severe downturn in economic conditions from 2009 through 2010.

What GAO Recommends
This report recommends that the
Federal Reserve complete a final 2year SCAP analysis, and apply
lessons learned from SCAP to
improve transparency of bank
supervision, examiner guidance,
risk identification and assessment,
and regulatory coordination. The
Federal Reserve agreed with our
five recommendations and noted
current actions that it has
underway to address them.
Treasury agreed with the report’s
findings.

SCAP largely met its goals of increasing the level and quality of capital held by
the 19 largest U.S. BHCs and, more broadly, strengthening market confidence
in the banking system. The stress test identified 9 BHCs that met the capital
requirements under the more adverse scenario and 10 that needed to raise
additional capital. Nine of the 10 BHCs were able to raise capital in the private
market, with the exception of GMAC LLC, which received additional capital
from the U.S. Department of the Treasury (Treasury). The resulting capital
adequacy of the 19 BHCs has generally exceeded SCAP’s requirements, and
two-thirds of the BHCs have either fully repaid or begun to repay their TARP
investments. Officials from the BHCs, credit rating agencies, and federal
banking agencies indicated that the Federal Reserve’s public release of the
stress test methodology and results in the spring of 2009 helped strengthen
market confidence. During the first year of SCAP (2009), overall actual losses
for these 19 BHCs have generally been below GAO’s 1-year pro rata loss
estimates under the more adverse economic scenario. Collectively, the BHCs
experienced gains in their securities and trading and counterparty portfolios.
However, some BHCs exceeded the GAO 1-year pro rata estimated 2009 losses
in certain areas, such as consumer and commercial lending. Most notably, in
2009, GMAC LLC exceeded the loss estimates in multiple categories for the
full 2-year SCAP period. More losses in the residential and commercial real
estate markets and further deterioration in economic conditions could
challenge the BHCs, even though they have been deemed to have adequate
capital levels under SCAP.

View GAO-10-861or key components.
For more information, contact Orice Williams
Brown at (202) 512-8678 or
williamso@gao.gov.

United States Government Accountability Office

Highlights of GAO-10-861 (continued)

SCAP provided a number of important lessons for
regulators about the benefits of increased transparency,
the need for regulators to strengthen bank supervision,
the need for regulators and BHCs to improve their risk
identification and assessment practices, and the need
for regulators to improve coordination and
communication. First, SCAP underscored the potential
benefits that increased transparency about the financial
health of the nation’s largest BHCs can provide. Many
experts have said that the lack of transparency about
potential losses from certain assets contributed
significantly to the instability in financial markets
during the current crisis. But transparency in the
banking supervisory process is a controversial issue.
Some observers say that publicly disclosing sensitive
bank information without a federal capital backstop
could have unintended negative effects, such as runs on
banks, that would disproportionately affect weaker
banks. However, other observers believe that more
transparency about banks’ asset valuations and losses
could help the public better understand the risk
exposures of BHCs, increase market discipline, and
improve the oversight of these institutions. A final
analysis by the Federal Reserve of BHCs’ performance
during the full 2-year SCAP period can help in this
regard. The Federal Reserve and other banking
regulators could benefit from developing a plan to
improve the transparency of bank supervision. Second,

SCAP showed that more robust regulatory oversight of
bank stress tests was necessary to better understand
banks’ capacity to withstand downturns in the
economy. Regulators and BHC officials commented
that internal bank stress tests prior to SCAP did not
comprehensively stress their portfolios. The Federal
Reserve is finalizing examiner guidance for assessing
capital adequacy, including stress testing, but it has not
established criteria for assessing the rigor of the BHCs’
stress test assumptions. Without more robust guidance,
ensuring that stress tests are being evaluated
thoroughly and consistently is difficult. Third, the SCAP
exercise highlighted opportunities to enhance both the
process and data inputs for conducting future stress
tests. The Federal Reserve has started to build a plan to
enhance its risk identification and assessment
infrastructure in response to the financial crisis, but
further planning is needed to reflect recent changes
under the Dodd-Frank Wall Street Reform and
Consumer Protection Act of 2010. Finally, SCAP
demonstrated the need for robust coordination and
communication among regulators in examining
complex institutions. While SCAP promoted
coordination and communication, further efforts are
needed to ensure the participation of relevant
regulators in multiagency examinations of banks.

Table 1: Indicative Loss Rates Estimates and Actual SCAP BHCs and Banking Industry Average Loss Rates, December 31, 2009
Percentage

Loan category

SCAP indicative loss rate estimates
GAO’s more adverse 1Federal Reserve’s more
a
b
year pro rata loss rate
adverse 2-year loss rate

First-lien mortgage

2009 actual loss rates
SCAP BHCs average
Banking industry
c
loss rate
average loss rate

7-8.5%

Credit cards

1.7%

4.4

3.9

2.5-4

2.5

2.3

9-12

Commercial real estate

1.9%

6-8

5-8

Commercial and industrial

3.5-4.25%

12-16

Second/junior lien mortgages

4.5-6

2.3

2.4
10.2

18-20

9-10

10.1

Other consumer

8-12

4-6

4.1

4.4

Other loans

4-10

2-5

1.4

1.1

Sources: Federal Reserve SCAP results report and GAO analysis of SNLFinancial Y-9C regulatory data.
a

Data as of December 31, 2010.

b

GAO calculated the more adverse 1-year pro rata loss rate by dividing the SCAP more
adverse 2-year loss rates by 2. A key limitation of this approach is that it assumes equal
distribution of losses, revenues, expenses, and changes to reserves over time, although these
items were unlikely to be distributed evenly over the 2-year period. Another important
consideration is that actual results were not intended and should not be expected to align with
the SCAP projections.

c

Data are for BHCs with greater than $1 billion in total assets.

United States Government Accountability Office

Contents

Letter

1
Background
SCAP Process Generally Viewed as Promoting Coordination,
Transparency, and Capital Adequacy
While SCAP Increased Capital Levels and Improved Confidence in
the Banking System, BHCs Could Face Ongoing Challenges
SCAP Provided Lessons That Could Help Regulators Strengthen
Supervisory Oversight and BHCs Improve Risk Management
Practices
Conclusions
Recommendations
Agency Comments and Our Evaluation

40
53
55
56

Appendix I

Objectives, Scope, and Methodology

61

Appendix II

Status of Bank Holding Companies’ TARP
Investments as September 22, 2010

75

One-Year Actual Performance Compared to GAO’s
Pro rata Stress Test Loss Projections for Each of
the 19 SCAP BHCs

78

Appendix III

Appendix IV

Appendix V

Appendix VI

4
9
19

Comments from the Board of Governors of the
Federal Reserve System

118

Comments from the Department of the Treasury’s
Office of Financial Stability

121

GAO Contact and Staff Acknowledgments

122

Page i

GAO-10-861 Lessons Learned from Bank Stress Test

Tables
Table 1: Indicative Loss Rates Estimates and Actual SCAP BHCs
and Banking Industry Average Loss Rates, December 31,
2009
Table 2: Summary of Capital Raised by 10 BHCs to Meet Their
SCAP Capital Buffer Amount, as of November 9, 2009
Table 3: Capital Measures for SCAP BHCs, December 31, 2008 and
December 31, 2009
Table 4: Percentage Change in Tier I Capital Ratios, December 31,
2008, and December 31, 2009
Table 5: Actual and GAO Pro Rata Estimates of Aggregate Losses
and Changes in Resources Other than Capital to Absorb
Losses across the 19 SCAP BHCs, December 31, 2009
Table 6: Items Used to Calculate Tier 1 Capital, Asset Losses,
PPNR, and ALLL
Table 7: Tier 1 Common Capital Calculation
Table 8: Crosswalk of Y-9C Net Charge-Offs and Asset
Classifications to Classifications Used by SCAP
Table 9: Crosswalk of Y-9C Loans and Lease Financing Receivables
to and Classifications used by SCAP
Table 10: Status of TARP Investments for the 19 BHCs
Participating in SCAP, as of September 22, 2010
Table 11: Identification of 19 BHCs Subject to the Stress Test
Table 12: American Express Company
Table 13: Bank of America Corporation
Table 14: BB&T Corporation
Table 15: The Bank of New York Mellon Corporation
Table 16: Capital One Financial Corporation
Table 17: Citigroup Inc.
Table 18: Fifth Third Bancorp
Table 19: GMAC LLC
Table 20: The Goldman Sachs Group, Inc.
Table 21: JPMorgan Chase & Co.
Table 22: KeyCorp
Table 23: MetLife, Inc.
Table 24: Morgan Stanley
Table 25: PNC Financial Services Group, Inc.
Table 26: Regions Financial Corporation
Table 27: State Street Corporation
Table 28: SunTrust Banks, Inc.
Table 29: U.S. Bancorp
Table 30: Wells Fargo & Company

Page ii

17
20
21
22

27
66
67
68
72
76
78
80
82
84
86
88
90
92
94
96
98
100
102
104
106
108
110
112
114
116

GAO-10-861 Lessons Learned from Bank Stress Test

Figures
Figure 1: Timeline of Key Activities Regarding Implementation of
SCAP, February 10, 2009, through December 31, 2009
Figure 2: Commercial Bank 2-Year Loan Loss Rates from 1921
through 2013 Compared to SCAP Loan Loss Rate
Figure 3: Actual Economic Performance to Date Versus SCAP More
Adverse Assumptions
Figure 4: Gross Common Equity Issuance by Banks and Thrifts,
2000 to First Quarter 2010
Figure 5: Stock Market Prices, October 2007 through March 2010
Figure 6: Bank Credit Default Swap Spreads, January 2007 through
March 2010
Figure 7: Comparison of Actual and GAO Pro Rata Estimated
Losses for Consumer and Commercial Loans, December
31, 2009
Figure 8: Comparison of Actual and GAO Pro Rata Estimated Gains
and Losses for Securities Available for Sale and Held to
Maturity, December 31, 2009
Figure 9: Comparison of Actual and GAO Pro Rata Estimated Gains
and Losses for Trading and Counterparty, December 31,
2009
Figure 10: Change in the Percentage of Nonperforming Loans for
Applicable SCAP BHCs, by Loan Type, First Quarter 2007
through Fourth Quarter 2009
Figure 11: Comparison of Actual and GAO Pro Rata Estimated
Resources Other Than Capital to Absorb Losses,
December 31, 2009

Page iii

5
16
18
24
25
26

30

34

36

38

39

GAO-10-861 Lessons Learned from Bank Stress Test

Abbreviations
ALLL
BHC
BB&T
CAP
CES
CPP
DTA
EESA
ESOP
FDIC
FSP
GDP
GMAC
ICAAP
OCC
PPNR
SCAP
TARP
TIP
Y-9C

allowance for loan and lease losses
bank holding company
BB&T Corporation
Capital Assistance Program
common equivalent securities
Capital Purchase Program
deferred tax asset
Emergency Economic Stabilization Act of 2008
Employee Stock Ownership Plan
Federal Deposit Insurance Corporation
Financial Stability Plan
gross domestic product
GMAC LLC
internal capital adequacy assessment process
Office of the Comptroller of the Currency
preprovision net revenue
Supervisory Capital Assessment Program
Troubled Asset Relief Program
Targeted Investment Program
Consolidated Financial Statements for Bank Holding
Companies–FR Y-9C

This is a work of the U.S. government and is not subject to copyright protection in the
United States. The published product may be reproduced and distributed in its entirety
without further permission from GAO. However, because this work may contain
copyrighted images or other material, permission from the copyright holder may be
necessary if you wish to reproduce this material separately.

Page iv

GAO-10-861 Lessons Learned from Bank Stress Test

United States Government Accountability Office
Washington, DC 20548

September 29, 2010
Congressional Committees
The recent financial crisis seriously undermined confidence in the nation’s
financial system and institutions. In February 2009, to help restore
confidence, the U.S. Department of the Treasury (Treasury) announced
the Financial Stability Plan, which established the Supervisory Capital
Assessment Program (SCAP). 1 SCAP, as implemented by the Board of
Governors of the Federal Reserve System (Federal Reserve) and other
federal banking regulators, 2 was to determine through a stress test
whether the largest 19 U.S. bank holding companies (BHC) 3 had enough
capital for the next 2 years (2009-2010) to support their lending activities
and survive a second similar economic shock. 4 As of December 31, 2008,
the largest 19 BHCs accounted for approximately 67 percent of the assets
and more than 50 percent of loans in the U.S. banking system. BHCs that
were found to need additional capital would be allowed, and were
encouraged, to raise the funds privately, but if they could not, Treasury
would provide capital infusions using funding available under the
Troubled Asset Relief Program’s (TARP) Capital Assistance Program

1

Treasury, Financial Stability Plan (Feb. 10, 2009). SCAP was a key component of the
Capital Assistance Program.
2

The other federal banking regulators involved in SCAP were the Federal Deposit Insurance
Corporation and the Office of the Comptroller of the Currency. The Office of Thrift
Supervision did not participate. The Federal Reserve led the SCAP stress test since it is the
primary federal bank regulator for bank holding companies.
3

The 19 BHCs each had at least $100 billion in risk-weighted assets as of December 31,
2008, meeting the established threshold for required participation in the SCAP stress test.
Risk-weighted assets are the total assets and off-balance sheet items, adjusted for risks that
institutions hold. A BHC is a company that owns or controls one or more banks or one that
owns or controls one or more BHCs. See 12 U.S.C. § 1841(a). Since a BHC may also own
another BHC, which in turn owns or controls a bank, the company at the top of the
ownership chain is commonly called the top holder. The Federal Reserve is responsible for
regulating and supervising BHCs, even if the bank owned by the holding company is under
the primary supervision of a different federal banking agency. For example, the Federal
Reserve is responsible for regulating and supervising Citigroup Inc. (the BHC) and the
Office of the Comptroller of the Currency is responsible for regulating and supervising
Citibank N.A. (the main bank in the holding company structure).
4

Capital is a source of long-term funding, contributed largely by a bank’s equity
stockholders and its own returns in the form of retained earnings that provides banks with
a cushion to absorb unexpected losses. A stress test is a “what-if” scenario that is not a
prediction or expected outcome of the economy.

Page 1

GAO-10-861 Lessons Learned from Bank Stress Test

(CAP). 5 However, Treasury made no investments under CAP and
terminated the program in November 2009. When SCAP was first
announced in February 2009, and again around the time the Federal
Reserve released the results of the stress test in May 2009, some
academics, market participants, and others raised concerns about the test,
noting that the assumptions used in the more adverse economic scenario
were not severe enough and that the test did not account for differences in
institutions’ business models.
As part of GAO’s continued analysis and monitoring of Treasury’s process
for implementing the Emergency Economic Stabilization Act of 2008, 6 this
report on the stress test expands on SCAP activities that we reported on in
June 2009. 7 Specifically, this report (1) describes the process used to
design and conduct the stress test and participants’ views on the process,
(2) describes the extent to which the stress test achieved its goals and
compares its estimates with the BHCs’ actual results, and (3) identifies the
lessons regulators and BHCs learned from SCAP and examines how each
are using those lessons to enhance their risk identification and assessment
practices.
To meet the report’s objectives, we reviewed the Federal Reserve’s The
Supervisory Capital Assessment Program: Design and Implementation
(SCAP design and implementation document) dated April 24, 2009, and
The Supervisory Capital Assessment Program: Overview of Results
(SCAP results document) dated May 7, 2009. In addition to the publicly
released BHC-level loss estimates, we analyzed the initial stress test
results that the Federal Reserve provided to each BHC, the subsequent
adjustments the Federal Reserve made to these results, and its reasons for
making them. We also reviewed the BHCs’ quarterly regulatory filings,
such as the Federal Reserve’s 2009 Consolidated Financial Statements for

5

GAO is required to report at least every 60 days on TARP activities and performance.
TARP was authorized under the Emergency Economic Stabilization Act of 2008 (EESA),
Pub. L. No. 110-343, 122 Stat. 3765 (2008), codified at 12 U.S.C. §§ 5201 et seq. EESA
originally authorized Treasury to purchase or guarantee up to $700 billion in troubled
assets. The Helping Families Save Their Homes Act of 2009, Pub. L. No. 111-22, Div. A, 123
Stat. 1632 (2009), codified at 12 U.S.C. § 5225(a)(3), amended ESSA to reduce the
maximum allowable amount of outstanding troubled assets under ESSA by almost $1.3
billion, from $700 billion to $698.741 billion.

6

12 U.S.C. § 5226.

7

GAO, Troubled Asset Relief Program: June 2009 Status of Efforts to Address
Transparency and Accountability Issues, GAO-09-658 (Washington, D.C.: June 2009).

Page 2

GAO-10-861 Lessons Learned from Bank Stress Test

Bank Holding Companies—FR Y-9C (Y-9C); form 10-Qs and annual form
10-Ks; speeches, testimonies, and articles regarding SCAP and stress
testing; and BHC presentations to shareholders and earnings reports. To
more completely understand the execution of SCAP, we completed a
literature search of stress tests that other entities have conducted, such as
the Committee of European Banking Supervisors and the International
Monetary Fund. We also reviewed the Congressional Oversight Panel’s
analysis of SCAP. In addition, we reviewed the capital plans of the 10
BHCs that were required to raise capital to satisfy their SCAP capital
requirement. We collected and analyzed data on the BHCs’ actual
performance from a private financial database of public information and
compared it with the 2-year SCAP estimates and with GAO’s 1-year pro
rata loss estimates for the more adverse scenario (pro rata loss estimate).
GAO calculated the pro rata loss estimates by dividing the SCAP more
adverse 2-year loss estimates by 2. This pro rata estimate methodology has
some limitations, because losses, expenses, revenues, and changes to
reserves are historically unevenly distributed and loss rates over a 2-year
period in an uncertain economic environment can follow an inconsistent
path. However, the Federal Reserve, the Office of the Comptroller of the
Currency (OCC), credit rating agencies, an SNL Financial analyst, and all
of the BHCs we interviewed that are tracking performance relative to
SCAP estimates are also using the same methodology. We obtained
Federal Reserve and BHCs comments on our performance comparison.
Further, we interviewed regulatory and BHC officials to get their views on
the SCAP stress test. Regulatory officials included bank examiners,
economists, and attorneys from the Federal Reserve; the Federal Reserve
district banks; the OCC; the Federal Deposit Insurance Corporation
(FDIC); the Office of Thrift Supervision; and BHC senior officials,
including chief financial officers and chief risk officers, who participated
in the SCAP stress test and were responsible for coordinating and
discussing the results with regulators. These officials represented several
types of BHCs, including traditional, custodial, investment, auto finance,
and credit card institutions. Finally, we met with credit rating agency
officials to get their views on SCAP and understand their own stress
testing practices for banks. For additional information on the scope and
methodology for this engagement, see appendix I.
We conducted this performance audit from August 2009 to September 2010
in accordance with generally accepted government auditing standards.
Those standards require that we plan and perform the audit to obtain
sufficient, appropriate evidence to provide a reasonable basis for our
findings and conclusions based on our audit objectives. We believe that

Page 3

GAO-10-861 Lessons Learned from Bank Stress Test

the evidence obtained provides a reasonable basis for our findings and
conclusions based on our audit objectives.

Background

Despite efforts undertaken by TARP to bolster capital of the largest
financial institutions, market conditions in the beginning of 2009 were
deteriorating and public confidence in the ability of financial institutions
to withstand losses and to continue lending were further declining. On
February 10, 2009, Treasury announced the Financial Stability Plan, which
outlined measures to address the financial crisis and restore confidence in
the U.S. financial and housing markets. The goals of the plan were to (1)
restart the flow of credit to consumers and businesses, (2) strengthen
financial institutions, and (3) provide aid to homeowners and small
businesses. Under SCAP, the stress test would assess the ability of the
largest 19 BHCs to absorb losses if economic conditions deteriorated
further in a hypothetical “more adverse” scenario, characterized by a
sharper and more protracted decline in gross domestic product (GDP)
growth, 8 a steeper drop in home prices, and a larger rise in the
unemployment rate than in a baseline consensus scenario. BHCs that were
found not to meet the SCAP capital buffer requirement under the “more
adverse” scenario would need to provide a satisfactory capital plan to
address any shortfall by raising funds, privately if possible. CAP, which
was a key part of the plan, would provide backup capital to financial
institutions unable to raise funds from private investors. Any of the 19
BHCs that participated in the stress test and had a capital shortfall could
apply for capital from CAP immediately if necessary. 9 The timeline in
figure 1 provides some highlights of key developments in the
implementation of SCAP.

8

Percent change in the annual average of real GDP. GDP is defined as the total market
value of goods and services produced domestically during a given period (i.e., one year).
9

Financial institutions that were not subject to the stress test could, after supervisory
review, also apply for capital from CAP if they were in need of additional capital.

Page 4

GAO-10-861 Lessons Learned from Bank Stress Test

Figure 1: Timeline of Key Activities Regarding Implementation of SCAP, February 10, 2009, through December 31, 2009

Feb. 25:
Treasury announces
the terms and
conditions for CAP.

April 24:
The Federal Reserve
releases information
regarding the design
and implementation of
the stress test.

Feb.

April

May 8:
Wells Fargo &
Company issues $8.6
billion and Morgan
Stanley Inc. issues $8
billion in equity and
debt to meet SCAP
capital requirements
and/or pay back
taxpayer money.

May

June 30:
GMAC LLC, a
Delaware limited
liability company,
was converted into a
Delaware corporation and renamed
GMAC Inc.

Dec. 30:
GMAC Inc. is given $3.8
billion in taxpayer money
under TARP’s Automotive
Industry Financing Program
to meet the level of capital
required by the stress test.

Sept. 30:
All BHCs but GMAC Inc.
have met or exceeded the
capital requirements of the
more adverse scenario.

June

Sept.

Nov.

Dec.

2009
Feb. 10:
Treasury announces
the Financial
Stability Plan (FSP)
to stabilize and
repair the financial
system and support
the flow of credit. A
key component of
FSP is CAP and a
key part of CAP is
SCAP.

Feb. 23:
Treasury announces
SCAP or stress test
of the largest 19
U.S. BHCs.

May 7:
The Federal
Reserve
announces the
results of the
stress test and 10
of the 19 BHCs
are found to need
capital.

June 17:
Nine of the 19 SCAP BHCs
repurchased their preferred stock
from Treasury.
June 8:
The 10 BHCs that required capital under
SCAP submit plans to raise capital.

Nov. 9:
Deadline for raising
capital. GMAC Inc. is the
only BHC unable to raise
the capital as required by
SCAP. Also, Treasury
closes CAP with no
investments having been
made.

May 21:
Treasury announces a $3.5 billion
investment in GMAC LLC to help meet its
stress test capital needs.
Source: GAO.

Note: On May 10, 2010, GMAC Inc. changed its name to Ally Financial Inc.

In a joint statement issued on February 10, 2009, Treasury, along with the
Federal Reserve, FDIC, and OCC (collectively referred to as the SCAP
regulators), committed to design and implement the stress test. According
to a Treasury official, the department generally did not participate in the
design or implementation of SCAP, but was kept informed by the Federal
Reserve during the stress test. The SCAP regulators developed economic
assumptions to estimate the potential impact of further losses on BHCs’
capital under two scenarios. The baseline scenario reflected the consensus
view about the depth and duration of the recession, and the more adverse
scenario reflected a plausible but deeper and longer recession than the
consensus view. Regulators then calculated how much capital, if any, was
required for each BHC to achieve the required SCAP buffer at the end of
2010 under the more adverse scenario.
The SCAP assessment examined tier 1 capital and tier 1 common capital,
and the BHCs were required to raise capital to meet any identified capital
shortfall (either tier 1 capital or tier 1 common capital). Tier 1 risk-based
capital is considered core capital—the most stable and readily available

Page 5

GAO-10-861 Lessons Learned from Bank Stress Test

for supporting a bank’s operations and includes elements such as common
stock and noncumulative perpetual preferred stock. 10 SCAP’s focus on tier
1 common capital, a subset of tier 1 capital, reflects the recent regulatory
push for BHCs to hold a higher quality of capital. 11 The focus on common
equity reflected both the long held view by bank supervisors that common
equity should be the dominant component of tier 1 capital and increased
market scrutiny of common equity ratios, driven in part by deterioration in
common equity during the financial crisis. Common equity offers
protection to more senior parts of the capital structure because it is the
first to absorb losses in the capital structure. Common equity also gives a
BHC greater permanent loss absorption capacity and greater ability to
conserve resources under stress by changing the amount and timing of
dividends and other distributions.
To protect against risks, financial regulators set minimum standards for
the capital that firms are to hold. 12 However, SCAP set a one-time
minimum capital buffer target for BHCs to hold to protect against losses
and preprovision net revenue (PPNR) that were worse than anticipated
during the 2009 to 2010 period. 13 For the purposes of SCAP, the one-time
target capital adequacy ratios are at least 6 percent of risk-weighted assets
in tier 1 capital and at least 4 percent in tier 1 common capital projected as
of December 31, 2010. For the purposes of the projection, the regulators
assumed that BHCs would suffer the estimated losses and earned revenues
in 2009 and 2010 in the more adverse scenario. SCAP regulators conducted
the stress test strictly on the BHCs’ assets as of December 31, 2008, 14
and—with the exception of off-balance sheet positions subject to

10

Common stock is a security that represents ownership in a company and gives the
stockholder the right to vote for the company’s board of directors and benefit from its
financial success. Noncumulative perpetual preferred stock is a security that has no fixed
maturity date and pays its stated dividend forever or “in perpetuity,” but any unpaid
dividends do not accumulate or accrue to stockholders.

11

In general, tier 1 common capital is voting common equity subject to certain deductions
from capital.
12
For example, to be well-capitalized under Federal Reserve definitions, on a consolidated
basis, a BHC must have a tier 1 risk-based capital ratio of at least 6 percent of total riskweighted assets, among other things, 12 C.F.R. § 225.2(r)(1)(ii).
13

PPNR is defined as net interest income plus noninterest income minus noninterest
expense.

14

Trading book positions and counterparty exposures were stress tested as of February 20,
2009.

Page 6

GAO-10-861 Lessons Learned from Bank Stress Test

Financial Accounting Statements No. 166 and 167, which assumed in the
analysis to come on balance sheet as of January 1, 2010—did not take into
account any changes in the composition of their balance sheets over the 2year time frame. 15
Stress testing is one of many risk management tools used by both BHCs
and regulators. Complex financial institutions need management
information systems that can help firms to identify, assess, and manage a
full range of risks across the whole organization arising from both internal
and external sources and from assets and obligations that are found both
on and off the BHC’s balance sheet. This approach is intended to help
ensure that a firmwide approach to managing risk has been viewed as
being crucial for responding to rapid and unanticipated changes in
financial markets. Risk management also depends on an effective
corporate governance system that addresses risk across the institution and
also within specific areas, such as subprime mortgage lending. 16 The board
of directors, senior management, audit committee, internal auditors,
external auditors, and others play important roles in effectively operating
a risk management system. The different roles of each of these groups
represent critical checks and balances in the overall risk management
system. However, the management information systems at many financial
institutions have been called into question since the financial crisis began
in 2007. Identified shortcomings, such as lack of firmwide stress testing,
have led banking organizations and their regulators to reassess capital

15

These statements became effective on January 1, 2010, and require banking organizations
to bring onto their balance sheets off-balance sheet positions. However, for regulatory
purposes, the BHCs and other institutions may defer bringing such positions onto their
balance sheets until the end of 2010.

16

Subprime mortgages are mortgages granted to borrowers whose credit history includes
significant impairments resulting in lower credit scores.

Page 7

GAO-10-861 Lessons Learned from Bank Stress Test

requirements, risk management practices, and other aspects of bank
regulation and supervision. 17
Stress testing has been used throughout the financial industry for more
than 10 years, but has recently evolved as a risk management tool in
response to the urgency of the financial crisis. The main evolution is
towards the use of comprehensive firmwide stress testing as an integral
and critical part of firms’ internal capital adequacy assessment processes.
In the case of SCAP, the intent of the stress test was to help ensure that
the capital held by a BHC is sufficient to withstand a plausible adverse
economic environment over the 2-year time frame ending December 31,
2010. The Basel Committee on Banking Supervision (Basel Committee)
issued a document in May 2009 outlining several principles for sound
stress testing practices and supervision. 18 The Basel Committee document
endorses stress testing by banks as a part of their internal risk
management to assess the following:
•

Credit risk. The potential for financial losses resulting from the failure of
a borrower or counterparty to perform on an obligation.

•

Market risk. The potential for financial losses due to an increase or
decrease in the value of an asset or liability resulting from broad price
movements; for example, in interest rates, commodity prices, stock prices,
or the relative value of currencies (foreign exchange).

17
For a more detailed discussion about risk management practices in place during the
market turmoil, see the following reports: Senior Supervisors Group, Observations on Risk
Management Practices during the Recent Market Turbulence (New York, Mar. 6, 2008);
The President’s Working Group on Financial Markets, Policy Statement on Financial
Market Developments (March 2008); International Monetary Fund, Global Financial
Stability Report: Containing Systemic Risk and Restoring Financial Soundness
(Washington, D.C., April 2008); Financial Stability Forum, Report of the Financial Stability
Forum on Enhancing Market and Institutional Resilience (April 2008); Institute of
International Finance, Final Report of the IIF Committee on Market Best Practices:
Principles of Conduct and Best Practice Recommendations (July 2008); Credit Risk
Management Policy Group III, Containing Systemic Risk: The Road to Reform (August
2008); and Senior Supervisors Group, Risk Management Lessons from the Global Banking
Crisis of 2008 (Oct. 21, 2009).
18

The Basel Committee seeks to improve the quality of banking supervision worldwide, in
part by developing broad supervisory standards. The Basel Committee consists of central
bank and regulatory officials from 27 member countries. The Basel Committee’s
supervisory standards are also often adopted by nonmember countries. See Basel
Committee on Banking Supervision, Principles for Sound Stress Testing Practices and
Supervision. (Basel, Switzerland, May 2009).

Page 8

GAO-10-861 Lessons Learned from Bank Stress Test

•

Liquidity risk. The potential for financial losses due to an institution’s
failure to meet its obligations because it cannot liquidate assets or obtain
adequate funding.

•

Operational risk. The potential for unexpected financial losses due to a
wide variety of institutional factors including inadequate information
systems, operational problems, breaches in internal controls, or fraud.

•

Legal risk. The potential for financial losses due to breaches of law or
regulation that may result in heavy penalties or other costs.

•

Compliance risk. The potential for loss arising from violations of laws or
regulations or nonconformance with internal policies or ethical standards.

•

Strategic risk. The potential for loss arising from adverse business
decisions or improper implementation of decisions.

•

Reputational risk. The potential for loss arising from negative publicity
regarding an institution’s business practices.

SCAP Process
Generally Viewed as
Promoting
Coordination,
Transparency, and
Capital Adequacy

According to SCAP regulators and many market participants we
interviewed, the process used to design and implement SCAP was
effective in promoting coordination and transparency among the
regulators and participating BHCs, but some SCAP participants we
interviewed expressed concerns about the process. The majority of
supervisory and bank industry officials we interviewed stated that they
were satisfied with how SCAP was implemented, especially considering
the stress test’s unprecedented nature, limited time frame, and the
uncertainty in the economy. SCAP established a process for (1)
coordinating and communicating among the regulators and with the BHCs
and (2) promoting transparency of the stress test to the public. In addition,
according to regulators, the process resulted in a methodology that yielded
credible results and by design helped to assure that the BHCs would be
sufficiently capitalized to weather a more adverse economic downturn.

SCAP Process Included
Coordination and
Communication among the
Federal Bank Regulators
and with the BHCs

Robust coordination and communication are essential to programs like
SCAP when bringing together regulatory staff from multiple agencies and
disciplines to effectively analyze complex financial institutions and
understand the interactions among multiple layers of risk. Moreover,
supervisory guidance emphasizes the importance of coordination and
communication among regulators to both effectively assess banks and

Page 9

GAO-10-861 Lessons Learned from Bank Stress Test

conduct coordinated supervisory reviews across a group of peer
institutions, referred to as “horizontal examinations.”
The regulators implemented each phase of SCAP in a coordinated
interagency fashion. Also, while some disagreed, most regulators and
market participants we interviewed were satisfied with the level of
coordination and communication. They also thought that the SCAP
process could serve as a model for future supervisory efforts. The
regulators executed the SCAP process in three broad phases:
•

In the first phase, the Analytical Group, comprising interagency
economists and supervisors, generated two sets of economic conditions—
a baseline scenario and a more adverse scenario with a worse-thanexpected economic outcome—and then used these scenarios to aid in
estimating industrywide indicative loan loss rates. To develop these
scenarios, the Analytical Group used three primary indicators of economic
health: the U.S. GDP, housing prices in 10 key U.S. cities, 19 and the annual
average U.S. unemployment rate. 20 The baseline scenario reflected the
consensus view of the course for the economy as of February 2009,
according to well-known professional economic forecasters. 21 The Federal
Reserve developed the more adverse scenario from the baseline scenario
by taking into account the historical accuracy of the forecasts for
unemployment and the GDP and the uncertainty of the economic outlook
at that time by professional forecasters. The Federal Reserve also used
regulators’ judgment about the appropriate severity of assumed additional
stresses against which BHCs would be required to hold a capital buffer,
given that the economy was already in a recession at the initiation of
SCAP.

•

In the second phase, several Supervisory Analytical and Advisory Teams—
comprising interagency senior examiners, economists, accountants,
lawyers, financial analysts, and other professionals from the SCAP
regulators—collected, verified, and analyzed each BHC’s estimates for

19

Regulators used the Case-Shiller 10-City Composite index to forecast changes in housing
prices.
20

The unemployment rate is the number of jobless people who are available for work but
not currently employed and are actively seeking jobs, expressed as a percentage of the
labor force.

21

According to the Federal Reserve’s SCAP design and implementation document, the
professionals are the Consensus Forecasts, Blue Chip survey, and Survey of Professional
Forecasters.

Page 10

GAO-10-861 Lessons Learned from Bank Stress Test

losses, PPNR, and allowance for loan and lease losses (ALLL). 22 The teams
also collected additional data to evaluate the BHC’s estimates, and to
allow supervisors to develop their own independent estimates of losses for
loans, trading assets, counterparty credit risk, and securities and PPNR for
each BHC.
•

In the third phase, the Capital Assessment Group, comprising interagency
staff, served as the informal decision-making body for SCAP. The Capital
Assessment Group developed a framework for combing the Supervisory
Analytical and Advisory Teams’ estimates with other independent
supervisory estimates of loan losses and resources available to absorb
these losses. 23 They evaluated the estimates by comparing across BHCs
and by aggregating over the 19 BHCs to check for consistency with the
specified macroeconomic scenarios to calculate the amount, if any, of
additional capital needed for each BHC to achieve the SCAP buffer target
capital ratios as of December 31, 2010, in the more adverse economic
environment. Lastly, the Capital Assessment Group set two deadlines: (1)
June 8, 2009, for BHCs requiring capital to develop and submit a capital
plan to the Federal Reserve on how they would meet their SCAP capital
shortfall and (2) November 9, 2009, for these BHCs to raise the required
capital.
A key component of this process was the involvement of multidisciplinary
interagency teams that leveraged the skills and experiences of staff from
different disciplines and agencies. The Federal Reserve, OCC, and FDIC
had representatives on each SCAP team (the Analytical Group,
Supervisory Analytical and Advisory Teams, and the Capital Assessment
Group). For example, OCC officials said that they contributed to the
development of quantitative models required for the implementation of
SCAP and offered their own models for use in assessing the loss rates of
certain portfolios. In addition, each of the SCAP regulators tapped
expertise within their organization for specific disciplines, such as
accounting, custodial banking, macroeconomics, commercial and industry
loan loss modeling, and consumer risk modeling. According to the FDIC,
the broad involvement of experts from across the agencies helped validate
loss assumptions and also helped improve confidence in the results.
Further, these officials noted that the SCAP process was enhanced

22

ALLL is the capital reserve set aside to cover anticipated losses.

23

Resources available to absorb losses is defined as PPNR less the change in ALLL from
December 31, 2008, to December 31, 2010.

Page 11

GAO-10-861 Lessons Learned from Bank Stress Test

because productive debate became a common event as team members
from different regulatory agencies and disciplines brought their own
perspectives and ideas to the process. For example, some SCAP staff
argued for a more moderate treatment of securities in BHCs’ available for
sale portfolios, which would have been consistent with generally accepted
accounting principles under a new change in accounting standards. 24 They
maintained that the modified accounting standard for declines in market
value (and discounting the impact of liquidity premia) that had been
implemented after the stress test was announced and before the numbers
had been finalized was in some ways more reflective of the realized credit
loss expectations for the affected securities. After significant discussion,
the regulators decided to allow for the accounting change in the baseline
loss estimates, but not in the more adverse scenario estimates. They
believed that under the more adverse scenario there was a heightened
possibility of increased liquidity demands on banks and that many
distressed securities would need to be liquidated at distressed levels.
Consequently, for securities found to be other than temporarily impaired
in the more adverse scenario, they assumed the firm would have to realize
all unrealized losses (i.e., write down the value of the security to market
value as of year end 2008). 25 Similarly, some staff argued against adopting
other changes in accounting standards that were expected to impact
BHCs’ balance sheets, including their capital adequacy. Primary among
these was the inclusion of previously off-balance sheet items. 26 As noted
above, ultimately, the more conservative approach prevailed and the
expected inclusion of these assets was addressed in SCAP.
To facilitate coordination, the Federal Reserve instituted a voting system
to resolve any contentious issues, but in practice differences among
regulators were generally resolved through consensus. When SCAP
regulators met, the Federal Reserve led the discussions and solicited input
from other regulators. For example, officials from OCC and FDIC both
told us that they felt that they were adequately involved in tailoring the

24

Financial Accounting Standards Board position numbers 115-2 and 124-2 focus on
whether firms with debt securities held in the available for sale and held to maturity
accounts intended or would be required to sell securities at a lower price than its cost
basis. Generally accepted accounting principles are a widely accepted set of rules,
conventions, standards, and procedures for reporting financial information established by
the Financial Accounting Standards Board.

25

Other than temporarily impaired write down is measured as the difference between a
security’s book value and market value.
26

See Financial Accounting Statements No. 166 and 167.

Page 12

GAO-10-861 Lessons Learned from Bank Stress Test

aggregate loss estimates to each BHC as part of the determination of each
BHC’s SCAP capital requirement. SCAP regulators were also involved in
drafting the design and results documents, which were publicly released
by the Federal Reserve.
Representatives from most of the BHCs were satisfied with the SCAP
regulators’ coordination and communication. Many of the BHC officials
stated that they were generally impressed with the onsite SCAP teams and
said that these teams improved the BHCs’ coordination and
communication with the regulators. BHC officials said that they usually
received answers to their questions in a timely manner, either during
conference calls held three times a week, through the distribution of
answers to frequently asked questions, or from onsite SCAP examiners.
Collecting and aggregating data were among the most difficult and timeconsuming tasks for BHCs, but most of them stated that the nature of the
SCAP’s requests were clear. At the conclusion of SCAP, the regulators
presented the results to each of the institutions showing the final numbers
that they planned to publish.

Market Participants
Generally Agreed that the
SCAP Process Was
Transparent

The SCAP process included steps to promote transparency, such as the
release of key program information to SCAP BHCs and the public.
According to SCAP regulators, BHCs, and credit rating agency officials we
interviewed, the release of the results provided specific information on the
financial health and viability of the 19 largest BHCs regarding their ability
to withstand additional losses during a time of significant uncertainty.
Many experts have said that the lack of transparency about potential
losses from certain assets contributed significantly to the instability in
financial markets during the current crisis. Such officials also stated that
publicly releasing the methodology and results of the stress test helped
strengthen market confidence. Further, many market observers have
commented that the Federal Reserve’s unprecedented disclosure of
sensitive supervisory information for each BHC helped European bank
regulators decide to publicly release detailed results of their own stress
tests in July 2010.
Not all SCAP participants agreed that the SCAP process was fully
transparent. For example, some participants questioned the transparency
of certain assumptions used in developing the stress test. According to
BHC officials and one regulator, the Federal Reserve could have shared

Page 13

GAO-10-861 Lessons Learned from Bank Stress Test

more detailed information about SCAP loss assumptions and calculations
with BHCs. 27 According to several BHC officials, the Federal Reserve did
not fully explain the methodology for estimating losses but expected BHC
officials to fully document and provide supporting data for all of their
assumptions. Without knowing the details of the methodology, according
to some BHC officials, they could not efficiently provide all relevant
information to SCAP examiners.

SCAP Was Designed to
Help Ensure That BHCs
Were Adequately
Capitalized Under the
More Adverse Economic
Scenario

SCAP regulators aimed to ensure that SCAP sufficiently stressed BHCs’
risk exposures and potential PPNR under the more adverse scenario. To
accomplish this, the regulators made what they viewed to be conservative
assumptions and decisions in the following areas. First, the regulators
decided to stress only assets that were on the BHCs’ balance sheets as of
December 31, 2008, (i.e., a static approach) without accounting for new
business activity. According to BHC officials, new loans were thought to
have generally been of better quality than legacy loans because BHCs had
significantly tightened their underwriting standards since the onset of the
financial crisis. 28 As a result, BHCs would have been less likely to chargeoff these loans within the SCAP time period ending December 31, 2010,
resulting in the potential for greater reported revenue estimates for the
period. By excluding earnings from new business, risk-weighted assets
were understated, charge-off rates were overstated, and projected capital
levels were understated.
Second, SCAP regulators generally did not allow the BHCs to cut expenses
to address the anticipated drop in revenues under the more adverse
scenario. However, some BHC officials told us that they would likely cut
expenses, including initiating rounds of layoffs, if the economy performed
in accordance with the more adverse economic scenario, especially if they
were not generating any new business. Federal Reserve officials noted that
BHCs were given credit in the stress test for cost cuts made in the first
quarter of 2009.

27
In its June 2009 SCAP analysis report, the Congressional Oversight Panel also noted that
there was a lack of transparency about the linkage between the loan losses and the three
macroeconomic assumptions.
28

According to the Federal Reserve, legacy loans refer to those bank loans made during the
2005 to 2007 period. Underwriting standards refer to guidelines used by lenders to ensure
that loans meet credit standards and that the terms and conditions of a loan are
appropriate to its risk and maturity.

Page 14

GAO-10-861 Lessons Learned from Bank Stress Test

Third, some BHCs were required to assume an increase in their ALLL as of
the end of 2010, if necessary, to ensure adequate reserves relative to their
year end 2010 portfolio. Some BHC officials believed that this requirement
resulted in the BHCs having to raise additional capital because the
required ALLL increases were subtracted from the revenue estimates in
calculating the resources available to absorb losses. This meant that some
BHCs judged to have insufficient year end 2010 reserve adequacy had to
account for this shortcoming in the calculation of capital needed to meet
the SCAP targeted capital requirements as of the end of 2010 while
maintaining a sufficient ALLL for 2011 losses under the more adverse
economic scenario. According to some BHCs, the size of the 2010 ALLL
was severe given the extent of losses are already included in the 2009 and
2010 loss estimates and effectively stressed BHCs for a third year.
Finally, according to many BHC officials and others, the calculations used
to derive the loan loss rates and other assumptions to stress the BHCs
were conservative (i.e., more severe). For example, the total loan loss rate
estimated by the SCAP regulators was 9.1 percent, which was greater than
the historical 2-year loan loss rates at all commercial banks from 1921 until
2008, including the worst levels seen during the Great Depression (see
figure 2). However, the macroeconomic assumptions of the more adverse
scenario, which we will discuss later in the report, did not meet the
definition of a depression. Specifically, a 25 percent unemployment rate
coupled with economic contraction is indicative of a depression. In
contrast, the more adverse scenario estimated approximately a 10 percent
unemployment rate with some economic growth in late 2010.

Page 15

GAO-10-861 Lessons Learned from Bank Stress Test

Figure 2: Commercial Bank 2-Year Loan Loss Rates from 1921 through 2013 Compared to SCAP Loan Loss Rate
Percentage
10
SCAP total loan loss rate = 9.1%
8

6

4

2

0
'21 '23 '25 '27 '29 '31 '33 '35 '37 '39 '41 '43 '45 '47 '49 '51 '53 '55 '57 '59 '61 '63 '65 '67 '69 '71 '73 '75 '77 '79 '81 '83 '85 '87 '89 '91 '93 '95 '97 '99 '01 '03 '05 '07 '09 '11 '13
Year
Source: International Monetary Fund.

Note: The solid line represents actual loss rates (1921-2008) and the dotted line represents estimated
loss rates (2009-2013).

SCAP regulators also estimated ranges for loan loss rates within specific
loan categories using the baseline and more adverse scenarios as guides.
They used a variety of methods to tailor loan losses to each BHC,
including an analysis of past BHC losses and quantitative models, and
sought empirical support from BHCs regarding the risk level of their
portfolios. However, some BHCs told us that the Federal Reserve made
substantial efforts to help ensure conformity with the indicative loan loss
rates while incorporating BHC-specific information where possible and
reliable. Table 1 compares the different indicative loan loss rate ranges
under the more adverse scenario for each asset category with actual losses
in 2009 for SCAP BHCs and the banking industry. 29 Some BHCs stated that
the resulting loan loss rates were indicative of an economy worse off than
that represented by the more adverse macroeconomic assumptions,
although they recognized the need for the more conservative approach.

29

Loss rate ranges under the more adverse scenario were later tailored to each BHC.

Page 16

GAO-10-861 Lessons Learned from Bank Stress Test

However, nearly all agreed that the loan loss rates were a more important
indication of the stringency of SCAP than the assumptions.
Table 1: Indicative Loss Rates Estimates and Actual SCAP BHCs and Banking Industry Average Loss Rates, December 31,
2009
Percentage
SCAP indicative loss rate estimates

Loan category

2009 actual loss rates

Federal Reserve’s
more adverse 2-year
loss ratea

GAO’s more
adverse 1-year pro
b
rata loss rate

SCAP BHCs
average
loss rate

Banking industry
average
c
loss rate

7-8.5%

3.5-4.25%

1.9%

1.7%

First-lien mortgage
•
Prime

3-4

1.5-2

n/a

0.5

•

Alt-A

9.5-13

4.75-6.5

n/a

3.6

•

Subprime

21-28

10.5-14

n/a

6.2

Second/junior lien mortgages
•
Closed-end junior liens

12-16

6-8

4.4

3.9

22-25

11-12.5

7.5

6.6

Home lines of credit

8-11

4-5.5

3.6

3.1

•

Commercial and industrial

5-8

•
•

Nonfarm, nonresidential

2.5

2.3

4.5-6

2.3

2.4

15-18

Multifamily

2.5-4

9-12

Commercial real estate
•
Construction

7.5-9

5.8

6.1

10-11

Credit cards

5-5.5

1.1

1.1

7-9

3.5-4.5

0.9

0.8

18-20

9-10

10.1

10.2

Other consumer

8-12

4-6

4.1

4.4

Other loans

4-10

2-5

1.4

1.1

Sources: Federal Reserve SCAP results report, GAO analysis of SNL Financial Y-9C regulatory data, and Moody’s Investors Service
for prime, Alt-A, and subprime mortgage loss rates data.

Note: N/a means not available.
a

Data as of December 31, 2010.

b

GAO calculated the more adverse 1-year pro rata loss rate by dividing the SCAP more adverse 2year loss rates by 2 (i.e., the straight-line method). A key limitation of this approach is that it assumes
equal distribution of losses, revenues, expenses, and changes to reserves over time, although these
items were unlikely to be distributed evenly over the 2-year period. Another important consideration is
that actual results were not intended and should not be expected to align with the SCAP projections.
c

Data are for BHCs with greater than $1 billion in total assets.

After the public release of the SCAP methodology in April 2009, many
observers commented that the macroeconomic assumptions for a more
adverse economic downturn were not severe enough given the economic
conditions at that time. In defining a more adverse economic scenario, the

Page 17

GAO-10-861 Lessons Learned from Bank Stress Test

SCAP regulators made assumptions about the path of the economy using
three broad macroeconomic indicators—changes in real GDP, the
unemployment rate, and home prices—during the 2-year SCAP period
ending December 2010. The actual performances of GDP and home prices
have performed better than assumed under the more adverse scenario.
However, the actual unemployment rate has more closely tracked the
more adverse scenario (see figure 3). Further, as noted earlier, some
regulatory and BHC officials have indicated that the loan loss rates that
the regulators subsequently developed were more severe than one would
have expected under the macroeconomic assumptions. While our analysis
of actual and SCAP estimated indicative loan losses (see table 1) is
generally consistent with this view, these estimates were developed at a
time of significant uncertainty about the direction of the economy and the
financial markets, as well as an unprecedented deterioration in the U.S.
housing markets.
Figure 3: Actual Economic Performance to Date Versus SCAP More Adverse Assumptions
GDP actual vs. SCAP adverse

Unemployment actual vs. SCAP adverse

Home price actual vs. SCAP adverse

Percentage
6

Percentage
12

Percentage
100

Actual

Actual

5
4

11

3
2

Actual

10

90

1
0

9

-1

SCAP more adverse

-2

8

-3

80

-4

7

SCAP more adverse

-5
-6

6

SCAP more adverse

-7
-8

Q4

Q1

Q2

Q3

2008 2009

Q4

Q1
2010

Q2

Q3

Q4

5

Q4

Q1

2008 2009

Q2

Q3

Q4

Q1
2010

Q2

Q3

Q4

70

Q4

Q1

2008 2009

Q2

Q3

Q4

Q1

Q2

Q3

Q4

2010

Quarter and year
Source: GAO analysis of Bureau of Economic Analysis, Bureau of Labor Statistics, and Standard and Poor's 10-City Case-Shiller data.

Page 18

GAO-10-861 Lessons Learned from Bank Stress Test

While SCAP Increased
Capital Levels and
Improved Confidence
in the Banking
System, BHCs Could
Face Ongoing
Challenges

SCAP largely met its goals of increasing the level and quality of capital
held by the 19 largest BHCs and, more broadly, of strengthening market
confidence in the banking system. The stress test identified 10 of the 19
BHCs as needing to raise a total of about $75 billion in additional capital.
The Federal Reserve encouraged the BHCs to raise common equity via
private sources—for example, through new common equity issuances,
conversion of existing preferred equity to common equity, and sales of
businesses or portfolios of assets. Nine of the 10 BHCs were able to raise
the required SCAP amount of new common equity in the private markets
by the November 9, 2009, deadline (see table 2). Some of these BHCs also
raised capital internally from other sources. 30 GMAC LLC (GMAC) was the
only BHC that was not able to raise sufficient private capital by the
November 9, 2009, deadline. 31 On December 30, 2009, Treasury provided
GMAC with a capital investment of $3.8 billion to help fulfill its SCAP
capital buffer requirement, drawing funds from TARP’s Automotive
Industry Financing Program. 32 A unique and additional element of the
estimated losses for GMAC included the unknown impact of possible
bankruptcy filings by General Motors Corporation (GM) and Chrysler LLC
(Chrysler). Thus, a conservative estimate of GMAC’s capital buffer was
developed in response to this possibility. The Federal Reserve, in
consultation with Treasury, subsequently reduced GMAC’s SCAP required
capital buffer by $1.8 billion—$5.6 billion to $3.8 billion—primarily to
reflect the lower-than-estimated actual losses from the bankruptcy

30

Other forms of raising capital included the use of deferred tax assets (DTA), employee
stock option awards, and restriction on dividend payments. A DTA represents the amount
by which taxes can be reduced in future years as a result of temporary tax differences for
financial reporting and tax reporting purposes. DTAs are includable as tier 1 capital up to
no more than 10 percent of a BHC’s tier 1 capital.

31

The SCAP results required GMAC to raise a total of $1l.5 billion in capital, of which $9.1
billion had to be in new equity capital. On May 21, 2009, Treasury made a capital
investment of $3.5 billion in GMAC via the TARP Automotive Industry Financing Program
to be applied as a down payment towards GMAC’s SCAP capital buffer of $9.1 billion in
new equity capital. GMAC had to raise the remaining amount of $5.6 billion by the
November 9, 2009, deadline from either the private markets or through further Treasury
assistance. In December 2009, Treasury converted its existing $5.25 billion of preferred
stock into mandatorily convertible preferred stock and converted $3 billion of existing
GMAC mandatorily convertible preferred securities into common equity that allowed
GMAC to meet its total SCAP capital requirement of $11.5 billion.

32

The Automotive Industry Financing Program was created in December 2008 by Treasury
under TARP to prevent a significant disruption of the American automotive industry.
Treasury determined that such a disruption would pose a systemic risk to financial market
stability and have a negative effect on the U.S. economy.

Page 19

GAO-10-861 Lessons Learned from Bank Stress Test

proceedings of GM and Chrysler. GMAC was the only company to have its
original capital buffer requirement reduced.
Table 2: Summary of Capital Raised by 10 BHCs to Meet Their SCAP Capital Buffer
Amount, as of November 9, 2009
Dollars in billions
BHC
Bank of America Corporation

Sources of capital
raised
New shares, asset
a
sales, and conversion

Required capital
buffer amount

Capital
raised

$33.9

$35.9

Citigroup Inc.

Conversion

5.5

5.6

Fifth Third Bancorp

New shares, asset
sales, and conversion

1.1

1.7

GMAC LLC

New shares

11.5

4.6

KeyCorp

New shares, asset
sales, and conversion

1.8

2.3

Morgan Stanley

New shares, asset
sales, and conversion

1.8

7.0

PNC Financial Services
Group, Inc.

New shares and asset
sales

0.6

1.1

Regions Financial
Corporation

New shares, asset
sales, conversion, and
b
other actions

2.5

2.5

SunTrust Banks, Inc.

New shares, asset
sales, conversion, and
other actions

2.2

2.2

Wells Fargo & Company

New shares and other
actions

13.7

13.7

$74.6

$76.6

Total
Source: Federal Reserve documentation.

Notes: The following nine BHCs were not required to raise SCAP capital because they had sufficient
capital to withstand a worse-than-expected economic downturn through the end of 2010 and continue
to meet the SCAP capital buffer targets: American Express Company; BB&T Corporation; The Bank
of New York Mellon Corporation; Capital One Financial Corporation; The Goldman Sachs Group, Inc.;
JPMorgan Chase & Co.; MetLife, Inc.; State Street Corporation; and U.S. Bancorp. Data in the
“capital raised” column is as of November 9, 2009, according to the Federal Reserve.
a

“New shares” indicates that BHC issued new common equity, “assets sales” represent business lines
or products sold to raise cash, and “conversion” shows BHC preferred equity that was converted to
common equity.

b

“Other action” indicates equity raised internally (e.g., sale of equity to employee stock options plans).

Page 20

GAO-10-861 Lessons Learned from Bank Stress Test

Capital adequacy generally improved across all 19 SCAP BHCs during
2009. As shown in table 3, the largest gains were in tier 1 common capital,
which increased by about 51 percent in the aggregate across the 19 BHCs,
rising from $412.5 billion on December 31, 2008, to $621.9 billion by
December 31, 2009. On an aggregate basis, the tier 1 common capital ratio
at BHCs increased from 5.3 percent to 8.3 percent of risk-weighted assets
(compared with the SCAP threshold of 4 percent at the end of 2010).33 The
tier 1 risk-based capital ratio also grew from 10.7 percent to 11.3 percent
of risk-weighted assets (compared with the SCAP threshold of 6 percent at
the end of 2010). 34 While these ratios were helped to some extent by
reductions in risk-weighted assets, which fell 4.3 percent from $7.815
trillion on December 31, 2008, to $7.481 trillion on December 31, 2009, the
primary driver of the increases was the increase in total tier 1 common
capital.
Table 3: Capital Measures for SCAP BHCs, December 31, 2008 and December 31,
2009
Dollars in billions
Capital measures
Capital levels
•
Tier 1 capital
•

Tier 1 common capital

•

Risk-weighted assets

Capital ratios
•
Tier 1 risk-based capital ratio
•

Tier 1 common capital ratio

2009

2008

Percent
difference

$846.2

$836.7

1.1%

$621.9

$412.5

50.8

$7,480.8

$7,814.8

-4.3

11.3%

10.7%

5.6

8.3%

5.3%

57.5

Sources: GAO analysis of Federal Reserve SCAP, SNL Financial, and company data.

The quality of capital—measured as that portion of capital made up of tier
1 common equity—also increased across most of the BHCs in 2009. The
tier 1 common capital ratio increased at 17 of the 19 BHCs between the
end of 2008 and the end of 2009 (see table 4). Citigroup Inc. (Citigroup)
and The Goldman Sachs Group, Inc. (Goldman Sachs) had the largest
increases in tier 1 common capital ratios—747 and 450 basis points,

33

Tier 1 common capital ratio equals tier 1 common capital divided by total risk-weighted
assets.

34

Tier 1 risk-based capital ratio equals tier 1 capital divided by total risk-weighted assets.

Page 21

GAO-10-861 Lessons Learned from Bank Stress Test

respectively. 35 However, GMAC’s tier 1 common capital ratio declined by
155 basis points in this period to 4.85 percent. MetLife, Inc. was the only
other BHC to see a drop in its tier 1 common capital ratio, which fell by 33
basis points to 8.17 percent and still more than double the 4 percent target.
Based on the SCAP results document, the 2008 balances in the table
include the impact of certain mergers and acquisitions, such as Bank of
America Corporation’s (Bank of America) purchase of Merrill Lynch & Co.
Inc. Further, the increase in capital levels reflects the capital that was
raised as a result of SCAP.
Table 4: Percentage Change in Tier I Capital Ratios, December 31, 2008, and December 31, 2009
Tier 1 common capital ratio

Tier 1 risk-based capital ratio

2009
(percentage)

Change from 2008
(basis points)

American Express Company

9.83%

13

9.84%

14

Bank of America Corporation

7.82

322

10.41

-19

Bank holding company

BB&T Corporation

2009 Change from 2008
(percentage)
(basis points)

8.50

140

11.48

-82

The Bank of New York Mellon Corporation

10.53

103

12.12

-118

Capital One Financial Corporation

10.62

152

13.75

105

Citigroup Inc.

9.77

747

11.67

-23

Fifth Third Bancorp

7.00

260

13.31

271

GMAC LLC

4.85

-155

14.15

405

12.20

450

14.97

237

JPMorgan Chase & Co.

8.79

229

11.10

90

KeyCorp

7.50

190

12.75

185

MetLife, Inc.

8.17

-33

8.91

-29

Morgan Stanley

6.71

101

15.30

10

PNC Financial Services Group, Inc.

6.00

130

11.42

182

Regions Financial Corporation

7.15

55

11.54

114

15.59

9

17.74

-246

7.67

187

12.96

206

The Goldman Sachs Group, Inc.

State Street Corporation
SunTrust Banks, Inc.

35

A basis point is a common measure used in quoting yields on bills, notes, and bonds and
represents 1/100 of a percent of yield. For example, an increase from 4.35 percent to 4.45
percent would be an increase of 10 basis points.

Page 22

GAO-10-861 Lessons Learned from Bank Stress Test

Tier 1 common capital ratio

Tier 1 risk-based capital ratio

2009
(percentage)

Change from 2008
(basis points)

U.S. Bancorp

6.76

166

9.61

-99

Wells Fargo & Company

6.46

336

9.25

125

8.31%

303

11.31%

60

Bank holding company

Average (weighted)

2009 Change from 2008
(percentage)
(basis points)

Sources: GAO analysis of Federal Reserve SCAP, SNL Financial, and company data.

As previously stated by interviewees, the unprecedented public release of
the stress test results helped to restore investors’ confidence in the
financial markets. Some officials from participating BHCs and credit rating
agencies also viewed the BHCs’ ability to raise the capital required by the
stress test as further evidence of SCAP’s success in increasing market
confidence and reducing uncertainty. But some expressed concerns that
the timing of the announcement of SCAP on February 10, 2009—nearly 3
months before the results were released on May 7, 2009—may have
intensified market uncertainty about the financial health of the BHCs.
A broad set of market indicators also suggest that the public release of
SCAP results may have helped reduce uncertainty in the financial markets
and increased market confidence. For example, banks’ renewed ability to
raise private capital reflects improvements in perceptions of the financial
condition of banks. Specifically, banks and thrifts raised significant
amounts of common equity in 2008, totaling $56 billion. Banks and thrifts
raised $63 billion in common equity in the second quarter of 2009 (see
figure 4). The substantial increase in second quarter issuance of common
equity occurred after the stress test results were released on May 7, 2009,
and was dominated by several SCAP institutions.

Page 23

GAO-10-861 Lessons Learned from Bank Stress Test

Figure 4: Gross Common Equity Issuance by Banks and Thrifts, 2000 to First Quarter 2010
Dollars in billions
120

Dollars in billions
70

Quarterly data, 2007-2010

60

100

50
40
80
30
20
60

10
0

40

Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1
(22) (18) (19) (27) (19) (31) (22) (30) (15) (51) (56) (55 (47)
2007
2008
2009
2010

20

Quarter, (number of banks who raised capital), year

0
2000

2001

2002

2003

2004'

2005

2006

2007

2008

2009

Source: GAO analysis of data from SNLFinancial.

Note: The spike in common equity issuance in the fourth quarter of 2009 primarily relates to Citigroup,
Wells Fargo & Company, and other banks raising capital to buy back their TARP capital investment
from Treasury. However, the quarterly data do not reflect $19.29 billion of common equivalent
securities issued in December 2009 by Bank of America that converted to common stock in February
2010.

Similarly, stock market prices since the release of the stress test results in
May 2009 through October 2009 improved substantially in the overall
banking sector and among the 18 public BHCs that participated in SCAP
(see figure 5). 36 The initial increase since May 2009 also suggests that
SCAP may have helped bolster investor and public confidence. However,
equity markets are generally volatile and react to a multitude of events.

36

GMAC is the only nonpublic BHC that was included in SCAP.

Page 24

GAO-10-861 Lessons Learned from Bank Stress Test

Figure 5: Stock Market Prices, October 2007 through March 2010
Percent change
10
0

October 2007 stock price
Announcement of
stress test as
part of FSP
(Feb. 10)

-10
-20

Release of
stress test
results
(May 7)

-30
-40
-50
-60
-70
-80
-90
Oct. Nov Dec. Jan. Feb. Mar. April May June July Aug. Sept. Oct. Nov. Dec. Jan. Feb. Mar. April May June July Aug. Sept. Oct. Nov. Dec. Jan. Feb. Mar.
2007

2008

2009

2010

S&P 500
Simple average of 18 public BHCs
Financial industry measure
Source: GAO analysis of Yahoo! Finance data.

Credit default swap spreads, another measure of confidence in the
banking sector, also improved. A credit default swap is an agreement in
which a buyer pays a periodic fee to a seller in exchange for protection
from certain credit events such as bankruptcy, failure to pay debt
obligations, or a restructuring related to a specific debt issuer or issues
known as the reference entity. Therefore, the credit default swap spread,
or market price, is a measure of the credit risk of the reference entity, with
a higher spread indicating a greater amount of credit risk. When the
markets’ perception of the reference entity’s credit risk deteriorates or
improves, the spread generally will widen or tighten, respectively.
Following the SCAP results release in May 2009, the credit default swap
spreads continued to see improvements (see figure 6). While many forces
interact to influence investors’ actions, these declining spreads suggest
that the market’s perception of the risk of banking sector defaults was
falling. Further, the redemption of TARP investments by some banking
institutions demonstrated that regulators believed these firms could
continue to serve as a source of financial and managerial strength, as well
as fulfill their roles as intermediaries that facilitate lending, while both

Page 25

GAO-10-861 Lessons Learned from Bank Stress Test

reducing reliance on government funding and maintaining adequate capital
levels. This positive view of the regulators may also have helped increase
market confidence in the banking system (see appendix II for details on
the status of TARP investments in the institutions participating in SCAP).
Figure 6: Bank Credit Default Swap Spreads, January 2007 through March 2010
Credit default swap spread in basis points
1,500

Announcement of
stress test
as part
of FSP
(Feb. 10)

GMAC
10,000

1,200

8,000

Announcement
of stress test
as part of FSP
(Feb. 10)

Release of
stress test
results
(May 7)

6,000
4,000
900

Release of
stress test
results
(May 7)

2,000
0
2007

600

2008

2009

2010

300

0
Jan.

Mar.

May

July

Sept.

Nov.

2007

Jan.

Mar.

May

July

Sept.

Nov.

2008

Jan.
2009

Mar.

May

July

Sept.

Nov.

Jan.

Mar.

2010

Bank index
Average of 12 of the 19 SCAP BHCs
Source: GAO analysis of Thomson Reuters Datastream.

The 19 Tested BHCs
Experienced Better
Performance Than a Pro
Rata Estimate under the
More Adverse Scenario

As of the end of 2009, while the SCAP BHCs generally had not experienced
the level of losses that were estimated on a pro rata basis under the stress
test’s more adverse economic scenario, concerns remain that some banks
could absorb potentially significant losses in certain asset categories that
would erode capital levels. Collectively, the BHCs’ total loan losses of
$141.2 billion were approximately 38 percent less than the GAO-calculated
$229.4 billion in pro rata losses under the more adverse scenario for 2009

Page 26

GAO-10-861 Lessons Learned from Bank Stress Test

(see table 5). 37 The BHCs also experienced significant gains in securities
and trading and counterparty credit risk portfolios compared with
estimated pro rata losses under SCAP. Total resources other than capital
to absorb losses (resources) were relatively close to the pro rata amount,
exceeding it by 4 percent.
Table 5: Actual and GAO Pro Rata Estimates of Aggregate Losses and Changes in
Resources Other than Capital to Absorb Losses across the 19 SCAP BHCs,
December 31, 2009
Dollars in billions

Asset category

Actual

GAO
pro rata
Percent
estimatea difference

Consumer and commercial loan losses
•
First-lien mortgages

$19.2

$51.2

-62%

•

Second/junior lien mortgages

26.1

41.6

-37

•

Commercial and industrial loans

21.2

30.1

-29

•

Commercial real estate loans

13.5

26.5

-49

•

Credit card loans

31.6

41.2

-23

•

Otherb

29.5

38.9

-24

$141.2

$229.4

-38%

(3.5)

17.6

-120

Total consumer and commercial loans losses
Securities—available for sale and held to maturity—
losses (gains)
Trading and counterparty losses (gains)

(56.9)

49.7

-215

Total asset losses

$80.8

$296.7

-73%

$188.4

$181.5

4%

Resources other than capital to absorb losses
Sources: GAO analysis of Federal Reserve SCAP and SNL Financial data.

Notes: A parenthetical number indicates a gain.

37

The asset categories are first-lien mortgages consisting of prime, Alt-A, and subprime
residential mortgages; second/junior lien mortgages consisting of closed-end junior liens
and home equity line of credit residential mortgages; commercial and industrial loans
consisting of large corporate and middle market, small business, and asset-based lending
loans; commercial real estate loans consisting of construction and land development,
multifamily, and nonfarm nonresidential loans; credit card loans, consisting of credit cards;
other loans consisting of auto, personal, and student loans, and farmland lending, loans to
depository institutions, loans to governments, and other categories; securities (available for
sale and held to maturity) consisting of a majority of Treasury securities, government
agency securities, sovereign debt, and high-grade municipal securities and corporate
bonds, equities, asset-backed securities, commercial mortgage-backed securities, and
nonagency residential mortgage-backed securities; and trading and counterparty, or trading
book positions (e.g., securities such as common stock and derivatives).

Page 27

GAO-10-861 Lessons Learned from Bank Stress Test

The trading and counterparty data in the Y-9C includes both customer derived revenue from
transactions for BHCs that operate as broker-dealers, as well as gains and losses from proprietary
trading and associated expenses. These items are presented on a net basis in the Y-9C. For the five
BHCs that had their trading portfolios stressed (Goldman Sachs, Morgan Stanley, Citigroup,
JPMorgan Chase & Co., and Bank of America), the trading and counterparty line item is based on
projections of gains or losses from proprietary trading, but preprovision net revenue (specifically
noninterest revenue) included projections of gains or losses from customer derived revenue from
transactions due to operations as a broker-dealer. These items cannot be segregated based on the
Y-9C data and therefore are included in the net amount in both the trading and counterparty and
noninterest income line items above. As a result of this limitation, the net amount of the trading gains
or losses and preprovision net revenue in the table may be overstated or understated.
a

GAO calculated 1-year pro rata loss estimates by dividing the SCAP more adverse 2-year loss
estimates by 2 (e., the straight-line method). A key limitation of this approach is that it assumes equal
distribution of losses, revenues, expenses, and changes to reserves over time, although these items
were unlikely to be distributed evenly over the 2-year period. Another important consideration is that
actual results were not intended and should not be expected to align with the SCAP projections.

b

For “Other” we excluded about $6 billion in losses for State Street Corporation realized in 2009.
Since this was a one-time charge that was realized in 2009, this effect was segregated from more
typical loss amounts for our tracking purposes.

In tracking BHCs’ losses and resources against the SCAP estimates, we
compared the actual results with those estimated under the more adverse
scenario. We used the 2-year estimates of the more adverse scenario from
the SCAP results and annualized those amounts by dividing them in half
(the “straight line” method) to get pro rata loss estimates for 2009 because
the SCAP regulators did not develop estimates on a quarterly or annual
basis. A key limitation of this approach is that it assumes equal
distribution of losses, revenues, expenses, and changes to reserves over
time, although these items were unlikely to be distributed evenly over the
2-year period. Another important consideration is that actual results were
not intended and should not be expected to align with the SCAP
projections. Actual economic performance in 2009 differed from the SCAP
macroeconomic variable inputs, which were based on a scenario that was
more adverse than was anticipated or than occurred, and other forces in
the business and regulatory environment could have influenced the timing
and level of losses. Appendix I contains additional details on our
methodology, including our data sources and calculations, for tracking
BHCs’ financial performance data.

Page 28

GAO-10-861 Lessons Learned from Bank Stress Test

Losses Varied by Individual
BHCs

Although the 19 BHCs’ actual combined losses were less than the 2009 pro
rata loss estimates for the more adverse scenario, the loss rates varied
significantly by individual BHCs. For example, most of the BHCs had
consumer and commercial loan losses that were below the pro rata loss
estimates, but three BHCs—GMAC, Citigroup, and SunTrust Banks Inc.
(SunTrust)—exceeded these estimates in at least one portfolio (see figure
7). GMAC was the only one with 2009 loan losses on certain portfolios that
exceeded SCAP’s full 2-year estimate. Specifically, GMAC exceeded the
SCAP 2-year estimated losses in the first-lien, second/junior lien, and
commercial real estate portfolios and the 1-year pro rata losses in the
“Other” portfolio; Citigroup exceeded the 1-year pro rata estimated losses
in the commercial and industrial loan portfolio; and SunTrust exceeded
the 1-year estimated losses in the first-lien and credit card portfolios.
Appendix III provides detailed data on the individual performance of each
of the BHCs.

Page 29

GAO-10-861 Lessons Learned from Bank Stress Test

Figure 7: Comparison of Actual and GAO Pro Rata Estimated Losses for Consumer and Commercial Loans, December 31,
2009
GAO’s 1-year
pro rata estimate

Full 2-year
SCAP estimate

American Express Co.
BB&T Corp.
The Bank of New York Mellon Corp.
Bank of America Corp.
Capital One Financial Corp.
Citigroup Inc.
Fifth Third Bancorp
GMAC LLC
JPMorgan Chase & Co.
KeyCorp
MetLife, Inc.
Morgan Stanley
PNC Financial Services Group Inc.
Regions Financial Corp.
State Street Corp.
SunTrust Banks. Inc.
U.S. Bancorp
Wells Fargo & Co.

Total losses for 19 BHCs
0

25

50

75

100

125

200

225

250

275

300

Percentage
Other

Commercial real estate

Commercial and industrial loans

First-lien mortgages

Second/junior lien mortgages

Credit cards

Source: GAO analysis of Federal Reserve and SNLFinancial data.

Notes: Figure shows only those loan loss categories that were applicable under SCAP and that
showed losses in 2009. In addition, Goldman Sachs was not included in the figure because it had no
losses or recoveries for these loan categories in 2009. The “Other” category for State Street
Corporation does not include one-time items in the actual or estimated amounts. See table 27 in
appendix III for additional details.

Page 30

GAO-10-861 Lessons Learned from Bank Stress Test

GAO calculated 1-year pro rata loss estimates by dividing the SCAP more adverse 2-year loss
estimates by 2 (i.e., the straight-line method). A key limitation of this approach is that it assumes
equal distribution of losses, revenues, expenses, and changes to reserves over time, although these
items were unlikely to be distributed evenly over the 2-year period. Another important consideration is
that actual results were not intended and should not be expected to align with the SCAP projections.

GMAC faced particular challenges in the first year of the assessment
period and posed some risk to the federal government, a majority equity
stakeholder. 38 GMAC’s loan losses in its first-lien portfolio were $2.4
billion, compared with the $2 billion projected for the full 2-year period. In
the second/junior lien portfolio, GMAC saw losses of $1.6 billion,
compared with the $1.1 billion estimated losses for the 2 years. GMAC
experienced losses of $710 million in its commercial real estate portfolio,
compared with $600 million projected for the full 2-year period. Further, in
its “Other” portfolio (which is comprised of auto leases and consumer auto
loans), GMAC’s losses were $2.1 billion, exceeding the 1-year pro rata $2
billion loss estimate. With a tier 1 common capital ratio of 4.85 percent—
just more than the SCAP threshold of 4 percent—at the end of 2009, GMAC
has a relatively small buffer in the face of potential losses.
GMAC’s position should be placed in context, however, because it is
relatively unique among the SCAP participants. It was the only nonpublicly
traded participant, and the federal government owns a majority equity
stake in the company as a result of capital investments made through the
Automotive Industry Financing Program under TARP. Further, GMAC’s
core business line—financing for automobiles—is dependent on the
success of efforts to restructure, stabilize, and grow General Motors
Company and Chrysler Group LLC. 39 Finally, the Federal Reserve told us
that because GMAC only recently became a BHC and had not previously
been subject to banking regulations, it would take some time before
GMAC was fully assimilated into a regulated banking environment. 40 To
improve its future operating performance and better position itself to
become a public company in the future, GMAC officials stated that the
company posted large losses in the fourth quarter of 2009 as result of
accelerating its recognition of lifetime losses on loans. 41 In addition, the

38

As of September 22, 2010, Treasury has a 56.3 percent ownership stake in GMAC.

39

General Motors Company and Chrysler Group LLC are the new names that the former GM
and Chrysler adopted, respectively, after emerging from bankruptcy.

40

The Federal Reserve approved GMAC’s application to become a BHC on December 24,
2008.

41

Lifetime losses are those losses which occur from origination to the life-end of the loans.

Page 31

GAO-10-861 Lessons Learned from Bank Stress Test

company has been restructuring its operations and recently sold off some
nonperforming assets. 42 However, the credit rating agencies we met with
generally believed that there could still be further losses at GMAC,
although the agencies were less certain about the pace and level of those
losses. Two of the agencies identified GMAC’s Residential Capital, LLC
mortgage operation as the key source of potential continued losses.

BHCs Are Generally Not
Experiencing the Level of
Securities and Trading
Losses That Were
Estimated under the Pro
Rata More Adverse
Scenario, and Some Have
Recorded Gains

Given that market conditions have generally improved, the BHCs’
investments in securities and trading account assets performed
considerably better in 2009 than had been estimated under the pro rata
more adverse scenario. 43 The SCAP assessment of the securities portfolio
consisted of an evaluation for possible impairment of the portfolio’s
assets, including Treasury securities, government agency securities,
sovereign debt, and private sector securities. In the aggregate, the
securities portfolio has experienced a gain of $3.5 billion in 2009,
compared with a pro rata estimated loss of $17.6 billion under the stress
test’s more adverse scenario. As figure 8 shows, 5 of the 19 BHCs recorded
securities losses in 2009, 44 13 recorded gains, and 1 (Morgan Stanley)
recorded no gain or loss. Losses were projected at 17 of the BHCs under
the pro rata more adverse scenario, and SCAP regulators did not consider
the remaining 2 BHCs (American Express Company and Morgan Stanley)
to be applicable for this category. In the securities portfolio, The Bank of
New York Mellon Corporation had losses greater than estimated under

42

According to an April 12, 2010, GMAC press release, GMAC’s mortgage subsidiary—
Residential Capital, LLC—agreed to sell its European mortgage assets and businesses to
Fortress Investment Group LLC. These transactions represent approximately 10 percent of
Residential Capital, LLC’s December 31, 2009, total assets and approximately 40 percent of
total assets on a pro forma basis, adjusted for the required accounting treatment for certain
off-balance sheet securitizations that are recorded on-balance sheet effective January 1,
2010, (see Financial Accounting Statement No. 167). The assets in the transactions are
valued at approximately the levels established in the fourth quarter of 2009, and there is no
material gain or loss expected. GMAC reported positive earnings for the first and second
quarters of 2010, although it continued to show losses in certain portfolios. These were its
first profits since the fourth quarter of 2008. GMAC’s tier 1 common capital ratio also
improved to 5 percent and 5.2 percent, respectively.

43

Trading account assets are assets held to hedge risks or speculate on price changes for
the bank or its customers. Because the more adverse scenario was plausible but unlikely to
occur, the actual results were not intended and should not be expected to align with such
scenario.
44

The five BHCs are The Bank of New York Mellon Corporation; Citigroup; MetLife, Inc.;
SunTrust; and U.S. Bancorp.

Page 32

GAO-10-861 Lessons Learned from Bank Stress Test

SCAP for the full 2-year period. 45 The variances could be due to a number
of factors, including the extent to which a BHC decides to deleverage, how
their positions react to changing market values, and other factors.

45
Based on discussion with The Bank of New York Mellon Corporation officials and as
stated in a October 20, 2009, company press release, the BHC’s securities portfolio
underwent a significant restructuring in the third quarter of 2009 in order to reduce its
balance sheet risk, causing it to recognize significant losses in that period. The officials
noted that the BHC sold off many of its riskiest holdings in that period, including many AltA residential mortgage-backed securities, so that they expect to see gains in this portfolio
in the future, keeping the final 2-year loss under the SCAP projected amount. As of the
second quarter of 2010 year to date, the BHC experienced a gain of $20 million in this
portfolio.

Page 33

GAO-10-861 Lessons Learned from Bank Stress Test

Figure 8: Comparison of Actual and GAO Pro Rata Estimated Gains and Losses for
Securities Available for Sale and Held to Maturity, December 31, 2009
American
Express Co.

GAO pro rata estimate n/a

BB&T Corp.
The Bank of
New York
Mellon Corp.
Bank of
America Corp.
Capital One
Financial
Corp.
Citigroup
Inc.
Fifth Third
Bancorp
GMAC LLC
The Goldman
Sachs Group
Inc.
JPMorgan
Chase & Co.
KeyCorp
MetLife, Inc.
PNC Financial
Services
Group Inc.
Regions
Financial Corp.
State Street
Corp.
SunTrust
Banks, Inc.
U.S.
Bancorp
Wells Fargo
& Co.

-6

-5

-4

-3

-2

-1

0

0

1

8

10

Dollars in billions
GAO’s pro rata estimates for more adverse scenario
Actual results
Source: GAO analysis of Federal Reserve and SNLFinancial data.

Notes: Morgan Stanley was not included in the figure because it has not had any available for sale or
held to maturity securities gains (losses) in 2009 and was deemed to be not applicable for this
category in SCAP. American Express Company was also deemed not applicable for this category in
SCAP, but was included in the figure because it had securities gains in 2009.

Page 34

GAO-10-861 Lessons Learned from Bank Stress Test

GAO calculated 1-year pro rata loss estimates by dividing the SCAP more adverse 2-year loss
estimates by 2 (i.e., the straight-line method). A key limitation of this approach is that it assumes
equal distribution of losses, revenues, expenses, and changes to reserves over time, although these
items were unlikely to be distributed evenly over the 2-year period. Another important consideration is
that actual results were not intended and should not be expected to align with the SCAP projections.

To estimate trading and counterparty losses, SCAP regulators assumed
that these investments would be subject to the change in value of a
proportional level as experienced in the last half of 2008. 46 The trading
portfolio shows an even greater difference between the 1-year pro rata
estimates and the actual performance—a gain of $56.9 billion in 2009
rather than the pro rata $49.7 billion estimated loss under the more
adverse scenario (see table 5). The stress test only calculated trading and
counterparty credit loss estimates for the five BHCs with trading assets
that exceeded $100 billion. 47 All five had trading gains as opposed to
losses, based on the publicly available data from the Y-9C. 48 These gains
were the result of a number of particular circumstances. First, the extreme
spreads and risk premium resulting from the lack of liquidity during the
financial crisis—especially in the second half of 2008—reversed in 2009,
improving the pricing of many risky trading assets that remained on BHCs’
balance sheets. Because the trading portfolio is valued at fair value, it had
been written down for the declines in value that occurred throughout 2008
and the first quarter of 2009 and saw significant gains when the market
rebounded through the remainder of 2009. Second, the crisis led to the
failure or absorption of several large investment banks, reducing the
number of competitors and, according to our analysis of Thomson Reuters

46

A counterparty loss is a loss resulting from a counterparty to a transaction failing to fulfill
its financial obligation in a timely manner or from a credit valuation adjustment.
47

These BHCs include Bank of America, Citigroup, Goldman Sachs, JPMorgan Chase & Co.,
and Morgan Stanley were stress tested in SCAP.

48

Limitations of the Y-9C make it difficult to compare the actual results to the projections of
SCAP. The trading and counterparty data in the Y-9C includes both customer derived
revenue from transactions for BHCs that operate as broker-dealers, as well as gains and
losses from proprietary trading and associated expenses. These items are presented on a
net basis in the Y-9C. For the five BHCs that had their trading portfolios stressed (Goldman
Sachs, Morgan Stanley, Citigroup, JPMorgan Chase & Co., and Bank of America), the
trading and counterparty line item is based on projections of gains or losses from
proprietary trading, but preprovision net revenue (specifically noninterest revenue)
included projections of gains or losses from customer derived revenue from transactions
due to operations as a broker-dealer. These items cannot be segregated based on the Y-9C
data and therefore are included in the net amount in both the trading and counterparty and
noninterest income line items above. As a result of this limitation, the net amount of the
trading gains or losses and preprovision net revenue in the table may be over- or
understated.

Page 35

GAO-10-861 Lessons Learned from Bank Stress Test

Datastream, increased market share and pricing power for the remaining
firms. 49 Finally, the Federal Reserve’s low overnight bank lending rates
(near 0 percent) have prevailed for a long period and have facilitated a
favorable trading environment for BHCs. This enabled BHCs to fund
longer-term, higher yielding assets in their trading portfolios with
discounted wholesale funding (see figure 9). 50
Figure 9: Comparison of Actual and GAO Pro Rata Estimated Gains and Losses for
Trading and Counterparty, December 31, 2009
Bank of
America Corp.
Citigroup Inc.
The Goldman
Sachs Group
Inc.
JPMorgan
Chase & Co.
Morgan
Stanley

-15

-10

-5

0

0

5

10

15

20

25

Dollars in billions
GAO’s pro rata estimates for more adverse scenario
Actual results
Source: GAO analysis of Federal Reserve and SNLFinancial data.

Notes: SCAP regulators only generated trading and counterparty estimates for the 5 BHCs with a
trading book (assets) greater than $100 billion, therefore this comparison is not applicable to the other
14 BHCs.
GAO calculated 1-year pro rata loss estimates by dividing the SCAP more adverse 2-year loss
estimates by 2 (i.e., the straight-line method). A key limitation of this approach is that it assumes
equal distribution of losses, revenues, expenses, and changes to reserves over time, although these
items were unlikely to be distributed evenly over the 2-year period. Another important consideration is
that actual results were not intended and should not be expected to align with the SCAP projections.

49

In 2008, Lehman Brothers Holdings Inc. failed, Merrill Lynch & Co. Inc. was acquired by
Bank of America, and The Bear Stearns Companies Inc. was sold to JP Morgan Chase & Co.

50

Wholesale funding describes a class of funding used by banks to meet their liquidity
needs. Wholesale funding providers include, but are not limited to, money market funds,
trust funds, pension funds, corporations, banks, government agencies, and insurance
companies. Wholesale funding instruments include, but are not limited to, federal funds,
public funds, Federal Home Loan Bank advances, the Federal Reserve’s primary credit
program, foreign deposits, brokered deposits, and deposits obtained through the Internet
or certificate of deposits listing services.

Page 36

GAO-10-861 Lessons Learned from Bank Stress Test

Potential Losses in
Consumer and
Commercial Credit
Continue to Pose a
Challenge

Potentially large losses in consumer and commercial loans continue to
challenge SCAP BHCs, and addressing these challenges depends on a
variety of factors, including, among other things, the effectiveness of
federal efforts to reduce foreclosures in the residential mortgage market.
The BHCs absorbed nearly $400 billion in losses in the 18 months ending
December 31, 2008. As they continue to experience the effects of the
recent financial crisis, estimating precisely how much more they could
lose is difficult. In March 2010, officials from two credit rating agencies
indicated that 50 percent or more of the losses the banking industry was
expected to incur during the current financial crisis could still be realized
if the economy were to suffer further stresses.
Data for the 19 BHCs show a rapid rise in the percentage of nonperforming
loans over the course of 2009 (see figure 10). 51 Specifically, total
nonperforming loans grew from 1 percent in the first quarter of 2007 to 6.6
percent in the fourth quarter of 2009 for SCAP BHCs. In particular,
increases in total nonperforming loans were driven by significant growth
in nonperforming first-lien mortgages and commercial real estate loans.
Standard & Poor’s Corporation noted that many nonperforming loans may
ultimately have to be charged-off, exposing the BHCs to further potential
losses. According to the credit rating agencies that we interviewed, federal
housing policy to aid homeowners who are facing foreclosures, as well as
time lags in the commercial real estate markets, will likely continue to
affect the number of nonperforming loans for the remainder of the SCAP
time frame (December 2010).

51

Nonperforming loans, for the purposes of this figure, represent the total of loans that are
either 90 plus days past due or in nonaccrual status. As defined by the instructions to the Y9C, an asset is in nonaccrual status if: (1) it is maintained on a cash basis because of
deterioration in the financial condition of the borrower, (2) payment in full of principal or
interest is not expected, or (3) principal or interest as been in default for a period of 90
days or more unless the asset is both well secured and in the process of collection. Per the
Y-9C instructions, an asset is 90 plus days past due if payment is due and unpaid for 90 days
or more, and if that asset is not in nonaccrual status.

Page 37

GAO-10-861 Lessons Learned from Bank Stress Test

Figure 10: Change in the Percentage of Nonperforming Loans for Applicable SCAP BHCs, by Loan Type, First Quarter 2007
through Fourth Quarter 2009
Percent of total loans
15

First-lien mortgages

12

9

Commercial real estate loans
Total loans

6
Commercial and industrial loans
Credit cards
3

Second/junior lien mortgages
Other

0
Q1

Q2

Q3

Q4

2007

Q1

Q2

2008

Q3

Q4

Q1

Q2

Q3

Q4

2009
Source: GAO analysis of Federal Reserve and SNLFinancial data.

Note: Because they converted to BHCs in late 2008, American Express Company, Goldman Sachs,
and Morgan Stanley did not submit Y-9Cs to the Federal Reserve until the first quarter of 2009, and
GMAC did not submit its Y-9C until the second quarter of 2009. As a result, the data do not include
information on these holding companies before those dates.

The Economic and
Regulatory Environment
Could Impact BHCs’ Net
Revenues and Loss
Reserves

The total amount of resources other than capital to absorb losses
(resources) has tracked the amount GAO prorated under the stress test’s
more adverse scenario. Resources measure how much cushion the BHCs
have to cover loans losses. As shown previously in table 5, the aggregate
actual results through the end of 2009 for resources showed a total of
$188.4 billion, or 4 percent more than GAO’s pro rata estimated $181.5
billion in the stress test’s more adverse scenario. Eleven of the 19 BHCs
tracked greater than the pro rata estimated amount in 2009, while the
remaining 8 tracked less than the estimate (see figure 11). GMAC and
MetLife, Inc. had negative resources in 2009, although only GMAC was
projected to have negative resources over the full 2-year period.

Page 38

GAO-10-861 Lessons Learned from Bank Stress Test

Figure 11: Comparison of Actual and GAO Pro Rata Estimated Resources Other Than Capital to Absorb Losses, December
31, 2009
American Express Co.
BB&T Corp.
The Bank of New
York Mellon Corp.
Bank of America Corp.
Capital One Financial Corp.
Citigroup Inc.
Fifth Third Bancorp
GMAC LLC
The Goldman
Sachs Group Inc.
JPMorgan Chase & Co.
KeyCorp
MetLife, Inc.
Morgan Stanley
PNC Financial
Services Group, Inc.
Regions Financial Corp.
State Street Corp.
SunTrust Banks, Inc.
U.S. Bancorp
Wells Fargo & Co.

-5

0

5

10

15

20

25

30

35

40

Dollars in billions
GAO’s pro rata estimates for more adverse scenario
Actual results
Source: GAO analysis of Federal Reserve and SNLFinancial data.

Notes: Resources other than capital to absorb losses are calculated as preprovision net revenue less
the change in allowance for loan and lease losses.
GAO calculated 1-year pro rata loss estimates by dividing the SCAP more adverse 2-year loss
estimates by 2 (i.e., the straight-line method). A key limitation of this approach is that it assumes
equal distribution of losses, revenues, expenses, and changes to reserves over time, although these
items were unlikely to be distributed evenly over the 2-year period. Another important consideration is
that actual results were not intended and should not be expected to align with the SCAP projections.

Page 39

GAO-10-861 Lessons Learned from Bank Stress Test

Our calculation considers increases in ALLL during 2009 to be a drain on
resources in order to mirror the regulators’ calculation for the full 2-year
projection. However, the ALLL may ultimately be used as a resource in
2010, causing available resources to be higher than they currently appear
in our tracking. PPNR is based on numerous factors, including interest
income, trading revenues, and expenses. The future course of this
resource will be affected by factors such as the performance of the general
economy, the BHCs’ business strategies, and regulatory changes, including
the Dodd-Frank Wall Street Reform and Consumer Protection Act of 2010
(Dodd-Frank Act) and the Credit Card Accountability, Responsibility, and
Disclosure Act of 2009. 52 Such regulatory changes could impose additional
costs or reduce future profitability, either of which would impact future
PPNR.

SCAP Provided
Lessons That Could
Help Regulators
Strengthen
Supervisory Oversight
and BHCs Improve
Risk Management
Practices

The SCAP stress test provided lessons in a number of areas that can be
incorporated in the bank supervision process and used to improve BHCs’
risk management practices. First, the transparency that was part of SCAP
helped bolster market confidence, but the Federal Reserve has not yet
developed a plan that incorporates transparency into the supervisory
process. Second, the SCAP experience highlighted that BHCs’ stress tests
in the past were not sufficiently comprehensive and we found that
regulators’ oversight of these tests has been generally weak. Third, we
identified opportunities to enhance both the process and data inputs for
conducting stress testing in the future. Finally, SCAP demonstrated the
importance of robust coordination and communication among the
different regulators as an integral part of any effective supervisory
process. By incorporating these lessons going forward, regulators will be
able to enhance their ability to efficiently and effectively oversee the risktaking in the banking industry.

52

Pub. L. No. 111-203, 124 Stat. 1376 (2010); Pub. L. No. 111-24, 123 Stat. 1734 (2009).

Page 40

GAO-10-861 Lessons Learned from Bank Stress Test

SCAP’s Transparency
Helped Bolster Market
Confidence, but the
Federal Reserve Has Yet to
Implement a Plan to
Incorporate Greater
Transparency into the
Supervisory Process

As stated earlier and as agreed generally by market participants, the public
release of the SCAP design and results helped restore confidence in the
financial system during a period of severe turmoil. Some agency officials
stated that their experience in implementing SCAP suggested that greater
transparency would also be beneficial in the supervisory process. In recent
statements, the chairman and a governor of the Federal Reserve have both
stated that, while protecting the confidentiality of firm-specific proprietary
information is imperative, greater transparency about the methods and
conclusions of future stress tests could benefit from greater scrutiny by
the public. 53 The Federal Reserve governor also noted that feedback from
the public could help to improve the methodologies and assumptions used
in the supervisory process. In addition, they noted that more transparency
about the central bank’s activities overall would ultimately enhance
market discipline and that the Federal Reserve is looking at ways to
enhance its disclosure policies. 54
Consistent with the goal of greater transparency, we previously
recommended that the Federal Reserve consider periodically disclosing to
the public the aggregate performance of the 19 BHCs against the SCAP
estimates for the 2-year forecast period. 55 Subsequently, the chairman and
a governor of the Federal Reserve have publicly disclosed 2009 aggregate
information about the performance of the 19 BHCs based on the Federal
Reserve’s internal tracking. As the 2-year SCAP period comes to a close at
the end of 2010, completing a final analysis that compares the performance
of BHCs with the estimated performance under the more adverse
economic scenario would be useful; however, at the time of the review,
Federal Reserve officials told us that they have not decided whether to
conduct and publicly release any type of analysis. Given that the chairman
and a governor of the Federal Reserve have already publicly disclosed
some aggregate BHC performance against the more adverse scenario for
2009, providing the

53

Ben S. Bernanke, “The Supervisory Capital Assessment Program—One Year Later,”
speech delivered at the Federal Reserve Bank of Chicago 2010 46th Annual Conference on
Bank Structure and Competition (Chicago, Illinois, May 6, 2010). Daniel K. Tarullo,
“Lessons from the Crisis Stress Tests,” speech delivered at the Federal Reserve Board 2010
International Research Forum on Monetary Policy (Washington, D.C., Mar. 26, 2010).

54

Daniel K. Tarullo, “Involving Markets and the Public in Financial Regulation,” speech
delivered at the Council of Institutional Investors Meeting (Washington, D.C., Apr. 13,
2010). Bernanke “The Supervisory Capital Assessment Program—One Year Later” (2010).
55

GAO, Troubled Asset Relief Program: June 2009 Status of Efforts to Address
Transparency and Accountability Issues, GAO-09-658 (Washington, D.C.: June 17, 2009).

Page 41

GAO-10-861 Lessons Learned from Bank Stress Test

2-year results would provide the public with consistent and reliable
information from the chief architect of the stress test that could be used to
further establish the importance of understanding such tests and consider
lessons learned about the rigor of the stress test estimates.
Increasing transparency in the bank supervisory process is a more
controversial issue to address. Supervisory officials from OCC (including
the then Comptroller) and the Federal Reserve question the extent to
which greater transparency would improve day-to-day bank supervision.
And, some BHCs we interviewed also were against public disclosure of
future stress tests results. They noted that SCAP was a one-time stress test
conducted under unique circumstances. Specifically, during the financial
crisis, Treasury had provided a capital backstop for BHCs that were
unable to raise funds privately. They expressed concern that public
disclosure of certain unfavorable information about individual banks in a
normal market environment could cause depositors to withdraw funds en
masse creating a “run” on the bank. In addition, banks that appear weaker
than their peers could be placed at a competitive disadvantage and may
encourage them to offer more aggressive rates and terms for new
depositors, thereby increasing their riskiness and further affecting their
financial stability. While these concerns are valid and deserve further
consideration, they have to be weighed against the potential benefits of
greater transparency about the financial health of financial institutions and
the banking system in general to investors, creditors, and counterparties.
The Dodd-Frank Act takes significant steps toward greater transparency.
For example, the act requires the Federal Reserve to perform annual stress
tests on systematically significant institutions and publicly release a
summary of results. Also, the act requires each of the systematically
significant institutions to publicly report the summary of internal stress
tests semiannually. 56 Given comments by its senior leadership, the Federal
Reserve is willing to engage in a constructive dialogue about creating a
plan for greater transparency that could benefit the entire financial sector.
The other federal bank regulators—FDIC, OCC, and the Office of the
Thrift Supervision—are also critical stakeholders in developing such a
plan. While Federal Reserve officials have discussed possible options for
increasing transparency, the regulators have yet to engage in a formal
dialogue about these issues and have not formalized a plan for the public

56

The act also establishes the Financial Stability Oversight Council and Treasury’s Office of
Financial Research in order to further the goals of effective systemic risk measurement.

Page 42

GAO-10-861 Lessons Learned from Bank Stress Test

disclosure of regulatory banking information or developed a plan for
integrating public disclosures into the ongoing supervisory process.
Without a plan for reconciling these divergent views and for incorporating
steps to enhance transparency into the supervisory process and practices,
including the public disclosure of certain information, bank regulators
may miss a significant opportunity to enhance market discipline by
providing investors, creditors, and counterparties with information such
as bank asset valuations.

Limited Use and Weak
Oversight of BHCs’ Stress
Tests Prior to SCAP
Highlights the Need for
More Rigorous Testing and
Better Monitoring of Tests

SCAP highlighted that the development and utilization of BHCs’ stress
tests were limited. Further, BHC officials noted that they failed to
adequately stress test for the effects of a severe economic downturn
scenario and did not test on a firmwide basis or test frequently enough. We
also found that the regulator’s oversight of these tests were weak,
reinforcing the need for more rigorous and firmwide stress testing, better
risk governance processes by BHCs, and more vigorous oversight of BHCs’
stress tests by regulators. Going forward, as stress tests become a
fundamental part of oversights of individual banks and the financial
system, more specific guidance needs to be developed for examiners.
BHCs and regulators stated that they are taking steps to address these
shortcomings.

BHCs Generally Did Not
Perform Firmwide Stress Tests
Prior to SCAP

Prior to SCAP, many BHCs generally performed stress tests on individual
portfolios, such as commercial real estate or proprietary trading, rather
than on a firmwide basis. SCAP led some institutions to look at their
businesses in the aggregate to determine how losses would affect the
holding company’s capital base rather than individual portfolios’ capital
levels. As a result, some BHC officials indicated that they had begun
making detailed assessments of their capital adequacy and risk
management processes and are making improvements. Officials from one
BHC noted that before SCAP their financial and risk control teams had run
separate stress tests, but had not communicate or coordinate with each
other about their stress testing activities. Officials from another BHC
noted that their senior management and board of directors were not
actively involved in the oversight of the stress testing process. These
officials said that since participating in SCAP, they have improved in these
areas by institutionalizing the internal communication and coordination
procedures between the financial risk and control teams, and by
increasing communication with senior management and board of directors
about the need for active involvement in risk management oversight,
respectively. These improvements can enhance the quality of the stress
testing process. Moreover, officials of BHCs that were involved in ongoing

Page 43

GAO-10-861 Lessons Learned from Bank Stress Test

bank mergers during the SCAP process credited SCAP with speeding up of
the conversion process of the two institutions’ financial systems since the
BHCs’ staff had to work together to be able to quickly provide, among
other things, the aggregate asset valuations and losses of the combined
firm’s balance sheets to the regulators.
BHC officials also stated that their stress tests would take a firmwide
view, that is, taking into account all business units and risks within the
holding company structure and would include updates of the economic
inputs used to determine potential losses and capital needs in adverse
scenarios. One BHC noted that it had developed several severe stress
scenarios for liquidity because the recent financial crisis had shown that
liquidity could deteriorate more quickly than capital, endangering a
company’s prospects for survival. This danger became evident in the
failures of major financial institutions during the recent financial crisis—
for example, IndyMac Bank, Lehman Brothers, and Bear Stearns.

BHCs Did Not Sufficiently
Stress Their Portfolios for
Unexpected Losses Prior to
SCAP

Officials from many SCAP BHCs and the Federal Reserve noted that
internal bank stress test models generally did not use macroeconomic
assumptions and loss rates inputs as conservative as those used in the
SCAP stress test. According to Federal Reserve officials, using the SCAP
macroeconomic assumptions, most of the 19 BHCs that took part in SCAP
initially determined that they would not need additional capital to weather
the more adverse scenario. However, the SCAP test results subsequently
showed that more than half of them (10 of 19) did need to raise capital to
meet the SCAP capital buffer requirements. Some BHCs indicated that
future stress tests would be more comprehensive than SCAP. BHCs can
tailor their stress test assumptions to match their specific business
models, while SCAP generally used a one-size-fits-all assumptions
approach. For example, some BHCs noted that they use macroeconomic
inputs (such as disability claims, prolonged stagflation, or consumer
confidence) that were not found in the SCAP stress test.
Although the Federal Reserve has required BHCs to conduct stress tests
since 1998, officials from several BHCs noted that their institutions had
not conducted rigorous stress tests in the years prior to SCAP, a statement
that is consistent with regulatory findings during the same period. To some
degree, this lack of rigorous testing reflected the relatively good economic
times that preceded the financial crisis. According to one credit rating
agency and a BHC, stress test assumptions generally tend to be more
optimistic in good economic times and more pessimistic in bad economic
times. In addition, one BHC noted that it had conducted stress tests on and
off for about 20 years, but usually only as the economy deteriorated. To

Page 44

GAO-10-861 Lessons Learned from Bank Stress Test

address this issue, many BHC officials said that they have incorporated or
are planning to incorporate more conservative inputs into their stress test
models and are conducting more rigorous, firmwide stress testing more
frequently.

Regulators Required Stress
Tests Prior to SCAP, but
Oversight Was Limited

Although regulators’ guidelines have required for over 10 years that
financial institutions use stress tests to assess their capacity to withstand
losses, we found that regulators’ oversight of these tests had been limited.
Horizontal examinations by the regulators from 2006 through 2008
identified multiple weaknesses in institutions’ risk management systems,
including deficiencies in stress testing. Areas of weaknesses found during
examination included that the BHCs’ stress testing of their balance sheets
lacked severity, were not performed frequently enough, and were not done
on a firmwide basis. Also, it was found that BHCs’ risk governance process
lacked the active and effective involvement of BHC senior management
and board of directors. The SCAP stress test and the financial crisis
revealed the same shortcomings in BHCs’ risk management and stress
testing practices.
However, we previously found that regulators did not always effectively
address these weaknesses or in some cases fully appreciate their
magnitude. 57 Specifically, regulators did not take measures to push
forcefully for institutions to better understand and manage risks in a
timely and effective manner. In addition, according to our discussions with
some SCAP participants, oversight of these tests through routine
examinations was limited in scope and tended to be discretionary. For
example, regulators would review firms’ internal bank stress tests of
counterparty risk and would make some suggestions, but reviews of these
tests were done at the discretion of the individual supervisory team and
were not consistently performed across teams. Even though BHCs have
for many years performed stress tests to one degree or another, they have
not been required to report the results of their testing to the Federal
Reserve unless it specifically requested the information.
The Federal Reserve recently issued a letter to the largest banking
organizations outlining its view on good practices with respect to the use
of stress testing in the context of internal capital adequacy assessment

57

GAO, Financial Regulation: Review of Regulators’ Oversight of Risk Management
System at a Limited Number of Large, Complex Financial Institutions, GAO-09-499T
(Washington, D.C.: Mar. 18, 2009).

Page 45

GAO-10-861 Lessons Learned from Bank Stress Test

practices (ICAAP). For example, some areas highlighted in the letter
include how frequent a stress test should be performed, the minimum time
frame that the test should cover, documentation of the process,
involvement of senior management and board of directors, and types of
scenarios and risks to include in such tests. Some BHC officials believed
that stress testing would become an integral part of future risk
management practices and noted that SCAP helped them see how bank
examiners would want them to stress their portfolios in the future. In
anticipation of future action by regulators, many BHCs were designing at
least part of their stress tests along the lines of SCAP. However, a few BHC
officials hoped that future stress tests would not be performed in the same
manner as SCAP, with the largest institutions tested simultaneously in a
largely public setting, but rather as part of the confidential supervisory
review process.

Regulatory Oversight Is to
Focus on More Rigorous Stress
Testing, but Examiners Need
More Specific Guidance

Federal Reserve officials stated that going forward, stress tests will
become a fundamental part of the agency’s oversight of individual banks
and the financial system. As a result of SCAP, Federal Reserve officials
stated that they are placing greater emphasis on the BHCs’ internal capital
adequacy planning through their ICAAP. This initiative is intended to
improve the measurement of firmwide risk and the incorporation of all
risks into firms’ capital planning assessment and planning processes. In
addition to enhanced supervisory focus on these practices across BHCs,
stress testing is also a key component of the Basel II capital framework
(Pillar 2). 58 Under Pillar 2, supervisory review is intended to help ensure
that banks have adequate capital to support all risks and to encourage that
banks develop and use better risk management practices. All BHCs,
including those adopting Basel II, must have a rigorous process of
assessing capital adequacy that includes strong board and senior
management oversight, comprehensive assessment of risks, rigorous
stress testing and scenario analyses, validation programs, and independent
review and oversight. In addition, Pillar 2 requires supervisors to review
and evaluate banks’ internal capital adequacy assessments and monitor
compliance with regulatory capital requirements. The Federal Reserve

58

Basel II is an international risk-based capital framework that aims to align minimum
capital requirements with enhanced risk measurement techniques and to encourage banks
to develop a more disciplined approach to risk management. It was organized with three
main principles of capital known as pillars: Pillar 1 relates to minimum capital
requirements. Pillar 2 relates to the supervisory review of an institution’s internal
assessment process and capital adequacy relative to the institution’s overall risk profile.
Pillar 3 relates to the effective use of disclosure to strengthen market discipline as a
complement to supervisory efforts.

Page 46

GAO-10-861 Lessons Learned from Bank Stress Test

wants the large banks to conduct this work for themselves and report their
findings to their senior management and boards of directors. According to
Federal Reserve officials, for BHCs to satisfy the totality of expectations
for ICAAP it may take 18 to 24 months, partly because the BHCs are taking
actions to enhance practices where needed—including with respect to the
use of stress testing and scenario analyses in internal capital
assessments—and the Federal Reserve then needs to evaluate these
actions across a relatively large number of BHCs.
In addition, the Federal Reserve is finalizing guidance for examiners to
assess the capital adequacy process, including stress testing, for BHCs.
Examiners are expected to evaluate how BHCs’ stress tests inform the
process for identifying and measuring risk and decisions about capital
adequacy. Federal Reserve officials stated that examiners are expected to
look closely at BHCs’ internal stress test methodologies and results. In a
letter to BHCs, the Federal Reserve also emphasized that institutions
should look at time frames of 2 or more years and considers losses
firmwide. It also suggested that BHCs develop their own stress test
scenarios and then review these scenarios and the results for appropriate
rigor and quantification of risk.
While these are positive steps, examiners do not have specific criteria for
assessing the quality of these tests. For example, the Federal Reserve has
not established criteria for assessing the severity of the assumptions used
to stress BHCs’ balance sheets. The Federal Reserve officials stated that
they intend to have technical teams determine the type of criteria that will
be needed to evaluate these assumptions, but they are in the early
planning stages. Development of such criteria will be particularly helpful
in ensuring the effective implementation of the stress test requirements
under the Dodd-Frank Act. Without specific criteria, Federal Reserve
examiners will not be able to ensure the rigor of BHCs’ stress tests—an
important part of the capital adequacy planning. Furthermore, the absence
of such guidance could lead to variations in the intensity of these
assessments by individual examiners and across regional districts.

Risk Identification and
Assessment Infrastructure
Needs to be Upgraded to
Improve Oversight

Following SCAP, regulatory and BHC officials we met with identified
opportunities to enhance both the process and data inputs for conducting
stress testing in the future. This would include processes for obtaining,
analyzing, and sharing data and capabilities for data modeling and
forecasting, which potentially could increase the Federal Reserve’s
abilities to assess risks in the banking system. According to the Federal
Reserve, an essential component of this new system will be a quantitative

Page 47

GAO-10-861 Lessons Learned from Bank Stress Test

surveillance mechanism for large, complex financial institutions that will
combine a more firmwide and multidisciplinary approach for bank
supervision. This quantitative surveillance mechanism will use supervisory
information, firm-specific data analysis, and market-based indicators to
identify developing strains and imbalances that may affect multiple
institutions, as well as emerging risks within specific institutions. This
effort by the Federal Reserve may also improve other areas of supervision
which rely on data and quantitative analysis, such as assessing the process
used by BHC’s to determine their capital adequacy, forecasting revenue,
and assessing and measuring risk, which is critical to supervising large,
complex banks. Officials at the Federal Reserve told us that examiners
should be analyzing BHC performances versus their stress test projections
to provide insight into the agency’s loss forecasting approach. Moreover,
Federal Reserve officials stated that they are always looking to increase
their analytical capabilities, and they have recently implemented a new
governance structure to address some of their management information
infrastructure challenges. However, not enough time has passed to
determine the extent to which such measures will improve banking
supervision.
In addition, some other deficiencies were found in the data reported to the
Federal Reserve by BHCs using the Y-9C, as well as the Federal Reserve’s
ability to analyze the risk of losses pertaining to certain portfolios that
were identified during the SCAP stress test. This led the Federal Reserve
to develop a more robust risk identification and assessment infrastructure
including internally developed models or purchased analytical software
and tools from data vendors. Going forward, such models and analytics
would facilitate improved risk identification and assessment capabilities
and oversight, including the oversight of systemic risk. Moreover, a risk
identification and assessment system that can gauge risk in the banking
sector by collecting data on a timelier basis is necessary to better ensure
the safety and soundness of the banking industry. Specific areas in which
data collection and risk identification and assessment could be enhanced
include mortgage default modeling to include more analysis of
nontraditional mortgage products, counterparty level exposures, country
and currency exposures, and commodity exposures. An example of where
the Federal Reserve used SCAP to significantly upgrade its ability to
assess risks across large BHCs is the development of a system that
allowed BHCs to submit their securities positions and market values at a
fixed date and apply price shocks. This process was enhanced during

Page 48

GAO-10-861 Lessons Learned from Bank Stress Test

SCAP to facilitate the stress analysis of securities portfolios held by SCAP
BHCs. 59 This system allowed the Federal Reserve to analyze approximately
100,000 securities in a relatively short time period. The Federal Reserve
intends to continue using this database to receive and analyze updated
positions from BHCs.
With other portfolios, the Federal Reserve contracted with outside data
and analytical systems providers. For multifamily loan portfolios, nonfarm
loans, and nonresidential loans with a maturity beyond 2 years, all of
which are subsets of commercial and industrial loans or commercial real
estate portfolios, the Federal Reserve used internal models and purchased
an outside vendor service that allowed it to estimate losses for these
portfolios. For the remaining commercial portfolios, the Federal Reserve
used different existing models found at both the Federal Reserve and
Federal Reserve district banks and new models developed to meet the
needs of SCAP. When analyzing BHCs’ mortgage portfolios, the consumer
loans Supervisory Analytical and Advisory Team provided templates to the
BHCs to collect granular data for such analysis, allowing the system to
separate BHCs’ mortgage portfolios into much more granular tranches
than would be possible using data from regulatory filings. The Federal
Reserve further used data from various sources, including a large
comprehensive loan-level database of most mortgages that have been
securitized in the United States to assist in developing its own loss
estimates to benchmark against the BHCs’ proprietary estimates.
These examples point to enhancements in the ability to assess risks to
individual firms and across the banking sector that resulted from the SCAP
stress test. The Federal Reserve has made clear that it views many of these
innovations in its ability to assess and model risks and potential losses as
permanent additions to its toolkit, and has also recognized the need for
more timely and granular information to improve its supervision of BHCs
and other institutions. However, the extent to which these models and
tools will be distributed across the Federal Reserve district banks and
other federal banking regulators is unclear. In addition, as the stress test
applied to trading positions was limited to those BHCs that held trading
positions of at least $100 billion as of February 20, 2009, the Federal
Reserve has not indicated that it will roll out its new system to BHCs with

59

These portfolios were the only ones tested under the SCAP for which the positions were
taken as of a different date than December 31, 2008. Positions were taken as of February
20, 2009, as it was both more relevant to the actual risk exposure of the BHCs at the time of
SCAP and easier for the BHCs to provide.

Page 49

GAO-10-861 Lessons Learned from Bank Stress Test

smaller trading positions. The Federal Reserve has taken steps to maintain
and enhance the tools and data used during SCAP. Further, improving the
Federal Reserve’s financial data collection and supervisory tools will
require additional resources, training for bank examiners, coordination in
the dissemination of new infrastructure across all U.S. financial regulators,
and, according to a Federal Reserve governor, would benefit from relief
from the Paperwork Reduction Act of 1980 as well.
The Federal Reserve lacks a complete plan on how it will achieve
permanent improvements in its risk identification and assessment
infrastructure, but according to officials, such a plan is in development.
The Federal Reserve has finalized a plan that describes a governance
structure for overseeing large, complex financial organizations. The plan
defines the roles and responsibilities of various committees and teams
within the Federal Reserve that will carry out its supervisory
responsibilities over these organizations. However, further planning is
needed to incorporate lessons learned from SCAP for addressing data and
modeling gaps that existed prior to the crisis and a structure for
disseminating improvements to risk identification and assessment.
Specifically, this plan will also be critical to addressing improvements to
data and modeling infrastructure in supervising not only large financial
holding companies but also smaller institutions. A fully developed plan
would also consider how to disseminate data, models, and other
infrastructure to the entire Federal Reserve System and bank regulatory
agencies, as well as the newly established Financial Stability Oversight
Council and Treasury’s Office of Financial Research. Without such a plan,
the agency runs the risk of not optimizing its oversight responsibilities,
especially in light of its new duties as the systemic risk regulator under the
Dodd-Frank Act.

More Coordination and
Communication across
Regulators Is Critical for
Understanding Risks to
Individual Institutions and
Financial Markets

Another critical lesson from SCAP was the need for robust coordination
and communication among the regulators in examining large, complex
financial institutions. Officials from the regulatory agencies and BHCs
stated that the degree of cooperation among the SCAP regulators was
unprecedented and improved the understanding of the risks facing the
individual BHCs and the financial market. Such coordination and
communication will become increasingly important as banking regulators
increase their oversight role. Even with recent major reform to the
financial regulatory structure, multiple regulatory agencies continue to
oversee the banking industry, and regulators will need to prioritize efforts
to promote coordination and communication among staff from these

Page 50

GAO-10-861 Lessons Learned from Bank Stress Test

agencies so that emerging problematic issues affecting the financial
industry are identified in a timely manner and effectively addressed.
Going forward, based on our discussions with various SCAP participants
and statements by Federal Reserve officials, including the chairman, the
regulators’ experience with SCAP is anticipated to lead to the expanded
use of horizontal examinations and multidisciplinary staff that will require
extensive interagency coordination. Horizontal examinations may involve
multiple regulators and underscore the importance of effective
coordination and communication.
Currently, regulators are conducting horizontal examinations of internal
processes that evaluate the capital adequacy at the 28 largest U.S. BHCs.
Their focus is on the use of stress testing and scenario analyses in ICAAP,
as well as how shortcomings in fundamental risk management practices
and governance and oversight by the board of directors for these
processes could impair firms’ abilities to estimate their capital needs.
Regulators recently completed the initial phase of horizontal examinations
of incentive compensation practices at 25 large U.S. BHCs. As part of this
review, each organization was required to submit an analysis of
shortcomings or “gaps” in its existing practices relative to the principles
contained in the proposed supervisory guidance issued by the Federal
Reserve in the fall of 2009 as well as plans—including timetables—for
addressing any weaknesses in the firm’s incentive compensation
arrangements and related risk-management and corporate governance
practices. In May 2010, regulators provided the banking organizations
feedback on the firms’ analyses and plans. These organizations recently
submitted revised plans to the Federal Reserve for addressing areas of
deficiencies in their incentive compensation programs. In a June 2010
press release, the Federal Reserve noted that to monitor and encourage
improvements in compensation practices by banking organizations, its
staff will prepare a report after 2010 on trends and developments in such
practices at banking organizations.
Our prior work has found that coordination and communication among
regulatory agencies is an ongoing challenge. 60 For example, in 2007, OCC
onsite examiners, as well as officials in headquarters, told us that

60

GAO, Financial Market Regulation: Agencies Engaged in Consolidated Supervision
Can Strengthen Performance Measurement and Collaboration, GAO-07-154 (Washington,
D.C.: Mar. 15, 2007).

Page 51

GAO-10-861 Lessons Learned from Bank Stress Test

coordination issues hampered the Federal Reserve’s horizontal
examinations. Also, in 2007, a bank told us that it had initially received
conflicting information from the Federal Reserve, its consolidated
supervisor, and the OCC, its primary bank supervisor, regarding a key
policy interpretation. Officials from the bank also noted that when the
Federal Reserve collected information, it did not coordinate with OCC, the
primary bank examiner of the lead bank, resulting in unnecessary
duplication. We noted that to improve oversight in the future, regulators
will need to work closely together to expedite examinations and avoid
such duplications.
Since the SCAP stress test was concluded, the following examples
highlight ongoing challenges in coordination and communication:
•

Officials from OCC and FDIC indicated that they were not always involved
in important discussions and decisions. For example, they were not
involved in the decision to reduce GMAC’s SCAP capital requirement, even
though they were significantly involved in establishing the original capital
requirement. Also, FDIC noted that it was excluded from such decision
even though it is the primary federal bank regulator for GMAC’s retail
bank (Ally Bank).

•

The Federal Reserve held an internal meeting to discuss lessons learned
from SCAP, but has yet to reach out to the other SCAP regulators. The
OCC and FDIC told us that they had not met with the Federal Reserve as a
group to evaluate the SCAP process and document lessons learned. As a
result, the FDIC and OCC did not have an opportunity to share their views
on what aspects of SCAP worked and did not work, as well as any
potential improvements that can be incorporated into future horizontal
reviews or other coordinated efforts.

•

In the recent horizontal examinations, both FDIC and OCC noted that the
interagency process for collaboration—especially in the initial design
stages—was not as effective as it was for SCAP. OCC commented that
more collaboration up front would have been preferable. Also, FDIC
stated that the Federal Reserve did not include it in meetings to formulate
aggregate findings for the horizontal examination of incentive
compensation programs, and it experienced difficulties in obtaining
aggregate findings from the Federal Reserve. The Federal Reserve
commented that the FDIC was involved in the development of findings for
those organizations that control an FDIC-supervised subsidiary bank and
that FDIC has since been provided information on the findings across the
full range of organizations included in the horizontal review, the majority
of which do not control an FDIC-supervised subsidiary bank.

Page 52

GAO-10-861 Lessons Learned from Bank Stress Test

These continued challenges in ensuring effective coordination and
communication underscore the need for sustained commitment and effort
by the regulators to ensure the inclusion of all relevant agencies in key
discussions and decisions regarding the design, implementation, and
results of multiagency horizontal examinations. As the SCAP process has
shown, active participation by all relevant regulators can strengthen
approaches used by examiners in performing their supervisory activities.
Without continuous coordination and communication, the regulators will
miss opportunities to leverage perspectives and experiences that could
further strengthen the supervision of financial institutions, especially
during horizontal examinations of financial institutions.

Conclusions

Publicly reporting a comparison of the actual performance of the SCAP
BHCs and the estimated performance under a more adverse scenario
provides insights into the financial strength of the nation’s largest BHCs.
Senior Federal Reserve officials have publicly disclosed select aggregate
information about the performance of the 19 BHCs consistent with the
recommendation in our June 2009 report. Specifically, we recommended
that the Federal Reserve consider periodically disclosing to the public the
performance of the 19 BHCs against the SCAP estimates during the 2-year
period. However, the Federal Reserve has yet to commit to completing a
final analysis that compares the BHCs’ actual performance with the
estimated performance under SCAP’s more adverse economic scenario for
the entire 2-year period and making this analysis public. Such an analysis
is important for the market and BHCs to assess the rigor of the stress test
methodology. Publicly releasing the results also would allow the public to
gauge the health of the BHCs that participated in SCAP, which is a strong
proxy for the entire U.S. banking industry. And public disclosure of this
analysis could act as a catalyst for a public discussion of the value of
effective bank risk management and enhance confidence in the regulatory
supervision of financial institutions.
The public release of the stress test methodology and results helped
improve market confidence in the largest BHCs during the recent financial
crisis and provided an unprecedented window into bank supervision
process. Subsequently, the Chairman of the Federal Reserve and a Federal
Reserve governor have publicly stated that greater transparency should be
built into the supervisory process and that feedback from the public could
help increase the integrity of the supervisory process. Increased
transparency can also augment the information that is available to
investors and counterparties of the institutions tested and enhance market
discipline. Despite these statements, the Federal Reserve and other bank

Page 53

GAO-10-861 Lessons Learned from Bank Stress Test

regulators have yet to start a formal dialogue about this issue, nor have
they developed a plan for integrating public disclosures into the ongoing
supervisory process. Such a plan could detail the types of information that
would benefit the markets if it were publicly released; the planned
methodology for the stress tests, including assumptions; the frequency
with which information would be made public; and the various means of
disseminating the information. Taking into account the need to protect
proprietary information and other market-sensitive information would be
an important part of such a plan. While regulators will undoubtedly face
challenges in determining how best to overcome skepticism about the
potential effects on the financial markets of disclosing sensitive
information on the financial health of banks, the Dodd-Frank Act requires
that the Federal Reserve and certain banks publicly release a summary of
results from periodic stress tests. Without a plan for enhancing the
transparency of supervisory processes and practices, bank regulators may
miss a significant opportunity to further strengthen market discipline and
confidence in the banking industry by providing investors, creditors, and
counterparties with useful information.
The SCAP stress test shed light on areas for further improvement in the
regulators’ bank supervision processes, including oversight of risk
management practices at BHCs. Prior to SCAP, regulatory oversight of
stress tests performed by the BHCs themselves was ineffective.
Specifically, although regulators required stress tests, the guidelines for
conducting them were more than a decade old, and the individual banks
were responsible for designing and executing them. The Federal Reserve’s
reviews of the internal stress tests were done at the discretion of the
BHCs’ individual supervisory teams and were not consistently performed.
Further, even though BHCs performed stress tests, they were not required
to report the results of their stress testing to the Federal Reserve without a
specific request from regulators. Post-SCAP, however, the Federal Reserve
has stated that stress testing will now be a fundamental part of their
oversight of individual banks. The Federal Reserve expects to play a more
prominent role in reviewing assumptions, results, and providing input into
the BHCs’ risk management practices. While the Federal Reserve has
begun to take steps to augment its oversight, currently Federal Reserve
examiners lack specific criteria for assessing the severity of BHCs’ stress
tests. Without specific criteria, Federal Reserve examiners will not be able
to ensure the rigor of BHCs’ stress tests. Furthermore, the absence of such
criteria could lead to variations in the intensity of these assessments by
individual examiners and across regional districts.

Page 54

GAO-10-861 Lessons Learned from Bank Stress Test

The experience with SCAP also showed that regulators needed relevant
and detailed data to improve oversight of individual banks and to identify
and assess risks. As the Federal Reserve and the other regulators conduct
more horizontal reviews, they will need a robust plan for quantitatively
assessing the risk in the banking sector. Collecting timely data for the
annual stress testing and other supervisory actions will be critical in order
to better ensure the safety and soundness of the banking industry. The
Federal Reserve has finalized a plan that describes a governance structure
for overseeing large, complex financial organizations. However, further
planning is needed to incorporate lessons learned from SCAP for
addressing data and modeling gaps and a structure for disseminating
improvements to risk identification and assessment. Further, efforts to
improve the risk identification and assessment infrastructure will need to
be effectively coordinated with other regulators and the newly established
Financial Stability Oversight Council and Treasury’s Office of Financial
Research in order to ensure an effective systemwide risk assessment.
Without fully developing a plan that can identify BHCs’ risks in time to
take appropriate supervisory action, the Federal Reserve may not be wellpositioned to anticipate and minimize future banking problems and ensure
the soundness of the banking system.
Despite the positive coordination and communication experience of the
SCAP stress test, developments since the completion of SCAP have
renewed questions about the effectiveness of regulators’ efforts to
strengthen their coordination and communication. For example, on
important issues, such as finalizing GMAC’s SCAP capital amount, the
Federal Reserve chose not to seek the views of other knowledgeable bank
regulators. While the Dodd-Frank Act creates formal mechanisms that
require coordination and communication among regulators, the
experiences from SCAP point to the need for a sustained commitment by
each of the banking regulators to enhance coordination and
communication. In particular, ensuring inclusion of relevant agencies in
key discussions and decisions regarding the design, implementation, and
results of multiagency horizontal examinations will be critical. If
regulators do not consistently coordinate and communicate effectively
during horizontal examinations, they run the risk of missing opportunities
to leverage perspectives and experiences that could further strengthen
bank supervision.

Recommendations

To gain a better understanding of SCAP and inform the use of similar
stress tests in the future, we recommend that the Chairman of the Federal
Reserve direct the Division of Banking Supervision and Regulation to:

Page 55

GAO-10-861 Lessons Learned from Bank Stress Test

•

Compare the performance of the 19 largest BHCs against the more adverse
scenario projections following the completion of the 2-year period covered
in the SCAP stress test ending December 31, 2010, and disclose the results
of the analysis to the public.
To leverage the lessons learned from SCAP to the benefit of other
regulated bank and thrift institutions, we recommend that the Chairman of
the Federal Reserve in consultation with the heads of the FDIC and OCC
take the following actions:

•

•

Develop more specific criteria to include in its guidance to examiners for
assessing the quality of stress tests and how these tests inform BHCs’
capital adequacy planning. These guidelines should clarify the stress
testing procedures already incorporated into banking regulations and
incorporate lessons learned from SCAP.

•

Fully develop its plan for maintaining and improving the use of data, risk
identification and assessment infrastructure, and requisite systems in
implementing its supervisory functions and new responsibilities under the
Dodd-Frank Act. This plan should also ensure the dissemination of these
enhancements throughout the Federal Reserve System and other financial
regulators, as well as new organizations established in the Dodd-Frank
Act.

•

Agency Comments
and Our Evaluation

Follow through on the Federal Reserve’s commitment to improve the
transparency of bank supervision by developing a plan that reconciles the
divergent views on transparency and allows for increased transparency in
the regular supervisory process. Such a plan should, at a minimum, outline
steps for releasing supervisory methodologies and analytical results for
stress testing.

Take further steps to more effectively coordinate and communicate among
themselves. For example, ensuring that all applicable regulatory agencies
are included in discussions and decisions regarding the development,
implementation, and results of multiagency activities, such as horizontal
examinations of financial institutions.

We provided a draft of this report to the Federal Reserve, FDIC, OCC,
OTS, and Treasury for review and comment. We received written
comments from the Chairman of the Federal Reserve Board of Governors
and the Assistant Secretary for Financial Stability. These comments are
summarized below and reprinted in appendixes IV and V, respectively. We

Page 56

GAO-10-861 Lessons Learned from Bank Stress Test

also received technical comments from the Federal Reserve, FDIC, OCC,
and Treasury, which we incorporated into the report as appropriate. OTS
did not provide any comments. In addition, we received technical
comments from the Federal Reserve and most of the 19 SCAP BHCs on the
accuracy of our tracking of revenues and losses in 2009 for each of the
SCAP BHCs and incorporated them into the report as appropriate.
In its comment letter, the Federal Reserve agreed with all five of our
recommendations for building on the successes of SCAP to improve bank
supervision. The Federal Reserve noted that our recommendations
generally relate to actions it is currently undertaking or planning to take
under the Dodd-Frank Act. It also cited that in coordination with FDIC and
OCC, it would provide a public assessment of BHCs’ performance relative
to the loss and preprovision net revenue estimates under the more adverse
scenario, taking into account the limitations of such an analysis. For our
remaining recommendations related to increased transparency, examiner
guidance, risk identification and assessment, and coordination and
communication of multiagency activities, the Federal Reserve generally
noted that it has taken step in these areas and will continue to consult
with the FDIC and OCC in implementing our recommendations and its
new responsibilities under the Dodd-Frank Act.
While our report recognizes the steps that the Federal Reserve has taken
related to transparency, examiner guidance, risk identification and
assessment, and coordination and communication of multiagency
activities, these areas warrant ongoing attention. For example, as we note
in the report, while the Federal Reserve is in the process of finalizing
examination guidance for reviewing stress tests, examiners currently do
not have specific criteria for assessing the severity of these tests nor have
they coordinated with the other bank regulators. Until this guidance is
completed, examiners will lack the information needed to fully ensure the
rigor of BHCs’ stress tests, and the Board will not be able to fully ensure
the consistency of the assessment by individual examiners. Our report also
notes the positive coordination and communication experience of the
SCAP stress test, but we continued to find specific instances since the
completion of SCAP that have renewed questions about the effectiveness
of regulators’ efforts to strengthen their coordination and communication.
For instance, while the Federal Reserve included relevant agencies in key
discussions and decisions regarding the design, implementation, and
results of SCAP, we found that the Federal Reserve missed opportunities
to include other bank regulators when planning more recent horizontal
examinations.

Page 57

GAO-10-861 Lessons Learned from Bank Stress Test

Treasury agreed with our report findings, noting that it appreciated our
acknowledgment that SCAP met its goals of providing a comprehensive,
forward-looking assessment of the balance sheet risks of the largest banks
and increasing the level and quality of capital held by such banks. It
further noted that the unprecedented public release of the stress test
results led to an increase in the market confidence in the banking system,
which aided in improving the capital adequacy of the largest banks.

We are sending copies of this report to the appropriate congressional
committees; Chairman of the Federal Reserve, the Acting Comptroller of
Currency, Chairman of the FDIC, the Acting Director of the Office of the
Thrift Supervision, and the Secretary of the Treasury. Also, we are sending
copies of this report to the Congressional Oversight Panel, Financial
Stability Oversight Board, the Special Inspector General for TARP, and
other interested parties. In addition, the report will be available at no
charge on the GAO Web site at http://www.gao.gov.
Should you or your staff have any questions on the matters discussed in
this report, please contact me at (202) 512-8678 or williamso@gao.gov.
Contact points for our Offices of Congressional Relations and Public
Affairs may be found on the last page of this letter. GAO staff who made
key contributions to this report are listed in
appendix VI.

Orice Williams Brown
Director, Financial Markets
and Community Investment

Page 58

GAO-10-861 Lessons Learned from Bank Stress Test

List of Committees
The Honorable Daniel K. Inouye
Chairman
The Honorable Thad Cochran
Vice Chairman
Committee on Appropriations
United States Senate
The Honorable Christopher J. Dodd
Chairman
The Honorable Richard C. Shelby
Ranking Member
Committee on Banking, Housing, and Urban Affairs
United States Senate
The Honorable Kent Conrad
Chairman
The Honorable Judd Gregg
Ranking Member
Committee on the Budget
United States Senate
The Honorable Max Baucus
Chairman
The Honorable Charles E. Grassley
Ranking Member
Committee on Finance
United States Senate
The Honorable David R. Obey
Chairman
The Honorable Jerry Lewis
Ranking Member
Committee on Appropriations
House of Representatives
The Honorable John M. Spratt Jr.
Chairman
The Honorable Paul Ryan
Ranking Member
Committee on the Budget
House of Representatives

Page 59

GAO-10-861 Lessons Learned from Bank Stress Test

The Honorable Barney Frank
Chairman
The Honorable Spencer Bachus
Ranking Member
Committee on Financial Services
House of Representatives
The Honorable Sander M. Levin
Chairman
The Honorable Dave Camp
Ranking Member
Committee on Ways and Means
House of Representatives

Page 60

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix I: Objectives, Scope, and
Methodology

Appendix I: Objectives, Scope, and
Methodology
The objectives of this report were to (1) describe the process used to
design and conduct the stress test and participants views’ of the process,
(2) describe the extent to which the stress test achieved its goals and
compare its estimates with the bank holding companies’ (BHC) actual
results, and (3) identify the lessons regulators and BHCs learned from the
Supervisory Capital Assessment Program (SCAP) and examine how each
are using those lessons to enhance their risk identification and assessment
practices.
To meet the report’s objectives, we reviewed the Board of Governors of
the Federal Reserve System’s (Federal Reserve) The Supervisory Capital
Assessment Program: Design and Implementation (SCAP design and
implementation document) dated April 24, 2009, and The Supervisory
Capital Assessment Program: Overview of Results (SCAP results
document) dated May 7, 2009. We analyzed the initial stress test data that
the Federal Reserve provided to each BHC, the subsequent adjustments
the Federal Reserve made to these estimates, and the reasons for these
adjustments. We reviewed BHC regulatory filings such as the Federal
Reserve’s 2009 Consolidated Financial Statements for Bank Holding
Companies—-FR Y-9C (Y-9C); 1 company quarterly 10-Qs and annual 10-Ks;
speeches and testimonies regarding SCAP and stress testing; BHCs’
presentations to shareholders and earnings reports; bank supervision
guidance issued by the Federal Reserve, Office of the Comptroller of the
Currency (OCC), and the Federal Deposit Insurance Corporation (FDIC);
and documents regarding the impact of SCAP and the financial crisis and

1
The Y-9C is a Federal Reserve reporting form that collects basic financial data from a
domestic BHC on a consolidated basis in the form of a balance sheet, an income statement,
and detailed supporting schedules, including a schedule of off balance-sheet items. The
information is used to assess and monitor the financial condition of BHC organizations,
which may include parent, bank, and nonbank entities. The Y-9C is a primary analytical tool
used to monitor financial institutions between on-site inspections and is filed quarterly as
of the last calendar day of March, June, September, and December. The Federal Reserve
used such format to collect data from the BHCs for purposes of conducting the SCAP
stress test.

Page 61

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix I: Objectives, Scope, and
Methodology

proposed revisions to bank regulation and supervisory oversight. 2 To
further understand these documents and obtain different perspectives on
the SCAP stress test, we interviewed officials from the Federal Reserve,
OCC, FDIC, and the Office of the Thrift Supervision, as well as members of
the multidisciplinary teams created to execute SCAP. 3
We also collected data from SNL Financial—a private financial database
that contains publicly filed regulatory and financial reports, including
those of the BHCs involved in SCAP—in order to compare the BHCs’
actual performance in 2009 against the regulators’ 2-year SCAP loss
estimates and GAO’s 1-year pro rata loss estimates. To obtain additional
background information regarding the tracking of the BHCs, perspectives
on their performance, anticipated loan losses, and the success of SCAP in
achieving its goals, we interviewed relevant officials (e.g., chief risk
officers and chief financial officers) from 11 of the 19 BHCs that
participated in the SCAP stress test. The BHCs we interviewed were the
American Express Company; Bank of America Corporation; The Bank of
New York Mellon Corporation; BB&T Corporation; Citigroup Inc.; GMAC
LLC; 4 The Goldman Sachs Group, Inc.; JPMorgan Chase & Co.; MetLife,
Inc.; Regions Financial Corporation; and Wells Fargo & Company. We
selected these BHCs to reflect differences in size, types of financial
services provided, geographic location, primary bank regulator, and
participation in the Troubled Asset Relief Program (TARP). In addition, we
met with credit rating agency officials from the Standard and Poor’s
Corporation, Moody’s Corporation, and Fitch Ratings Inc. for their

2
For a more detailed discussion about risk-management practices in place during the
market turmoil, see the following reports: Senior Supervisors Group, Observations on Risk
Management Practices during the Recent Market Turbulence (New York, Mar. 6, 2008);
International Monetary Fund, Global Financial Stability Report: Containing Systemic
Risk and Restoring Financial Soundness (Washington, D.C.: April 2008); Financial
Stability Forum, Report of the Financial Stability Forum on Enhancing Market and
Institutional Resilience (April 2008); Institute of International Finance, Final Report of the
IIF Committee on Market Best Practices: Principles of Conduct and Best Practice
Recommendations (July 2008); Credit Risk Management Policy Group III, Containing
Systemic Risk: The Road to Reform (August 2008); Senior Supervisors Group, Risk
Management Lessons from the Global Banking Crisis of 2008 (Oct. 21, 2009); Basel
Committee on Banking Supervision, Principles for Sound Stress Testing Practices and
Supervision (Basel, Switzerland, May 2009); and the President’s Working Group on
Financial Markets, Policy Statement on Financial Market Developments (March 2008).
3

The Office of the Thrift Supervision did not participate in conducting the stress test.

4

On June 30, 2009, GMAC LLC changed its corporate structure and became GMAC Inc., and
on May 10, 2010, GMAC Inc. changed its name to Ally Financial Inc.

Page 62

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix I: Objectives, Scope, and
Methodology

perspective on SCAP and their own stress test practices. To more
completely understand the execution of SCAP, we completed a literature
search of stress tests conducted by others—for example, the Committee
on European Banking Supervisors and the International Monetary Fund.
We also reviewed relevant credit rating agency reports and the reports of
other oversight bodies such as the Congressional Oversight Panel and the
Special Inspector General for the Troubled Asset Relief Program on topics
related to stress testing and TARP. We also reviewed our past work on the
bank supervisory process and SCAP. 5
In addition, to track the actual performance of the 19 BHCs, we collected
data from several sources. We then compared the BHCs’ actual
performance to the December 31, 2008, capital levels presented in SCAP
and the projections made under the more adverse scenario for estimated
losses for loans, securities (available for sale and held to maturity), trading
and counterparty, and resources other than capital to absorb losses. 6 Our
primary source for SCAP estimates was the May 7, 2009, SCAP results
document, which contained the estimates for each of the 19 BHCs and
aggregate data for all BHCs. We also reviewed the confidential April 24,
2009, and May 5, 2009, presentations that the SCAP regulators made to
each of the 19 BHCs to identify estimates of preprovision net revenue
(PPNR) and changes in allowance for loan and lease losses (ALLL) for the
2 years ended 2010. Our primary source for the actual results at the BHCs
was the Federal Reserve’s Y-9C. In doing so, we used the SNL Financial
database to extract data on the Y-9C and the Securities and Exchange
Commission forms 10-K and 10-Q. These data were collected following the
close of the fourth quarter of 2009, the halfway point of the SCAP’s 2-year
time frame.

5
See GAO, Financial Market Regulation: Agencies Engaged in Consolidated Supervision
Can Strengthen Performance Measurement and Collaboration, GAO-07-154 (Washington,
D.C.: Mar. 15, 2007); Financial Regulation: A Framework for Crafting and Assessing
Proposals to Modernize the Outdated U.S. Financial Regulatory System, GAO-09-216
(Washington, D.C.: Jan. 8, 2009); Financial Regulation: Review of Regulators’ Oversight of
Risk Management Systems at a Limited Number of Large, Complex Financial
Institutions, GAO-09-499T (Washington, D.C.: Mar. 18, 2009); and Troubled Asset Relief
Program: June 2009 Status of Efforts to Address Transparency and Accountability
Issues, GAO-09-658 (Washington, D.C.: June 2009). Also, see Congressional Oversight
Panel’s Stress Testing and Shoring Up Bank Capital (June 9, 2009).
6
The BHCs had to maintain a tier 1 capital ratio of at least 6 percent of risk-weighted assets
and a tier 1 common capital ratio of at least 4 percent of risk-weighted assets at the end of
2010. PPNR is defined as net interest income plus noninterest income minus noninterest
expense. Allowance for loan and lease losses is defined as the capital reserve set aside to
cover anticipated losses.

Page 63

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix I: Objectives, Scope, and
Methodology

Since losses were not estimated on a quarter-by-quarter or yearly basis but
projected for the full 2-year period, we assumed that losses and revenue
estimates under the more adverse scenario were distributed at a constant
rate across the projection period. Thus, we compared the actual 2009 year
end values with half of the Federal Reserve’s 2-year SCAP projections.
This methodology has some limitations because losses, expenses,
revenues, and changes to reserves are historically unevenly distributed
and loss rates over a 2-year period in an uncertain economic environment
can follow an inconsistent path. However, the Federal Reserve, OCC,
credit rating agencies, an SNL Financial analyst, and most of the BHCs we
interviewed who are tracking performance relative to SCAP estimates are
also using the same methodology. We assessed the reliability of the SNL
Financial database by following GAO’s best practices for data reliability
and found that the data was sufficiently reliable for our purposes. 7 To
confirm the accuracy of our BHC tracking data, we shared our data with
the Federal Reserve and the 19 SCAP BHCs. We received comments and
incorporated them as appropriate.
Some of the data that we collected were not in a form that was
immediately comparable to the categories used in the SCAP results, and
we had to make adjustments in order to make the comparison. For tier 1
common capital, most asset categories, and resources other than capital to
absorb losses, we had to find a methodology suited to aggregating these
data so that we could compare it to the corresponding SCAP data. For
example, net-charge offs for the various loan categories are broken out
into more subcategories in the Y-9C than those listed in the SCAP results.
In addition, we calculated “Resources Other than Capital to Absorb
Losses” to correspond to the SCAP definition of PPNR minus the change
in ALLL, which required obtaining data from multiple entries within the Y9C. When calculating noninterest expense we removed the line item for
goodwill impairment losses because this item was not included in the
SCAP regulators’ projections. We also used the calculation of a change in
ALLL until December 31, 2009. But the SCAP regulators considered an
increase in ALLL over the full 2-year period to be a drain on resources,
because the provisions made to increase the ALLL balance would not be
available to absorb losses during the 2-year SCAP time frame. This notion
creates a problem in using the formula for 1-year tracking purposes
because an increase in ALLL during 2009 would require provisions for that

7

See GAO, Assessing the Reliability of Computer-Processed Data, Version 1, GAO-02-15G
(Washington, D.C.: September 2002).

Page 64

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix I: Objectives, Scope, and
Methodology

increase, but those added reserves could ultimately be used to absorb
losses during 2010. To maintain consistency, our calculation considers
ALLL increases during 2009 to be a drain on resources, but we recognize
that this money could act as a resource to absorb losses rather than a
drain on those resources.
We faced an additional limitation pertaining to the ALLL calculation and a
challenge with regard to the treatment of trading and counterparty
revenues. In our review of SCAP documentation, we found that SCAP
regulators used two different ALLL calculations—1 calculation for 4 of the
BHCs that included a reserve for off-balance sheet items and another for
the remaining 15 BHCs that did not include off-balance sheet reserves. The
Federal Reserve confirmed that there were two different calculations that
were not adjusted for consistency. In order to be consistent across the
BHCs, we applied the same methodology that the regulators used for 15 of
the BHCs to the 4 that remained. The treatment of trading and
counterparty revenue created a challenge because the data in the Y-9C
includes both customer derived revenue from transactions for BHCs that
operate as broker-dealers and gains (or losses) from proprietary trading
and certain associated expenses. These items are presented only in net
form in the Y-9C. However, for the five BHCs (Bank of America
Corporation; Citigroup, Inc.; Goldman Sachs Group, Inc.; JPMorgan Chase
& Co.; and Morgan Stanley) that had their trading portfolios stressed, the
trading and counterparty line is based on projections of gains (losses)
from proprietary trading, but PPNR (specifically noninterest revenue) is
based on gains from customer derived revenue from transactions for BHCs
that operate as broker-dealers. Because we could not segregate these
items based on the Y-9C, we have included the net amount in both the
trading and counterparty and noninterest income line items. This means
that the net amount of the trading gains or losses as reported in the Y-9C
are included in two places in our tracking table for those five BHCs. For
the remaining 14 BHCs, we included the entire line item in noninterest
income, as that is where it was located in the SCAP projections.
Table 6 shows the items we used to calculate tier 1 capital, asset losses,
PPNR, and ALLL as of December 31, 2009 and the specific sources we
used. We also included specific references to the sources we used. Some
elements within the table required a more detailed aggregation or
calculation and are therefore explained further in tables 7 and 8 below.
For reporting these capital measures and asset balances for the year
ending December 31, 2008, we generally relied on the figures published in
various SCAP documents.

Page 65

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix I: Objectives, Scope, and
Methodology

Table 6: Items Used to Calculate Tier 1 Capital, Asset Losses, PPNR, and ALLL
Actual at 12/31/09 and source
Capital measures
Tier 1 capital

From line 11 (“tier 1 capital”) of Schedule HC-R (page 40) of the FR Y-9C

Tier 1 common capital

See table 7

Risk-weighted assets

From line 62 (“total risk-weighted assets”) of Schedule HC-R (page 43) of the
FR Y-9C

Tier 1 risk-based ratio

Calculated as tier 1 capital divided by risk-weighted assets

Tier 1 common capital ratio

Calculated as tier 1 common capital divided by risk-weighted assets

a

Asset category

First-lien mortgages

See table 8

Second/junior lien mortgages

See table 8

Commercial and industrial loans

See table 8

Commercial real estate loans

See table 8

Credit card loans

See table 8

Securities (available for sale and held to maturity)

Calculated as: total of line 6a (“realized gains (losses) on held-to-maturity
securities”) and line 6b (“realized gains (losses) on available-for-sale
securities”) of Schedule HI (page 2) of the FR Y-9C

Trading and counterparty

For the 5 BHCs that had their trading and counterparty portfolio stressed: line
5c (“trading revenue”) of Schedule HI (page 1) of the FR Y-9C. For all other
BHCs, this line item was left blank.

Other

See table 8

One-time items (included in “Other” in SCAP results) If one-time losses (gains) could be identified, they were located here and
removed from the respective category above. This only applies to State Street
Corporation.
Resources other than capital to absorb losses (Total PPNR less change in ALLL)b
Preprovision net revenue (PPNR)c
Net interest income (expense)

Line 3 (“net interest income”) of Schedule HI (page 1) of the FR Y-9C.

Noninterest income

Line 5m (“total noninterest income”) of Schedule HI (page 2) of the FR Y-9C.

Less noninterest expense

Calculated as: line 7e (“total noninterest expense”) less Line 7c (1) (“goodwill
impairment losses”) of Schedule HI (page 2) of the FR Y-9C.

Change in allowance for loan and lease losses (ALLL)d
ALLL at 12/31/08

Line 1 (“balance most recently reported at the end of previous year”) of Part II
of the Schedule HI-B (page 7) of the 12/31/09 FR Y-9C.

ALLL at 12/31/09

Line 7 (“balance at end of current period”) of Part II of the Schedule HI-B (page
7) of the 12/31/09 FRY-9C.
Source: GAO analysis of Federal Reserve 2009 Y-9C information.
a

Calculated as the total of the losses (gains) below. Categories that were n/a in the SCAP were
included in this total.

b

Calculated as Total PPNR less the change in ALLL.

Page 66

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix I: Objectives, Scope, and
Methodology

c

Calculated as net interest income (expense) plus noninterest income less noninterest expense.

d

Calculated as ALLL at 12/31/09 less ALLL at 12/31/08.

Table 7 shows our methodology for calculating tier 1 common capital,
including the part of the Y-9C in which the data can be found. Currently,
there is no defined regulatory method for calculating tier 1 common
capital, and it is not a required data field for BHCs to file in their Y-9C
submissions. As a result, we developed a formula consistent with the
Federal Reserve’s by reviewing the guidance available in the SCAP design
and implementation and SCAP results documents and consulting with SNL
Financial regarding its methodology.
Table 7: Tier 1 Common Capital Calculation
Location within the FR Y 9-C

Tier 1 common capital calculation

Line 11 of Schedule HC-R (page 40)

Tier 1 capital

Line 23 of Schedule HC (page 12)

Less: perpetual preferred stock and related surplus

Line 6a of Schedule HC-R (page 40)

Less: qualifying Class A noncontrolling (minority) interests in consolidated
subsidiaries

Line 6b of Schedule HC-R (page 40)

Less: qualifying restricted core capital elements (other than cumulative perpetual
preferred stock)

Line 6c of Schedule HC-R (page 40)

Less: qualifying mandatory convertible preferred securities of internationally
active bank holding companies

Line 1 of the “notes to the balance
sheet-other” (page 49)

Less: amount of excess restricted core capital elements included in Schedule HCR, item 10

Line 5 of Schedule HC-R (page 40)

Add: nonqualifying perpetual preferred stock
Total = Tier 1 common capital
Source: Federal Reserve 2009 Y-9C documentation.

Table 8 provides a crosswalk for the asset classification we used to group
the various charge-off categories listed in the Y-9C.

Page 67

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix I: Objectives, Scope, and
Methodology

Table 8: Crosswalk of Y-9C Net Charge-Offs and Asset Classifications to Classifications Used by SCAP
Classification used in GAO analysis
Y-9C classification

Overall category

Primary category

Sub-category

(1) One to four family residential construction loans

Commercial real
estate

Commercial real
estate

Construction

(2) Other construction loans and all land development and
other land loans

Commercial real
estate

Commercial real
estate

Construction

Other

Other loans

n/a

Second/junior lien

Second/junior lien

Home equity line of
credit

(a) Secured by first liens

First lien

First lien

n/a

(b) Secured by junior liens

Second/junior lien

Second/junior lien

Closed-end junior
liens

Commercial real
estate

Commercial real
estate

Multifamily

(1) Loans secured by owner-occupied nonfarm
nonresidential properties

Commercial real
estate

Commercial real
estate

Nonfarm,
nonresidential

(2) Loans secured by other nonfarm nonresidential
properties

Commercial real
estate

Commercial real
estate

Nonfarm,
nonresidential

Other

Other loans

n/a

a. To U.S. banks and other U.S. depository institutions

Other

Other loans

n/a

b. To foreign banks

Other

Other loans

n/a

Other

Other loans

n/a

a. To U.S. addresses (domicile)

Commercial and
industrial

Commercial and
industrial

n/a

b. To non-U.S. addresses (domicile)

Commercial and
industrial

Commercial and
industrial

n/a

1. Loans secured by real estate
a. Construction, land development, and other land loans in
domestic offices:

b. Secured by farmland in domestic offices
c. Secured by one to four family residential properties in
domestic offices:
(1) Revolving, open-end loans secured by one to four family
residential properties under lines of credit
(2) Closed-end loans secured by one to four family
residential properties in domestic offices:

d. Secured by mutlifamily (five or more) residential properties in
domestic offices
e. Secured by nonfarm nonresidential properties in domestic
offices:

f. In foreign offices
2. Loans to depository institutions and acceptances of other
banks:

3. Loans to finance agricultural production and other loans to
farmers
4. Commercial and industrial loans:

5. Loans to individuals for household, family, and other personal
expenditures

Page 68

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix I: Objectives, Scope, and
Methodology

Classification used in GAO analysis
Y-9C classification

Overall category

Primary category

Sub-category

a. Credit cards

Credit cards

Credit cards

n/a

b. Other (includes single payment, installment, all student
loans, and revolving credit plans other than credit cards)

Other

Other consumer

n/a

6. Loans to foreign governments and official institutions

Other

Other loans

n/a

7. All other loans

Other

Other loans

n/a

a. Leases to individuals for household, family, and other
personal expenditures

Other

Other consumer

n/a

b. All other leases

Other

Other loans

n/a

8. Lease financing receivables:

Source: Federal Reserve 2009 Schedule HI-B of the Y-9C.

Note: N/a means not applicable.

To ensure additional comparability with SCAP, we attempted to identify
any unique circumstances that could skew the results. For example, after
we shared our initial tracking estimates with the 19 BHCs, one BHC had
identified an issue with our calculation of tier 1 common capital that
resulted from the way information is reported on the Y-9C. After
discussing the issue with the BHC and verifying their explanation, we
adjusted our calculation to more accurately reflect their position. Another
BHC also had a one-time charge that had been included in the “Other” loss
category, and we decided to segregate this item as a separate line item. We
have also submitted our tracking spreadsheet to the Federal Reserve and
to each BHC to give them an opportunity to provide input and ensure the
accuracy and comparability of our numbers. Appropriate adjustments to
2009 numbers based on information received from the Federal Reserve
and individual BHCs are noted, where applicable, in the tables in appendix
III.
Some items that impact precise comparisons between actual results and
the pro rata estimates are disclosed in our footnotes, rather than as
adjustments to our calculations. For example, the stress test was applied
to loan and other asset portfolios as of December 31, 2008, without
including a calculation for ongoing banking activities. Because the Y-9C
data includes ongoing activity as of the date of the report, the actual
results are slightly different than the performance of the stressed assets as
the BHCs were treated as liquidating concerns rather than going concerns
in the SCAP stress test. Distinguishing between the gains (losses) from
legacy assets and those that resulted from new assets is not possible using
public data. Other examples are that SCAP did not include the impact of
the owned debt value adjustment or one-time items (occurring subsequent

Page 69

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix I: Objectives, Scope, and
Methodology

to SCAP) in their projections of PPNR. 8 As credit default swap spreads
narrowed in 2009, 9 liability values increased at most banks, causing a
negative impact on revenue at those banks that chose to account for their
debt at fair value; but these losses were not included in the SCAP
estimates. One-time items, such as sales of business lines, were also not
included in the SCAP estimates of PPNR, as these events occurred
subsequent to the stress test and, in part, could not be fully predicted as a
part of SCAP. Rather than remove the losses from the owned debt value
adjustments and the gains (or losses) due to one-time items from the
BHCs’ 2009 PPNR results, we disclosed the amounts in footnotes for the
applicable BHCs. We chose this treatment so that PPNR would reflect
actual results at the BHCs, while still disclosing the adjustments needed
for more precise comparability to SCAP.
We identified the TARP status of each of the 19 BHCs that participated in
SCAP by reviewing data from the Treasury’s Office of Financial Stability’s
TARP Transactions Report for the Period Ending September 22, 2010
(TARP Transactions Report) and the SCAP results document. We used the
SCAP results document to identify BHCs that were required to raise
capital. The TARP Transactions Report, was then used to identify the
program under which TARP funds were received (if any), the amount of
funds received, capital repayment date, amount repaid, and warrant
disposition date and to determine whether the warrants were repurchased
or sold by Treasury in a public offering.
To gain a better understanding of future potential losses, we determined
the percentage of BHCs’ total loans that are either nonaccrual or more

8
The owned debt value adjustment is an adjustment made to BHC financial statements if
the BHC chose to value its own debt on a mark-to-market basis rather than book value. As
the BHC’s debt becomes cheaper, this creates a positive impact on its financial statements,
while, as seen in 2009, the debt becomes more expensive, it has a negative impact on the
BHC’s financial statements.
9

A credit default swap spread is one measure of investors’ confidence in the banking
sector. It is an agreement in which a buyer pays a periodic fee to a seller, in exchange for
protection from certain credit events such as bankruptcy, failure to pay debt obligations, or
a restructuring related to a specific debt issuer or issues known as the reference entity.
Therefore, the credit default swap spread, or market price, is a measure of the credit risk of
the reference entity, with a higher spread indicating a greater amount of credit risk. When
the markets’ perception of the reference entity’s credit risk deteriorates or improves, the
spread generally will widen or tighten, respectively.

Page 70

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix I: Objectives, Scope, and
Methodology

than 90 days past due using Y-9C data from the SNL Financial database. 10
We used quarterly data for the period 2007 through 2009 on nonaccrual
loans and past due balances of more than 90 days, for each of the BHCs.
We aggregated the data into the same six loan categories used in SCAP:
first-lien mortgages, second/junior-lien mortgages, commercial and
industrial loans, commercial real estate loans, credit card balances, and
“Other.” (See tables 8 and 9 for details.) Once the data were aggregated,
we divided that data by the applicable total loan balance for each category
at each point in time (i.e., quarterly basis). One limitation is that Y-9C data
were not available for all periods for four of the BHCs (American Express
Company; GMAC LLC; The Goldman Sachs Group, Inc.; and Morgan
Stanley) because they had recently became BHCs. 11 As a result, we did not
include these BHCs in the calculation during those periods where their Y9Cs were not available (fourth quarter of 2008 and earlier for all except
GMAC LLC, which also did not have a Y-9C in the first quarter of 2009).
We collected Y-9C data from the SNL Financial database to calculate the
loan loss rates across BHCs with more than $1 billion of assets and
compare the 19 BHCs with the indicative loss rates provided by the SCAP
regulators. We used annual data for the year ended December 31, 2009, on
loan charge-offs. We also used average total loan balances. In the Y-9C
total loan balances were categorized somewhat differently from chargeoffs. Table 9 provides a crosswalk for the asset classification. We
aggregated loan balance data into the same categories that were used in
the indicative loss rate table in SCAP: first-lien mortgages, prime
mortgages, Alt-A mortgages, subprime mortgages, second/junior lien
mortgages, closed-end junior liens, home equity lines of credit,
commercial and industrial loans, commercial real estate loans,
construction loans, multifamily loans, nonfarm nonresidential loans, credit
card balances, other consumer, and other loans. Once the data were
aggregated into these categories, we divided the net charge-offs by the

10

Nonperforming loans represent the total of loans that are either 90 plus days past due or
in nonaccrual status. As defined by the instructions to the Y-9C, an asset is in nonaccrual
status if: (1) it is maintained on a cash basis because of deterioration in the financial
condition of the borrower, (2) payment in full of principal or interest is not expected, or (3)
principal or interest as been in default for a period of 90 days or more unless the asset is
both well secured and in the process of collection. Per the Y-9C instructions, an asset is 90
plus days past due if payment is due and unpaid for 90 days or more, and if that asset is not
in nonaccrual status.

11
The Goldman Sachs Group, Inc. and Morgan Stanley were approved by the Federal
Reserve to become BHCs on September 22, 2008; American Express Company was
approved on November 10, 2008; and GMAC LLC was approved on December 24, 2008.

Page 71

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix I: Objectives, Scope, and
Methodology

applicable average loan balance. This calculation showed the loss rate for
each category (e.g., first-lien mortgages and commercial real estate) for
the year ended December 31, 2009. This methodology was applied to
calculate the loss rates for the 19 SCAP BHCs and all BHCs with more than
$1 billion of assets, respectively. Because those institutions had recently
converted to being BHCs, Y-9C data on loan balances was not available for
the fourth quarter of 2008 for American Express Company; The Goldman
Sachs Group, Inc.; and Morgan Stanley, and was not available for GMAC
LLC for both the first quarter of 2009 and the fourth quarter of 2008.
Therefore, we approximated the loan balances in these periods for GMAC
LLC and American Express Company based on their Form 10-Q for these
time periods. Because The Goldman Sachs Group, Inc. and Morgan Stanley
have considerably smaller loan balances, in general, than the other BHCs;
the fourth quarter of 2008 balance was not approximated for these BHCs.
Instead, the average loan balance was simply based on the available data
(e.g., first quarter of 2009 through fourth quarter of 2009).
Table 9: Crosswalk of Y-9C Loans and Lease Financing Receivables to and Classifications used by SCAP
Classification used in GAO analysis
Overall
category

Primary
category

(1) One to four family residential construction loans

Commercial
real estate

Commercial real
estate

Construction

(2) Other construction loans and all land development and other land
loans

Commercial
real estate

Commercial real
estate

Construction

Other

Other loans

n/a

Second/junior
lien

Second/junior lien

Home equity lines
of credit

Y-9C classification

Sub-category

1. Loans secured by real estate
a. Construction, land development, and other land loans in domestic
offices:

b. Secured by farmland
c. Secured by one to four family residential properties:
(1) Revolving, open-end loans secured by one to four family
residential properties and extended under lines of credit
(2) Closed-end loans secured by one to four family residential
properties:
(a) Secured by first liens

First lien

First lien

n/a

(b) Secured by junior liens

Second/junior
lien

Second/junior lien

Closed-end junior
liens

Commercial
real estate

Commercial real
estate

Multifamily

Commercial
real estate

Commercial real
estate

Nonfarm,
nonresidential

d. Secured by mutlifamily (five or more) residential properties
e. Secured by nonfarm nonresidential properties:
(1) Loans secured by owner-occupied nonfarm nonresidential
properties

Page 72

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix I: Objectives, Scope, and
Methodology

Classification used in GAO analysis
Overall
category

Y-9C classification
(2) Loans secured by other nonfarm nonresidential properties

Primary
category

Sub-category

Commercial
real estate

Commercial real
estate

Nonfarm,
nonresidential

Other

Other loans

n/a

2. Loans to depository institutions and acceptances of other banks:
a. To U.S. banks and other U.S. depository institutions
b. To foreign banks

Other

Other loans

n/a

Other

Other loans

n/a

a. To U.S. addresses (domicile)

Commercial
and industrial

Commercial and
industrial

n/a

b. To non-U.S. addresses (domicile)

Commercial
and industrial

Commercial and
industrial

n/a

a. Credit cards

Credit cards

Credit cards

n/a

b. Other revolving credit plans

Other

Other consumer

n/a

b. Other consumer (includes single payment, installment, and all
student loans)

Other

Other consumer

n/a

7. Loans to foreign governments and official institutions (including foreign
central banks)

Other

Other loans

n/a

9. a. Loans for purchasing and carrying securities (secured and
unsecured)

Other

Other loans

n/a

Other

Other loans

n/a

a. Leases to individuals for household, family, and other personal
expenditures

Other

Other consumer

n/a

b. All other leases

Other

Other loans

n/a

3. Loans to finance agricultural production and other loans to farmers
4. Commercial and industrial loans:

6. Loans to individuals for household, family, and other personal
expenditures

b. All other loans
10. Lease financing receivables (net of unearned income):

11. Less: Any unearned income on loans reflected in items 1-9 above.

-

a

-

a

-a

Source: Federal Reserve 2009 Schedule HC-C of the Y-9C.

Notes: Foreign office real estate was also included in our calculation of the total loans, but is not
distinguishable in the table above. We pulled it directly from the SNL Financial database. This amount
equates to the difference, in Schedule HC-C, between line item 1 for the “Consolidated” and “In
Domestic Offices” columns (these columns are not depicted above). The classification of these loans
in our calculations was as “Other” and the primary category was “Other loans.”
N/a means not applicable.
a

For calculations for the 19 SCAP BHCs, unearned income was distributed to all loan balances based
on the percent that each line item represented of total loans for that BHC (excludes lease financing
receivables). For calculations for all BHCs with total assets greater than $1 billion, unearned income
was distributed to the aggregate balances for each line item based on the respective percentage that
each balance represented of the total.

Page 73

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix I: Objectives, Scope, and
Methodology

We conducted this performance audit from August 2009 to September 2010
in accordance with generally accepted government auditing standards.
Those standards require that we plan and perform the audit to obtain
sufficient, appropriate evidence to provide a reasonable basis for our
findings and conclusions based on our audit objectives. We believe that
the evidence obtained provides a reasonable basis for our findings and
conclusions based on our audit objectives.

Page 74

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix II: Status of Bank Holding
Companies’ TARP Investments as September
22, 2010

Appendix II: Status of Bank Holding
Companies’ TARP Investments as September
22, 2010
Twelve of the 19 bank holding companies (BHC) that participated in the
Supervisory Capital Assessment Program (SCAP) had redeemed their
Troubled Asset Relief Program (TARP) investments and had their
warrants disposed of as of September 22, 2010, and most of them were not
required to raise capital under SCAP (table 10). Six of the 19 BHCs tested
under SCAP have not repaid TARP investments or disposed of warrants,
and one, MetLife, Inc., did not receive any TARP investments. BHCs
participating in SCAP must follow specific criteria to repay TARP funds. In
approving applications from participating banks that want to repay TARP
funds, the Federal Reserve considers various factors. Some of these
factors 1 include whether the banks can demonstrate an ability to access
the long-term debt market without relying on the Federal Deposit
Insurance Corporation’s (FDIC) Temporary Liquidity Guarantee Program
and whether they can successfully access the public equity markets,
remain in a position to facilitate lending, and maintain capital levels in
accord with supervisory expectations. 2 BHCs intending to repay TARP
investments must have post repayment capital ratios that meet or exceed
SCAP requirements.

1

See Federal Reserve’s June 1, 2009, press release that sets forth the criteria that SCAP
BHCs must meet before they can pay back their TARP investments.
2

FDIC created this facility in November 2008 to encourage liquidity in the banking system
by guaranteeing newly issued senior unsecured debt of banks, thrifts, and certain holding
companies and by providing full coverage of noninterest-bearing deposit transaction
accounts. The facility is scheduled to end in 2010.

Page 75

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix II: Status of Bank Holding
Companies’ TARP Investments as September
22, 2010

Table 10: Status of TARP Investments for the 19 BHCs Participating in SCAP, as of September 22, 2010
Dollars in billions

Bank holding company

Required to
raise capital
under SCAP?

Type of
TARP
received

Capital
amount
received

Capital
repayment
date

Capital
amount
repaid

Warrant
disposition
date

Warrants
repurchased (R)
or sold via
a
auction (A)?

BHC that was not a recipient of TARP funding
MetLife, Inc.

No

n/a

n/a

n/a

n/a

n/a

n/a

BHCs that were recipients of TARP funding and have exited TARP
American Express
Company

No

CPPb

$3.4

06/17/09

$3.4

07/29/09

R

Bank of America
Corporation

Yes

CPP

25.0

12/09/09

25.0

03/03/10

A

TIPc

20.0

12/09/09

20.0

03/03/10

A

BB&T Corporation

No

CPP

3.1

06/17/09

3.1

07/22/09

R

The Bank of New York
Mellon Corporation

No

CPP

3.0

06/17/09

3.0

08/05/09

R

Capital One Financial
Corporation

No

CPP

3.6

06/17/09

3.6

12/03/09

A

The Goldman Sachs
Group, Inc.

No

CPP

10.0

06/17/09

10.0

07/22/09

R

JPMorgan Chase & Co.

No

CPP

25.0

06/17/09

25.0

12/10/09

A

Morgan Stanley

Yes

CPP

10.0

06/17/09

10.0

08/12/09

R

PNC Financial Services
Group, Inc

Yes

CPP

7.6

02/02/10

7.6

04/29/10

A

State Street Corporation

No

CPP

2.0

06/17/09

2.0

07/08/09

R

U.S. Bancorp

No

CPP

6.6

06/17/09

6.6

07/15/09

R

Wells Fargo & Company

Yes

CPP

25.0

12/23/09

25.0

05/21/10

A

BHCs that have not fully repaid TARP funding or disposed of warrants
Citigroup Inc.d

Yes

CPP

25.0

-

TIP

20.0

12/23/09

-

-

-

20.0

-

-

Fifth Third Bancorp

Yes

CPP

3.4

-

-

-

-

GMAC LLCe

Yes

AIFP

16.3

-

-

-

-

KeyCorp

Yes

CPP

2.5

-

-

-

-

Regions Financial
Corporation

Yes

CPP

3.5

-

-

-

-

SunTrust Banks, Inc.

Yes

CPP

4.9

-

-

-

-

Sources: Federal Reserve’s SCAP results document and Treasury’s TARP Transactions Report for the Period Ending September XX,
2010.

Note: N/a means not applicable since MetLife, Inc. did not receive any TARP funding.

Page 76

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix II: Status of Bank Holding
Companies’ TARP Investments as September
22, 2010

a

“R” indicates that the warrants were repurchased by the financial institution via negotiations with
Treasury. “A” indicates that Treasury sold the warrants in a registered public auction.
b

The Capital Purchase Program (CPP) is a program in which Treasury invests in preferred securities
issued by qualified financial institutions.

c

Treasury created the Targeted Investment Program (TIP) to stabilize the financial system by making
investments in institutions that are determined to be critical to the functioning of the financial system.

d

As part of an exchange offer designed to strengthen Citigroup Inc.’s capital, in June 2009, Treasury
agreed to exchange its $25 billion of CPP preferred stock in Citigroup for 7.7 billion shares of
Citigroup Inc. common stock at a price of $3.25 per common share. In May 2010, Treasury sold 1.5
billion of its 7.7 billion common shares. In June 2010, Treasury sold 1.1 billion shares and has a
remaining ownership of 5.1 billion common shares.

e

On June 30, 2009, GMAC LLC changed its corporate structure and became GMAC Inc., and on May
10, 2010, GMAC Inc. changed its name to Ally Financial Inc.

Page 77

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs
Table 11 shows the names, location, and total assets as of December 31,
2008, of the 19 bank holding companies (BHC) subject to the Supervisory
Capital Assessment Program (SCAP) stress test that was conducted by the
federal bank regulators in the spring of 2009. The stress test was a
forward-looking exercise intended to help federal banking regulators
gauge the extent of the additional capital buffer necessary to keep the
BHCs strongly capitalized and lending even if economic conditions are
worse than had been expected between December 2008 and December
2010.
Table 11: Identification of 19 BHCs Subject to the Stress Test
Dollars in thousands
Table
number

Name of bank holding company

Location of
headquarters

Total assets as of
December 31, 2008

12

American Express Company

New York, NY

$126,074,000

13

Bank of America Corporation

Charlotte, NC

1,817,943,000

14

BB&T Corporation

Winston-Salem, NC

152,015,000

15

The Bank of New York Mellon Corporation

New York, NY

237,512,000

16

Capital One Financial Corporation

McLean, VA

165,913,452

17

Citigroup Inc.

New York, NY

1,938,470,000

18

Fifth Third Bancorp

Cincinnati, OH

119,764,000

19

GMAC LLC

Detroit, MI

189,476,000

20

The Goldman Sachs Group, Inc.

New York, NY

884,547,000

21

JPMorgan Chase & Co.

New York, NY

2,175,052,000

22

KeyCorp

Cleveland, OH

104,531,000

23

MetLife, Inc.

New York, NY

501,678,000

24

Morgan Stanley

New York, NY

676,764,000

25

PNC Financial Services Group, Inc.

Pittsburgh, PA

291,081,000

26

Regions Financial Corporation

Birmingham, AL

146,247,810

27

State Street Corporation

Boston, MA

173,631,000

28

SunTrust Banks, Inc.

Atlanta, GA

189,137,961

29

U.S. Bancorp

Minneapolis, MN

265,912,000

30

Wells Fargo & Company

San Francisco, CA

1,309,639,000

Source: GAO.

The following tables (12 through 30) compare the 2009 performance of the
19 BHCs involved in SCAP to the 2-year SCAP estimates and the GAO 1year pro rata estimates for the more adverse economic scenario.
Specifically, these tables include comparison of actual and estimates of
losses and gains associated with loans, securities, trading and

Page 78

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

counterparty, resources, preprovision net revenue (PPNR), and allowance
for loan and lease losses (ALLL). These tables also include a comparison
of actual capital levels at December 31, 2009, and December 31, 2008.
Totals may not add due to rounding. For a more detailed explanation of
the calculations made in constructing this analysis, see appendix I.

Page 79

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

Table 12: American Express Company
Dollars in billions
Actual at 12/31/09
Tier 1 capital

12/31/08 balance
per SCAP

Difference

12/31/09 as a percent of
the 12/31/08 balance

$11.5

$10.1

$1.4

113.5%

Tier 1 common capital

$11.5

$10.1

$1.4

113.5%

Risk-weighted assets

$116.6

$104.4

$12.2

111.6%

Tier 1 risk-based ratio

9.8%

9.7%

0.1%

101.4%

Tier 1 common capital ratio

9.8%

9.7%

0.1%

101.4%

2-year SCAP GAO 1-year pro
estimate
rata estimate Difference

Actual as a percent of
the pro rata estimate

Actual for year
ended 12/31/09
Total asset losses
•
First-lien mortgages

$4.4

$11.2

$5.6

$(1.2)

78.0%

0.0

n/a

n/a

n/a

n/a

•

Second/junior lien
mortgages

0.0

n/a

n/a

n/a

n/a

•

Commercial and industrial
loans

0.0

n/a

n/a

n/a

n/a

•

Commercial real estate
loans
Credit card loans

0.0

n/a

n/a

n/a

n/a

3.4

8.5

4.3

(0.9)

79.9

•
•

Securities (available for
sale and held to maturity)

(0.2)

n/a

n/a

n/a

n/a

•

Trading and counterparty

-a

n/a

n/a

n/a

n/a

•

Other

1.2

2.7

1.4

(0.2)

88.0

Resources other than capital
to absorb losses (Total
PPNR less change in ALLL)

$7.4

$11.9

$6.0

$1.4

123.8%

PPNR

$7.8

-

-

-

-

5.3

-

-

-

-

•

Net interest income
(expense)
Noninterest income

19.0a

-

-

-

-

•

Less: noninterest expense

16.5

-

-

-

-

Change in allowance for
loan and lease losses (ALLL)
•
ALLL at 12/31/08

$0.4

-

-

-

-

3.4

-

-

-

-

3.8

-

-

-

-

•

•

ALLL at 12/31/09

Source: GAO analysis of Federal Reserve and SNL Financial data.

Note: N/a means not applicable.

Page 80

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

a

Trading and counterparty positions were not stressed because the total portfolio is less than the
$100 billion required for stress testing in SCAP, but trading (gain) loss information for this BHC was
included in the “trading revenue” line of Schedule HI of the Y-9C in 2009. In SCAP, the projections of
trading gains or losses for this BHC were included in the estimate of PPNR rather than the trading
and counterparty line. Therefore, we have included the actual trading results in PPNR (specifically
noninterest income).

Page 81

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

Table 13: Bank of America Corporation
Dollars in billions

Actual at 12/31/09

12/31/08 balance
per SCAP

Difference

12/31/09 as a percent
of the 12/31/08
balance

$160.6

$173.2

$(12.6)

92.7%

Tier 1 common capital

a

$120.6

$74.5

$46.1

161.9%

Risk-weighted assets

$1,541.6

$1,633.8

$(92.2)

94.4%

Tier 1 risk-based ratio

10.4%

10.6%

(0.2)%

98.3%

7.8%

4.6%

3.2%

170.1%

Difference

Actual as a percent of
the pro rata estimate

Tier 1 capital

Tier 1 common capital ratio

Actual for year
ended 12/31/09
Total asset losses
•
First-lien mortgages
•

Second/junior lien
mortgages

•

Commercial and
industrial loans
Commercial real estate
loans

•
•

Credit card loans

•
•

Securities (available for
sale and held to maturity)
Trading and counterparty

•

Other

Resources other than
capital to absorb losses
(total PPNR less change in
ALLL)
PPNR
•
Net interest income
(expense)

2-year SCAP GAO 1-year pro
estimate
rata estimate

$12.3

$136.6

$68.4

$(56.0)

18.1%

3.6

22.1

11.1

(7.5)

32.5

7.6

21.4

10.7

(3.1)

70.9

5.0

15.7

7.9

(2.8)

63.8

3.3

9.4

4.7

(1.4)

69.4

7.8

19.1

9.6

(1.8)

81.5

(9.3)

8.5

4.3

(13.5)

(218.4)

(12.1)b

24.1

12.1

(24.1)

(100.1)

6.4

16.4

8.2

(1.8)

78.7

$29.5

$74.5

$37.3

$(7.7)

79.3%

c,d

-

-

-

-

$43.7

47.8

-

-

-

-

•

Noninterest income

62.6b

-

-

-

-

•

Less: noninterest
expense

66.7

-

-

-

-

Change in allowance for
loan and lease losses
(ALLL)
•
ALLL at 12/31/08

$14.1

-

-

-

-

23.1

-

-

-

-

ALLL at 12/31/09

37.2

-

-

-

-

•

Source: GAO analysis of Federal Reserve and SNL Financial data.

Note: N/a means not applicable.

Page 82

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

a

Tier 1 common capital includes $19.29 billion of common equivalent securities (CES) issued in
December 2009. As described in Bank of America Corporation’s (Bank of America) Form 10-K for the
year ended December 31, 2009, CES are included in tier 1 common capital based upon applicable
regulatory guidance and the expectation that the underlying securities would convert to common
stock following shareholder approval of additional authorized shares. Shareholders approved the
increase in the number of authorized shares of common stock at the special meeting of shareholders
held on February 23, 2010, and the CES converted to common stock on February 24, 2010.

b

The trading and counterparty data in the Y-9C includes both customer derived revenue from
transactions for BHCs that operate as broker-dealers as well as gains and losses from proprietary
trading and associated expenses. These items are presented in net form only in the Y-9C. For the
five BHCs that had their trading portfolios stressed (including Bank of America), the trading and
counterparty line is based on projections of (gains) losses from proprietary trading, but PPNR
(specifically noninterest revenue) included projections of gains (losses) from customer derived
revenue from transactions due to operations as a broker-dealer. Because we could not segregate
these items based on the Y-9C, we have included the net amount in the trading and counterparty and
noninterest income line items above. As a result of this limitation, the net amount of the trading gains
or losses and PNNR in the table may be overstated or understated.
c

PPNR includes an owned debt value adjustment of ($4.80) billion, which was not stressed in SCAP.
As Bank of America’s credit spreads narrowed during 2009, this caused the liability values to
increase. This offsets the gains Bank of America experienced in 2008 when its credit spreads
widened.
d

PPNR includes one-time items totaling $4.90 billion, which were not included in SCAP.

Page 83

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

Table 14: BB&T Corporation
Dollars in billions

Difference

12/31/09 as a percent
of the 12/31/08
balance

Tier 1 capital

$13.5

$13.4

$0.1

100.4%

Tier 1 common capital

$10.0

$7.8

$2.2

127.7%

Risk-weighted assets

$117.2

$109.8

$7.4

106.7%

Tier 1 risk-based ratio

11.5%

12.3%

(0.8)%

93.4%

8.5%

7.1%

1.4%

119.7%

Difference

Actual as a percent of
the pro rata estimate

12/31/08 balance
per SCAP
Actual at 12/31/09

Tier 1 common capital ratio

Actual for year
ended 12/31/09
Total asset losses
•
First-lien mortgages
•

Second/junior lien
mortgages

•

Commercial and industrial
loans
Commercial real estate
loans

•
•

Credit card loans

•
•

Securities (available for
sale and held to maturity)
Trading and counterparty

•

Other

Resources other than capital
to absorb losses (total PPNR
less change in ALLL)

2-year SCAP GAO 1-year pro
estimate
rata estimate

$1.6

$8.7

$4.4

$(2.8)

36.2%

0.3

1.1

0.6

(0.3)

47.8

0.1

0.7

0.4

(0.2)

40.9

0.2

0.7

0.4

(0.2)

56.5

0.7

4.5

2.3

(1.5)

32.3

0.1

0.2

0.1

(0.0)

81.6

(0.2)

0.2

0.1

(0.3)

-199.3

-a

n/a

n/a

-

-

0.4

1.3

0.7

(0.3)

55.6

$2.6

$5.5

$2.8

$(0.1)

94.9%

PPNR
•
Net interest income
(expense)

$3.6

-

-

-

-

4.8

-

-

-

-

•

Noninterest income

a

3.5

-

-

-

-

•

Less: noninterest expense

4.7

-

-

-

-

Change in allowance for loan
and lease losses (ALLL)

$1.0

-

-

-

-

•

ALLL at 12/31/08

1.6

-

-

-

-

•

ALLL at 12/31/09

2.6

-

-

-

-

Source: GAO analysis of Federal Reserve and SNL Financial data.

Note: N/a means not applicable.

Page 84

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

On August 14, 2009, BB&T Corporation (BB&T) entered into a purchase and assumption agreement
with the Federal Deposit Insurance Corporation (FDIC) to acquire certain assets and assume
substantially all of the deposits and certain liabilities of Colonial Bank, an Alabama state-chartered
bank headquartered in Montgomery, Alabama. As further discussed in BB&T’s Form 10-K for the year
ended December 31, 2009, BB&T entered into loss sharing agreements with the FDIC related to
certain loans, securities, and other assets. The actual results include BB&T’s performance including
the Colonial Bank acquisition.
a

Trading and counterparty positions were not stressed because the total portfolio is less than the
$100 billion required for stress testing in SCAP, but trading (gain) loss information for this BHC was
included in the “trading revenue” line of Schedule HI of the Y-9C in 2009. In SCAP, the projections of
trading gains or losses for this BHC were included in the estimate of PPNR rather than the trading
and counterparty line. Therefore, we have included the actual trading results in PPNR (specifically
noninterest income).

Page 85

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

Table 15: The Bank of New York Mellon Corporation
Dollars in billions

Actual at 12/31/09

12/31/08 balance
per SCAP

Difference

12/31/09 as a percent
of the 12/31/08
balance

Tier 1 capital

$12.9

$15.4

$(2.5)

83.7%

Tier 1 common capital

$11.2

$11.0

$0.2

101.8%

Risk-weighted assets

$106.3

$115.8

$(9.5)

91.8%

Tier 1 risk-based ratio

12.1%

13.3%

(1.2)%

91.1%

Tier 1 common capital ratio

10.5%

9.5%

1.0%

110.8%

Actual for year
ended 12/31/09

2-year SCAP
estimate

GAO 1-year pro
rata estimate

Difference

Actual as a percent of
the pro rata estimate

$5.6

$5.4

$2.7

$2.9

207.6%

0.1

0.2

0.1

(0.0)

60.0

0.0

n/a

n/a

0.1

0.4

0.2

(0.1)

45.0

0.1

0.2

0.1

(0.1)

50.0

3.3

255.7

Total asset losses
•
First-lien mortgages
•

Second/junior lien
mortgages

•

Commercial and industrial
loans
Commercial real estate
loans

•
•

Credit card loans

0.0

n/a

n/a

•

5.4

a

4.2

2.1

•

Securities (available for
sale and held to maturity)
Trading and counterparty

-b

n/a

n/a

•

Other

0.0

0.4

0.2

(0.2)

17.0

Resources other than capital
to absorb losses (total PPNR
less change in ALLL):

$3.4

$6.7

$3.4

$(0.0)

100.2%

PPNR
•
Net interest income
(expense)

$3.5

-

-

-

-

2.9

-

-

-

-

•

Noninterest income

b

10.1

-

-

-

-

•

Less: noninterest expense

9.6

-

-

-

-

Change in allowance for loan
and lease losses (ALLL)

$0.1

-

-

-

-

•

ALLL at 12/31/08

.4

-

-

-

-

•

ALLL at 12/31/09

.5

-

-

-

-

Source: GAO analysis of Federal Reserve and SNL Financial data.

Note: N/a means not applicable.

Page 86

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

a

Based on discussions with Bank of New York Mellon Corporation (Bank of New York Mellon)
officials, the company’s securities portfolio underwent a significant restructuring in the third quarter of
2009 in order to de-risk this portfolio, causing the recognition of significant losses in that period.
However, The Bank of New York Mellon sold off many of its riskiest holdings in that period, including
many Alt-A residential mortgage-backed securities, so that it expects to see gains in this portfolio in
the future, causing the final 2-year loss amount to be less than the amount projected under SCAP in
the securities (available for sale and held to maturity) portfolio. See the Bank of New York Mellon’s
Form 10-K for the year ended December 31, 2009, and other public disclosures for additional
information. As a result of the write downs taken due to the restructuring of the securities portfolio,
Bank of New York Mellon expects an increase in net interest revenue of $200 million in 2010. As of
the second quarter of 2010 year to date, the BHC experienced a gain of $20 million in this portfolio.
b

Trading and counterparty positions were not stressed because the total portfolio is less than the
$100 billion required for stress testing in SCAP, but trading (gain) loss information for this BHC was
included in the “trading revenue” line of Schedule HI of the Y-9C in 2009. In SCAP, the projections of
trading gains or losses for this BHC were included in the estimate of PPNR rather than the trading
and counterparty line. Therefore, we have included the actual trading results in PPNR (specifically
noninterest income).

Page 87

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

Table 16: Capital One Financial Corporation
Dollars in billions

Difference

12/31/09 as a percent
of the 12/31/08
balance

Tier 1 capital

$16.0

$16.8

$(0.8)

95.2%

Tier 1 common capital

$12.4

$12.0

$0.4

103.0%

Risk-weighted assets

$116.3

$131.8

$(15.5)

88.2%

Tier 1 risk-based ratio

13.7%

12.7%

1.0%

108.2%

Tier 1 common capital ratio

10.6%

9.1%

1.5%

116.7%

Actual for year
ended 12/31/09

2-year SCAP
estimate

GAO 1-year pro
rata estimate

Difference

Actual as a percent of
the pro rata estimate

$4.4

$13.4

$6.7

$(2.3)

65.1%

0.1

1.8

0.9

(0.8)

7.8%

0.2

0.7

0.4

(0.1)

63.0

0.6

1.5

0.8

(0.1)

81.5

0.3

1.1

0.6

(0.2)

62.4

1.8

3.6

1.8

(0.0)

98.0

(0.2)

0.4

0.2

(0.4)

(103.1)

-a

n/a

n/a

-

-

1.6

4.3

2.2

(0.6)

72.5

12/31/08 balance
per SCAP
Actual at 12/31/09

Total asset losses
•
First-lien mortgages
•

Second/junior lien
mortgages

•

Commercial and industrial
loans
Commercial real estate
loans

•
•

Credit card loans

•
•

Securities (available for
sale and held to maturity)
Trading and counterparty

•

Other

Resources other than capital
to absorb losses (total PPNR
less change in ALLL):

$5.8

$9.0

$4.5

$1.3

128.0%

PPNR
•
Net interest income
(expense)

$5.4

-

-

-

-

7.7

-

-

-

-

•

Noninterest income

a

5.0

-

-

-

-

•

Less: noninterest expense

7.3

-

-

-

-

Change in allowance for loan
and lease losses (ALLL)

$(0.4)

-

-

-

-

•

ALLL at 12/31/08

4.5

-

-

-

-

•

ALLL at 12/31/09

4.1

-

-

-

-

Source: GAO analysis of Federal Reserve and SNL Financial data.

Note: N/a means not applicable.

Page 88

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

a

Trading and counterparty positions were not stressed because the total portfolio is less than the
$100 billion required for stress testing in SCAP, but trading (gain) loss information for this BHC was
included in the “trading revenue” line of Schedule HI of the Y-9C in 2009. In SCAP, the projections of
trading gains or losses for this BHC were included in the estimate of PPNR rather than the trading
and counterparty line. Therefore, we have included the actual trading results in PPNR (specifically
noninterest income).

Page 89

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

Table 17: Citigroup Inc.
Dollars in billions

Difference

12/31/09 as a percent
of the 12/31/08
balance

Tier 1 capital

$127.0

$118.8

$8.2

106.9%

Tier 1 common capital

$106.4

$22.9

$83.5

464.5%

Risk-weighted assets

$1088.5

$996.2

$92.3

109.3%

Tier 1 risk-based ratio

11.7%

11.9%

(0.2)%

98.1%

9.8%

2.3%

7.5%

424.9%

Actual for year
ended 12/31/09

2-year SCAP
estimate

$27.2

$104.7

$52.4

$(25.1)

52.0%

4.2

15.3

7.7

(3.5)

54.5

4.7

12.2

6.1

(1.4)

76.8

5.1

8.9

4.5

0.6

113.6

0.7

2.7

1.4

(0.6)

52.2

6.5

19.9

10.0

(3.4)

65.4

12/31/08 balance
per SCAP
Actual at 12/31/09

Tier 1 common capital ratio

Total asset losses
•
First-lien mortgages
•

Second/junior lien
mortgages

•

Commercial and industrial
loans
Commercial real estate
loans

•
•

Credit card loans

•
•

Securities (available for
sale and held to maturity)
Trading and counterparty

•

Other

Resources other than capital
to absorb losses (total PPNR
less change in ALLL):
PPNR
•
Net interest income
(expense)

GAO 1-year pro
Actual as a percent of
rata estimate Difference the pro rata estimate

0.9

2.9

1.5

(0.5)

62.8

(4.4)a

22.4

11.2

(15.6)

(39.7)

9.6

20.4

10.2

(0.6)

94.3

$25.5

$49.0

$24.5

$1.0

103.9%

b,c

-

-

-

-

50.4

-

-

-

-

a

$31.9

•

Noninterest income

32.1

-

-

-

-

•

Less: noninterest expense

50.6

-

-

-

-

Change in allowance for loan
and lease losses (ALLL)

$6.4

-

-

-

-

•

ALLL at 12/31/08

29.6

-

-

-

-

•

ALLL at 12/31/09

36.0

-

-

-

-

Source: GAO analysis of Federal Reserve and SNL Financial data.

Note: N/a means not applicable.

Page 90

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

a

The trading and counterparty data in the Y-9C includes both customer derived revenue from
transactions for BHCs that operate as broker-dealers as well as gains and losses from proprietary
trading and associated expenses. These items are presented in net form only in the Y-9C. For the
five BHCs that had their trading portfolios stressed (including Citigroup Inc.), the trading and
counterparty line is based on projections of (gains) losses from proprietary trading, but PPNR
(specifically noninterest revenue) included projections of gains (losses) from customer derived
revenue from transactions due to operations as a broker-dealer. Because we could not segregate
these items based on the Y-9C, we have included the net amount in both the trading and
counterparty and noninterest income line items above. As a result of this limitation, the net amount of
the trading gains or losses and PPNR in the table may be overstated or understated.

b

PPNR includes an owned debt value adjustment of ($4.23) billion, which was not included stressed
in SCAP. As Citigroup’s credit spreads narrowed during 2009, this caused the liability values to
increase. This offsets the gains Citigroup Inc. experienced in 2008 when its credit spreads widened.

c

PPNR includes one-time items totaling $2.73 billion, which were not included in SCAP.

Page 91

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

Table 18: Fifth Third Bancorp
Dollars in billions

Difference

12/31/09 as a percent
of the 12/31/08
balance

$13.4

$11.9

$1.5

112.8%

$7.1

$4.9

$2.2

144.0%

Risk-weighted assets

$100.9

$112.6

$(11.7)

89.6%

Tier 1 risk-based ratio

13.3%

10.6%

2.7%

125.6%

7.0%

4.4%

2.6%

159.0%

Difference

Actual as a percent of
the pro rata estimate

12/31/08 balance
per SCAP
Actual at 12/31/09
Tier 1 capital
Tier 1 common capital

Tier 1 common capital ratio

Actual for year
ended 12/31/09
Total asset losses
•
First-lien mortgages
•

Second/junior lien
mortgages

•

Commercial and industrial
loans
Commercial real estate
loans

•
•

Credit card loans

•
•

Securities (available for
sale and held to maturity)
Trading and counterparty

•

Other

Resources other than capital
to absorb losses (total PPNR
less change in ALLL):
PPNR
•
Net interest income
(expense)

2-year SCAP GAO 1-year pro
estimate
rata estimate

$2.5

$9.1

$4.6

$(2.1)

54.6%

0.3

1.1

0.6

(0.3)

49.1

0.3

1.1

0.6

(0.2)

57.8

0.6

2.8

1.4

(0.8)

39.4

1.0

2.9

1.5

(0.4)

70.8

0.2

0.4

0.2

(0.0)

84.6

(0.1)

0.1

0.0

(0.1)

-227.5

-a

n/a

n/a

0.2

0.9

0.5

(0.2)

54.6

$3.3

$5.5

$2.8

$0.5

119.7%

b

-

-

-

-

3.5

-

-

-

-

$4.3

a

-

•

Noninterest income

4.6

-

-

-

-

•

Less: noninterest expense

3.8

-

-

-

-

Change in allowance for loan
and lease losses (ALLL)

$1.0

-

-

-

-

•

ALLL at 12/31/08

2.8

-

-

-

-

•

ALLL at 12/31/09

3.7

-

-

-

-

Source: GAO analysis of Federal Reserve and SNL Financial data.

Note: N/a means not applicable.

Page 92

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

a

Trading and counterparty positions were not stressed because the total portfolio is less than the
$100 billion required for stress testing in SCAP, but trading (gain) loss information for this BHC was
included in the “trading revenue” line of Schedule HI of the Y-9C in 2009. In SCAP, the projections of
trading gains or losses for this BHC were included in the estimate of PPNR rather than the trading
and counterparty line. Therefore, we have included the actual trading results in PPNR (specifically
noninterest income).

b

Fifth Third Bancorp’s PPNR includes one-time items totaling $2.05 billion, which were not included in
SCAP.

Page 93

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

Table 19: GMAC LLC
Dollars in billions

Difference

12/31/09 as a percent
of the 12/31/08
balance

$22.4

$17.4

$5.0

128.7%

$7.7

$11.1

$(3.4)

69.2%

Risk-weighted assets

$158.3

$172.7

$(14.4)

91.7%

Tier 1 risk-based ratio

14.1%

10.1%

4.0%

140.1%

4.8%

6.4%

(1.6)%

75.8%

Difference

Actual as a percent of
the pro rata estimate

12/31/08 balance
per SCAP
Actual at 12/31/09
Tier 1 capital
Tier 1 common capital

Tier 1 common capital ratio

Actual for year
ended 12/31/09
Total asset lossesa
•
First-lien mortgages

2-year SCAP GAO 1-year pro
estimate
rata estimate

$6.9

$9.2

$4.6

$2.3

150.7%

2.4

2.0

1.0

1.4

239.9

•

Second/junior lien
mortgages

1.6

1.1

0.6

1.0

287.3

•

Commercial and industrial
loans
Commercial real estate
loans

0.4

1.0

0.5

(0.1)

71.2

0.7

0.6

0.3

0.4

236.7

•

Credit card loans

0.0

n/a

n/a

•

Securities (available for
sale and held to maturity)
Trading and counterparty

(0.2)

0.5

0.3

(0.4)

(66.4)

-b

n/a

n/a

-

-

2.1

4.0

2.0

0.1

102.7

•

•
•

Other

c

Resources other than capital
to absorb losses (total PPNR
less change in ALLL):

$(1.1)

$(0.5)

$(0.3)

$(0.8)

(429.6)%

PPNR
•
Net interest income
(expense)

$(2.1)

-

-

-

-

0.1

-

-

-

-

•

Noninterest income

10.1

-

-

-

-

•

Less: noninterest expense

12.3

-

-

-

-

Change in allowance for loan
and lease losses (ALLL)

$(1.0)

-

-

-

-

•

ALLL at 12/31/08

3.4

-

-

-

-

•

ALLL at 12/31/09

2.4

-

-

-

-

Source: GAO analysis of Federal Reserve and SNL Financial data.

Notes: N/a means not applicable.

Page 94

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

GMAC LLC (GMAC) experienced a loss from discontinued operations totaling $2.4 billion in 2009.
The item was not included in our calculation of PPNR.
GMAC changed its corporate name to GMAC Inc. on June 30, 2009. On May 10, 2010, GMAC Inc.
changed its name to Ally Financial Inc.
a

According to GMAC officials, in order to be positioned for better future performance the company
pulled losses forward into the fourth quarter of 2009 by recognizing the lifetime losses on assets in
that period; and as a result of the accelerated loss recognition less losses would be expected in 2010.
GMAC had a profit in the first and second quarters of 2010, its first profits since the fourth quarter of
2008. GMAC’s tier 1 common capital ratio also improved to 5 percent and 5.2 percent, respectively.
b

Trading and counterparty positions were not stressed because the total portfolio is less than the
$100 billion required for stress testing in SCAP, but trading (gain) loss information for this BHC was
included in the “trading revenue” line of Schedule HI of the Y-9C in 2009. In SCAP, the projections of
trading gains or losses for this BHC were included in the estimate of PPNR rather than the trading
and counterparty line. Therefore, we have included the actual trading results in PPNR (specifically
noninterest income).

c

GMAC’s “Other” loans category per SCAP included only automobile-related loans. However, our
classification of “Other” using Y-9C data includes automobile loans and other loans such as European
home mortgages, which had substantial losses in 2009. Automobile loan losses totaled about $600
million in 2009 compared to the $2.0 billion prorated SCAP estimate, according to GMAC officials. On
April 12, 2010, GMAC’s mortgage subsidiary, Residential Capital, LLC agreed to sell its European
mortgage assets and business. The assets in the transactions are valued at approximately the levels
established in the fourth quarter of 2009, and there is no material gain or loss expected.

Page 95

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

Table 20: The Goldman Sachs Group, Inc.
Dollars in billions

Difference

12/31/09 as a percent
of the 12/31/08
balance

Tier 1 capital

$64.6

$55.9

$8.7

115.6%

Tier 1 common capital

$52.7

$34.4

$18.3

153.2%

Risk-weighted assets

$431.9

$444.8

$(12.9)

97.1%

Tier 1 risk-based ratio

15.0%

12.6%

2.4%

118.8%

Tier 1 common capital ratio

12.2%

7.7%

4.5%

158.4%

Actual for year
ended 12/31/09

2-year SCAP
estimate

$(23.3)

$17.8

$8.9

$(32.2)

(261.3)%

0.0

n/a

n/a

n/a

-

0.0

n/a

n/a

n/a

-

0.0

0.0

0.0

(0.0)

0.0%

0.0

n/a

n/a

n/a

-

0.0

n/a

n/a

n/a

-

12/31/08 balance
per SCAP
Actual at 12/31/09

Total asset losses
•
First-lien mortgages
•

Second/junior lien
mortgages

•

Commercial and industrial
loans
Commercial real estate
loans

•
•

Credit card loans

•
•

Securities (available for
sale and held to maturity)
Trading and counterparty

•

Other

Resources other than capital
to absorb losses (total PPNR
less change in ALLL):
PPNR
•
Net interest income
(expense)

GAO 1-year pro
Actual as a percent of
rata estimate Difference the pro rata estimate

(0.0)

0.1

0.1

(0.1)

(72.0)

(23.2)a

17.4

8.7

(31.9)

(267.10)

0.0

0.3

0.2

(0.2)

0.0

$19.4

$18.5

$9.3

$10.2

209.9%

b

-

-

-

-

7.4

-

-

-

-

a

$19.4

•

Noninterest income

37.3

-

-

-

-

•

Less: noninterest expense

25.3

-

-

-

-

Change in allowance for loan
and lease losses (ALLL)

$0.0

-

-

-

-

•

ALLL at 12/31/08

0.0

-

-

-

-

•

ALLL at 12/31/09

0.0

-

-

-

-

Source: GAO analysis of Federal Reserve and SNL Financial data.

Note: N/a means not applicable.

Page 96

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

a

The trading and counterparty data in the Y-9C includes both customer derived revenue from
transactions for BHCs that operate as broker-dealers as well as gains and losses from proprietary
trading and associated expenses. These items are presented in net form only in the Y-9C. For the
five BHCs that had their trading portfolios stressed (including Goldman Sachs Group, Inc.), the
trading and counterparty line is based on projections of (gains) losses from proprietary trading, but
PPNR (specifically noninterest revenue) included projections of gains (losses) from customer derived
revenue from transactions due to operations as a broker-dealer. Because we could not segregate
these items based on the Y-9C, we have included the net amount in both the trading and
counterparty and noninterest income line items above. As a result of this limitation, the net amount of
the trading gains or losses and PPNR in the table may be overstated or understated.

b

PPNR includes an owned debt value adjustment of ($770) million, which was not stressed in SCAP.
As Goldman Sachs Group, Inc.’s credit spreads narrowed during 2009, this caused the liability values
to increase. This offsets the gains Goldman Sachs Group, Inc. experienced in 2008 when its credit
spreads widened.

Page 97

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

Table 21: JPMorgan Chase & Co.
Dollars in billions

Difference

12/31/09 as a percent
of the 12/31/08
balance

Tier 1 capital

$133.0

$136.2

$(3.2)

97.6%

Tier 1 common capital

$105.3

$87.0

$18.3

121.0%

Risk-weighted assets

$1,198.0

$1,337.5

$(139.5)

89.6%

Tier 1 risk-based ratio

11.1%

10.2%

0.9%

108.8%

8.8%

6.5%

2.3%

135.2%

Difference

Actual as a percent of
the pro rata estimate

12/31/08 balance
per SCAP
Actual at 12/31/09

Tier 1 common capital ratio

Actual for year
ended 12/31/09
Total asset losses
•
First-lien mortgages
•

Second/junior lien
mortgages

•

Commercial and industrial
loans
Commercial real estate
loans

•
•

Credit card loans

•
•

Securities (available for
sale and held to maturity)
Trading and counterparty

•

Other

Resources other than capital
to absorb losses (total PPNR
less change in ALLL):
PPNR
•
Net interest income
(expense)

2-year SCAP GAO 1-year pro
estimate
rata estimate

$12.0

$97.4

$48.7

$(36.7)

24.6%

3.5

18.8

9.4

(5.9)

37.7

4.7

20.1

10.1

(5.4)

46.3

3.6

10.3

5.2

(1.5)

70.8

0.8

3.7

1.9

(1.0)

45.4

8.1

21.2

10.6

(2.5)

76.1

(1.1)

1.2

0.6

(1.7)

(185.0)

(9.9)a

16.7

8.4

(18.2)

(118.2)

2.2

5.3

2.7

(0.4)

83.7

$38.3

$72.4

$36.2

$2.1

105.9%

b

-

-

-

-

51.3

-

-

$46.8

-

-

a

•

Noninterest income

48.5

-

-

-

-

•

Less: noninterest expense

53.0

-

-

-

-

Change in allowance for loan
and lease losses (ALLL)

$8.4

-

-

-

-

•

ALLL at 12/31/08

23.2

-

-

-

-

•

ALLL at 12/31/09

31.6

-

-

-

-

Source: GAO analysis of Federal Reserve and SNL Financial data.

Page 98

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

a

The trading and counterparty data in the Y-9C includes both customer derived revenue from
transactions for BHCs that operate as broker-dealers as well as gains and losses from proprietary
trading and associated expenses. These items are presented in net form only in the Y-9C. For the
five BHCs that had their trading portfolios stressed (including JPMorgan Chase & Co.), the trading
and counterparty line is based on projections of (gains) losses from proprietary trading, but PPNR
(specifically noninterest revenue) included projections of gains (losses) from customer derived
revenue from transactions due to operations as a broker-dealer. Because we could not segregate
these items based on the Y-9C, we have included the net amount in both the trading and
counterparty and noninterest income line items above. As a result of this limitation, the net amount of
the trading gains or losses and PPNR in the table may be overstated or understated.

b

PPNR includes an owned debt value adjustment of ($1.57) billion, which was not stressed in SCAP.
As JPMorgan Chase & Co.’s credit spreads narrowed during 2009, this caused the liability values to
increase. This offsets the gains JPMorgan Chase & Co. experienced in 2008 when its credit spreads
widened.

Page 99

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

Table 22: KeyCorp
Dollars in billions

Actual at 12/31/09

12/31/08 balance
per SCAP

Difference

12/31/09 as a percent
of the 12/31/08
balance

$11.0

$11.6

$(0.6)

94.4%

$6.4

$6.0

$0.4

107.4%

Risk-weighted assets

$85.9

$106.7

$(20.8)

80.5%

Tier 1 risk-based ratio

12.8%

10.9%

1.9%

117.0%

7.5%

5.6%

1.9%

133.9%

Actual for year
ended 12/31/09

2-year SCAP
estimate

GAO 1-year pro
rata estimate

Difference

Actual as a percent of
the pro rata estimate

$2.3

$6.7

$3.3

$(1.0)

69.3%

0.0

0.1

0.1

(0.0)

55.7

0.2

0.6

0.3

(0.1)

52.3

0.6

1.7

0.9

(0.2)

76.0

1.0

2.3

1.2

(0.2)

85.8

0.0

0.0

0.0

(0.0)

44.4

(0.1)

0.1

0.1

(0.2)

-225.7

-a

n/a

n/a

n/a

n/a

0.6

1.8

0.9

(0.3)

64.6

Tier 1 capital
Tier 1 common capital

Tier 1 common capital ratio

Total asset losses
•
First-lien mortgages
•

Second/junior lien
mortgages

•

Commercial and industrial
loans
Commercial real estate
loans

•
•

Credit card loans

•
•

Securities (available for
sale and held to maturity)
Trading and counterparty

•

Other

Resources other than capital
to absorb losses (total PPNR
less change in ALLL):

$0.1

$2.1

$1.1

$(1.0)

5.3%

PPNR
•
Net interest income
(expense)

$0.9

-

-

-

-

2.4

-

-

-

-

•

Noninterest income

a

1.8

-

-

-

-

•

Less: noninterest expense

3.3

-

-

-

-

Change in allowance for loan
and lease losses (ALLL)

$0.9

-

-

-

-

•

ALLL at 12/31/08

1.8

-

-

-

-

•

ALLL at 12/31/09

2.7

-

-

-

-

Source: GAO analysis of Federal Reserve and SNL Financial data.

Note: N/a means not applicable.

Page 100

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

a

Trading and counterparty positions were not stressed because the total portfolio is less than the
$100 billion required for stress testing in SCAP, but trading (gain) loss information for this BHC was
included in the “trading revenue” line of Schedule HI of the Y-9C in 2009. In SCAP, the projections of
trading gains or losses for this BHC were included in the estimate of PPNR rather than the trading
and counterparty line. Therefore, we have included the actual trading results in PPNR (specifically
noninterest income).

Page 101

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

Table 23: MetLife, Inc.
Dollars in billions

Difference

12/31/09 as a percent
of the 12/31/08
balance

Tier 1 capital

$28.8

$30.1

$(1.3)

95.6%

Tier 1 common capital

$26.4

$27.8

$(1.4)

94.8%

Risk-weighted assets

$322.8

$326.4

$(3.6)

98.9%

Tier 1 risk-based ratio

8.9%

9.2%

-0.3%

96.9%

Tier 1 common capital ratio

8.2%

8.5%

-0.3%

96.1%

Actual for year
ended 12/31/09

2-year SCAP
estimate

GAO 1-year pro
rata estimate

Difference

Actual as a percent of
the pro rata estimate

$1.7

$9.6

$4.8

$(3.1)

35.1%

0.0

0.0

0.0

(0.0)

24.0

0.0

0.0

0.0

(0.0)

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.8

0.4

(0.4)

6.9

12/31/08 balance
per SCAP
Actual at 12/31/09

Total asset losses
•
First-lien mortgages
•

Second/junior lien
mortgages

•

Commercial and industrial
loans
Commercial real estate
loans

•
•

Credit card loans

0.0

n/a

n/a

n/a

n/a

•

1.6

8.3

4.2

(2.5)

39.3

•

Securities (available for
sale and held to maturity)
Trading and counterparty

-a

n/a

n/a

n/a

n/a

•

Other

0.0

0.5

0.3

(0.2)

10.7

Resources other than capital
to absorb losses (total PPNR
less change in ALLL):

$(1.1)

$5.6

$2.8

$(3.9)

(38.9)%

PPNR
•
Net interest income
(expense)

$(0.7)

-

-

-

-

14.1

-

-

-

-

•

Noninterest income

a

29.4

-

-

-

-

•

Less: noninterest expense

44.1b

-

-

-

-

Change in allowance for loan
and lease losses (ALLL)

$0.4

-

-

-

-

•

ALLL at 12/31/08

0.3

-

-

-

-

•

ALLL at 12/31/09

0.7

-

-

-

-

Source: GAO analysis of Federal Reserve and SNL Financial data.

Note: N/a means not applicable.

Page 102

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

a

Trading and counterparty positions were not stressed because the total portfolio is less than the
$100 billion required for stress testing in SCAP, but trading (gain) loss information for this BHC was
included in the “trading revenue” line of Schedule HI of the Y-9C in 2009. In SCAP, the projections of
trading gains or losses for this BHC were included in the estimate of PPNR rather than the trading
and counterparty line. Therefore, we have included the actual trading results in PPNR (specifically
noninterest income).

b

MetLife, Inc. (MetLife) experienced high noninterest expense in 2009 largely due to derivative losses
from interest rate hedging, which protects MetLife against lower interest rates among other things.
Similar to the owned debt value adjustment, as MetLife’s credit spreads narrowed during 2009, this
caused the liability values to increase. This offsets the gains MetLife experienced in 2008 when its
credit spreads widened.

Page 103

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

Table 24: Morgan Stanley
Dollars in billions

Difference

12/31/09 as a percent
of the 12/31/08
balance

Tier 1 capital

$46.7

$47.2

$(0.5)

98.9%

Tier 1 common capital

$20.5

$17.8

$2.7

115.0%

Risk-weighted assets

$305.0

$310.6

$(5.6)

98.2%

Tier 1 risk-based ratio

15.3%

15.2%

0.1%

100.7%

6.7%

5.7%

1.0%

117.8%

Difference

Actual as a percent of
the pro rata estimate

12/31/08 balance
per SCAP
Actual at 12/31/09

Tier 1 common capital ratio

Actual for year
ended 12/31/09
Total asset losses
•
First-lien mortgages
•

Second/junior lien
mortgages

•

Commercial and industrial
loans
Commercial real estate
loans

•
•

Credit card loans

•
•

Securities (available for
sale and held to maturity)
Trading and counterparty

•

Other

Resources other than capital
to absorb losses (total PPNR
less change in ALLL):
PPNR
•
Net interest income
(expense)

2-year SCAP GAO 1-year pro
estimate
rata estimate

$(7.1)

$19.7

$9.8

$(16.9)

-72.7%

0.0

n/a

n/a

n/a

n/a

0.0

n/a

n/a

n/a

n/a

0.0

0.1

0.1

(0.0)

20.0

0.1

0.6

0.3

(0.2)

46.7

0.0

n/a

n/a

n/a

n/a

0.0

n/a

n/a

n/a

n/a

(7.3)a

18.7

9.4

(16.6)

-77.9

0.0

0.2

0.1

(0.1)

0.0

$1.0

$7.1

$3.6

$(2.5)

28.4%

b,c

-

-

-

-

0.9

-

-

-

-

a

$1.1

•

Noninterest income

22.7

-

-

-

-

•

Less: noninterest expense

22.5

-

-

-

-

Change in allowance for loan
and lease losses (ALLL)

$0.1

-

-

-

-

•

ALLL at 12/31/08

0.0

-

-

-

-

•

ALLL at 12/31/09

0.2

-

-

-

-

Source: GAO analysis of Federal Reserve and SNL Financial data.

Note: N/a means not applicable.

Page 104

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

a

The trading and counterparty data in the Y-9C includes both customer derived revenue from
transactions for BHCs that operate as broker-dealers as well as gains and losses from proprietary
trading and associated expenses. These items are presented in net form only in the Y-9C. For the
five BHCs that had their trading portfolios stressed (including Morgan Stanley), the trading and
counterparty line is based on projections of (gains) losses from proprietary trading, but PPNR
(specifically noninterest revenue) included projections of gains (losses) from customer derived
revenue from transactions due to operations as a broker-dealer. Because we could not segregate
these items based on the Y-9C, we have included the net amount in both the trading and
counterparty and noninterest income line items above. As a result of this limitation, the net amount of
the trading gains or losses and preprovision net revenue in the table may be overstated or
understated.

b

PPNR includes an owned debt value adjustment of ($5.30) billion, which was not included as a
stress in SCAP. As Morgan Stanley’s credit spreads narrowed during 2009, this caused the liability
values to increase. This offsets the gains Morgan Stanley experienced in 2008 when its credit
spreads widened.

c

PPNR includes one-time items totaling $710 million, which were not included in SCAP.

Page 105

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

Table 25: PNC Financial Services Group, Inc.
Dollars in billions

Difference

12/31/09 as a percent
of the 12/31/08
balance

Tier 1 capital

$26.5

$24.1

$2.4

110.1%

Tier 1 common capital

$13.9

$11.7

$2.2

119.2%

Risk-weighted assets

$232.3

$250.9

$(18.6)

92.6%

Tier 1 risk based ratio

11.4%

9.6%

1.8%

119.0%

6.0%

4.7%

1.3%

127.7%

Actual for year
ended 12/31/09

2-year SCAP
estimate

GAO 1-year pro
rata estimate

Difference

Actual as a percent of
the pro rata estimate

$2.7

$18.8

$9.4

$(6.6)

29.3%

12/31/08 balance
per SCAP
Actual at 12/31/09

Tier 1 common capital ratio

Total asset losses
•
First-lien mortgages

0.1

2.4

1.2

(1.1)

4.6

•

Second/junior lien
mortgages

0.4

4.6

2.3

(1.9)

19.4

•

Commercial and industrial
loans
Commercial real estate
loans

0.9

3.2

1.6

(0.7)

57.8

0.8

4.5

2.3

(1.5)

33.4

•

Credit card loans

0.2

0.4

0.2

(0.0)

85.6

•

0.0

1.3

0.7

(0.6)

4.2

•

Securities (available for sale
and held to maturity)
Trading and counterparty

-a

n/a

n/a

n/a

n/a

•

Other

0.4

2.3

1.2

(0.8)

31.4

$6.2

$9.6

$4.8

$1.4

128.6%

b

-

-

-

-

9.1

-

-

-

-

•

Resources other than capital
to absorb losses (Total PPNR
less change in ALLL):
PPNR
•
Net interest income
(expense)

$7.3

•

Noninterest income

7.9a

-

-

-

-

•

Less: noninterest expense

9.6

-

-

-

-

Change in allowance for loan
and lease losses (ALLL)

$1.2

-

-

-

-

•

ALLL at 12/31/08

3.9

-

-

-

-

•

ALLL at 12/31/09

5.1

-

-

-

-

Source: GAO analysis of Federal Reserve and SNL Financial data.

Note: N/a means not applicable.

Page 106

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

a

Trading and counterparty positions were not stressed because the total portfolio is less than the
$100 billion required for stress testing in SCAP, but trading (gain) loss information for this BHC was
included in the “trading revenue” line of Schedule HI of the Y-9C in 2009. In SCAP, the projections of
trading gains or losses for this BHC were included in the estimate of PPNR rather than the trading
and counterparty line. Therefore, we have included the actual trading results in PPNR (specifically
noninterest income).

b

PPNR includes one-time items totaling $1.08 billion, which were not included in SCAP.

Page 107

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

Table 26: Regions Financial Corporation
Dollars in billions
12/31/09 as a percent
of the 12/31/08
balance
Difference

12/31/08 balance
per SCAP
Actual at 12/31/09
Tier 1 capital

$11.9

$12.1

$(0.2)

98.5%

$7.4

$7.6

$(0.2)

97.2%

Risk-weighted assets

$103.3

$116.3

$(13.0)

88.8%

Tier 1 risk-based ratio

11.5%

10.4%

1.1%

111.0%

7.1%

6.6%

0.5%

108.3%

Actual for year
ended 12/31/09

2-year SCAP
estimate

GAO 1-year pro
rata estimate Difference

Actual as a percent
of the pro rata
estimate

$2.2

$9.2

$4.6

$(2.4)

48.8%

0.2

1.0

0.5

(0.3)

36.7

0.4

1.1

0.6

(0.2)

72.0

0.3

1.2

0.6

(0.3)

43.2

Tier 1 common capital

Tier 1 common capital ratio

Total asset losses
•
First-lien mortgages
•
•

Second/junior lien
mortgages
Commercial and industrial
loans

•

Commercial real estate
loans

1.1

4.9

2.5

(1.3)

46.7

•

Credit card loans

0.0

n/a

n/a

n/a

n/a

•

Securities (available for sale
and held to maturity)

(0.0)

0.2

0.1

(0.1)

(6.4)

a

n/a

n/a

n/a

n/a

0.3

0.8

0.4

(0.1)

67.9

$1.1

$3.3

$1.7

$(0.6)

64.9%

b

-

-

-

-

•

Trading and counterparty

•

Other

Resources other than capital
to absorb losses (total PPNR
less change in ALLL):
PPNR
•
Net interest income
(expense)

-

$2.4

3.3

-

-

-

-

•

Noninterest income

3.5a

-

-

-

-

•

Less: noninterest expense

4.5

-

-

-

-

Change in allowance for loan
and lease losses (ALLL)
•
ALLL at 12/31/08

$1.3

-

-

-

-

•

ALLL at 12/31/09

1.8

-

-

-

-

3.1

-

-

-

-

Source: GAO analysis of Federal Reserve and SNL Financial data.

Note: N/a means not applicable.

Page 108

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

a

Trading and counterparty positions were not stressed because the total portfolio is less than the
$100 billion required for stress testing in SCAP, but trading (gain) loss information for this BHC was
included in the “trading revenue” line of Schedule HI of the Y-9C in 2009. In SCAP, the projections of
trading gains or losses for this BHC were included in the estimate of PPNR rather than the trading
and counterparty line. Therefore, we have included the actual trading results in PPNR (specifically
noninterest income).

b

PPNR includes one-time items totaling $140 million, which were not included in SCAP.

Page 109

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

Table 27: State Street Corporation
Dollars in billions
12/31/09 as a percent
of the 12/31/08
balance
Difference

12/31/08 balance
per SCAP
Actual at 12/31/09
Tier 1 capital

$12.0

$14.1

$(2.1)

85.1%

Tier 1 common capital

$10.6

$10.8

$(0.2)

97.7%

Risk-weighted assets

$67.7

$69.6

$(1.9)

97.3%

Tier 1 risk-based ratio

17.7%

20.2%

(2.5)%

87.8%

Tier 1 common capital ratio

15.6%

15.5%

0.1%

100.6%

Difference

Actual as a percent
of the pro rata
estimate

Actual for year
ended 12/31/09
Total asset losses
•
First-lien mortgages
•
•

Second/junior lien
mortgages
Commercial and industrial
loans

2-year SCAP GAO 1-year pro
estimate
rata estimate

$(0.1)

$2.2

$1.1

$(1.2)

(4.7)%

0.0

n/a

n/a

n/a

n/a

0.0

n/a

n/a

n/a

n/a

0.0

0.0

0.0

(0.0)

0.0

•

Commercial real estate
loans

0.1

0.3

0.2

(0.1)

46.5

•

Credit card loans

0.0

n/a

n/a

n/a

n/a

•

Securities (available for sale
and held to maturity)

(0.1)

1.8

0.9

(1.0)

-15.6

a

n/a

n/a

n/a

n/a

•

Trading and counterparty

•

Other

•

One-time items in SCAPb

Resources other than capital
to absorb losses (total PPNR
less change in ALLL):

-

0.0

0.1

0.1

(0.0)

37.5

$6.1

$5.9

n/a

$0.2

103.4%

$2.5

$4.3

$2.2

$0.3

115.1%

$2.5

-

-

-

-

2.6

-

-

-

-

a

5.9

-

-

-

-

Less: noninterest expense

6.0

-

-

-

-

Change in allowance for loan
and lease losses (ALLL)
•
ALLL at 12/31/08

$0.1

-

-

-

-

0.0

-

-

-

-

0.1

-

-

-

-

PPNR
•
Net interest income
(expense)
•
Noninterest income
•

•

ALLL at 12/31/09

Source: GAO analysis of Federal Reserve and SNL Financial data.

Note: N/a means not applicable.

Page 110

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

a

Trading and counterparty positions were not stressed because the total portfolio is less than the
$100 billion required for stress testing in SCAP, but trading (gain) loss information for this BHC was
included in the “trading revenue” line of Schedule HI of the Y-9C in 2009. In SCAP, the projections of
trading gains or losses for this BHC were included in the estimate of PPNR rather than the trading
and counterparty line. Therefore, we have included the actual trading results in PPNR (specifically
noninterest income)

b

We broke out “other” losses into two categories—”Other” and “One-time items.” As discussed in
State Street Corporation’s (State Street) May 7, 2009, press release, $5.9 billion of the amount listed
in the “Other” category in the SCAP results was a pretax charge that was expected to occur when
certain asset-backed commercial paper conduits administered by State Street were consolidated onto
its balance sheet in 2009. Since this was a one-time charge that was realized in 2009, this effect was
segregated from more typical loss amounts for tracking purposes. Upon consolidation, the actual
amount realized was $6.1 billion, as reported in State Street’s Form 10-Q for the second quarter of
2009.

Page 111

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

Table 28: SunTrust Banks, Inc.
Dollars in billions

Difference

12/31/09 as a percent
of the 12/31/08
balance

Tier 1 capital

$18.1

$17.6

$0.5

102.7%

Tier 1 common capital

$10.7

$9.4

$1.3

113.7%

Risk-weighted assets

$139.4

$162.0

$(22.6)

86.0%

Tier 1 risk-based ratio

13.0%

10.9%

2.1%

118.9%

7.7%

5.8%

1.9%

132.3%

Actual for year
ended 12/31/09

2-year SCAP
estimate

$3.1

$11.8

$5.9

$(2.8)

53.1%

1.1

2.2

1.1

0.0

101.5

0.8

3.1

1.6

(0.8)

48.8

0.5

1.5

0.8

(0.2)

69.9

0.6

2.8

1.4

(0.8)

39.8

0.1

0.1

0.1

0.0

115.3

(0.1)

0.0

0.0

(0.1)

(980.2)

-a

n/a

n/a

n/a

n/a

0.2

2.1

1.1

(0.8)

21.6

$1.4

$4.7

$2.4

$(0.9)

61.3%

b,c

-

-

-

-

4.5

-

-

-

-

12/31/08 balance
per SCAP
Actual at 12/31/09

Tier 1 common capital ratio

Total asset losses
•
First-lien mortgages
•

Second/junior lien
mortgages

•

Commercial and industrial
loans
Commercial real estate
loans

•
•

Credit card loans

•
•

Securities (available for
sale and held to maturity)
Trading and counterparty

•

Other

Resources other than capital
to absorb losses (total PPNR
less change in ALLL):
PPNR
•
Net interest income
(expense)

$2.2

a

GAO 1-year pro
Actual as a percent of
rata estimate Difference the pro rata estimate

•

Noninterest income

3.6

-

-

-

-

•

Less: noninterest expense

5.9

-

-

-

-

Change in allowance for loan
and lease losses (ALLL)

$0.8

-

-

-

-

•

ALLL at 12/31/08

2.4

-

-

-

-

•

ALLL at 12/31/09

3.1

-

-

-

-

Source: GAO analysis of Federal Reserve and SNL Financial data.

Note: N/a means not applicable.

Page 112

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

a

Trading and counterparty positions were not stressed because the total portfolio is less than the
$100 billion required for stress testing in SCAP, but trading (gain) loss information for this BHC was
included in the “trading revenue” line of Schedule HI of the Y-9C in 2009. In SCAP, the projections of
trading gains or losses for this BHC were included in the estimate of PPNR rather than the trading
and counterparty line. Therefore, we have included the actual trading results in PPNR (specifically
noninterest income).

b

PPNR includes an owned debt value adjustment of ($150) million, which was not stressed in SCAP.
As SunTrust Banks Inc.’s credit spreads narrowed during 2009, this caused the liability values to
increase. This offsets the gains SunTrust experienced in 2008 when its credit spreads widened.

c

PPNR includes one-time items totaling $110 million, which were not included in SCAP.

Page 113

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

Table 29: U.S. Bancorp
Dollars in billions
12/31/09 as a percent
of the 12/31/08
balance
Difference

12/31/08 balance
per SCAP
Actual at 12/31/09
Tier 1 capital

$22.6

$24.4

$(1.8)

92.7%

Tier 1 common capital

$15.9

$11.8

$4.1

134.7%

Risk-weighted assets

$235.2

$230.6

$4.6

102.0%

Tier 1 risk-based ratio

9.6%

10.6%

(1.0)%

90.7%

Tier 1 common capital ratio

6.8%

5.1%

1.7%

132.5%

Actual for year
ended 12/31/09

2-year SCAP
estimate

GAO 1-year pro
rata estimate Difference

Actual as a percent
of the pro rata
estimate

$4.3

$15.7

$8.0

$(3.6)

54.3%

0.5

1.8

0.9

(0.4)

54.3

0.3

1.7

0.9

(0.5)

39.8

0.6

2.3

1.2

(0.6)

50.9

Total asset losses
•
First-lien mortgages
•
•

Second/junior lien
mortgages
Commercial and industrial
loans

•

Commercial real estate
loans

0.6

3.2

1.6

(1.0)

38.6

•

Credit card loans

1.0

2.8

1.4

(0.4)

73.6

•

Securities (available for
sale and held to maturity)

0.5

1.3

0.7

(0.2)

69.4

a

n/a

n/a

n/a

n/a

0.8

2.8

1.4

(0.6)

57.7

$7.1

$13.7

$6.9

$0.2

103.3%

PPNR
•
Net interest income
(expense)

$8.6

-

-

-

-

8.5

-

-

-

-

•

Noninterest income

8.4a

-

-

-

-

•

Less: noninterest expense

8.3

-

-

-

-

Change in allowance for loan
and lease losses (ALLL)
•
ALLL at 12/31/08

$1.6

-

-

-

-

3.5

-

-

-

-

5.1

-

-

-

-

•

Trading and counterparty

•

Other

Resources other than capital
to absorb losses (total PPNR
less change in ALLL):

•

ALLL at 12/31/09

-

Source: GAO analysis of Federal Reserve and SNL Financial data.

Note: N/a means not applicable.

Page 114

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

a

Trading and counterparty positions were not stressed because the total portfolio is less than the
$100 billion required for stress testing in SCAP, but trading (gain) loss information for this BHC was
included in the “trading revenue” line of Schedule HI of the Y-9C in 2009. In SCAP, the projections of
trading gains or losses for this BHC were included in the estimate of PPNR rather than the trading
and counterparty line. Therefore, we have included the actual trading results in PPNR (specifically
noninterest income).

Page 115

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

Table 30: Wells Fargo & Company
Dollars in billions

Difference

12/31/09 as a percent
of the 12/31/08
balance

$93.8

$86.4

$7.4

108.6%

a

$65.5

$33.9

$31.6

193.2%

Risk-weighted assets

$1,013.6

$1,082.3

$(68.7)

93.7%

Tier 1 risk-based ratio

9.3%

8.0%

1.3%

115.7%

Tier 1 common capital ratio

6.5%

3.1%

3.4%

208.4%

Actual for year
ended 12/31/09

2-year SCAP
estimate

GAO 1-year pro
rata estimate

Difference

Actual as a percent of
the pro rata estimate

$18.0

$86.1

$43.1

$(25.1)

41.7%

3.0

32.4

16.2

(13.2)

18.4

4.9

14.7

7.4

(2.5)

66.1

2.8

9.0

4.5

(1.7)

61.5

1.5

8.4

4.2

(2.7)

35.8

2.6

6.1

3.1

(0.5)

83.7

(0.2)

4.2

2.1

(2.3)

(9.8)

-b

n/a

n/a

n/a

n/a

3.5

11.3

5.7

(2.1)

62.1

12/31/08 balance
per SCAP
Actual at 12/31/09
Tier 1 capital
Tier 1 common capital

Total asset losses
•
First-lien mortgages
•

Second/junior lien
mortgages

•

Commercial and industrial
loans
Commercial real estate
loans

•
•

Credit card loans

•
•

Securities (available for
sale and held to maturity)
Trading and counterparty

•

Other

Resources other than capital
to absorb losses (total PPNR
less change in ALLL):

$36.1

$60.0

$30.0

$6.1

120.5%

PPNR
•
Net interest income
(expense)

$39.6

-

-

-

-

46.9

-

-

-

-

•

Noninterest income

b

41.5

-

-

-

-

•

Less: noninterest expense

48.8

-

-

-

-

Change in allowance for loan
and lease losses (ALLL)

$3.5

-

-

-

-

•

ALLL at 12/31/08

21.0

-

-

-

-

•

ALLL at 12/31/09

24.5

-

-

-

-

Source: GAO analysis of Federal Reserve and SNL Financial data.

Note: N/a means not applicable.

Page 116

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix III: One-Year Actual Performance
Compared to GAO’s Pro rata Stress Test Loss
Projections for Each of the 19 SCAP BHCs

a

The tier 1 common calculation has been adjusted to provide for appropriate treatment of preferred
shares Wells Fargo & Company (Wells Fargo) issued as a part of its Employee Stock Ownership Plan
(ESOP). Each share of ESOP preferred stock released from the unallocated reserve of the 401(k)
plan is converted into shares of Wells Fargo’s common stock based on the stated value of the ESOP
preferred stock and the current market price of Wells Fargo’s common stock. Wells Fargo sells ESOP
preferred stock to its 401(k) plan and lends the 401(k) plan cash to purchase those shares. The loan
is recorded as “Unearned ESOP Preferred Shares.” While the ESOP preferred shares are counted as
an addition to equity, the loans recorded as Unearned ESOP Preferred Shares are treated as a
reduction to equity, and so there is no net impact on the equity accounts (including tier 1 capital).
However, the tier 1 common capital calculation removes the ESOP preferred shares without also
removing the corresponding loans recorded as Unearned ESOP Preferred Shares. After consulting
with Wells Fargo, GAO adjusted the tier 1 common capital calculation by removing the $442 million of
Unearned ESOP Preferred Shares outstanding as of December 31, 2009 (the Unearned ESOP
Preferred Shares is a negative amount; thus, removing this item leads to the addition of $442 million
in tier 1 capital), which is consistent with SCAP’s treatment.

b

Trading and counterparty positions were not stressed because the total portfolio is less than the
$100 billion required for stress testing in SCAP, but trading (gain) loss information for this BHC was
included in the “trading revenue” line of Schedule HI of the Y-9C in 2009. In SCAP, the projections of
trading gains or losses for this BHC were included in the estimate of PPNR rather than the trading
and counterparty line. Therefore, we have included the actual trading results in PPNR (specifically
noninterest income).

Page 117

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix IV: Comments from the Board of
Governors of the Federal Reserve System

Appendix IV: Comments from the Board of
Governors of the Federal Reserve System

Page 118

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix IV: Comments from the Board of
Governors of the Federal Reserve System

Page 119

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix IV: Comments from the Board of
Governors of the Federal Reserve System

Page 120

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix V: Comments from the Department
of the Treasury’s Office of Financial Stability

Appendix V: Comments from the Department
of the Treasury’s Office of Financial Stability

Page 121

GAO-10-861 Lessons Learned from Bank Stress Test

Appendix VI: GAO Contact and Staff
Acknowledgments

Appendix VI: GAO Contact and Staff
Acknowledgments
GAO Contact

Orice Williams Brown, (202) 512-8678 or williamso@gao.gov

Staff
Acknowledgments

Daniel Garcia-Diaz (Assistant Director), Michael Aksman, Emily Chalmers,
Rachel DeMarcus, Laurier Fish, Joe Hunter, William King, Matthew
McDonald, Sarah M. McGrath, Timothy Mooney, Marc Molino, Linda Rego,
and Cynthia Taylor made important contributions to this report.

(250507)

Page 122

GAO-10-861 Lessons Learned from Bank Stress Test

GAO’s Mission

The Government Accountability Office, the audit, evaluation, and
investigative arm of Congress, exists to support Congress in meeting its
constitutional responsibilities and to help improve the performance and
accountability of the federal government for the American people. GAO
examines the use of public funds; evaluates federal programs and policies;
and provides analyses, recommendations, and other assistance to help
Congress make informed oversight, policy, and funding decisions. GAO’s
commitment to good government is reflected in its core values of
accountability, integrity, and reliability.

Obtaining Copies of
GAO Reports and
Testimony

The fastest and easiest way to obtain copies of GAO documents at no cost
is through GAO’s Web site (www.gao.gov). Each weekday afternoon, GAO
posts on its Web site newly released reports, testimony, and
correspondence. To have GAO e-mail you a list of newly posted products,
go to www.gao.gov and select “E-mail Updates.”

Order by Phone

The price of each GAO publication reflects GAO’s actual cost of
production and distribution and depends on the number of pages in the
publication and whether the publication is printed in color or black and
white. Pricing and ordering information is posted on GAO’s Web site,
http://www.gao.gov/ordering.htm.
Place orders by calling (202) 512-6000, toll free (866) 801-7077, or
TDD (202) 512-2537.
Orders may be paid for using American Express, Discover Card,
MasterCard, Visa, check, or money order. Call for additional information.

To Report Fraud,
Waste, and Abuse in
Federal Programs

Contact:

Congressional
Relations

Ralph Dawn, Managing Director, dawnr@gao.gov, (202) 512-4400
U.S. Government Accountability Office, 441 G Street NW, Room 7125
Washington, DC 20548

Public Affairs

Chuck Young, Managing Director, youngc1@gao.gov, (202) 512-4800
U.S. Government Accountability Office, 441 G Street NW, Room 7149
Washington, DC 20548

Web site: www.gao.gov/fraudnet/fraudnet.htm
E-mail: fraudnet@gao.gov
Automated answering system: (800) 424-5454 or (202) 512-7470

Please Print on Recycled Paper